Big Data Surveillance and Security Intelligence

This page intentionally left blank

Big Data Surveillance and Security Intelligence
The Canadian Case

Edited by David Lyon and David Murakami Wood

 © UBC Press 2021
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, without prior written permission of
the publisher.
Library and Archives Canada Cataloguing in Publication
Title: Big data surveillance and security intelligence : the Canadian case /
  edited by David Lyon and David Murakami Wood.
Names: Lyon, David, editor. | Wood, David Murakami, editor.
Description: Includes bibliographical references and index.
Identifiers: Canadiana (print) 20200333143 | Canadiana (ebook) 20200333887 |
  ISBN 9780774864176 (hardcover) | ISBN 9780774864190 (PDF) |
  ISBN 9780774864206 (EPUB)
Subjects: LCSH: Electronic surveillance – Canada. | LCSH: Intelligence service – Canada. |
  LCSH: Cyber intelligence (Computer security) – Canada. | LCSH: National security –
  Canada. | LCSH: National security – International cooperation. | LCSH: Data
  protection – Canada.
Classification: LCC JL86.I58 B54 2020 | DDC 327.1271—dc23

UBC Press gratefully acknowledges the financial support for our publishing program of
the Government of Canada (through the Canada Book Fund) and the British Columbia
Arts Council.
Set in Helvetica Condensed and Minion by Apex CoVantage, LLC
Copy editor: Francis Chow
Proofreader: Judith Earnshaw
Indexer: Delano Aragão Vaz
Cover designer: Alexa Love
UBC Press
The University of British Columbia
2029 West Mall
Vancouver, BC V6T 1Z2
www.ubcpress.ca

Lyon & Murakami Wood_18164-0090q4-4Pass - e-book.indd 4 10/17/2020 10:40:59 AM

 Contents

List of Figures and Tables / viii

Preface / ix
List of Abbreviations / x
Introduction / 3
David Lyon and David Murakami Wood

Part 1: Understanding Surveillance, Security, and Big Data

1 Collaborative Surveillance with Big Data Corporations:

Interviews with Edward Snowden and Mark Klein / 21
Midori Ogasawara

2 On Denoting and Concealing in Surveillance Law / 43

Christopher Prince

3 Big Data against Terrorism / 57

Stéphane Leman-Langlois

4 Algorithms as Suspecting Machines: Financial Surveillance

for Security Intelligence / 68
Anthony Amicelle and David Grondin

Part 2: Big Data Surveillance and Signals Intelligence

in Canadian Security Organizations

5 From 1967 to 2017: The Communications Security Establishment’s

Transition from the Industrial Age to the Information Age / 89
Bill Robinson
 vi Contents

6 Pixies, Pop-Out Intelligence, and Sandbox Play: The New Analytic

Model and National Security Surveillance in Canada / 112
Scott Thompson and David Lyon

7 Limits to Secrecy: What Are the Communications Security

Establishment’s Capabilities for Intercepting Canadians’ Internet
Communications? / 126
Andrew Clement

Part 3: Legal Challenges to Big Data Surveillance in Canada

8 Gleanings from the Security Intelligence Review Committee about

the Canadian Security Intelligence Service’s Bulk Data Holdings
and the Bill C-59 “Solution” / 149
Micheal Vonn

9 Bill C-59 and the Judicialization of Intelligence Collection / 167

Craig Forcese

10 The Challenges Facing Canadian Police in Making Use of Big Data

Analytics / 180
Carrie B. Sanders and Janet Chan

Part 4: Resistance to Big Data Surveillance

11 Confronting Big Data: Popular Resistance to Government

Surveillance in Canada since 2001 / 197
Tim McSorley and Anne Dagenais Guertin

12 Protesting Bill C-51: Reflections on Connective Action against Big

Data Surveillance / 218
Jeffrey Monaghan and Valerie Steeves

Part 5: Policy and Technical Challenges of

Big Data Surveillance

13 Horizontal Accountability and Signals Intelligence: Lessons

Drawing from Annual Electronic Surveillance Reports / 237
Christopher Parsons and Adam Molnar
 Contents vii

14 Metadata – Both Shallow and Deep: The Fraught Key to Big Data
Mass State Surveillance / 253
Andrew Clement, Jillian Harkness, and George Raine

Afterword / 269
Holly Porteous
Contributors / 275
Index / 281
Figures and Tables

Figures
1.1 New Collection Posture NSA slide / 22
7.1 Canadian cyber sensor grid / 133
7.2 Boomerang route originating and terminating in Toronto / 136
14.1 “Context” as metadata category in XKEYSCORE / 261

Tables
7.1 Top border cities for Internet traffic entering Canada / 137
7.2 Top carriers bringing Internet data into Canada / 137
7.3 Principal concentrations of Internet routers by metropolitan area
and carrier / 139
Preface

This book examines one of the most pressing issues in the organization of
society today: the application of new data practices to both global systems and
everyday devices – in this case, those that have to do with national security. The
editors have been involved in research on surveillance in relation to security
for over two decades, but it is difficult to think of innovations as far-reaching
and consequential as those involving so-called big data. This is because they
rest on novel ways of using data, enabled by massive computing power, and
because they touch the lives of everyone. Intelligence gathering in the era of
social media and the Internet of Things cannot but implicate and involve all
citizens, domestically as well as in other countries, not just those conventionally
thought of as suspicious, risky, or threatening.
The book is the product of a large-scale research project involving several
Canadian universities as well as partners and collaborators in other countries,
with its central node at the Surveillance Studies Centre at Queen’s University
in Kingston, Ontario. Other lead researchers are from Université Laval, the
University of Ottawa, and the University of Victoria, along with St. Andrew’s
University in Scotland. This project is underwritten by a five-year Partnership
Grant from the Social Sciences and Humanities Research Council (SSHRC), in
which academic researchers work collaboratively with partners from both public
policy and civil society. More specifically, it comprises expert papers shared at
a very stimulating and fruitful research workshop held in Ottawa, in which both
academics and members of privacy commissions and civil liberties groups
contributed papers for discussion.
We are very grateful to the SSHRC both for the funding opportunity and for
the doors it opened for our research to be carried out in active cooperation with
our Canadian government policy and compliance bodies (the Office of the
Privacy Commissioner in Ottawa and the British Columbia Office of the Infor-
mation and Privacy Commissioner) and non-governmental organization
partners (the International Civil Liberties Monitoring Group in Ottawa and the
BC Civil Liberties Association). These partnerships are evident in the chapters
of this book.
Abbreviations

ARPANET Advanced Research Projects Agency Network

ASD Australian Signals Directorate
ATIP Access to Information and Privacy
BCCLA British Columbia Civil Liberties Association
BCOIPC British Columbia Office of the Information and Privacy
Commissioner
BRUSA Britain-USA [Agreement]
CANUSA Canada-USA [Agreement]
CBNRC Communications Branch of the National Research Council
CNE computer network exploitation
COMINT COMmunications INTelligence
COMSEC Communications Security (part of the CBNRC)
COTC Canadian Overseas Telecommunications Corporation
CRA Canada Revenue Agency
CRTC Canadian Radio-television and Telecommunications
Commission
CSE Communications Security Establishment
CSIS Canadian Security Intelligence Service
DARPA Defense Advanced Research Projects Agency
DEW Defense Early Warning
DGI Director General Intelligence
DNI Digital Network Intelligence
DNR Dial Number Recognition
DPI deep packet inspection
EFF Electronic Frontier Foundation
FINTRAC Financial Transactions and Reports Analysis Centre of
Canada
FISC Foreign Intelligence Surveillance Court
FLQ Front de libération du Québec
FVEY Five Eyes
GAC Global Affairs Canada
 Abbreviations xi

GCHQ Government Communications Headquarters

GCSB Government Communications Security Bureau (NZ)
HUMINT HUMan INTelligence
IAC Intelligence Advisory Committee
ICAMS International Campaign Against Mass Surveillance
ICLMG International Civil Liberties Monitoring Group
ICT information and communications technology
ILC International Licensed Carriers
ILP intelligence-led policing
IMINT IMagery INTelligence
ISIL Islamic State of Iraq and the Levant
ISP Internet service provider
LESAs law enforcement and security agencies
LTAP Long-Term Accommodation Project
MITACS Mathematics of Information Technology and Complex Systems
consortium
NAE New Analytic Environment
NAM New Analytic Model
NDA National Defence Act
NIM National Intelligence Model
NSA National Security Agency (US)
NSICOP National Security and Intelligence Committee of
Parliamentarians
NSIRA National Security and Intelligence Review Agency
OCSEC Office of the Communications Security Establishment
Commissioner
ODAC Operational Data Analysis Centre
OPC Office of the Privacy Commissioner
PCLOB Privacy and Civil Liberties Oversight Board
PIPEDA Personal Information Protection and Electronic Documents Act
SCISA Security of Canada Information Sharing Act
SIGINT SIGnals INTelligence
SIRC Security Intelligence Review Committee
SIS Secret Intelligence Service (MI6)
SLA specialized lawful access
SMO Social Movement Organization
SOCMINT SOCial Media INTelligence
SOIA Security of Information Act
xii Abbreviations

SSO Special Source Operations

TIA Total (later, Terrorism) Information Awareness
TIMC Tutte Institute for Mathematics and Computing
TSP telecommunications service provider
UKUSA UK-USA [Agreement]
Big Data Surveillance and Security Intelligence
This page intentionally left blank
Introduction
David Lyon and David Murakami Wood

“Everything has changed” in the wake of big data, declared the Canadian
Communications Security Establishment (CSE).1 While some more skeptical
analysts may raise an eyebrow at this, it is undoubtedly the case that the modes
of analysis of communications described as “big data” have produced huge
changes in the pursuit of national security. Far from being an exclusively Can-
adian phenomenon, this is a global process in which agencies worldwide work
together using new methods of data analytics to try to identify risks to national
security with a view to preventing or pre-empting the possibility that those risks
might become realities. The title of our book, Big Data Surveillance and Security
Intelligence: The Canadian Case, thus sums up a crucially important trend, one
that calls for serious and critical analysis.
While this process is global, one cannot hope to grasp the import and impact
of what is happening by attempting to capture the complete global scene.
Thus, this book is primarily about what is happening in Canada, to follow
how the big data changes came about in one country. Of course, this pulls us
right back into the broader picture because the Canadian experience is deeply
influenced by others, especially through the so-called Five Eyes partnership
of Australia, Canada, New Zealand, the United Kingdom, and the United
States. But the case of Canada remains central to what follows. It is not at all
identical to the experience of others and contains important markers for
ongoing analysis.
First, it is vital that the two terms “security intelligence” and “surveillance”
appear together. The former, a growing activity of any nation-state, requires the
latter, which has to do with the purposeful collection and examination of per-
sonal data. Of course, all manner of information handling may be carried out
for national security, but the aim is to discover human activities potentially
detrimental to the proper functioning of the nation-state. One way this occurs
is through open-source data gathering, changes in which indicate the enor-
mously enlarged scope of personal data gathering and sifting in the twenty-first
century. During the Second World War, “open-source intelligence” referred to
the monitoring of enemy radio stations for clues about hostile activities. Today
it has expanded exponentially to include the monitoring of the Internet and
especially social media. This is the primary origin of the tremendous troves of
data referred to as “big.”
4 David Lyon and David Murakami Wood

Second, as the following chapters indicate, parallel linked trends are apparent
in national security intelligence. One is the expansion of the notion of security
to cover a range of fields not previously designated as such. In the field of
international relations, “securitization” refers to the designation by govern-
ments of their citizens and those of other countries as matters of security. This
means that extraordinary means – rendition to other countries for questioning
or even torture, for example, as happened to Maher Arar and other Canadians
after the terrorist attacks of 11 September 2001 – may be used in the name of
security. Not only persons but also certain kinds of events, such as musical or
athletic activities, or sites, such as sidewalks with no vehicle barriers, may also
be newly considered as security risks. The second trend is the one already
alluded to, of gathering data from sources that only recently – in the twenty-
first century – have become available.
Each of these trends underscores the importance of discussing the activ-
ities of communications security and security intelligence in Canada and
around the world. And they also point to the need for a deeper understand-
ing of the ways in which surveillance has become such an essential player
in each area, along with others, such as domestic policing, that are also
securitized and also increasingly depend on big data.2 As far as national
security is concerned, the big data connections started to become publicly
clear with the whistle-blowing activities of American security operatives
such as William Binney, Thomas Drake, Mark Klein, and, especially since
2013, Edward Snowden. What they demonstrated was both the widened
sweep of surveillance – often called “mass surveillance” – and the profound
dependence of the national security agencies of all countries on telecom-
munications and Internet corporations for the acquisition and then the
promiscuous use of personal data.
Third, then, is the notion of a big data age. While it may be premature or
misleading to adopt big data as a key descriptor for an historical era, it is none-
theless essential to recognize and assess the impact of big data practices in
multiple areas of corporate and governmental life. If one turns to Google’s
Internet dictionary for a definition of big data, you may find this: “extremely
large data sets that may be analyzed computationally to reveal patterns, trends,
and associations, especially relating to human behavior and interactions.” This
is very instructive because it speaks not only to the volume of data but also to
the dependence on massive computer power, the production of correlations
between disparate factors, and the predominant focus on human activity. What
is missing is a sense of the enthusiasm with which big data “solutions” are often
sought, and the relative lack of critical reflection on the limits on and ethics of
what can be achieved using such methods.
 Introduction 5

The “big data” buzzword may have a limited shelf life but what it points to is
highly significant. At CSE, the relevant phrase is “New Analytic Model” (Chapter 6).
The key idea is to take advantage of the availability of rapidly growing quantities
of data available through the Internet but especially consequent on the rise of
social media. The fact that the platforms also learned to monetize what was
previously referred to as “data exhaust” also meant that such data were sought
more vigorously. Rather than relying on conventional modes of analysis, we
can now mine and analyze such data, using algorithms, to discover previously
unnoticed patterns of activity. While big data is often summarized by attributes
such as volume (of data), velocity (speed of analysis), and variety (the expanding
range of usable datasets), its core is practices. As danah boyd and Kate Crawford
note, it is the “capacity for researchers to search, aggregate and cross-reference
large data-sets.” 3 They also ask some telling questions about such data
practices.
It is hardly surprising, however, that security intelligence services would wish
to exploit new possibilities for learning from the mass of metadata that actually
answers queries a private detective agency might have – such as location, time,
and type of transaction and communication, along with identifying details of
participants. From about 2009 in Canada, it became clear that legal warrant
was sought for such data gathering and analysis. However, as hinted at in the
designation of big data as a buzzword, hard ontological and epistemological
questions are easily glossed over. Such practices are all too often marred by what
Jose van Dijck calls “dataism,” a secular belief in the “objectivity of quantification
and the potential for tracking human behaviour and sociality through online
data,” along with the presentation of such data as “raw material” to be analyzed
and processed into predictive algorithms.4 The potential for new analytics and
the potential problems that arise from this are discussed throughout this book.
It is also worth noting, however, that the processes involved are ones that are
also visible in everyday life, not only in the arcane world of national security
intelligence and surveillance. After all, the data concerned are frequently gleaned
from what David Lyon calls “user-generated surveillance,” referring to the ways
in which contemporary new media encourage the sharing of information and
images and the proliferation of online communications and transactions that
yield highly consequential metadata.5 This parallels Bernard Harcourt’s account,
which explores processes of quotidian self-exposure, where people “follow”
others, “sharing” information as they roam the web and as they themselves are
“followed” by multiple commercial and governmental organizations that obtain
access to the same data.6 The corporate use of such data has now reached such
major proportions that theorists such as Shoshana Zuboff now dub this “surveil-
lance capitalism,” a process that has far-reaching and profound political,
6 David Lyon and David Murakami Wood

economic, and social repercussions. This is the crucial context for the present
study.7
The words that form the title of our book sum up what its authors are at pains
to reveal as they debate the big data surveillance involved in today’s security
measures, seen especially in intelligence gathering and analysis dependent on
big data practices. The book does not question the need for forms of national
security or policing as such, but it does raise many questions about the ways in
which new kinds of data production and analysis challenge conventional and
time-honoured practices within such agencies. The book queries the rapid and
wholesale departure from earlier traditions of national security activity, along
with the ethics and legal instruments that have governed these areas in the past,
but it does so with a view to the future of national security, not with a nostalgic
view of what went before. After all, during the twentieth century many questions
were also raised about national security and policing measures. There is a long
history of reasoned and principled assessment of such endeavours, to which
this book makes a further contemporary contribution.

Situating Security Intelligence

Security intelligence encompasses a large number of diverse state activities.
Broadly speaking, we are talking about espionage, or spying, although much of
what takes place under this banner is far from the popular images of Cold War
espionage, whether John le Carré’s cynical British Secret Intelligence Service
(SIS, or MI6) spooks or the antics of Mad magazine’s “Spy vs. Spy.” Such
HUMINT (HUMan INTelligence) operations remain important, but with the
rise of, first, telecommunications, and then computing, and their now inescap-
able and ubiquitous combination, HUMINT has increasingly been overtaken
by SIGINT (SIGnals INTelligence) and associated specialist domains, including
COMINT (COMmunications INTelligence), IMINT (IMagery INTelligence),
and SOCMINT (SOCial Media INTelligence).
Much security intelligence activity is conducted in secret, and for long periods
of the history of security intelligence this secrecy has been felt to be both neces-
sary and inevitable. There have been occasional voices raised, questioning or
criticizing such assumptions and what Clive Ponting, in the context of the British
state, called the “culture of secrecy.”8 Therefore, only some of what we know of
security intelligence agencies and their history and activities has come via
official channels in often belated formal releases of historical information (after
thirty, fifty, or seventy years). Some more has come through the Canadian Access
to Information Act (ATIA) or the British and American Freedom of Information
Acts (FOIA). As we have already noted, however, much of the most significant
revelations have come through the actions of activists and whistle-blowers in
 Introduction 7

acquiring and publicizing information that would otherwise have remained

hidden.9
While espionage dates back to the beginnings of government, there are some
generally accepted milestones in the history of Anglo-American intelligence,
into which the Canadian system fits. A dedicated HUMINT office was first set
up under Francis Walsingham during the reign of Elizabeth I in England, but
such initiatives remained both sporadic and specific in their targeting (Wal-
singham’s efforts were mainly targeted at the Catholic threat to England’s nascent
Protestant regime). State intelligence activity gradually increased, especially
during the long period of warfare in Europe in the eighteenth century and into
the Napoleonic Wars of the early nineteenth century. The eighteenth and nine-
teenth centuries also saw the emergence of internal political policing with the
“new police,” whose roles included the breaking up of reform meetings and
demonstrations and the “moving on” of the urban poor and homeless. The
nineteenth century also saw the first British legislative interest in the targeting
of emerging communications systems by state surveillance. Of particular note
was the 1844 scandal in which an Italian exile, Joseph Mazzini, who was resident
in London, discovered that the British government was secretly opening his
mail, prompting parliamentary discussion and demands for accountability and
reform.10
The story of intelligence in the Anglo-American world cannot be treated
separately from settler-colonial control and policing. It was in India, for example,
where British colonial officers pioneered the use of fingerprinting and databases
of personal biometric information for identification. This was mainly to ensure
prompt payment of taxes imposed by the British Raj, but the use of fingerprints
to combat resistance and insurgency as well as more conventional crimes fol-
lowed. Fingerprinting spread from India to Ireland, Britain’s nearest colony, and
then to the imperial metropole of London and other major cities in the United
States and in the Dominion of Canada.
Early security concerns in what would become Canada centred on the
conflict between France and the United Kingdom over territory that belonged
to neither but was claimed by both, as well as on Indigenous peoples, whose
land it was and who continued in many cases to fight against the tide of settler-
colonialism. Within British colonial territory, the additional internal threat
of the Fenians, Irish republicans seeking to free their country from British
rule, became an additional concern, and as the nineteenth century progressed,
strongly racialized concerns about immigration from China and India, as well
as freed African slaves from the United States and Canada, moved to the fore.
However, there was hardly anything resembling a security intelligence appar-
atus at this time.
8 David Lyon and David Murakami Wood

Meanwhile, in the United States arguments about the nation’s own potential
overseas empire beyond the frontier that had its terminus at the Pacific appeared
to have been settled almost by circumstance with the results of the Spanish-
American War and the US acquisition of the Philippines. The United States
instituted an intensive military governance program that conflated both crime
and terrorism and saw a “solution” to the problem in counter-insurgency oper-
ations.11 Counter-insurgency thereafter proved a rich vein for new intelligence
services that were created in the twentieth century, notably the Army Intelligence
Agency (later the Defense Intelligence Agency) and the Central Intelligence
Agency (CIA), deployed with debatable success in Europe after the Second
World War via the Gladio network, and over a longer period in Latin America,
following the Monroe Doctrine, which asserted US hegemony over the American
hemisphere. The tactic of essentially manufacturing threats that could be dis-
rupted was also the modus operandi of the Federal Bureau of Investigation
(FBI) on the US mainland.
The postwar period in the United States saw the migration and ascent of
security intelligence from a fringe activity often run largely by enthusiasts and
amateurs to the core of the state during the Cold War, with intelligence agencies – as
expert bureaucracies with the greatest access to secret information, burgeoning
budgets, and the least accountability to elected bodies – constituting the core
of what has been variously called the permanent, secret, or deep state.
With Britain’s economy grievously damaged by the war and its imperial power
broken, the postwar period also saw the United States confirmed as the new
centre of the anglophone world, and military and security intelligence arrange-
ments were among the first to recognize this new reality. Conventional histories
still emphasize NATO cooperation. This was (and is) undoubtedly important,
but in the security intelligence field there is no doubt that secret and unacknow-
ledged accords have long marked the new reality. The signing of the Britain-USA
Agreement (BRUSA) in 1943 set the stage for the UK-USA Agreement (UKUSA)
in 1946, which confirmed American dominance over the still extensive British
imperial intelligence networks. The United Kingdom was vital not just for its
personnel and expertise but also because one major legacy of British imperial
power was its centrality in international undersea cable networks. Any SIGINT
agency with global aspirations was going to want the ability to tap those cables,
and the integration of UKUSA agencies into existing and new cable and domestic
telecommunications systems was a priority from this time.
The United States made it a priority to bring in British white settler-colonial
dominions (Canada, Australia, and New Zealand). For SIGINT, a Canada-USA
Agreement (CANUSA) was signed in 1948, and Australia and New Zealand’s
intelligence services (originally as a single entity) were brought into the emerging
 Introduction 9

system in 1956. These were the founding “second parties” to what later became
known as the Quadripartite Agreement and later, more informally, the Five
Eyes (FVEY), with the United States as “first party.” The reconstruction of post-
war intelligence agencies in the defeated former Axis powers (West Germany,
Italy, and Japan in particular) was also achieved with US and Allied oversight
(often involving many of the same people who had worked for the former fascist
security intelligence agencies), and the new security intelligence agencies became
“third parties” in the emerging US-dominated international security intelligence
system. Other eventual third parties included members of NATO and countries
propped up with US military aid during the Cold War, such as Thailand and
Israel.
The postwar period also saw the rapid rise of SIGINT, facilitated by both this
international system and the spread of existing technologies like the telephone
and new technologies of computing. The historian of intelligence Bernard Porter
has argued that SIGINT became increasingly central from the 1950s onward,
and that the “the future of intelligence clearly lay with ‘Sigint’ and the new
technology.”12
Compared with the United States and the United Kingdom, Canadian par-
ticipation was relatively insignificant during this period. Canadian universities
contributed to CIA research projects in psychological manipulation and torture.
Canadian SIGINT did not really become important until the emergence of the
National Research Bureau in the 1960s and its eventual more public identity as
the Communications Security Establishment of Canada (CSEC, latterly CSE).13
For much of this time, the most important Canadian contribution consisted of
hosting multiple tiny Arctic installations that formed part of what was called
the DEW (Defense Early Warning) Line, designed to protect the United States,
and to a lesser extent Canada and Britain, against Soviet long-range bomber
and later ICBM (intercontinental ballistic missile) attack, and its successors.14
While Canada struggled in many ways to establish a meaningful international
security intelligence role independent of the United States, internally it adopted
tactics that were strongly reminiscent of the FBI’s domestic version of counter-
insurgency. The Royal Canadian Mounted Police (RCMP), by global standards
a rather strange hybrid policing and internal security organization more akin
to London’s Metropolitan Police Service – with its dual local/regional conven-
tional policing and national anti-terrorism/internal security roles – than to any
other organizational equivalent, was infamously revealed to have been involved
in multiple illicit activities in order to uncover enemies within.15
This sparked a national inquiry, the McDonald Commission (Royal Com-
mission of Inquiry into Certain Activities of the RCMP), which led to the
removal of the RCMP’s internal security intelligence role and the creation of
10 David Lyon and David Murakami Wood

the Canadian Security Intelligence Service (CSIS) in 1984. However, the post-
9/11 environment has once again muddied the never entirely clear blue waters
that separate CSIS’s security intelligence role from the RCMP’s policing mandate.
As CSIS has grown in power and influence, pressure has grown for it to have
more active capabilities to – in the words of the National Security Act, 2017
(formerly Bill C-59) – “disrupt” terrorist plots and so on. In addition, the RCMP
still maintains what can only be described as a political policing role, akin to
the Special Branch of London’s Metropolitan Police, targeting what it describes
as “domestic” or “multi-issue extremists” (who include environmental activists,
Indigenous peoples’ rights organizations, Quebecois separatists, and more).16
As a result, CSIS has developed its own extensive databases, in line with the US
Department of Homeland Security’s “fusion centers,” created after 9/11. There
are multiple proposals in Canada for similar organizations to share data and
cooperate operationally, following a model that was tested for the 2010 Winter
Olympics in Vancouver.17
By 2010, Canada was four years into the decade-long government of Stephen
Harper, who had promised to change Canada “beyond recognition.” This seems
far-fetched, particularly in retrospect and in the long-term historical view taken
here, but there were a number of key changes in the area of security intelligence
with which the country is still wrestling. Harper’s approach in general was to
favour increasing the legal powers of state security intelligence organizations,
with a corresponding reduction in human rights, particularly privacy. Argu-
ments over what became known as “lawful access” were not a consistent factor
over Harper’s ten years in power but became increasingly important, particularly
after 2009.
Successive public safety ministers, Peter van Loan and Steven Blaney, and
Justice Ministers Rob Nicholson and Peter MacKay, and above all Vic Toews,
who held both portfolios at different times, joined Harper in these attempts
to enact legislation giving police and security intelligence greater powers.
The initial push in 2009 came through two bills: Bill C-46, the Investigative
Powers for the 21st Century Act, and Bill C-47, the Technical Assistance for
Law Enforcement in the 21st Century Act. Key features, which were to recur
in almost all later bills, centred on allowing warrantless police access to
many aspects of Internet communications. Security fears associated with
the Vancouver Winter Olympics may have had something to do with it, but
terrorism, serious crime, and other threats were cited. Although this pair
of bills did not attract enough support, a second attempt came in 2011 with
Bill C-30, the Protecting Children from Internet Predators Act, whose title
demonstrates the kind of rhetoric deployed to justify lawful access provi-
sions. This again failed.
 Introduction 11

The Harper government did not limit itself to parliamentary avenues. In 2012,
the CSIS was effectively given more leeway with the elimination of the CSIS
Inspector-General’s Office, the body responsible for monitoring CSIS. Instead,
it was to be held accountable by the Security Intelligence Review Committee
(SIRC), which was made up of part-time federal appointees. The federal Office
of the Privacy Commissioner (OPC), although with no direct oversight of the
security intelligence services, is important in setting a tone with regard to privacy
and related human rights; here too Prime Minister Harper attempted to stack
the deck, appointing a career government security lawyer, Daniel Therrien,
when the opportunity arose, over more directly qualified candidates who were
favoured in the privacy law community. Both SIRC and Therrien gradually
discovered their teeth, however, and have bitten back to some extent against
both the government that appointed them and its Liberal successor.
In 2013, there was another attempt to introduce lawful access legislation: Bill
C-13, the Protecting Canadians from Online Crime Act, purportedly to tackle
cyber-bullying but containing provisions that were almost identical to the previ-
ous unsuccessful attempts. It again struggled, although a revised version was
eventually enacted as SC 2014, c 31. Of course, 2013 was a landmark year for
security intelligence because of the revelations of US National Security Agency
(NSA) contractor Edward Snowden. He was far from the first nor was he the
most highly placed NSA whistle-blower, but the huge cache of internal training
documents and slides he revealed was carefully curated with the assistance of
major newspapers and resonated with the public as no previous revelation had.
Canada initially seemed far from the centre of the action, even though the
documents confirmed that CSE was in fact a long-time junior partner of the
NSA. It was also revealed, however, that CSE had its own mass surveillance
programs. This should have been a national scandal, but for several years after-
wards CSE managed to avoid the consequences of the Snowden revelations that
other Five Eyes agencies faced. Instead, it moved into a slick new building whose
basement would hold massive supercomputing power – essential for the move-
ment to big data analysis.18 Far from its being reined in, the year after the
Snowden revelations saw a more concerted attempt to extend CSE’s powers.
This time the rationale was “lone wolf ” terrorist attacks in Quebec and Ottawa
and the rise of ISIS in the Middle East. Bill C-51 used tactics that the Conserva-
tive government had previously used to bury difficult legislation: it was included
as part of an omnibus bill, making it difficult to deal with the quantity and detail.
Bill C-51 provided CSIS with greater foreign and domestic powers and more
explicit immunity from accountability in the use of these powers. Documents
released in response to access to information requests to the Canadian security
services revealed that the state fears that drove Bill C-51 were much more related
12 David Lyon and David Murakami Wood

to environmental and Indigenous opposition to the oil industry in Alberta

(reflected in both the grassroots Idle No More and anti-pipeline movement),
resulting in RCMP infiltration and surveillance of such groups throughout this
period.
Bill C-51 sparked massive opposition, much more so than the previous failed
attempts, but was ultimately successful even though the Harper government
looked tired and out-of-date compared with the Instagram-friendly new Liberal
leader, Justin Trudeau. Trudeau had refused to oppose Bill C-51, claiming instead
that his government would review the provisions of the bill in a consultation
process. The resulting Bill C-59 (National Security Act, 2017) finally received
royal assent in June 2019. Such is the importance of this bill and its predecessor
that they are considered in multiple chapters of this book.
As for the intelligence relationship between Canada and the United States,
the situation is less certain in the age of President Donald Trump. In 2011, Canada
and the United States signed a special declaration titled Beyond the Border: A
Shared Vision for Perimeter Security and Economic Competitiveness, aiming “to
enhance our security and accelerate the legitimate flow of people, goods and
services.” Had this vision progressed, it would have tied Canadian sovereignty
further to that of the United States in line with post-9/11 trends. There were
hints of expansion of the extraterritorial powers of the US Customs and Border
Protection agency in places such as Pearson International Airport in Toronto,
but the aggressive anti-migrant and anti–free trade policies of President Trump
seem to have at least temporarily derailed such continental security efforts.
Further, Trump has struck at the other foundational elements of the postwar
settlement (e.g., international law, the United Nations, NATO, and the European
Union) but, despite his frequent denunciations of the deep state and the intel-
ligence agencies that he claims are conspiring against his government, not yet
the Five Eyes network. There can be no doubt that security intelligence co-
operation continues, but it is difficult to assess how much long-standing arrange-
ments may have been (or may yet be) damaged by the authoritarian populist
turn in the United States.
It is in this nested series of contexts – from technological transformation and
global networks to political machinations in Ottawa – that we present these
reflections on Canadian security intelligence and surveillance in the big data
age. For the longest time, security intelligence has depended on classic know-
ledge practices of the modern era, such as deductive reasoning and explanatory
logic, augmented since the mid-twentieth century by successive phases of
computer and communications development. Such information technologies
became significant in new ways with the creation of the Internet – itself origin-
ating in military defence – which facilitated both more rapid and global
 Introduction 13

surveillance networks and also, in the twenty-first century, the rise of platforms
and social networking. These generated massive amounts of data as the partici-
pation of ordinary users grew exponentially, which major corporations learned
to harvest for profit. It is these data above all that gave the impetus to big data
intelligence gathering and analysis, which is the subject of the rest of this book.

Security, Surveillance, and Big Data: A Road Map

This book is divided into five parts, each tackling a significant aspect of security
intelligence and surveillance in a big data age. The chapters are written both by
practically minded academics who wish to understand the issues through the
prism of actual processes and events, and by others from policy fields who are
very aware of the significant debates. Part 1, “Understanding Surveillance,
Security, and Big Data,” begins on a large canvas. In Chapter 1, Midori
Ogasawara paints a picture of how security surveillance is seen from within
national security agencies by operatives who courageously questioned their
practices. Her interviews with Mark Klein and Edward Snowden show clearly
how such agencies collaborate with big data corporations and the kinds of
impact this has for everyday surveillance practices. Such collaboration is a global
phenomenon, as clearly visible in Canada as in the United Kingdom or the
United States and elsewhere – one that raises crucial questions about where
responsibility lies for the kinds of pre-emptive surveillance, along with profit
making, characteristic of today’s situation.
This is followed by Chapter 2, from Christopher Prince, a researcher at the
Office of the Privacy Commissioner of Canada, who notes that law and policy
have become more complex and contested, that surveillance powers are difficult
to grasp, and that when it comes to intelligence gathering the view is positively
murky and mysterious. He offers a commentary on how this works out in
Canada, concluding appropriately that clarity is called for at every level. He
neatly summarizes some ways this could happen, and warns that while surveil-
lance may be necessary, it is not a solution to everything.
Chapter 3 focuses on anti-terrorism features as a key rationale for intensified
surveillance within both security agencies and policing services. The question
must be asked, however: Do the shifts towards data analytics offer genuine
benefits to the process of finding reliable and rights-respecting ways of combat-
ting terrorism? According to a long-time expert in the field, the answers are not
at all straightforward. Civil liberties are likely to take a hit, and without the
hoped-for benefits. So why pursue this route? There are many reasons, suggests
Stéphane Leman-Langlois, but none is compelling.
The question of big data may be pursued in even greater detail if a particular
case – here, financial tracking – is examined. In Chapter 4, Anthony Amicelle
14 David Lyon and David Murakami Wood

and David Grondin note the ways in which algorithms form part of the essential
infrastructure of security surveillance, especially as it applies to terrorist finan-
cing and money laundering. In Canada, the Financial Transactions and Reports
Analysis Centre of Canada (FINTRAC) uses algorithm-based alerts to sniff out
suspicious activity and govern the movements of suspects. It is not clear, how-
ever, whether the limitations of such methods are understood by their
practitioners.
Part 2 focuses on “Big Data Surveillance and Signals Intelligence in the Can-
adian Security Establishment.” Chapter 5, by Bill Robinson, carefully tracks the
specificities of Canadian SIGINT history through different phases, showing what
is distinctive about Canadian approaches as well as what Canada shares with other
nations. It is a fascinating and important story, especially as the recently discerned
tendencies of governments to turn their attention to internal “security” have
antecedents. Today’s CSE has grown greatly since the postwar days, however, and
no longer passively receives but actively hunts for data. Among other things, this
makes it more difficult to tell “targets” and “non-targets” apart.
Chapter 6 explores in more detail one recent, crucial aspect of these changes –
the shift within CSE to a “New Analytic Model” starting in 2012. Scott Thompson
and David Lyon use material provided by Access to Information and Privacy
(ATIP) requests to show that, in the same period that the United Kingdom and
United States made similar moves, big data analysis became the order of the
day at CSE. Along with growing dependence on tech corporations and their
software, there is a much greater reliance on computing expertise and practices –
“sandbox play,” for instance – and a reduced role for legal and ethical inter-
vention appears to be the perhaps unintended consequence of these
developments.
Another angle of the work of CSE – examining the interception of Internet
communications – is investigated by Andrew Clement in Chapter 7. The agency
itself is secretive and, unlike its US partner, the NSA, no truth-tellers have come
forward, as Edward Snowden did, to indicate more precisely what goes on in
the Edward Drake Building in Ottawa. Clement’s evidence strongly suggests
that CSE intercepts Canadians’ domestic Internet communications in bulk – as
do the NSA in the United States and the Government Communications Head-
quarters (GCHQ) in the United Kingdom – which is not legally permitted. The
“airport Wi-Fi” case from 2014 is just the first of several telling examples explored
here. Clement’s case is one that should give pause not only to anyone profes-
sionally concerned with privacy or those seeking stronger digital rights or data
justice but also to CSE and, indeed, every single Canadian citizen.
Part 3 focuses on the “Legal Challenges to Big Data Surveillance in Canada.”
In Chapter 8, Micheal Vonn sets the tone for the section with her analysis of
 Introduction 15

what can be learned from SIRC reports about the conduct of CSIS, an agency
as shrouded in secrecy as CSE. One report suggests that CSIS data acquisition
practices are “essentially unmoored from law.” Vonn cuts sharply through the
language used by CSIS, showing that data collection is not collection, a threshold
is not a threshold, and guidelines are not guidelines. Is this data collection
proportionate, necessary, and relevant? If not, it may be unconstitutional, and
the Bill C-59 “solution” to these problems may not be a solution.
This issue segues neatly into Craig Forcese’s Chapter 9, which is devoted to
Bill C-59, although readers may conclude that this analysis is slightly more
sanguine about the bill than the previous chapter. Nonetheless, it is a trenchant
critique from a leading legal scholar. He carefully distinguishes, for example,
between surveillance as watching on the one hand and the “potential watching”
enabled by new mass data-gathering methods on the other. The chapter clearly
understands the challenges of big data surveillance but concludes that despite
its limitations, Bill C-59 is a definite improvement on current legal measures.
These difficulties are echoed in a different area of surveillance – policing – that
nonetheless raises issues very similar to those that must be explored with national
security. In Chapter 10, Carrie Sanders and Janet Chan look at how big data
methods are actually used by police (much more is known about their activities
than about CSE and CSIS). Their findings are very instructive for grasping the
scope of this shift not only within the police but also within security agencies.
The connecting word is “security,” which each body has a mandate to protect.
The agencies’ desire to pre-empt and prevent security breaches, such as terror-
ism, is matched by the police claim to be detecting and disrupting crime – in
each case, leaning more heavily on big data. Like some other authors discussing
security agencies in this book, Sanders and Chan query police services’ know-
ledge and capacity to fully understand “the capabilities and limitations of big
data and predictive policing.” Responsible representatives of the security agencies
acknowledge this deficit too.
Part 4 then moves beyond formal legal challenges to consider active “Resist-
ance to Big Data Surveillance” by security intelligence agencies on the part of
civil society. In Chapter 11, Tim McSorley and Anne Dagenais Guertin survey
three revealing cases of resistance to government surveillance in Canada since
2001: Stop Illegal Spying (2013), Stop Online Spying (2011), and the International
Campaign Against Mass Surveillance (2004). They argue that each campaign
did make a difference because each was clearly targeted, created coalitions of
interest, and used diverse tactics to make its claims and mount its cases.
These three case studies are complemented by another – the protests against
government surveillance enshrined in Bill C-51 from 2014. In Chapter 12,
Jeffrey Monaghan and Valerie Steeves see this as unprecedented grassroots
16 David Lyon and David Murakami Wood

opposition to surveillance in Canada. It succeeded, they argue, due to new

forms of “connective action” using social media as well as conventional tactics,
and because it not only built on previous protests but also took advantage of
external events – notably the Snowden disclosures – to buttress its case. On
the other hand, Monaghan and Steeves recognize that future success will
depend once again on a variety of circumstances and tactics.
Finally, Part 5, “Policy and Technical Challenges of Big Data Surveillance,”
considers these challenges with a view to showing how the debates over big
data surveillance might be pursued towards appropriate goals. This book
began with the issues faced at CSE and this theme recurs with Christopher
Parsons and Adam Molnar’s Chapter 13, which analyzes its accountability
for signals intelligence. They propose new requirements for reporting CSE
activities – transparency at the legal, statistical, narrative, and proportional-
ity levels that would help accountability be not only vertical, towards over-
sight bodies, but also horizontal, for citizens in general, particularly those
who have concerns because of their own awareness, through personal
experience of how data issues affect them.
In Chapter 14, Andrew Clement, Jillian Harkness, and George Raine return
to the unresolved controversies over “metadata,” seeing it as the “fraught key
to big data mass state surveillance.” It is good that this is tackled by individuals
with computer science expertise because all too often the black box remains
unopened. As they note, in the hands of state actors, “metadata can provide the
basis for highly intrusive intervention into people’s lives.” The authors draw on
a study of more than 500 documents released by Snowden that show how
metadata may be used to build detailed life profiles of those surveilled. It is
clearly not “context, not content,” as often claimed. Surveillance researchers,
security intelligence agencies, and critics of mass surveillance are each offered
vital guidance in light of this.
The book concludes with an afterword by Holly Porteous, who agreed to
reflect on both the original research workshop and the chapters of this book
from her unique perspective as a former employee at CSE. Limited as it is by
what she is able to say in the context of her former employment, the value of
this chapter lies in its offering of a rare glimpse into the thinking of CSE
personnel, acknowledging the correctness of many conclusions reached in
our book while also questioning others in a very constructive fashion and
providing a number of fruitful avenues for further research that could fill gaps
and develop new insights. We are delighted to have this valuable coda to pull
together so neatly the central themes of the book while simultaneously offer-
ing a constructively critical view of where informed scholarship should focus
its analysis.
 Introduction 17

Conclusion
We started with CSE’s assertion that “everything has changed” in an age of big
data. Our brief historical sketch shows that this is far from the first time that
this argument has been made, and that the changes that are undoubtedly occur-
ring have deeper roots themselves, as well as potentially profound consequences.
It is also worth emphasizing that while a turn to a New Analytic Model would
seem to indicate a further shift from what has traditionally been understood as
HUMINT to SIGINT, there are two main caveats. First, security intelligence
agencies, whether SIGINT or HUMINT, are established bureaucratic organiza-
tions subject to the self-perpetuating logic of bureaucracies identified by Max
Weber early in the twentieth century.19 HUMINT agencies persist even as a lot
of what they now do and will increasingly do is indistinguishable technically
from the ostensible function of SIGINT agencies. Second, and despite the first
caveat, a lot of what happens from policing up to national security still has
nothing directly to do with big data. Human sources, tipoffs, infiltration, pro-
vocation, and much more besides remain central modes of operation for security
intelligence and political policing.
Many questions remain as to data’s centrality and “bigness.” As Carrie Sand-
ers shows, in the world of policing, big data practices are often marginalized
compared with these older, more trusted human methods.20 It appears that
this also holds for national security HUMINT work. Perhaps the officers who
doubt big data’s universal usefulness are right to do so: Stéphane Leman-
Langlois is profoundly skeptical about the historical effectiveness of big data
analysis techniques, arguing that most of the lessons that have supposedly
been learned relate to very specific situations that are not applicable in the
national security context.21 The effective use of data is often narrower and
smaller than the hype.
And finally, there are many questions about how the movement towards big
data affects how security intelligence agencies can be controlled and held
accountable, and their powers and reach in some cases rolled back. The cases
offered here present some contradictory lessons. Perhaps it is unlikely that any
legal, regulatory, or political oppositional activities on their own are going to
prevent the accumulation of ever larger collections of data and the application
of ever more “intelligent” analytic techniques, but that does not provide a carte
blanche for the collection of all or any data, for all or any data-mining processes,
and for all or any applications. Above all, the pernicious technocentric story of
endless and unstoppable technical progress must be challenged when it comes
to security intelligence agencies, because their activities can profoundly yet
often silently and secretly affect human rights, civil liberties, and the conditions
for human flourishing in Canada.
 18 David Lyon and David Murakami Wood

Notes
1 CSE refers to this not as big data but as the “New Analytic Model.” See Chapter 6.
2 See, e.g., Chapter 10.
3 danah boyd and Kate Crawford, “Critical Questions for Big Data,” Information, Com-
munication and Society 15, 5 (2012): 662–79.
4 Jose van Dijck, “Datafication, Dataism and Dataveillance,” Surveillance and Society 12, 2
(2014): 197–208.
5 David Lyon, The Culture of Surveillance: Watching as a Way of Life (Cambridge: Polity,
2018).
6 Bernard Harcourt, Exposed: Desire and Disobedience in the Digital Age (Cambridge, MA:
Harvard University Press, 2015).
7 Shoshana Zuboff, The Age of Surveillance Capitalism (New York: Public Affairs, 2018).
8 Clive Ponting, Secrecy in Britain (Oxford: Basil Blackwell, 1990), 10.
9 Such revelations have a long history in themselves; see David Murakami Wood and Steve
Wright, “Before and After Snowden,” Surveillance and Society 13, 2 (2015): 132–38.
10 David Vincent, The Culture of Secrecy in Britain, 1832–1998 (Oxford: Oxford University
Press, 1998).
11 Alfred W. McCoy, Policing America’s Empire: The United States, the Philippines, and the
Rise of the Surveillance State (Madison: University of Wisconsin Press, 2009).
12 Bernard Porter, Plots and Paranoia: A History of Political Espionage in Britain, 1790–1988
(London: Routledge, 1992), ix.
13 See Chapter 5.
14 P. Whitney Lackenbauer and Matthew Farish, “The Cold War on Canadian Soil: Milita-
rizing a Northern Environment,” Environmental History 12, 4 (2007): 920–50.
15 Reg Whitaker, Secret Service: Political Policing in Canada: From the Fenians to Fortress
America (Toronto: University of Toronto Press, 2012).
16 Andrew Crosby and Jeffrey Monaghan, Policing Indigenous Movements: Dissent and the
Security State (Toronto: Fernwood, 2018).
17 For more on CSIS, see Chapter 8.
18 See Chapter 6.
19 Max Weber, Economy and Society, ed. Guenther Roth and Claus Wittich (Berkeley:
University of California Press, 1978).
20 See Chapter 10.
21 See Chapter 3.
Part 1
Understanding Surveillance, Security, and Big Data
This page intentionally left blank
1
Collaborative Surveillance with Big Data Corporations
Interviews with Edward Snowden and Mark Klein
Midori Ogasawara

When reporters asked, they [AT&T] would give this strange statement, “we
don’t comment on matters of national security,” which implicates them right
there. National security? I thought you were a telephone company!
– Mark Klein, from my interview in May 2017

One of the most striking facts about today’s security intelligence is an extensive
collaboration with technology companies, which traffic, store, and use people’s
digital footprints, often employing so-called big data practices. It is worth
remembering that people who accused tech companies of cooperating with
governments for state surveillance were usually seen as paranoid or labelled
conspiracy theorists without evidence, until Edward Snowden’s revelations in
June 2013. Although there were a few whistle-blowers prior to Snowden on
tech-intelligence collaborative surveillance, such as former AT&T employee
Mark Klein, their claims were neither understood nor accepted by the public
to the extent that Snowden’s were.1
By disclosing top-secret documents of the US National Security Agency
(NSA), the former contractor unveiled the systematic way tech giants like
Google, Microsoft, Apple, and Facebook have worked with NSA to provide
volumes of personal data on their customers. The Snowden documents have
also shown in subsequent research by investigative reporters that major tele-
communications enterprises, such as AT&T and Verizon, have helped the NSA
set up mass surveillance facilities at landing points for transoceanic cables.
Through these specifically named Internet and telecommunications companies
and their documented involvement, people finally realized that governments
could actually seize their personal data and communications, and that it mat-
tered to them. Surveillance became everyone’s issue because most of us were
familiar with Google and Apple and used the private services they provided,
while having only a murky view of state surveillance.
The names of secret partners were deeply hidden. Their extensive cooperation
with security intelligence was the vital key to “Collect It All,” the NSA’s new
imperative, established since the US “War on Terror.”2 In one secret NSA slide,
22 Midori Ogasawara

Figure 1.1  A top secret NSA slide for the 2011 conference of the Five Eyes, showing NSA’s new
organization for “Collect It All.” | Source: Glenn Greenwald, No Place to Hide: Edward Snowden,
NSA, and the US Surveillance State (Toronto: Signal, 2014), 97.

the full circle of the “New Collection Posture” is completed with “Partner It All”
and enables the starting point of “Sniff It All” (Figure 1.1). But how and why did
private companies become good partners of security intelligence for numerous,
unwarranted wiretappings? What made the two work together for mass surveil-
lance behind the scenes?
This chapter examines the development of collaborative surveillance between
security intelligence and tech companies, and the effect of the collaboration on
the political and judicial spheres. Security intelligence and tech companies
rapidly developed strategic relationships induced by political economic incen-
tives. The main resources for my analysis are two personal interviews I had with
whistle-blowers from both the security intelligence side and the corporate side,
Snowden in 2016 and Klein in 2017. It should be noted that the main focus of
these interviews was on Japan-US relations, because the activities of the NSA
had rarely been contextualized in Japanese politics before. However, Snowden
and Klein explained in the interviews the mechanisms of worldwide surveillance
networks, of which Japan, Canada, and many other countries are part. Because
the worldwide networks are basically American systems, technically supported
 Collaborative Surveillance with Big Data Corporations 23

by American big data companies, and the NSA shares the data collected from
those networks with other foreign agencies, it can be said that the NSA systems
are also used as a major vehicle for security intelligence in other countries to
obtain data. In this sense, Canada’s foreign intelligence agency, the Communica-
tions Security Establishment (CSE), has been getting more data from the NSA
than the Japanese Ministry of Defense has, as the NSA categorizes Canada in
its closest group of the Five Eyes network (United States, United Kingdom,
Australia, New Zealand, Canada), or the “second parties” in sharing classified
data, while placing Japan among the “third parties” with the rest of the US allies.
Thus, the basic mechanisms of collaborative surveillance between the NSA and
tech companies to be described in this chapter are relevant to Canadian intel-
ligence agencies, to which the NSA and tech companies have been feeding the
data. Furthermore, because of the long-standing Five Eyes relationship, the
NSA’s collaborative techniques with tech companies can imply similar relations
and methods that CSE might have developed with Canadian tech
companies.
Although worldwide surveillance networks, enabled by state collaboration
with private actors, are continuously hidden, the expanded networks have been
affecting global political landscapes and redrawing judicial borders of state
surveillance. To demonstrate this, I will provide a Japanese example, and my
main argument will be the apparent global tendency to legalize the currently
illegal state surveillance of citizens. Snowden elaborated on the NSA’s strategies
to compel the Japanese government to pass certain surveillance legislation while
he worked as the NSA’s secret contractor in Japan. This legislation, called the
Secrecy Act, helped the NSA remain covert and practise further illegal surveil-
lance in Japan, and hid the Japanese government’s involvement.3 Similar stories
about other US allies are unreported as yet. But the Five Eyes and other European
countries have also passed legislation to expand state surveillance over citizens
under the political anti-terror discourse of the past two decades, including
Canada’s Bill C-51 and France’s “Big Brother Laws.”4 Together, they create a
global wave of legalization of previously illegal surveillance of citizens. This
phenomenon strongly resonates with and confirms NSA’s policy-making strat-
egies as explained to me first-hand by Snowden.
In what follows, I will first describe a dominant format of the NSA’s dragnet
surveillance, which emerged in the early stages of the War on Terror and which
Klein witnessed in 2004. It was built in part on telecommunications infrastruc-
ture in the private sector and presumably proliferated to other locations, as told
in Snowden’s detailed story of the Special Source Operations (SSO). The part-
nerships with tech companies are shrouded in secrecy, and if the secrecy is
breached, immunity follows retroactively, in order not to punish the NSA
24 Midori Ogasawara

partners that aided the illegal tapping. Second, what Snowden calls “natural
incentives” are present in every chain of collaboration, rather than coercive
orders. I will lay out the political economic interests tying the two entities
together before analyzing two kinds of collaboration, which the NSA categorizes
as “witting” and “unwitting” relationships. In the former, the information and
communications technology (ICT) companies know that they are delivering
data to the NSA; in the latter, the data are delivered without the consent of the
collaborators. The unwitting relationships are often established outside the
United States, through a technological “backdoor,” with foreign intelligence
agencies. The two together keep pushing the boundaries of data collection and
secrecy globally. As a significant outcome of the globalized collaboration, I
examine the Japanese case of surveillance laws – the Secrecy Act, the Wiretapping
Act, and the Conspiracy Act – and discern a judicial trend towards legalizing
previously illegal mass surveillance, in relation to the NSA’s collaborative strat-
egies.5 The early format of retroactive immunity is sustained and reinforced in
the trend, which allows the extra-judicial activities to replace the rule of law.

Built-in Format of Dragnet Wiretapping

Mark Klein was one of the earliest whistle-blowers regarding the NSA’s
unwarranted wiretapping programs. He worked for AT&T, one of the oldest
and largest telecommunications companies in the United States, as a communi-
cation technician who dealt with hardware and troubleshooting on long distance
digital circuits.6 When he was transferred to a branch located in downtown San
Francisco in 2004, he came across the NSA’s covert operation and collaboration
with AT&T.7
At the “611 Folsom Street office,” as it was known after Klein’s whistle-blowing,
he discovered a device called a “splitter cabinet” while troubleshooting a long
distance circuit. The circuit did not work well when connecting to the splitter
cabinet. Klein noticed that the splitter cabinet, when copying something,
degraded the signal, and this could cause problems. He called a management
technician who knew about the cabinet. The technician came from a secret
room downstairs, cleared by the NSA. He helped Klein solve the problem, but
Klein became aware of how the splitter worked. It was copying data and sending
them down to the NSA’s server. The splitter was not helping the data flow but
duplicating data for the NSA.
For Klein, the shock lay in the fact that the circuits connected to the splitter
cabinet carried local and domestic communications. The US government had
occasionally admitted that the NSA eavesdropped on international communica-
tions, but denied spying on domestic ones between US citizens. “What I revealed
was that they were collecting purely domestic traffic in San Francisco, so that
 Collaborative Surveillance with Big Data Corporations 25

meant they were collecting everyone’s data. The story about, ‘We’re just getting
international,’ was just a cover story,” said Klein.
Why everybody? Klein also found the points at which the NSA accessed the
networks: peering links. He explained to me:

“Peering links” are a typical term for the links that connect one network with
others, and that’s how you get the Internet. So AT&T’s domestic fiber network
connects to other companies’ networks, like Sprint or what have you, with
peering links so that people who are not AT&T, their messages can get to the
AT&T network. By tapping into peering links, you get a lot more of every-
body’s communications, not just AT&T’s. The fact that they did this in San
Francisco and they were tapping into peering links, people were really upset
when I revealed that.

Klein did not want to be part of the illegal unwarranted wiretapping, but could
not take immediate action for fear of losing his decent job. Later in 2004, the
company offered a financial package for the employees around his age who
wanted to retire. He took this opportunity and retired, taking with him engin-
eering documents that proved how the splitter cabinet was connected to NSA
equipment. He brought the documents to the Electronic Frontier Foundation
(EFF), an advocacy group for privacy and free expression in the electronic age.
In 2006, the EFF sued AT&T on behalf of its customers, for violating privacy
law by collaborating with the NSA in an illegal program to intercept citizens’
communications.8 Klein supported the EFF lawsuit as a witness and his docu-
ments were submitted to the court as evidence in Hepting v AT&T.
The Bush administration quickly intervened in this private case, however.
The government petitioned the court to dismiss the case on the grounds of state
secret privilege. Though the court rejected the petition, the government eventu-
ally succeeded in getting the controversial Foreign Intelligence Surveillance Act
[FISA] of 1978 Amendments Act of 2008 enacted by Congress, under which
AT&T was awarded so-called retroactive immunity. Law is principally non-
retroactive, but it was an exception. The 2008 amendments allow the Attorney
General to require the dismissal of lawsuits over a company’s participation in
the warrantless surveillance program if the government secretly certifies to the
court that the surveillance did not occur, was legal, or was authorized by the
president, whether legal or illegal.9 As a result, in 2009, a federal judge dismissed
Hepting and dozens of other lawsuits over AT&T’s collaboration with the NSA
in illegal wiretapping.
The EFF also sued the NSA and other government agencies in 2008, and
Klein’s documents from AT&T were again provided as evidence that AT&T had
26 Midori Ogasawara

routed copies of Internet traffic to the NSA’s secret room. By then, there was
more public exposure from other whistle-blowers regarding NSA mass surveil-
lance. However, the Obama administration also moved to dismiss this case,
claiming that litigation over the wiretapping program would require the govern-
ment to disclose privileged “state secrets” and that the wiretapping program
was immune from suit, using the same logic the Bush administration had
pursued.10 Though the case is not completely over, Klein believes that President
Obama effectively contained it. “The government has a million ways to kill a
lawsuit against the government. So I don’t hold out any hope for lawsuits,” he
commented.
Klein’s early revelations of NSA mass surveillance systems, in the first few
years after the United States declared the War on Terror in 2001, illuminate two
aspects of the collaboration between the NSA and ICT companies. One is that
the collaboration appears to have started at the foundation of the existing facili-
ties built by the telecommunication giants. When I visited 611 Folsom Street
after my interview with Klein, this reality became more tangible, because the
building is located at a busy street corner in downtown San Francisco. The secret
room was set up at a location familiar to locals and visitors, but not many pay
attention to the black building, which has few windows and looms large, like a
huge data box in the city centre. The NSA began a massive wiretap within the
telecom’s facility in this populated area, and the same systems were disclosed
in other AT&T facilities by the New York Times and the Intercept.11 The more
wiretapping points there were, the more personal data were acquired. This
built-in format of dragnet tapping would grow to sites beyond the telecoms’
branches, as Snowden later revealed, to a worldwide scale.
The other important aspect of the early collaboration between the NSA and
its partners is how devotedly the government protected the private partners
that had delivered the data. The government created a very exceptional, retro-
active law that turned illegal actions of the telecoms into legal ones. Retroactive
immunity nullified the law that banned warrantless wiretapping against citizens,
and the presidential order, whether legal or illegal, replaced the law. No respon-
sibility and accountability could be demanded if the rule was retroactively
changed after the game had started. Retroactive immunity is fundamentally
opposed to the rule of law and democratic oversight by Congress and the
people. This type of immunity has become an increasingly common strategy
used by governments to expand partnerships for mass surveillance worldwide,
as I discuss later. The strategy has drastically changed the boundaries between
legal and illegal state surveillance. My interview with Mark Klein highlights
the early format that enabled collaborative surveillance, both physically and
judicially.
 Collaborative Surveillance with Big Data Corporations 27

Natural Incentives for Collaboration

Though Klein clearly exposed the collaborative relationships between the NSA
and telecommunications companies for mass surveillance systems, the pioneer-
ing whistle-blower found it difficult to publicly prove what the NSA had spied
on. He did not have NSA security clearance. “I was working down in the nitty-
gritty, down at the hardware connecting things up. I didn’t have the big view.”
Edward Snowden walked into the right spot, in Klein’s view, to show the big
picture. “Snowden had the best evidence possible, actual documents from inside
NSA. What more could you ask for?”
Snowden explained to me various methods of NSA mass surveillance, includ-
ing the Special Source Operations, which he believes plays a central role in
today’s covert data acquisition.12 As he told me, “The government calls this bulk
collection, and everyone else calls it mass surveillance.” The SSO is typically set
up at the landing sites of the international transoceanic cables. The telecom-
munications companies from different countries jointly build and operate these
cables, and locate the landing stations on the shore to sort data traffic. According
to Snowden, the NSA requires the telecoms to set up a room for the NSA to
copy all data going through the landing sites. Based on classified NSA docu-
ments that Snowden disclosed in 2013, it has been revealed that the major landing
sites were part of their wiretapping programs, such as those code-named
FAIRVIEW, STORMBREW, or BLARNEY.13 The landing sites are called “choke
points” by the NSA and this is where tapping devices are embedded in the
communication hubs owned by ICT companies. This method has a very similar
format to the one that was built into the major communications infrastructure
at 611 Folsom Street in 2004, and appears to have evolved to a global scale.
Snowden said: “This is the real central problem because this is the way the
majority of spying happens today. It’s not done through satellite, although of
course that’s still done, but not majority. It comes from these ocean cables.” He
argues that SSO is more intrusive than the NSA’s other highly controversial
program, the PRISM program. PRISM shocked the world in 2013, revealing
how covertly well-known Internet service providers like Microsoft, Yahoo!,
Google, Apple, and others have cooperated with the NSA. According to Snow-
den, however, the government is more likely to pursue knowledge of individual
cases than knowledge from all the customers in PRISM, and there is an inter-
mediate step where the government sends requests for certain online accounts
to the company, and then the company pulls all the information directly from
the servers and gives an exact copy to the government. With SSO, on the other
hand, once the NSA has a telecom company set up the splitter cabinet, “then
the company doesn’t really have to do anything ever again, because NSA has
basically a perfect copy of everything that’s coming over those lines, all the time,”
28 Midori Ogasawara

just like duplicating all data going through the domestic communications cables
in San Francisco.
Why did the NSA develop this method as today’s major source of security
intelligence? The spy agency asked itself that in one of the Snowden documents,
which also indicates the reasons. “Why SSO?” Because “SSO is ‘Big Data.’” It
can capture up to “60% of content” and “75% of metadata” in a digital data
stream.14 There is no definition of big data in this document, but it is easy to
assume that big data includes not only actual correspondence among people,
such as telephone calls, emails, and text messages, but also human behaviour,
both online and offline, such as what people searched online and where they
went. Big data provides the NSA with a wider scope of information to predict
what people would like to do next.
Snowden described the actual process behind the scenes – how the NSA
negotiated with its partners to set up choke points at the landing sites. The NSA
normally pays the company for a room where it puts all its equipment, including
its servers. The NSA attaches special locks to the room and allows only staff
with security clearance to enter. It then asks the company to run the cable
through the NSA’s equipment, to copy all the data it wants. The telecoms do not
deal with what the NSA does with that data. “But, there is no real case here
where these companies don’t know about it,” asserted Snowden.
Why do big data corporations cooperate with these illegal wiretapping pro-
grams? Snowden points out that these companies take the NSA’s requests into
their business calculations. To expand the service areas and access new networks,
the company needs to obtain permissions and approvals from the government.
If the company got into trouble with the NSA by refusing the requests, it might
lose an opportunity to increase profits, resulting in a shrinking of its business.
“So, they’ve got a little bit of leverage with the government,” says Snowden.

Not that they are really threatened by the government, but the government
goes, “Well, it’s in our interests to help them build this new site or do whatever,
because they’ll give us access too.” And so, it’s this kind of culture that nobody
sees, because this all happens secretly. But it’s the nature of how intelligence
works ... you don’t want to think about villains. What you want to think is what
are the natural incentives for all of the people engaged at every level of the pro-
cess? What, if you were in their position, what would you do? And, suddenly, it
becomes obvious why everything works out the way that it does; everybody is
just scratching each other’s back.

On the government side, where Snowden once belonged, he experienced a

similar culture of political economic incentives within the organization, which
 Collaborative Surveillance with Big Data Corporations 29

encouraged avoiding trouble and following the existing tacit rules, for job secur-
ity and promotion. If any lawmakers attempted to stop the process, they could
find themselves in a politically vulnerable situation. If any incidents occurred,
they could be blamed or retaliated against by the intelligence agency that can
find anyone’s weak spot and leak personal data to achieve its goals. It would be
a safe choice for lawmakers to let the intelligence agency do what it wants, so
they suffer no consequences at all. “It’s not a product of evil ... It’s a product of
incentives in the system. Secrecy corrupts the process of democracy.” Because
few challenge the secret power of security intelligence, the intelligence agencies
transcend democratic power and simultaneously undermine democratic deci-
sion making.
Snowden attributes the rapid expansion of collaborative surveillance to pol-
itical economic incentives active at both ends – the NSA and ICT companies.
From a cautious perspective, delivering customers’ data to the NSA does not
directly profit the ICT companies. Rather, it violates customers’ privacy and
other rights and organizational compliance, and may damage business by harm-
ing public trust. This would be another story if the collaborations were all hidden.
Secrecy would nudge the ICT companies to avoid problems with the authorities,
and to choose the political economic benefits of conformity. In fact, when the
secret was revealed by the whistle-blowers, angry customers began accusing
the companies, as the EFF sued AT&T on behalf of its customers. Thus, secrecy
is a necessary condition for “natural incentives” to come into effect and for
growing the political economic incentives. The NSA and ICT companies can
“scratch each other’s backs” while people don’t know that their data are actually
used in the deal.
If all these processes had been transparent from the very beginning, the
political economic incentives would have been more likely to turn into business
risks because customers would stop using the services or sue the companies.
Collaboration based on illegal surveillance inherently requires secrecy, and
secrecy lets the incentives work for both ends – the government and the private
sector. However, an end to the secrecy initiated by Klein, Snowden, and other
whistle-blowers did not end the collaboration. With secrecy removed, the gov-
ernment took formal action to wrench the legal standard towards illegal prac-
tices, by passing laws to legalize the unprecedented scale of state surveillance.
Instead of secrecy, the government invented and provided retroactive immunity
to its partners, to protect their illegal involvement in mass surveillance

Witting and Unwitting Relationships

For the NSA, protecting and expanding its partners has been an important
keystone of the War on Terror imperative of Collect It All. The diagram of this
30 Midori Ogasawara

imperative consists of six stages in a cycle: Sniff It All, Know It All, Collect It
All, Process It All, Exploit It All, and Partner It All (see Figure 1.1).15 The last
stage is particularly relevant, as it guarantees the means of collecting personal
data through communication infrastructure as much as possible.
Snowden’s analysis is this: “In many cases, the way it works, the biggest ICT
companies are the ones that are working the most closely with the government.
That’s how they got to be big and that’s why they don’t have any trouble with
regulatory problems, and they got all the permits they need.” The largest Amer-
ican telecoms, AT&T and Verizon, have been reported as having “extreme
willingness to help” the NSA.16 A smooth relationship with the government
helps business grow, so political economic incentives are convincing. The NSA
aims to eventually set up an SSO at all landing sites of all international trans-
oceanic cables to gear up for Collect It All.
Snowden told me that these SSO collaborations between the government and
businesses are called “witting” relationships in NSA terms: the executives know
that they are working with the intelligence agency and that information is closely
shared with other intelligence agencies, such as the Five Eyes or the second
parties.17 For example, participants in PRISM, such as Microsoft, Facebook,
Google, and Apple, are involved with witting relationships, as are AT&T and
Verizon in the SSO.
Much less known are the “unwitting” relationships, where companies are
unaware that they are providing data to the NSA. These are unilateral programs
in which the NSA sets up tapping devices within ICT equipment and uses them
as a “backdoor” to absorb data to its servers.18 For example, according to Snow-
den, the NSA encourages foreign governments to set up their own network
equipment, whether in Afghanistan or Japan, so the foreign governments order
these high-tech appliances from ICT companies, many of which are based in
the United States. When the equipment is shipped from or transits the United
States, the NSA attaches tapping devices to the product. “So we made them into
a kind of Trojan Horse where you install it on your network and it works per-
fectly for you but it also works for us without you knowing.”
Those more deeply hidden unilateral programs directly assist anti-terror
tactics of identifying targets by location technologies and attacking them
remotely by drones. The point here is that the NSA sneakily develops unwitting
relationships in order to push the limits of witting relationships towards Partner
It All. In fact, the NSA has established both types of relationships with thirty-
three countries called third parties, including Japan, Germany, France, Brazil,
and Mexico.19 These countries are diplomatically US allies, so they often cooper-
ate with NSA surveillance by offering and exchanging data. Simultaneously,
however, the NSA surveilles these countries, including German chancellor
 Collaborative Surveillance with Big Data Corporations 31

Angela Merkel’s mobile phone,20 former Brazilian president Dilma Rousseff ’s

calls,21 and Japan’s thirty-five telephone lines connected to ministries.22 Thus,
witting and unwitting relationships were combined and often against the same
targets for various purposes. This shows the inherently unequal, ambiguous
relationships with allies as partners.
One NSA document Snowden and I discussed in the Japanese context clearly
indicates that the NSA had been collecting data from Japan in the SSO program
called FAIRVIEW.23 The piece of correspondence, dated 23 August 2011,
reported:

On 5 Aug 2011, collection of DNR and DNI traffic at the FAIRVIEW CLIFF-
SIDE trans-pacific cable site resumed, after being down for approximately five
months. Collection operation at CLIFFSIDE had been down since 11 March
2011, due to the cable damage as result of the earthquake off of the coast of
Japan ... FAIRVIEW operations will continue to task collection for all new and
restored circuits.24

“DNI” refers to Digital Network Intelligence and “DNR” is apparently Dial

Number Recognition. According to the investigative news organization Pro-
Publica, which reported on this document jointly with the New York Times,
CLIFFSIDE is a cable station operated by AT&T in Point Arena, California.25
The submarine cable landing at Point Arena is connected to Japan, where AT&T’s
counterpart, Japan’s NTT Communications, admitted one of the cables was
repaired on that date.
Another transoceanic cable, Trans-Pacific Express, is also a part of the SSO
program STORMBREW, according to the New York Times.26 The Trans-Pacific
Express was jointly constructed by telecommunications companies from China,
Taiwan, Korea, Japan, and the United States. In the United States, it is operated
by Verizon; in Japan, the landing station is again run by NTT Communications.
NTT is the oldest and largest telecom in Japan, which fits Snowden’s formula:
“the largest telecom works most closely with the government for surveillance.”
I asked Snowden: “Do you think that the Japanese telecoms are aware that this
cable is being used by NSA?” He carefully responded:

I don’t have any specific understanding of a Japanese corporation cooperating

directly with the American government, or NSA to spy on these things. How-
ever, I would be shocked if these corporations weren’t cooperating with Japanese
intelligence services to get exactly the same kind of information. It’s simply a
case of where they trade. The US feeds information from what they get to the
Japanese, and the Japanese feed information from what they get to the US.
32 Midori Ogasawara

The NSA proudly declared in a top-secret document the establishment of

“alliances with over 80 major global corporations supporting” its mission, as
strategic partnerships.27 Japanese collaborators have not been reported, but NTT
has, at least unwittingly, cooperated with the NSA in constructing, operating,
and maintaining submarine cables at the company’s cost. The collaborative
networks extend in a combination of witting and unwitting, and absorb data
from the foreign communications infrastructure. It remains unknown whether
or not the NSA’s choke points are constructed only within the United States,
but the NSA has been pushing the boundaries of dragnet surveillance outside
US soil through unwitting relations with foreign ICT companies. Before the
Internet became the main arena of mass surveillance, satellites were the major
means of intercepting foreign communications, most famously known as ECH-
ELON.28 ECHELON started with the UKUSA Agreement in 1947, and soon
came to be operated by the Five Eyes. It constructed satellite facilities outside
the United States, such as the radomes in Menwith Hill, UK, and at the US
airbase in Misawa, Japan. The NSA had already established international cor-
respondence on these foreign sites, not only for Cold War tactics but also for
corporate espionage against allied countries. Historically and legally, the NSA
has had more of a free hand to access foreign citizens’ information than that of
American citizens.

Legalize the Illegal

The complex of witting and unwitting relationships also highlights foreign
governments as another major group of NSA collaborators, besides ICT com-
panies, as they deliver and exchange data with the NSA. To protect and enhance
the surveillance networks, the NSA has been developing collaborative relations
with foreign intelligence agencies and putting political pressure on foreign
governments. This was the most eye-opening part of my interview with Edward
Snowden, who worked as an NSA contractor at the American Yokota Airbase
in Tokyo in 2009–11. With an insider’s perspective, Snowden discussed the
detailed, unknown background of a highly controversial law, the Secrecy Act,
passed by the Japanese Diet in 2013, stating that “this new state secrets law was
actually designed by the United States.”
According to Snowden, the NSA has a group of roughly 100 lawyers who
work for the Office of General Counsel. They partnered with a group called the
Foreign Affairs Directory, which researches different countries’ legal limits on
collaboration with the NSA and how to get around the legal protections that
prevent these countries from spying on their citizens or classifying the informa-
tion. The legal experts’ work includes hiding NSA mass surveillance from the
public eye so that the NSA can further expand its surveillance networks. Japan’s
 Collaborative Surveillance with Big Data Corporations 33

Secrecy Act was needed in this respect, and was proposed to Japanese
counterparts.

Again, they are not doing this for [an] evil purpose. They see this as necessary for
national defense. But the bottom line is, you start to get a group of countries, this
Five Eyes network, that had been creating this system of spying and judicial run-
arounds, legal runarounds for many years. And they start to export it to other
countries. They go, if you pass these laws ... of course, you can rewrite it. You
don’t have to pass the law exactly as we say. But, in our experience, (we say,) this
is what you should aim for, you should do this, you should do this. Those other
countries go, well, hey, we should do this. And, this is exactly what happened
with the state secrets law in Japan. When I was in Japan, we would have Japanese
partners come over to our building at Yokota. They knew we were there. It was
like the world’s greatest secret, because we were sharing information that we
were gaining from all this spying.

Characterizing them as typical NSA tactics, Snowden elaborated on how the

NSA usually motivated its Japanese or foreign partners to legitimize their secret
relationships:

And, we would say we can only share information at this level of classification
with you. We use it as sort of a carrot and a stick, where the Japanese military
want this piece of information, or they want that piece of information. Then we
say, well, we can’t give you that because your laws aren’t the way we want them
to be. We’ll give you this smaller piece of something else that might be helpful,
just kind of a token gift to help them save face ... So we would share things at the
secret level in Japan. We said, if you change your laws, we’ll share things at the
top-secret level with you.

These conversations all take place behind the curtain of intelligence agencies,
which I had never heard of as a background of the Secrecy Act. I asked Snowden
with suspicion how these conversations finally became law, what the diplomatic
process was. He explained that the conversation normally began with division
chiefs of intelligence agencies, then went up to the heads of the agencies and
eventually to the Department of State to formalize the agreement at a policy
level (this process was evidenced by an NSA document disclosed in April 2017).29
The NSA is aware that conducting mass surveillance is illegal according to Jap-
anese constitutional rights and judicial standards. “So it’s in violation of the law.
But, it doesn’t really matter at this point because no one can find out, at least in
terms of the government’s political calculation.” Secrecy, in this case about the
34 Midori Ogasawara

diplomatic process, enables security agents and bureaucrats to discuss the

unconstitutional bill against citizens:

And, then, eventually, on the basis of this, if this program has been continuing
for too long without legal authorization, we’ve proved that you need it. This is
how they get their foot in the door, a kind of press government. And I’m talking
about intelligence agencies in general, not just here in the States. In Japan, they
get their foot in the door by saying we’ve already operationalized this policy
and it allowed us to learn this, this, and do this and this, whatever, but we don’t
have the legal authorization we need to continue this program. So, if you don’t
force a law through your government that authorizes this, we’ll have to shut the
program down.

Then, according to Snowden, the NSA shows some examples of immediate

“threats,” such as that foreign agents are active in Japan, a terrorist attack is
planned, or Japan is losing its advantage in trade negotiations. Whatever the
stories are, these threats are used as “evidence” to convince not only the intel-
ligence agencies but also lawmakers, who are not familiar with the realities of
mass surveillance and how data are collected, retained, and used. Lawmakers
tend to think that, though it’s ambiguous how surveillance systems help them,
the systems look valuable and can be useful. Apparently, there is no pushback
against intelligence agencies’ having more surveillance and control over infor-
mation belonging to the citizenry.
The hidden processes Snowden described to me aptly correspond to recent
studies about secret Japan-US treaties, negotiated at the Japan-US Joint Com-
mittee, which was established by Japanese bureaucrats and American military
officers, not the government, after the Second World War.30 We do not have
space in this chapter to examine Japan-US relations further, but these historical
studies support Snowden’s first-hand information that the United States pres-
sured, if not forced, Japan’s government to legitimize the hiding of American
illegal surveillance in Japan, in order to maintain and enhance it. The Secrecy
Act enabled the Japanese government to designate any administrative files as
secret, upon which those files become immune to opening by the media, jour-
nalists, and citizens. Anyone who leaks secrets, and any journalist who reports
on it, can be sentenced to up to ten years of penal servitude.
The harsh punishment secures the government’s illegal practices from public
view. A rapid chilling effect emerged among the Japanese media and caught
international attention. The World Press Freedom Index, released by Reporters
Without Borders in Paris, ranked Japan 72nd out of 180 countries in 2016 and
2017, rated for freedom of news access, transparency, and protection of
 Collaborative Surveillance with Big Data Corporations 35

journalists. This was an extreme drop from 11th place in 2010.31 The index reflects
the serious dampening influence of the Secrecy Act, which induced self-
censorship in the media. The United Nations Special Rapporteur on Freedom
of Expression, David Kaye, visited Japan in 2016 and also reported on the nega-
tive effects of the Secrecy Act.32 The act has surely prevented the public from
knowing about the government’s illegal practices, including its support for the
NSA’s illegal surveillance. And this is not the end of the story.
After passing the Secrecy Act by an undemocratic procedure in the Diet, Prime
Minister Shinzō Abe’s government proposed and enacted two other surveillance
laws. One is a revision of the Wiretapping Act of 2016 that greatly expanded the
categories of crimes subject to police wiretapping investigations. The other is
the 2017 Conspiracy Act. The Conspiracy Act had failed thrice in the Diet in the
past because of its extreme stance against privacy and free speech. It criminal-
izes subjects for communications in which they agree on alleged crimes, whether
or not they actually commit any criminal acts. It replaces Japan’s postwar prin-
ciple of criminal justice without declaration, under which no one is charged for
a crime before taking action. It enables the police to tap potentially everyone’s
conversations, in order to identify who is talking about crimes.33 We have not
heard that the United States has been involved in crafting these two laws, but
they obviously help both the Japanese and American governments enlarge their
surveillance capacities. The Wiretapping Act legitimized the means of wiretap-
ping in criminal investigations, and the Conspiracy Act created the new reason
for the police to monitor everyone’s conversations, including emails, text mes-
sages, video chats, and so on. The serial enactment of the three laws contributed
first to protecting the NSA’s operations and the Japanese government’s collab-
oration in the first place, and then gave real teeth to their extension of mass
surveillance legally. Together, the three laws vitiate the once established legal
standards of privacy and right to confidential correspondence in the Japanese
Constitution, and modify them to accommodate the illegal surveillance prac-
tices.34 The previously illegal reality of mass surveillance has now been trans-
formed into lawful data collection, where extra-judicial and judicial power are
indistinguishable.
Significantly, retroactive immunity, the early means of collaborative surveil-
lance to protect the NSA’s private partners in the United States, has been
developed to a more legitimate form in Japanese surveillance laws, to legalize
illegal state surveillance. It is important to recognize this transformation because
retroactive immunity is an exceptional measure particularly deployed to absolve
the tech companies of violation of privacy. It had to be made retroactive, contrary
to the legislative principle of non-retroactivity, otherwise those companies might
have been found guilty in American courts. Retroactive immunity reveals the
36 Midori Ogasawara

illegal origin of mass surveillance in once established legal standards, while

governments try to blur the boundaries between legal and illegal surveillance,
place practices in a “grey zone,” and push the boundaries.
Taking into consideration Snowden’s comments that the NSA first created
the strategies in Five Eyes and began exporting them to other countries, the
tendency to legalize illegal surveillance does not appear to be limited to Japan.
For example, we can reconsider the backgrounds of Canada’s Bill C-51 or the
anti-terror measures enacted in France after 2015 in the state of emergency.35
Although it is beyond the scope of this chapter whether this happens within
the NSA’s second or third parties, such practices would be highly problematic
for the independence of the state and would show the United States as an imperial
power that writes other nations’ laws. Only whistle-blowers can reveal to us the
political processes behind the scenes, just as Snowden disclosed the Japanese
story. As he reiterated, the NSA has been using the same tactics in other coun-
tries, so some common explanation should be provided for its actions in different
countries to bring about discrete changes in laws in the global War on Terror.
In this sense, the Council of Europe’s Convention on Cybercrime, signed on
23 November 2001, can be seen as the first international move for surveillance
legislation during the War on Terror.36 Forty-three countries, including the
non-member states of Canada, Japan, South Africa, and the United States, signed
the convention, which requires participating nations to enact legislation that
facilitates investigation and prosecution of crimes committed through the
Internet. In the name of harmonizing national legislation and improving inter-
national cooperation, the signatories committed to granting law enforcement
authorities “lawful access” to data traffic, which compels Internet service pro-
viders and telecoms to assist interceptions by the state. Similar legislation had
been previously proposed in the United States and the United Kingdom but
had been dropped because of public concerns about privacy.37 Thus, the conven-
tion provided a “policy-laundering” detour for those governments38 because,
although it focused on particular crimes such as fraud, forgery, child pornog-
raphy, and copyright infringement, it opened up a much larger sphere for state
policing and intelligence gathering. Over time, compliance with the convention
was invoked to justify legislation that significantly weakened the legal standards
of data protection in each signatory country.39

Conclusion
In this chapter, I have described how the ICT industries have worked with the
NSA to collect and deliver data worldwide, based on the accounts of two whistle-
blowers. In summary, the NSA has developed a format of mass surveillance
systems built into existing digital communications infrastructure, from telecoms’
 Collaborative Surveillance with Big Data Corporations 37

branches in populated areas to the landing sites of the transoceanic cables on

remote shores. This format captured enormous amounts of big data, up to “60%
of content” and “75% of metadata” in a data flow.40 The alliance between security
intelligence and ICT industries was constructed with “natural incentives,” as
Snowden calls them, for avoiding regulatory trouble with the government and
expanding business on the business side, and for avoiding disputes in conformity
with the existing organizational imperatives on the government/military side.
These are the political economic ties, concealed and unprecedentedly expanded,
between the two entities.
The political economic incentives were effectively developed in secrecy
because giving consumers’ data to the government without their consent would
cause public outrage and damage business. Secrecy is vital to the functioning
of political economic incentives. Then when the secret was uncovered, the
government legitimized the companies’ involvement in illegal surveillance by
granting them retroactive immunity. It legalized the illegal surveillance practices,
and enabled further alliance and cooperation for more secret surveillance by
protecting the partners.
The NSA calls these relationships with the private sector “witting” relation-
ships, and has also constructed “unwitting” relationships by secretly installing
technological backdoors. With both witting and unwitting relationships, it has
achieved a global network of mass surveillance incorporating American ICT
companies, foreign governments, and foreign ICT companies.
As a result, my serious concern over expanding collaborations is the worldwide
domino effect of legalizing the previously illegal surveillance by states over the
population. Edward Snowden elaborated on the Japanese case in my interview,
which revealed that foreign legislative change has been one of the goals of the
NSA, to expand its surveillance networks. Japan’s Secrecy Act first provided the
NSA with official secrecy to cover illegal intelligence activities in Japan, and
then allowed the NSA to maintain and enhance illegal data collection with
support from the Japanese government. In turn, the law veiled the Japanese
government’s involvement with NSA surveillance, and its own illegal data col-
lection, and enabled the government to punish any whistle-blowers or journalists
who attempted to reveal these to the public. Two more surveillance laws followed
the Secrecy Act. The revised Wiretapping Act and the Conspiracy Act legalized
the previously illegal means and reasons of state surveillance over the last few
years. The expansion of global surveillance networks has been pushing Japan
to redraw the judicial borders of state surveillance and undermine the rights
established by the Japanese Constitution – such as privacy, confidential corres-
pondence, and freedom of expression – as a consequence of the invasive state
surveillance against the population in the wars until 1945.
38 Midori Ogasawara

Situating the NSA’s successful attempt in “designing” Japan’s Secrecy Act of

2013 in a global context, we find a transformation of retroactive immunity,
deployed by the US government and Congress in 2008, to protect the collab-
orators. Finding legal runarounds and eventually legalizing illegal surveillance
are parts of the NSA’s legal strategy to Partner It All and Collect It All that has
been initially created and tested within Five Eyes, according to Snowden. The
NSA’s global strategy of expanding collaborative networks should be further
researched and discussed in the context of different countries. As each country
has different relations with the United States, it is important to reconsider sur-
veillance legislation from the perspective of a country’s relations with the United
States. Together, we can identify the common dynamics involved in widening
the circles of digitized mass surveillance, which nullifies legal protections for
individuals, undermines constitutional rights, and presents itself as lawful
afterwards. Defining borders between legal and illegal surveillance is important
in order to resist the avalanche of illegal surveillance that seems to grow with
every technological advancement.
With respect to the use of big data, the tech companies have taken more
initiative than the intelligence agencies. It is the ICT companies that own the
infrastructure to collect, use, and sell personal data, and that create new
algorithms and exploit future markets based on collected data. Though big
tech companies are unwilling to disclose their relationships with security
intelligence, they are not shy to speak out about the predictive technologies
generated by big data, which go beyond existing legal boundaries for a “better
world.”41 Google co-founder Larry Page commented in 2013: “There are many,
many exciting and important things we can do, but we can’t, because they are
illegal.”42 Illegal here refers to violations of human rights and regulations in
democratic state systems. Marc Dugain and Christophe Labbé also refer to
other tech business leaders, such as Facebook’s Mark Zuckerberg or former
Google CEO Eric Schmidt, showing their beliefs that technologies can draw
out individual potential, achieve social equality, and overcome famine and
unemployment.43 Their comments represent not only the neoliberal ideologies
of unlimited deregulation and marketization but also the view that big data
is more efficient than politics in solving social problems. Democratic debates
from various viewpoints seem just inefficient obstacles in their view of the
world under the digital revolution. The algorithm is the new law governing
this “exciting” era. Dugain and Labbé point out that traditional politics deals
with causes of problems, whereas Silicon Valley deals only with results, ignor-
ing causes, processes, and especially inconvenient facts for their businesses:
“Big data is making politics unnecessary. This is an invisible coup, aimed at
taking out the essence of democracy and leaving only the outline.”44 As a result,
 Collaborative Surveillance with Big Data Corporations 39

big data corporations have been changing citizens into consumers, purely and
forcibly.
Thus, political and judicial standards to protect human rights are under attack
from technological solutionism, in a larger picture of a data-driven society. The
collaboration between security intelligence and big data corporations takes
place outside official judicial institutions and takes over political spheres. There
is a fundamental ideological sync between the two in transcending democratic
oversight and subverting judicial rules. Big data is currently exploited in this
depoliticized technological zone, where the two players (and other players too)
envision controlling the future, without political and judicial interventions.
As a closing note, whistle-blowers are a very rare resource in unfolding the
murky military/surveillance-industrial complex today. Their protection is an
urgent matter for any society. On the other hand, Mark Klein told me that
whistle-blowers are not sufficient to bring about political change: “What you
really need is a combination of whistleblowers and a party that’s determined to
bring down the government. And, you need large masses of people angry and
marching the streets.” Even as I was nodding, I was curious about why he decided
to begin this process alone. He answered in a husky voice, “I thought it [war-
rantless wiretapping by the NSA and AT&T] was illegal and immoral and
dangerous, and I thought I had a chance to do it and get away with it ... Well, I
had some principles.” At least, democratic “principles” may offer everyone a
starting point for holding the secretive collaborators of big data surveillance
accountable and countering illegal state practices.

Notes
1 I have to admit with shame that I was part of the ignorant public despite having
researched surveillance studies for several years.
2 Glenn Greenwald, No Place to Hide: Edward Snowden, NSA, and the US Surveillance State
(Toronto: Signal, 2014), 97.
3 Midori Ogasawara, スノーデン、監視社会の恐怖を語る [Snowden talks about
the horrors of the surveillance society: The complete record of an exclusive interview]
(Tokyo: Mainichi Shimbun, 2016).
4 Canada, Bill C-51, An Act to amend the Criminal Code and the Department of Justice Act
and to make consequential amendments to another Act, 2017, https://www.justice.gc.ca/
eng/csj-sjc/pl/cuol-mgnl/c51.html; Ewen MacAskill, “How French Intelligence Agencies
Failed before the Paris Attacks,” Guardian, 15 November 2015, https://www.theguardian.
com/world/2015/nov/19/how-french-intelligence-agencies-failed-before-the-paris
-attacks; Kim Willsher, “France Approves ‘Big Brother’ Surveillance Powers despite UN
Concern,” Guardian, 24 July 2015, https://www.theguardian.com/world/2015/jul/24/
france-big-brother-surveillance-powers; Angelique Chrisafis, “France Considers Extend-
ing National State of Emergency,” Guardian, 22 January 2016, https://www.theguardian.
com/world/2016/jan/22/france-considers-extending-national-state-of-emergency;
 40 Midori Ogasawara

Lizzie Dearden, “Paris Attacks: France’s State of Emergency Is Imposing ‘Excessive’

Restrictions on Human Rights, UN Says,” Independent, 20 January 2016, http://www.
independent.co.uk/news/world/europe/paris-attacks-frances-state-of-emergency-is
-imposing-excessive-restrictions-on-human-rights-un-says-a6822286.html.
5 Act on the Protection of Specially Designated Secrets, Act No. 108 of 2013, Japanese
Law Translation, http://www.japaneselawtranslation.go.jp/law/detail/?id=2543&vm=
04&re=01 (Secrecy Act); Act on Communications Interception during Criminal
Investigations, Act No. 137 of 1999, amended by Act No. 54 of 2016, https://elaws.e-
gov.go.jp/search/elawsSearch/elaws_search/lsg0500/detail?lawId=411AC0000000137
(Wiretapping Act); Revised Act on Punishment of Organized Crimes and Restriction of
Criminal Profits, Act No. 67 of 2017, https://www.sn-hoki.co.jp/article/pickup_hourei/
ph728/ (Conspiracy Act).
6 Jeff Larson, Julia Angwin, Henrik Moltke, and Laura Poitras, “A Trail of Evidence
Leading to AT&T’s Partnership with NSA,” ProPublica, 15 August 2015, https://www.
propublica.org/article/a-trail-of-evidence-leading-to-atts-partnership-with-the-nsa.
7 I interviewed Mr. Klein for about two hours at his house in a suburb of San Francisco in
May 2017.
8 Electronic Frontier Foundation, Hepting v AT&T, https://www.eff.org/cases/hepting.
9 Ibid.
10 Electronic Frontier Foundation, Jewel v NSA, https://www.eff.org/cases/jewel.
11 Mark Klein, Wiring Up the Big Brother Machine ... and Fighting It (Charleston, SC: Book-
Surge, 2009); Julia Angwin, Charlie Savage, Jeff Larson, Henrik Moltke, Laura Poitras,
and James Risen, “AT&T Helped U.S. Spy on Internet on a Vast Scale,” New York Times,
15 August 2015, https://www.nytimes.com/2015/08/16/us/politics/att-helped-nsa-spy
-on-an-array-of-internet-traffic.html; Ryan Gallagher and Henrik Moltke, “The NSA’s
Hidden Spy Hubs in Eight U.S. Cities,” Intercept, 25 June 2018, https://theintercept.
com/2018/06/25/att-internet-nsa-spy-hubs/.
12 I interviewed Mr. Snowden for two and a half hours in May 2016, via an encrypted video
channel from Queen’s University, Ontario.
13 Greenwald, No Place to Hide.
14 “Newly Disclosed N.S.A. Files Detail Partnerships with AT&T and Verizon,” New York
Times, 15 August 2015, https://www.nytimes.com/interactive/2015/08/15/us/documents.
html?_r=0.
15 Greenwald, No Place to Hide, 97.
16 Julia Angwin, Jeff Larson, Charlie Savage, James Risen, Henrik Moltke, and Laura Poitras,
“NSA Spying Relies on AT&T’s ‘Extreme Willingness to Help,’” ProPublica, 15 August 2015,
https://www.propublica.org/article/nsa-spying-relies-on-atts-extreme-willingness-to-help.
17 Paul Farrell, “History of 5-Eyes – Explainer,” Guardian, 2 December 2013, https://www.
theguardian.com/world/2013/dec/02/history-of-5-eyes-explainer.
18 Greenwald, No Place to Hide; Bruce Schneier, Data and Goliath: The Hidden Battles to
Collect Your Data and Control Your World (New York: W.W. Norton, 2015).
19 Ryan Gallagher, “How Secret Partners Expand NSA’s Surveillance Dragnet,” Intercept, 18 June
2014, https://theintercept.com/2014/06/18/nsa-surveillance-secret-cable-partners-revealed
-rampart-a/.
20 Kevin Rawlinson, “NSA Surveillance: Merkel’s Phone May Have Been Monitored ‘for
Over 10 Years,’” Guardian, 26 October 2013, https://www.theguardian.com/world/2013/
oct/26/nsa-surveillance-brazil-germany-un-resolution.
21 Julian Borger, “Brazilian President: US Surveillance a ‘Breach of International Law,’”
Guardian, 24 September 2013, https://www.theguardian.com/world/2013/sep/24/brazil
-president-un-speech-nsa-surveillance.
 Collaborative Surveillance with Big Data Corporations 41

22 WikiLeaks, “Target Tokyo,” news release, 30 July 2015, https://wikileaks.org/nsa-japan/.

23 Larson et al., “A Trail of Evidence.”
24 Ibid.
25 Ibid.
26 “Newly Disclosed N.S.A. Files.”
27 Greenwald, No Place to Hide, 102.
28 Toshimaru Ogura, エシュロンー暴かれた全世界盗聴網 [ECHELON: The unveiled
global tapping network] (Tokyo: Nanatsumori-shokan, 2002); Duncan Campbell,
“GCHQ and Me: My Life Unmasking British Eavesdroppers,” Intercept, 3 August 2015,
https://theintercept.com/2015/08/03/life-unmasking-british-eavesdroppers/.
29 See the NSA document found in “The Story Behind the Move,” Intercept, 24 April
2017, https://theintercept.com/document/2017/04/24/the-story-behind-the-move/,
published as part of the article by Ryan Gallagher, “Japan Made Secret Deals with the
NSA That Expanded Global Surveillance,” Intercept, 24 April 2017, https://theintercept.
com/2017/04/24/japans-secret-deals-with-the-nsa-that-expand-global-surveillance/.
30 Ukeru Magosaki, 戦後史の正体 [The true face of the postwar history] (Tokyo: Sougen-
sha, 2013); Koji Yabe, 日本はなぜ、「戦争ができる国」になったのか [Why Japan
became the country that can conduct war] (Tokyo: Shueisha International, 2015); Toshi-
hiro Yoshida, 「日米合同委員会」の研究 [The research on the Japan–United States
Joint Committee] (Tokyo: Sougen-sha, 2016).
31 Reporters Without Borders, “World Press Freedom Index,” https://rsf.org/en/ranking.
32 Ogasawara, [Snowden talks about the horrors of the surveillance society].
33 Midori Ogasawara, “Surveillance at the Roots of Everyday Interactions: Japan’s Con-
spiracy Bill and Its Totalitarian Effects,” Surveillance and Society 15, 3/4 (2017):
477–85, https://ojs.library.queensu.ca/index.php/surveillance-and-society/article/view/
6626/6442.
34 The Constitution of Japan, 1946, Japanese Law Translation, http://www.japaneselaw
translation.go.jp/law/detail/?id=174.
35 Daniel Leblanc and Chris Hannay, “Privacy, Security and Terrorism: Everything You
Need to Know about Bill C-51,” Globe and Mail, 10 March 2015, https://beta.the
globeandmail.com/news/politics/privacy-security-and-terrorism-everything-you-need
-to-know-about-bill-c-51/article23383976/?ref=http://www.theglobeandmail.com&;
MacAskill, “How French Intelligence Agencies Failed before the Paris Attacks”; Willsher,
“France Approves ‘Big Brother’ Surveillance Powers”; Chrisafis, “France Considers Extend-
ing National State of Emergency”; Dearden, “Paris Attacks.”
36 Kristin Archick,  Cybercrime: The Council of Europe Convention  (Washington, DC:
Congressional Research Service, 2006); Convention on Cybercrime, 2001, ETS No.
185 (entered into force 1 July 2004), Council of Europe, https://www.coe.int/en/web/
conventions/full-list/-/conventions/treaty/185.
37 Terry Palfrey, “Surveillance as a Response to Crime in Cyberspace,”  Information and
Communications Technology Law 9, 3 (2000): 173–93.
38 American Civil Liberties Union, “ACLU Announces International Project to Stop ‘Policy
Laundering,’” 13 April 2005, https://www.aclu.org/news/aclu-announces-international
-project-stop-policy-laundering?redirect=technology-and-liberty/aclu-announces
-international-project-stop-policy-laundering; Colin J. Bennett, Kevin D. Haggerty,
David Lyon, and Valerie Steeves, Transparent Lives: Surveillance in Canada (Edmonton:
Athabasca University Press, 2014).
39 Laura Huey and Richard Rosenberg, “Watching the Web: Thoughts on Expanding Police
Surveillance Opportunities under the Cyber-Crime Convention,” Canadian Journal of
Criminology and Criminal Justice 46, 5 (2004): 597–606; Colin J. Bennett and Charles
 42 Midori Ogasawara

Raab, The Governance of Privacy: Policy Instruments in Global Perspective (Cambridge,

MA: MIT Press, 2006).
40 “Newly Disclosed N.S.A. Files.”
41 Greenwald, No Place to Hide.
42 Marc Dugain and Christophe Labbé, L’homme nu: La dictature invisible du numérique
[ビッグデータという独裁者], trans. Kinuko Tottori (Tokyo: Chikuma shobo, 2017), 33.
43 Ibid.
44 Ibid., 101.
2
On Denoting and Concealing in
Surveillance Law
Christopher Prince

Accumulate, then distribute.

Of the mirror of the universe, be the part that is densest,
most useful and least apparent.
– René Char, “Leaves of Hypnos”

In the context of security intelligence, three premises can usefully kick-start

discussion of how the law operates. First, the law itself, as an instrument of
policy, has generally grown more complex (and contested) in the past decade.
At the same time, citizens, companies, and civil society groups stand ready to
contest new laws before they even come into force. Second, surveillance provi-
sions, as a substratum of law, can be labyrinthine. And, finally, when those
authorities are exercised in a national security context, they tend towards the
most opaque statutory powers of all.1
To be clear, long before preoccupation with counterterrorism, national security
exceptions cast a long shadow over the law; in this domain, many aspects of law
and its mechanics “went dark” at the very outset of the Cold War.2 Governments
today are still remedying those past excesses.3 So, while official secrecy is nothing
new, more recent criticism has begun to consider how law that is both covert and
complex can not only impact individuals but also erode confidence in systemic
notions of justice and rule of law as a whole.4 Facts withheld are one thing, but
fundamentals abandoned are a problem of another magnitude altogether.5
There are two separate but mutually reinforcing issues here. The first issue is
secrecy. Laura Donahue, in her Future of Foreign Intelligence (2016), describes
a whole legal corpus hidden from view, “a separate body of secret law” that has
developed in the national security courts of the United States since the 1970s
that few outside the classified legal system are able to grasp.6 The second issue
is complexity. David Anderson, in his 2015 review of British surveillance law,
made a related observation when he described the legal provisions as so complex
and entangled that probably only a dozen or so persons outside government
truly grasped how they all fit together.7 And finally, here in Canada, we have
had two high-profile decisions from our national security court making pointed
observations as to the lack of clarity and candour in the warrants they have
authorized or the operations they have been called upon to review.8
44 Christopher Prince

In Canada, transparency and oversight are perennial issues. Risks labelled by

security agencies as the judicialization of intelligence consumed an entire volume
of the federal Commission of Inquiry into the role of the Canadian Security
Intelligence Service (CSIS), the RCMP, and the Communications Security
Establishment (CSE) in the 1984 bombing of Air India Flight 182.9 This study
made much (over 300 pages) of the inherent tension between openness and due
process commitments (under the principle of open courts) and the clandestine
requirements of intelligence sharing and source protection (that accompany
highly classified national security operations).10 More broadly, there are also
rule-of-law questions concerning justiciability in the national security context,
where it becomes unclear where the jurisdiction of a given court extends and
where executive authority maintains.
This chapter does not aim to revisit these issues; the three studies just cited
provide an extensive comparative view of open courts versus covert methods.
Neither does it argue for any particular legal solution, because published statute
does not definitively delimit surveillance practice as a whole.11 Instead, it provides
a sketch of the particular situation in Canada by mapping the array of federal
law that intersects and combines when drawn into focus by a specific, complex
investigation. Flowing from that, one can then ask how core legal principles
(such as openness, intelligibility, or predictability) or privacy protections (such
as accountability and accuracy) withstand the overarching framework of powers
described. Finally, the rest of the contributions in this book are related to these
core protections and principles.

Canadian Surveillance Law in Brief

Numerous, complex federal laws come into play in this domain. The Security
of Information Act (SOIA), CSIS Act, Customs Act, and Canada Evidence Act
can come into play early in investigations. As information gathering ramps up,
the Criminal Code, National Defence Act, RCMP Act, Personal Information
Protection and Electronic Documents Act (PIPEDA), and Privacy Act tend to be
cited, either as sources of authority, as germane to issues flagged in legal opin-
ions, or as provisions dictating relevant disclosures between organizations to
produce evidence or advance investigations. Finally, where electronic data was
conveyed over networks as part of the offence, certain foundational clauses of
the Radiocommunication Act and Telecommunications Act could also be con-
sidered as part of the legal landscape.12
To be clear, these are just the laws implicated in authorizing specific electronic
surveillance or forensic evidence gathering. The black-letter law alone provides
an incomplete picture and the parameters that govern surveillance and intel-
ligence gathering are nested in numerous other forms of regulation that operate
 On Denoting and Concealing in Surveillance Law 45

alongside each statute: the Solicitor General Enforcement Standards for Lawful
Interception,13 CSE Ministerial Authorizations,14 Canadian Radio-television and
Telecommunications Commission (CRTC) Licensing Requirements, CSIS
Operational Directives,15 classified opinions and recommendations from the
National Security Branch at the Department of Justice,16 redacted jurisprudence
from the Federal Court,17 and reviews determined to be Secret or Top Secret
previously conducted by the Security Intelligence Review Committee (SIRC)
and the Office of the Communications Security Establishment Commissioner
(OCSEC) and now conducted by the National Security Intelligence Review
Agency (NSIRA). All such documentation can impact operations significantly,
even though they flow beneath the visible surface of the law(s). For example:
• Security of Information Act – provisions on document-handling infractions
[section 4(4)(b)] and offences for unauthorized entry of classified areas (section 6)
• Criminal Code – the investigative powers utilized in national security investiga-
tions, including production orders (section 487.014), tracking devices (section
492.1), wiretaps (section 186),18 and/or search warrants (section 487)19
• CSIS Act – sections 21 and 21.1, providing authorities to investigate threats to
Canadian systems20 and/or alleged involvement of foreign powers
• Customs Act – section 99.2, providing for broad powers of inspection and/or
seizure for any persons leaving or entering Canada
• Canada Evidence Act – section 38, providing for secrecy provisions and sealing
of court records used to cover sensitive evidence revealed, as well as to conceal
particular surveillance techniques
• National Defence Act – section 273.3, allowing CSE to provide specialized lawful
access (SLA) assistance to other federal agencies [paragraph 273.64(1)(c)]21
• PIPEDA – subparagraphs 7(3)(c.1)(i) or 7(3)(d)(ii), wherein companies such
as airlines or telecommunications service providers (TSP) holding information
relevant to an investigation bearing on Canada’s national defence or security are
given specific discretion to disclose information to authorities
• Privacy Act – paragraph 8(2)(c) or (e), whereby other government departments
may also disclose information to investigators, either under a court order or in
the enforcement of any law of Parliament22
• Telecommunications Act [section 15(1)] and Radiocommunication Act [section
5(1)(d)], which set out provisions23 allowing agencies to specify through regu-
lation or standards the maintenance of particular surveillance capabilities on
networks.24

Any or all of the above-noted provisions may figure in the court orders,
ministerial authorizations, supporting opinions, or warrant procedures tied to
either initiating specific surveillance or laying the foundation for approving
46 Christopher Prince

such activities. Arguably, they form the core legal parameters that responsible
ministers, members of the judiciary, or both expect to review before they sign
off on the particulars of intrusive surveillance or authorize broad collection of
digital evidence.25 Specific targets and suspected crimes may change from case
to case, but the underlying legal ground rules should be fairly familiar to
officials, even if they are not widely known outside security intelligence
organizations.26
So while surveillance conducted in the context of security intelligence is
almost always clandestine, it would be difficult to argue that these activities are
somehow unmoored from the law.27 Nor is it automatically malicious for gov-
ernment to keep secrets, if only to safeguard the lives of certain sources and
citizens. In fact, in any given operation conducted in Canada, by the RCMP,
CSIS, or CSE, the overlapping legal requirements accumulate quickly. Despite
commentary to the contrary – that surveillance and security operations some-
how unfold in a legal vacuum – the opposite argument can be made. The legal
framework around national security investigations in Canada offers practical
examples of surveillance assemblage, given the number of statutes that mutually
reinforce and legitimize each other.28 That theoretical term – developed by Kevin
Haggerty and Richard Ericson – refers to the elaborate arrangement of technical,
legal, personnel, and commercial systems required to operationalize ongoing,
persistent surveillance.

Law as Context, Complexity, or Complacency

To outside observers, these interlaced authorities appear both opaque and
daunting.29 As we are all aware, however, urban life at this particular moment
of modernity is very highly regulated. Ultimately, the qualitative difference in
law around surveillance arises from the highly sensitive nature of state surveil-
lance (for individuals) as well as the potential dangers posed to democracy (for
society collectively). Certainly police often remark that even just a generation
ago (before disclosure requirements and Charter issues) it was far simpler to
conduct investigations and advance cases to court. One can debate this point
in several ways; anyone reading Rebecca West’s New Meaning of Treason – from
the British espionage trials of the 1950s – might conclude it has in fact never
been simple.30 Many of the provisions cited above are exceptions and carve-outs
specifically placed into newer statutes as they were drafted within government
(or debated in Parliament) in order to maintain the efficacy of pre-existing tools
and practices used by intelligence agencies or police.
To be more explicit, in the Canadian context, it is easy to forget that CSE had
existed and operated for forty years before the first Charter jurisprudence began
to emerge from the Supreme Court or the federal Privacy Act came into force.
 On Denoting and Concealing in Surveillance Law 47

Similarly, CSIS had been gathering information through Canadian telecom-

munications networks for fifteen years before Canada had a full-fledged com-
mercial privacy law (albeit under a warrant scheme). As a result, the complex,
interstitial nature of the laws outlined above can present some very puzzling,
historical artifacts. Like layers of accrued sediment, the security apparatus of a
government might well reach back decades or longer, but on such an extended
timeline, new jurisprudence, new rights, new social practice, and new technolo-
gies also emerge and overlap. While cross-referential legal texts may not be an
intentional product, they can produce confusion in language and interpretation
that even seasoned lawyers and long-serving jurists struggle with.31 Examples
of these controversies in the surveillance and security context are well docu-
mented in subsequent chapters.
A skeptic might ask at this point whether it has not been ever thus: the
British law establishing the Royal Mail three centuries ago carried an oblique
loophole for the interception of letters and packets under the order of
the minister of state.32 What of it? Here a critical reader can turn back to the
contemporary discussion around the re-emergence of “secret law,” a line of
discussion wherein constitutional scholars and international jurists remind
us that our whole edifice of justice is defined by avoiding the secretive hall-
marks of Star Chambers and general warrants.33 As we have discovered in
the past few years, in the United States, the United Kingdom, and Canada
it is not simply that the legal machinery is complex (to a point approaching
arcane).34 It is not solely that the investigative authorities and government
counsel have exceptionally nuanced interpretations and definitions for terms
and techniques that can greatly stretch their legal powers.35 It is not strictly
speaking that much of the court oversight in the national security domain
is conducted in camera and ex parte, allowing little openness to the public
or adversarial challenge.36
These are all dangers in their own way, but what is of truly grave concern
is that there is now in many countries an entire body of jurisprudence,
administrative review, and legal interpretation of which the free press, Parlia-
ment, and the public are wholly ignorant.37 If there is a singular bedrock
principle nested in the term “rule of law,” it is that courts be open so that
their decisions are available for scrutiny and the law itself be accessible and
predictable (granted, those are three principles, but they are related).38 It is
very difficult to see how any of these conditions hold at present.39 So if we
accept the premise that surveillance law is highly fragmented, where does
this lead? How do those conditions impact conditions of fundamental justice?
Besides the interpretative confusion and bafflement of the general public,
what is the actual harm?
48 Christopher Prince

From Rights to Bare Harms

Canada has been through the experience of constraining wayward intelligence
work – through an assessment of impacts upon rights and principles of justice –
a good many times before. These questions have been the source of near-
perennial inquiry – from the aftermath of Igor Gouzenko’s defection40 to the
Munsinger spy scandal; from the RCMP’s CATHEDRAL and CHECKMATE
programs41 to the Air India tragedy and the Arar Affair. Each generation has
seen its own challenge and response to the question of how much discretion
we as a citizenry are prepared to allow our security agencies in their use of
intrusion and deception. The debate over Bill C-51 (2015) and the remedies
offered by Bill C-59 (2017) are but the most recent echoes of that same indis-
pensable question: How far to go?42
The complexity and opacity described above enormously affects how sensibly
that question can even be asked or answered by the public, by parliamentarians,
by civil society writ large.43 The density of the material creates obvious procedural
hurdles for participating in the debate.44 Worse, over time, the inaccessibility
created by those conditions produces a corrosive effect upon individuals and
their rights.45 Specifically, this occurs whenever we compromise the justiciability
of government decision making, the clear accountability for errors and missteps,
and the plain-language predictability of the law itself.46 As these are significant
ripple effects of the complexity inherent in surveillance law especially, it is better
to examine them one at a time.
The principle of justiciability (i.e., court appeal or judicial review) men-
tioned at the outset of this chapter is a core element of due process within a
system of justice tethered to rule of law.47 Without this, systemic mistakes,
political meddling, or simply arbitrary use of power go unchecked; that is
point one.48 The principle of accountability (i.e., state power overstepping its
bounds will be answered for) within a rule-of-law framework requires some
degree of ongoing openness or transparency for that process of questioning
to even begin.49 Without a channel for feedback and challenge, the security
process leads effectively to a parallel system of justice that operates under
different controls, categories, and conclusions; that is point two. Finally, such
incoherence leads to a final compromise, by thwarting the principle of pre-
dictability.50 When a classified system of complaint, review, and interpretation
operates beneath the veneer of public law, practices and operations can be set
in motion that would very likely fall by the wayside under more rigorous,
open scrutiny.
The Foreign Intelligence Surveillance Court (FISC) in the United States pro-
duced just such an effect in its year-over-year endorsement of creative and
expansive interpretations by the executive drawn from the USA PATRIOT Act
 On Denoting and Concealing in Surveillance Law 49

provisions in the decade after 9/11.51 The Independent Reviewer of Terrorism

Legislation in the United Kingdom uncovered the same “creativity” in reading
the statutes in his system. And, arguably, the irritation made clear in recent
decisions by Justice Mosley52 and Noël53 (to say nothing of the basic-law-school
tone of the Spencer decision by the Supreme Court of Canada)54 also represented
considerable reversals of securitized legal reasoning here in Canada. Either way,
it is clear that for arbitration of security matters to be effective in an ongoing
fashion, they must be wholly independent of government influence and the
courts need to have all the facts beforehand (not just the arguments of the
security authorities themselves).55
Those are the formalist “rule-of-law” objections to overclassification and
muddying of complexity in security surveillance law. They are significant
criticisms, if only because they affect not only the quality and resilience of
our justice systems but also their global repute and influence. If that sounds
abstract and idealist, there are other recent trends that indicate that the prob-
lems spelled out above will also tend to manifest not just in human rights
cases and constitutional challenges – eventually these will overflow into the
economic arena as well.
To cite one prominent example focused on Canada (well described by Gabe
Maldoff and Omer Tene), recent developments in the European Union show a
clear linkage between trade and commerce issues, on the one hand, and respect
for basic human rights protections and clear channels for judicial review, on
the other.56 The courts and other legal authorities of the EU have been ruling
along these lines for several years now, with the collapse of the US Safe Harbor
privacy framework being just one recent instance.57 The rejection in 2017 of the
EU-Canada Passenger Name Record (PNR) Agreement is another.58 To put it
bluntly, where security and intelligence powers are at play, complexity quickly
becomes a bug, not a feature.
At least within the EU model, when other countries are being examined for
their commitments to core rights such as privacy and legal redress, one of the
criteria courts are flagging as non-negotiable is accessibility. Another is
independence. A third is transparency.59 On those pillars alone, the measures
in the United States and the United Kingdom to contain surveillance powers
have been found wanting, which has in turn created considerable uncertainty
for many companies. It is no coincidence that Silicon Valley firms lobbied very
hard for Congress to speed passage of the Judicial Redress Act after EU authorities
threatened to halt transfers of personal data on their citizens.60 While it is dif-
ficult to accurately tally the costs for business and investment, certain sectors
such as banking, transportation, and data services report contracting losses and
higher operational costs in the billions of dollars.61
50 Christopher Prince

The Case for Openness

Accretion of secrecy and porosity of interpretation leads to a basic question:
How well does law actually control surveillance, or conversely, does it simply
facilitate it? Take a specific example like data analytics or machine learning:
Where are the statutory limitations? Until very recently, at least in Canada, law
was silent there. Mute is, after all, the default for security organizations when
sources and methods are at play. For the past half-century, describing their
capabilities, technologies, or techniques in published writing or public comment
was strictly forbidden, a crime to describe unless authorized, carrying prison
terms of ten years or more.
Yet most media-aware citizens are generally aware that security agencies, who
are in the business of surveillance, have access to powerful tools, advanced
technologies, massive computing power, and shared global information net-
works.62 These have been treated in leaked documents, media coverage, and
academic work around surveillance for the past decade.63 Until quite recently,
however, these capabilities were not specified through legislation.64
Recently, though, a federal government push towards big data has finally
resulted in public law that actually bears scrutiny. We now have in Canada
specific clauses in law passed by Parliament regarding the use of tracking
devices, communications tracing, and transmission data recorders,65 in addi-
tion to specific warrants to compel production of location data, transaction
records, and communications metadata, all in the Criminal Code. We now
even provide stand-alone authorizations in the law for frequency jammers,
under the Radiocommunication Act.66 All have come into force in just the past
few years. More recently still, now before Parliament, are a bevy of new legal
authorities governing how CSIS will obtain and analyze datasets as well as
how the CSE will conduct a wide range of collection activities and cyber
operations.67
It has taken over a decade to arrive at this point – where federal authorities
see the general benefit of being ostensibly open about their surveillance activities
versus the cost of too much information being set before the public – but it is
heartening at last to be able to survey the legal landscape with more precision
and less paranoia.68 Canada has been through a protracted, trying debate these
past fifteen years; academics, civil society, security officials, and the human
rights community all grew deeply divided. Lawful access, metadata collection,
Snowden, Spencer, the Security of Canada Information Sharing Act (SCISA) –
each was a bruising argument on its own. What they in succession have pro-
duced, however, if nothing else, is a clarification of both the terms of art as well
as the high stakes involved for all.69
 On Denoting and Concealing in Surveillance Law 51

And in this midst of these controversies, our courts have rightly maintained
an independent skepticism towards surveillance solutionism; they have instead
asked government to reconsider first principles. The Supreme Court of Canada
treated these questions in recent cases such as Tse, Wakeling, Spencer, and
TELUS.70 Our Federal Court underscored related points through decisions of
Justice Mosley and Noël.71 The US Courts of Appeals did likewise in their
review of mobile device tracing and Jones.72 The European Court of Justice
did so in annulling the Safe Harbor Agreement. Proper democratic practices
and constitutional safeguards against intrusive state powers have been insisted
upon, and that is, ultimately, the purpose of public law, underscored by the
very due process discussions that courts around the world have been urging
us to have.

Acknowledgments
The views presented are solely those of the author. Thanks to both Christopher Parsons
and Lisa Austin for kind suggestions.

Notes
1 Henry A. Crumpton, The Art of Intelligence Lessons from a Life in the CIA’s Clandestine
Service (New York: Penguin, 2012), 7; Heidi Boghosian, Spying on Democracy: Govern-
ment Surveillance, Corporate Power and Public Resistance (San Francisco: City Lights
Books, 2013), 22.
2 Frederick A.O. Schwarz, Democracy in the Dark: The Seduction of Government Secrecy
(New York: New Press, 2015), 36–37.
3 The most recent official apology and work towards compensation concerned the target-
ing and harassment of public servants and military staff who were gay, a practice that
continued into the early 1990s; see John Paul Tasker, “Ottawa to Apologize Formally to
LGBT Community for Past Wrongs,” CBC News, 17 May 2017, http://www.cbc.ca/news/
politics/ottawa-apologize-lgbt-community-past-wrongs-1.4120371.
4 For recent discussion of individual effects, see Fen Osler Hampson and Eric Jardine,
Look Who’s Watching: Surveillance, Treachery and Trust Online (Waterloo, ON: CIGI,
2016).
5 G.T. Marx, Windows of the Soul: Privacy and Surveillance in the Age of High Technology
(Chicago: University of Chicago Press, 2016), 321.
6 Laura Donahue, The Future of Foreign Intelligence: Privacy and Surveillance in the Digital
Age (New York: Oxford University Press, 2016), 145.
7 David Anderson, A Question of Trust: Report of the Investigatory Powers Review (Lon-
don: The Stationery Office, 2015), https://terrorismlegislationreviewer.independent.gov.
uk/wp-content/uploads/2015/06/IPR-Report-Web-Accessible1.pdf, 8, 148, 218.
8 2014 FCA 249, Federal Court of Appeal, https://www.canlii.org/en/ca/fca/doc/2014/
2014fca249/2014fca249.pdf.
9 Commission of Inquiry into the Investigation of the Bombing of Air India Flight
182, Air India Flight 182: A Canadian Tragedy, vol 3, The Relationship between Intel-
ligence and Evidence and the Challenges of Terrorism Prosecutions (Ottawa: Public
Works and Government Services Canada, 2010), http://publications.gc.ca/collections/
 52 Christopher Prince

collection_2010/bcp-pco/CP32-89-2-2010-3-eng.pdf; Kent Roach, “The Unique

Challenges of Terrorism Prosecutions: Towards a Workable Relation between Intel-
ligence and Evidence,” in Air India Flight 182: A Canadian Tragedy, vol 4, The Unique
Challenges of Terrorism Prosecutions (Ottawa: Public Works and Government Services
Canada, 2010), http://publications.gc.ca/collections/collection_2010/bcp-pco/CP32-
89-5-2010-4-eng.pdf.
10 Canada (Citizenship and Immigration) v Harkat, 2014 SCC 37, para 85, https://scc-csc.
lexum.com/scc-csc/scc-csc/en/item/13643/index.do.
11 In support of that point (to cite just one example), there are by one estimate some-
where approaching a thousand informal arrangements, agreements, letters, memoranda
of understanding, and formal treaties between the Five Eyes states alone governing
exchange of information. See J.T. Richelson and D. Ball, The Ties That Bind: Intelligence
Cooperation between the UKUSA Countries, the United Kingdom, the United States of
America, Canada, Australia, and New Zealand (Boston: Unwin, 1985), 141–43, cited in
Ashley Decks, “Intelligence Services, Peer Constraints and the Law,” in Global Intelli-
gence Oversight, edited by Z.K. Goldman and S.J. Rascoff (New York: Oxford University
Press, 2016), 6.
12 Security of Information Act, RSC 1985, c O-5; Canadian Security Intelligence Service
Act, RSC 1985, c C-23; Customs Act, RSC 1985, c 1 (2nd Supp.); Canada Evidence
Act, RSC 1985, c C-5; Criminal Code, RSC 1985, c C-46; National Defence Act, RSC
1985, c N-5; Royal Canadian Mounted Police Act, RSC 1985, c R-10; Personal Infor-
mation Protection and Electronic Documents Act, SC 2000, c 5; Privacy Act, RSC
1985, c P-21; Radiocommunication Act, RSC 1985, c R-2; Telecommunications Act,
SC 1993, c 38.
13 Available at https://cippic.ca/uploads/ATI-SGES_Annotated-2008.pdf.
14 K. Walby and S. Anaïs, “Communications Security Establishment Canada (CSEC) Struc-
tures of Secrecy, and Ministerial Authorization after September 11,” Canadian Journal
of Law and Society 27, 3 (2012): 365–80, doi:10.1017/S0829320100010553; Wesley
Wark, “CSE and Lawful Access after Snowden” (working paper, Centre for Interna-
tional Policy Studies, University of Ottawa, 2016), http://www.cips-cepi.ca/wp-content/
uploads/2011/09/WWP-FINALfinal.pdf.
15 Public Safety Canada, “Ministerial Directions,” https://www.publicsafety.gc.ca/cnt/trnsprnc/
ns-trnsprnc/index-en.aspx.
16 Public Safety Canada, “Memorandum for the Minister: Next Steps on a Federal Court
Decision on CSIS Warrant Powers,” 13 December 2013, https://www.scribd.com/
document/211668176/DIFTS-in-Jeopardy.
17 2016 FC 1105, https://decisions.fct-cf.gc.ca/fc-cf/decisions/en/item/212832/
index.do.
18 “Section 186,” in Martin’s Annual Criminal Code 2018, edited by Edward L. Greenspan,
QC, The Honourable Justice Marc Rosenberg, and Marie Henein, LLB (Toronto: Thomson
Reuters Canada, 2017), 384–94.
19 “Section 487,” in ibid., 928–39.
20 Robert W. Hubbard, Peter M. Brauti, and Scott K. Fenton, “Electronic Surveillance
under the CSIS Act,” in Wiretapping and Other Electronic Surveillance: Law and Proce-
dure (Toronto: Canada Law Book, 2009), 12–13.
21 Stanley Cohen, “Communications Security Establishment,” in Privacy, Crime and Terror
(Markham, ON: LexisNexis, 2005), 226–28.
22 Craig Forcese, “Privacy Act,” in National Security Law (Toronto: Irwin Law, 2008),
390–93.
 On Denoting and Concealing in Surveillance Law 53

23 Robert Howell, “Regulatory Structures and Features,” in Canadian Telecommunications

Law (Toronto: Irwin Law, 2011), 79–80.
24 In the case of use of radio-spectrum, in connection with standards developed with
Public Safety Canada, these can entail identification, data capture, location, and elec-
tronic surveillance features on their networks, as a condition of licence. See Law
Library of Congress, “Government Access to Encrypted Communications: Canada,”
https://www.loc.gov/law/help/encrypted-communications/canada.php; also Christo-
pher Parsons, “Canada’s Quiet History of Weakening Communications Encryption,”
Technology, Thoughts and Trinkets, 11 August 2015, https://christopher-parsons.com/
canada-quiet-history-of-weakening-communications-encryption/.
25 Stanley Cohen, “The Administration of Justice, the Charter and Canada’s National Secu-
rity,” in In the Balance: The Administration of Justice and National Security in Democracies
(Toronto: Irwin Law, 2007), 99.
26 David K. Shipler, The Rights of the People : How Our Search for Safety Invades Our Liber-
ties (New York: Knopf, 2011), 28.
27 Gary T. Marx, Windows in the Soul: Surveillance and Society in the Age of High Technol-
ogy (Chicago: University of Chicago Press, 2016), 17. Note that despite all the criticisms
and gaps noted above, it should be restated that at a certain level of granularity, surveil-
lance in the intelligence and security context will of necessity remain secret. This must be
so often – in point of fact, even largely – to protect the targets of surveillance as opposed
to the technical tradecraft involved; the secrecy in that sense is not malicious but rather
one that protects the reputation and livelihood of those under scrutiny.
28 K.D. Haggerty and R.V. Ericson, “The Surveillant Assemblage,” British Journal of Sociol-
ogy 51 (2000): 605–22. See also R.V. Ericson, Crime in an Insecure World (Cambridge:
Polity, 2007).
29 Boghosian, Spying on Democracy, 118–19.
30 “The law should be at once the recognition of an eternal truth and the solution by a com-
munity of one of its temporal problems; for both conceived that the divine will was mir-
rored in nature, which man could study by the use of his reason. This is the faith which
has kept jurisprudence an honest and potent exercise through the ages”: Rebecca West,
New Meaning of Treason (New York: Viking, 1964), 40.
31 Peter Gill, Policing Politics: Security Intelligence and the Liberal Democratic State (Lon-
don: Cass, 1994), 64, 78.
32 Jon Moran and Clive Walker, “Intelligence Powers and Accountability in the UK,” in
Global Intelligence Oversight, edited by Z.K. Goldman and S.J. Rascoff (New York: Oxford
University, 2016), 293.
33 Daniel Solove, Nothing to Hide (New Haven, CT: Yale University Press, 2011), 178–79;
also Commission of Inquiry into Certain Activities of the Royal Canadian Mounted
Police, Second Report, vol 1, Freedom and Security under the Law (Ottawa: Minister
of Supply and Services Canada, 1981), 841, 881, http://publications.gc.ca/collections/
collection_2014/bcp-pco/CP32-37-1981-2-1-1-eng.pdf.
34 Dorle Helmuth, Counterterrorism and the State (Philadelphia: University of Pennsylva-
nia Press, 2016), 58–59.
35 Schwarz, Democracy in the Dark, 124–25.
36 Steven Aftergood, “Privacy and the Imperative of Open Government,” in Privacy in the
Modern Age: Search for Solutions (New York: New Press, 2015), 22–23.
37 David Wise and Thomas Ross, Invisible Government (New York: Random House, 1964), 6.
38 William Arkin, American Coup: Martial Life and the Invisible Sabotage of the Constitu-
tion (Boston: Little, Brown, 2013), 33.
 54 Christopher Prince

39 Tamir Israel, “Foreign Intelligence in an Inter-Connected World,” in Law, Privacy, and

Surveillance in Canada in the Post-Snowden Era, edited by Michael Geist (Ottawa:
University of Ottawa Press, 2015), 73, https://ruor.uottawa.ca/handle/10393/32424.
40 Stuart Farson and Mark Pythian, Commissions of Inquiry and National Security (Denver:
Praeger, 2011), 35.
41 Greg Marquis, The Vigilant Eye: Policing in Canada from 1867 to 9/11 (Black Point, NS: Fern-
wood, 2016), 96; see also Richelson and Ball, The Ties That Bind, 290–94; Morris Manning,
The Protection of Privacy Act: Analysis and Commentary (Toronto: Butterworths, 1974), 78.
42 H.H. Ransom, “Can the Intelligence Establishment Be Controlled in a Democracy?” in
Surveillance and Espionage in a Free Society, edited by Richard H. Blum (New York: Praeger,
1972), 215.
43 William W. Keller, Democracy Betrayed: The Rise of the Security Surveillance State
(Berkeley, CA: Counterpoint, 2017), 19, 192.
44 David Tucker, The End of Intelligence: Espionage and State Power in the Information Age
(Stanford, CA: Stanford University Press, 2014), 33.
45 Brian Tamanaha, Laws as a Means to an End (Cambridge: Cambridge University Press,
2006), 2; also Joseph Stiglitz, “Liberty, the Right to Know and Public Discourse” (1999),
in Government Secrecy: Classic and Contemporary Readings, edited by Jan Goldman and
Susan L. Maret (Westport, CT: Libraries Unlimited, 2009), 697–98.
46 F.A. Hayek, Constitution of Liberty (Chicago: University of Chicago Press, 2011), 239,
319–23.
47 Ian Kerr, “Prediction, Pre-emption and Presumption,” in Privacy, Due Process and the
Computational Turn, edited by Katja de Vries (New York: Routledge, 2013), 107–8; see also
Peter W. Hogg and Allison A. Thorton, “The Charter Dialogue between the Courts and the
Legislatures,” in Judicial Power and Canadian Democracy, edited by Paul Howe and Peter
H. Russell (Montreal and Kingston: McGill-Queen’s University Press, 2001), 106.
48 Lisa Austin, “Lawful Illegality,” in Geist, Law, Privacy, and Surveillance in Canada, 108.
49 Andre Nollkaemper, National Courts and the International Rule of Law (Oxford: Oxford
University Press, 2011), 3–5.
50 Raymond Plant, The Neo-liberal State (Oxford: Oxford University Press, 2012), 16.
51 Gabriel Schoenfeld, Necessary Secrets: National Security, the Media and the Rule of Law
(New York: Norton, 2010), 40; see also Donahue, The Future of Foreign Intelligence, 32,
and Elaine Scarry, Rule of Law, Misrule of Men (Cambridge, MA: MIT Press, 2010), xvii.
52 2013 FC 1275, CanLII, https://www.canlii.org/en/ca/fct/doc/2013/2013fc1275/2013fc1275.pdf.
53 2016 FC 1105, https://decisions.fct-cf.gc.ca/fc-cf/decisions/en/item/212832/index.do.
54 R v Spencer, 2014 SCC 43, https://scc-csc.lexum.com/scc-csc/scc-csc/en/item/14233/
index.do.
55 Law Reform Commission of Canada, Electronic Surveillance (Ottawa: Information
Canada, 1986), 12, 17.
56 Gabe Maldoff and Omer Tene, “‘Essential Equivalence’ and European Adequacy after
Schrems: The Canadian Example,” Wisconsin International Law Journal 34 (2016): 211–83,
SSRN, https://ssrn.com/abstract=2896825.
57 Michael Warner, The Rise and Fall of Intelligence: An International Security History
(Washington, DC: Georgetown University Press, 2014), 263.
58 Gabriela Baczynska, “Deal to Share Passenger Info between EU and Canada Struck
Down on Privacy Concerns,” 26 July 2017, http://www.cbc.ca/news/business/airline
-travel-deal-1.4222074.
59 Zachary K. Goldman and Samuel J. Rascoff, “The New Intelligence Oversight,” in Global
Intelligence Oversight: Governing Security in the Twenty-first Century, edited by Zachary
K. Goldman and Samuel J. Rascoff (New York: Oxford University Press, 2016), xxix.
 On Denoting and Concealing in Surveillance Law 55

60 Judicial Redress Act of 2015, 5 USC.

61 Daniel Castro and Alan McQuinn, “Beyond the USA Freedom Act: How US Surveil-
lance Still Subverts US Competitiveness,” June 2015, Information Technology and Inno-
vation Foundation, http://www2.itif.org/2015-beyond-usa-freedom-act.pdf?_ga=1.
114044933.369159037.1433787396; see also Claire Groden, “NSA Spying Is Going to
Cost the Tech Sector Much More Than We Thought,” Fortune, June 2015, http://fortune.
com/2015/06/09/surveillance-tech-sector/.
62 David Omand, Learning from the Secret Past: Cases in British Intelligence History
(Washington, DC: Georgetown University Press, 2011), 7.
63 Internet Archive, “Global Surveillance Disclosures,” https://archive.org/details/nsia-
snowden-documents; Electronic Frontier Foundation, “NSA Primary Sources,” https://
www.eff.org/nsa-spying/nsadocs; Snowden Surveillance Archive, https://snowdenarchive.
cjfe.org; Technology, Thoughts and Trinkets, “Canadian SIGINT Summaries,” Technol-
ogy, Thoughts and Trinkets, https://christopher-parsons.com/writings/cse-summaries/;
Guardian, “The NSA Files,” https://www.theguardian.com/us-news/the-nsa-files; ACLU,
“NSA Documents,” https://www.aclu.org/nsa-documents-search. See also the related
WikiLeaks, “The Spy Files,” https://wikileaks.org/the-spyfiles.html, and National Secu-
rity Agency Central Security Service, “Declassification and Transparency Index,” https://
www.nsa.gov/news-features/declassified-documents/.
64 Ian Leigh and Laurence Lustgarten, In from the Cold: National Security and Parliamen-
tary Democracy (Oxford: Clarendon, 1994), 372–73.
65 Robert W. Hubbard, Peter M. Brauti, and Scott K. Fenton, “IMSI Catchers and MDI
Devices,” in Wiretapping and Other Electronic Surveillance, 96–100.
66 Radiocommunication Act (Subsection 4(4) and Paragraph 9(1)(b)) Exemption Order,
No. 2015-1, SOR/2015-36, http://laws-lois.justice.gc.ca/eng/regulations/SOR-2015-36/
page-1.html.
67 Canada, Bill C‑59, An Act respecting national security matters, 1st Sess, 42nd Parl, 2017,
http://www.parl.ca/Content/Bills/421/Government/C-59/C-59_1/C-59_1.PDF; see also
“Spy Bill Allows Government Security Agency to Collect ‘Publicly Available’ Info on
Canadians,” The Star, 21 June 2017, https://www.thestar.com/news/canada/2017/06/21/
spy-bill-allows-government-security-agency-to-collect-publicly-available-info-on
-canadians.html.
68 Rahal Sagar, Secrets and Leaks: The Dilemma of State Security (Princeton, NJ: Prince­
ton University Press, 2013), 184; see also Schwarz, Democracy in the Dark, 224. The
next logical step would be to provide annual reporting and statistics on the use of all
these new powers – as is already done for earlier forms of electronic surveillance such as
wiretapping, hidden microphones, and covert video surveillance. This may not cure the
complexity problem noted above, but it could bring a good deal more coherence to the
understanding of the law and increase efficacy in the review of operations.
69 One can pick from the official nomenclature – “mass surveillance, bulk collection,
unspecified acquisition” – but the real question is how the law can be said to operate in an
environment respecting democratic norms. What the courts in the United States, Canada,
and the European Union are saying (the highest courts in the land, one might add) has
been rather chastening. Government collection of big data, metadata, bulk data, personal
datasets, identity data have all met with stern rebuke from our most senior judges.
70 R v Tse, 2012 SCC 16, https://scc-csc.lexum.com/scc-csc/scc-csc/en/item/8002/index.
do; Wakeling v United States of America, 2014 SCC 72, https://scc-csc.lexum.com/scc
-csc/scc-csc/en/item/14439/index.do; R v Spencer, 2014 SCC 43, https://scc-csc.lexum.
com/scc-csc/scc-csc/en/item/14233/index.do; R v TELUS Communications Co., 2013
SCC 16, https://scc-csc.lexum.com/scc-csc/scc-csc/en/item/12936/index.do.
 56 Christopher Prince

71 Canadian Press, “CSIS Slammed for End-Running Law to Snoop on Canadians Abroad,”
CBC News, December 2013, http://www.cbc.ca/news/politics/csis-slammed-for-end
-running-law-to-snoop-on-canadians-abroad-1.2472843.
72 “United States v. Jones,” SCOTUSblog (blog), http://www.scotusblog.com/case-files/
cases/united-states-v-jones/; “Police Use of ‘Sting Ray’ Cellphone Tracker Requires
Search Warrant, Appeals Court Rules,” Washington Post, 21 September 2017, https://
www.washingtonpost.com/news/true-crime/wp/2017/09/21/police-use-of-stingray
-cellphone-tracker-requires-search-warrant-appeals-court-rules.
3
Big Data against Terrorism
Stéphane Leman-Langlois

In the realm of national security, big data analytics aim at two things. The
first is pre-emption, or stopping an attack at the preparation stage. This might
involve communications analysis or the recording of purchases of dangerous
materials. It may also be possible to detect behaviour patterns that reveal the
“casing” of particular targets. The second aim is prevention, which needs to
occur further upstream, before violent action is being considered. Usual preven-
tion tactics involve network disruption or the identification of persons “at risk.”
The analytical tools used to extract “actionable” information from large data-
sets have yet to show results, however. Agencies tasked with maintaining order
and with law enforcement have mostly tackled terrorism through traditional
patrol and investigation techniques, and mostly starting with citizen reports.
Such so-called low policing organizations are quickly adopting high-tech “pre-
dictive policing” approaches that make use of big data analysis, but those are
most useful with common, high-frequency crimes. On the other hand, entities
tasked with national security missions outside of officially defined crimes, or
“high policing” organizations,1 have enthusiastically embraced big data surveil-
lance for the protection of the state, both in detection and interference missions.
This also remains less than conclusive to date, for reasons listed below.
Meanwhile, since the data in question involve large numbers of citizens, col-
lecting it has raised many questions about state power and individual autonomy.
Despite the overwhelming costs and the fleeting benefits, however, big data
approaches to terrorism flourish and are definitely here to stay. In this chapter,
I review the problems and then the costs of big data for national security, con-
cluding with the incentives that keep it alive.

Hurdles for Big Data Analytics in Terrorism Prevention

Rarity. Terrorism, even if one uses the widest possible definition, is vanishingly
rare in the West. In Canada, it is rare enough to be considered statistically non-
existent. This of course means that there simply is no big data directly about
terrorism, and that if trends continue, there never will be. There are three types
of big data solutions to this “problem.”
The first is to identify patterns of terrorist or “pre-terrorist” (early preparation)
activity. Here, the method consists of matching observed sets of activity with
known terrorist behaviour. In short, the analytical process rests on a distillation
58 Stéphane Leman-Langlois

of what terrorists, in general, have done when planning past attacks. Unfortu-
nately, this distillation is vulnerable in many ways. First, our knowledge of
terrorist biographies is incomplete at best. Most were collected through state-
ments given to police, court proceedings, and direct or indirect testimony. Each
of these sources is weighted by different, but highly consequential, biases. So
these biographies are reconstructions, framed by the demands of the circum-
stances in which they are produced. This is true of even the most factual aspects
of the stories being told.
Second, much external data on terrorist activity comes from sources whose
objectivity is overestimated. Official reports, stories in the media, and police
reports are in fact filters that process and classify data according to common
(usually wrong) knowledge about terrorism. In the case of police reports, we
must also take into account the dozens of cases involving non-trivial interven-
tion by undercover agents in the plotting.2
In Canada, the Korody-Nuttal case in British Columbia illustrates this prob-
lem. After months of investigation and trial, the court finally found that RCMP
personnel were so closely involved in the plotting of the attack that the two
defendants had been entrapped. Both were immediately freed. Intelligence
reports are also commonly used in the building of terrorism or extremism
models and are often taken as highly reliable. Yet they are often heavily flawed
and rest on highly questionable sources. In reality, untold amounts of “intelli-
gence” are based on show trials held in dictatorships around the world, security
services manipulation, and politically expedient designations used to neutralize
individuals, groups, and populations. For instance, the famous “returnees from
Albania” show trial in Egypt in 1999 gave rise to hundreds of wrongful identi-
fications in the West, including the issuance of security certificates in Canada.
One Mohamed Mahjoub was held in custody for over twelve years as the head
of the Vanguards of Conquest in Canada, an extremely dangerous Egyptian
terror group, which also happens to have never actually existed, according to
expert testimony in the case. Finally, even scientific, apolitical databases are
coloured by politically loaded definitions of “terrorism.” This is not a semantic
problem: it can multiply or divide the number of cases in extreme proportions.3
In turn, this variation is bound to affect any identification of possible “pre-
terrorist” behaviours.
Originally, the next step was to use these historical sources to identify behav-
iours or sequences of behaviours that had preceded past terrorist acts. These,
in turn, would become the red flags that automated search tools would look for
in current masses of data. This is what the Total (later, Terrorism) Information
Awareness (TIA) program at the US Defense Advanced Research Projects Agency
(DARPA) had planned to achieve in the frantic months following 11 September
 Big Data against Terrorism 59

2001.4 Current big data analytics are no longer looking for patterns in masses
of data. The idea is to develop both the patterns and the search strategy in
combination, in real time, one adjusting the other continuously. Though this
solves the post facto problem, it does almost nothing to counter the “garbage
in, garbage out” problem of unreliable data. At best, it can contribute in flagging
particularly egregious unintentional or deliberate errors.
The second strategy is to look at statistically atypical clusters of data traces
and transactions. The need for pre-established patterns disappears, as the target
moves from the needle to the haystack and the goal is to filter out whatever is
statistically normal. Of course, the parameters for “normal” behaviour are also
far from consistent, but outliers are relatively easy to identify. Large amounts
of fertilizer being ordered to a suburban bungalow, or multiple one-way airline
tickets being purchased at the same time, can stand out.
One approach is to start by a clustering method, first sorting out the base
rates for various behaviours in order to identify statistical outliers. Neural net-
works are especially adept at automatically forming and testing hypotheses
about regularities, through machine learning. This takes many subjective,
cultural, and simply erroneous assumptions, theories, and classifications out of
the loop.
However, both terrorist and non-terrorist behaviours are extremely diverse
and do not fall in clear categories, meaning that there is no way to identify the
extent of the intersection between the “abnormal but acceptable” and the “ter-
rorist” sets. Because of the rarity of terrorism, and the profusion of eccentric
behaviours, the number of false positives is likely to skyrocket. The result could
easily overwhelm policing organizations, who might be able to follow up only
a very small minority of computerized notifications.
Finally, big data analytics may also seek to identify clandestine, or “dark”
social networks. Such networks are different from those based on family, work
or leisure, friendships, and the like, because they are attempting to evade detec-
tion. This means that their communication structure is significantly different,
and therefore might be detected. At the moment, most link analysis approaches
begin not with big data but with very small data: once a person is labelled a
terrorist, his or her contacts are simply singled out within the masses of captured
metadata (usually with a court-issued warrant). This is at once simpler and
more reliable, but rests heavily on the trustworthiness of the initial labelling.
When successful, this approach significantly narrows down the populations
that should be given additional attention by high policing, national security
intelligence organizations. In that way it is far more efficient, but it is not big
data analysis per se. Big data, or bottom-up/inductive graph analysis, has a dif-
ferent set of hurdles. One common strategy of network analysis is to daisy chain
60 Stéphane Leman-Langlois

communications through metadata. When sender, recipient(s), time of day,

duration of call, and frequency of communications are fed into the algorithm,
various types of network analysis statistical techniques can be used. Combined
with social psychology theory, this technique can reveal both the structure of
the network and the role of individuals within it. The proper big data way to
start the process is to access large amounts of bulk communications, and from
the data attempt to identify the networks that have common characteristics not
shared by most “normal” groups of telecom users (mostly time of day of the
calls, duration, and locality; statistically, they also tend towards low link density,
or fewer contacts among members). The problem with this analytical strategy
is that there is no base rate of legitimate dark networks (of political dissidents,
for instance). Each statistical study uses only confirmed illegitimate or criminal
networks in its analysis (for instance, Jennifer Xu and Hsinchun Chen simply
treat “dark” as synonymous with “illegitimate”).5
Each of these strategies circumvents the rarity problem, but in so doing also
generates new problems. There are two additional fundamental hurdles.

Math
Using big data analytics to prevent rare events is an overly optimistic scenario
for two other, more fundamental reasons. The first one is the Bayes theorem,
which simply uses basic probability calculations to underline the impracticality
of such predictions. Simply put, and only considering false positives, even a
near-magical 99 percent accurate algorithm would wrongly identify 1 percent,
or 320,000 random Canadians, as suspects (out of 32 million adults). Again,
this is not a realistic, practical scenario for policing. Of course, that still leaves
false negatives, those whom the algorithm falsely labelled non-dangerous but
who will in fact still commit crimes. Given this, it is difficult to detect any added
value in the use of big data analytics in the so-called War on Terror. In fact, it
is obviously not going to work as the proverbial “added layer” of security or as
an additional tool. In practice, it would drain disproportionate resources away
from the other layers into an effort that is extremely unlikely to pay off.
The other mathematical problem is the explosion of spurious correlations. It
was first raised by data scientist Jeff Jonas6 and has grown into a humour industry
on the web.7 Fortuitous or spurious correlations of variables increase rapidly
with the amount of available data, become plentiful, and eventually form an
entirely new type of haystack. They are not identifiable with statistical tools and
can show extremely high statistical significance. Only human logic can tell that,
for instance, the high correlation between the marriage rate in Kentucky and
the number of drownings after falling out of a fishing boat in the last decade is
probably meaningless (Tyler Vigen has an inventory of over 30,000 such
 Big Data against Terrorism 61

examples).8 Put another way, just as it has been said that 600 monkeys on 600
typewriters, given enough time, would eventually rewrite the works of Shake-
speare, given enough data it seems that any hypothesis, however eccentric, can
be proven (and, given enough data, in no time at all). Since much data mining
is the identification of patterns and the algorithmic comparison of digital traces
of behaviour, this problem is likely to become intractable with the ever-increasing
mountains of data. This is sure to shatter hopes that big data is a direct path to
reality: on the contrary, the more data, the greater the need for theory.

Civil Liberties
The acquisition and analysis of masses of private data is sure to raise more than
a few eyebrows. It may spur resistance within the public, or at least within
subgroups with sufficient political capital, who might successfully block or
hamper efforts to exploit big data analytics. Such attacks may diminish the
political and social capital of organizations that use or promote big data analytics
as “moral entrepreneurs.” Moral entrepreneurs establish themselves as “owners”
of social problems, in the sense that they have the authority to define them and
to elaborate, to suggest – if not to directly apply – corresponding solutions. Such
organizations are facing a difficult dilemma. On the one hand, there is a risk
that the occurrence of a terrorist attack could diminish the power of their
claimed expertise in national security. On the other, aggressive surveillance and
social control tactics might undermine their legitimacy with the public and
increase their vulnerability to criticism.

Adverse Consequences of Big Data Analytics in Terrorism Prevention

Sources of data are already influenced by the intense cultural focus on selected
populations of “terrorists.” Most police forces, for instance, identify “Islamist”
terrorism as the most important source of terrorist risk. The Canadian Security
Intelligence Service (CSIS), which informs the federal government on risks to
national security, has consistently rated Islamist extremists as the most import-
ant threat to Canada and Canadians, even when the actual number of related
attacks and plots has remained very low for decades. It is no surprise, then, that
Arab, Muslim, or Arab Muslim populations are overrepresented in any database,
which in turn further skews the models towards the same populations.
Police reports produce much of the information included in databases, espe-
cially the confidential ones used by national security organizations. This creates
a feedback loop where the data are collected and analyzed under a closed set of
assumptions, expectations, and definitions dictated by police objectives. This
problem is common to all crime statistics, but it is especially deep with terror-
ism cases because a significant proportion of the latter are the product of
62 Stéphane Leman-Langlois

preventive investigations led by undercover police. From the start, such inves-
tigations tend to format situations with conventional categories. The resulting
data, then, form a very specific type of social construction determined by
policing objectives. It is difficult to see how any deep learning activated within
this closed universe might avoid assimilating and reproducing the structural
bias.
In the case of algorithmic profiling, we have seen that even the best-designed
system will generate massive numbers of false suspects. This will certainly have
the effect of swamping police and security forces, making them less, not more,
efficient. The year 2014 was an unusually busy one for counterterrorism in
Canada, with two separate “lone wolf ” attacks. Michael Zehaf-Bibeau invaded
the Canadian Parliament with a firearm and Martin Couture-Rouleau hit mil-
itary personnel with his car. This prompted RCMP management to reassign
hundreds of investigators who were working on organized crime to terrorism
investigations. The case of Couture-Rouleau was particularly damaging for
police, who had interviewed him many times in the weeks prior to his attack.
This clearly indicates that a sudden influx of hundreds of leads would either
destabilize the system or require massive new investments in policing.
The glut of false suspects would subject thousands of computer-selected
persons to excessive scrutiny from police organizations. This scrutiny might
extend to the international level. As shown by a recent Federal Appeal Court
case (X(Re), 2013 FC 1275), Canadian intelligence agencies have subcontracted
the surveillance of Canadians to foreign Five Eyes partners, this without
informing the court when obtaining the original interception warrants. The law
has since been amended to allow the practice, which should make watching
suspects around the globe much easier. Hundreds of Canadians will be travel-
ling with secret “terrorism” labels affixed to their passports, the consequences
of which are well illustrated by the cases of Canadians Maher Arar, Abdullah
Almalki, Muayyed Nureddin, and Ahmed Abou-Elmaati. Each was the victim
of torture abroad due to misidentification by Canadian police as a terror
suspect.9
As David Lyon notes, data mining and especially clustering and network
analysis amplify social sorting.10 Neutral technology, once deployed in a socially
stratified world, can only reinforce and exaggerate the stratification. Automated
network analysis categorizes users according to their communication habits.
This includes frequency of use, time of day, chain of contacts, duration of calls,
attachments in emails, and so on, and produces a typology based on the esti-
mated risk of belonging to a radical network.
Many chilling effects associated with mass surveillance will also appear.
Research has shown that surveillance modifies the way we express ourselves,
 Big Data against Terrorism 63

even if we are convinced we have “nothing to hide.”11 Elizabeth Stoycheff found

that fewer of those polled were willing to discuss the power of the National
Security Agency (NSA) once they were informed that the NSA might intercept
their words.12 Surveillance, even without intervention, affects our feeling of
autonomy from the state apparatus. It lowers our ability, and in fact our desire,
to speak freely. This reduced autonomy is certain to affect areas other than
speech. For instance, geolocation may deter people from visiting urban areas
associated with illegal or potentially embarrassing behaviours.
In principle, big data analytics are driven by algorithms: it is robotic surveil-
lance. This form of machine watching is often presented as free from the potential
for abuse. Machines are devoid of personal interest and are not prone to preda-
tory behaviours such as blackmail. In the case of human-designed algorithms,
there subsists a potential for misuse at the programming stage. But learning
machines, such as neural networks, start with only basic programming and
improve themselves. One hope is that this could neutralize any original bias
built in at the programming stage, such as ethnic profiling, if it does not fit the
data. As we have seen, however, most databases that have to do with terrorism
contain structural biases. To the extent that robots “learn” from these data, they
could exponentially increase the effects of the original bias.

Incentives for Adoption of Big Data Analytics in Terrorism Prevention

Given the hurdles and adverse consequences, the accelerated expansion of big
data analytics in the context of national security is puzzling. The paradox is not
unlike that which exists in the world of camera surveillance. Cameras are also
accepted, if not defined, as the ready solution to almost any particular security
problem, even when clear evidence of lack of results is available. Cameras flour-
ish because of several important factors: their non-threatening appearance – they
are sometimes called “protection cameras” in order to enhance their inoffensive
aspect; their relatively low cost, especially since maintenance or operational
costs are never taken into account; their “democratization,” which lowers the
perceived risk to privacy and autonomy through familiarity; their assumed
objectivity in showing reality; the conviction that surveillance solves all mis-
behaviour problems – or, in fact, that undesirable behaviours are caused by
deficient surveillance; the high-tech “disruptive,” “innovative,” “smart,” and
“connected” sales pitch; and the risk mitigation approaches promoted by the
insurance industry, which often require the installation of cameras.
Many of the same general factors are at work with big data analytics. Though
they are part of the suspicious, unfamiliar world of programmers, mathemat-
icians, and cutting-edge computer science, they are not threatening because
they are invisible. Second, they are usually sold as a low-cost replacement to the
64 Stéphane Leman-Langlois

heavy manpower necessary to watch massive amounts of information. Third,

automated systems and artificial intelligence are increasingly used in many
devices in the consumer market, including toys, home appliances, and of course
cellphones, which increases their familiarity and lowers the potential dread
associated with non-human intelligence. Fourth, big data analytics are presented
as the apex of pure objectivity, free from human desires, assumptions, personal
and institutional agendas, and politics. Fifth, as physical surveillance solves
problems in the real world, automated surveillance is seen as solving behavioural
problems in cyberspace. There is much demand for monitoring behaviours such
as bullying, terrorism, sexting, suicidal thoughts, and so on. Finally, the very
word “algorithm” has come to symbolize progress, innovation, and “disruptive-
ness,” as in “the Uber algorithm” or “the Amazon algorithm.”

Practical Incentives
Beyond conventional wisdom, there are a few more tangible incentives for the
adoption of massive data analytics in security policing. Most modern organiza-
tions, whether public, private, or hybrid, are already collecting large amounts
of data on every facet of their day-to-day operations. With exponentially more
to come in the near future, most are trying to keep afloat by improving their
data centres or moving them to the cloud. And it so happens that every data
centre management software vendor, as well as all cloud providers, now sell
their services with “big data analytics” included in some way.
Policing organizations have not been immune to either the data glut or the
sales pitch. They contribute to the latter by claiming that better analytics will
lead to better security. This is taking place as most police organizations are in
the course of adopting in part or in whole the language, the tactics, the analytical
methodologies, the policies, and the management style associated with the
“intelligence-led policing” (ILP) model. ILP rests heavily on the efficient use of
the data available to police in order to maximize their impact on crime statistics.
As such, it is made to measure for the world of big data.
One of the most powerful factors of acceptance of big data surveillance is its
claimed security function. For the time being, however, this is more a promise
than an actual result, as security successes are few and far between. The NSA
claims that its massive interception and analysis capabilities were used in some
fifty-four investigations since 2001. This seems like a rather meagre return on
investment, considering the astounding resources engaged. On close scrutiny,
the technology played what might charitably be called a peripheral role at best.13
Of course, it could be argued that the data involved are simply not yet big enough:
most of the NSA holdings are communications intercepts. The original concept
of the TIA project was to gather all kinds of data, from everywhere, including
 Big Data against Terrorism 65

financial institutions, utilities, transportation, hospitals, clinics and pharmacies,

schools, large retail enterprises, distributors of specific services, goods (such as
books, chemicals, tools, weapons), and information. With enough data, security
would finally be attainable.
Meanwhile, the “security” provided by this colossal apparatus remains a rather
vague concept. As adopted by police, intelligence agencies, politicians, and the
media, the concept blurs the distinction between security outcomes and security-
labelled activities. For instance, most of the debates about mass surveillance
that followed the Snowden papers centred on “balancing” rights not with actual
security outcomes but with (always increasing) security-labelled powers,
budgets, and tactics. The effect of this blurring, other than wrapping unproven
police tactics in the indisputable legitimacy blanket of “security,” is to confuse
the means and the goal and to transform the mere implementation of new
strategies into security successes. This is politically and administratively useful
because it depicts policing organizations as knowledgeable, diligent, and respon-
sible, stable rocks in the morass of global uncertainty and insecurity.
What is more, massive data is often presented as freeing organizations from
the equally murky world of the overly politicized social sciences.14 This is espe-
cially important in this era of evidence-based policing. Under this view, the
social sciences rest on a defective scientific model involving subjective theories,
incomplete, cherry-picked observations, and, too often, the value judgment of
the researcher. This scientific process should be replaced by the direct contact
with reality and its workings that big data analytics now offer us. In other words,
big data is defined not only by its size, speed, variety, and so on but by the way
it bridges the gap between the data and the thing, between the map and the
land. This seems like a rather obvious misunderstanding of the nature and
function of theory in understanding the world, one that allows its replacement
by the spurious correlations noted above.
Finally, it is often argued that big data brings us not a fearsome new era of
total surveillance but one of renewed privacy. As noted above, the machine-led
“privacy” protecting data from human misuse was a cornerstone of the defence
of the TIA program. It was also an important response offered by the NSA to
the Snowden revelations. When Director of National Intelligence James Clapper
claimed that information was not being “collected,” he merely meant that humans
could never see it all. Only a small slice became visible, with court-issued per-
mission – even though gargantuan flows of data were finding their way to NSA
servers. Since the future of all forms of technical surveillance will in fact be big
data analytics, machine-led privacy is bound to become the leading response
to claims of loss of privacy or autonomy. Of course, one flaw is that machines
tend to be programmed with human politics and agendas in the first place. Part
 66 Stéphane Leman-Langlois

of the promise of big data is to circumvent this shortcoming, and we should

expect that it will soon be possible to argue convincingly that this has been
achieved to a satisfactory extent. Then, in the near future, we will have to decide
if we collectively feel that our privacy and autonomy are invaded or reduced by
machines programmed for outcomes we have set ourselves.

Conclusion
It seems rather unlikely that national security will be better protected with the
adoption of big data analytics. The extreme rarity of national security threats
virtually ensures that unwanted consequences will outnumber prevention “suc-
cesses.” What it will achieve is always-on or ambient surveillance, where omni-
present robotic systems collect continuous, fine-grained information about our
daily activities. The promise of complete security through machine analytics
will become impossible to resist, especially if new signal crimes alert public
opinion to the police complaint of “going dark” because of increasingly prevalent
anonymization and encryption in communications.

Notes
1 Jean-Paul Brodeur and Stéphane Leman-Langlois, “Surveillance-Fiction: High and Low
Policing Revisited,” in The New Politics of Surveillance and Visibility, edited by Kevin
Haggerty and Richard Ericson (Toronto: University of Toronto Press, 2006), 171–98.
2 John Mueller, Terrorism since 9/11: The American Cases (Columbus, OH: Mershon Center
for International Security Studies, Ohio State University, 2011).
3 Stéphane Leman-Langlois and Jean-Paul Brodeur, Terrorisme et antiterrorisme au Can-
ada (Montreal: Presses de l’Université de Montréal, 2009).
4 Jean-Paul Brodeur and Stéphane Leman-Langlois, “Surveillance-Fiction,” 171–98.
5 Jennifer Xu and Hsinchun Chen, “The Topology of Dark Networks,” Communications of
the ACM 51, 10 (2008): 58–65.
6 Jeff Jonas and Jim Harper, “Effective Counterterrorism and the Limited Role of Predictive
Data Mining,” Policy Analysis 584 (2006), https://www.cato.org/sites/cato.org/files/pubs/
pdf/pa584.pdf.
7 For instance, “Spurious Correlations,” tylervigen.com, http://www.tylervigen.com/
spurious-correlations.
8 Ibid.
9 The Honourable Frank Iacobucci, Internal Inquiry into the Actions of Canadian Officials
in Relation to Abdullah Almalki, Ahmed Abou-Elmaati and Muayyed Nureddin (Ottawa:
Public Works and Government Services Canada, 2008).
10 David Lyon, “Surveillance, Snowden, and Big Data: Capacities, Consequences, Critique,”
Big Data and Society 1, 13 (2014), https://doi.org/10.1177%2F2053951714541861.
11 See Elizabeth Stoycheff, “Under Surveillance: Examining Facebook’s Spiral of Silence
Effects in the Wake of NSA Internet Monitoring,” Journalism and Mass Communica-
tion Quarterly 1, 16 (2016), https://doi.org/10.1177%2F1077699016630255; Jonathon
Penney, “Chilling Effects: Online Surveillance and Wikipedia Use,” Berkeley Technology
Law Journal 31, 1 (2016): 117–83, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2769645.
12 Stoycheff, “Under Surveillance.”
 Big Data against Terrorism 67

13 Bailey Cahall, David Sterman, Emily Schneider, and Peter Bergen, “Do NSA’s Bulk Sur-
veillance Programs Stop Terrorists?” (policy paper, New America, Washington, DC,
2014), https://www.newamerica.org/international-security/do-nsas-bulk-surveillance
-programs-stop-terrorists/.
14 Chris Anderson, “The End of Theory: The Data Deluge Makes the Scientific Method
Obsolete,” Wired, 27 June 2008, https://www.wired.com/2008/06/pb-theory/.
4
Algorithms as Suspecting Machines
Financial Surveillance for Security Intelligence
Anthony Amicelle and David Grondin

The concept of mobilities encompasses both the large-scale movements of

people, capital and information across the world, as well as the more local
processes of daily transportation, movement through public space and the
travel of material things within everyday life ... Fears of illicit mobilities and
their attendant security risks increasingly determine logics of governance and
liability protection within both the public and private sectors.
– Kevin Hannam, Mimi Sheller, and John Urry,
“Mobilities, Immobilities and Moorings”

Mobilities raise security concerns, and technology is seen as crucial to

ensuring the everyday securing of mobilities in the digital age. Technological
innovations in digital computing and big data analytics now play a central role
in managing the security/mobility nexus, the “dynamic tension between free-
dom of mobility and the provision of security.”1 Controlling mobility increas-
ingly involves “new” technologies to deal with what is presented as the dilemma
of how to facilitate movements of people, money, and data while enforcing the
laws against illicit mobilities. Despite its importance, however, little is known
about the set of problems posed by the choice and use of the new technologies
that make mobility controls operational for law enforcement and counterterror-
ism purposes.
This chapter examines what it means to govern through algorithms in the
age of big data surveillance. To that end, we look at how algorithms work as
logistical media to monitor mobilities for crime control and national security
purposes. We then introduce our case study of security intelligence and big
“dataveillance” in relation to financial mobilities. We propose the idea of “sus-
pecting machines” to explore the use of algorithmic devices that produce alerts
regarding suspicious financial flows and conduct, which are usually focused
on money laundering and terrorism. Our case study involves document analysis
and over thirty interviews with a representative sample of officials from domes-
tic banks in Canada as well as Financial Transactions and Reports Analysis
Centre of Canada (FINTRAC) officials in Ottawa, Toronto, and Montreal in
2015 and 2016.
 Algorithms as Suspecting Machines 69

Policing Financial Flows with the Security/Mobility Nexus

We use the term “security/mobility nexus” to suggest how security and mobility
have been increasingly stitched together to function as part of a governmental
regime that manages and polices mobility flows through digital technologies
and surveillance configurations.2 The mobility regimes of people, money, and
data, while often distinguished heuristically for analytical purposes, are in fact
closely related, and we contend that analysis should acknowledge their connec-
tion and interdependence in security discourses and practices. It is no accident
that security programs with dataveillance capabilities – such as the US terrorist
finance tracking program3 – are officially promoted for their ability to recognize
interconnecting mobilities. According to the US Treasury Department and the
European Commission:

Terrorist finance tracking program-derived information may be used to provide

leads that assist in identifying and locating persons involved with terrorist net-
works and providing evidence of financial activities in aid of terrorist attacks.
For example, it is possible to locate a suspect by checking when and where the
suspect closed and/or opened a new bank account in a city or country other
than his or her last known place of residence. This is a clear indicator that the
person may have moved. However, even when a suspect does not change bank
accounts but rather moves and continues using the “old” account (e.g., through
e-banking), it has been possible to detect the change of location by, for example,
identifying payments for specific goods or services (e.g., for repairs or mainte-
nance or other activities which are usually carried out where a person lives) ...
The terrorist finance tracking program can provide key information about the
movements of suspected terrorists and the nature of their expenditures.4

The surveillance-related connection to mobilities is explicit in this context. On

the basis of digital data about financial transactions, this security program traces
the movement of money to track terrorist suspects. In other words, it relies on
the digital traces of financial mobilities to find and follow the physical traces of
human mobilities.

Governing Finance with Algorithmic Infrastructures of Security

In this chapter, we are specifically interested in the policing of financial flows
to understand how people and data are governed and controlled through algo-
rithmic infrastructures of security. Using the “haystack metaphor” at the core
of current security intelligence discourses,5 providers of surveillance devices
describe the detection of illicit financial flows as the “ultimate search for the
70 Anthony Amicelle and David Grondin

needle in the haystack.”6 To see how that goal is achieved, we investigate the
policies and practices involved in financial surveillance as well as the detection
and reporting of “suspicious transactions.” But who is in charge of finding the
needle of dirty money in the Canadian financial haystack? As Jef Huysmans
suggests in his work on contemporary policing, financial policing is partly
detached from the institution of the police as it deals with the more general
associative practices found in networks of risk knowledge, technologies, and
agencies.7
In Canada, over 31,000 businesses – from the real estate sector to the banking
industry – must comply with legal obligations dealing with dirty money, includ-
ing reporting suspicious transactions to the relevant state authority, namely,
FINTRAC. As Canada’s financial intelligence unit, “FINTRAC receives reports
from financial institutions and intermediaries, analyzes and assesses the reported
information, and discloses suspicions of money laundering or of terrorist finan-
cing activities to police authorities and others as permitted by the Act. FINTRAC
will also disclose to CSIS [Canadian Security Intelligence Service] information
that is relevant to [a] threat to the security of Canada.”8 FINTRAC receives over
200,000 reports of suspicious transactions annually, mainly from banks, which
have become the main source of denunciation to the state,9 practising financial
dataveillance in the name of security through “the systematic use of [financial-
related] personal data systems in the investigation or monitoring of the actions
or communications of one or more persons.”10
In an attempt to meet their legal responsibilities regarding anti–money laun-
dering and counterterrorism financing, numerous banks have implemented
algorithmic infrastructures to monitor suspicious activity and help make sense
of the daily avalanche of transaction-related digital data. Our aim in discussing
the seemingly invisible work of these security/mobility infrastructures is to shed
some light on the impact and productive power of governing with algorithms
in an era characterized by big data surveillance. We seek to show how doing so
changes former practices, highlighting how socio-technical devices have become
part of security apparatuses, as we explore financial big dataveillance practices
through the algorithmic instrumentation that makes everyday surveillance and
denunciation possible and stable over time.

Making Sense of Algorithms as Logistical Media

of Big Financial Dataveillance
“Logistical media – as technologies, infrastructure, and software – coordinate,
capture, and control the movement of people, finance and things. Infrastructure
makes worlds. Logistics governs them.”11 This quotation captures Ned Rossiter’s
logistical media theory of infrastructures and suggests how thinking about
 Algorithms as Suspecting Machines 71

algorithms and security brings the issue of infrastructures to the fore.12 To

understand security/mobility, we must understand what infrastructures do in
the background – what they “do” when they work. As Keller Easterling says,
infrastructures actually do something: they make “certain things possible and
other things impossible.”13 To grasp how mobilities of people, money, data, or
objects are governed according to an algorithmic security rationality,14 we must
be keenly attentive to the power of infrastructures, particularly to algorithms
as digital infrastructures acting invisibly while governing mobilities through the
digital socio-technical environment.
“Infrastructure is defined by the movement or patterning of social form. It is
the living mediation of what organizes life: the lifeworld of structure.”15 This
relational understanding echoes Susan Leigh Star’s groundbreaking framing of
“infrastructure [a]s a fundamentally relational concept, becoming real infra-
structure in relation to organized practices.”16 As John Durham Peters points
out, infrastructures set the terms for operations: “The job of logistical media is
to organize and orient, to arrange people and property, often into grids. They
both coordinate and subordinate, arranging relationships among people and
things.”17 Algorithms, understood as logistical media governing mobilities, act
as infrastructures that organize life. Recognizing this underscores the import-
ance of understanding how governance is achieved through algorithms. As
argued by Rob Kitchin:

Algorithms cannot be divorced from the conditions under which they are devel-
oped and deployed ... What this means is that algorithms need to be understood
as relational, contingent, contextual in nature, framed within the wider context
of their socio-technical assemblage. From this perspective, “algorithm” is one
element in a broader apparatus which means it can never be understood as a
technical, objective, impartial form of knowledge or mode of operation.18

Louise Amoore and Rita Raley construe the algorithm “as both technical process
and synecdoche for ever more complex and opaque socio-technical assem-
blages.”19 To put it simply, an algorithm is a mathematical formula or a set of
instructions or rules that enable its users to obtain a specific result using its
computational ability to sift through large amounts of data and accomplish tasks
that human agents could not perform in a timely manner.20 More tellingly,
algorithms have become highly publicized artifacts that populate our everyday
life, with their logic woven “into the very fabric of all social processes, inter-
actions and experiences that increasingly hinge on computation to unfold.”21
Looking at the security/mobility nexus enables us to analyze surveillance and
infrastructure in the digital world. As part of a secretive security infrastructure,
72 Anthony Amicelle and David Grondin

algorithms operate in the background, not unlike the “black box society” that
Frank Pasquale describes in relation to the hidden algorithms that control
money and information, affecting decisions made on the mistaken assumption
that the information they are based on is neutral and technical. Pasquale uses
the black box to recognize both its function as a recording device for monitoring
technical data in transportation (cars, planes, and trains) and its metaphorical
use suggesting the obscurity and opacity of complex systems whose inner work-
ings we cannot easily decipher or reveal.22 Thinking about what algorithms do
with dataveillance means taking into account how vast streams of digital data
about people’s life, conduct, and mobility are sifted and sorted.

On Suspecting Machines: Surveillance in the Canadian Financial System

Our case study of security intelligence and big financial dataveillance in Canada
provides clear examples of the critical importance of algorithms in contempor-
ary “modes of policing that combine risk communication and surveillance to
technologically assemble and circulate suspicion.”23 We introduce the expression
of “suspecting machines” to account for the pivotal role played by technologic-
ally driven digital surveillance devices in monitoring, detecting, and reporting
suspicious activity. Financial technology providers sell financial institutions
what they describe as “suspicious activity monitoring” solutions, using adver-
tisements such as this:

The Actimize Suspicious Activity Monitoring (SAM) solution combines cutting-

edge technology with years of anti-money laundering human-expertise, help-
ing to ensure accurate alert detection, increased team productivity and lowered
compliance program costs. Anti-money laundering automation brings together
AI, machine learning and robotic process automation, the solution enables end-
to-end coverage for the detection, scoring, alerting, workflow processing and
suspicious activity reporting. As a result, anti-money laundering departments
can more effectively monitor suspicious activities, be confident they’re focusing
on the right issues and risks, and automate processes while retaining the power
to make the final decision.24

Before examining the algorithms in action and looking at their specific findings
beyond those described in the promotional material, let us briefly paint a rough
picture of our case study.

The Context for the New Financial Intelligence Systems

On one side, we have 31,000 reporting entities, which include banks as well as
life insurance providers, dealers in precious stones, real estate brokers and
 Algorithms as Suspecting Machines 73

developers, accountants, casinos, and securities dealers. These entities are

responding to legal requirements regarding anti–money laundering and counter-
terrorism. The banks, for example, wish to be seen as compliant with their
specific security-related obligations but, as for-profit providers of financial
services, not security services, they also wish to remain profitable. On the other
side is FINTRAC. “As Canada’s financial intelligence unit and anti-money
laundering and anti-terrorist financing regulator, FINTRAC plays a critical role
in helping to combat money laundering, terrorism financing and threats to the
security of Canada.”25 FINTRAC relies first and foremost on non-security-
oriented businesses (the reporting entities), which act as the eyes and ears of
the state in the field of finance. As summed up by a FINTRAC spokesperson,
“the reporting to us is absolutely critical. Without those reports, FINTRAC is
out of business.”26
Banks report both suspicious and certain transactions to FINTRAC. Accord-
ing to official guidelines at the time of our study, suspicious transactions are
financial transactions that reporting entities have “reasonable grounds to sus-
pect” are related to the commission or attempted commission of a money
laundering offence and/or the commission or attempted commission of a ter-
rorist activity financing offence.27 According to FINTRAC, “‘reasonable grounds
to suspect” is determined by what is reasonable in your circumstances, taking
into account normal business practices and systems within your industry. While
the Act and Regulations do not specifically require you to implement or use an
automated system for detecting suspicious transactions, you may decide that
such a system would be beneficial to your business.”28 While the last sentence
suggests that using an automated system is voluntary, a FINTRAC officer stated
in an interview: “They have no choice. Listen, all the big banks I know, they
have an IT system for that.”29 The legal framework creates a context of practice
that pressures institutions such as banks to rely on suspecting machines to meet
compulsory security demands. The regulator wants those regulated to comply
with the law, while recognizing that a small real estate agency does not need –
and cannot afford – the apparatus used by the biggest banks in the country to
identify clients, keep records, monitor financial conducts, and, ultimately, report
suspicious transactions. Faced with a daily avalanche of transaction-related
digital data – over 150 million transactions per month at each of the top six
financial institutions in Canada – institutions hoping to make sense of the
totality of digital noise generated by business relationships and account activities
have no choice but to implement technologically mediated “solutions” if they
wish to be stamped compliant. As a result, the non-binding nature of the formal
reference to automated surveillance devices becomes, in practice, a categorical
obligation for banks in particular, both in Canada and internationally:
74 Anthony Amicelle and David Grondin

“Regulators are pressuring smaller financial institutions to adopt more auto-

mated solutions while pushing larger financial institutions to upgrade their
current solutions’ sophistication.”30
This process has created tensions within the banking sector at two levels: (1)
internal debates about the conflicting demands of security versus profit, espe-
cially over the costs of meeting the financial surveillance requirement; and (2)
the actual socio-technical challenges posed by suspecting machines and their
interaction with human resources. As with any informational infrastructure,
this requires “making visible an infrastructure that, in other contexts, has been
rendered invisible” and making it meaningful and useful for “radically different
users.”31 Meeting these requirements is a challenge as much for researchers as
it is for financial and security practitioners; our findings highlight the tensions
and difficulties in big financial dataveillance. We show that looking at how
logistics govern “practice” makes clear how algorithmic infrastructures, as
logistical media, are part of governance. Only when we focus on how these
socio-technical devices are called upon, enlisted, solicited, and made part of
security infrastructures can we recognize the “human and institutional choices
[as well as the science, rationalities, and technologies] that lie behind these cold
mechanisms.”32

Between the Hammer of Defensible Compliance

and the Anvil of Financial Burden
Regardless of their degree of willingness to help combat crime and terrorism,
banks are, above all, for-profit organizations. Trying to reconcile these two goals
has contradictory consequences for their use of financial surveillance equip-
ment. On the one hand, as extensively discussed in the academic literature,
bank representatives are very sensitive about their reputation in relation to
organized crime and terrorism issues.33 “Protect your reputation” is still one of
the main advertising slogans used by vendors to promote financial surveillance
devices. As one bank’s chief anti–money laundering officer reported:

I think the motivation to be compliant is, you know, you want to protect the
bank’s brand. And you want customers and the shareholders to feel as if their
bank is a safe bank in which they can conduct transactions and they go without
having to worry about. You know, is this bank going to be open tomorrow? Are
there issues associated with whom they are lending money to? Do they have good
business practices? And to comply allows us to give that type of confidence to
our customers, internally to our employees and also to our shareholders. We
look at it as a value proposition whereas before compliance was looked as a cost
of doing business or an expense. But, you know, we treat compliance as our
 Algorithms as Suspecting Machines 75

ability to contribute to the brand and to the safety and soundness of the organ-
isation. We think it is a competitive advantage to comply.34

The focus on possible fines and criminal or administrative penalties associated

with non-compliance complements the selling point of reputational protection.
Chief anti–money laundering officers also mention the fear of losing part or all
of their banking licences and the possibility of losses due to the reaction of the
other financial institutions – the critical business partners of correspondent
banks – if they are sanctioned for regulatory breaches. Public penalties for non-
compliance have a large impact on a bank’s reputation and risk score within the
industry. Indeed, any bank sanctioned for non-compliance with surveillance
and reporting obligations will be seen as high-risk across the industry, which
has tremendous financial consequences for maintaining or building vital rela-
tionships with other banks.
The internalization of compliance requirements as an integral part of business
in finance means that these requirements are seen as an inevitable regulatory
burden at worst and a competitive advantage at best. The irony of this internal-
ization is that the main concern of financial institutions is related to non-compli-
ance rather than money laundering and terrorist financing. Surveillance
technology, however necessary it may be for defensible compliance,35 is seen as
a costly investment that is not expected to provide a fair return to for-profit
organizations whose core businesses are clearly not involved with crime control
and national security issues. In this context, the distinction between saving
money (i.e., avoiding sanctions and reputational damage) and making money
(i.e., business as usual) is often mentioned to qualify and clarify the role of
technological developments:

There is still ... hum ... a bit of push back in cost. There is always a bottom line
cost in that, the requirement is you have to do X, Y and Z, now they have to adjust
their system and internal technology in order to meet that requirement ... And they
say, “you know, we cannot just go and do so many technological changes plus we
have a business to run. So, it is not all about you [the regulator] all the time.” So,
it is always about money and it is always about the time to do it [technological
improvement]. And they say, “we can’t do it now, then we will do it in a year.”
Well, a lot can happen in a year. Right? So, they push back on us, and their tech-
nology department pushes back on them, because the priority is making money.
For instance, they have to put a system in to collect service fees. That probably
takes priority over what we want them to do, because it is going to affect their
bottom line. We are costing them money. We don’t make them money. We can
save their reputation or risk, right! But they don’t see it that way. They don’t see
76 Anthony Amicelle and David Grondin

it that way at all, every time. And it is always, and it is not just the banks. It is
all of them [reporting entities]. They all push back because of the finance, the
financial burden. Burden. That’s what it is: a financial burden.36

The work of surveillance is also a cost challenge in a different way for bank
tellers who are the other main – and the traditional – source of internal alerts
in any financial institution. As a former compliance officer interviewed noted:

The bottom line is always the dollar. So, at the branch level, the customer service
representatives, or the loan manager, their major focus is to make money. To sell
by opening up accounts, sell credit cards, sell mortgages, sell loans, sell this ...
that is their job. It is to make that money. So, they have quota. They will say to
their tellers in one day-time you have to do X amount of loans or X amount of
mortgages, or X amount of credit cards. They push from the top. So you don’t
spend the time that you need to the KYC [Know Your Customer compliance
obligations] because the requirement of making money comes first again. So, if
as a teller I can sell a mortgage, if I can do two deals or sell a credit card in half an
hour that would be great, as oppose [sic] to doing one in an hour because I have
taken the time to have a conversation, do the KYC, file out everything properly.
The branches don’t have the time to do it. They are just under the gun of making
more revenue.37

The emphasis on making money also explains the diversity in complexity of

algorithmic infrastructure in the banking industry, with the level of technological
involvement widely dispersed on a continuum of financial surveillance equip-
ment that ranges from a minimalist high-tech setting to a maximalist one.

Uneven Distribution of Technological “Solutions”

with Suspecting Machines

The Minimalist High-Tech Version

Certain financial institutions have not fully implemented suspicious activity
monitoring devices. Their first technological investment has been in automated
devices to help comply with reporting obligations for monetary thresholds and
targeted sanctions rather than those for suspicious financial conduct.
For financial transactions that must be reported on the basis of monetary
threshold, such institutions have implemented software that is designed to detect
electronic fund transfers “of $10,000 or more out of or into Canada in a single
transaction or in two or more transactions totalling $10,000 or more made
within 24 consecutive hours by or on behalf of the same individual or entity,
 Algorithms as Suspecting Machines 77

through any electronic, magnetic or optical device, telephone instrument or

computer.”38 In 2018–19, FINTRAC received over 17 million electronic cross-
border fund transfer reports.39
For financial transactions that must be reported on the basis of national and
international sanctions of particular regimes, the financial institutions’ priority
has been to create consolidated digital lists of targeted entities using screening
and watch-list filtering devices that can compare customers and their financial
relations to the listed parties.40 In particular, banking actors must submit “terrorist
property reports” when they know or believe that financial assets are related to a
designated terrorist group or a listed person suspected to be associated with enti-
ties such as the Taliban, al-Qaeda, or ISIL (Islamic State of Iraq and the Levant).
A routine part of surveillance and filtering practices, with or without dedicated
technology, also consists in looking at news and open-source information.41
Finally, with regard to suspicious financial transactions, most financial insti-
tutions without fully operational suspicious activity monitoring devices have
begun implementing such procedures but they take time to complete, as noted
by one chief anti–money laundering officer we interviewed:

Initially it took three years just getting all the data feeds right, and going
through the banking system, data and figuring ... and I know we are not the
only one that has this issue ... but actually figuring where all the data is, and
how to get it into the right format, and be able to identify the type of transac-
tion ... So there is a lot of manipulation that goes on in the background to get
the data that you need in the transaction monitoring system, to be able to effec-
tively get meaningful alerts.42

The idea that technologically mediated suspicious activity monitoring is the

culmination of a long implementation process is accepted by FINTRAC
officials:

Two years, you are very optimistic! We are talking about banks, they have mil-
lions of customers, whether involved individuals, corporations, trusts, and so
on. And it is even worse when we talk about big financial institutions. The IT
system is not only for the bank but for all the subsidiaries of the group, insur-
ance, life insurance, investment activity and many more, in Canada and abroad.
Thus, the system often begins with the bank and it is gradually extended to the
other subsidiaries too. The task is enormous. It is a work in progress. Even when
you are there, you are still in the process of calibration because there are new
financial products, you have bought a competitor or you have forgotten a market
and so on.43
78 Anthony Amicelle and David Grondin

Before or during the implementation of the algorithmic device, the socio-

technical process of suspicion production requires much more manual labour
to generate internal alerts on financial conducts. Consider the example of a
financial institution with 10 million clients. Before implementation, surveillance
of financial conduct and production of related internal alerts must be largely
hand-done in two main ways.
First, bank tellers and bank advisers are in direct contact with clients. In
addition to the commercial objectives at the core of their work, they are encour-
aged to report any “unusual transaction” to their line manager or to the bank’s
anti–money laundering/counterterrorism financing compliance department.
Financial surveillance and internal reporting is thus largely focused on trans-
actions that depend on face-to-face interaction in a bank branch. Furthermore,
it relies on banking employees whose main tasks, priorities, and training are
not related to the fight against dirty money. Internal alerts are normally based
on specific behavioural indicators, although FINTRAC’s official guidelines note
that gut feelings matter: “As a general guide, a transaction may be connected to
money laundering or terrorist activity financing when you think that it (or a
group of transactions) raises questions or gives rise to discomfort, apprehension
or mistrust.”44 Unusual transaction reports, as internal alerts, are assigned to
dedicated analysts within the compliance department for further investigation.
This investigative process “results in [one of] two opposite outcomes with the
translation of the detected abnormality into a false-alert or a suspicious behav-
iour that must be reported” to the state authorities.45
Second, in addition to vigilance over direct business, one unit of the bank
compliance department is dedicated to monitoring the transactions and account
activities of clients “scored” as high-risk on a regular basis. According to one
bank’s risk-scoring practices, about 20,000 clients out of 10 million are con-
sidered high-risk in relation to money laundering and/or terrorist financing.46
Such financial surveillance and internal reporting is thus focused on a rather
small percentage of customers, i.e., the predefined high-risk minority. Looking
at the process after the implementation of a suspicious activity monitoring
device highlights how the new algorithmic infrastructure makes a difference
in everyday financial surveillance.

The Maximalist High-Tech Version

In the maximalist high-tech version of a suspicious monitoring device, previous
methods of surveillance of financial conduct and internal reporting of suspi-
cious activity still exist. However, the implementation of a dedicated algorithmic
device transforms the scope and systematization of financial surveillance. In
addition to filtering based on monetary threshold and watch lists, every single
 Algorithms as Suspecting Machines 79

financial transaction by each of the bank’s clients is now technologically mon-

itored on an everyday basis, leading to internal alerts about potentially suspicious
actions. These alerts are produced automatically overnight, ready for human
analysis the next morning. This blanket, technologically driven surveillance of
financial conduct does not mean that risk scoring no longer matters, as behav-
ioural and transactional indicators are weighted according to risk indicators.
In other words, risk score affects the production of alerts to the extent that,
depending on clients’ risk scores, the same financial conduct will or will not
generate an automated alert. More generally, suspicious activity monitoring
relies on a combination of indicator-related scenarios and analytics, plus addi-
tional features. In this respect, algorithmic devices, acting as suspecting
machines, rely on the deductive and/or inductive determination of abnormality,
unusualness, and, ultimately, suspicion.
Suspicious activity monitoring occurs automatically based on one of two sets
of algorithms. In the first set, the algorithms of detection are designed to uncover
transactions that match a predefined set of scenarios. These scenarios – also
called rules – are based on a combination of indicators from internal and external
sources, beginning with those provided by FINTRAC, which are intended

to help assess whether or not transactions might give rise to reasonable grounds
for suspicion. They are examples of common and industry-specific indica-
tors that may be helpful when evaluating transactions, whether completed or
attempted. They include indicators based on certain characteristics that have
been linked to money laundering or terrorist activities in the past. These indi-
cators were compiled in consultation with reporting entities, law enforcement
agencies and international financial intelligence organizations.47

FINTRAC’s hundreds of indicators are selected, combined with others, and

then tested – including fine-tuning or deletion – on a more or less regular basis
either by the technology providers or by individuals on the site where they will
be used:

There is continuous feedback from the triage group [which receives the auto-
matically generated alerts] to the analytics people [who deal with tuning on-site
scenarios]. They say, “You know, we keep seeing this and we keep seeing this. It
is nothing, it is nothing. Can we optimize the rules?” That, as a process, continu-
ally takes place from that perspective, in order to optimize the rules while, at
the same time, through typologies and other things, we build new rules to bring
in new alerts. It is a continuous process to maintain and keep your rules up to
date.48
80 Anthony Amicelle and David Grondin

In the second set, suspicious activity monitoring is automatically driven by

algorithms designed to uncover unusual and potentially suspicious transactions
rather than using comparison with a predefined set of information. This second
production-alert generation process is based on recognizing a customer’s devia-
tion from peer-group behaviour and/or historical and expected individual
behaviour. Such an alert is a by-product of social and behavioural sorting strat-
egies as well as comparison with historical data for account activity. Link analysis
is also used to explore a bank’s databases to identify associations among clients
(both individuals and businesses), bank accounts, and transactions (electronic
fund transfers, cash deposits, and so on). This social network approach both
supplements analysis of already detected suspicious financial conduct or discov-
ers new cases. The work of suspicious activity monitoring devices (and link
analysis) can be combined with the technologically mediated analysis of unstruc-
tured data sources such as websites (media, social networks, and so on) or “free-
text” narratives related to previous suspicious transaction reports.
Ultimately, it goes without saying, there is wide variation between financial
institutions along the continuum from the minimalist version to the maximalist
version of financial surveillance equipment. Despite the technological asym-
metries, however, every bank faces the same challenge in attempting to find a
way to make money while avoiding the pain of being caught between the ham-
mer of defensible compliance and the anvil of financial burden resulting from
false-positive alerts.

Managing False Positives

“[Five percent] to 10% of all alerts actually turn into cases. So, 90% to 95% are
false positives. And then, from a case to a suspicious transaction report, it is
less than 10% ... less than 10% of the cases actually turn into suspicious trans-
action reports.”49 According to this chief anti–money laundering officer (working
with a system close to the maximalist version of high-tech equipment), at best
1 alert out of 100 leads to a suspicion-based denunciation from his financial
institution, which still results in thousands of suspicious transaction reports
being sent to FINTRAC per year. The most time-consuming task for any bank
compliance unit is related to the “triage” of alerts – annually from tens of thou-
sands to millions for the largest financial institutions – most of which are gener-
ated by automated surveillance devices. One chief anti–money laundering officer
underscored the challenge facing all bank compliance units in trying to reduce
the rate of “obvious” false alerts as much as possible:

It is still the big issue, there is a lot of false positives, and there is always the bal-
ance between spending money on improving this system and also the resources
 Algorithms as Suspecting Machines 81

available on improving the system, versus going through the false positives and
spending the time to go through the false positives, so that is an issue and it is
always going to be an issue until I can find someone to give me more money
[laughs]. Anyway, false positives will always stay but the number of false posi-
tives, to get that down to a reasonable size so that you are getting alerts on what
you should be getting alerts on, and covering off the activity without just getting
a lot of noise and having to go through that noise, because once the alert is gen-
erated, you have to at least look at it. And every single one takes time.50

Looking at suspecting machines and the management of the problem of false

positives helps us understand how algorithms act as a logistical media of techno-
logical financial surveillance. To paraphrase Evelyn Ruppert, alerts do not simply
pop up but are the result of multiple socio-technical arrangements between
technological and human actors that configure agency and action.51 As with
other contemporary policing practices, the capacity to produce alerts emerges
from the relationships – the connections – between a range of equipment and
social actors, from the designers of surveillance devices to their end-users.52
While “the internal characteristics (technical, logical, and cognitive) of devices
both constrain and enable the action of their users,”53 it is critical to recognize
the variable features of suspicious activity monitoring devices in order to under-
stand how they are appropriated within banks. Specifically, how do banking
actors fine-tune their surveillance and detection practices according to the
specific characteristics of their technological “solutions”? Part of the answer to
this question requires looking at banking actors’ relationships with vendors of
surveillance devices, which range from high dependency to relative autonomy.
It is also where we see power relations at work, revealing that the choices made
are contingent, based on human decisions. Returning to the black box metaphor,
the critical issue is to know who – as well as how and to what extent – is in
favour of preserving the black box rather than opening it.
High dependency refers to cases in which algorithmic devices remain largely
a black box to the banking actors themselves, with few customized elements.
Actors must rely on the vendors for any changes, especially for the creation and
modification of scenarios:

The problem with actually changing the algorithms is that every time you engage
the vendor it costs money, and the organisation is not willing to pay for extras ...
And that is how the technical model [of the vendors] is set up. It is set up so
that once they become your provider, that is when they start charging you the
extras because they know that you cannot go anywhere else. And to implement a
new system is not cost-effective, especially every couple of years when you really
82 Anthony Amicelle and David Grondin

need to upgrade because the technology is changing. Once you have decided on
one provider, you pretty much are going to stay with them unless there is some-
thing really significant that takes place.54

Relative autonomy refers to cases in which the ability to conduct fine-tuning

operations such as scenario authoring is done primarily through in-house
expertise using dedicated human resources (the “data junkies”):

The analytics people, this is where we develop the rules, the thresholds, where we
basically identify alerts that could be cases that could be suspicious transactions.
They actually are the ones who are programming our transactions monitoring
rules, the algorithms. The stats and the math guys who develop the algorithms
for transactions, sanctions, names matching, everything ... My analytics people
are all, not all but most of them, masters of science or PhD analytics, so they
grew either in the computer science or the quantitative sciences world. They are
the data junkies. They love the data, they love to work with the data.55

The “data junkies” are hired to connect with “data-hungry beasts” (i.e., algo-
rithmic devices)56 and refine algorithmic monitoring in accordance with testing
sessions, internal feedback, and new indicators. Although the “white box” or
“open box [transparent] environment” approach is more expensive than black-
box algorithms, its selling point is still about “saving money”: “Can you really
afford a bigger staff to handle growing alert volumes?”57 In either case, the
reduction of false positives often becomes an end in itself, thus obscuring the
original financial surveillance mission for security intelligence.

Conclusion
Ultimately, the automated production of alerts turns out to be more negotiated
than automated, given the competing objectives of making money and fighting
crime and terrorism. The outcome of these negotiations in terms of alerts and
false positives not only is based on the algorithmic design used in transaction
monitoring but depends equally on the quality and verifiability of the data being
processed. The changes that algorithms as logistical media lead to in the inner
workings of financial surveillance are significant.
The algorithmic production of alerts differs from bank tellers’ alerts in many
ways, one of which deserves further attention. In both cases, knowledge about
the client is part of the process of alert production, but there is a difference in
the type of knowledge that can be used. With regard to clients, the form and
content of knowledge may vary from one bank to another, from one type of
clientele or business relationship to another, and from one type of bank employee
 Algorithms as Suspecting Machines 83

to another (i.e., bank tellers in bank branches versus data analysts in headquar-
ters). According to Lyn Loftland and John Dewey, human beings can develop
a knowledge of people (acquaintance-knowledge) or a knowledge about people
(knowledge-about).58 Clive Norris sums up the importance of this distinction
in connection with surveillance and control by noting that the “basis of knowing
of, rather than about, people is face-to-face interaction. When we only know
about people our knowledge is second-hand.”59 De facto, analytics and data
analysts are further removed from the local branch context than the front-line
staff – acting on the basis of face-to-face interaction and face-to-face knowledge
is no longer an option. The resulting surveillance at a distance through data can
be seen either as a critical lack of knowledge or as a key element of objectivity
in controlling mobility.
The first interpretation highlights the inability of analytics “to look at some-
body and understand if they are avoiding eye contact, if they are being really
fidgety, or if they are hurrying you along. Those are indicators that something
is not quite right.”60 The second interpretation highlights the connivance bias
that may be introduced by face-to-face interactions:

That is often the case, especially if, in some cases, the managers have more a
relationship model with the clientele. So if you get into, like, wealth management
businesses where the whole nature of the business is the relationship, then they
may take a more personal view, you know, I know the client, the client is good,
I have known him for ten years, blah-blah-blah, there should not be anything
wrong with what they are doing. So you always got to balance that aspect with
what you see and sometimes stark realities have to come out depending on what
you find.61

More generally, these two interpretations illustrate the tension between

“Ancients” and “Moderns” in policing financial conduct and mobility, with the
process for identifying cases that trigger suspicion and denunciation oscillating
between a focus on human hunches or on technological processing of numbers
and big data. In practice, human hunches and big data devices are often
entwined, in everything from the tuning of suspicious activity monitoring
devices to the finalization of suspicious activity reporting.
The changes in the inner workings of big (financial) dataveillance created by
algorithms as logistical media are significant, particularly in normalizing the
use of financial surveillance as a way to protect banks’ reputations and business
relationships. However, the imbalance created by the resulting number of false-
positive alerts, which must be analyzed by human agents, suggests that we need
to go further down a more sociological path and consider the security practices
 84 Anthony Amicelle and David Grondin

of both technologies and agents. It also reveals the need to reconcile a technical
problem – i.e., the work of suspecting machines – with a business rationale that
seeks to minimize as much as possible the costs of using such machines (as
banks, after all, want to make money). The challenge we faced in this chapter
was making visible the complex process of algorithms as logistical media, in a
context in which they govern data. Doing so also involved understanding how
this process is part of a human/non-human assemblage, a socio-technical
infrastructure in which the algorithm is embedded. Looking at how suspecting
machines operate enables us to see where power relations and preferences play
a role in what may appear to be neutral technical operations, through human
decisions.

Notes
1 Louise Amoore, Stephen Marmura, and Mark B. Salter, “Smart Borders and Mobilities:
Spaces, Zones, Enclosures,” Surveillance and Society 5, 2 (2008): 96–101.
2 This chapter is drawn from a larger project funded by an Insight Social Sciences and
Humanities Research Council (SSHRC) grant on the central role of algorithms in the
governance of North American borderlands (NAB) and the related policing of mobilities
in the digital age of big data. The project deals with the question of how algorithms “act”
and make a difference in the security process, specifically to what extent algorithms have
come to serve as critical operators and “gatekeepers” given the security/mobility nexus
that has come to define how security is processed and delivered in the digital era.
3 Marieke de Goede and Mara Wesseling, “Secrecy and Security in Transatlantic Terror-
ism Finance Tracking,” Journal of European Integration 39, 3 (2017): 253–69; Anthony
Amicelle, “The Great (Data) Bank Robbery: The Terrorist Finance Tracking Program
and the Swift Affair,” CERI Questions de recherche/Research Questions 36 (2011): 1–27.
4 European Commission, Joint Report from the Commission and the U.S. Treasury Department
regarding the Value of TFTP Provided Data (Brussels, 2013), 5.
5 Claudia Aradau, “The Signature of Security: Big Data, Anticipation, Surveillance,” Radi-
cal Philosophy 191 (2015): 1–8.
6 Julie Conroy, Global AML Vendor Evaluation: Managing Rapidly Escalating Risk (Boston:
Aite, 2015), https://www.aitegroup.com/report/global-aml-vendor-evaluation-managing
-rapidly-escalating-risk.
7 Jef Huysmans, Security Unbound: Enacting Democratic Limits (Abingdon, UK: Routledge,
2014).
8 FINTRAC, “What Is FINTRAC?” 16 August 2019, http://www.fintrac-canafe.gc.ca/
questions/FAQ/1-eng.asp.
9 FINTRAC, FINTRAC Annual Report 2018–19 (Ottawa: FINTRAC, 2019), 37, https://
www.fintrac-canafe.gc.ca/publications/ar/2019/ar2019-eng.pdf.
10 Roger Clarke, “Information Technology and Dataveillance,” Communications of the
ACM 31, 5 (1998): 498–512.
11 Ned Rossiter, Software, Infrastructure, Labor: A Media Theory of Logistical Nightmares
(New York: Routledge, 2016), 4–5.
12 Joseph Masco,  The Theater of Operations: National Security Affect from the Cold War
to the War on Terror (Durham, NC: Duke University Press, 2014); Brian Larkin, “The
Politics and Poetics of Infrastructure,” Annual Review of Anthropology 42, 1 (2013):
 Algorithms as Suspecting Machines 85

327–43; Gabrielle Hecht, “Introduction,” in Entangled Geographies: Empire and Techno­

politics in the Global Cold War, edited by Gabrielle Hecht (Cambridge, MA: MIT Press,
2011), 1–12; Claire Waterton,  “Experimenting with the Archive: STS-ers as Analysts
and Co-constructors of Databases and Other Archival Forms,” Science, Technology, and
Human  Values  35 (2010): 645–76; Susan Leigh Star, “The Ethnography of Infrastruc-
ture,” American Behavioral Scientist 43, 3 (1999): 377–91.
13 Keller Easterling,  Extrastatecraft: The Power of Infrastructure Space (London: Verso,
2014), 14.
14 David Grondin., “Mobilité, vie algorithmique et société de surveillance dans Person of
Interest: La traque du national security state cyberspatial,” in Représentations politiques,
luttes de pouvoir et science-fiction, edited by Isabelle Lacroix and Karine Prémont (Quebec
City: Presses de l’Université du Québec, 2016), 165–202; Frank Pasquale, The Black Box
Society: The Secret Algorithms That Control Money and Information (Cambridge, MA: Har-
vard University Press, 2015); David Lyon, “Surveillance, Snowden, and Big Data: Capaci-
ties, Consequences, Critique,” Big Data and Society 1 (2014): 1–13; Antoinette Rouvroy and
Thomas Berns, “Gouvernementalité algorithmique et perspectives d’émancipation: Le dis-
parate comme condition d’individuation par la relation?” Réseaux 177, 1 (2013): 163–96.
15 Lauren Berlant, “The Commons: Infrastructures for Troubling Times,” Environment and
Planning D: Society and Space 34, 3 (2016): 39.
16 Star, “The Ethnography of Infrastructure,” 388.
17 John Durham Peters, The Marvelous Clouds: Toward a Philosophy of Elemental Media
(Chicago: University of Chicago Press, 2015), 37.
18 Rob Kitchin, “Thinking Critically about and Researching Algorithms,” Information,
Communication and Society 20, 1 (2016): 18.
19 Louise Amoore and Rita Raley, “Securing with Algorithms: Knowledge, Decision, Sov-
ereignty,” Security Dialogue 48, 1 (2016): 1.
20 Dominique Cardon, “Deconstructing the Algorithm: Four Types of Digital Information
Calculations,” in Algorithmic Cultures: Essays on Meaning, Performance and New Technolo-
gies, edited by Robert Seyfert and Jonathan Roberge (New York: Routledge, 2016), 95–110.
21 Seyfert and Roberge, Algorithmic Cultures.
22 Pasquale, The Black Box Society.
23 Huysmans, Security Unbound, 91.
24 NICE Actimize, “Transforming Transaction Monitoring and Reporting of Suspicious
Activity,” https://www.niceactimize.com/anti-money-laundering/suspicious-activity
-monitoring.
25 FINTRAC, FINTRAC Annual Report 2017–18, 2.
26 Jim Bronskill, “Canadian Bank Fined $1.1M for Failing to Report Suspicious Transaction,
Money Transfers,” The Star, 5 April 2016, https://www.thestar.com/business/2016/04/05/
canadian-bank-fined-11m-for-failing-to-report-suspicious-dealings.html.
27 FINTRAC, “Guideline 2: Suspicious Transactions,” June 2017, http://fintrac-canafe.
gc.ca/guidance-directives/transaction-operation/Guide2/2-eng.asp.
28 Ibid.
29 Interview with a FINTRAC official, Canada, 2015.
30 Conroy, Global AML Vendor Evaluation.
31 Patrice Flichy, “Rendre visible l’information,” Réseaux 178–79 (2013): 55–89.
32 Tarleton Gillespie, “The Relevance of Algorithms,” in Media Technologies: Essays on Com-
munication, Materiality, and Society, edited by Tarleton Gillespie, Pablo Boczkowski, and
Kirsten Foot (Cambridge, MA: MIT Press, 2014), 167–94.
33 Anthony Amicelle, “Towards a ‘New’ Political Anatomy of Financial Surveillance,”
Security Dialogue 42, 2 (2011): 161–78; Kirstie Ball, E. Daniel, S. Dibb, A. Canhoto, M.
 86 Anthony Amicelle and David Grondin

Meadows, and Keith Spiller, The Private Security State? Surveillance, Consumer Data
and the War on Terror (Frederiksberg: Copenhagen Business School Press, 2015); Gilles
Favarel-Garrigues, Thierry Godefroy, and Pierre Lascoumes, “Reluctant Partners? Banks
in the Fight against Money Laundering and Terrorism Financing in France,” Security
Dialogue 42, 2 (2011): 179–96; Lyliya Gelemerova, “On the Frontline against Money-
Laundering: The Regulatory Minefield,” Crime, Law and Social Change 52 (2009): 33–55;
Eric Helleiner, “State Power and the Regulation of Illicit Activity in Global Finance,” in
The Illicit Global Economy and State Power, edited by Peter Andreas and Richard Friman
(Lanham, MD: Rowman and Littlefield, 1999), 53–89; Béatrice Hibou, The Bureaucrati-
zation of the World in the Neoliberal Era (London: Palgrave Macmillan, 2015).
34 Interview with a compliance officer, Canada, 2016.
35 Richard Ericson, Crime in an Insecure World (London: Polity, 2007).
36 Interview with a FINTRAC official, Canada, 2015.
37 Interview with a former compliance officer, Canada, 2015.
38 FINTRAC, “Financial Transactions That Must Be Reported,” 2020, https://www.fintrac
-canafe.gc.ca/reporting-declaration/rpt-eng.
39 FINTRAC, FINTRAC Annual Report 2018–19.
40 Anthony Amicelle and Elida Jacobsen, “The Cross-Colonization of Finance and Security
through Lists: Banking Policing in the UK and India,” Environment and Planning D:
Society and Space 34, 1 (2016): 89–106.
41 Anthony Amicelle and Vanessa  Iafolla, “Suspicion-in-the-Making: Surveillance and
Denunciation in Financial Policing,” British Journal of Criminology 58, 4 (2018): 845–63.
42 Interview with a bank compliance officer, Canada, 2015.
43 Interview with a FINTRAC official, Canada, 2015.
44 FINTRAC, “Financial Transactions That Must Be Reported,” item 6.1.
45 Amicelle and Iafolla, “Suspicion-in-the-Making.”
46 Ibid.
47 FINTRAC, “Guideline 2: Suspicious Transactions,” item 6.3.
48 Interview with a compliance officer, Canada, 2015.
49 Ibid.
50 Interview with a bank compliance officer, Canada, 2015.
51 Evelyn Ruppert, “The Governmental Topologies of Database Devices,” Theory, Culture
and Society 29, 4–5 (2012): 116–36.
52 Anthony Amicelle, Claudia Aradau, and Julien Jeandesboz, “Questioning Security
Devices: Performativity, Resistance, Politics,” Security Dialogue 46, 4 (2015): 293–306.
53 Ibid., 294.
54 Interview with a bank compliance officer, Canada, 2015.
55 Interview with a bank compliance officer, Canada, 2015.
56 Conroy, Global AML Vendor Evaluation.
57 SAS, “SAS Anti-Money Laundering” (fact sheet, 2016), 2, https://www.sas.com/content/
dam/SAS/en_us/doc/factsheet/sas-anti-money-laundering-105623.pdf; BAE Systems, “Net-
Reveal AML Transaction Monitoring,” https://www.baesystems.com/en/cybersecurity/
product/aml-transaction-monitoring.
58 Lyn H. Lofland, A World of Strangers: Order and Action in Urban Public Space (New York:
Basic Books, 1973); John Dewey, Logic: The Theory of Inquiry (New York: Holt, 1938).
59 Clive Norris, “From Personal to Digital: CCTV, the Panopticon, and the Technological
Mediation of Suspicion and Social Control,” in Surveillance as Social Sorting: Privacy, Risk,
and Digital Discrimination, edited by David Lyon (New York: Routledge, 2003), 251.
60 Interview with a bank compliance officer, Canada, 2015.
61 Ibid.
Part 2
Big Data Surveillance and Signals Intelligence
in Canadian Security Organizations
This page intentionally left blank
5
From 1967 to 2017
The Communications Security Establishment’s Transition
from the Industrial Age to the Information Age
Bill Robinson

Concern about the possibility of domestic surveillance has long surrounded

the Communications Security Establishment (CSE), Canada’s signals intelligence
(SIGINT) agency. The validity of this concern has been difficult to assess, how-
ever, as information about the highly secretive agency is hard to find. This
chapter pulls together scattered, little-known, and recently released information
on the history of CSE to provide for the first time a detailed – although still
incomplete – overview of the evolution of the agency and its operations over
the last half-century. At the beginning of the period covered by this chapter, the
Communications Branch of the National Research Council (CBNRC), as CSE
was then called, paid almost no attention to Canadians. In the years since 1967,
however, changes in intelligence priorities, budget resources, legal authorities,
and, most important, communications and computing technologies have led
to a growing overlap between the lives of Canadians and the operations of CSE.
Today, CSE remains primarily a foreign intelligence agency, but its operations
take place in a context where the communications of Canadians and non-
Canadians are thoroughly intermixed, targets and non-targets are often difficult
to distinguish, counterterrorism concerns make communications involving
individuals in Canada some of the agency’s highest collection priorities, and
CSE’s analysts increasingly look to a wide range of big data pertaining to both
Canadians and non-Canadians to help answer the agency’s intelligence ques-
tions. These changes have fundamentally altered the relationship between the
agency and Canadians.

CBNRC at the End of the 1960s

In the summer of 1967, as Canadians celebrated the centennial of Confederation,
the Soviet Union conducted the largest strategic nuclear forces exercise ever
detected by the West,1 the Six-Day War threw the Middle East into turmoil, and
French president Charles de Gaulle, visiting Expo 67, delivered his notorious
Vive le Québec libre speech and was promptly invited to go home. CBNRC had
spent most of its twenty-one years of existence focused primarily on monitoring
the Soviet Union. For the last ten of those years, its collection and processing
activities had been directed principally at the Soviet Arctic, providing
90 Bill Robinson

intelligence to the National Security Agency (NSA) and Government

Communications Headquarters (GCHQ), CBNRC’s American and British
partners, as well as to Canadian intelligence customers. In addition to its own
reporting, the agency served as the funnel through which the Canadian govern-
ment received the voluminous SIGINT reporting produced by those partners.
SIGINT on the Middle East crisis flooded the agency’s communications centre
that summer.2 Domestic Canadian communications, by contrast, were almost
entirely ignored. The agency did monitor the cross-border communications of
foreign delegations in Ottawa and the clandestine radio communications of
foreign intelligence agents, but otherwise paid little attention to communica-
tions originating or terminating in Canada.3
About 80 percent of CBNRC’s 600 employees were engaged in SIGINT-related
activities, while the remaining 20 percent worked on the Communications
Security (COMSEC) side of the agency.4 Although a member of the five-country
UKUSA (UK-USA agreement) SIGINT partnership (now commonly called the
Five Eyes), CBNRC worked almost exclusively with NSA and GCHQ, with only
occasional contact with its Australian and New Zealand partners. Canadian
documents from the period refer mostly to “Tripartite” or “CANUKUS” SIGINT
arrangements.
Soviet high-level encryption systems had been almost entirely resistant to
cryptanalysis since the late 1940s,5 and thus Canadian SIGINT activities focused
on low-echelon military and civilian communications, which could still provide
important information when collected in bulk and analyzed in conjunction
with the “external” data (originator, destination, date, time, etc.) that accom-
panied both encrypted and unencrypted communications. (Today we call these
externals “metadata.”) Monitoring the movements of Soviet aircraft, for example,
could provide vital information about the status of Soviet nuclear forces, includ-
ing changes in their level of alert and preparations for possible operations.
By the 1960s, however, many Soviet communications were migrating away
from long-range high-frequency (HF) radio circuits to shorter-range VHF and
UHF circuits, or in the case of fixed installations, to highly directional microwave
and troposcatter systems, satellite links, and domestic cable systems, none of
which were accessible from Canada. CBNRC was facing the prospect of going
dark.6 Between 1971 and 1975, the number of Canadian SIGINT reports issued
per year fell from 6,813 to 4,591. The small scale of the Canadian contribution
is illustrated by the fact that in 1972 the five UKUSA partners produced in total
approximately 167,000 reports.7 The drop in the already small Canadian con-
tribution did not go unnoticed.
Other kinds of change were also in the air as the 1960s came to an end. The
capacity of transoceanic telecommunications cables was expanding rapidly,
 From 1967 to 2017 91

with an increasing number of voice circuits being carried.8 Communications

satellites had begun to populate the skies. In 1965, INTELSAT’s Early Bird, the
first commercial transatlantic communications satellite, was placed into orbit.
Just four years later, in 1969, an even more momentous event occurred: the first
packet-switched message transited the Advanced Research Projects Agency
Network (ARPANET), marking the birth of what would eventually become the
Internet.
Canadian intelligence priorities were under examination as well. Pierre Tru-
deau, who succeeded Lester Pearson as prime minister in April 1968, came into
office determined to increase Canada’s independence from the United States.
Greater production of political and economic intelligence might have a part to
play in that program. The Trudeau government also faced growing domestic
tensions. The Front de libération du Québec (FLQ) was becoming increasingly
active, and student radicalism was spreading on the country’s university cam-
puses. In October 1968, the Royal Commission on Security (the Mackenzie
Commission) reported that “the most important communist activities in Canada
are largely directed from abroad, although domestic adherents of and sympa-
thizers with communism pose considerable problems in themselves; the sep-
aratist movement is primarily of domestic origin, although there has been some
indication of foreign involvement.”9 Resolving the question of foreign influence
was also an area where the Canadian intelligence program might have a con-
tribution to make.

The Ears Turn Inward

In November 1969, Clerk of the Privy Council Gordon Robertson suggested to
the Intelligence Policy Committee that “there was a need to formulate new
Objectives and Priorities for the Intelligence Program, since the present program
did not seem to be meeting governmental requirements adequately.” A senior
public servant, Claude Isbister, was subsequently appointed to act as an “outside
examiner” of the Canadian intelligence program.10 Isbister’s report, delivered
in November 1970, endorsed the main elements of the existing program, but
also backed Robertson’s concerns, recommending that “an effort be made ... to
give more emphasis to areas of direct interest to Canada.”11 Submitted in the
immediate wake of the October Crisis, the report reinforced the government’s
interest in better utilizing the foreign intelligence program to address domestic
security questions.
Cross-border cable traffic was one potential source of intelligence. NSA and
GCHQ had long monitored the communications carried by commercial telecom-
munications companies (International Licensed Carriers, or ILCs) on the world’s
transoceanic cable systems, most of which made landfall on the territory of one
92 Bill Robinson

or both of the partners,12 and some of that material was already being made avail-
able to Canada. But the question remained whether more specific Canadian
intelligence needs might be served by direct Canadian intelligence collection.
By April 1972, a survey was underway to “determine the value of arrangements
with the Canadian Overseas Telecommunications Corporation (COTC) by
which we would have the opportunity to examine messages between Canada
and locations abroad which would (1) clarify links between revolutionary
activities abroad, (2) provide information concerning known revolutionary
elements, and (3) contribute intelligence about foreign and economic affairs
of direct interest to Canada.”13 The location of the test operation was the
COTC’s Montreal office,14 which served as the gateway for all Canadian trans-
atlantic telecommunications.
Later in 1972, CBNRC sought the budget authority to place the COTC oper-
ation on a full operational footing beginning in fiscal year 1973–74.15 The proposal
was put on hold, however, when cabinet froze the entire Canadian intelligence
program at its existing level pending completion of a review of the program.16
Not long afterwards, passage of the Protection of Privacy Act made the intercep-
tion of “private communications” (communications with one or more ends in
Canada) illegal except under specific defined circumstances, quashing the
CBNRC’s cable-monitoring plan in the process. Nobody had asked the agency
what effect the new act might have on SIGINT collection:

The Department of External Affairs queried some aspects of the draft bill which
might affect legitimate intelligence interests adversely. Copies of the queries, which
went out over the signature of the Under-Secretary of State for External Affairs
(USSEA), went to the Cabinet Secretary, the Chief of the Defence Staff (CDS)
and the Director General of the Security Service (DGSS) in the RCMP, but not to
Director CB. After receiving answers from Justice, the USSEA told his Minister in
June [1971] that “radio communications would not come within the prohibition”
against the interception of “private communications.” Also, the “use or disclosure”
of communications intercepted outside Canada would not constitute an offence,
because such an act would not be “wilful,” as being based on selection from a mass
of material picked up without any “mens rea.” He also told his Minister that rel-
evant information would be made available to intelligence authorities (presumably
including CB), even if obtained for specifically security purposes under a warrant
issued by the Solicitor General. However, these did not all turn out to be the inter-
pretations of several subsequent Solicitors General and Ministers of Justice.17

Among its other effects, the Protection of Privacy Act denied CBNRC access to
telephone, telegraph, and telex messages transmitted by cable if one or both
 From 1967 to 2017 93

ends were in Canada, as these were by definition private communications.

CBNRC was prohibited both from collecting the messages itself and from asking
its partners to collect the traffic. “The effect on RCMP interests such as terror-
ism, drug trafficking and illegal immigration could be judged by the angry
reaction of the DGSS to the cutting off of the UK and US intercepted cable
traffic.”18

CBNRC Becomes CSE

On 1 April 1975, CBNRC was transferred to the Department of National Defence
and renamed the Communications Security Establishment. Cross-border tele-
communications traffic continued to grow as cable and satellite connections
improved, but CSE’s exploitation of ILC traffic remained stalled by the effects
of the Protection of Privacy Act. During 1977 and 1978, CSE and the RCMP
pushed within the Intelligence Advisory Committee (IAC) for approval of a
special collection project apparently focused on ILC traffic. But the fraught
politics of national unity following the election in November 1976 of the first
Parti Québécois government, compounded by revelations of RCMP wrongdoing
in Quebec, made the timing poor for a new intelligence operation centred on
that province. Commenting on the project in March 1977, IAC chair John
Hadwen wrote:

I doubt if we can get everything the RCMP and CSE want, but we can certainly
get something if we come up with a practical proposal. My own assessment is
that we could get permission at least to tap the US and UK data banks and lately
I have been wondering whether we could not take some action in Toronto and
Vancouver which would be less difficult to organize and explain than action in
Montreal.19

(Toronto and Vancouver were the locations of the COTC’s other gateways for
cross-border telecommunications.) Within a few months, however, both the
Keable Inquiry and the McDonald Commission had been launched to investigate
the RCMP’s misdeeds.20 In January 1978, the IAC “generally agreed the present
atmosphere associated with the McDonald Commission and the Keable Inquiry
is not conducive at this time for government approval of the [special collection]
project.”21 Nevertheless, it remained on the wish list. Just one month later, in its
Review of 1977 and Forecast for 1978, the IAC put the Interdepartmental Com-
mittee on Security and Intelligence on notice that a proposal would likely be
forthcoming: “At present very little intelligence is available which would lend
itself to the production of analyses on foreign attitudes, intentions and activities
relating to Canadian unity. A special collection project may be required to
94 Bill Robinson

improve prospects in this regard, and it is anticipated that a specific proposal

concerning the project will be submitted later.”22
Renewed access to US and UK ILC collection was probably secured over the
next few years,23 but the politics of establishing a Canadian cross-border col-
lection ability remained unfavourable throughout the rest of the 1970s and
early 1980s. The May 1980 referendum on sovereignty-association and the
reports of the Keable Inquiry (1981) and the McDonald Commission (one in
1979 and two in 1981) ensured that questions of intelligence gathering in Quebec
remained politically delicate, and the effort to draft and then pass legislation
to replace the RCMP Security Service with the Canadian Security Intelligence
Service (CSIS) took priority during the remaining years of the Trudeau
government.
One section of the resulting CSIS Act did, however, pertain to the collection
of foreign intelligence within Canada. Section 16 was probably intended primar-
ily to provide legal authority for foreign-intelligence monitoring of the com-
munications of foreign diplomats and diplomatic premises in Canada, but it
may also have been intended to enable the collection of economic and other
intelligence in communications crossing Canada’s borders. The volume of
“transit traffic” – ILC traffic that passed through Canada during transmission
between foreign originators and foreign recipients – was growing rapidly during
this time. Transiting telephone traffic, for example, more than doubled between
1982 and 1986, totalling nearly 150,000 hours in 1986.24 In August 1987, the
minister of national defence, the secretary of state for external affairs, and the
solicitor general signed a memorandum of understanding governing the use of
section 16. However, few section 16 operations were conducted prior to 1990.
The Security Intelligence Review Committee commented in its 2001–02 Annual
Report:

Since 1990, collection activities under section 16 have gradually increased. The
Committee believes several factors are behind this trend. First, the notion of col-
lecting foreign intelligence in the early years of the Act was novel and untested. It
was only after the signing of the Tri-Ministerial MOU that the details of exactly
how to proceed were established. Second, there has been a growing awareness
within government of the utility of the kind of information that tasking under
section 16 can generate.25

Thus, Canadian foreign-intelligence exploitation of cross-border ILC traffic,

first explored at the beginning of the 1970s, is likely to have begun, if at all, only
in the 1990s. By that time, the Berlin Wall had fallen, and CSE was already a
much different agency than it had been at the beginning of the 1970s.
 From 1967 to 2017 95

Collection Goes Global

Despite fears of going dark, the continuing effects of budget and resource pres-
sures, and a series of embarrassing public disclosures, the 1970s were good to
the UKUSA SIGINT agencies. The CANYON and RHYOLITE families of
geosynchronous SIGINT satellites, first orbited in 1968 and 1970, respectively,
gave NSA better access to Soviet communications, including internal microwave
and troposcatter networks, than the agency had ever had before. In 1971, the
CIA, which controlled access to the data collected by the RHYOLITE satellites,
agreed to bring Canada into the program; by 1974, 109 “slots” had been allotted
to CSE voice transcribers and analysts.26 Additional slots were provided to
analysts in the Department of National Defence’s Directorate of Scientific and
Technical Intelligence to process Soviet missile telemetry intercepted by the
satellites.27 NSA had been an especially strong proponent of bringing Canada
into the program. In a meeting of the US Intelligence Board in November 1970,
NSA’s long-serving deputy director, Lou Tordella,

discussed the contribution that could be made by Canada, under the CAN­
UKUS agreement, in support of NSA processing of “R” take. He covered the fol-
lowing points: a. Canadian responsibilities under the CANUKUS agreement are
to process all raw voice intercepts in the Soviet Arctic and the northern military
districts. In order to accomplish this, the Canadians use linguists, transcribers
and analysts. At the present time, with the switch of the Soviets away from HF,
this Canadian capacity is not being used. b. The “R” system is now intercept-
ing the troposcatter system in the Soviet Arctic. NSA, therefore, proposes that
Canada be authorized to process this in order to take up the slack resulting from
the HF dry-up.28

CSE probably also helped to process traffic from the CANYON satellites, which
were controlled by NSA.29
Meanwhile, the advent of the supercomputer, heralded by the arrival of the
Cray-1 in 1976, revolutionized UKUSA cryptanalysis. Within a year, NSA began
breaking into high-echelon Soviet encryption systems, obtaining a degree of
access not seen since the end of the Second World War. NSA historian Thomas
R. Johnson wrote that “for the cryptologists, it was their finest hour since 1945.”30
Access to the encryption systems of other countries was even more complete:
a 1979 assessment of NSA’s G Group, which was responsible for non-Soviet
targets, concluded that its cryptanalytic achievements were “at an all-time peak.”31
CSE and the other partners in the UKUSA community were the beneficiaries
of NSA’s successes. But the shrinking importance of their own contributions
also placed the partners under pressure to justify their privileged access to the
96 Bill Robinson

output of the US intelligence system. US pressure probably contributed signifi-

cantly to the decisions made at the end of the 1970s and beginning of the 1980s
to revitalize and expand Canadian SIGINT capabilities. Other factors that may
have played a role include the deteriorating state of East-West détente, especially
in the wake of the Soviet intervention in Afghanistan; significant increases in
the budget of the Department of National Defence, which opened space for
comparable increases in the CSE budget; and the government’s continuing
interest in expanding the range of Canadian foreign intelligence targets, includ-
ing its desire to collect more economic-related intelligence. Whatever the relative
weights these and other factors may have had, it is clear that major new initia-
tives began to be approved around 1980.
These initiatives included a series of increases in CSE’s authorized establish-
ment that enabled the agency to grow from approximately 600 employees at
the end of the 1970s to 900 at the beginning of the 1990s; approval in 1980 to
conduct site surveys to determine the best locations for embassy-based intercept
sites (Project PILGRIM);32 and the first steps, also taken in 1980, towards the
revitalization of CSE’s cryptanalytic capabilities, which led to acquisition in 1985
of a Cray X-MP/11, Canada’s first cryptanalytic supercomputer.33 The crypt-
analysis initiative seems to have been part of a UKUSA-wide effort, as the
Australian SIGINT agency obtained its first supercomputer, also a Cray, in
1986,34 and New Zealand’s agency established its first cryptanalysis unit around
the same time.35
Another collection initiative involved satellite communications. Around 1985,
CSE began installing satellite monitoring dishes at Canadian Forces Station
Leitrim, just south of Ottawa. Four dishes, all covered by radomes, were in place
by 1990.36 Some of the dishes seem to have been targeted on the Atlantic INTEL-
SAT satellites as part of the UKUSA agencies’ ECHELON program. The specific
users targeted may have been located in Latin America or the Caribbean. Other
satellites, such as the Brazilian and Mexican domestic communications satellites
and the INMARSAT maritime satellites, may also have been monitored at vari-
ous times.
In December 1987, External Affairs reported that legal concerns threatened
to block CSE’s participation in the program:

ECHELON is a CSE project which was designed to collect Intelsat communica-

tions but which has been stopped by a legal opinion from Justice which states
that such collection activity would be, if not illegal, at least “imprudent.” We
understand that [CSE Chief Peter] Hunt wishes to brief the [Security and Intel-
ligence Consultative Meeting] on the project and the problems it encountered
with Justice. Basically Mr. Hunt will emphasize that there appears to be no real
 From 1967 to 2017 97

awareness in the legal system of the needs of the S&I community and there may
be implications down the road for other projects (e.g., PILGRIM, MADRIGAL).
Our position on ECHELON has been to support the project as a valuable con-
tribution to the overall Canadian and allied effort. We regret that it appears it
will not go forward.37

The Department of Justice’s concerns, probably related to the possibility that

Canadian private communications would inadvertently be intercepted, were
evidently resolved not too long afterwards, as a June 1988 memo noted that the
project was “going forward” after all.38 Satellite monitoring activities continue
at the Leitrim station today.
In October 1987, CSE’s workforce was augmented by the creation of 771 Com-
munications Research Squadron,39 a military unit whose members were inte-
grated into CSE’s SIGINT directorate, boosting the combined workforce at CSE
to as many as 1,000 by the end of the decade, about 800 of whom worked on
the SIGINT side of the house. The increase in collection and processing resources
during the 1980s led to a noticeable increase in SIGINT output. By 1990, Canada
was producing about 10,000 end product reports per year, more than double
its 1975 figure. By comparison, the total number of reports available to Canada
in 1990 was about 200,000, or about 20 percent more than the number produced
in the early 1970s.40

New Targets and New Technology

CSE had just about completed its decade-long buildup when its primary target,
the Soviet Union, disintegrated. The end of the Cold War did not eliminate all
interest in events in Russia and other former Soviet republics, but it did wipe
out much of the operational military traffic that CSE had formerly monitored,
and it reduced the apparent importance of the traffic that remained, both to
Canada and to its UKUSA partners. Canadian foreign intelligence priorities
were revisited from the ground up. Political and economic intelligence concern-
ing other parts of the world had already begun to be collected through the
PILGRIM and ECHELON programs, and these subjects quickly grew in import-
ance. The 1990s also saw the growth of CSIS Act section 16 collection. Former
CSE analyst Jane Shorten revealed in 1995 that section 16 targets included the
Japanese, Mexican, and South Korean embassies in Canada,41 but as noted above,
such operations may also have targeted foreign transit traffic. Economic intel-
ligence was a major goal of these activities. CSE chief John Adams confirmed
in 2007 that “in the time between the end of the cold war and 2001, CSE’s
reporting concentrated mostly on prosperity issues.”42 Monitoring the activities
of the fishing fleets off Canada’s East Coast also remained important,43 and
98 Bill Robinson

international narcotics smuggling became a significant target. In 1996, Canada

formally joined the US SANDKEY program, which provided SIGINT support
to the US War on Drugs in the Caribbean and Latin America.44 Support to
peacekeeping and other military operations also featured in the priorities list.
Counterterrorism and other security issues were on the list too, but accounted
for only 3 percent of reporting at the end of the 1990s.45
The 1990s also saw the arrival of the Internet as a significant phenomenon.
Initially, the volume of email, file-sharing, and web traffic was tiny in comparison
to traditional teletype, fax, and, especially, telephone services. The UKUSA
partners began hacking into network-connected computers to acquire informa-
tion (computer network exploitation, or CNE), but collection and processing
of the still-growing volume of voice traffic remained the main challenge facing
the UKUSA community in the mid-1990s. CSE and its partners focused a lot
of attention on developing better computer technology for speaker recognition,
language recognition, and speech recognition for keyword and topic
spotting.46
By one set of estimates, voice traffic represented 98 percent of the volume
of global telecommunications in 1993, while Internet traffic represented just
1.3 percent. By the year 2000, however, the figures were 44 percent and 51
percent, respectively, and by 2007 voice traffic accounted for just 2 percent
of global telecommunications while the Internet accounted for 97 percent.
This reversal was not the result of a drop in voice traffic, which more than
tripled during the intervening fourteen years; rather, it was a consequence
of the explosive growth in Internet traffic over that period.47 Towards the end
of the 1990s, it had become clear that the Internet was the future of telecom-
munications. The NSA’s Transition 2001 report, published in December 2000,
declared the agency’s determination to “master and operate in the global net
of tomorrow”:48

SIGINT in the Industrial Age meant collecting signals, often high frequency
(HF) signals connecting two discrete and known target points, processing the
often clear text data and writing a report. eSIGINT in the Information Age
means seeking out information on the Global Net, using all available access
techniques, breaking often strong encryption, again using all available means,
defending our nation’s own use of the Global net [sic], and assisting our war­
fighters in preparing the battlefield for the cyberwars of the future. The Fourth
Amendment is as applicable to eSIGINT as it is to the SIGINT of yesterday and
today. The Information Age will however cause us to rethink and reapply the
procedures, policies and authorities born in an earlier electronic surveillance
environment ... senior leadership must understand that today’s and tomorrow’s
 From 1967 to 2017 99

mission will demand a powerful, permanent presence on a global telecommuni-

cations network that will host the “protected” communications of Americans as
well as the targeted communications of adversaries.49

The term “eSIGINT” fortunately never caught on, but the transition to Infor-
mation Age SIGINT certainly did. In 2000, CSE “embarked upon an important
strategic exercise to identify alternative approaches to delivering its mandate.
As a starting point, it defined its vision thus: ‘to be the agency that masters the
global information network to enhance Canada’s safety and prosperity.’”50 But
its efforts remained hamstrung by a lack of suitable legal authorities. Internet
traffic could be monitored for foreign intelligence purposes, but CSE had to
ensure that no private communications were intercepted by its systems. The
difficulty of determining whether a particular Internet user was in Canada at
the moment of communication made this an extraordinarily challenging task.
The occasional inadvertent intercept might be forgiven, but any lapse in dili-
gence would open the agency to the charge of violating the ban on wilful
interception.51 Moreover, if a private communication did end up inadvertently
intercepted, the information in it could be neither used nor disclosed, even if
it concerned the proverbial ticking time bomb. A new watchdog position, the
CSE Commissioner, was created in 1996 to keep CSE’s compliance with the
law under continuing review. Successive commissioners demonstrated little
inclination to declare CSE in non-compliance,52 but their activities certainly
forced the agency to tighten up its privacy-related practices. Behind-the-scenes
work was begun to draft an Information Age legal regime for CSE, but no bill
was put before Parliament.
In the meantime, budget resources had become tight. Despite having lost its
primary target at the end of the Cold War, CSE avoided the major program cuts
that swept through Ottawa in the mid-1990s, but it did suffer minor cuts. “Pro-
gram integrity” top-ups for fiscal years 2000–01 and 2001–02 enabled the agency
to grow to nearly 950 personnel by mid-2001.53 However, this increase probably
served only as partial compensation for the loss of 771 Communications
Research Squadron, which was disbanded shortly afterwards.54
According to CSE chief Keith Coulter, by the end of the 1990s the agency was
facing a serious erosion of its SIGINT capabilities:

[When] the events of 9/11 took place, CSE was ... facing a tough scenario. Simply
put, in a kind of perfect storm situation, the 1990s saw the global revolution in
communications technologies, resource shortages and the lack of an updated
authority framework combine to create a serious erosion of CSE’s SIGINT
capabilities.55
100 Bill Robinson

9/11 Changes Everything

All of that changed following the terrorist attacks of 11 September 2001. Within
months, CSE’s technological capabilities, staff and budget resources, legal
authorities, and target priorities were all transformed. The government of Jean
Chrétien took the broader CSE powers that had been under discussion within
the bureaucracy and bundled them into an omnibus Anti-terrorism Act (Bill
C-36) that was quickly pushed through Parliament, receiving royal assent just
nine weeks after first reading.56 Bill C-36 amended the National Defence Act to
give CSE a statutory mandate for the first time and create a ministerial author-
ization regime that enabled the agency to intercept private communications
during its foreign intelligence and IT security activities. This made it legal for
CSE to intercept, use, and disclose the communications of foreign targets even
when the other end of the communication was located in Canada. However,
the agency remained prohibited from directing its activities at persons in
Canada or Canadians anywhere, except when providing support to a federal
law enforcement or security agency, when it would be subject to that agency’s
legal authorities, including the requirement for judicial warrants to intercept
private communications. Passage of the bill also gave CSE the power to engage
in computer network exploitation activities; the ministerial directive issued
on 14 January 2002 was probably the order that formally established the agency’s
CNE program.
The doors to the treasury were thrown open at the same time. In October,
CSE’s $100 million fiscal year 2001–02 budget was boosted by an immediate
injection of $37 million, $31 million of which was for computer purchases and
other technology upgrades.57 The final amount CSE spent that year was even
higher, totaling $189 million. The budget for the following year, fiscal year
2002–03, increased CSE’s baseline funding to $175 million and authorized the
agency to boost its workforce by 35 percent, to about 1,300 employees.
CSE’s intelligence priorities were also transformed. Overnight, counterterrorism
became its top priority. In October 2001, the agency’s Office of Counter Terrorism
was created “to centralize foreign signals intelligence efforts relating to inter-
national terrorism threats.”58 Counter-proliferation, defence against cyber threats,
and support to military operations also moved up on the priorities list. In 2003,
CSE informed the NSA that it planned to “dedicate 40 percent of its SIGINT re-
sources to security initiatives,”59 and in 2005 Coulter testified that security issues,
broadly defined, accounted for over 80 percent of the agency’s activities.60 By
2006, with Canadian combat operations heating up in Afghanistan, over one-
quarter of CSE security reporting was said to be related to that country.61
One consequence of CSE’s reshuffled intelligence priorities and new legal
authorities was that Canadians began to show up a lot more in CSE reporting.
 From 1967 to 2017 101

Canadian identity information is normally “suppressed” (replaced by a generic

reference such as “a named Canadian”) in end product reports derived from
CSE’s foreign intelligence activities, but those identities can subsequently be
released to customers who affirm a need for the information and can show that
it is relevant to the customer’s mandate and operational program. An indication
of the degree to which communications concerning Canadians had come to
figure in CSE’s operations can be found in the fact that during the first six months
of 2005, 300 to 400 Canadian identities, or about two a day, were released to
CSE customers.62
The aftermath of 9/11 also transformed information-sharing arrangements
within the Five Eyes community: “Canada had always benefitted from sharing
arrangements with the US, but after 9/11, Canada was told that ‘the taps were
being opened.’ The increased flow of foreign intelligence reporting and the
volume of US assessments was so immense that a special capacity had to be
created to handle it.”63 The principle of “need to know,” which had dominated
access to SIGINT during the Cold War, was augmented by “need to share.” With
minor exceptions for especially sensitive topics or accesses, all SIGINT report-
ing became shared by default among the Five Eyes partners. A push was made
to share raw traffic, including metadata, much more extensively as well.64

Data Gets Big

In the years that followed, CSE and its partners struggled to take advantage of
the growing torrent of data flooding through the world’s communications
networks. In March 2004, CSE received a 25 percent increase in its baseline
budget and personnel establishment, enabling it to expand its workforce to 1,650
employees. A new ministerial directive, signed in the same month as the budget
increase, probably provided new guidance on Internet collection activities. It
also provided the agency with its first ministerial guidance on the use of com-
munications metadata.65 It may have marked the beginning of CSE’s EONBLUE
program, which would eventually see over 200 intercept and forward-processing
systems deployed around the world.66 Chief Coulter was probably referring to
that program when he testified in 2005 that “we are trying to go out there into
those haystacks and electronic highways outside our borders and find informa-
tion that is of value to government decision makers.”67
A similar evolution was underway around the same time at NSA, which
cancelled its troubled TRAILBLAZER program in mid-2004 and in 2005
launched a replacement program called TURBULENCE.68 By late 2006, NSA
was anticipating the imminent deployment of TURBULENCE’s TURMOIL
subsystem, which would improve the agency’s Internet filtering and selection
processes and move them to forward intercept points to help save its analysts
102 Bill Robinson

from the “tsunami of intercept” threatening to overwhelm them.69 That tsunami

was still building. Between 2000 and 2007, the spread of broadband Internet
“effectively multiplied the world’s telecommunication capacity by a factor of 29,
from 2.2 exabytes in 2000 to 65 in 2007.”70 And it had only just begun. In 2016,
annual Internet traffic surpassed 1,000 exabytes (one zettabyte). Much of that
traffic consisted of streaming video and other feeds of little or no interest to
SIGINT agencies, but the flood still had to be filtered and processed in order
to find the traffic that was of interest.
In April 2007, John Adams, who became CSE chief in 2005, declared his
agency’s intent to “master the Internet” in partnership with the other Five Eyes
agencies:

That is a challenge that no one institution – be it ours or the National Security

Agency, NSA, for that matter – can manage on their own. We try to do that
in conjunction with our allies. At the same time, we have a threat that is very
diverse, very distributed around the world – similar to needles in haystacks. We
have the combination of the technology and the threat that, together, make it
virtually impossible for any one organization to manage it on its own ... If we
are to master that Internet, we will have to do it together; and we are focusing
on that.71

GCHQ announced its own “Mastering the Internet” project during the same
year,72 and NSA’s SIGINT Mission Strategic Plan FY2008–2013, promulgated in
October 2007, declared that it too sought “to utterly master the foreign intel-
ligence implications of networks and network technology.”73 One way NSA
sought to do so was to collect target communications at the Internet companies
that handled them, through the PRISM program, instead of intercepting them
in the wild. In September 2007, Microsoft became the first participant in PRISM.
Yahoo, Google, Facebook, and others quickly followed.74 Within a few years,
data obtained through PRISM appeared in nearly one in seven of the first-,
second-, and third-party reports produced or received by NSA.75 The NSA vision
also called for improved “collection operations around the globe” to identify
and collect the most important traffic wherever it could be found, using “fast,
flexible, front-end processors [to] spot targets based on patterns, events, or
metadata rather than pre-defined selectors or brute force scanning of
content.”76
The key to understanding and managing the deluge of data streaming through
the Five Eyes worldwide monitoring systems was the collection and analysis of
telephone and Internet metadata, which could be used both to monitor specific
targets and, at least potentially, to identify previously unknown individuals and
 From 1967 to 2017 103

activities of intelligence interest. In 2004, CSE began working with the Math-
ematics of Information Technology and Complex Systems (MITACS) consor-
tium, a Canadian network of academia, industry, and the public sector, to
improve the agency’s ability to exploit metadata. A 2006 description of the
MITACS Semi-Supervised Learning in Large Graphs project provides a rare
public look into CSE’s interests:

As part of ongoing collaborations with the Communications Security Establish-

ment (CSE), we are applying unsupervised and semi-supervised learning meth-
ods to understand transactions on large dynamic networks, such as telephone and
email networks. When viewed as a graph, the nodes correspond to individuals
that send or receive messages, and edges correspond to the messages themselves.
The graphs we address can be observed in real-time, include from hundreds to
hundreds of thousands of nodes, and feature thousands to millions of transac-
tions ... For reasons of efficiency, we have restricted our attention to meta-data
of message transactions, such as the time, sender, and recipient, and ignored the
contents of messages themselves. In collaboration with CSE, we are studying the
problem of counter-terrorism, a semi-supervised problem in which some terror-
ists in a large network are labeled, but most are not ... Another common feature
of counter-terrorism problems is the fact that large volumes of data are often
“streamed” through various collection sites, in order to provide maximal infor-
mation in a timely fashion. A consequence of efficient collection of transactions
on very large graphs is that the data itself can only be stored for a short time. This
leads to a nonstandard learning problem, since most learning algorithms assume
that the full dataset can be accessed for training purposes. Working in conjunc-
tion with CSE, we will devise on-line learning algorithms that scale efficiently
with increasing volume, and need only use each example once.77

Much of this research was later placed under the aegis of the Cryptologic
Research Institute (now called the Tutte Institute for Mathematics and Comput-
ing), which was created in 2009 to help CSE bring outside mathematical exper-
tise to bear on cryptanalytic and data-mining questions. The agency acquired
a Cray XMT, a supercomputer optimized for data-mining operations, around
the same time.78 The XMT excels at two types of problems: “The first is the
finding-the-needle-in-a-haystack problem, which involves locating a particular
piece of information inside a huge dataset. The other is the connecting-the-dots
problem, where you want to establish complex relationships in a cloud of seem-
ingly unrelated data.”79
CSE also sought big data analysis techniques that could run on “non-extra-
ordinary” hardware. In late 2012, it began implementing what it called a New
104 Bill Robinson

Analytical Model designed to help its intelligence analysts keep up with and
better exploit the metadata and other target-related information, such as finan-
cial and travel data, increasingly available to them.80
Additional growth in CSE’s workforce was also on the agenda. Between 2009
and 2014, the agency grew from 1,650 employees to roughly 2,200, an increase
of 33 percent over five years.81 CSE’s Long-Term Accommodation Project (LTAP)
saw the construction of a brand-new headquarters complex located next to the
CSIS headquarters over the same period. CSE asserts that the Edward Drake
Building contains both the “largest concentration of supercomputers in
Canada”82 and the “largest volume databases in the country.”83

Future Prospects
Whether the Five Eyes partners can truly be said to have “mastered” the Internet
is open to question. In February 2012, however, the NSA felt justified in declar-
ing it had successfully made the transition to the Information Age: “As the world
has changed, and global interdependence and the advent of the information
age have transformed the nature of our target space, we have adapted in innova-
tive and creative ways that have led some to describe the current day as ‘the
golden age of SIGINT.’”84
The gradual spread of encryption in email, web browsing, and messaging
apps – almost certainly accelerated to some degree by the Snowden leaks in
2013 – may have taken some of the lustre off the gold in the years since that
statement. CSE maintains that Edward Snowden’s “unauthorized disclosures
have diminished the advantage that we have had, both in the short term but
more worryingly in the long term.”85 It is certainly the case that encryption is
becoming more common. In 2016, the United Kingdom’s Independent Reviewer
of Terrorism Legislation reported that “about 50% of Internet traffic was now
encrypted, and 100% of emails from major email providers.”86 In June 2017,
Australian attorney-general George Brandis lamented that over 40 percent of
counterterrorism investigations were encountering encrypted communications,
compared with less than 3 percent in mid-2013: “Within a short number of years,
effectively, 100 per cent of communications are going to use encryption ... This
problem is going to degrade if not destroy our capacity to gather and act upon
intelligence unless it’s addressed.”87
Such claims are almost certainly exaggerated, however. Depending on when
and in what form quantum computing makes its appearance, existing Internet
encryption technologies could be rendered obsolete in less than a decade.88 In
the meantime, the continuing migration of Internet traffic to mobile devices,
the pervasive vulnerability of existing software, and the growth of the Internet
of Things may be making targeted surveillance even more difficult to evade – as
 From 1967 to 2017 105

long as you can identify your target. The Five Eyes agencies were already work-
ing hard on smartphone exploitation techniques in 2010, and they are likely to
have made progress in the years since.89 As ubiquitous computing becomes ever
more deeply embedded in daily life, and governments and corporations collect
the growing data trails thus generated, big data analysis is also likely to take on
increasing importance, both as a means of identifying and tracking individual
targets and for generating unique intelligence on target activities and
connections.
Between June 2011 and May 2012, NSA and its partners shared approximately
180,000 end product reports, not much lower than the approximately 200,000
shared in 1990. Of those 180,000, 7,511 were issued by Australia, 11,257 were
issued by the United Kingdom, and approximately 150,000 were issued by the
United States.90 A Canadian figure was not provided, but the number was prob-
ably somewhat lower than the 10,000 Canadian reports issued in 1990.91 How-
ever, more recent documents suggest that the Canadian total has returned to
the five-digit range.92 On their face, these numbers do not support the suggestion
that Canada’s SIGINT output has suffered in the post-Snowden era, but it is
possible that their continuing quantity masks a decline in quality.

An Agency Transformed
Today’s CSE is very different from the CBNRC of 1967. It has more than four
times the workforce it had fifty years ago, a much larger budget, and immeasur-
ably greater information collection, storage, and processing capabilities. Its
SIGINT activities remain focused on foreign intelligence, but the agency now
has a much larger role in support of domestic law enforcement and security
agencies than it had in the past. It also has a much more extensive role in infor-
mation security than it had in 1967. CSE is even more tightly bound into the
Five Eyes transnational SIGINT partnership, and it considers that partnership
to be more valuable now than it has ever been.93 The agency has evolved from
a passive collector of radio signals received in Canada to an active hunter of
data stored on information systems or passing through the air or fibre-optic
cables at locations far removed from Canadian soil. It now has legal authority
to intercept Canadian communications during its operations, although it can
target Canadians or persons in Canada only when operating under the aegis of
judicial warrants obtained by the law enforcement and security agencies to
which it provides support. In the course of its activities, it collects and processes
vast amounts of metadata pertaining to Canadians. And, perhaps most import-
ant, it now operates in a domain where foreign communications and Canadian
communications are deeply and inevitably intermixed, its desired targets are
frequently difficult to distinguish from non-targets, and activities such as
 106 Bill Robinson

terrorism that have an important nexus with Canadian domestic life are much
higher on its intelligence priorities. It also operates under the eyes of a watchdog
agency to ensure that CSE complies with the law and, since 2017, a new parlia-
mentary review body, the National Security and Intelligence Committee of
Parliamentarians (NSICOP). Further change came in June 2019 with the passage
of Bill C-59, which gave CSE the authority to conduct Computer Network Attack
operations for both defensive and offensive purposes, extended its cybersecurity
mandate to the protection of Canadian private sector infrastructures (subject
to request by those entities), and created an entirely new oversight and review
structure for the agency.94 These changes in priorities, budget resources, legal
authorities, and worldwide communications and information technologies have
transformed the relationship between CSE and Canadians. As this chapter has
demonstrated, early concerns about domestic surveillance by CSE were largely
unfounded, but the potential for Canadians to be drawn into the agency’s dragnet
is now much greater, highlighting the importance of oversight, review, and
transparency measures for preventing abuse of the agency’s extraordinarily
intrusive capabilities.

Notes
1 Matthew Aid, Secret Sentry: The Untold History of the National Security Agency (New
York: Bloomsbury Press, 2009), 139.
2 Kevin O’Neill and Ken Hughes, “History of CBNRC,” Communications Security Estab-
lishment, August 1987, vol 4, ch 14, 43–45, released in redacted form under Access to
Information Request Number A-2015-00045.
3 “The Current Canadian Intelligence Program – Objectives and Activities,” attachment
to draft memorandum to the Cabinet Committee on Security and Intelligence, 20 April
1972, Library and Archives Canada (LAC), RG 25, box 10, file 1-3-12-1. I am indebted
to the Canadian Foreign Intelligence History Project for access to this and the other LAC
documents cited in this chapter.
4 O’Neill and Hughes, “History of CBNRC,” vol 6, ch 25, 8.
5 Aid, Secret Sentry, 17–18.
6 “Supplementary Radio Activities Consolidation Plan,” Department of National Defence,
30 May 1966, released in redacted form under an Access to Information request.
7 O’Neill and Hughes, “History of CBNRC,” vol 1, ch 4, 16–17.
8 Prior to 1956, all Canadian overseas telephone calls were transmitted by high-frequency
radio; the last overseas call transmitted from Canada by commercial radio was made in
1975.
9 Report of the Royal Commission on Security (Abridged) (Ottawa: Queen’s Printer, June
1969), 5.
10 O’Neill and Hughes, “History of CBNRC,” vol 1, ch 2, 24.
11 Claude M. Isbister, “Intelligence Operations in the Canadian Government,” Privy Coun-
cil Office, 9 November 1970, 51, released in redacted form under Access to Information
Request Number A-2011-00010.
12 GCHQ alone processed about 1 million ILC messages a month at this time: O’Neill
and Hughes, “History of CBNRC,” vol 3, ch 13, 16–17. NSA’s effort was similar in scale
 From 1967 to 2017 107

– in 1975 it was estimated that 2.8 million of the 2 billion telegrams that passed over
ILC channels every month were forwarded to NSA headquarters, where analysts pro-
cessed about 1 million of them: Letter from Frederick A.O. Schwarz to Thomas Latimer,
Tab A, 16 September 1975, 5, National Security Archive, http://nsarchive2.gwu.edu//
dc.html?doc=4058229-Document-10-Letter-from-Frederick-A-O-Schwarz-to.
13 “The Canadian Intelligence Program,” draft memorandum to the Cabinet Committee
on Security and Intelligence (CCSI), 20 April 1972, 13–14, LAC, RG 25, box 10, file 1-3-
12-1. The final version of this document was considered by the CCSI in May 1972. The
Canadian Overseas Telecommunications Corporation was a Crown corporation that
had a monopoly on overseas telephone, telex, and telegraph services from Canada. It
was renamed Teleglobe Canada in 1975 and was later privatized, eventually becoming
part of Tata Communications.
14 A.F. Hart, “Meeting of Interdepartmental Committee on Security and Intelligence (ICSI)
– Tuesday, May 2, 2:30 pm,” 27 April 1972, LAC, RG 25, box 10, file 1-3-12-1.
15 Ibid.
16 O’Neill and Hughes, “History of CBNRC,” vol 6, ch 25, 13.
17 Ibid., vol 6, ch 26, 44–45. See also vol 1, ch 2, 30.
18 Ibid., vol 6, ch 26, 45–46.
19 Memorandum from John Hadwen to Major-General Reg Weeks, 25 March 1977, LAC,
RG 25, vol 29022, file 29-4-IAC, pt 2.
20 The Quebec government’s Keable Inquiry was announced on 15 June 1977, and the fed-
eral government’s McDonald Commission followed on 6 July 1977.
21 “Extract of the Minutes of the 1th [sic] Meeting of the Intelligence Advisory Committee
Held on Wednesday, 11 January 1978,” Intelligence Advisory Committee, LAC, RG 25,
vol 29022, file 29-4-IAC, pt 3.
22 Intelligence Advisory Committee (IAC) Review of 1977 and Forecast for 1978, Intelli-
gence Advisory Committee, 22 February 1978, LAC, RG 25, vol 29022, file 29-4-IAC,
pt 3.
23 The current Department of Justice view is that it is legal for CSE to receive one-end
Canadian traffic intercepted by Canada’s allies. However, it is not permitted to ask those
allies to target the communications of specific Canadians or persons in Canada except
at the request of a federal law enforcement or security agency operating under a suit-
able warrant. Access to second- and third-party intercepts might explain former CSE
employee Mike Frost’s claim that there was a “French problem” section within CSE dur-
ing the 1970s. See Mike Frost and Michel Gratton, Spyworld: Inside the Canadian and
American Intelligence Establishments (Toronto: Doubleday, 1994), 96.
24 Teleglobe Canada, 37th Annual Report, for the Year Ended December 31, 1986 (Ottawa:
Teleglobe Canada, 1987), 24.
25 Security Intelligence Review Committee, SIRC Report 2001–2002: An Operational Audit
of the Canadian Security Intelligence Service (Ottawa: Public Works and Government
Services Canada, 2002), 14–15, http://www.sirc-csars.gc.ca/pdfs/ar_2001-2002-eng.pdf.
26 O’Neill and Hughes, “History of CBNRC,” vol 3, ch 11, 87; vol 6, ch 26, 43.
27 Ibid., vol 1, ch 4, 20–21.
28 Lieutenant General Donald Bennett, “Executive Session of USIB [US Intelligence
Board], Thursday, 5 November 1970,” Memorandum for the record, 9 November 1970. I
am grateful to the late Jeffrey Richelson for providing a copy of this document to me.
29 According to Jeffrey Richelson, both Canada and the United Kingdom assisted in process-
ing CANYON traffic: Jeffrey Richelson, “Eavesdroppers in Disguise,” AIR FORCE Maga-
zine, August 2012, 58–61, http://www.airforcemag.com/MagazineArchive/Documents/
2012/August%202012/0812eavesdroppers.pdf.
 108 Bill Robinson

30 Thomas R. Johnson, “A Cryptologist Encounters the Human Side of Intelligence,” Stud-

ies in Intelligence, 8 February 2007, 2, https://www.cia.gov/library/readingroom/docs/
DOC_0001407027.pdf.
31 Thomas R. Johnson, American Cryptology during the Cold War 1945–1989 (Fort Meade,
MD: National Security Agency, 1998), bk. III, 223. A large part of this success was due to
secret US/West German control over Crypto AG, the company that supplied the encryp-
tion machines used by many non-Soviet Bloc countries: Greg Miller, “The Intelligence
Coup of the Century,” Washington Post, 11 February 2020.
32 Frost and Gratton, Spyworld, 112.
33 Bill Robinson, “The Fall and Rise of Cryptanalysis in Canada,” Cryptologia 16, 2 (1992): 23–38.
34 Defence Signals Directorate, “DSD and Supercomputers,” Internet Archive, http://web.
archive.org/web/20020103063013/http://www.dsd.gov.au:80/dsd/supercomp.html.
DSD was renamed the Australian Signals Directorate in 2013.
35 Nicky Hager, Secret Power: New Zealand’s Role in the International Spy Network (Nelson,
NZ: Craig Potton, 1996), 110–11.
36 Personal observations.
37 John M. Fraser, “S&I Consultative Meeting: December 18,” 17 December 1987, LAC, RG
25, file 29-4-ICSI, pt 5. MADRIGAL was the section 16 project.
38 P.R. Anderson, “ICSI Meeting June 28 15:00 hrs.: Agenda Items,” 28 June 1988, LAC, RG
25, file 29-4-ICSI-2, pt 4.
39 Lynn Wortman and George Fraser, History of Canadian Signals Intelligence and Direction
Finding (London: Nanlyn Press, 2005), 15.
40 “The Canadian Intelligence Community,” 16 March 1990, app. C, LAC, RG 25, BAN
2016-0149, box 2, file 3-5-5, pt 1.
41 “Canada Spied on Allies, Former CSE Agent Says: Embassies Said to Be Subjects of Sur-
veillance,” Globe and Mail, 13 November 1995.
42 John Adams, “CCSE Speech to Université Laval Students,” 6 February 2007. On file with
author.
43 Colin Freeze, “How CSEC Became an Electronic Spying Giant,” Globe and Mail, 30
November 2013.
44 Dave Pugliese, “Canadian Spies Join U.S. Drug War,” Ottawa Citizen, 17 May 2001.
45 “Integrated SIGINT Operational Model (ISOM)” presentation, slide 8, 652 in CSEC
Foundational Learning Curriculum (Ottawa: Communications Security Establishment,
January 2013), released in redacted form to journalist Colin Freeze under an Access to
Information request.
46 Duncan Campbell, “Development of Surveillance Technology and Risk of Abuse of
Economic Information” (working document for the STOA Panel, European Parliament,
Luxembourg, October 1999), http://www.duncancampbell.org/menu/surveillance/ech-
elon/IC2000_Report%20.pdf.
47 Martin Hilbert and Priscila López, “The World’s Technological Capacity to Store, Com-
municate, and Compute Information,” Science, 1 April 2011, 60–65, and supporting
online material.
48 Transition 2001 (Fort Meade, MD: National Security Agency, December 2000), 3.
49 Ibid., 32.
50 Office of the Communications Security Establishment Commissioner, Annual Report
2000–2001 (Ottawa: Public Works and Government Services Canada, 2001), 7, https://
www.ocsec-bccst.gc.ca/a83/ann-rpt-2000-2001_e.pdf.
51 The CSE Commissioner commented in 2001 that “CSE is well aware that it must con-
tinually upgrade its capabilities to screen out Canadian communications or risk acting
unlawfully if it does not make every effort to do so”: ibid., 13.
 From 1967 to 2017 109

52 But it has happened on one occasion. See Bill Robinson, “CSE Commissioner: CSE
Violated Law,” Lux Ex Umbra (blog), 28 January 2016, https://luxexumbra.blogspot.
ca/2016/01/cse-commissioner-cse-violated-law.html.
53 Keith Coulter, “CSE’s Post-9/11 Transformation” (speech to the Canadian Association
of Security and Intelligence Studies conference, 15 October 2004), Internet Archive,
https://web.archive.org/web/20060502140839/http://www.cse-cst.gc.ca:80/documents/
publications/casis-speech.pdf.
54 Wortman and Fraser, History of Canadian Signals Intelligence and Direction Finding, 131.
771 Communications Research Squadron was disbanded in December 2002: Christine
Grimard, “15 Years of Service Remembered,” Maple Leaf, 9 April 2003.
55 Coulter, “CSE’s Post-9/11 Transformation.”
56 Canada, Bill C-36, An Act to amend the Criminal Code, the Official Secrets Act, the Can-
ada Evidence Act, the Proceeds of Crime (Money Laundering) Act and other Acts, and to
enact measures respecting the registration of charities in order to combat terrorism, 1st
Sess, 37th Parl, LEGISinfo, http://www.parl.ca/LegisInfo/BillDetails.aspx?Language=en
&Mode=1&billId=73328.
57 Canadian Press, “Secretive Federal Spy Agencies Get $47 Million for New Technology,”
19 October 2001.
58 Office of the Communications Security Establishment Commissioner, Annual Report
2013–2014 (Ottawa: Public Works and Government Services Canada, 2014), 32, https://
www.ocsec-bccst.gc.ca/a37/ann-rpt-2013-2014_e.pdf.
59 “Communications Security Establishment (CSE) – Our Good Neighbor to the North,”
7 August 2003, SIDtoday (internal NSA publication), https://theintercept.com/
snowden-sidtoday/3008306-communications-security-establishment-cse-our/.
60 Keith Coulter, testimony to Special Senate Committee on the Anti-terrorism Act, 11
April 2005, Senate of Canada, https://www.sencanada.ca/en/Content/SEN/Commit-
tee/381/anti/07evb-e. In later testimony, he used the figure “over 75%” and included
counter-intelligence activities in the count: Coulter, testimony to Subcommittee on
Public Safety and National Security of the Standing Committee on Justice, Human
Rights, Public Safety and Emergency Preparedness, House of Commons, http://www.
ourcommons.ca/DocumentViewer/en/38-1/SNSN/meeting-11/evidence. In both cases,
the figure likely included the 20–25 percent of the CSE budget then spent on the IT
Security program.
61 John Adams, testimony to the Standing Senate Committee on National Security and
Defence, 30 April 2007, Senate of Canada, https://sencanada.ca/en/Content/Sen/
committee/391/defe/15evb-e.
62 Office of the Communications Security Establishment Commissioner, “Role of the
CSE’s Client Relations Officers and the Operational Policy Section (D2) in the Release
of Canadian Identities,” 30 March 2007, 7, released in redacted form. The actual number
was redacted from the document, but it can easily be seen that it consists of three digits
and begins with a three.
63 Greg Fyffe, “The Canadian Intelligence Community after 9/11,” Journal of Military and
Strategic Studies 13, 3 (Spring 2011): 6.
64 “The Global Network Forum (Update #1),” 22 October 2004, SIDToday, https://the
intercept.com/snowden-sidtoday/3676087-the-global-network-forum-update-1/; see also
“Coming Soon: A SID Classification Guide,” 1 March 2005, SIDToday, https://theintercept.
com/snowden-sidtoday/3991126-coming-soon-a-sid-classification-guide/.
65 The directive was signed on 15 March 2004. The first ministerial directive specifically
on metadata was signed on 9 March 2005. According to the latter directive, “metadata
is defined as information associated with a telecommunication to identify, describe,
 110 Bill Robinson

manage or route that telecommunication or any part of it as well as the means by which
it was transmitted, but excludes any information or part of information which could
reveal the purport of a telecommunication, or the whole or any part of its content.”
66 Communications Security Establishment (CSE), “CSEC SIGINT Cyber Discovery:
Summary of the Current Effort” (slide deck, November 2010), 13, https://christopher-
parsons.com/Main/wp-content/uploads/2015/02/cse-csec-sigint-cyber-discovery.pdf.
67 Keith Coulter, testimony to the Special Senate Committee on the Anti-terrorism
Act, 11 April 2005, Senate of Canada, https://www.sencanada.ca/en/Content/SEN/
Committee/381/anti/07evb-e.
68 Fred Kaplan, Dark Territory: The Secret History of Cyber War (New York: Simon and
Schuster, 2016), 156–57.
69 “Dealing with a ‘Tsunami’ of Intercept,” SIDtoday, 29 August 2006, https://www.eff.org/
files/2015/05/26/20150505-intercept-sidtoday-tsunami-of-intercept-final.pdf.
70 Hilbert and López, “The World’s Technological Capacity,” 63.
71 John Adams, testimony to the Standing Senate Committee on National Security and
Defence, 30 April 2007, Senate of Canada, http://www.parl.gc.ca/Content/SEN/Committee/
391/defe/15evb-e.htm.
72 Christopher Williams, “Jacqui’s Secret Plan to ‘Master the Internet,’” The Register, 3 May
2009.
73 National Security Agency (NSA), “SIGINT Mission Strategic Plan FY2008–2013,” 3
October 2007, 4, https://www.eff.org/files/2013/11/15/20131104-nyt-sigint_strategic_plan.
pdf.
74 NSA, “PRISM/US-984XN Overview” (slide deck, April 2013), 6, https://snowden
archive.cjfe.org/greenstone/collect/snowden1/index/assoc/HASH01f5/323b0a6e.dir/
doc.pdf.
75 NSA, “PRISM Expands Impacts: FY12 Metrics,” 19 November 2012, https://www.aclu.
org/foia-document/prism-expands-impacts-fy12-metrics.
76 NSA, “SIGINT Mission Strategic Plan FY2008–2013,” 8.
77 “Mathematics of Information Technology and Complex Systems: Research,” Inter-
net Archive, https://web.archive.org/web/20070519133815/http://www.iro.umontreal.
ca:80/~bengioy/mitacs/Research.htm.
78 CSE, “CSEC ITS/N2E: Cyber Threat Discovery” (slide deck, 2010), 47, https://christo-
pher-parsons.com/Main/wp-content/uploads/2015/03/csec-its-dsco-2010-20101026-
final.pdf.
79 Michael Feldman, “Cray Pushes XMT Supercomputer into the Limelight,” HPCwire, 26 Jan-
uary 2011, https://www.hpcwire.com/2011/01/26/cray_pushes_xmt_supercomputer_
into_the_limelight/.
80 For further discussion of the New Analytical Model, see Chapter 6.
81 As of 2020, the total had grown to approximately 2,900.
82 “Introduction to CSE Deck,” CSE, November 2015, 3, released in redacted form under
Access to Information Request Number A-2015-00067.
83 CSE, “Experienced Professionals and New Graduates,” 24 April 2012, Internet Archive,
https://web.archive.org/web/20130527193541/http://www.cse-cst.gc.ca/home-accueil/
careers-carrieres/professionals-professionnels-eng.html.
84 SIGINT Strategy 2012–2016 (Fort Meade, MD: National Security Agency, 23 February
2012), 2, https://www.eff.org/files/2013/11/25/20131123-nyt-sigint_strategy_feb_2012.
pdf.
85 “Unauthorized Disclosures,” CERRID #20084275, CSE, contained in briefing binder
prepared for the Chief of CSE in March 2015, released under Access to Information
Request Number A-2015-00021.
 From 1967 to 2017 111

86 David Anderson, Report of the Bulk Powers Review (London: Williams Lea Group,
August 2016), 105.
87 David Wroe, “How the Turnbull Government Plans to Access Encrypted Messages,” Sydney
Morning Herald, 11 June 2017.
88 Ian MacLeod, “Quantum Computing Will Cripple Encryption Methods within Decade,
Spy Agency Chief Warns,” Ottawa Citizen, 23 September 2016.
89 Government Communications Headquarters, “Mobile Theme Briefing: May 28 2010”
(slide deck, 28 May 2010), 2, https://christopher-parsons.com/Main/wp-content/uploads/
2014/12/gchq-mobile-theme-briefing.pdf.
90 NSA, “PRISM Based Reporting June 2011–May 2012” (slide deck, 13 June 2012), https://
www.aclu.org/foia-document/prism-based-reporting-june-2011-may-2012. See also NSA,
“PRISM Expands Impacts.”
91 A briefing note produced by CSE appears to show that a four-digit number of reports
were produced in FY 2011–12, i.e., from April 2011 to March 2012: “CSEC Metadata
Collection,” CSEC ref: 1327209, 18 June 2013, released in redacted form under Access to
Information Request Number A-2013-00058.
92 For example, “In 2013–14, CSE issued [redacted five-digit number] intelligence reports
(known as End Product Reports, or EPRs) in line with GC intelligence priorities.”
Annual Report to the Minister of National Defence 2013–2014 (Ottawa: Communications
Security Establishment, 2014), 2, released in redacted form under Access to Information
Request Number A-2015-00086.
93 CSE, “Risk Assessment: Information Sharing with the Second Parties” (draft, 18 Decem-
ber 2015), 3, released in redacted form under Access to Information Request Number
A-2015-00052.
94 For more on Bill C-59 as it pertains to CSE, see Christopher Parsons, Lex Gill, Tamir
Israel, Bill Robinson, and Ronald Deibert, “Analysis of the Communications Security
Establishment Act and Related Provisions in Bill C-59 (An Act respecting national secu-
rity matters), First Reading (December 18, 2017)” (Citizen Lab/Canadian Internet Policy
and Public Interest Clinic report, December 2017), https://citizenlab.ca/wp-content/
uploads/2018/01/C-59-Analysis-1.0.pdf.
6
Pixies, Pop-Out Intelligence, and Sandbox Play
The New Analytic Model and National Security
Surveillance in Canada
Scott Thompson and David Lyon

“Everything has changed,” proclaims a Canadian Communications Security

Establishment (CSE) document, obtained using an Access to Information and
Privacy (ATIP) request. It refers to the remarkable shift from older intelligence-
gathering methods that had evolved steadily since the agency’s founding in 1946
to an embrace of big data as a fresh set of practices. The new practices, collectively
described as a “New Analytic Model (NAM),” include data mining – for example,
interception from deep packet inspection devices (see Chapter 7), posted
material from social media, apps, and the “mountains of raw data” stored or
generated from Internet-connected devices1 – and machine learning, in which
both modes of data capture and modes of analysis are radically different from
those of the past. Drawing on internal CSE documents obtained though ATIP
requests,2 this chapter demonstrates a significant shift in how CSE approaches
knowledge production and speaks to the broader implications of a big data
approach to national security work.
Sometimes history moves in a steady arc where the trajectory remains more
or less the same over many decades or even centuries. From time to time, how-
ever, a historical turn occurs in which past practices are abandoned or allowed
to shrink in importance, to be replaced with new practices that come to dominate
the scene and, cascade-like, have effects in a broad range of other areas. A his-
toric turn has been made at CSE. Some aspects of this are plain to see, such as
the iconic multibillion-dollar Edward Drake Building in Ottawa and the part-
nership with the Tutte Institute for Mathematics and Computing (TIMC). Others
are closely guarded secrets – there is notable continuity rather than rupture
here – that may be gleaned only from non-redacted sections of documents
released in response to specific requests.
Several other factors might lead one to suspect that there is a new modus
vivendi at CSE. The Snowden scandal in 2013 prompted serious questions about
how exactly the Canadian partner of the US National Security Agency (NSA)
conducted its intelligence gathering. Two questions preoccupied journalists
and others after the disclosures about the NSA: (1) were Canadians being
targeted despite CSE’s mandate to deal with foreign intelligence? and (2) how
was metadata treated, as personal data or as something else, not covered by
 Pixies, Pop-Out Intelligence, and Sandbox Play 113

any privacy legislation? There was a recognition that massive amounts of data
were becoming available, especially following the rapid development of Web
2.0, and that CSE was working towards exploiting those new data sources3 (see
also Chapter 14).
What was not fully recognized, however, was that a wholesale shift to big data
practices was underway, creating a watershed in data handling. As Carrie Sand-
ers and James Sheptycki put it (in relation to policing), this new development
amounts to the “algorithmic administration of populations and territory ...
based on morally neutral technology.”4 This is a new mode of doing intelligence
gathering and analysis, based in fact on global neoliberalism, seen, for instance,
in the public-private partnerships that help to drive the direction of the agency.
The disclosures by Edward Snowden in 2013 and later make it very clear that
among national security agencies, and especially at the NSA, there is a shift to
big data practices.5 The attempts to legalize certain data-gathering activities in
Canada, notably in the Anti-terrorism Act, known generally as Bill C-51, display
a desire to normalize big data in this realm.6 While these two factors of data
handling and public-private partnerships are clearest, there are others.
The establishment of the Tutte Institute for Mathematics and Computing in
2009 and its partnership with CSE are a reminder of a long-term association
with top researchers. CSE was created as such in 1975, from the former Com-
munications Branch of the National Research Council. Computer scientists,
mathematicians, engineers, linguists, and analysts are all found at CSE. Hints
of a new approach were available a few years ago. For example, Colin Freeze
found that new recruits at CSE were told not to emulate Hollywood’s James
Bond style of secret agents. Rather, they should act like the “traffic fairy,” a “tiny
pixie who apparently flits through computer traffic in search of secrets.”7
Here we comment first on the nature of the shift from the evolutionary
approach since 1946 to the big data turn starting in 2012. From the 1940s to
1974, the existence of what became CSE was secret, but it is now known that
signals intelligence (SIGINT) was its main mandate from the Second World
War onward. This meant telephone, radio (“wireless”), and telegraph – indeed
any system of communication used to send classified, and thus encoded or
encrypted, information. The primary user of SIGINT was the Department of
National Defence, although various leaks and scandals showed that sometimes
Canadian citizens, including prominent ones, could be in view.
The shift to big data practices depends heavily on very large-scale computing
facilities as well as technical, mathematical, and statistical expertise – hence
CSE depends not on the tens of workers with which it began in the 1940s but
now on more than 2,000 operating staff. At CSE, as elsewhere, big data is not
so much suspicion-driven as data-driven. That is, rather than a process of
114 Scott Thompson and David Lyon

targeted scrutiny of groups and individuals, big data engages in what was called
after Snowden “mass surveillance,” that is monitoring of communications to
discern patterns of relationship that may be deemed “actionable intelligence.”
It should of course be recalled that the kinds of data sought and analyzed are
millions or billions of bits of data generated by everyday transactions and com-
munications; numbers, text, graphics, videos, images, and sensor information.
This is what the “pixie” is trained to sift through.
Second, we note some key features of CSE big data practices and comment
on their broader meanings. The term “big data” usually refers to the volume of
data in use, which can now be handled by large-scale computing facilities that
contribute to the velocity of calculation and analysis possible, along with the
variety of datasets that may be drawn into any given analysis. Each of these
characteristics is visible in the New Analytic Model. Although this name seems
to lend weight to the analytical dimensions of big data, it is no less the case that
CSE depends on new modes of data capture, especially those associated with
“data exhaust” from everyday communications and transactions and so-called
user-generated content from massive social media sources.
Third, we observe that the changes at CSE are not by any means limited to
that body. Rather, some deliberate strategies indicate a plan to influence many
or even all government departments to use big data methods, with which
CSE expertise will be available to assist. Moreover, through potential
legal changes and alteration of the protocols of data management policy, the
idea  is to catalyze a cascading effect of these new practices throughout
government.
In what follows, we show what sorts of pressures and opportunities produced
the radical shift to big data at CSE, the discernible patterns of development of
big data that are becoming visible, and the likely consequences of the adoption
of big data practices in intelligence services in Canada. This final item has a
necessarily critical edge, but is intended to prompt those engaged with CSE’s
mission to reflect more deeply on certain matters that have to do with ethics,
citizenship, and democratic oversight of which CSE is already aware, as seen in
the ATIP documents made available to us.

The CSE’s “New Analytic Model”

A significant shift in how intelligence work is done within CSE is described in
internal documents as resulting from the new production, availability, and
institutional ingestion of large, machine-readable datasets. In the “New Analytic
Environment” (NAE), CSE documents from 2012 specifically identified a spike
and shift in the volume and type of data ingested.8 In response, the latter part
of 2012 saw the radical adoption of new analytics-based methods and
 Pixies, Pop-Out Intelligence, and Sandbox Play 115

applications under a new initiative described as the “New Analytic Model.” In

its application, the model put aside the traditional “target, translate, report”
approach and replaced it with one in which CSE analysts would be “rebuilt” to
take up “new tasks” in a new “working environment” with an expanded skillset.9
In short, analysts were to become data scientists – to “compose tests (tradecraft)
through the composition of analytics.”10
With changes in how national security knowledge development was defined
in order to place a primary focus on data analytics, work at CSE would have to
change as well. The NAM policy documents assert that “to address this challenge
requires CSE and DGI [the Director General for Intelligence] to improve our
skills, and shift our SIGINT systems development focus toward a new form of
architecture and environment.”11 As these internal policy documents note, the
shift in approach to SIGINT would be significant and fundamentally reshape
the work done at CSE: “to be clear what we’re talking about here is a revolution
in the field of SIGINT analysis,” as the NAM called for “a departure from the
old way of doing things and there are tremendous implications for all of us.”12
In taking up this data-driven, fluid, and flexible model, program designers
identified six key benefits to SIGINT data collection and analysis from the NAM:

1) Provide a greater visibility into all collected data that w[ould] allow for the
rapid development of analytical capabilities to enrich, discover and to analyze
trends and anomalies across all relevant repositories at once; 2) Ensure that the
technological evolution supports the analysts’ quest to explore, experiment, dis-
cover, analyse and assert findings; 3) [REDACTED]; 4) Enable more effective
and efficient implementation and performance of analytic methods by lever-
aging computer automation and machine learning in order to aid in the for-
mulation of reasonable and actionable conclusions from sometimes conflicting
information; 5) Evolve the deliverables beyond static one-way interface and aim
towards an interactive mechanism through cutting edge interactive visualiza-
tion that present valuable insights in impactful ways; 6) Provide the ability to
plan, share and collaborate, monitor and measure performance, analyze results,
predict outcomes and strategize among peers, across teams, across units and,
ideally, set the stage for continued extension of these features.13

Where previously a single analyst or team would work to target, translate, and
report on identified targets, acting in more siloed, specialist areas, the NAE
would call for “a complete reworking of the DGI analyst’s task, working environ-
ment, and skill set,” acknowledging “that the role of analysis needs to undergo
a revolution of sorts, focused on innovation and sharing and collaboration.”14
Analysts would shift “from being a unique service provider, to being just one
116 Scott Thompson and David Lyon

tool in the box, from being a collector of facts, to an assessor or creator of

information.”15
Ultimately, the goal of the shift at CSE was to “re-build the intelligence analyst
through the NAM and the NAE” by shifting from a specialized research area,
with a targeted focus and specialty, to one that instead engaged with a sandbox
of available data, moving flexibly from project to project exploring, experiencing,
testing, and predicting, with machine learning and big data analytics tools.16

The New Analytic Model, Big Data, Machine Learning,

and Data Sandbox Analysts
As noted above, CSE’s response to the availability and ingestion of new types
and massively increased volumes of machine-readable data was the development
and adoption of the New Analytic Model policy initiative. This “analytical
revolution,” which significantly shifted the work of national security intelligence
analysts, also reshaped the forms of knowledge being produced as part of
SIGNIT.17 What did these new analytic practices do to shift the expectations of
analysts? How did they change the work done within CSE?
The NAM initiative started in late 2012, though as of 2014 “the initiative
continue[d] to be informed by changes in ... priorities, client requirements,
access, operations and technology environment.”18 The initiative’s stated goals
are “to use the model to develop an environment to fulfill CSE’s mandate
through cutting edge analysis,” which would “ensur[e] an ability to detect,
prevent, predict and respond to events crucial to Canada’s security and sover-
eignty at scales consistent with our evolving data collection.”19 In envisioning
what the NAM would look like in practice, its designers established “the solu-
tion aim” as being able to “ensure the creation of an agile, scalable and sustain-
able dynamic environment that can be easily stood up, torn down, shared,
[and] modified.”20
NAM policy was designed to achieve three central goals: (1) to “take advantage
of the increasing volume of available data”; (2) to “facilitate the growth of analytic
capacity”; and (3) to “better enable the full spectrum of analysis underway across
the organization.”21 In establishing how intelligence work would change under
the NAM, internal documents identify a required shift in technological and
methodological approaches at CSE. At a technological level, work would be
reconceptualized into an “Infrastructure Layer, Data Management Layer, Ana-
lytics and Service Layer, and Human Insights and Visualization Layer,” while
methodologically, the shift would reorganize SIGINT knowledge production
into “interactive cycles of data gathering and examination through which mean-
ing [would be] derived and questions answered, culminating in higher order
understanding.”22 For analysts, this meant an emphasis on big data analytics
 Pixies, Pop-Out Intelligence, and Sandbox Play 117

and the adoption of machine learning initiatives into their development of

actionable intelligence.23
Internally, CSE defines big data as “the amount of data that exceeds the pro-
cessing capacity and analytic capabilities of conventional systems,” while char-
acterizing the classification of big data through the dimensions of the four Vs:
Volume, or “quantities of collected data”; Variety, “encompass[ing] the many
different data types and sources”; Velocity, or “the speed at which the data is
created, collected and processed”; and finally, Veracity, “the level of reliability
associated with a given type of data” – this final dimension being noted by CSE
as the “least researched” aspect of big data.24 As part of the NAM, big data was
understood as providing “a new way of asking questions” and “organizing data,”
as well as providing “new tools to use” for analysts; more importantly, big data
would initiate “a different way of thinking, since you learn to understand from
summaries of data rather than individual items.”25 That is, “intelligence may
‘pop out’ of exploratory analysis, rather than being prompted by specific analytics
questions.”26 In very plain language, big data analytics would work by enabling
analysts to “understand what is ‘normal’ for [their] data set (which is some
information in itself),” and “what is ‘abnormal’ (doesn’t fall within the range of
what is normal).”27
In this way, the NAM would shift the work of analysts into a type of virtual
data sandbox, through which they would build, test, assess, and engage with
big data and machine learning tools in order to develop actionable SIGNIT
knowledge. As a 2012 internal document that explains this data-sandbox-type
research environment notes, CSE under the NAM

can be conceptualized as a virtual world for exploration and development by

analysts. It’s not unlike one of the more popular online games right now. In
MINECRAFT, users have their own universe to explore, develop and survive
in. Players start off with a completely blank slate. They are free to explore, mine
for minerals, build their own tools, buildings, cities, sculptures etc. The environ-
ment is contained, operates according to the rules of physics etc. and players are
free to do whatever they want within those bounds, without fear of breaking the
system.28

In addition to opening up frictionless interactions with data, the quick assembly

and disassembly of teams, fluid organizational nature at CSE, and incorporation
of elements of play associated with the new analyst of the NAM were each built
directly into the new architecture of CSE’s new Cdn$1.2 billion dollar headquar-
ters, the Edward Drake Building.29 The building was specifically designed to
“promote efficient workflow while being adaptable to enable a sense of
118 Scott Thompson and David Lyon

collaboration so communities of interest can flourish without impacting on

overall coherence.”30 In a practical sense, this meant modular workspace designs,
the inclusion of special-purpose spaces, and a capacity to reorient how work-
spaces are laid out within the building.

Big Data and Machine Learning at CSE, 2012–14

The New Analytic Model was formally adopted at CSE in late 2012 and was to
be implemented incrementally in a series of stages – from the internal identi-
fication of educational needs of CSE personnel to the application of big data
and machine learning analytical practices in operational settings. This section
reviews the phases in which the NAM was implemented and identifies within
CSE internal documents what can be known about the current use of big data
and machine learning in national security work at CSE.
By 2014, internal documents had already asserted the centrality of big data
within CSE’s national security intelligence work. As noted in a presentation to
other Canadian government departments on big data, “CSE can be seen as one
of the biggest consumers of ‘big data’ in Canada,” and “‘Big Data’ is critical to
everyday CSE operations,” from how “CSE collects, manages, samples and
analyses [REDACTED] volume of data on a regular basis,” to how, by 2014, “this
data [wa]s the primary feed for CSE’s reporting on intelligence and cyber secur-
ity.”31 When approached by other Canadian government departments interested
in adopting big data practices in 2014, CSE presented itself as an expert organ-
ization, explaining that it “has significant experience in the use of Big Data and
Machine Learning to fulfill its operational requirements” and that “there is the
potential for CSE to leverage Big Data beyond operations to other areas such
as policy and reporting.”32 Although precise operational details regarding the
current use of big data at CSE remain redacted in ATIP release documents,
overarching NAM policy, CSE presentations, reports to other departments, and
public leaks do provide some details on how big data and machine learning
work is being done at CSE.
At a more general level, CSE’s work regarding big data and machine learning
under the NAM policy initiative does retain a central role for the specialized
knowledge of analysts and stresses the need for human actors within data-driven
knowledge production. In particular, internal documents explain that “it is
important for subject-matter experts to be involved with the entire process and
interpretation as much as possible,” in order “to make sure the questions are
meaningful.”33 Interestingly, the analyst within the NAM is also to play a role in
the identification and selection of relevant data, as “using ‘convenient’ data, may
actually lead you astray and to incorrect conclusions,” while analysts are also
to manually review outputs since datasets “are not ‘clean’ (decision tree is not
 Pixies, Pop-Out Intelligence, and Sandbox Play 119

perfect; will classify some incorrectly)” – with the NAM approach, the import-
ance of the analyst is highly stressed, as “one needs to start from the types of
questions we want to answer, then obtain the relevant data, even if it seems
expensive to do this at the beginning of the project.”34 Beyond simply involving
the analyst, CSE internal documents on big data and machine learning specific-
ally assert that fully automated approaches are incompatible with effective
knowledge development, noting that

by themselves, data analysis tools will not be useful unless there are analysts
who understand both the tools and the data ... Instead the tools will be part of a
process, which includes people analyzing the results produced by the tools and
using their own knowledge of the business and data to draw conclusions. This
analysis will include using good data visualization and other techniques to bet-
ter understand the results. Usually those conclusions will be hypotheses which
will need to be further tested in an ongoing process where the users continu-
ously refine the results. In CSE’s experience, pursuing the “Star Trek” vision has
consistently led to a dead end.35

Consequences and Critique

Having established that, in their own words, “everything has changed” in terms
of data analysis at CSE, it is important that the consequences of this shift be
exposed and examined. Given the unlikelihood of obtaining any kind of account
of the ongoing critical assessment of the New Analytic Method at CSE, we turn
to the kinds of likely outcomes at CSE and the questions that are prompted by
the use of big data practices in other fields. The language used in the documents
obtained makes clear that (1) the shift to new methods is decisive and far-
reaching; (2) its catalyst comprises developments taking place among partner
organizations, especially with the NSA and Five Eyes, which in turn is predicated
on (3) the veritable explosion of data availability from the exponential expansion
of digital communications in the early twenty-first century. The continuity
suggested by naming the new CSE building for Edward Drake, pioneer of the
Canadian signals intelligence community from 1945, is ironically belied by the
wholesale shift to methods that abandon Drake’s approach.
The widespread public realization that security agencies rely extensively on
big data occurred in 2013 with the publication of the first documents copied
by Edward Snowden, subcontractor to the NSA, with Booz Allen Hamilton.36
But had that public read James Bamford’s account of the NSA’s Utah Data
Center in 2012, this was already evident. One of the key purposes of the Utah
Data Center is cryptanalysis, for breaking complex encryption. Bamford
explained its effects as “everybody is a target; everyone with communication
120 Scott Thompson and David Lyon

is a target.”37 In Bamford’s account, having been caught off guard by a series of

terrorist attacks culminating in 9/11, the NSA was “quietly reborn” and the mas-
sive new data centre in Bluffdale, Utah, was the symbol of that rebirth.38
At Bluffdale and CSE, in intelligence production, big data is named and valid-
ated as the New Analytic Model in conjunction with machine learning and new
forms of reliance on statistical techniques. It is important to recall, with danah
boyd and Kate Crawford, that “Big Data is less about data that is big than it is
about a capacity to search, aggregate, and cross-reference large data sets.”39 It is
precisely that capacity that CSE was establishing with its shift to NAM. As boyd
and Crawford also argue, big data is a combination of technology, for gathering,
analyzing, and linking datasets; analysis, drawing on datasets to identify pat-
terns, such that new claims can be made; and mythology, that large datasets offer
a higher form of knowledge, accurately generating insights that were previously
impossible.40 As noted above, CSE compares the NAM shift to a “scientific
revolution.” In this case, NAM challenges the classic deductive scientific method
that underlay the prior CSE approach.41
As a result, new types of knowledge are created from the data-driven approach,
the likely direction of which is from a targeted causal logic to a bulk, correlational
logic. The quest is for patterns, insights, and, as shown in CSE documents,
intelligence that “pops out” of exploratory analysis. This is in effect a new form
of empiricism that downplays theory and claims that meaningful knowledge
may be produced from patterns in so-called raw data.42 In this view, one no
longer needs to test hypotheses, because the correlations generated by new
computational, machine learning, algorithmic, and statistical methods produce
superior knowledge when datasets are mined for patterns. From there, without
human framing one may move forward to prediction, rather than backward to
explanation. Anyone with some statistical sense or the means to visualize data
can interpret the patterns.
The problems associated with this approach are manifold, including the
following:43

• The data will always remain no more than a sample, taken from a certain van-
tage point.
• The newly developed systems are designed to capture certain kinds of data, and
the analytics and algorithms must already have been scientifically tested in spe-
cific ways.
• Data never speak for themselves without being framed in some way and are not
inherently meaningful – the correlations may well be random.44
• Information that “pops out” of the data is inadequate insofar as it lacks contex-
tual and domain-specific knowledge.
 Pixies, Pop-Out Intelligence, and Sandbox Play 121

This suggests that the hesitations hinted at in the ATIP documents are appropri-
ate. Deep knowledge of prior modes of intelligence gathering would appear to
be vital for correctly interpreting data generated by the NAM.
Surveillance is always and everywhere a means of making visible certain
individuals or groups such that, in the case of national security intelligence-
gathering operations, they may be assessed for their possible connections with
criminal, violent, or terrorist activities. The shift to the NAM at CSE means that
new kinds of visibility emerge, different from older targeted methods and
involving “bulk” data. This is where the often missing “V” of big data – vulner-
ability – becomes acutely significant.45 Evidence from many other areas of big
data practice suggests that unless extreme care is taken, big data practices can
make some already marginalized or disadvantaged groups more vulnerable
than others.46
In the organization of intelligence work, the shift to big data practices may
be seen in the use of new gaming metaphors, new partnerships, and, significantly,
new concerns about whether the NAM will completely eclipse older practices.
Again, these kinds of shifts are not uncommon in big data practices, especially
in their corporate settings. Gaming is seen as a way of organizing teams of
workers and as a model of how progress is made in the search for actionable
data. This in turn plays into the kinds of partnerships evident among a variety
of players, especially among those who on their own would lack the resources
to mount big data centres of analysis on their own account. Universities and
technology companies frequently form such alliances in order to share ideas
and expertise. Lastly, while there is much hype surrounding big data practices,
some sober assessments acknowledge the need to combine older with newer
methods in order to make reliable claims about what the data indicate.
The gaming metaphors are significant. Those highlighted in the ATIP docu-
ments include the “sandbox” and the “Minecraft” models. Based originally
on a children’s exploratory playspace, the idea of a sandbox in software
development is for production development, for learning, testing, and experi-
menting. But the sandbox also hints at limits, the ways in which certain kinds
of experimentation may be constrained to protect items such as data, code,
and servers from potential damage resulting from changes to the system. So
while play is present, in this case it is clearly purposeful play that is in view.
The same is true of the Minecraft analogy, where, in the SIGINT lab, “starting
with a blank slate,” analysts have opportunities to “experiment, explore, create
and innovate in this universe.” The documents state, curiously, that the analysts
are bound “by the rules of physics” (really, no others?) and that within the
“Minecraft” lab they are free to innovate at will without fear of doing
damage.
122 Scott Thompson and David Lyon

Another crucial factor is the public-private partnerships that are evident in

the very building now housing CSE and the academic-corporate connections
visible in the relationship with the Tutte Institute. The latter has the focus of
“research in fundamental mathematics and computer science” and is “continu-
ally working to strengthen and develop partnerships with research institutes,
governmental agencies, industry and universities.”47 It works with the Institute
for Defense Analyses in the United States, associated with universities including
Princeton, and the Heilbronn Institute for Mathematical Research in the United
Kingdom. It is formally partnered with the University of Calgary and with
Carleton University in Ottawa. The participants in the Tutte programs are from
the fields of mathematics, computing science, and engineering. There is no
mention of whether these persons or partners involve legal scholars or persons
from the social sciences, such as political science or international relations, or
from the humanities, such as history.
All the emphasis at present appears to be on the mathematics, computing,
and engineering aspects of surveillance, intelligence gathering, and analysis.
For CSE, a shift of this magnitude means that new analysts must be found and
trained to read the outputs of big data and machine learning. Some residual
questions need to be addressed about whether older forms of expertise will be
lost, and if so, whether or not this loss will reduce the capacity of CSE to do its
work appropriately and efficiently. This does not seem to be a key priority of
CSE and the NAM, however. While CSE says that they resist a fully automated
system, which is why they want analysts to direct aspects of the process of intel-
ligence gathering and analysis, the analysts appear to be drawn almost exclusively
from disciplines that are unlikely to include expertise relating to classic intel-
ligence gathering for national security purposes.

Conclusion
From the evidence presented here, everything has changed at CSE in the wake
of decisions to switch to big data practices, known at CSE as the New Analytic
Model. The aim is to improve national security in an age of exploding com-
munications media. While some aims of the NAM are worthy ones, given the
urgency of dealing with global crime, violence, and terrorism, the methods and
modes of expertise chosen are very heavily weighted towards mathematics,
computing, and engineering. While this is appropriate for grappling with the
immense quantity of data available, it is not clear that sufficient attention is
being paid to the quality of the intelligence thus gleaned. The problem with
abandoning old methods in favour of new is that the risk of threatening situa-
tions may be misconstrued, with negative consequences for individuals and
groups. In CSE’s zeal to avoid human framing and “bias,” blinkers appear around
 Pixies, Pop-Out Intelligence, and Sandbox Play 123

the very nature of the data and the inevitability that they are “framed” from the
outset.
Along with this – and possibly CSE is aware of the problem, however dimly –
is the ironic fact that without human involvement in the process of data analysis, the
likelihood of successful utilization of big data for security intelligence and sur-
veillance will remain slim. The focus on algorithms and machines precisely takes
attention away from the crucial matters of context. As a 2017 article states, “these
tools cannot replace the central role of humans and their ability to contextualize
security threats. The fundamental value of big data lies in humans’ ability to under-
stand its power and mitigate its limits.”48 And, we might add, not just any humans
will do. While those trained in mathematics, computing, and engineering are essential
to the task, if these skills are not complemented with ones from, for example, law,
the social sciences, and humanities, the essential task of contextualizing analysis not
only will be poorer but could raise human rights, civil liberties, and privacy problems
for vulnerable individuals, for whole classes of persons, and indeed for everyone.

Notes
1 Communications Security Establishment (CSE), “Opportunities for Professionals –
Foreign Language Intelligence Analysts,” 1 September 2017, https://web.archive.org/web/
20170825040050/https://cse-cst.gc.ca/en/node/1402.
2 Under Canada’s federal Access to Information Act or Privacy Act, individuals (either citi-
zens, permanent residents, or those currently present in Canada) and corporations are
able to make requests to federal government departments and organizations to obtain
documents relating to a given subject (the provinces also have similar legislation to cover
their departments and organizations). The acts do place some limitations regarding what
kinds of information can be released, and it is at times challenging to obtain documents
on subjects like national security or policing. Requests need to be written in such a way
as to avoid asking for current practices or ongoing investigations; they also need to be
made with the proper language or keywords of the institution. The language for the
request upon which much of this chapter is based (#A-2016-00068), for example, was
for all “high level briefing notes, presentations, policy framework documents or reports
specifically related to CSE’s definition of ‘Big Data,’ ‘Knowledge Discovery,’ and/or ‘Data
Mining’ and their impact on the mission.” The collected volume by Jamie Brownlee and
Kevin Walby, Access to Information and Social Justice: Critical Research Strategies for
Journalists, Scholars, and Activists (Winnipeg: ARP Books, 2015), is an excellent start-
ing point for those looking to better understand the use of Access to Information and
Privacy (ATIP) legislation for research, journalism, or social activism in Canada. ATIP
requests can be made online (https://atip-aiprp.apps.gc.ca/atip/welcome.do), and the
preceding two years of completed requests are also catalogued. The Surveillance Stud-
ies Centre, along with Queen’s University, is currently working to develop a repository
where the full texts of completed ATIP requests will be made available to researchers and
the general public, including those used in this chapter.
3 CSE, “NAM Episode II: Rise of the New Analytical Environment, CSE PowerPoint Pre-
sentation Deck,” 193, released under Access to Information Request Number A-2016-
00068 (2016).
 124 Scott Thompson and David Lyon

4 Carrie Sanders and James Sheptycki, “Policing, Crime and ‘Big Data’: Towards a Critique
of the Moral Economy of Stochastic Governance,” Crime, Law and Social Change 68, 1–2
(2017): 1–15.
5 David Lyon, “Surveillance, Snowden, and Big Data: Capacities, Consequences, Critique,”
Big Data and Society 1, 2 (2014): 1–13.
6 Craig Forcese and Kent Roach, False Security: The Radicalization of Canadian Anti-
Terrorism (Toronto: Irwin Law, 2016).
7 Colin Freeze, “Canadian Spy Manual Reveals How New Recruits Are Supposed to Con-
ceal Their Identities,” Globe and Mail, 22 December 2013.
8 CSE, “NAM Episode II,” 193.
9 CSE, “[REDACTED TITLE], CSE PowerPoint Presentation Deck,” 175, released under
Access to Information Request Number A-2016-00068 (2016); Communications Secu-
rity Establishment, “NAM Episode II,” 193.
10 CSE, “[REDACTED TITLE],” 184.
11 CSE, “Analytic Environment for DGI: Business Case Proposal – Draft,” 98, released
under Access to Information Request Number A-2016-00068 (2016).
12 CSE, “NAM Episode II,” 191.
13 CSE, “Analytic Environment for DGI,” 100–1.
14 CSE, “NAM Episode II,” 193.
15 CSE, “[REDACTED TITLE],” 174.
16 CSE, “NAM Episode II,” 205.
17 CSE, “[REDACTED TITLE],” 189.
18 CSE, “Analysis and Production Evolution: A Simplified Take on the NAM, CSE Power-
Point Presentation Deck,” 162, released under Access to Information Request Number
A-2016-00068 (2016).
19 Ibid.; CSE, “Analytic Environment for DGI,” 96.
20 Ibid., 100.
21 CSE, “NAM Episode II,” 205.
22 Ibid.
23 CSE, “[REDACTED TITLE],” 189.
24 CSE, “Analytic Environment for DGI,” 97.
25 CSE, “Machine Learning, CSE PowerPoint Presentation Deck,” 142, released under
Access to Information Request Number A-2016-00068 (2016).
26 Ibid.
27 Ibid.
28 CSE, “NAM Episode II,” 197–98.
29 Greg Weston, “Inside Canada’s Top-Secret Billion-Dollar Spy Palace,” CBC News, 8 Octo-
ber 2013, http://www.cbc.ca/news/politics/inside-canada-s-top-secret-billion-dollar-spy
-palace-1.1930322.
30 Canadian Council for Public-Private Partnerships, The Canadian Council for Public-
Private Partnerships 2011 National Award Case Study Silver Award for Project Financ-
ing: Communication Security Establishment Canada Long-Term Accommodation Project
(Ottawa: Canadian Council for Public-Private Partnerships, 2011), 8.
31 CSE, “Big Data for Policy Development 2014, CSE PowerPoint Presentation Deck,” 30,
released under Access to Information Request Number A-2016-00068 (2016); CSE, “Big
Data Discussion Paper,” 4–5, released under Access to Information Request Number
A-2016-00068 (2016).
32 CSE, “Big Data Discussion Paper.”
33 CSE, “Machine Learning, CSE PowerPoint Presentation Deck,” 142, released under
Access to Information Request Number A-2016-00068 (2016).
34 Ibid., 129, 142.
 Pixies, Pop-Out Intelligence, and Sandbox Play 125

35 CSE, “Big Data Discussion Paper,” 6–7.

36 David Lyon, Surveillance after Snowden (Cambridge: Polity, 2014).
37 James Bamford, “The NSA Is Building the Country’s Biggest Spy Center,” Wired, 15
March 2012, https://www.wired.com/2012/03/ff_nsadatacenter/.
38 Ibid.
39 danah boyd and Kate Crawford, “Critical Questions for Big Data,” Information, Com-
munication and Society 15, 5 (2012): 663.
40 Ibid.
41 Rob Kitchin, “Big Data, New Epistemologies and Paradigm Shifts,” Big Data and Society
1, 1 (2014): 1–12.
42 Geoff Bowker, Memory Practices in the Sciences (Cambridge, MA: MIT Press, 2005); Lisa
Gitelman, ed., Raw Data Is an Oxymoron (Cambridge, MA: MIT Press, 2013).
43 Kitchin, “Big Data, New Epistemologies,” 1–12.
44 One sees this even in non–big data contexts, such as the National Security actions taken
against Ahmad Abou El-Maati, based on, inter alia, a map of government buildings in
Ottawa. He was tortured in Syria as a result, but the simple explanation of the correla-
tions between his driving a truck in Ottawa and a map showing government buildings
was that he was on a delivery run for a company.
45 David Lyon, “The Missing ‘V’ of Big Data: Surveillance and Vulnerability,” in Reading
Sociology: Canadian Perspectives, edited by Patrizia Albanese, Lorne Tepperman, and
Emily Alexander (Toronto: Oxford University Press, 2017), 324–27.
46 Neil Richards and Jonathan King, “Three Paradoxes of Big Data,” Stanford Law Review
41 (2013): 41–46; Oscar Gandy, Coming to Terms with Chance: Engaging Rational Dis-
crimination and Cumulative Disadvantage (Farnham, UK: Ashgate, 2009).
47 CSE, “Tutte Institute for Mathematics and Computing – About the Institute,” 8 May
2020, https://www.cse-cst.gc.ca/en/tutte-institute.
48 Damien Van Puyvelde, Stephen Coulthart, and M. Shahriar, “Beyond the Buzzword: Big
Data and National Security Decision-Making,” International Affairs 93, 6 (2017): 1397–
1416, https://academic.oup.com/ia/article-abstract/doi/10.1093/ia/iix184/4111109/
Beyond-the-buzzword-big-data-and-national-security.
7
Limits to Secrecy
What Are the Communications Security Establishment’s Capabilities
for Intercepting Canadians’ Internet Communications?
Andrew Clement

The largest, most significant big data surveillance operations are arguably
the global Internet interception and analysis activities of the Five Eyes security
alliance. Consisting of the US National Security Agency (NSA) and its signals
intelligence partners in the United Kingdom, Canada, Australia, and New
Zealand, this alliance is capable of intercepting and analyzing much of the
world’s Internet communications as they flow through major switching centres.
While the public has learned a great deal from whistle-blowers such as Edward
Snowden about how the American and British agencies conduct surveillance
within their domestic networks, Canadians know relatively little about similar
operations by Canada’s own Communications Security Establishment (CSE).
Even at the best of times, secretive security agencies like CSE pose an inherent
dilemma for liberal democracies. Public transparency and accountability of
state institutions are fundamental tenets of democratic governance. This inevit-
ably creates tension with the secrecy that such agencies require to be effective
in their missions around national security. Central to achieving national security
is upholding the democratic rights upon which the integrity of the nation
ultimately depends. If citizens fear that their personal communications may be
intercepted unjustifiably and feel at risk of being treated more as threats than
rights holders, they are more likely to withdraw from the public sphere, with-
hold support for government initiatives, or even subvert them. Pursuing total
secrecy is not a viable long-term option. Security agencies need to be sufficiently
open about their activities to demonstrate that they respect the privacy, freedom
of expression, and other rights of individuals. With an implicit duty of public
candour, the onus is on them to hold themselves accountable and thereby earn
public trust. Achieving an appropriate balance of secrecy and transparency is
thus difficult and contingent.
Being publicly transparent is especially important when there are clear
indications that an agency has violated public trust. The Snowden revelations
of 2013 together with the recently legislated expansion of CSE’s mandates
make questioning CSE’s highly secretive posture as well as its dubious sur-
veillance practices particularly urgent. Yet when leaked documents have
indicated that CSE is conducting domestic Internet surveillance comparable to
 Limits to Secrecy 127

that of its Five Eyes partners, its official responses generally go no further than
bland assertions of legal compliance:

CSE is prohibited by law from directing its activities at Canadians anywhere or

at anyone in Canada.1

Our activities are guided by a robust framework of Ministerial Directives and oper-
ational policies. CSE’s activities, as well as its operational directives, policies and
procedures, are reviewed by the CSE Commissioner, to ensure they are lawful.2

In the context of its partner agencies caught lying publicly about their domestic
surveillance activities, stretching legal definitions and mandates far beyond
conventional interpretations and engaging in activities that in Canada could
arguably be considered unconstitutional, CSE’s vague statements provide little
reassurance. Without being more open and offering persuasive details, CSE
invites questions about its integrity as well as whether existing laws and other
regulatory measures are sufficiently robust to keep such a powerful agency
within established democratic norms.
Canadians have shown a keen interest in issues of privacy, freedom of expres-
sion, and other democratic rights in relation to domestic surveillance, law
enforcement, and national security. Recent legislative initiatives around lawful
access (2011) and anti-terrorism (2015) triggered politically potent controversies.
In 2017, Bill C-59 (An Act respecting national security matters)3 raised new con-
cerns and reinvigorated the public debate. At that time CSE exercised secret
powers based on three broad mandates:
A. to acquire and use information from the global information infrastructure
for the purpose of providing foreign intelligence, in accordance with Govern-
ment of Canada intelligence priorities;
B. to provide advice, guidance and services to help ensure the protection of elec-
tronic information and of information infrastructures of importance to the
Government of Canada;
C. to provide technical and operational assistance to federal law enforcement
and security agencies in the performance of their lawful duties.4

Bill C-59 expanded these already formidable powers, adding two mandates that
give CSE brand-new offensive and defensive cyber authorities. It also created
the National Security and Intelligence Review Agency (NSIRA) and the Intel-
ligence Commissioner. With these novel review and oversight bodies, passage
of Bill C-59 in 2019 far from ended the national security debate but shifted it to
a new, promising phase.
128 Andrew Clement

Much of the debate reasonably revolves around legal principles, democratic

rights, and governance mechanisms. But in the absence of concrete insights
into how they apply in practice to CSE operations, the debate remains relatively
abstract. It is very difficult for people to relate CSE activities to their lived experi-
ence as individuals within a national community without knowing more about
whether CSE may be intercepting their communications, and if so how, where,
when, and with what legitimate justification. With CSE insistently shrouding
every feature of its operations in obscurity, public dialogue regarding what it
actually does and how its activities are relevant to personal Internet communi-
cations has been largely non-existent.
More specifically, CSE’s meagre provision of public information about its
surveillance techniques and practices hinders consideration of such essential
policy questions as:
• Is mass surveillance of domestic electronic communication an appropriate
means of ensuring national security and is it in the public interest?
• What is a reasonable scope for CSE secrecy?
• Where secrecy and surveillance may be justifiable, what mandates and oversight
measures can ensure that authorized agencies operate this surveillance infra-
structure within the civil liberties and governance norms fundamental to the
integrity of Canada’s “free and democratic society”?

For insight into these questions, and more generally in order for Canadians to
participate meaningfully in the national security debate, a deeper understanding
of CSE’s practical capabilities is needed.
This chapter seeks to equip Canadians more fully for this ongoing debate.5
The broad aim is to articulate “capability transparency” as a principle that
security agencies such as CSE should be held to as a core feature of their
democratic accountability. It makes the case that CSE has the capability to
intercept Canadians’ domestic Internet communications in bulk. It does not
seek to establish this claim “beyond a reasonable doubt,” as might be expected
of conventional scholarly writing or in a criminal prosecution. Given the
current degree of secrecy and obfuscation, this claim cannot be settled one
way or another solely with publicly available information. Rather, the immedi-
ate goal is more modest – to reach the lower standard of “reasonable suspi-
cion.” This means that where evidence of mass surveillance is ambiguous,
the benefit of the doubt does not go to CSE but instead can add weight to
reasonable suspicion. In light of the discussion above about secretive security
agencies operating within democratic norms, this mode of argumentation
should be sufficient to place the burden of proof squarely on CSE, which
should provide clear public evidence that either contradicts the claims here
 Limits to Secrecy 129

or else acknowledges their validity and justifies secretly developing such an

interception capability.
The chapter begins by closely examining several classified CSE documents
that the Canadian Broadcasting Corporation (CBC) published based on those
that Snowden released to journalists. Together these documents strongly suggest
that CSE has sought and at times deployed the capability to conduct mass sur-
veillance of Canadians, and that it is hiding this fact from the public. To
strengthen these findings and render them more concretely relatable to everyday
Internet practices, a growing body of Canadian Internet routing data is examined
to identify the cities and telecom carriers through which CSE would most likely
conduct its interception operations. Given the highly concentrated nature of
both Internet switching facilities and service providers, it is unsurprising that
the routing data show that it would be feasible for CSE to capture a large pro-
portion of Canadian communications with relatively few interception points,
accessed through partnerships with just a handful of Internet service providers
(ISPs).
The chapter concludes by challenging CSE’s current practice of total secrecy
about its domestic interception capabilities. As an alternative, it proposes a basic
“capability transparency” approach. To become more accountable and earn the
trust of Canadians, CSE should adopt an openness policy that appropriately
and demonstrably balances legitimate secrecy needs with disclosure of basic
surveillance capabilities.

Does CSE Intercept Canadians’ Internet Communications in Bulk?

While the Snowden documents offer significant public insight into the domestic
mass surveillance operations of the NSA and of the Government Communications
Headquarters (GCHQ) in the United Kingdom, comparatively little is known
about similar activities of CSE. Fewer than forty of the hundreds of Snowden
documents published to date relate directly to CSE. However, a careful reading
of these documents, in combination with the more abundant materials about
its Five Eyes partners and a basic understanding of Internet routing technique,
offers sufficient basis for the reasonable suspicion that CSE is routinely inter-
cepting Canadians’ domestic Internet communications in bulk.

The “Airport Wi-Fi” Story – No Tracking Canadians

at Airports, Just “Normal Global Collection”
Among the media exposés of CSE surveillance, the CBC’s 2014 report “CSEC
Used Airport Wi-Fi to Track Canadian Travellers: Edward Snowden Documents”
sparked the greatest controversy.6 The classified documents behind this story
are also the most revealing of the reach and intensity of the domestic Internet
130 Andrew Clement

interception capabilities that CSE had achieved by 2012. The CBC’s central claim,
that the CSE could track individual Canadians, backward and forward in time,
who accessed the Internet via Wi-Fi hotpots as they passed through airports,
hotels, libraries, and other popular locations, was based on a CSE presentation
describing a trial program in which CSE took up the challenge “to develop [a]
new needle-in-a-haystack analytic.”7 This involved linking travellers’ user IDs
with the IP addresses locally assigned to their devices. However, this was done
not by intercepting Wi-Fi signals, as the article ambiguously suggests, but
through the untargeted (i.e., bulk) interception of all Internet traffic passing
through the switching centres of major ISPs. Only a small fraction of this traffic
would include communications originating at public Wi-Fi spots, such as at an
airport. To extract the communications of individuals using the airport Wi-Fi,
analysts looked for originating IP addresses that showed specific characteristic
usage patterns. From this, they developed a signature profile for various kinds
of public Wi-Fi hotspots and inspected all the user IDs of individuals who
moved between them.
To understand what data CSE had access to from its interception operations
and how it could track individuals requires a closer look at the technical char-
acteristics of Internet communications. An IP address, such as for the router at
a Wi-Fi gateway, is included in the header of every packet originating from or
destined for that router. These IP addresses used for routing fit the conventional
definition of communication metadata, that is, of the “shallow” kind – “infor-
mation used by computer systems to identify, describe, manage or route com-
munications across networks.”8 However, the user IDs that CSE needed for
tracking individuals from one site to another are not found in packet headers
used for routing but only within the packet “payload.”
The CSE presentation does not mention what was used as the basis for user
IDs, but almost certainly it included some combination of individualized online
account names (e.g., for accessing Facebook, Yahoo, or Gmail) and “cookies.”
To reliably extract these IDs from Internet communication collected in bulk,
as would have been needed for the experiment described in the presentation,
CSE would have had to use deep packet inspection (DPI) facilities to reassemble
packets into the original full messages for analysis. This is because an account
name or cookie may be broken across more than one packet. In other words,
to achieve this fine degree of granularity, CSE must have access to message
content, not just the metadata needed for routing.
Once user IDs have been extracted from intercepted communications, track-
ing them across the various sites from which they access the Internet, and
especially back in time before becoming an ID of interest, involves a further
impressive but ominous computational feat. Foremost, it requires the ability to
 Limits to Secrecy 131

intercept all the Internet communications in a target region, referred to as “full-

take” capture. While much traffic, such as Netflix streaming video content, needs
to be filtered out immediately to avoid overloading storage facilities, it still
generates vast quantities of (meta)data to be sifted through. Furthermore, this
form of tracking cannot simply be conducted in real time, just by inspecting
data as they flow past, but requires storage of some duration and retrospective
analysis of extraordinarily large volumes of intercepted communications. And
this comprehensive Internet interception was not confined to the Canadian city
with the international airport. For its proof of concept test, the CSE team also
“swept a modest size city ... over 2 weeks.”9
CSE helped confirm this interpretation of its capabilities. Within days of the
CBC story, the Senate  Standing Committee on National Security and
Defence  invited then CSE head John Forster, along with other government
officials, to appear before it. Especially revealing is this statement in Forster’s
testimony: “This [CSE surveillance] exercise involved a snapshot of historic
metadata collected from the global internet. There was no data collected through
any monitoring of the operations of any airport. Just a part of our normal global
collection” (emphasis added).10 No doubt intended to reassure Canadians that
they were not tracked individually at airports, this statement should heighten
concerns about the extent of domestic Internet surveillance capabilities. Here
Forster confirmed that CSE’s analytic experiment did not require any new data
collection. It wasn’t just data from specific airports, Wi-Fi hotspots, or one
“modest size city” covering a period of a few weeks, but “normal,” “historic”
data collection, presumably captured in pursuit of some combination of CSE’s
three mandates prior to 2019 mentioned earlier. The repeated use of the term
“global” gives the impression that the collection was conducted outside Canada,
but this of course doesn’t exclude portions of the Internet on Canadian soil.
Indeed, as subsequent releases of Snowden documents reveal, given the need
for signals intelligence on threatening foreign agents that may be within Canada
(Mandate A),11 for IT security (Mandate B), or to provide assistance to internally
focused federal law enforcement (e.g., RCMP) and security agencies (e.g., CSIS)
(Mandate C), CSE’s interception capabilities are very likely to be well represented
inside Canada and cover several cities.

The “Cyberwarfare” Story – Integrating Domestic

Interception Capabilities
Just over a year after the airport Wi-Fi story, the CBC published a series of
further stories based on leaked CSE documents that provided a deeper insight
into its domestic interception capabilities. Two top-secret PowerPoint presenta-
tions in particular revealed that by 2011 CSE operated or planned to operate
132 Andrew Clement

several Internet interception programs or “sensors,” including within Canada –

Photonic Prism and CASCADE, the cover name for the EONBLUE,
INDUCTION, THIRD-EYE, and CRUCIBLE programs.12 EONBLUE is the
most prominent and relevant of these. Both it and INDUCTION, its multi-line
version, are capable of “full-take” collection of Internet communications in
service of CSE’s three broad mandates. Typically located on the premises of
telecommunications companies, EONBLUE’s and INDUCTION’s deep packet
inspection devices process data streams at 10 gigabits per second for each tapped
line. Sensors in Canada accumulate in their repositories up to 300 terabytes (3
million gigabits) of full-take then filtered interception, “equivalent to ‘months’
of traffic.” In addition, EONBLUE sensors forward both metadata and content
to CSE central facilities for further analysis and storage, such as that required
in the airport Wi-Fi example.13
These CSE presentations also point out limitations in the agency’s surveillance
capabilities. The slide deck “CASCADE: Joint Cyber Sensor Architecture” notes
that CSE’s IT Security deployments for protecting Government of Canada
systems were “currently performing full take and storage of all monitored Traf-
fic,” but experiencing “System performance issues.”14 “While both ITS/SIGINT
currently leverage EONBLUE Software ... The architectures are not aligned,”
making it “difficult to manage [the] current sensor environment” and “costly
to grow.” While maintaining the full-take strategy, the solution proposed is to
standardize the various sensors and through interoperability synchronize them
into a “Single Interconnected Sensor Grid” that could serve all three mandates
(see Figure 7.1).
Looking beyond this sensor unification to 2015, the presentation identifies
several strategic priorities. Prominent among them is “Expand Our Access
Footprint ... We will increase SPECIAL SOURCE [i.e., telecommunications
companies] access to include all international gateways accessible from Canada.”
Asserting that protecting gateways, devices, and end nodes is essential but not
sufficient, the presentation calls for embedding sensors throughout the domestic
Internet with the following goals:
• Detect threats as they enter our national networks, not at the Gateway
• Identify Exfiltration, Command and Control, anywhere in our national
networks
• The network is your defence for all infrastructure

For the CSE presenters, this means that “EONBLUE will be integrated into the
Network ... [to enable] Monitoring Core Infrastructure (Special Source)
extending the reach to view national infrastructure.” Under the catch phrase
“The Network is the Sensor,” the presentation anticipates that for domestic
 Limits to Secrecy 133

Figure 7.1  Canadian cyber sensor grid. This is a slide from a top-secret CSE presentation that
shows the location of CSE’s sensors in various parts of the global information infrastructure,
including within the Canadian Internet Space and at entry points into Government of Canada
networks. | Source: “CASCADE: Joint Cyber Sensor Architecture,” https://snowdenarchive.cjfe.
org/greenstone/collect/snowden1/index/assoc/HASH9a6d.dir/doc.pdf.

defence, “the same capabilities [as deployed for foreign signals intelligence
(SIGINT) interception] will be integrated into the CORE of the [Canadian]
Internet.”
Some caveats are in order here about interpreting these CSE documents.
These slide decks are not simply factual reports but are intended in part to
impress colleagues with accomplishments and argue for more resources. And
of course they are intended to accompany a verbal presentation within a wider
security agency milieu to which the public has no access; they assume knowledge
of obscure technical, tradecraft, and codeword terms; and they are several years
old. Capabilities that may have existed at the time may have been discontinued,
and new ones added. Many of the statements in the 2011 and earlier documents
refer to interception capabilities in the planning stage, some of which may
not have come to fruition. However, sensor performance reported in the
134 Andrew Clement

2012 document on which the airport Wi-Fi story was based is consistent with
the operational and projected capabilities mentioned previously, and strongly
suggests that CSE may have accomplished many of its goals around intercepting
Internet communications within Canada.
Given CSE’s clear ambitions, growing funding, and expanded collection dur-
ing this recent period, it is reasonable to suspect that by 2020 CSE had success-
fully developed extensive domestic interception capabilities – that is, it is likely
routinely intercepting within Canada, filtering, storing and analyzing Internet
communications in bulk. To accomplish this, CSE would have needed the co-
operation, whether willing or coerced, of major Canadian telecommunication
providers in installing interception equipment. Furthermore, the data captured
would go well beyond metadata, in the conventional meaning of CSE’s definition
above, and draw on message content. This kind of activity constitutes mass,
population-wide, content-based, suspicion-less surveillance that would intrude
upon the communications of millions of Canadians.

Where Might CSE Intercept Canadians’ Internet Communications,

and with Whose Assistance?
Building on this conclusion, that there are reasonable grounds to suspect that
CSE is intercepting Internet communications in bulk throughout the Canadian
Internet, the analysis turns to estimating the geographic locations of this inter-
ception. It further assesses which telecom service providers CSE would be most
likely to make arrangements with for accessing the Internet infrastructure to
accomplish this bulk interception.
The principal motivation for making these assessments within the broad aims
of this chapter is to provide further support for the central claim – that CSE
indeed has the capability to conduct such interception – by demonstrating that
CSE can feasibly intercept a large fraction of Canadian Internet communications
with relatively few interception points and partners. More provocatively, open-
ing up the technical and institutional aspects of Internet surveillance to public
scrutiny is an attempt to illustrate a form of capability transparency that CSE
should consider adopting when reforming its total secrecy regime in the interests
of becoming more accountable and trusted.

Interception at Canada’s Border Cities

This investigation begins by asking where CSE would most likely place sensors
to fulfill its missions to provide foreign intelligence (Mandate A) and protect
national IT systems (Mandate B).15 Most obviously, this means intercepting
Internet traffic entering Canada from foreign countries and analyzing it for
potential threats. Based on what the Snowden documents have revealed about
 Limits to Secrecy 135

NSA and GCHQ surveillance operations, this interception is likely not done
right at the border but at the first Internet exchange where the cross-border
traffic is next routed to its various destinations.
An obvious way to identify the cross-border switching centres in Canada is
to examine the route maps that various major telecom companies provide
publicly for promoting their Internet businesses. These maps paint a consistent
picture of the main fibre routes that cross the border and where they connect
with the Canadian Internet backbone. Collating the information from the route
maps of major Canadian carriers as well as large foreign carriers providing
transit services in the Canadian market indicates that nine cities host nearly all
the cross-border Internet connections: Victoria, Vancouver, Calgary, Winnipeg,
Windsor, Hamilton, Toronto, Montreal, and Halifax/Dartmouth.
It would be reasonable to expect that CSE would seek to install interception
devices at the main Internet switching centres in each of these cities, and thereby
capture close to 100 percent of inbound Internet traffic. However, especially if
resources are limited, some cities will be prioritized over others, depending on
their relative volumes of traffic and which telecom providers will need to be
enrolled. For this it is helpful to analyze the actual routes that data follow in
practice. In the absence of public reporting of such detailed Internet traffic
statistics, the IXmaps Internet mapping and analysis tool can provide useful
insights.

IXmaps – An Internet Transparency Tool

As part of broader Internet transparency initiatives, the IXmaps project seeks
to promote greater public understanding of Internet operations and related
governance issues.16 In particular, the IXmaps Internet mapping platform pro-
vides lay users with a means of probing hitherto obscure technical and institu-
tional aspects of Internet routing, notably potential sites of mass surveillance.
In order to produce Internet data that reflect individual users’ online activity
and preferences, IXmaps takes a crowd-sourced approach to generating the
traceroute data that it requires for mapping Internet routes. Traceroutes measure
the paths data take across the Internet as a series of hops from one router to the
next. IXmaps enables Internet users to contribute to a shared database the routes
their data take when they visit websites of their choice. Based on estimates of
the geographic location of the routers involved, users can then selectively map
their own or others’ traceroutes and check which ones pass through suspected
security agency interception sites, as well as which Internet service providers
carry their data en route.17 (See Figure 7.2 for an example.)
The IXmaps database also provides a valuable research resource for estimating
the most likely sites of CSE domestic Internet interception. In particular, it can
136 Andrew Clement

Figure 7.2  Boomerang route originating and terminating in Toronto. In this map showing
southern Ontario and the northeastern United States, a line, representing a data path, is
drawn from the origin in Toronto to New York City, then to Chicago, and finally to Toronto,
the final destination. Both New York City and Chicago are marked with an icon indicating
that they are sites of NSA surveillance. | Source: Adapted from Andrew Clement and Jonathan
Obar, “Canadian Internet ‘Boomerang’ Traffic and Mass NSA Surveillance: Responding to
Privacy and Network Sovereignty Challenges,” in Law, Privacy and Surveillance in Canada in
the Post-Snowden Era, edited by Michael Geist (Ottawa: University of Ottawa Press, 2015), 21.
Cartography by Eric Leinberger.

show the main Internet routes entering and traversing Canada, and reveal
the main network operators and their switching centres. The following
analysis considers the more than 250,000 traceroutes users contributed to
the database in the twelve-month period from 1 December 2016 to 30 Nov-
ember 2017. Of these, over 75,000 routes entered Canada. Table 7.1 shows in
ranked order the top five metropolitan areas where the first router in Canada
along the route is located. The percentages refer to the proportion of routes
in relation to the number of routes for which a city can be identified. In other
words, over 90 percent of incoming Internet routes make their first Canadian
hop in just these five metropolitan areas before being routed to their ultim-
ate destinations within Canada. It is these cities that CSE would be most likely
 Limits to Secrecy 137

Table 7.1 
Top border cities for Internet traffic entering Canada

City % Inbound routes Cumulative %

Toronto 34 34
Montreal 19 53
Vancouver 16 70
Ottawa 16 85
Calgary 7 92

Table 7.2
Top carriers bringing Internet data into Canada

Carrier Country % Inbound routes Cumulative %

Hurricane Electric USA 24 24
Bell Canada 18 42
Cogent USA 12 54
Peer 1 Canada 7 61
Zayo/Allstream USA 5 66
TATA India 5 70
Shaw Canada 4 74
GTT USA 2 76
Telus Canada 2 79
Level 3 USA 2 81

to prioritize when developing interception capabilities for monitoring foreign

threats.
As revealed in the documents released by Snowden and Mark Klein,18 an
earlier NSA whistle-blower, the preferred approach of signals intelligence
agencies for intercepting Internet data is through arranging directly with
telecom service providers to gain access to the switching equipment for install-
ing their sensors. A similar analysis of traceroute data to that above indicates
which carriers will be the most promising targets for CSE to approach. IXmaps
reports the officially registered carrier for each router’s IP address. Table 7.2
shows in ranked order the top ten carriers whose routers are the first encoun-
tered upon entry to Canada and that collectively account for over 80 percent
of inbound traffic.
138 Andrew Clement

Boomerang Routing
It is important to note that within the stream of Internet data entering Canada
from the United States, a significant portion (~10 percent) originated in Canada.
These are known as “boomerang” routes because of their characteristic pattern
of leaving Canada before returning to reach their destination. Figure 7.2 offers
an example of a boomerang route that both originates and terminates in Toronto
but transits via New York City and Chicago. While estimates vary, such routes
constitute approximately one-quarter or more of domestic routes (Canada to
Canada). Furthermore, nearly all Internet users in Canada that visit prominent
Canadian websites, including those of federal and provincial governments, are
likely to have their personal data transmitted in this way.19 This is significant
because CSE may have a legal basis to treat the data it intercepts in these inbound
channels as foreign-based by default, which provides a lower level of protection.
Only packet analysis after interception can indicate whether or not the data
originate with someone in Canada.

Interception within the Core of the Canadian Internet

As noted above, CSE planned to integrate the same capabilities deployed for
foreign SIGINT interception into “the CORE of the [Canadian] Internet” and,
based on the airport Wi-Fi case, succeeded to a considerable extent. IXmaps
data for routes that both originate and terminate in Canada help identify the
most likely sites for sensor installation to accomplish CSE’s ambition of
“extending the reach [of EONBLUE] to view national infrastructure.” Examining
domestic Canadian Internet communications during the same period produces
a similar picture. The following ten cities (metro areas), in decreasing order of
approximate traffic volumes, account for over 90 percent of the hops routed
through Canada’s Internet core: Toronto, Vancouver, Montreal, Ottawa, Calgary,
Edmonton, Winnipeg, Saskatoon, Saint John, and Thunder Bay.
Focusing on the hops by Canadian facilities-based carriers operating routers
in the Internet core, over 90 percent are handled by just five carriers and their
subsidiaries: Bell, Cogeco, Rogers, Shaw, and Telus. The leading foreign carriers
providing transit but no retail services within Canada account for about 20
percent of the Canadian domestic hops where a carrier is known. These are
Cogent, Hurricane Electric, TATA Communications, and Zayo (formerly
Allstream).
Table 7.3 indicates the cities where each of the major carriers, both Canadian
and foreign, predominantly operate their core Internet routers. While this
analysis is partial and approximate, the pattern is clear. If CSE is routinely sur-
veilling Canadians’ Internet communications within Canada, it would be able
to capture a large fraction (roughly estimated at 80 percent) of this
 Limits to Secrecy 139

Table 7.3 
Principal concentrations of Internet routers by metropolitan area and carrier

Carrier Vancouver Calgary Edmonton Winnipeg Toronto Ottawa Montreal

Canadian              
Bell         x x x
Shaw x x x   x    
Rogers x       x    
Telus x x x   x    
Cogeco/Peer 1 x       x   x
Foreign              
Hurricane x x   x x   x
Cogent x       x   x
Zayo x     ? x x ?
TATA x       x   x

communication with relatively few, strategically positioned interception devices.

The nine carriers listed in Table 7.3, as well as the cities where they handle the
largest volumes of traffic, are likely to be those CSE would target first in this
regard. An EONBLUE sensor installed in each of these 30 sites would represent
just 15 percent of the 200 sensors that CSE had deployed globally by 2010.

Concentration of Canada’s Physical Internet Infrastructure and

Telecom Industry Facilitates Mass Interception
The finding that CSE could achieve effective and efficient Internet interception
by arranging with a handful of carriers to access sensors in a small number of
cities is not surprising given the highly concentrated character of Internet rout-
ing. The switching facilities located in relatively few centres connected by high-
capacity fibre-optic trunk lines, which constitute the physical choke points of
Internet routing, offer attractive sites for intercepting the greatest amount
of traffic with the least effort. In this respect, the cloud metaphor for Internet
operations is highly misleading. While Canada covers a vast terrain, its Internet
infrastructure, like its population, is concentrated in large urban centres strung
along the US border, each with its own well-established historical north-south
and east-west transportation corridors along which fibre cable can be relatively
easily strung.
Another factor enabling bulk interception is Canada’s highly concentrated
telecom industry. The top five groups – Bell, Quebecor (Videotron),
140 Andrew Clement

Rogers, Telus, and Shaw – account for approximately 83 percent of total

industry revenues.20 All of these are long-standing incumbents in the com-
munications arena and depend on government licensing approval to stay in
business. For years, they routinely provided individual subscriber information
without a warrant when requested by law enforcement agencies, at least until
the 2014 Supreme Court of Canada ruling in R v Spencer.21 When lawful access
legislation (Bill C-30) was introduced in 2012, Bell led a group of telecom
companies that cooperated with the federal government around implementa-
tion issues while reportedly expressing no concerns about privacy, only about
how they were to be compensated. These five therefore constitute “low-hanging
fruit” from CSE’s perspective when seeking interception “partners.” Govern-
ment of Canada relationships with foreign transit providers are less clear, but
at least in the case of Cogent, Hurricane, and Zayo there is good reason to
believe the NSA already conducts surveillance on their networks and facilitates
CSE access.
The point of this analytic exercise is not to claim with certitude that CSE is
intercepting Internet traffic in any particular Canadian city or that any particular
carrier has agreed to become a Special Source Operation (SSO) by providing
CSE with access to its network infrastructure. Rather it is to make a well-
grounded case for suspecting the existence of bulk domestic Internet surveillance –
one sufficiently strong to place the onus on CSE to be significantly more forth-
coming with the Canadian public than it has been so far about what it is and is
not doing in this respect.

Implications for CSE and Its Telecom Partners

The foregoing analysis of Snowden documents and Canadian Internet routing
patterns provides reasonable grounds to suspect that:
• CSE is capable of intercepting the Internet communications of millions of
Canadians within Canada.
• CSE conducts this interception at the Internet exchanges in the following cities:
Calgary, Edmonton, Halifax, Hamilton, Montreal, Ottawa, Toronto, Vancouver,
Victoria, Windsor, and Winnipeg.
• CSE accesses the Internet switching operations of leading facilities-based tele-
communications providers such as Bell, Rogers, Shaw, Telus, and Videotron to
conduct its bulk Internet interception.

These claims are not conclusive, nor are they allegations of illegal behaviour on
the part of CSE or its telecommunications partners. They are, however, suffi-
ciently well founded to call into serious question CSE’s stock responses regarding
possible domestic surveillance. CSE repeatedly asserts that it follows the law,
 Limits to Secrecy 141

but has yet to categorically deny that it intercepts Canadians’ Internet com-
munications in bulk. Since CSE is ultimately answerable to the Canadian public,
the mounting evidence of domestic interception places a clear onus on CSE to
be much more transparent and accountable about its Internet surveillance
capabilities and activities.
In response to the airport Wi-Fi story, CSE mentioned only its foreign intel-
ligence mandate for collecting communications, leaving open the possibility
that it may have been collecting Canadian data used for the tracking experiment
under another of its mandates. Such collection would be consistent with what
we have seen earlier about the integration of sensors in the CASCADE program
to serve multiple mandates. CSE further denied “targeting” Canadians’ com-
munication (i.e., focusing attention on already specified individuals), but what
about data that is inevitably collected “incidentally” as part of full-take intercep-
tion? It also denied “collection,”22 but what does this actually mean? Is it equiva-
lent to interception in the usual sense adopted here, or, following the NSA, does
CSE regard collection as occurring only at the point that an analyst actually
reads or listens to message content?23
An essential move for CSE to maintain the trust of Canadians would be to
state plainly whether or not it has the capability to intercept Canadians’
Internet communications in bulk, whether it is routinely doing so, and under
what mandate(s). If CSE does intercept Canadians’ communications, it would
also need to be clearer about what personal communication data it captures
and what measures are in place to ensure that personal and democratic rights
are well protected. What content is intercepted? If “only” metadata is extracted
and stored, what does this consist of? If not all message content is discarded
immediately, is metadata extraction performed later?24 While beyond the
scope of this chapter, this opens a host of other questions concerning how
CSE processes the data after the point of interception. Where are the data
stored and for how long? What organizational entities have access to them
and for what purposes? What can CSE and its partner agencies infer about
individuals’ lives from the metadata? What “minimization” and other pro-
tective procedures apply? While some of these questions may infringe on the
sources and methods that security agencies legitimately keep secret, there
should at least be some public understanding of how the boundary is drawn.
Most fundamentally, Canadians need demonstrable assurance that any
secrecy or other security measure is consistent with the Canadian Charter
of Rights and Freedoms as well as established international human rights
norms.25 In particular, it should satisfy the four-part constitutional test based
on Oakes, that any secrecy or other security measure be minimal, propor-
tionate, necessary, and effective.26
142 Andrew Clement

Another aspect of the interception process calling for greater transparency

is the role telecom carriers play in enabling CSE to access their facilities or the
data flowing through them. Is it not reasonable to expect some degree of dis-
closure about whether ISPs are cooperating with CSE to facilitate surveillance?
It would contribute to a better-informed public debate about national security
to know whether Internet service providers are facilitating mass domestic sur-
veillance. In particular, following the model of AT&T’s likely cooperation with
the NSA, does CSE have access to fibre-optic splitters and deep packet inspection
devices installed on the premises of telecom carriers? If not, by what other
means do CSE sensors access Canadians’ data?
It is not surprising that CSE has been extremely reluctant to reveal anything
publicly about its operational capabilities. It is typical of security-oriented organ-
izations to treat any information they expose about themselves as potentially
providing advantage to those who may pose a threat. Furthermore, CSE was
founded in the wake of the Second World War and shaped by Cold War condi-
tions for over half its life, so a deeply ingrained culture of secrecy and suspicion
of outsiders is to be expected. As CSE pursues potential terrorist threats that are
more difficult to distinguish within the civilian population than the foreign state
agents that the agency was created to track (see Chapter 3), the tensions between
secrecy and democratic rights, such as personal privacy, freedom of expression,
and public accountability, are becoming more acute. This requires CSE to be
more transparent in ways that imply significant organizational change.
There are undoubtedly areas of CSE activity that rightfully need to remain at
the highest levels of secrecy for the agency to meet its legitimate mission. But
the degree of secrecy and obscurity that it currently maintains appears to far
exceed what is necessary or justified in this regard. The official response to the
airport Wi-Fi story is typical of CSE’s resistance to transparency about its activ-
ities: “The unauthorized disclosure of tradecraft puts our techniques at risk of
being less effective when addressing threats to Canada and Canadians.”27 CSE
has yet to make a substantive public case, however, that there is nothing that it
can reveal about what it does without incurring risks. The Snowden revelations
about NSA domestic surveillance provide abundant counter-evidence. Millions
of AT&T and Verizon customers know that their ISP has enabled the NSA to
intercept their personal communications, but there is no indication that this
has compromised US national security. Much more obvious is that this new
knowledge has fostered a vigorous and ongoing debate about the authority and
legitimacy of this surveillance, even leading to modest legislative reform.
Also typical of CSE’s position of secrecy is its response to a journalist reporting
on its EONBLUE program: “CSE cannot comment further on operations, methods
or capabilities, as that would constitute a breach of the Security of Information
 Limits to Secrecy 143

Act.”28 However, the Security of Information Act prohibits only unauthorized

disclosure of “special operational information,” defined as that which the “Gov-
ernment of Canada is taking measures to safeguard” (emphasis added).29 In other
words, there appears to be nothing in this act to prohibit CSE from being more
transparent if government policy were to change to provide the necessary author-
izations and selectively declassify information for which current safeguards cannot
be justified. Unfortunately, CSE’s statements so far do not even acknowledge that
there may be a public interest in disclosing some of its basic capabilities.
It is very difficult to see how clarifying this issue would so undermine CSE’s
effectiveness as to override the right and value of citizens knowing what their
government is doing, and the risks that such surveillance may pose to consti-
tutionally protected rights. Any serious adversary will already presume such
surveillance, especially in light of the published Snowden documents, and adopt
countermeasures. In the absence of a compelling rationale for why basic trans-
parency about surveillance capabilities would harm the public interest, the main
losers with CSE’s ongoing refusal to be more forthcoming on this issue will be
Canadian citizenry and democracy. This refusal fuels suspicion that CSE is
hiding from public scrutiny not so much out of legitimate operational needs
but to protect itself from exposing its own shortcomings, which, once revealed,
will lead to embarrassment or worse. Postponing this through continued secrecy
may be tempting as a short-term tactic but trust will continue to erode.
Between CSE’s current total secrecy about all surveillance activities and secrecy
that is demonstrably necessary for effectiveness, there is a broad middle ground
of transparency and accountability worth exploring, especially related to domes-
tic surveillance capabilities. Based on the preceding analysis, initial candidates
for basic capabilities that belong in this middle ground, to be considered for
public disclosure, include responses to such questions as:

• What proportion of Canadians’ Internet communications can or does CSE rou-

tinely intercept?
• What types of information can CSE derive from these intercepted communications?
• What is the nature of the relationship between CSE and the telecommunica-
tions carriers it relies on for domestic interception?
• What safeguards does CSE adopt to ensure compliance with Canadian law and
the Constitution?
• How operationally does CSE define the key terms that it uses in its official state-
ments, such as metadata, content, target, collection, direct, intercept, monitor,
use, surveil, personal information, private communication?
• How do these various capabilities and definitions relate specifically to CSE’s for-
mer and current mandates?
 144 Andrew Clement

This approach may be termed capability transparency, which intersects with

and complements the approach outlined in Chapter 13, on horizontal account-
ability and signals intelligence.
CSE has significantly outgrown its Cold War signals intelligence agency role
focusing its analogue transmission spying capabilities on a foreign state adver-
sary (see Chapter 5). In the past two decades, it has greatly expanded its surveil-
lance capabilities to include digital transmissions and it is now able to intercept
and process enormous volumes of Internet communications, both domestically
and internationally.30 Along with the emergence of more diffuse threats under
cybersecurity and anti-terrorism labels, CSE has also broadened the scope of
its operations in ways that potentially implicate all Canadians and their everyday
communications. However, there is little sign that it has acknowledged the new
obligations that come with these transitions, especially regarding public account-
ability and reform of its legacy culture of adamant secrecy.31 Good first steps in
this direction would be for CSE to acknowledge its responsibilities for greater
transparency and to clarify its capabilities for intercepting the Internet com-
munications of millions of Canadians. The newly formed National Security and
Intelligence Review Agency (NSIRA) and the Intelligence Commissioner offer
some promise this might actually happen.

Acknowledgments
An earlier extended version of this chapter, with additional technical details and more
extensive endnotes, is available at SSRN, https://ssrn.com/abstract=3206875.
I am grateful for the feedback on a preliminary presentation of this work from partici-
pants in the Security and Surveillance workshop held in Ottawa on 18–19 October 2017.
In particular, my ongoing conversations with Chris Parsons as well as his “Canadian
SIGINT Summaries” (https://christopher-parsons.com/writings/cse-summaries/) have
provided invaluable insights into CSE activities. Bill Robinson, whose Lux Ex Umbra blog
(https://luxexumbra.blogspot.com/) is an important resource for better public understand-
ing of CSE, also contributed helpful feedback on an earlier draft of this chapter. The
Snowden Surveillance Archive (https://snowdenarchive.cjfe.org/greenstone/cgi-bin/
library.cgi), designed and built by archivists George Raine and Jillian Harkness, likewise
assisted in finding and interpreting relevant Snowden documents. I also appreciate the
contributions over many years of Colin McCann and other collaborators in the IXmaps
(https://ixmaps.ca/) research project that provided the basis for the empirical analysis of
Internet routing patterns. IXmaps research has received funding from Canada’s Social
Sciences and Humanities Research Council, the Office of the Privacy Commissioner of
Canada, the Canadian Internet Registration Authority, and the Centre for Digital Rights.
As is the norm for academic research, the views expressed here are those of the author
alone and not of the funding organizations or others who have assisted in the research.

Notes
1 Communications Security Establishment (CSE), “Frequently Asked Questions” (13.
Does CSE target Canadians?), 3 July 2015, https://www.cse-cst.gc.ca/en/about-apropos/
faq#q13.
 Limits to Secrecy 145

2 CSE, “CSE Statement re: January 30 CBC Story – January 30, 2014,” 13 December 2016,
https://www.cse-cst.gc.ca/en/media/media-2014-01-30.
3 Bill C-59, An Act respecting national security matters, 1st Sess, 42nd Parl, 2019 (assented
to 21 June 2019), SC 2019, c 13.
4 Communication Security Establishment, “What We Do and Why We Do It,” 1 August
2019, http://www.cse-cst.gc.ca/en/inside-interieur/what-nos.
5 See also Chapters 2, 5, 8, 13, and 14.
6 Greg Weston, “CSEC Used Airport Wi-Fi to Track Canadian Travellers: Edward Snowden
Documents,” CBC News, 31 January 2014, http://www.cbc.ca/news/politics/csec-used
-airport-wi-fi-to-track-canadian-travellers-edward-snowden-documents-1.2517881.
Unless otherwise indicated, quotes in this section are from this article. Since publication
of the article, CSE no longer uses the acronym “CSEC.”
7 CSE, “IP Profiling Analytics & Mission Impacts” (slide deck, 10 May 2012), http://www.
cbc.ca/news2/pdf/airports_redacted.pdf.
8 CSE, “Metadata and Our Mandate,” 1 August 2019, https://www.cse-cst.gc.ca/en/
inside-interieur/metadata-metadonnees. See Chapter 14 in this book for the distinction
between shallow and deep metadata.
9 The slide deck gives the impression that this was another Canadian city, but CSE
denies this. See Amber Hildebrandt, “CSE Worried about How Its Use of Cana-
dian Metadata Might Be Viewed,” CBC News, 22 April 2015, http://www.cbc.ca/
news/canada/cse-worried-about-how-its-use-of-canadian-metadata-might-be-
viewed-1.3040816.
10 Laura Payton, “Spy Agencies, Prime Minister’s Adviser Defend Wi-Fi Data Collec-
tion,” CBC News, 3 February 2014, http://www.cbc.ca/news/politics/spy-agencies-prime
-minister-s-adviser-defend-wi-fi-data-collection-1.2521166.
11 CSE, “What We Do and Why We Do It.”
12 The principal CSE documents are “CSEC Cyber Threat Capabilities/SIGINT and ITS: An
End-to-End Approach” (slide deck, October 2009), https://assets.documentcloud.org/
documents/1690224/doc-6-cyber-threat-capabilities.pdf; and “CASCADE: Joint Cyber
Sensor Architecture” (slide deck, 2011), https://s3.amazonaws.com/s3.documentcloud.
org/documents/1690204/cascade-2011.pdf.
13 For more on CSE’s sensor programs, see Chapter 13.
14 CSE, “CASCADE.”
15 CSE, “What We Do and Why We Do It.”
16 See http://IXmaps.ca.
17 Andrew Clement, “IXmaps – Tracking Your Personal Data through the NSA’s War-
rantless Wiretapping Sites,” in Proceedings of the 2013 IEEE International Symposium
on Technology and Society (ISTAS) (Toronto, 27–29 June 2013), 216–23, doi: 10.1109/
ISTAS.2013.6613122.
18 Mark Klein, Wiring Up the Big Brother Machine ... and Fighting It (Charleston, SC: Book-
Surge, 2009).
19 Andrew Clement and Jonathan Obar, “Canadian Internet ‘Boomerang’ Traffic and Mass
NSA Surveillance: Responding to Privacy and Network Sovereignty Challenges,” in
Law, Privacy and Surveillance in Canada in the Post-Snowden Era, edited by Michael
Geist (Ottawa: University of Ottawa Press, 2015), 13–44. Available for free, open-access down-
load at http://www.press.uottawa.ca/law-privacy-and-surveillance or http://hdl.handle.
net/10393/32424.
20 Canadian Radio-television and Telecommunications Commission (CRTC), “Communi-
cations Monitoring Report 2017: The Communications Industry,” https://crtc.gc.ca/eng/
publications/reports/PolicyMonitoring/2017/cmr3.htm.
 146 Andrew Clement

21 R v Spencer, 2014 SCC 43, https://scc-csc.lexum.com/scc-csc/scc-csc/en/item/14233/

index.do; Alex Boutilier, “Government Agencies Seek Telecom User Data at ‘Jaw-Dropping’
Rates,” The Star, 29 April 2014, http://www.thestar.com/news/canada/2014/04/29/
telecoms_refuse_say_how_often_they_hand_over_customers_data.html.
22 CSE, “CSE Statement re: January 30 CBC Story”: “No Canadian communications were
(or are) targeted, collected or used.”
23 For more discussion of the contested and ambiguous interpretations of key terms, see
Chapters 2 and 8.
24 See Chapter 14 for a discussion of CSE’s possible interpretation of metadata.
25 Necessary and Proportionate: International Principles on the Application of Human Rights
to Communications Surveillance, May 2014, https://necessaryandproportionate.org/files/
2016/03/04/en_principles_2014.pdf.
26 R v Oakes, [1986] 1 SCR 103, http://www.canlii.org/en/ca/scc/doc/1986/1986canlii46/1
986canlii46.html.
27 CSE, “CSE Statement re: January 30 CBC Story.”
28 Matthew Braga, “How Canadian Spies Infiltrated the Internet’s Core to Watch What You
Do Online,” Motherboard, 11 February 2015, https://motherboard.vice.com/en_us/article/
9ak8qa/how-canadian-spies-infiltrated-the-internets-core-to-watch-what-you-do-online.
29 Security of Information Act, RSC, 1985, c O-5, esp. ss. 8, 14, 16, and 17, http://laws-lois.
justice.gc.ca/PDF/O-5.pdf.
30 In Chapter 6, Scott Thompson and David Lyon illuminate how CSE’s expanded technical
capabilities for bulk data collection have been accompanied by a significant transforma-
tion of the agency in terms of its conceptualization of data, policies, organizational struc-
ture, recruiting, and training. They conclude that the New Analytic Model’s overreliance
on technical skills at the expense of other disciplines may impair CSE’s actual capabilities
to appropriately interpret its newly available data and thereby create fresh problems.
31 Indeed, as Bill Robinson’s Lux Ex Umbra blog post of 16 April 2018, “And Still Darker:
CSE Stops Reporting Budget Breakdown,” suggests, CSE appears to be heading for
increased secrecy. https://luxexumbra.blogspot.ca/2018/04/and-still-darker-cse-stops-
reporting.html#links.
Part 3
Legal Challenges to Big Data Surveillance
in Canada
This page intentionally left blank
8
Gleanings from the Security Intelligence Review
Committee about the Canadian Security
Intelligence Service’s Bulk Data Holdings and
the Bill C-59 “Solution”
Micheal Vonn

The issue of mass surveillance by national security agencies in Canada has

probably seen greater prominence in the last three years than ever before.
The national security landscape of Canada was radically altered in 2015. That
year saw the enactment of an omnibus national security bill so notorious that
it was generally known not by the name of the act but simply as “Bill C-51.” It
generated widespread public opposition. Greatly increased surveillance capacity
and the potential for bulk data acquisition by national security agencies by
means of new powers of “information sharing” between federal government
bodies were among the most prominent of the concerns.1
The government that brought in Bill C-51 lost the next election. The new
government (the Liberals) had supported Bill C-51, but with the promise that,
if elected, they would make appropriate and needed amendments to the new
law. In the run-up to the introduction of Bill C-59, which was the government’s
delivery of its promised changes, a series of reports and revelations brought
critically important insight into the issue of Canada’s national security surveil-
lance more generally and bulk data acquisition and data analytics more
specifically.
The revelations that were reported from the cache of documents disclosed in
2013 by Edward Snowden about the surveillance activities of the United States
and allies brought the issue of bulk data acquisition prominently into public
discourse in the Anglo-American press. More familiar terms like “mass surveil-
lance” were generally used to discuss the practices that had been revealed, such
as the secret interception of communications traffic through undersea fibre-optic
cables. The question of how various countries conduct their national security
intelligence, and particularly the scale, rationale, and legality of domestic spying
activities, came into sharp focus in many countries, notably the United States
and the United Kingdom.
By contrast, Canada’s reaction to the Snowden revelations was muted. While
the American and British governments had vigorous debates about domestic
bulk surveillance powers after the Snowden revelations, the Canadian
150 Micheal Vonn

government was unresponsive to calls for an emergency debate on the surveil-

lance activities of Canada’s national electronic intelligence-gathering agency,
the Communications Security Establishment Canada (now the Communications
Security Establishment, or CSE).2
In 2013, the British Columbia Civil Liberties Association (BCCLA) filed a
lawsuit against the CSE claiming unjustifiable infringement of rights under the
Canadian Charter of Rights and Freedoms in respect of the statutory powers
under the National Defence Act permitting the interception, retention, and use
of private communications (including metadata) of persons in Canada as
authorized by ministerial directives and authorizations.3 As of 2020, the case
has not been heard in court on its merits. In contrast, by 2019 the European
Court of Human Rights had rendered a historic decision on bulk data surveil-
lance by the British government and appeals had been filed.4
However, judicial and quasi-judicial initiatives, most notably those issuing
from the Federal Court, have been a critical aspect of the limited transparency
that has been brought to bear on Canada’s intelligence agencies’ bulk data
acquisition. These decisions, alongside reports from oversight bodies, provide
arguably the best insight available to the public on Canadian intelligence agen-
cies’ mass domestic surveillance.
The first part of this chapter discusses both what is reported and what can
be gleaned from two recent annual reports by the Security Intelligence Review
Committee (SIRC) about the Canadian Security Intelligence Service’s (CSIS)
bulk data holdings and use of big data surveillance and big data analytics.
The second part outlines how Bill C-59, An Act respecting national security
matters, proposes to address the problems identified with CSIS’s bulk data
program.5
In the fall of 2016, the Federal Court issued an explosive decision, X (Re),
about information that had been gathered by CSIS.6 It found that CSIS had
breached its duty of candour to the court in failing, for over ten years, to alert
the judges that had been issuing warrants under section 21 of the Canadian
Security Intelligence Service Act (CSIS Act) that CSIS was indefinitely retaining
information from these warrants about third parties unrelated to threats to the
security of Canada.7 The court found that CSIS was holding this information
illegally. The information had been added to the data holdings of a program
that appears to have since changed its name but was once known as the Oper-
ational Data Analysis Centre (ODAC). ODAC started in 2006, but CSIS had
not disclosed its existence to the court. The court discovered its existence only
inadvertently, through an annual report by SIRC.
SIRC is an external independent review body that conducts reviews of
CSIS’s activities and issues reports to Parliament. SIRC’s 2015–16 annual
 Gleanings from the Security Intelligence Review Committee 151

report was particularly notable for including the findings of its first exam-
ination into the CSIS data acquisition program.8 SIRC itself has broad
authority to examine information in the control of CSIS, but its public
reports on these examinations are terse, high-level, and tightly edited to
protect national security. Consequently, these reports typically require some
“unpacking” to extract the importance of the information provided by the
review and to use it to sketch more of the picture (or of the likely picture)
of CSIS activities.
SIRC’s 2016 report of its first examination of the CSIS data acquisition
program, in its characteristically measured, understated tone, managed to
convey that CSIS’s activities in this realm were essentially unmoored from
law.

The Challenge of Bulk Data Acquisition for CSIS

Canadians have a much better idea about what kinds of mass surveillance
activities are occurring among some of Canada’s intelligence partners in the
“Five Eyes” alliance than about the Canadian situation. Ongoing revelations
about the activities of the US National Security Agency (NSA) and extensive
information about mass surveillance in the United Kingdom disclosed through
a historic legal challenge9 have given Canadians fodder for speculation about
what national security agencies in Canada might be doing, or trying to do, with
respect to bulk data collection.
It is clear that the United Kingdom’s security intelligence agency has been
collecting vast troves of government, commercial, and telecommunications
records on millions of people, but Canada’s practices are still comparatively
unknown. Regarding CSIS specifically, the question has been not only what
bulk data it might be collecting but under what ostensible legal authority such
collection could be occurring.
The CSIS Act sets a demanding test of “strict necessity” for data collection:

s. 12(1) The Service shall collect, by investigation or otherwise, to the extent

that it is strictly necessary, and analyse and retain information and intelligence
respecting activities that may on reasonable grounds be suspected of constitut-
ing threats to the security of Canada and, in relation thereto, shall report to and
advise the Government of Canada.10

As the 2016 SIRC annual report confirms that CSIS has a program of bulk data-
sets, this naturally invites the question of what kinds of bulk data collection –
generally construed as “haystack hoovering” just in case any needles might be
found – could meet the test for strict necessity.
152 Micheal Vonn

Collection That Isn’t Collection

One way in which CSIS justified its bulk data holdings was by claiming that the
standard of strict necessity did not apply to certain classes of datasets that, on
the CSIS interpretation, the agency had acquired without ever “collecting.”
The 2016 SIRC report was the first public insight into the legal interpretation
by CSIS that creates a category of acquired but “non-collected” datasets. As the
report sets out, CSIS’s bulk data holdings are used in many ways and are con-
sidered to be of two broad types. One type consists of information used mainly
for verifying identification. CSIS calls these datasets “referential.” The other type
(“non-referential”) consists of bulk data on a wide variety of individuals that
are supposed to be retained only if they are determined to be relevant to an
actual investigation. CSIS’s position is that their referential datasets are not
“collected” (and therefore not governed by section 12 of the CSIS Act) because
they are all openly sourced and publicly available.
SIRC did not contest the principle that some “purely referential datasets would
not constitute ‘collection’ per se” and cited the phonebook as an example of a
dataset that would meet the criteria of being openly sourced and publicly avail-
able.11 This statement by SIRC appears carefully worded. It is not a legal inter-
pretation that is agreed to, but a “principle.” Likewise, there appears to be a
distinction introduced between referential and “purely referential” datasets.
And SIRC does not precisely concede that some datasets held were not collected,
but says rather that they were not collected “per se.”
However, even assuming the validity of this “non-collected” categorization,
which at least looks open to some dispute, SIRC finds that these bulk data hold-
ings include data that are not publicly available and openly sourced, and so do
not meet the criteria for inclusion in this category regardless of the category’s
validity.

A Threshold That Isn’t a Threshold

Turning to the non-referential datasets, CSIS does not dispute that they are
“collected” and thus subject to the collection threshold of “strict necessity.”
SIRC’s findings in its audit of the non-referential datasets of CSIS are contained
in a single, astonishing sentence:

Despite this [CSIS’s agreeing that section 12 of the act applied], SIRC found no
evidence to indicate that CSIS had appropriately considered the threshold as
required in the CSIS Act.12

Again, it is helpful to pay careful attention to the language. This was not a matter
of a dispute about how the threshold was interpreted or even how an
 Gleanings from the Security Intelligence Review Committee 153

interpretation was applied in some cases. This was a case of there being no
evidence to indicate that CSIS had even appropriately considered the threshold.
It is arguable that this is a failure so vast and otherwise inexplicable as to suggest
a contempt for the applicable legal requirements. While it is never expressly
stated, the inescapable conclusion of SIRC’s findings is that a very large amount
(and, minus the phonebook, possibly all) of the CSIS bulk data holdings were
not lawful.
SIRC issued recommendations (it has no order-making powers) that CSIS
re-evaluate all its referential bulk datasets; undertake an assessment of its non-
referential datasets to ensure they meet the “strictly necessary” threshold; and
halt all acquisition of bulk datasets until a formal process of assessment exists
to ensure that required standards are being met.13

Guidelines That Are (Probably) Not the Guidelines

SIRC then went beyond these recommendations and made public a proposal
for guidelines to help interpret the strict necessity standard in the context of
bulk data. In essence, SIRC weighed in to suggest that section 12, despite its
high threshold, was not incompatible with bulk data collection and proposed
criteria it implicitly advocated as principled and workable for making the neces-
sary evaluation.
SIRC’s proposal was a three-part test:

First, for any bulk information, a clear connection to a threat to the security
of Canada as defined in section 2 of the CSIS Act must be established. Sec-
ond, it must be established that less intrusive means that would satisfy the
intelligence requirements are not available as an alternative to bulk collection,
consistent with the principle of proportionality. Third, if there is no reason-
able alternative to bulk collection, CSIS needs to provide an objective assess-
ment of how closely connected the bulk information is to intelligence of value;
the broader the intended collection, the more strictly CSIS must establish the
connection between the bulk information and the threat-related intelligence.
[Emphasis added]14

It is stated in the 2016 SIRC report’s very brief “CSIS response” sidebar that
CSIS agreed to review its bulk data holdings and to suspend further bulk data
acquisition pending the implementation of its newly approved governance
framework.15 However, the “response” is silent as to whether that governance
framework includes any of the criteria proposed by SIRC. The read-between-
the-lines assessment would be that the SIRC guideline recommendations are
not found in the new governance framework for bulk data acquisitions.
154 Micheal Vonn

SIRC noted in this report that it had previously seen references in CSIS docu-
ments to “the need to validate the authority to collect and manage the risk of
over-collection by confining collection to that which is ‘strictly necessary.’”16
SIRC reported that it was told that a (then) two-year-old draft of a governance
framework existed but had never been finalized.17 The CSIS “response” statement
would suggest that it is this old governance document – drafted before the
decision in X (Re) – that CSIS agreed to finalize and implement. It is difficult
to understand why SIRC would be publishing its own proposal for criteria to
evaluate bulk data collection in compliance with section 12 of the CSIS Act if
no proposal was required because the CSIS governance framework already
contained these criteria. The suggestion is that the governance framework does
not contain guidance and interpretation of the kind proposed by SIRC.

The Program Previously Known as ODAC and Currently

Known as [Redacted]
SIRC and the Federal Court have given us some inkling of what might be occur-
ring in the data analytics program(s) of CSIS.
The 2016 SIRC report notes that

CSIS uses bulk datasets in multiple ways. They can be used to conduct indices
checks by taking information already connected to a potential threat – such as
an address, phone number or citizen identification number – and using it to
search for “hits” in the data. Datasets can also be used to enhance knowledge
of a target by searching the data for previously undetected trends, links or pat-
terns between and among data points. And data is used to support more focused
inquiries, such as “data mining” to identify leads. Finally, SIRC was told that the
data can be used to try to identify previously unknown individuals of interest
by linking together types of information which have mirrored threat behaviour.

Overall, the addition of more datasets was expected to enrich CSIS’s analytical
capacity and enhance its ability to provide support for CSIS investigations.18
This indicates that CSIS uses bulk datasets in all of the following ways: to
confirm identity, to learn more about targets, to detect networks and patterns,
and to profile.
From X (Re) we learn:

[37] In the early 2000’s, the CSIS considered that the information it collected
through investigations was underutilised as it was not processed through mod-
ern analytical techniques ... The ODAC was designed to be “a centre for excel-
lence for the exploitation and analysis” of a number of databases.
 ...
 Gleanings from the Security Intelligence Review Committee 155

[41] ... The present reasons should not give the impression that the Court is
well informed of the [redacted] program; only very limited evidence was pro-
vided. Given that the program was still called the ODAC at the time of the appli-
cation, I will use that term and not [redacted].
[42] The ODAC is a powerful program which processes metadata resulting in
a product imbued with a degree of insight otherwise impossible to glean from
simply looking at granular numbers. The ODAC processes and analyses [sic]
data such as (but not limited) to: [redacted]. The end product is intelligence
which reveals specific, intimate details on the life and environment of the per-
sons CSIS investigates. The program is capable of drawing links between vari-
ous sources and enormous amounts of data that no human being is capable of
[redacted].19

Later, in its discussion of the arguments put forward by counsel about the privacy
interests of the data at issue, specifically with respect to “information gleaned
from granular metadata and from the products of its aggregation and analysis,”
the court noted that “the products of the CSIS’s analytical methods are much
more elaborate than methods or types of information at issue in prior Supreme
Court of Canada cases.”20
This signals that we are in a grey zone about what level of constitutional
privacy protection the data – but, more specifically, the data processed in this
way – requires. Whatever else this statement might be referring to, it almost
certainly would be referring to national security profiling, which was cited in
the 2016 SIRC report as one of the uses of the bulk data holdings.

Security of Canada Information Sharing Act in Action –

Global Affairs Canada and Canada Revenue Agency
It is important to note that these revelations about CSIS’s bulk data collection
occurred against the backdrop of an increased ability of many government
agencies to “share” their information, including entire databases, with national
security agencies such as CSIS.
The Anti-terrorism Act, 201521 (also known as Bill C-51) is omnibus legislation
that, among other things, created a free-standing statute called the Security of
Canada Information Sharing Act (SCISA).22 SCISA allows seventeen federal
government institutions to share information under a standard broader than
any legal language in the Canadian national security context, namely, that the
information relates to an “activity that undermines the security of Canada”23
(versus the usual language of “threatens the security of Canada”).
The Canadian government conducted a national consultation on
security issues writ large in 2016 and the results of that consultation
156 Micheal Vonn

demonstrated “a great deal of concern” about SCISA.24 As the consultation

report noted:

Many organizations recommended SCISA be repealed or fundamentally

revised, with concerns – particularly among human rights, legal and commu-
nity organizations – the current definitions of information that can and cannot
be shared are too vague and that existing review mechanisms do not provide
enough accountability.25

Despite the wide scope for expansion of information sharing under SCISA, SIRC
reports that there has not been a large increase in the volume of sharing with
CSIS. In its 2017 report, SIRC “noted that the volume of exchanges [between
CSIS and other federal institutions] under SCISA has been modest.”26 There were
two agencies – Global Affairs Canada (GAC) and the Canada Revenue Agency
(CRA) – that did most of the SCISA-authorized exchanges with CSIS, and the
2017 SIRC report gives an overview of the information sharing with each.
These brief, high-level overviews provide only information subject to the
limitations that all of SIRC’s reporting is subject to. Even within these limits,
SIRC manages to give a good indication of the formidable entanglement of legal
and policy thresholds at play for information sharing and how these require-
ments from different bodies and in different pieces of legislation are supposed
to mesh, but likely do not.
Walking through this with some of the information provided in the findings
with respect to GAC, we note that CSIS and GAC share information relating to
activities that undermine the security of Canada, as defined in SCISA. However,
CSIS is allowed to “collect” information only “to the extent that it is strictly
necessary,”27 and it is not at all clear that the threshold for information to exit
GAC (under SCISA) and the threshold for information to enter CSIS (under
the CSIS Act) are the same thresholds, and if not, how such incompatibility is
assessed and what happens as a result.
We learn from the 2017 SIRC report that CSIS and GAC signed an information-
sharing agreement in 2016, but that discussions between the agencies are
ongoing, in part due to disagreements on thresholds and triggers for disclosures
of information (CSIS arguing for more, not less, disclosure).28 The 2017 SIRC
report also notes that a small number of disclosures of consular information
were recorded as having been made under the Privacy Act,29 which is perhaps
unsurprising given that legal commentators have often noted that interplay
between SCISA and the Privacy Act is extremely unclear.30 Alternatively or
additionally, it might signal that targeted information is being sought under
one scheme and bulk data under the other.
 Gleanings from the Security Intelligence Review Committee 157

From the 2017 report’s findings about the CRA we learn that at least some of
the information that is being shared by CRA with CSIS previously required a
judicially authorized warrant for disclosure to CSIS, and that there is no memo-
randum of understanding in place with respect to information sharing of tax-
payer information. In terms of aligning the thresholds for disclosing information
with the thresholds for receiving information, the Income Tax Act was amended
to allow for the disclosure of information to bodies like CSIS if “there are rea-
sonable grounds to suspect that the information would be relevant to (i) an
investigation of whether the activity of any person may constitute threats to the
security of Canada as defined in section 2 of the Canadian Security Intelligence
Service Act.”31 This is a low threshold requiring only suspicion (not belief) of
information relevant (not necessarily evidence) to an investigation of whether
the activity of any person may constitute threats to the security of Canada (not
the activities of targeted persons, but whether any person’s activities may pos-
sibly justify targeting). There does not appear to be any bar to bulk data sharing
on this threshold.

Questions of Proportionality, Necessity, and Relevance

SIRC’s discussion about the CRA data gives glimpses of ongoing disagreements
about data access. The 2017 SIRC report notes that CSIS has had an internal
directive since April 2016 that stipulates the specific internal authority that is
required for CSIS to ask the CRA for taxpayer information.32 In light of “the
type of constitutional protections that have been found to apply to taxpayer
information,” SIRC makes a recommendation that CSIS “increase the required
threshold for receiving taxpayer information from CRA” and “consider the
appropriateness of seeking a Department of Justice case-by-case analysis of the
proportionality of each request.”33
These recommendations certainly signal that SIRC’s view is that CSIS is over-
collecting from the CRA and that this over-collection may be unconstitutional.
CSIS “agreed to increase the required threshold for requesting taxpayer infor-
mation from CRA,” but did not agree to seek legal advice from the Department
of Justice on a case-by-case basis, stating the view that its agreement to raise
the threshold of the requests “better addresses the issue of proportionality from
a CSIS perspective.”34
This is an intriguing reference to “the issue of proportionality.” Recall that
SIRC’s proposed guidelines for the “strict necessity” test incorporated a pro-
portionality principle, specifically that “it must be established that less intrusive
means that would satisfy the intelligence requirements are not available as an
alternative to bulk collection.”35 It is not clear (and this chapter argues that it is
unlikely) that CSIS has adopted SIRC’s guidelines. Regardless of the guidelines
158 Micheal Vonn

CSIS is operating from, however, there is an “issue of proportionality” on which

CSIS has a perspective.
One aspect of the “perspective” may be on the meaning of the proportionality
principle. It is possible, for example, that CSIS’s perspective does not incorporate
a “least intrusive means” test in its proportionality principle. CSIS may be
interpreting “proportionality” along other lines, such as whether the nature of
the intrusion is proportionate to the seriousness of the threat being investigated.
Were that to be the case, it would effectively negate “proportionality” as a
meaningful limit on data acquisition. If it is a given that national security threats
are very serious, then any weighing of the data acquisition against the serious-
ness of the threat would tend to license very expansive acquisition.
Thus, it is important that “perspectives” on the meaning of the proportionality
principle be drawn out and, where needed, challenged. At this juncture, it
appears that “proportionality” is some component of the internal criteria being
applied, but that the meaning of the term (that is, what is meant to be propor-
tionate to what) may or may not align with ordinary understandings and
expectations.
But proportionality is not the only principle that we can infer forms part of
the criteria being used to assess the “strict necessity” test for collection. It also
appears likely that there is a “relevance” criterion on the evidence that the 2017
SIRC report announces that two aspects of its ongoing work are reviewing CSIS’s
response to the Federal Court decision in X (Re) and “examining CSIS data set
holdings to determine their relevance and necessity.”36
The examinations with respect to relevance are likely of critical importance. The
secret operational meaning of the standard of “relevance” in the national security
context of our close ally the United States illustrates this point very graphically.
The Snowden revelations in 2013 made public a large trove of information
about NSA surveillance programs. Among other outcomes, the revelations
spurred a report by the US Privacy and Civil Liberties Oversight Board (PCLOB)
into the NSA’s telephone records program.37
The NSA program in question was an ongoing collection of the details of
most Americans’ telephone records. It did not collect the content of the calls,
but did collect a rich repository of data about the calls. The ostensible authority
for the program was section 215 of the USA PATRIOT Act, which authorizes the
collection of certain business records where there are reasonable grounds to
believe that the records are relevant to an authorized investigation. The US
government’s justification for its ongoing mass surveillance of the nation’s
telephone activity turned on the interpretation of “relevant.”
To justify its position that the telephone records of the entire country are
relevant to all its counterterrorism activities, the US government advanced two
 Gleanings from the Security Intelligence Review Committee 159

rationales. The first rationale was that the data are relevant on the basis that the
NSA has analytical tools that are likely to generate investigative leads and that
these tools require bulk data for analysis. Because bulk data is necessary to
operate these analytical tools, the government argued that bulk data is relevant
for the purposes of the statute. The second rationale was that “relevance” should
have a very expansive interpretation drawing on analogous legal contexts, such
as the discovery process. The PCLOB soundly rejected both rationales.
With respect to the argument that relevance is founded in analytical tools
generating the necessity for bulk data, the PCLOB rejected the “elastic defin-
ition”38 that “supplies a license for nearly unlimited governmental acquisition
of other kinds of transactional information.”39

In the Board’s view, this interpretation of the statute is circular and deprives the
word “relevant” of any interpretative value. All records become relevant to an
investigation under this reasoning, because the government has developed an
investigative tool that functions by collecting all records to enable later search-
ing. The implication of this reasoning is that if the government develops an
effective means of search through everything in order to find something, then
everything becomes relevant to its investigations. The word “relevant” becomes
limited only by the government’s technological capacity to ingest information
and sift through it efficiently.40

The PCLOB also rejected the rationale that “relevance” has a particularized
meaning in legal contexts that is more expansive than the ordinary dictionary
definition of the term, arguing that while other legal processes demonstrate that
“relevance” can have legitimately expansive interpretations, no definition can
be so expansive as to effectively be no limit whatsoever.

Simply put, analogous precedent does not support anything like the principle
that necessity equals relevance, or that a body of records can be deemed relevant
when virtually all of them are known to be unrelated to the purpose for which
they are sought.
...
Relevance limitations are a shield that protects against overreaching, not a
sword that enables it.41

The disclosure of the NSA’s mass surveillance of telephone records and the
paradoxical interpretation of relevance that purported to authorize the program
generated controversy on several fronts. The massive sweep of the program
affected the privacy rights of nearly everyone in the United States. But further,
160 Micheal Vonn

it clearly demonstrated the dangers to the democratic process of secret legisla-

tive interpretations. Congressman Jim Sensenbrenner, a primary author of the
PATRIOT Act, made public his views that the legislative intent of the impugned
provision of the law had been fundamentally undermined by the administra-
tion’s secret interpretations: “As I have said numerous times, I did not know the
administration was using the Patriot Act for bulk collection, and neither did a
majority of my colleagues.”42
These are more than cautionary tales for Canada. The bulk data programs of
CSIS are in service of data analytics of the kind that US counterparts have used
to justify nearly unlimited data acquisition. Additionally, our democratic rep-
resentatives may be just as in the dark as the general public as to the nature of
the data acquisition that is occurring now or will occur under legislative pro-
posals in the future. And the analytics in question can present their own prob-
lems. For example, CSIS appears to be conducting profiling of some kind, and
yet there is a great deal of evidence to suggest that efficacious profiling for
terrorism or serious criminality is not even possible.43
But the failures of democratic process run deeper still, in that Canada has
never had a meaningful democratic debate about the shift from targeted surveil-
lance to bulk data acquisition and analytics. We are currently forced to rely on
the very thin disclosures provided through SIRC to piece together glimpses of
the likely activities of CSIS in this respect and the promise that new legislative
proposals will bring greater accountability.

Meanwhile, Bill C-59’s “Solution” to the Bulk Data Controversy

The centrepiece of the response of the government to Canada’s national security
consultation is Bill C-59. It is a very complex, omnibus legislative scheme that,
among other things, significantly revises the oversight and accountability of
national security and intelligence agencies. In response to many calls for effective
and integrated review, dating back many years, part one of Bill C-59 creates the
National Security and Intelligence Review Agency (NSIRA). NSIRA is a kind
of “super-SIRC” with a mandate that includes reviewing any activity carried
out by CSIS, the Communications Security Establishment (CSE; Canada’s signals
intelligence agency), and other specified federal bodies whose activities relate
to national security or intelligence.
The enactment of Bill C-59 means that NSIRA, rather than SIRC, has respon-
sibility for examining CSIS’s data holdings. NSIRA has the ability to make
findings and recommendations that relate to legal compliance as well as the
“reasonableness and necessity” of the exercise of the powers of CSIS and must
report on these annually. Under Bill C-59, NSIRA is folded into a larger account-
ability architecture that includes, in part two of the bill, the creation of an
 Gleanings from the Security Intelligence Review Committee 161

Intelligence Commissioner. The Intelligence Commissioner is a retired judge

who, among other duties, must review several of the new powers that Bill C-59
gives to CSIS with respect to the collection of bulk data.
Part four of Bill C-59 sets out a raft of amendments to the CSIS Act. Particularly
notable are the provisions about datasets. These changes expressly allow for
datasets (defined as “a collection of information stored as an electronic record
and characterized by a common subject matter”)44 to be acquired and used in
a number of different ways. The two types of datasets probably most germane
to Canadians and people in Canada are the “Canadian Datasets” and the “Pub-
licly Available Datasets.”
The minister of public safety can determine “classes” of Canadian Datasets
that CSIS is authorized to collect. These are datasets that contain personal
information that does not directly and immediately relate to activities that
represent a threat to the security of Canada.45 These datasets can be authorized
for collection by CSIS if the minister concludes that the querying (conducting
a specific search with respect to a person or entity for the purpose of obtaining
intelligence) or exploitation (a computational analysis of one or more datasets
for the purpose of obtaining intelligence that would not otherwise be apparent)
of any dataset in the class could lead to results that are relevant to the perform-
ance of CSIS’s duties and functions.46 The Intelligence Commissioner reviews
whether the minister’s conclusions in determining what classes of datasets are
authorized for collection are reasonable.
When CSIS acquires a Canadian Dataset, there is an evaluation period (ninety
days) during which a designated employee must delete any information that relates
to personal information that in CSIS’s opinion is not relevant to the performance
of its duties and functions and may be deleted without affecting the integrity of
the dataset. CSIS must also delete any information in respect of which there is a
reasonable expectation of privacy that relates to the physical or mental health of
an individual, and delete any information subject to solicitor-client privilege.
At the end of the evaluation period, CSIS must make an application for judicial
authorization to retain the Canadian Dataset. The judge may authorize the
retention of the dataset if it is “likely to assist” CSIS in the performance of its
duties or functions and the legislated obligations (authorized class, Intelligence
Commissioner’s review, etc.) have been fulfilled. A judicial authorization is valid
for up to two years and the authorization can set out terms and conditions,
including with regard to the querying or exploitation of the dataset or the
destruction or retention of the dataset or a portion of the dataset. A judicially
authorized Canadian Dataset can then be queried or exploited by CSIS “to the
extent that it is strictly necessary” to assist in the performance of its functions
and duties.47 The results of these queries and exploitations may be retained if it
162 Micheal Vonn

is “strictly necessary” to assist CSIS in the performance of its duties and func-
tions. NSIRA can review these activities.
Publicly Available Datasets are defined in the amendments to the CSIS Act
as datasets “publicly available at the time of collection.”48 CSIS can collect a
Publicly Available Dataset if it is satisfied that it is relevant to the performance
of its duties and functions and is evaluated by a designated employee. CSIS can
query and exploit Publicly Available Datasets and retain the results. NSIRA can
review these activities.

Addressing or Excusing the Controversy of Bulk Data?

It is instructive to compare SIRC’s proposals for standards and criteria applic-
able to bulk data collection with the scheme authorized by Bill C-59.
Recall that SIRC proposed a three-part test for the bulk data acquisition:
1 Clear connection to a threat to the security of Canada. A clear connection to a
threat to the security of Canada as defined in section 2 of the CSIS Act must be
established.
2 No less intrusive means available. It must be established that less intrusive means
that would satisfy the intelligence requirements are not available as an alterna-
tive to bulk collection, consistent with the principle of proportionality.
3 Objective assessment of intelligence value. If there is no reasonable alternative to
bulk collection, CSIS needs to provide an objective assessment of how closely
connected the bulk information is to intelligence of value. The broader the in-
tended collection, the more strictly CSIS must establish the connection between
the bulk information and the threat-related intelligence.

Here is the “test” as set out in C-59.

For Publicly Available Datasets, the first item of concern is that what consti-
tutes “publicly available” is not defined. In the referential databases that were
reviewed by SIRC, SIRC disagreed with CSIS that it had collected datasets that
were “publicly available and openly sourced.” The proposed amendments to the
CSIS Act leave the term “publicly available” undefined, leaving open a myriad
of possible interpretations, some expansive enough to include hacked informa-
tion that is posted to the Internet or personal information purchased from data
brokers. It gives no comfort that elsewhere in C-59 we do see definitions, and
when the new Communications Security Establishment Act uses the term “pub-
licly available information,” the definition is very expansive: “information that
has been published or broadcast for public consumption, is accessible to the
public on the global information infrastructure or otherwise or is available to
the public on request, by subscription or by purchase.”49 C-59 allows CSIS to
 Gleanings from the Security Intelligence Review Committee 163

collect “publicly available” datasets (with no definition of that term) on the basis
of a bare “relevance” standard.
As to Canadian Datasets, the personal information they contain is expressly
acknowledged as not directly and immediately relating to activities that represent
a threat to the security of Canada. The test is simply that the results of querying
or exploiting this information could be relevant and that this assessment must
be reasonable.
In theory, the privacy impact of this extremely wide-open funnel could be
slightly moderated by the “evaluation” that requires that irrelevant personal
information be deleted if this can be done without affecting the integrity of
the dataset, but it is difficult to imagine how this would ever be effectively
used. One of the main purposes of the dataset is for “searching the data for
previously undetected trends, links of patterns between and among data
points”50 and providing “products” that “[draw] links between various sources
and enormous amounts of data that no human being is capable of.”51 How
could an evaluator assess whether a piece of personal information is irrelevant
to a big data analytics process in which it is impossible to know the “relevance”
of any particular type of data because the queries aren’t set, predictable, or
even decisions made by humans, in the case of analytics that use machine
learning? In practice, it is likely that “the integrity of the dataset” will be cited
in almost every case as the reason for keeping the dataset intact and not
removing personal information.
The judicial authorization for the retention of the Canadian Datasets sounds
like significant gatekeeping, but in fact it simply compounds the effect of all of
the very low standards that lead up to it. Personal information that does not
directly and immediately relate to activities that represent a threat to the security
of Canada is allowed to be collected if it “could be relevant”; this assessment
must be reasonable and the judge decides whether the dataset can be retained
on the standard that it is “likely to assist.”52
It is only at the point that the now fully approved dataset gets queried or
exploited by CSIS that we see the introduction of the language that guided
the entirety of SIRC’s test, the language and interpretation of “strict necessity.”
NSIRA can review those decisions, and make findings as to reasonableness
and necessity and report those findings, but NSIRA is not likely to be report-
ing non-compliance with the law. To be clear, it is obliged to report even
possible non-compliance with the law to the attorney general, but the law as
it applies to CSIS collection, use, and retention of bulk data is so broadly
permissive that it is unlikely that CSIS will be exceeding such capacious
boundaries with any frequency.
164 Micheal Vonn

A “Solution” That Isn’t a Solution

The matter of CSIS’s unlawful bulk data collection, as revealed in X (Re), relates
to more than the rule of law. Certainly, it is critical to have confidence that CSIS
is abiding by the law, and legislating express powers where they are needed is
an obvious means of addressing some of the serious concerns that have arisen
in the context of SIRC’s reporting on bulk data acquisitions and the decision in
X (Re). However, it is arguable that what Bill C-59 contains does not even suf-
ficiently address this component of the problem. The absence of definitions of
such critically important and historically slippery and contested terms such as
“publicly available” are serious invitations to the kind of “secret-law-by-secret-
interpretation” problem that has historically plagued national security in many
countries, Canada included.53 Beyond the rule-of-law issue, however, there is
of course the substantive issue of what kind and degree of bulk data surveillance
is being authorized. Comparing the standards and criteria proposed by SIRC
for bulk data collection with those contained in C-59, it is evident that the
opportunity for a meaningful and measured approach to bulk data surveillance
for national security purposes has been squandered. Under C-59, CSIS would
have vast scope for bulk data surveillance and the national security account-
ability architecture, notwithstanding its importance, cannot provide an effective
means of restraining excessive bulk data surveillance when the thresholds for
surveillance it is empowered to enforce are shockingly low.
When the US Privacy and Civil Liberties Oversight Board undertook its
assessment of the NSA’s telephone records program and found that it was “not
sustainable from a legal or a policy perspective,”54 it warned against a response
that would address only the legality and not the policy:

The Board also recommends against the enactment of legislation that would
merely codify the existing program or any other program that collected bulk
data on such a massive scale regarding individuals with no suspected ties to ter-
rorism or criminal activity. While new legislation could provide clear statutory
authorization for a program that currently lacks a sound statutory footing, any
new bulk collection program would still pose grave threats to privacy and civil
liberties.55

In essence, this is what is occurring in Canada. While much remains opaque,

the evidence suggests that the proposal for a new legal regime for bulk data
acquisition will essentially codify former practices of surveillance on a massive
scale regarding individuals who present no basis for any suspicion of wrong-
doing. In short, Canada is addressing the process of bulk data acquisition but
failing to meaningfully assess the substance of the rights violations at issue.
 Gleanings from the Security Intelligence Review Committee 165

Notes
1 For further discussion of the security intelligence practices and agencies in Canada, see the
section “Situating Security Intelligence” in the Introduction to this volume, pages 6–13.
2 Simon Davies, A Crisis of Accountability: A Global Analysis of the Impact of the Snowden
Revelations (Privacy Surgeon, 2014), 24.
3 Canadian Charter of Rights and Freedoms, Part 1 of the Constitution Act, 1982, being
Schedule B to the Canada Act 1982 (UK), 1982, c 11, s 7; British Columbia Civil Liber-
ties Association, Notice of Civil Claim, Supreme Court of British Columbia, 2013, http://
bccla.org/wp-content/uploads/2013/10/2013-10-22-Notice-of-Civil-Claim.pdf.
4 Big Brother Watch and others v United Kingdom (Applications nos 58170/13, 62322/14
and 24960/15) (2018), ECHR 722; Rebecca Hill, “Bulk Surveillance Is Always Bad, Say
Human Rights Orgs Appealing against Top Euro Court,” The Register, 12 December
2018.
5 Canada, Bill C‑59, An Act respecting national security matters, 1st Sess, 42nd Parl, 2017
(first reading 20 June 2017) [Bill C‑59].
6 X (Re), 2016 FC 1105.
7 Canadian Security Intelligence Service Act, RSC 1985, c C‑23 [CSIS Act].
8 Canada, Security Intelligence Review Committee, Annual Report 2015–2016: Maintain-
ing Momentum (Ottawa: Public Works and Government Services Canada, 2016).
9 Privacy International v Secretary of State for Foreign and Commonwealth Affairs & others,
[2017] UKIPTrib IPT_15_110_CH (UK).
10 CSIS Act, s 12(1).
11 Canada, Security Intelligence Review Committee, Annual Report 2015–2016, 24.
12 Ibid., 24.
13 Ibid., 24–25.
14 Ibid., 25.
15 Ibid.
16 Ibid., 24.
17 Ibid.
18 Ibid., 23–24.
19 X (Re), paras 37, 41–42.
20 Ibid., para 79.
21 Anti‑terrorism Act, 2015, SC 2015, c 20.
22 Security of Canada Information Sharing Act, SC 2015, c 20, s 2 [SCISA]. SCISA was later
amended and renamed Security of Canada Information Disclosure Act [SCIDA].
23 Ibid., s 2.
24 Public Safety Canada, National Security Consultations: What We Learned Report
(Hill+Knowlton Strategies, 2017), 8.
25 Ibid., 4.
26 Canada, Security Intelligence Review Committee, Annual Report 2016–2017: Accelerat-
ing Accountability (Ottawa: Public Works and Government Services Canada, 2017), 22.
27 CSIS Act, s 12(1).
28 Canada, Security Intelligence Review Committee, Annual Report 2016–2017, 22–23.
29 Privacy Act, RSC 1985, c P‑21.
30 Canada, Security Intelligence Review Committee, Annual Report 2016–2017, 23.
31 Income Tax Act, RSC 1985, c 1 (5th Supp), s 9(b)(i).
32 Canada, Security Intelligence Review Committee, Annual Report 2016–2017, 24.
33 Ibid., 24.
34 Ibid., 24–25.
35 Canada, Security Intelligence Review Committee, Annual Report 2015–2016, 25.
 166 Micheal Vonn

36 Canada, Security Intelligence Review Committee, Annual Report 2016–2017, 2.

37 Privacy and Civil Liberties Oversight Board, Report on the Telephone Records Program
Conducted under Section 215 of the USA PATRIOT Act and on the Operations of the For-
eign Intelligence Surveillance Court (2014).
38 Ibid., 62.
39 Ibid.
40 Ibid.
41 Ibid., 65.
42 Jim Sensenbrenner, “How Obama Has Abused the Patriot Act,” Los Angeles Times, 19
August 2013.
43 See, for example, the comprehensive report on this subject from the Council of Europe:
Consultative Committee of the Convention for the Protection of Individuals with Regard to
Automatic Processing of Personal Data (Directorate General of Human Rights and Rule
of Law, 27 June 2016).
44 Bill C-59, cl 92.
45 Ibid., cl 97.
46 Ibid.
47 Ibid.
48 Ibid.
49 Ibid., cl 76.
50 Canada, Security Intelligence Review Committee, Annual Report, 2015–2016, 23–24.
51 X (Re), para 42.
52 Bill C‑59, cl 97.
53 Elizabeth Goitein, The New Era of Secret Law (New York: Brennan Center for Justice at
New York University School of Law, 2016).
54 Privacy and Civil Liberties Oversight Board, Report on the Telephone Records Program,
168.
55 Ibid., 169.
9
Bill C-59 and the Judicialization of
Intelligence Collection
Craig Forcese

Canada is remaking its national security law through Bill C-59.1 This law
project constitutes the country’s largest national security law reform since 1984
and the creation of the Canadian Security Intelligence Service (CSIS). And with
its 150 pages of complicated legislative drafting, C-59 follows the pattern in other
democracies of codifying once murky intelligence practices into statute.
On the cusp of being enacted in Parliament at the time of this writing, the
bill responds to quandaries common to democracies in the early part of the
twenty-first century. Among these questions: How broad a remit should intel-
ligence services have to build pools of data in which to fish for threats? And
how best can a liberal democracy structure its oversight and review institutions
to guard against improper conduct by security and intelligence services in this
new data-rich environment?
This chapter examines how Bill C-59 proposes to reshape the activities of both
CSIS and the Communications Security Establishment (CSE) in fashions
responding to these dilemmas. Specifically, it highlights C-59’s proposed changes
to CSIS’s capacity to collect bulk data as part of its security intelligence mandate,
and also the new oversight system proposed for CSE’s foreign intelligence and
cybersecurity regimes. The chapter examines the objectives motivating both
sets of changes and suggests that in its architecture C-59 tries to mesh together
the challenges of intelligence in a technologically sophisticated, information-
rich environment with privacy protections derived from a simpler age but
updated to meet new demands.

The Evolution of Intelligence Law

In 2008, then CSIS director Jim Judd raised concerns about “the judicialization
of intelligence.”2 His focus was on anti-terrorism and the propensity after 9/11
for greater overlap between intelligence and police investigations. But his expres-
sion has a broader meaning, capturing a core disconnect between a covert
intelligence function and the more overt culture of law and legal accountability.
Intelligence needs are fluid. Law is rigid. Intelligence needs are immediate and
exigent. Law can be laborious.
Nevertheless, law has inevitably encroached on intelligence. Wiretap powers
constitute a case in point. In the 1950s, national security domestic intercept
168 Craig Forcese

warrants were issued by the prime minister as an exercise of discretionary power

under a Korean War–period instrument, the Emergency Powers Act.3 This provided
vague statutory standards for such surveillance4 and certainly did not oblige
independent judicial oversight.5 In 1954, the deputy minister of justice issued the
so-called Varcoe opinion, concluding that telephone communications could
be intercepted for security purposes with a search warrant issued by a justice of
the peace, under the then Official Secrets Act.6 Twenty years later, in 1974, Parlia-
ment legislated the Protection of Privacy Act of 1974, now found as Part VI of the
Criminal Code.7 Part VI governs the intercept of “private communications” – in
essence, any oral communication or telecommunication made or received in
Canada in which the originator has a reasonable expectation of privacy.8 Intercep-
tion almost always requires advance authorization by a judge. And in 1984, Parlia-
ment followed this pattern in the CSIS Act and built CSIS search-and-seizure
powers around a judicial warrant process, with the practical result that CSIS
intercepts of private communications also require judicial preauthorization.9
The same year, the Supreme Court of Canada decided Hunter v Southam,10
still the leading case on section 8 of the Canadian Charter of Rights and Free-
doms. Section 8 provides, simply, that “everyone has the right to be secure against
unreasonable search or seizure.” Since Hunter, section 8 has protected against
unreasonable invasions of reasonable expectations of privacy. The gold standard
for a reasonable search in a criminal law context has been: “(1) prior authoriza-
tion; (2) granted by a neutral and impartial arbiter capable of acting judicially;
(3) based on reasonable and probable grounds to believe that an offence has
been committed and there is evidence to be found at the place to be searched.”11
Since then, in cases like the Federal Court of Appeal’s decision in Atwal,12
through to the Federal Court’s recent decision in the In the Matter of Islamist
Terrorism case,13 intelligence search-and-seizure expectations (in these cases,
conducted by CSIS) have been placed on a constitutional footing largely indis-
tinguishable from that of criminal law. In the Islamist Terrorism case, for instance,
the court pointed to the injuries that might befall an individual subject to an
intelligence investigation and held that “the investigation of threats to the secur-
ity of Canada ... and the collection of information or intelligence ... [is] closer
in nature to the purposes of criminal legislation than to the purposes underlying
the types of public welfare, regulatory or economic legislation in respect of
which low expectations of privacy have been found to exist.”14

Technological Conundrums in Information-Rich Environments

This “judicialization of intelligence” model has, to date, depended on judges
serving as gatekeepers, obliging the state to show reasonable grounds to believe
(or in some instances, suspect) that the information collected is tied to (in the
 Bill C-59 and the Judicialization of Intelligence Collection 169

case of the police) an offence or (in the case of CSIS) a threat to the security of
Canada. Warrants also oblige a large measure of specificity, targeting individuals
(or perhaps classes of individuals) who themselves are linked to these offences
or threats.
Oversight has, therefore, been front-ended, in advance of the intercept or
collection. And authorized information collection has then been relatively
focused. To date, therefore, the judicialization model has not accommodated
“bulk powers,” an expression borrowed from the United Kingdom. A bulk power
is one that allows intelligence agencies access to a large quantity of data, most
of which is not associated with existing targets of investigation. In other words,
it is the mass access to data from a population not itself suspected of threat-
related activity. The commonplace example, since Edward Snowden’s revelations,
is Internet or telephony metadata for entire populations of communications
users. But bulk powers can also involve content, and not just the metadata sur-
rounding that content.
Bulk powers are controversial – they are the heart of the post-Snowden pre-
occupations. They inevitably raise new questions about privacy and, in the
Canadian context, Charter rights, not least because bulk powers are irreconcil-
able with the requirements of classic warrants. There is no specificity. By def-
inition, bulk powers are not targeted; they are indiscriminate.
However, whether bulk powers amount to “dragnet” or “mass” surveillance
is a closely contested issue. Collection does not – and likely cannot, given
resource constraints – mean real-time, persistent observation. It does mean,
however, a sea of data that may then be queried and exploited. The distinction
between querying of collected and archived data and a permanent, panoptic
form of surveillance may be a distinction without a difference for members of
the public and privacy advocates, but it is one that David Anderson, former UK
Independent Reviewer of Terrorism Legislation, viewed as compelling in his
2016 report on bulk powers:

It should be plain that the collection and retention of data in bulk does not
equate to so-called “mass surveillance.” Any legal system worth the name will
incorporate limitations and safeguards designed precisely to ensure that access
to stores of sensitive data (whether held by the Government or by communica-
tions service providers [CSPs]) is not given on an indiscriminate or unjustified
basis.15

Put another way, surveillance means “watching,” and not “potential watching.”
And “potential” is controlled by safeguards that mean collection does not morph
seamlessly into watching. This is the philosophy that animated the United
170 Craig Forcese

Kingdom’s 2016 Investigatory Powers Act (IPA),16 and now is reflected also in
Bill C-59. Under the IPA, the world of bulk powers can be divided into bulk
interception, bulk equipment interference, bulk acquisition, and bulk personal
datasets. Canada’s Bill C-59 addresses issues relating to bulk interception and
bulk personal datasets. The bill does two things of note: for both CSE and CSIS,
it superimposes new quasi-judicial controls on collection of certain sorts of
information. For CSIS, it also creates judicial controls on retention, exploitation,
and querying of at least some sorts of information.

Bulk Interception by CSE

Bulk interception is what it sounds like: the collection of transiting communica-
tions passing through communications providers or otherwise through the
ether. Canadian law permits bulk collection by the Communications Security
Establishment. Among other things, CSE is Canada’s “signals intelligence”
service, charged with acquiring foreign intelligence from the “global informa-
tion infrastructure,” that is, electronic emissions and now also information from
other technology networks such as the Internet. It also has a cybersecurity
mandate: “to provide advice, guidance and services to help ensure the protection
of electronic information and of information infrastructures of importance to
the Government of Canada.”17
In conducting its foreign intelligence and cybersecurity function, CSE is to
cast its eyes outward, beyond Canada: it cannot direct its activities at Canadians
or any person in Canada. It also must take steps to protect the privacy of Can-
adians.18 This privacy requirement responds to a technical problem: the inevit-
ability of incidental acquisition of Canadian information. In acquiring
information from the global information infrastructure or performing its
cybersecurity role, CSE cannot know in advance whether Canadian or Canadian-
origin data will be swept within its acquisition activities.
Under the current National Defence Act, therefore, CSE may (and does) obtain
special “ministerial authorizations” under which it might inadvertently acquire
Canadian “private communications” within the meaning of Part VI of the
Criminal Code – essentially “telecommunications” with a nexus to Canada.19
There are presently three authorizations for foreign intelligence and one for
cybersecurity. The authorizations are broad – involving classes of activities and
not individual activities.
The current rules suffer from two key problems, however. First, technology
has evolved considerably since the original enactment of CSE’s powers in 2001.
Now, the focus is on metadata – the information that surrounds a communica-
tion, such as email addresses, routing information, duration and place of cellular
calls, and the like. The government’s view has been that these metadata are not
 Bill C-59 and the Judicialization of Intelligence Collection 171

a component of a private communication for which a ministerial authorization

must be sought – a conclusion dependent on a narrow reading of the definition
of “telecommunication” in the Interpretation Act.20
Second, whether or not CSE obtains a ministerial authorization, there are
evident constitutional issues under section 8 of the Charter, ones anticipated
many years ago but never resolved.21 As noted, section 8 usually means that
authorities may interfere with a reasonable expectation of privacy only under
a warrant authorized in advance by an independent judicial officer, that is,
someone able to act judicially.22 As suggested, wiretaps of communications must
be authorized by judicial warrant in almost all circumstances.
Whatever else he or she may be, the minister of defence is not an independent
judicial officer, and yet under the current act it is his or her authorization that
permits CSE’s collection of private communication, a form of data in which the
originator has, by definition, a reasonable expectation of privacy. For their part,
metadata do not include the content of a communication. But pieced together
(and even alone) they can be remarkably revealing of a person’s habits, beliefs,
and conduct. Metadata are often information in which there is a reasonable
expectation of privacy, especially when compiled as a mosaic.23 This conclusion
is supported, if not quite decided, by the Supreme Court of Canada’s decision
in R v Spencer, holding that even the most innocuous of nameplate information
tied to a digital trail – subscriber information associated with an IP address –
attracts constitutional protection.24
The risk, therefore, is that CSE now acquires information that enjoys consti-
tutional protection without going through the independent judicial officer
process (or anything approximating the process) that the Constitution requires
before the state acquires this information. That is, at core, the issue in a consti-
tutional challenge brought by the British Columbia Civil Liberties Association
(BCCLA) to CSE’s law and metadata practices.25
The fact that CSE’s acquisition of private communications and metadata is
incidental to its outward-looking mandates should not matter, since the collec-
tion of at least some constitutionally protected information is foreseeable and
inevitable. Canada’s constitutional standards for search and seizure do not say,
“You are protected against unreasonable search and seizures, except when the
search and seizure is simply a predictable, foreseen accident stemming from
other activities.” Put another way, the fact that information in which Canadians
have a reasonable expectation of privacy is incidentally but foreseeably (rather
than intentionally) collected by the state should not abrogate the constitutional
right (although I accept it may shape the precise protections that the Charter
will then require; see below). As the government itself has now acknowledged,
“because the authority to acquire private information through the GII [global
172 Craig Forcese

information infrastructure] has the potential to interfere with privacy interests,

it may engage section 8.”26

Bill C-59 and a Hunter Approximation

Bill C-59 responds to these constitutional concerns primarily through a new
system of quasi-judicial preauthorizations for CSE’s recrafted foreign intelligence
and cybersecurity mandates. Whereas under the current system, the defence
minister alone issues an authorization, C-59 anticipates a quasi-judicial intel-
ligence commissioner (IC) who will review the ministerial authorization before
its execution. To be clear: this is not a warrant. It will lack specificity. It will be
issued for classes of activities, not specific activities or operations. It is review
of the reasonableness of a ministerial authorization, not the more hands-on
warrant process. Moreover, the intelligence commissioner is not a sitting judge
but rather a former judge occupying an administrative office at arm’s length
from executive government, and is not clothed with as robust independence as
a court. Will this meet Hunter’s section 8 standards? Some have argued it will
not.27 I am inclined to suggest it will pass constitutional muster, because the
warrant cookie cutter cannot possibly apply to a form of bulk intercept in which
intercept of section 8 rights-bearer communications is entirely incidental, and
not targeted.
From past cases, we know that section 8 does not require an actual judge – but
instead a person capable of acting judicially.28 Nor does it oblige one-size-fits-all
warrants for all forms of search and seizure. As the Federal Court of Appeal
decided (in applying different criteria to a CSIS warrant than to a police wiretap):
“To conclude ... a different standard should apply where national security is
involved is not necessarily to apply a lower standard but rather one which takes
account of reality” (emphasis added).29 And so in that case, it made no sense to
require CSIS to show that it was investigating a criminal offence – its mandate is
to investigate threats to the security of Canada. This suggests that there is at least
some flexibility in design, so long as we preserve the core essentials of the section
8 jurisprudence: advance authorization by an independent judicial officer.

CSIS Datasets and Bulk Personal Datasets

Turning to domestic-facing bulk powers, Bill C-59 creates new bulk powers for
CSIS. Under the bill, CSIS would be empowered to collect bulk personal datasets.
The British understanding of this expression also describes what is at issue in
Canada:

A bulk personal dataset includes personal data relating to a number of individu-

als, and the nature of that set is such that the majority of individuals contained
 Bill C-59 and the Judicialization of Intelligence Collection 173

within it are not, and are unlikely to become, of interest to the intelligence ser-
vices in the exercise of their statutory functions. Typically these datasets are very
large, and of a size which means they cannot be processed manually.30

The C-59 approach to bulk personal datasets is a response, in part, to the Federal
Court’s 2016 decision on what was known as “ODAC.”31 But it also responds to
a broader concern about the ambit of CSIS’s threat investigation mandate.32 That
mandate is anchored in section 12 of the CSIS Act.
Under its section 12 mandate, CSIS collects, to the extent it is strictly neces-
sary, and analyzes and retains information and intelligence on activities it has
reasonable grounds to suspect constitute threats to the security of Canada. This
passage has several “magic words”: “to the extent that it is strictly necessary”;
“reasonable grounds to suspect”; and “threats to the security of Canada.”
“Threats to the security of Canada” is the only passage defined in the CSIS
Act (in section 2). “Reasonable grounds to suspect” has a generally well-
understood meaning: “suspects on reasonable grounds” is a suspicion based
on objectively articulable grounds that may be lower in quantity or content
than the requirement of reasonable belief, but must be more than a subjective
hunch.33 It amounts to a possibility the threat exists, based on cogent evidence
(and not simply a hunch).
Under section 12, CSIS commences an investigation on this standard. But
where the means of that collection are sufficiently intrusive to trigger section 8
of the Charter or the Part VI Criminal Code prohibition against unauthorized
intercept of private communications (for instance, a wiretap), it must get a
Federal Court warrant. A judge will issue a warrant only if persuaded that CSIS
has reasonable grounds to “believe” that it is required to investigate threats to
the security of Canada.
“Reasonable grounds to believe” is a higher standard than the reasonable
grounds to suspect standard that must be met for CSIS to begin an information
collection investigation under section 12. Sometimes called “reasonable and
probable grounds” in the constitutional case law, reasonable grounds to believe
is less demanding than the criminal trial standard of “beyond a reasonable
doubt.” Instead, it is defined as a “credibly-based probability” or “reasonable
probability.”34
CSIS obtains warrants in a closed-court (i.e., secret) process in which only
the government side is represented. The warrants can, and often do, impose
conditions on CSIS investigations. There are templates for standard warrant
applications. These templates are occasionally updated, a process that requires
CSIS to apply to the Federal Court. The 2006 ODAC case came about through
a belated updating process.
174 Craig Forcese

Operational Data Analysis Centre (ODAC)

CSIS collects much data in its section 12 investigations, including those aspects
of the investigation conducted pursuant to a warrant. Not unreasonably, it wants
to keep these data to pool them in a manner that it can then search to further
investigations in the future. And so, it created ODAC in 2006. It did not tell the
Federal Court about ODAC, at least not in any real concrete manner.
This is important because ODAC was pooling information collected via war-
rant. And that information included not only content and metadata produced
by an investigative target’s own communications (the collection of which was
authorized by warrant), but also so-called associated data. As the court defined
it, associated data are data “collected through the operation of the warrants
from which the content was assessed as unrelated to threats and of no use to
an investigation, prosecution, national defence, or international affairs.” Pre-
sumably a lot of these would-be data from third parties – that is, communication-
related information involving non-targets – are swept into the CSIS surveillance
net. For telephony, this might include the speech of the person on the other end
of a conversation, or the accompanying metadata (e.g., telephone number,
geolocation of a cellphone, etc.). For email, this could be content and metadata
totally unrelated to the target’s communication. Email travels in packets across
the Internet and packets bundle unrelated segments of individual emails. And
so, intercepting a target’s emails generally means intercepting all the packets,
and the accompanying content and metadata of other people’s communications
bundled with them.
CSIS chose, in the ODAC, to retain some of this “associated data,” specifically,
the metadata, although not the actual content of the communication. The reten-
tion of these metadata raised a legal issue. For one thing, as noted above, it now
seems certain that at least some metadata are protected by section 8 of the
Charter. In addition, the CSIS Act determines what CSIS can do with the infor-
mation it collects.
In the 2016 ODAC case, the court did not reach the section 8 issue, although
it acknowledged that the matter had been argued before it. Instead it focused
on the CSIS Act issue. And there, the key consideration was whether CSIS can
retain the information it collects through its investigations. On this point, there
are now two answers.
First, as per the Supreme Court of Canada’s holdings in Charkaoui II,35 CSIS
has a constitutional duty to retain information related to its targets, or to threats
to the security of Canada. As the Federal Court summarized this rule: “infor-
mation that is indeed linked to threats to the security of Canada or to the target
of a warrant must be retained in its original state by the CSIS to comply with
the protected rights under section 7 of the Charter.”36 Or in lay terms: CSIS
 Bill C-59 and the Judicialization of Intelligence Collection 175

cannot destroy information collected on targets/threats, because people impli-

cated in those threats may subsequently be subject to legal proceedings that
oblige full government disclosure allowing for a fair trial. And if CSIS has
destroyed the original collected information and (the argument would go)
simply kept a cherry-picking summary, then no fair trial can be had.
Second, this Charkaoui II rule does not apply to information unrelated to the
target or threats – that is, associated data. Charkaoui II was not about associated
data, and so the Federal Court looked to the CSIS Act and concluded as follows:
associated data, by definition, is non-threat-related. It is therefore not something
that is “strictly necessary” to the investigation of threats to the security of Canada.
Thus, collecting it is something CSIS should not be in the business of doing.
Technology means it cannot help but collect it while undertaking its bona fide
“strictly necessary” collection of threat-related information. And so, court war-
rants allow for this incidental collection. But authorizing incidental collection
does not bless indefinite retention. And indeed, indefinite retention is not
something any court could authorize without effectively usurping the “strictly
necessary” standard found in section 12. And so CSIS retention of the associated
metadata was illegal, because of the way the CSIS Act was drafted.

The Policy Issue

In this manner, the present CSIS Act ties information collection, retention, and
analysis to a narrow band of threat investigations. The analogy is a fishing craft
that must use a sonar with limited range and scope. Bill C-59 proposes relaxing
these limitations and permitting the netting of more data, of a sort that would
exceed the “strictly necessary” standard.
A spy service fishing in more ocean is, in some eyes, the stuff of Big Brother
and nightmares. On the other hand, an intelligence service that cannot have
access to the ocean when performing its functions is also not likely to perform
its functions very well. And there is a lot of ocean out there in the digital era.
As the government has urged:

Today’s threats to Canada’s national security are fast, complex and dynamic, and
threat actors are highly connected and mobile. The ease of movement across
international borders and spread of social media networks and modern com-
munications technology can be used by individuals and groups seeking to harm
Canada. This creates some very real challenges for CSIS.37

The dilemma lies in reconciling oceans full of data generated by innocents with
the intelligence function of clearing the fog of uncertainty and revealing not
just the known threats but also the unknown threats.
176 Craig Forcese

Bill C-59 proposes its solution: judicializing a more sweeping fish-detecting

sonar through a dataset system. Datasets are currently defined as collections of
“information stored as an electronic record and characterized by a common
subject matter.” More importantly, this is information that CSIS would not be
able to collect, retain, or analyze under its regular mandates because, for instance,
it will not know they contain information significant to a threat investigation
without searching them. Under C-59, new rules allow for the collection of
publicly available, foreign, and Canadian datasets, subject to an array of checks
and balances that vary in strictness. The most robust accountability regime
concerns Canadian datasets. Here, the minister of public safety and the intel-
ligence commissioner may approve collection of “classes” of Canadian datasets –
datasets primarily comprising Canadian information. Once ingested, the bill
then authorizes limited, firewalled vetting by CSIS to assess the usefulness of
the dataset. And then any subsequent retention for actual use must be approved
by the Federal Court, which is empowered to impose conditions on that sub-
sequent use. Querying and exploitation of the retained dataset must thereafter
generally be done with an eye to CSIS’s mandates – for example, only where
strictly necessary in performance of CSIS’s section 12 mandate.
The system is enormously complex, sparking skepticism among those who
look askance at the intrusion of law into intelligence. Privacy advocates, for
their part, may question any expansion of the state’s ability to collect personal
information. The system does constitute a quid pro quo, however: CSIS’s trad-
itional section 12 constraints are loosened to the extent that it may consume a
broader ocean of data, but retention (at least for Canadian datasets) requires
judicial supervision.38 This system recognizes that privacy interests extend
beyond the point of collection and include retention and use. In so doing, it
short-circuits inevitable frontier section 8 issues; namely, does section 8 attach
to the big data analysis of information, the individual bits of which trigger no
reasonable expectation of privacy (and thus, the initial collection of which
attracts no constitutional protection)? C-59 anticipates (and pre-empts) this
issue by superimposing independent judicial authorization guiding and con-
ditioning the big data analysis.
By bundling the issue of collection, retention, and use of electronic informa-
tion in the hands of CSIS into a specialized model, C-59 creates a system much
more demanding than the now antiquated rules found in the federal Privacy
Act. The latter includes no independent, direct oversight of personal type of
information collected, retained, and used by the federal government, instead
depending on periodic review audits by the privacy commissioner.
The C-59 system compares favourably, therefore, with other controls on
government use of information. That said, I have one lingering doubt about this
 Bill C-59 and the Judicialization of Intelligence Collection 177

constitution proofing of CSIS dataset retention. The Federal Court retention

authorization (and the IC approval of dataset classes) is limited to “Canadian
datasets.” Datasets primarily comprising information on foreign individuals
outside Canada are processed under a separate regime, in which the IC decides
the retention issue. Since Charter privacy rights are largely geographic in scope,
this more relaxed system is probably justifiable. However, a third class of datasets
comprises personal information “publicly available at the time of collection.”
Publicly available datasets are not subject to any independent oversight regime.
It matters, therefore, into which bucket information is placed. Some informa-
tion may be publicly available (e.g., hacked private information dumped on the
Internet) but still raises considerable privacy implications. CSIS has indicated
before Parliament that it will not treat hacked information as publicly available.
This is a policy decision, however, not one required by law. Should CSIS adopt
an underinclusive policy that steers information about which a Canadian still
has a reasonable expectation of privacy into the publicly available bucket, the
constitutionality of this practice would be suspect. This would then become a
high-risk practice, and raise the prospect of an ODAC controversy rerun.
The obvious solution would be to amend Bill C-59 to define “publicly available”
as excluding “information in which a Canadian or person in Canada retains a
reasonable expectation of privacy.” This would have the effect of steering such
information into the “Canadian dataset” bucket, with its more constitutionally
robust oversight system. Since that amendment was not made in Parliament,
the minister of public safety should, at minimum, issue a ministerial direction
with the same effect – and ensure that this direction and any of its successors
are public to create confidence in otherwise opaque internal procedures within
CSIS.

Conclusion
Privacy is among the least absolute rights in international and Canadian human
rights law. It has always been about equilibrium, from its inception as a common
law concept in the eighteenth century. Balancing has depended, in the Canadian
law tradition, on advance oversight by judicial officers. In relation to CSE,
changes in technology have placed that agency out of alignment with this
expectation. Bill C-59 tries to restore a more traditional pattern, albeit in cir-
cumstances where a classic judicial warrant model would prove unworkable.
Meanwhile, CSIS has laboured with an act designed for an analogue period.
It has become an intelligence service largely incapable of fishing in an electronic
sea. The challenge is to permit reasonable fishing but not dragnetting. Bill C-59
attempts to strike this balance by superimposing a judge, not to police CSIS’s
dataset fishing net but rather to determine what information captured within
 178 Craig Forcese

the net may be retained and analyzed. We are right to be wary of such a system,
since it depends on close adherence to a complicated set of checks and balances.
On the other hand, those checks and balances cannot become so burdensome
that intelligence services are left to obtain, essentially, a warrant to obtain a
warrant.
Put another way, C-59 seeks balance. Not everyone will agree, but in my view
(and subject to my doubts about publicly available datasets), C-59 succeeds
reasonably well in reconciling the “nightwatchman” role of the state’s security
services with the individual’s right to be left alone.

Notes
1 Canada, Bill C‑59, An Act respecting national security matters, 1st Sess, 42nd Parl, 2017,
http://www.parl.ca/DocumentViewer/en/42-1/bill/C-59/first-reading. This chapter deals
with C-59 as it existed at the time of writing: after first reading in the House of Commons.
2 Canadian Security Intelligence Service (CSIS), “Remarks by Jim Judd, Director of CSIS,
at the Global Futures Forum Conference in Vancouver,” 15 April 2008, Internet Archive,
https://web.archive.org/web/20081012174022/http://www.csis-scrs.gc.ca:80/nwsrm/
spchs/spch15042008-eng.asp.
3 RSC 1952, c 96.
4 Under the act, the minister of justice could require a communications agency to pro-
duce any communication “that may be prejudicial to or may be used for purposes that
are prejudicial to the security or defence of Canada.” David C. McDonald, “Electronic
Surveillance – Security Service and C.I.B.,” in Reports of the Commission of Inquiry Con-
cerning Certain Activities of the Royal Canadian Mounted Police (Ottawa: Privy Council
Office, 1981), vol 2–1, 158, http://publications.gc.ca/collections/collection_2014/bcp
-pco/CP32-37-1981-2-1-2-eng.pdf.
5 The document is archived as Canada, “Privy Council Wiretap Order (St-Laurent Gov-
ernment)” (unpublished document, 1951), http://secretlaw.omeka.net/items/show/69,
and was obtained by Dennis Molinaro under the Access to Information Act.
6 McDonald, Reports of the Commission of Inquiry, vol 2–1, 158.
7 RSC, 1985, c C-46.
8 Ibid., s 183.
9 RSC, 1985, c C-23, s 21.
10 [1984] 2 SCR 145.
11 Justice Canada, “Section 8 – Search and Seizure,” https://www.justice.gc.ca/eng/csj-sjc/
rfc-dlc/ccrf-ccdl/check/art8.html.
12 Atwal v Canada, [1988] 1 FC 107 (FCA).
13 2017 FC 1047.
14 Ibid., para 171.
15 David Anderson, Report of the Bulk Powers Review (London: Williams Lea Group,
2016), para 1.9, https://terrorismlegislationreviewer.independent.gov.uk/wp-content/
uploads/2016/08/Bulk-Powers-Review-final-report.pdf.
16 2016, c 25.
17 National Defence Act, RSC, 1985, c N-5, ss 274.61 and 273.64(1) [NDA]. These mandates
are preserved in Bill C-59, Part 3, Communications Security Establishment Act (CSE Act),
ss 2, 16, 17, and 18 [CSE Act].
18 NDA, s 273.64(2); CSE Act, ss 17, 18, 23, and 25.
 Bill C-59 and the Judicialization of Intelligence Collection 179

19 NDA, ss 273.65, 273.61; Criminal Code, RSC 1985, c C-46, s 183.

20 Interpretation Act, RSC, 1985 c I-21, s 35 (defining “telecommunications” as “the emis-
sion, transmission or reception of signs, signals, writing, images, sounds or intelligence of
any nature by any wire, cable, radio, optical or other electromagnetic system, or by any
similar technical system” [emphasis added]).
21 See, e.g., Stanley A. Cohen, Privacy, Crime and Terror: Legal Rights and Security in a Time
of Peril (Markham, ON: LexisNexis Butterworths, 2005), 232.
22 Hunter v Southam, [1984] 2 SCR 145, 162.
23 For a fuller discussion of metadata and privacy rules, see Craig Forcese, “Law, Loga-
rithms and Liberties: Legal Issues Arising from CSEC’s Metadata Collection Initiatives,”
in Law, Privacy and Surveillance in Canada in the Post-Snowden Era, edited by Michael
Geist (Ottawa: University of Ottawa Press, 2015), https://ssrn.com/abstract=2436615.
24 2014 SCC 43.
25 See BCCLA, “BCCLA v. CSEC: Stop Illegal Spying,” 23 May 2014, https://bccla.org/
our_work/stop-illegal-spying/. (In the interest of full disclosure: on behalf of BCCLA, I
provided factual background information for use by the court in that proceeding.)
26 Justice Canada, “Charter Statement – Bill C-59: An Act respecting national security mat-
ters,” 20 June 2017, http://www.justice.gc.ca/eng/csj-sjc/pl/charter-charte/ns-sn.html.
27 See, e.g., Christopher Parsons, Lex Gill, Tamir Israel, Bill Robinson, and Ronald Deib-
ert, “Analysis of the Communications Security Establishment Act and Related Provisions
in Bill C-59 (An Act respecting national security matters), First Reading (December 18,
2017) (Citizen Lab/Canadian Internet Policy and Public Interest Clinic report, Decem-
ber 2017), 18, https://citizenlab.ca/wp-content/uploads/2018/01/C-59-Analysis-1.0.pdf.
28 Hunter v Southam, 162.
29 Atwal v Canada, para 35.
30 United Kingdom, Home Office, Security and Intelligence Agencies’ Retention and Use of
Bulk Personal Datasets: Draft Code of Practice (2016), para 2.2, https://www.gov.uk/government/
uploads/system/uploads/attachment_data/file/557860/IP_Bill_-_Draft_BPD_code_of_
practice.pdf.
31 X (Re), 2016 FC 1105.
32 See CSIS, “Amendments to the CSIS Act – Data Analytics,” 20 June 2017, https://www.
canada.ca/en/security-intelligence-service/news/2017/06/amendments_to_thecsisact
-dataanalytics.html. (“When it was written, the CSIS Act could not have anticipated the
technological changes of the last 30 years. The Federal Court acknowledged the age of
the CSIS Act and that it may not be keeping pace with changing technology.”)
33 R v Kang-Brown, 2008 SCC 18.
34 R v Debot, [1989] 2 SCR 1140. In the administrative law context, courts have described it
as a bona fide belief of a serious possibility, based on credible evidence. Chiau v Canada
(Minister of Citizenship and Immigration), [2001] 2 FC 297 (FCA).
35 Charkaoui v Canada (Citizenship and Immigration), 2008 SCC 38.
36 X (Re), para 195.
37 CSIS, “Amendments to the CSIS Act.”
38 The Privacy Commissioner makes (essentially) this same point, and offered no recom-
mendations for changes to the CSIS dataset regime in Bill C-59. See Canada, Office of the
Privacy Commissioner of Canada, “Brief to the Standing Committee on Public Safety
and National Security,” 5 March 2018, 12.
10
The Challenges Facing Canadian Police in
Making Use of Big Data Analytics
Carrie B. Sanders and Janet Chan

Advanced information technology is playing a progressively more import-

ant role in policing and security. This is in part driven by the growing complexity
of the crime and security environment, and partly by the lure of new technology
such as big data technology, which promises greater efficiency and innovation.
The advent of big data technologies appears to be a continuation of a trend to
combine statistical thinking and communication technology to present “risk
data.”1 As Lucia Zedner points out, the impact of 9/11 and the changing and
volatile nature of organized crime have put pressure on governments to “think
and act pre-emptively,” leading to an acceleration of the “trajectory towards
anticipatory endeavor, risk assessment and intelligence gathering.”2 Big data
technology has been increasingly deployed by policing and national security
agencies to detect and/or prevent complex crime such as terrorism and organ-
ized crime.3 For example, the apprehension and arrest of the 2013 attempted
Via Rail bombers in Ontario not only illustrate the heightened role information,
data, and analytics play in public safety decision making but also demonstrate
the blurring boundaries between policing and national security intelligence
practices.4 Information and intelligence sharing among the RCMP, the Can-
adian Security Intelligence Service (CSIS), municipal and provincial police
forces in Ontario and Quebec, the US Department of Homeland Security, and
the FBI led to the early apprehension of the attempted Via Rail bombers. Indeed,
Patrick Walsh’s analysis of intelligence frameworks across the “Five Eyes” intel-
ligence enterprises (United States, Britain, Australia, Canada, and New Zealand)
demonstrates how intelligence frameworks are “evolving post 9/11” and
concludes that “each framework ... demonstrates the blurring between what
has been traditionally thought of as a ‘policing and national security
intelligence.’”5
The pursuit of security within policing is clearly evident in the adoption of
new policing initiatives – such as intelligence-led and predictive policing – that
rely upon “the collection and analysis of information to produce an intelligence
end product designed to inform law enforcement decision making at both the
tactical and strategic levels.”6 As Greg Ridgeway argues, “the demand for intel-
ligence on offenders incentivized the creation of information technology that
could feed the demand ... At the same time, with ready access to information,
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 181

police have an opportunity to make links and connections that used to be more
labor intensive.”7
New technology such as big data analytics promises opportunities for police
services to work more efficiently and effectively by identifying and predicting
crime patterns in large datasets with the hope that such practices will enable
“the opportunity to enter the decision cycle of [their] ... adversaries’ [sic] in
order to prevent and disrupt crime.”8 For example, the Vancouver Police Depart-
ment has implemented a “city-wide ‘Predictive Policing’ system that uses
machine learning to prevent break-ins by predicting where they will occur
before they happen – the first of its kind in Canada.”9 The system is said to have
80 percent accuracy in directing officers to locations of break-ins.10 While
Vancouver is the first service in Canada to implement predictive policing soft-
ware, other services are implementing technologies that enable them to collect
and store large sets of data.11 For example, Calgary Police Services, the Royal
Canadian Military Police, the Ontario Provincial Police, and Winnipeg Police
Services have all implemented Mobile Device Identifier technology (which
mimics cellular towers) that enables them to intercept cellphone data.12 Such
examples demonstrate how big data and data analytics are being integrated in
Canadian police services.
At present, the use of big data in policing has been largely limited to the col-
lection and storage of DNA information, mass surveillance, and predictive
policing.13 While there is a lot of speculation about the possibilities (both good
and bad) of these technologies, there is little empirical research available on
how big data is used and experienced in policing.14 Ridgeway identifies specific
applications of big data in American policing, such as pushing real-time infor-
mation to officers, tracking police locations, performance measurement, and
predictive policing, but like others, notes that the “evidence so far is mixed on
whether police can effectively use big data.”15 To date, much of the available
literature on big data and policing is focused on predictive policing and origin-
ates in the United States, United Kingdom, and Australia – where police intel-
ligence frameworks, policies, and practices differ.16 For example, the United
Kingdom has a National Intelligence Model (NIM) that provides police services
with a standardized approach for gathering and analyzing intelligence for
informing strategic and tactical decision making.17 Canada, however, does not
have a standardized model or approach to intelligence work, so significantly
less is known about the use of technology and data science in the Canadian
context.
This chapter presents empirical data on the challenges facing Canadian police
in making effective use of technologies and data science, including big data
technology, for intelligence practices. While Canadian police services are
182 Carrie B. Sanders and Janet Chan

adopting data analytic practices, the extent of their integration and use varies
signficantly across the country. Some large services in Canada, for example, are
working actively to implement an organizational intelligence framework that
facilitates the adoption of new technologies and standardized analytic practices
throughout their services, while others, particularly smaller municipal services,
are in the early phases of technological adoption and appropriation that facilitate
data science practices. Drawing directly from interviews with thirteen Canadian
police services, we identify technological, organizational, and cultural challenges
for the integration and effective uptake of big data technologies. It is important
to note that, similar to the variance in the integration and adoption of data
science practices across Canadian services, the extent to which services experi-
ence these challenges will also vary.

Police Technology and Big Data: A Review of the Literature

Police technologies and data science – particularly big data technologies – are
perceived to enhance work efficiency and effectiveness by improving the pre-
dictive capabilities of police services while simultaneously reducing subjectivity
and bias.18 Big data technology is generally defined using the three Vs: high
volume (amount of data), high velocity (speed at which data are generated),
and high variety (diversity of sources and formats).19 More recently, other com-
mon features of big data technologies in policing have been identified, such as
data being “exhaustive” (representing a census of all activity), “relational,” and
of “fine resolution” (which for police means greater precision in identifying the
time and location), and data systems that are “flexible and scalable.”20 Yet, defin-
ing big data in these ways focuses analytical attention on the technological
elements while ignoring the way broader contexts – such as culture and
organizations – shape big data.21 For example, research available on big data ana-
lytics has shown how “analytic platforms and techniques are deployed in pre-
existing organizational contexts ... and embody the purposes of their creators.”22
In order to analytically capture the “social shaping”23 of big data technology,
we use danah boyd and Kate Crawford’s definition of big data as a “cultural,
technological and scholarly phenomenon” that incorporates technology, analysis
and mythology.24 We extend this definition to include a fourth dimension con-
cerning “actionability,” which refers to the “action taken on the basis of the
inferences drawn,” because “the performing of algorithms and calculations,
the drawing of inferences and the taking of action are all important stages in
the use of big data analytics in ... decision-making.”25 The research available on
police use of big data suggests that the capability of policing agencies to take
advantage of big data technologies is uneven.26 Conducting in-depth interviews
with intelligence and security agencies in Australia, Janet Chan and Lyria
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 183

Bennett Moses found that most personnel did not have knowledge of, or first-
hand experience with, big data. When asking their participants about the value
and purpose of big data, they found that most law enforcement and security
personnel focused on the value of these technologies for investigative or crime
disruption purposes rather than for predictive purposes or for understanding
broader crime trends.27 Unlike Chan and Bennett Moses, Sarah Brayne’s ethno-
graphic study on the use of big data surveillance by the Los Angeles Police
Department (LAPD) found that the LAPD increasingly used big data analytics
for predictive rather than reactive or explanatory purposes.28 Further, she found
that big data analytics amplify prior surveillance practices, while also facilitating
fundamental transformations in surveillance activities.29 For example, she found
that the integration of big data technologies, such as “alert-based technologies”
instead of the old query-based technologies, made it possible for the LAPD to
survey an unprecedentedly large number of people – people who would not
normally be found in traditional police records systems.30
Whereas there are only a few empirical studies available on the in situ use of
big data technologies, there are studies that discuss their potential uses. Alex-
ander Babuta, for example, wrote a report exploring the potential uses of big
data analytics in British policing. His report identified four ways in which big
data analytics presently are, or could be, used: (1) predictive crime mapping;
(2) predictive analytics for identifying individuals at risk of reoffending or being
victimized; (3) advanced analytics “to harness the full potential of data collected
through visual surveillance”; and (4) the use of “open-source data, such as that
collected from social media, to gain a richer understanding of specific crime
problems, which would ultimately inform the development of preventative
policing strategies.”31 Yet, like other scholars in the field, he found that the
empirical evidence on the use of big data analytics in these four ways is uneven.32
There have also been studies that identify a number of “fundamental limita-
tions” to the implementation and effective use of big data technologies.33 For
example, Babuta notes that the “lack of coordinated development of technology
across UK policing,” fragmentation of databases and software, lack of organiza-
tional access to advanced analytic tools, and legal constraints governing data
usage impede the successful use of big data.34 In synthesizing the available
research on the uptake and impact of police technology, Chan and Bennett
Moses identify the importance of other factors – apart from technological
capacity, such as police leadership, management of technology, organizational
politics, and cultural resistance – for understanding technological adoption and
use.35 The research of Carrie Sanders, Crystal Weston, and Nicole Schott on the
integration of intelligence-led policing practices in Canada demonstrates how
cultural issues and management of innovation issues are intertwined. For
184 Carrie B. Sanders and Janet Chan

example, in the six Canadian police services studied, they found that the use of
analytic technologies to support intelligence-led policing was more rhetorical
than real. In particular, the “occupational culture of information hoarding ...
has shaped the use and functioning of police innovations.”36 In line with previ-
ous research on the “poorly understood and appreciated” role of crime and
intelligence analysts,37 the lack of knowledge and training about crime analysis
on the part of police managers and officers has left analysts to engage in crime
reporting instead of predictive analytics so that new technologies are used to
support traditional policing practices.38

Methods
The empirical data driving our argument comes from sixty-four semi-structured
interviews with personnel from thirteen Canadian police services. Our sample
includes forty-one crime and intelligence analysts, three senior/lead analysts,
six civilian managers of analytic units, three support personnel (including one
junior support analyst and two policy developers), and eleven sworn officers
who work with, or supervise, crime and intelligence analysts. Most of the inter-
views were conducted face to face, while a small portion (10 percent) were
conducted over the telephone. Interviews were supplemented with participation
at the Massachusetts Association of Crime Analysts training conference (2017);
Canadian chapter of the International Association of Crime Analysts training
conference (2017); International Association of Crime Analysts annual training
conference (2017); South Western Ontario Crime Analysis Network (2015); and
two meetings of the Canadian Association of Chiefs of Police (2016 and 2017).
Interviews and field observations were conducted by the lead author and a
doctoral research assistant. All data were stored and analyzed in NVivo 10, a
qualitative data analysis software program, using a constructivist grounded
theory approach.39 The data were thematically coded by the doctoral research
assistant, with continual collaboration and consultation with the lead author.
The lead author then used writing as an analytical device to make sense of, and
theorize, the data. Through analytic memoing,40 we saw many similarities
between the state of police technology and data science in Canadian policing
and that found in the United States, United Kingdom, and Australia. Of interest
for this chapter are the ways in which technological, organizational, and cultural
contexts create challenges for Canadian police in making effective use of big
data technologies. Such challenges include fragmented technological platforms
and information silos; resources and practical barriers; emphasis on tactical
and operational intelligence (i.e., creating workups on offenders or cases) over
strategic intelligence (identifying future offenders or emerging crime trends);
and the uneven professionalization of analytics and user capacities.
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 185

Fragmented Technological Platforms and Information Silos

Police services work with a variety of information and communication compan-
ies (such as Sunguard, Niche, Versaterm, Palantir), technological platforms,
and datasets (e.g., open-source media, Ministry of Transportation, internal
records management systems, provincial police records systems, national police
records systems, closed-circuit televisions, court databases, etc.). Presently, there
are no uniform policies or guidelines for police services when it comes to
technological adoption and appropriation for intelligence work, leaving each
service to act independently.41 As in the United Kingdom, the highly localized
and devolved governance structure of Canadian policing has left services to
work in relative isolation with autonomy over local policing strategies and
technological adoption.42 As a result, Canadian police services lack a standard-
ized and systematic approach to the collection and analysis of intelligence, which
has left data to be managed across multiple, separate systems that are not always
compatible. Further, many traditional police data warehouses do not have the
capacity to manage all of the structured and unstructured data available to them,
and, for many, their warehouses “are not sufficient to meet the requirements of
advanced analytics.”43
While all Canadian analysts stress how they have access to a large history of
data within their own service’s records management system, they also identify
how fragmented databases and technological platforms are a significant impedi-
ment to conducting crime and intelligence analysis:

Back in 2007, the Solicitor General Office in [a Canadian province] went and
talked to all of the police agencies and said we would like to get you on the same
software with the same tools ... They selected [a private IT company] to deliver
the computer aided dispatch and records management systems, business intel-
ligence and [intelligence analysis] for doing analytics. We started working on
that project with that vendor in 2010 – by 2012 it lost momentum. Policing is a
world where people like their privacy and they like to do their own thing – like
even within our own walls we have silos – so trying to get twelve agencies to
work together and agree on something is impossible. (I27)44

Fragmented databases have left police services having to create “Band-Aid

solutions” (I22) by purchasing off-the-shelf data management systems to facili-
tate interoperability – systems that often have proprietary regulations attached
that create additional challenges for technological integration.45 Without an
integrated system, officers have to input the same data across different systems.
Entering information in duplicate can lead to data entry errors – such as the
misspelling of a person’s name or an incorrect home address. When a name is
186 Carrie B. Sanders and Janet Chan

spelt incorrectly or a person is linked to multiple addresses, separate entities

are created in the data system. Further complicating the technological terrain
is the lack of a standardized lexicon and system for querying databases. There
are no strict guidelines or rules that officers must follow when writing reports,
which means that officers can use different words to describe the same incidents.
The lack of technological integration and standardization creates challenges for
computerized searching and querying of databases, thereby requiring analysts
to manually search, collate, and clean data from different systems.
Finally, the lack of a standardized process for gathering and analyzing intel-
ligence has also led services to develop and implement different intelligence
organizational structures. Of importance here is the way services have adopted
crime and intelligence analysis. Some services have created clear and often
hierarchical divisions between these roles and their responsibilities. For example,
crime analysts are frequently employed (1) by administration and used largely
for comp statistics (comp stat) policing activities (such as developing monthly
crime reports and statistics), or (2) by front-line officers where they spend a
great deal of their time “mapping crime,” “making pretty pictures,” and doing
“workups” on people of interest.46 On the other hand, intelligence analysts are
assigned to specialized units (such as Gang Crime, Homicide, Sex Crime Unit,
etc.) and tasked with doing tactical and operational intelligence analysis (such
as “linkage charts,” cellphone analysis). These organizational divisions have led
analysts to work in isolation from each other and created informal “information
silos” (fieldnotes).47

Resource and Practical Barriers

As described above, the multiplicity of informal information networks, data
systems, repositories, and technological platforms, coupled with data quality
issues, creates impediments for computational systems to effectively conduct
advanced predictive analytics. The use of manual computation and analysis did
not result solely from the fragmented technological platforms, although they
do significantly contribute to them, but also stemmed from a series of organiza-
tional factors, such as a lack of organizational policies and legal frameworks,
and user competencies. Many of the analysts interviewed did not have knowledge
of technological platforms available to them that would enable them to analyze
unstructured data, such as videos, images, and social media. Where analysts
did have access to social media analysis, they identified a number of challenges
to its integration. First, the costs associated with purchasing an open-source
analytic software licence – that wouldn’t leave a trace – were very high. Police
services could afford to purchase only a small number of licences, which in turn
left only a few computers in the service available for doing such work. This
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 187

means that analysts and police personnel have to request access to the social
media computer terminal – requiring them to physically move locations in the
service – or request the assistance of a different analyst. Neither of these was
perceived as an acceptable or ideal option.
Second, legislative barriers to accessing information and scraping data struc-
ture the types of open-source data and analytics that can be utilized for law
enforcement purposes. In services where open-source data were available, there
was often a lack of familiarity with, and awareness of, organizational policies
or legal frameworks for analyzing them. As one analyst explained:

We’re very behind in our policy ... It’s really due to a lack of understanding
from upper management. They are slow to understand that most of our crime
is moving online, or a lot of our evidence is moving to digital. And they don’t
understand the need for a policy ... We’re not CSIS and we don’t have those
resources, but I’m sure that within there, there is what law enforcement can
and cannot do (I6).

This analyst suggested that the lack of organizational policies concerning open-
source analytics are the result of a broader problem around organizational
knowledge and understanding of open-source data and the legalities of working
with such data for intelligence purposes.
Lack of organizational training on open-source analytics was also identified
as a challenge to the effective uptake of big data analytics. An analyst explained:

We had this program [redacted] that was social media related ... Unfortunately,
really, I didn’t have the training to be able to fully know what I was doing with
that program. So, as an example, we had a major incident that somebody said
“[CITY] Shopping Centre had a bomb threat.” There had been a number of
tweets, as an example, coming from this area. Can you pinpoint exactly where
these tweets are coming from? I didn’t have the training to do that – which really
to work that – you should really have it (I1).

This analyst explains how her service did provide her with a technological
platform that made possible open-source big data analytics, but the service did
not provide the training required to effectively integrate and use the software.
Our findings mirror those of Jyoti Belur and Shane Johnson, who identified
knowledge and process gaps that inhibit the integration of analysis. Specifically,
they found that “knowledge gaps existed not only on the part of analysts as
individuals and collectively, but also in terms of what policy-makers and senior
leaders knew about analysis and analytical capability of the organization.”48
188 Carrie B. Sanders and Janet Chan

Emphasis on Tactical and Operational Analysis over

Strategic Analysis
Research on police intelligence work identifies different types of intelligence,
such as tactical, strategic, and operational, which serve different purposes.
However, there appears to be a lack of consensus and agreement – among both
practitioners and academics – on how to classify intelligence and what the labels
mean.49 Marilyn Peterson defines tactical intelligence as information used to
target specific cases and people, strategic intelligence as information used
to identify broader crime trends and to recommend how resources should be
allocated, and operational intelligence as work with larger cases to target crime.50
When we spoke with police personnel – in both information technology depart-
ments and analytic roles – it quickly became apparent that procurement needs
were shaped by traditional police uses of information, specifically tactical and
operational needs, which left analysts having to work with “flat files” (I30) that
required them to “pull the unstructured data (intelligence narratives) out of
[the] records management system, compile it into a relational database, and
reorganize it” (I39).51 Police records management systems are structured around
the information requirements for inputting information about a single incident
or individual. Yet, as Ridgeway notes, police services “need to expand to data
designs with a Big Data strategy in mind, which means having the ability to
filter and link on data features.”52 Often, when describing their analytic process,
analysts described a very investigative and targeted approach. For example,
analysts explained how social media sites, such as Facebook, were frequently
used in a “targeted” way “to see if they’re [known suspects] posting photos with
other people – certain people” (I2). In this way, analysts who were on social
media were often using it to find evidence or information about people they
already suspected.
While traditional uses of police information have shaped technological design
and appropriation, we further argue that the tactical and operational focus of
front-line policing – on reaction and arrest – has led to an emphasis on, and value
for, tactical and operational intelligence over strategic intelligence. The following
excerpt from an interview with a strategic intelligence analyst illuminates the
service’s emphasis on targeted, investigative tactical and operational analysis:

I don’t feel that I have moved to the analysis of anything – because we start
out and say there are targets and all I do is run history, like I check all of the
databases to identify their address, associates, vehicles and so on and so forth,
but there are times that at least it is not required of me to go further and try to
see the bigger picture, of the group because of the mandate for my team ... we
don’t go up on wire, we do basic investigative files ... I think one area that we are
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 189

lacking is looking at the final aspects of it ... there is a bigger network and we
don’t seem to focus on that as much – we are more just reactive – and we do very
low level investigations. (I20)

Using analytics for “targeted” and “investigative purposes,” this analyst argued,
led to “low level” investigations that do not enable police services to identify
and interrupt larger crime trends.
When we inquired about the use of predictive analytics and broader strategic
intelligence analysis, all analysts noted the organizational emphasis on tactical
and operational intelligence over strategic intelligence:

Even [in] our bigger investigations there is less of an appetite [for strategic anal-
ysis] because of the resources that are involved. So, as an analyst, we could pool
all of that information and start seeing the connections and identifying what
works – pull financial[s] of these individuals and see where the money is com-
ing and going ... We don’t have the resources because we don’t have forensic
accountants that are capable. We are requesting regular members to look into
these documents ... so there is no appetite for that ... but they [police service]
don’t consider the impact that it has on the economy ... you can show, go to the
media and say “look at all these drugs and weapons that we have seized and all
this cash”[;] these are tangible things that you can show, but bank documents
and wire transfers, no one cares for the paper. (I20)

This analyst contended that the organizational emphasis on tactical intelligence

over strategic intelligence was shaped by the service’s desire to demonstrate its
ability to take guns and drugs off the street and make arrests in order to address
visible street-level crime. The emphasis on tactical and operational intelligence,
we argue, is also shaped by the fact that many analysts are tasked by uniform
officers who do not possess analytic training and “are reactive and short-term
goal driven.”53 Yet, for police services to effectively use big data technologies will
require an organizational shift away from targeted investigative data gathering
towards bulk data collection and analysis.

Uneven Professionalization of Analysts and User

Capacity and Capability
A final challenge facing the integration of big data technologies in Canadian
policing is connected to the uneven professionalization of crime and intelligence
analysts. While the field has changed significantly over the past ten years, there
are still very few post-secondary programs and recognized accreditation pro-
grams available on criminal intelligence analysis in Canada. Further, there are
190 Carrie B. Sanders and Janet Chan

no recognized core competencies associated with this position.54 The lack of

formalized education and recognized core competencies has created challenges
for the professionalization of the field, which has led to great variation in the
educational background and disciplinary training of analysts.
The costs associated with analytic training means that many services are unable
to send their analysts to advanced training courses. For many analysts, “experience
and expertise were developed” (I06) through on-the-ground peer learning. The
lack of high-quality training provided to analysts raises serious concerns about user
capacity and capability. For example, a lead intelligence analyst from a large police
service noted that they had hired an internationally recognized intelligence analyst
to come and train their service on doing cellphone analysis, but the trainer was
unable to run the training because the analysts did not have enough foundational
training in Microsoft Excel to do advanced analytics (fieldnotes).
Training for, and knowledge of, big data analytics was also limited among the
analysts and police personnel interviewed. When we asked analysts whether
they use big data or conduct big data analytics, we received mixed responses:

I don’t know. I guess – because I don’t do a whole lot of – well at least to my

knowledge of what big data refers to – because I do more of the intelligence stuff,
so more research on individuals, as opposed to any large data – like, in terms of
going over 10 years of information on volumes of shootings, stuff like that ... I
don’t personally use big data. (I1)

I haven’t had to do anything that’s been extensive, yet. But, in the sense of [inter-
nal records management system] being a large database, you have a lot of infor-
mation in there. So, usually I am kind of going through that, yeah. (I3)

We are tactical analysts, and they are strategic analysts, and of course we feed off
of each other ... But as for doing all the research, anything with big data, is done
by [strategic analysts]. We don’t have anything to do with it. (I12)

Thus, there are mixed understandings and definitions of what big data is. Over
half of the analysts interviewed noted that their internal police records systems
constituted big data, while the other half did not believe they worked with big
data. Also, of interest in the quotes above, is the delineation between the value
of big data for strategic versus tactical intelligence analysis. Both participant I1
and I12 believed that they did not access or conduct analysis with big data
because they were focused on tactical intelligence. The mixed definitions of
what big data is combined with mixed understandings of using it speaks to
broader user capacity and capability issues. We believe the current situation is
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 191

that a large number of analysts do not have the skills to match the functionality
of the principal software packages supported by, or available to, police services.

Conclusion
The growing complexity of the post-9/11 security environment has blurred the
boundaries of policing and national security intelligence. The changing security
environment, combined with the digital revolution of the 1990s and a permis-
sive legislative environment, has also shaped the ways in which police collect,
store, analyze, and interpret intelligence for the purpose of detecting and dis-
rupting crime.55 There are growing claims in the United States and the United
Kingdom, and more recently in Canada, about a movement towards predictive
policing practices through the integration of big data technologies. The findings
presented in this chapter problematize many of the claims surrounding big data
and predictive policing by identifying how big data technologies are socially
shaped by technological, organizational, and cultural contexts, which impede
their effective integration and uptake by Canadian law enforcement.
The findings raise concerns about police services’ knowledge and capacity to
fully understand and utilize data analytics and big data technologies in police
intelligence work. In fact, the empirical data informing this chapter demonstrate
that few analysts have the technical skills and data literacy to use the technical
tools to their full potential.56 Yet, as Weston, Bennett Moses, and Sanders argue,
“the increasing complexity of data science methods associated with pattern-
recognition and prediction, particularly in the context of large volumes of
diverse, evolving data, requires a relatively high level of technical sophistica-
tion.”57 Andrew Ferguson contends that while big data technology provides
innovative potential for police services, choosing a big data system is a political
decision rather than a policing decision. He argues that police administrators
must be able to (1) identify the risks they are trying to address; (2) defend the
inputs into the system (i.e., data accuracy, methodological soundness); (3) defend
the outputs of the system (i.e., how they will impact policing practice and com-
munity relationships); (4) test the technology in order to offer accountability
and some measure of transparency; and (5) answer whether their use of the
technology is respectful of the autonomy of the people it will impact.58 Our
findings raise serious concerns about the risks associated with low-level data
literacy skills and a police service’s understanding of the capabilities and limita-
tions of big data and predictive policing.
Lastly, throughout this chapter, we have demonstrated different ways in which
police adoption and use of big data and data analytics can be plagued with
“black data” – which results from “data bias, data error, and the incompleteness
of data systems.”59 Further complicating the issues is that algorithmic predictive
 192 Carrie B. Sanders and Janet Chan

technologies can also have technical and subjective bias built in that can, if their
users are not cautious, lead to discriminatory practices.60 While the outputs of
predictive technologies can be easily attained, due to proprietary algorithms
the outputs lack transparency and accountability. Thus, while big data technolo-
gies hold great promise for police services to become more efficient, effective,
and accountable, without strong data literacy and a sophisticated understanding
of the political, effective implementation their success is questionable, and more
importantly, the socio-political implications of their use are difficult to
determine.

Notes
1 Richard Ericson and Kevin Haggerty, Policing the Risk Society (Toronto: University of
Toronto Press, 1997).
2 Lucia Zedner, “Pre-Crime and Post Criminology?” Theoretical Criminology 11 (2007):
264.
3 Patrick F. Walsh, Intelligence and Intelligence Analysis (Oxford: Routledge, 2011).
4 Keeley Townsend, John Sullivan, Thomas Monahan, and John Donnelly, “Intelligence-
Led Mitigation” Journal of Homeland Security and Emergency Management 7 (2010):
1–17.
5 Walsh, Intelligence and Intelligence Analysis, 130.
6 Jerry Ratcliffe, Intelligence Led Policing (Cullompton: Willan Publishing, 2008), 81.
7 Greg Ridgeway, “Policing in the Era of Big Data,” Annual Review of Criminology 1 (2017):
409.
8 Charlie Beck and Colleen McCue, “Predictive Policing: What Can We Learn from Wal-
Mart and Amazon about Fighting Crime in a Recession?” Police Chief 76 (2009): 19.
9 Matt Meuse, “Vancouver Police Now Using Machine Learning to Prevent Property
Crime: ‘Predictive Policing’ Technology Uses Past Trends to Determine Where Break-
ins Are Likely to Occur,” CBC News, 22 July 2017.
10 Ibid.
11 Ibid.
12 Meghan Grant, “Calgary Police Cellphone Surveillance Device Must Remain Top Secret,
Judge Rules: Alleged Gangsters Barakat Amer and Tarek El-Rafie Were Targets of the
Cellphone Interception Tool,” CBC News, 30 October 2017.
13 Alexander Babuta, “An Assessment of Law Enforcement Requirements, Expectations
and Priorities” (RUSI Occasional Paper, ISSN 2397-0286, 2017).
14 Sarah Brayne, “Big Data Surveillance: The Case of Policing,” American Sociological Asso-
ciation 82, 3 (2017), doi: 10.1177/0003122417725865 i.org/10.1177/0003122417725865.
15 Ridgeway, “Policing in the Era of Big Data,” 408.
16 Babuta, “An Assessment of Law Enforcement”; Ridgeway, “Policing in the Era of Big
Data”; Walt Perry, Brian McInnis, Carter Price, Susan Smith, and John Hollywood, Pre-
dictive Policing: The Role of Crime Forecasting in Law Enforcement Operations (Santa
Monica, CA: RAND Corporation, 2013).
17 Jyoti Belur and Shane Johnson, “Is Crime Analysis at the Heart of Policing Practice? A
Case Study,” Policing and Society (2016): 2, doi: 10.1080/10439463.2016.1262364.
18 Brayne, “Big Data Surveillance.”
19 Janet Chan and Lyria Bennett Moses, “Making Sense of Big Data for Security,” British
Journal of Criminology 57 (2017): 299–319; Adam Crawford, “Networked Governance
 The Challenges Facing Canadian Police in Making Use of Big Data Analytics 193

and the Post-Regulatory State? Steering, Rowing and Anchoring the Provision of Polic-
ing and Security,” Theoretical Criminology 10, 6 (2006): 449–79.
20 Ridgeway, “Policing in the Era of Big Data,” 408.
21 Lisa-Jo Van den Scott, Carrie Sanders, and Andrew Puddephatt, “Reconceptualizing
Users through Rich Ethnographic Accounts,” in Handbook of Science and Technology
Studies, 4th ed., edited by Clark Miller, Urike Felt, Laurel Smith-Doerr, and Rayvon
Fouche (Cambridge, MA: MIT Press, 2017).
22 Brayne, “Big Data Surveillance,” 6.
23 Wiebe E. Bijker, “How Is Technology Made? – That Is the Question!” Cambridge Journal
of Economics 34 (2010): 63–76.
24 danah boyd and Kate Crawford, “Critical Questions for Big Data: Provocations for a
Cultural, Technological, and Scholarly Phenomenon,” Information, Communication and
Society 15 (2012): 663.
25 Lyria Bennett Moses and Janet Chan, “Using Big Data for Legal and Law Enforcement
Decisions: Testing the New Tools,” University of New South Wales Law Journal 37 (2014):
652.
26 Chan and Bennett Moses, “Making Sense of Big Data.”
27 Ibid.
28 Brayne, “Big Data Surveillance.”
29 Ibid.
30 Ibid.
31 Babuta, “An Assessment of Law Enforcement.”
32 Ibid.
33 Ibid; Janet Chan and Lyria Bennett Moses, “Can ‘Big Data’ Analytics Predict Policing
Practice?” in Security and Risk Technologies in Criminal Justice, edited by S. Hannem
et al. (Toronto: Canadian Scholars’ Press, 2019); Ridgeway, “Policing in the Era of Big
Data.”
34 Babuta, “An Assessment of Law Enforcement.”
35 Chan and Bennett Moses, “Can ‘Big Data’ Analytics Predict Policing Practice?”
36 Carrie B. Sanders, Crystal Weston, and Nicole Schott, “Police Innovations, ‘Secret Squir-
rels’ and Accountability: Empirically Studying Intelligence-Led Policing in Canada,”
British Journal of Criminology 55 (2015): 718.
37 Colin Atkinson, “Patriarchy, Gender, Infantilisation: A Cultural Account of Police Intel-
ligence Work in Scotland,” Australian and New Zealand Journal of Criminology 50, 2
(2017): 234–51; Nina Cope, “Intelligence Led Policing or Policing Led Intelligence? Inte-
grating Volume Crime Analysis into Policing,” British Journal of Criminology 44, 2 (2004):
188–203; Belur and Johnson, “Is Crime Analysis at the Heart of Policing Practice?”
38 Walsh, Intelligence and Intelligence Analysis; Patrick F. Walsh, “Building Better Intelligence
Frameworks through Effective Governance,” International Journal of Intelligence and
Counter Intelligence 28, 1 (2015): 123–42; Anthony A. Braga and David L. Weisburd, Police
Innovation and Crime Prevention: Lessons Learned from Police Research over the Past 20
Years (Washington, DC: National Institute of Justice, 2006); Christopher S. Koper, Cynthia
Lum, and James Willis, “Optimizing the Use of Technology in Policing: Results and Impli-
cations from a Multi-Site Study of the Social, Organizational, and Behavioural Aspects of
Implementing Police Technologies,” Policing 8, 2 (2014): 212–21; Carrie B. Sanders and
Camie Condon, “Crime Analysis and Cognitive Effects: The Practice of Policing through
Flows of Data,” Global Crime 18, 3 (2017): 237–55.
39 Kathy Charmaz, Constructing Grounded Theory: A Practical Guide through Qualitative
Analysis (London: Sage Publications, 2006).
40 Ibid.
 194 Carrie B. Sanders and Janet Chan

41 Sanders, Weston, and Schott, “Police Innovations.”

42 Babuta, “An Assessment of Law Enforcement.”
43 Ibid., 5.
44 “I” denotes interviewee number.
45 James Sheptycki, “Organizational Pathologies in Police Intelligence Systems: Some Con-
tributions to the Lexicon of Intelligence-Led Policing,” European Journal of Criminology
1, 3 (2004): 307–32.
46 Sanders and Condon, “Crime Analysis and Cognitive Effects.”
47 “Fieldnotes” refers to data acquired through direct field observations.
48 Belur and Johnson, “Is Crime Analysis at the Heart of Policing Practice?”
49 Walsh, Intelligence and Intelligence Analysis.
50 Marilyn Peterson, Intelligence-Led Policing: The New Intelligence Architecture (Washing-
ton, DC: US Department of Justice, 2005).
51 Sanders and Condon, “Crime Analysis and Cognitive Effects.”
52 Ridgeway, “Policing in the Era of Big Data,” 405.
53 Belur and Johnson, “Is Crime Analysis at the Heart of Policing Practice?” 8.
54 Crystal Weston, Lyria Bennett Moses, and Carrie Sanders, “The Changing Role of the
Law Enforcement Analyst: Clarifying Core Competencies for Analysts and Supervisors
through Empirical Research,” Policing and Society: An International Journal, https://doi.
org/10.1080/10439463.2018.1564751.
55 Walsh, Intelligence and Intelligence Analysis.
56 M. Burcher and Chad Whelan, “Intelligence-Led Policing in Practice: Reflections
from Intelligence Analysts,” Police Quarterly 22, 2 (2018): 139–60, doi: 10.1177/
1098611118796890; Weston, Bennett Moses, and Sanders, “The Changing Role of the
Law Enforcement Analyst.”
57 Weston, Bennett Moses, and Sanders, “The Changing Role of the Law Enforcement Ana-
lyst,” 8.
58 Andrew Ferguson, The Rise of Big Data Policing: Surveillance, Race and the Future of Law
Enforcement (New York: New York University Press, 2017), 439.
59 Ibid., 446.
60 Janet Chan and Lyria Bennett Moses, “Is Big Data Challenging Criminology?” Theoreti-
cal Criminology 20, 1 (2016): 21–39.
Part 4
Resistance to Big Data Surveillance
This page intentionally left blank
11
Confronting Big Data
Popular Resistance to Government Surveillance in Canada
since 2001
Tim McSorley and Anne Dagenais Guertin

State surveillance has been a hallmark of government activities since before

the official founding of Canada. As many have noted, the project of pacification –
a more palatable term for colonization – of Indigenous populations was under-
taken with thorough, widespread surveillance, ranging from tribal rolls to church
records.1
Observers, notably Mark Neocleous, have pointed out that the first use of the
word “pacification” traces back to the sixteenth century, and the concept was
quickly incorporated into the approach of colonial countries in their quest for
greater territory and resources.2 King Philip II of Spain stated explicitly that in
order to better achieve its colonial mission, his empire should adopt a policy of
pacification, including to “gather information about the various tribes, languages
and divisions of the Indians in the province.”3 This, of course, did not mean
understanding the local populations in order to enter into a reciprocal relation-
ship but rather to better quell unrest and integrate them into the colonial project.
As Neocleous notes, “it implies the construction of a new social order as well
as the crushing of opposition to that construction.”4
As the colonial project has continued over centuries, with Indigenous com-
munities continuing to face high levels of state surveillance, similar tools and
tactics have been applied to other segments of the Canadian population. The
greatest attention of the state surveillance apparatus continues to be directed
towards communities who present – or are viewed as presenting – the greatest
challenge to the status quo. This has alternately included trade unionists, com-
munists, Quebec independence activists, LGBTQI+ people, anti-globalization
activists, environmental activists, and Muslims and Arabs.5
Surveillance also goes beyond the targeting of identified populations, to the
monitoring of society as a whole. Numerous scholars have detailed how over-
arching methods of surveillance, particularly through law enforcement and
state bureaucracy, serve to keep a watchful – and acknowledged – eye on the
population at large, serving to (once again) pacify while at the same time caus-
ing a chilling effect.6 Canadians are sharing greater and greater amounts of their
information with both private companies and the government. At the same
time, new forms of surveillance technology have allowed government and private
interests to collect information about people with or without their consent.7
198 Tim McSorley and Anne Dagenais Guertin

This increase in surveillance has been motivated by several objectives, not

the least of which has been “national security.”8 Indeed, the idea of national
security – the protection of the Canadian state from external and internal threats –
has been a prime motivator for surveillance since the earliest days of colonialism,
through both domestic and international upheavals, until today.9
The idea of what constitutes a national security threat has changed greatly
over the decades, and has been a topic of research in itself. This chapter will
examine examples of government surveillance of Canadians but not the alleged
threats that motivate them. More specifically, we will examine recent popular/
grassroots responses to government surveillance, analyzing their impact and
the lessons they may hold for the future. To do so, we have selected three cases
where we have interviewed a participant, carried out research on the campaign,
and examined the results. The three cases are:
1 Stop Illegal Spying – The British Columbia Civil Liberties Association (BCCLA)
launched an ongoing lawsuit against the Communications Security Establish-
ment (CSE) in 2013, challenging CSE’s work as infringing on Canadians’ rights.
Along with the court case, the BCCLA has engaged in a popular education cam-
paign to raise awareness of problems with the CSE’s work.
2 Stop Online Spying – A 2011 campaign initiated by OpenMedia against lawful
access laws. At the time, the federal government framed the need for lawful ac-
cess laws as necessary for the enforcement of criminal law.
3 International Campaign Against Mass Surveillance (ICAMS) – Begun in 2004,
ICAMS was an international movement, modelled after the International
Campaign to Ban Landmines, with the goal of bringing about internationally
agreed-to limits on mass surveillance.

These three examples have been chosen because they address three different
issues pertaining to mass surveillance: the actions of a specific surveillance
agency, laws granting governments easier access to personal information, and
global systems of mass surveillance. They were also chosen for their differing
approaches: use of the legal arena, mass online mobilization, and an international
coalition of civil society around a text-based campaign.
As we interviewed only one person per campaign, and as impact is sometimes
difficult to determine and attribute accurately, our analysis has clear limitations.
Thus, this chapter is not meant to be read as an authoritative guide to the most
and least effective campaigning methods. As activists, we often do not have the
time to reflect on past experiences, especially those of others. Thus, this chapter
aims to present information about past campaigns that will hopefully be useful
to today’s anti-surveillance campaigners in determining their preferred approach
and actions.
 Confronting Big Data 199

It has been nineteen years since Canada’s first Anti-terrorism Act (Bill C-36),10
and national security and surveillance legislation has only continued to multiply:
by our count, no fewer than twelve bills have been passed to bring in new
national security and surveillance powers over that time, culminating with Bill
C-51, Canada’s second Anti-terrorism Act, in 2015.11 In June 2019, Parliament
passed a new piece of national security legislation, Bill C-59 (the National Secur-
ity Act, 2017), which has introduced sweeping changes, including to the powers
and regulations surrounding mass surveillance in Canada.12 Along with what
we have learned through the revelations made by Edward Snowden and other
whistle-blowers and journalists, it is an opportune time to look at popular
responses to surveillance and their impact.

Case 1: Stop Illegal Spying

In 2013, the BC Civil Liberties Association filed suit in Federal Court against
the Communications Security Establishment, Canada’s cryptologic and signals
intelligence agency. The suit alleges that CSE violates Canadians’ Charter rights
through its mass surveillance practices.13 It followed long-standing and multiple
concerns expressed by civil liberties organizations across the country. This case
is based on an interview with Caily DiPuma, who at the time was the acting
litigation director at the BCCLA and a member of the legal team for the associa-
tion’s lawsuit against CSE.
At the time, CSE’s mandate was defined in a brief section of the National
Defence Act. It allowed for the minister of defence to grant broad authorizations
for, among other things, the interception of the private conversations of Can-
adians without any judicial oversight.14 The adoption of the National Security
Act, 2017 has since created a new CSE Act, which establishes more explicit rules,
as well as oversight, for the agency. Despite this, concerns remain about the
scope of CSE surveillance, and an examination of the BCCLA’s suit still holds
lessons for today. The change in the law also does not eliminate the need for
accountability for possible wrongdoing under the previous legal regime.

The Court Case

The case in Federal Court alleges that two aspects of CSE’s activities violate the
Canadian Charter of Rights and Freedoms’ protections against unreasonable
search and seizure and infringe on free expression: (1) the interception of the
private communications of Canadians, and (2) the sweeping collection of meta-
data information produced by Canadians in their everyday activities online and
through phone conversations.15
In 2014, the BCCLA followed up with a companion class action suit.16 The
goal of the class action is to provide a way for Canadians who have been impacted
200 Tim McSorley and Anne Dagenais Guertin

by CSE surveillance to seek remedy if the initial suit is successful in striking

down the laws that allow CSE to collect Canadians’ private communications
and metadata. The action is being brought on behalf of all people who have
used a wireless device – laptops, cellphones, etc. – in Canada since 2001.
The federal government and CSE have responded that CSE’s mandate prohibits
it from targeting Canadians’ communications and that it would be unlawful
for it to do so. They also argue that due to the nature of CSE’s work, it is impos-
sible for the agency to know whether or not Canadians’ communications are
being swept up in its surveillance activities, so “there may be circumstances in
which incidental interception of private communications or information about
Canadians will occur.”17

Litigating against Mass Surveillance

Throughout its history, the BCCLA has used numerous tactics in its defence of
civil liberties from the impact of national security and anti-terrorism laws,
including lobbying, petitions, and public education.
In this instance, according to DiPuma, the BCCLA decided upon litigation
because the CSE provisions in the National Defence Act were particularly weak
and amenable to challenge in court. This is because the act did not establish any
real statutory authority beyond the ministerial directives that delineated what
CSE could do with Canadians’ data.
The lack of details contained in the National Defence Act regarding what CSE
could and could not do meant that, among other things, the authorizations
granted to CSE for its surveillance activities were issued (1) without restrictions
on the breadth of communications to be collected, (2) without a reasonable
time limit, and (3) without adequate safeguards to ensure the accountability of
CSE in the course of its collection of private communications.18 The suit alleges
that these issues violate section 8 of the Charter (freedom from unreason-
able search and seizure). The BCCLA’s statement of claim also argues that CSE
violates section 2(b) of the Charter, which protects freedom of thought, belief,
opinion, and expression.19
While the goal of the suit is to bring substantial change to the rules governing
CSE’s activities, the BCCLA has coupled it with a public outreach campaign,
using infographics, social media sharing tools, backgrounders, and video.20 The
hope, said DiPuma, is that such court cases will also help bring more public
attention to government surveillance and CSE, and that this will put political
pressure on the government to act.
The timing of the case was also meant to capitalize on public awareness. The
suit was filed shortly after the publication of information and documents leaked
by former CIA contractor Edward Snowden, revealing global networks of mass
 Confronting Big Data 201

surveillance, particularly on behalf of the United States and the United Kingdom,
but also Canada.21 According to DiPuma, the BCCLA hoped to take advantage
of “an historical moment when people were paying attention to the issue of
mass surveillance.” Further revelations from the Snowden files would eventually
implicate CSE in spying on Canadians as well as engaging in global mass sur-
veillance operations.22
As the suit has yet to go to trial, there is no way to predict the outcome. The
BCCLA is hoping for a favourable judgment, but DiPuma says that it is open
to other outcomes too, including unilateral steps by the government to ensure
that CSE’s activities do not violate Canadians’ rights. “Litigation can add to
the overall policy discussion in a way that affects meaningful change in the
law,” says DiPuma, adding that if the government were to introduce laws that
appropriately changed CSE’s legal framework, the BCCLA would reconsider
its suit.

Impact
It is difficult to attest to the impact of a lawsuit that is still before the courts.
However, DiPuma points to what she sees as some impacts already:
• The suit has contributed to the public policy debate around CSE and govern-
ment surveillance.
• The case has brought public awareness to a secretive regime.
• The BCCLA has learned new details about the operations of CSE, but which
cannot be shared publicly yet.

These impacts are difficult to measure, but there are a few indications we can
look to.

Public Awareness
Through its outreach and public discussions, the BCCLA has found that people
were “shocked” and agreed that change was needed when the details of the case
and the association’s concerns with CSE’s surveillance practices were explained
to them.
In her experience, DiPuma said, there are mixed reactions among the public
to the question of whether or not we need to be concerned about protecting
privacy rights. In this case, however, the issues “resonated deeply.” The secrecy
around CSE makes it particularly important in this case to get people engaged,
and to understand it, DiPuma said.
At the time the lawsuit was filed, the case garnered considerable media cover-
age across the country, including articles in the Globe and Mail, CBC News, and
the Vancouver Sun.23 Each time the case has come up in court, including the
202 Tim McSorley and Anne Dagenais Guertin

filing of the class action suit and the arguments over disclosure in the summer
of 2016, there was a resurgence in media coverage of CSE and its surveillance
practices.24
Although it is impossible to measure the exact impact of this lawsuit on
public awareness, it is safe to say that such mainstream coverage at each stage
of the lawsuit would have educated more than a few people. However, as CSE
remains little known or understood by the Canadian public, it is also safe to
say that much more public education and media coverage is necessary.

Disclosure
Another impact of such a lawsuit is disclosure of information. Information that
to date has remained secret, unpublished, or inaccessible through other means
may come to light through the disclosure process, once it is entered into evidence
and become public. The disclosure of these documents helps to inform advocates
and the public at large of new issues and concerns, and can lead to further
action.
The BCCLA’s suit has gone through the disclosure stage, including argu-
ments over whether certain documents deemed sensitive to national security
should be disclosed to the BCCLA and made available to be entered into
evidence. While the bulk of disclosure is still being considered and will be
made public only if and when it is entered into evidence, we have already been
given a glimpse of the kind of information such cases can provide. In June
2016, the Globe and Mail published a comparison of two versions of a docu-
ment detailing metadata collection by CSE.25 The first version, received by the
Globe through access to information requests, is heavily redacted and of
modest value. The second version, received by the BCCLA through disclosure,
is significantly less redacted and contains information on how metadata is
defined and what analysts are able to do with the information. As the Globe
and Mail notes:

The BCCLA version of the document shows how CSE is under orders to indis-
criminately collect and share metadata with allied agencies. But also revealed is
the order to scrub out any “Canadian identifying information” after the fact.26

It is also important to note that it is difficult to challenge the operations of

national security agencies when their regulations are kept secret. By simply
obtaining more public disclosure of the rules that these agencies play by, civil
liberties advocates are better able to identify potential flaws and weaknesses.
It is to be hoped that as the case heads to trial, more information that shines
a light on CSE’s mass surveillance activities will become public.
 Confronting Big Data 203

Policy
As with other impacts, it is difficult to show a direct relation between the pres-
sure that a lawsuit puts on policy-makers and their decisions. This is especially
true when it comes to national security, since so much of what drives govern-
ment decision making is kept confidential. At the same time, one of the BCCLA’s
stated public goals with this suit is to change the practice and law around
government surveillance. The association’s suit also comes at a time of pressure
from other organizations and sectors to reform CSE’s activities to ensure that
they do not violate Canadians’ Charter rights (or engage in mass surveillance
internationally, for that matter).
In June 2017, four years after the BCCLA filed its initial lawsuit, the federal
government proposed a major revamp of CSE’s legislation. Bill C-59, the National
Security Act, 2017, includes the newly titled CSE Act. The act lays out in detail
the policies and procedures that would govern the work of CSE, including
parameters for surveillance and retention of data.27 The bill was granted royal
assent in June 2019.
The creation in the National Security Act, 2017 of a new Intelligence Com-
missioner (IC), with quasi-judicial status (the IC must be a retired Federal
Court judge), to approve surveillance authorizations before they are carried
out may also be seen as a response to some of the issues addressed in the
lawsuit (although without more information it is impossible to point to
causality, and further research would be necessary to reach any conclusion).
Such an approval process could, in theory, provide more certainty that
authorizations are issued on reasonable and probable grounds, or at least
under a clear and articulable legal standard (which, the lawsuit argues, does
not currently occur).
While the BCCLA published a brief on the provisions of Bill C-59, including
those related to CSE, it did not express an opinion on whether these changes
address the concerns brought up in its court challenge.28

Analysis
Litigation can be an effective tactic to protect Charter rights: rather than rely
solely on public pressure and advocacy, Charter challenges can result in concrete
changes in law and put pressure on the government to act before the courts
issue their decision. They can also draw media attention to an issue and provide
an opportunity for public awareness campaigns. At the same time, they can be
resource-intensive and drawn out, and there is little guarantee that a court rul-
ing will support, in full or in part, the goal a campaign hopes to achieve. The
decision then centres on whether to continue to pursue it through the courts
or attempt to bring about change through other means.
204 Tim McSorley and Anne Dagenais Guertin

The advantage of litigation is that a government with a parliamentary major-

ity, as the Liberals are at the time of writing and the Conservatives were before
them, is able to withstand public pressure to amend legislation unless it threatens
its electability at the next polls. When dealing with surveillance and national
security, which is often little known to the public and rarely determines how
people vote, relying on pressure at the polls is difficult. The impact of a court
ruling could force the government’s hand to take action that other public advo-
cacy campaigns could not achieve. Furthermore, court rulings, even if not
completely favourable, can lead to momentum and openings for advocacy
campaigns to push for important changes to the law, with the weight of a court
decision behind them.
The possibility of a negative ruling can also push a government to act before
it is issued, in the hope that the suit is dropped; even the possibility of dis-
closures during trial that could further damage the government’s reputation
or raise other legal concerns could motivate it to take action and reform the
law.
Disclosure and access to information is another positive, if secondary, result
of this kind of legal action. Often the most difficult part of understanding – let
alone challenging – national security activities is accessing clear, comprehensive
information about those activities, and the authorizations and analyses that
underlie them. Disclosures in court can provide much more pertinent and
complete information than the current access to information laws provide.
While such disclosure would never be the primary reason for this kind of suit,
it nonetheless helps complete the picture and lays the foundation for future
work on these issues.
Finally, the fact that the government saw a need to finally detail the rules
governing CSE’s work in Bill C-59 could be seen as motivated by the BCCLA’s
suit – or at least a recognition that the lack of clear policy leaves the agency
open to such challenge. It is of course impossible to attribute any kind of caus-
ality without further research. We would suggest that further investigation be
carried out, including access to information requests and interviews with
policy-makers, to ascertain what role this suit has played. The other question,
of course, is whether the reforms contained in C-59 actually address the con-
cerns raised in the lawsuit. This remains a matter of debate, but overall, civil
society groups have judged the new CSE Act to be insufficient in addressing
current concerns as well as granting troubling new powers to the agency.29 The
CSE Act in fact enshrines some of CSE’s more controversial powers, including
those related to the retention of Canadians’ private information, in law. Instead
of limiting CSE’s activities, the adopted act may simply result in legitimizing
them. In that case, continuation of the lawsuit with the possibility of a verdict
 Confronting Big Data 205

favourable to the BCCLA’s arguments could provide additional pressure to

improve a problematic law.

Case 2: Stop Online Spying

In February 2012, the Conservative government introduced Bill C-30, the
Protecting Children from Internet Predators Act.30 Originally titled the Lawful
Access Act, the bill would have given law enforcement authorities and national
security agencies sweeping new powers to access private information without
a warrant.31 OpenMedia was central to organizing a coalition and online cam-
paign, including a petition, that led to withdrawal of the bill.32 In 2014, the
government enacted a similar but significantly weakened bill, which, however,
still expanded lawful access powers. This case is based on an interview with
David Christopher, OpenMedia’s interim communications and campaigns
director.

The Campaign
Many will remember the fight against lawful access in Canada for Public Safety
Minister Vic Toews’s infamous words just before Bill C-30 was introduced – that
Canadians “can either stand with us or with the child pornographers.”33 Within
a year, the bill was dead.
The battle against lawful access had been ongoing for several years already,
dating back to the 2005 Liberal government. Despite many attempts, no lawful
access legislation was successfully adopted.34 The Conservative party made it a
key plank of its proposed omnibus crime bill, which it promised to enact within
the first 100 days following its election with a parliamentary majority on 2 May
2011.35 That June, however, a coalition of thirty public interest organizations,
businesses, and concerned academics assembled by OpenMedia launched the
Stop Online Spying campaign (www.stopspying.ca), calling for an end to lawful
access legislation and for the Conservatives to exclude it from the proposed
omnibus crime bill.
At the time, OpenMedia was known especially for its Stop the Meter campaign,
which by mid-2011 had racked up more than 500,000 signatures on an e-petition
calling on the government to put an end to usage-based billing (or metered
Internet access). Such a large number of online petition signatures was unheard
of in Canada at the time, and the campaign was a factor in having the govern-
ment make the CRTC rule against the practice.36 The campaign’s success played
a role in OpenMedia’s taking on of lawful access issues.
OpenMedia believed that fighting lawful access in the same way could suc-
ceed. According to Christopher, the organization set out to bring together a
wide coalition that bridged the political spectrum in order to put pressure on
206 Tim McSorley and Anne Dagenais Guertin

all political parties. Why would OpenMedia weigh in on online surveillance

when, to that point, it was known for its focus on access to the Internet? As it
stated in a press release:

“Every single provincial privacy commissioner has spoken against this bill,” says
OpenMedia.ca’s Executive Director, Steve Anderson. “Law-abiding Canadians
should be able to use the Internet and mobile devices without Big Telecom and
government looking over their shoulders. These invasive surveillance bills will
transform the Internet into a closed, rigid, and paranoid space.”37

On 9 August 2011, coalition members followed up with an open letter to Prime

Minister Stephen Harper outlining their concerns.38
The campaign appears to have met with some early success, with the Con-
servatives introducing their omnibus crime bill, C-10, without mention of lawful
access.39 Christopher attributes this outcome to cross-sector approach. The
government, however, promised to provide what it described as necessary tools
for law enforcement to fight child pornographers and organized crime. The
campaign kept up the pressure by producing public service announcements
and a short documentary explaining the dangers of lawful access legislation to
Canadians’ privacy; these garnered around 25,000 and 75,000 views, respectively.
The Conservatives persevered, unsurprisingly since, with a majority govern-
ment, they had just passed the omnibus crime bill that was also highly unpopular
with civil liberties advocates. On 14 February 2012, they introduced Bill C-30
to a flurry of criticism.40
That same day, Toews made his infamous statement. His speech in the House
of Commons led to the #TellVicEverything hashtag as Canadians bombarded
him on Twitter with the minutest details of their lives, trending in Canada and
worldwide on 16 February.41 As Steve Anderson wrote on the OpenMedia web-
site, OpenMedia and its coalition members had prepared for the tabling of the
bill, pushing the online petition launched with the start of the campaign.42
Besides the online petition and the August 2011 open letter, OpenMedia also
divulged documents demonstrating that police forces had difficulty finding
examples of cases that were hindered due to the lack of lawful access
legislation.43
The controversy caused by Toews’s statement helped propel the campaign
forward, but, according to Christopher, the coalition also benefited from hav-
ing prepared tools in advance and from the confidence gained from its success
in the Stop the Meter campaign. At the time, said Christopher, OpenMedia,
which now operates internationally, had a much smaller team with fewer re-
sources, and was attempting to punch above its weight. The online petition, key
 Confronting Big Data 207

to the Stop the Meter campaign and viewed as key to the new campaign, took
off after Toews’s statement and the momentum generated by the hashtag. By
April, the petition had more than 130,000 signatures.44
Christopher points out that it is important to see this number in context. At
the time, online petitions receiving mass support were something new, especially
in Canada. Thus, as the petition passed each milestone – say, 25,000 signatures –
the media reported on it, resulting in ever-growing momentum. On 30 April,
the petition hits its peak, with 135,000 signatures – fewer than for Stop the
Meter but still significant for a campaign on issues like online surveillance
and lawful access, topics that were not often seen as causing strong public
outcry.
Bill C-30 had passed only first reading and while opposition parties brought
it up in debate, the government never returned it to the House of Commons.
A year after its introduction, the government withdrew the bill. Justice Minister
Rob Nicholson stated:

We will not be proceeding with Bill C-30 and any attempts we will have to mod-
ernize the Criminal Code will not contain the measures in C-30 – including
the warrantless mandatory disclosure of basic subscriber information, or the
requirement for telecommunications service providers to build intercept capa-
bilities within their systems ... Any modernization of the Criminal Code ... will
not contain those.45

OpenMedia and other anti-surveillance activists greeted this as a major victory

against lawful access at the time, having forced a majority government to roll
back a key law – which rarely happens.

Analysis
While the Stop Online Spying campaign is remembered for the #TellVicEverything
hashtag and the online petition, it was a multi-faceted campaign that also
included coalition building, lobbying, and popular education through online
videos and community screenings. These other tactics arguably helped build
momentum towards the 135,000-signature petition.
Along with the multiple tactics, it is also important to acknowledge what
Christopher called the “moment in time.” This includes the resounding success
of the Stop the Meter campaign (the largest online petition campaign in Canada
to date), the novelty and newsworthiness of a viral online petition, and the
government’s miscalculations, particularly Vic Toews’s “with us or with the child
pornographers” proclamation.
208 Tim McSorley and Anne Dagenais Guertin

It would appear that OpenMedia and its partners were able to take advantage
of this moment to pressure the government and make it politically unpalatable
for the Conservatives to proceed with Bill C-30. The online petition with 135,000
signatures served as visual evidence of the widespread disapproval of the bill
that was necessary to cause the government to retreat.
Unfortunately, just as there had already been multiple attempts to bring in
lawful access, the Conservative government did not give up. Several months
later, Justice Minister Peter MacKay reintroduced lawful access legislation with
Bill C-13. This time, the government framed the issue as an anti-bullying law to
address cases like those of Amanda Todd and Rehtaeh Parsons – teenage girls
bullied online to the point of suicide. Many, including Amanda Todd’s mother,
criticized the bill for combining two unrelated issues and decried it as a political
manoeuvre.46 The new amendments were scaled back, and did not, to the word,
break Nicholson’s earlier promise: the new legislation did not allow for war-
rantless access or force Internet service providers (ISPs) to build in technology
that would allow back-door entrance for law enforcement. It did, however, lower
the threshold for approval of certain kinds of access warrants, thereby making
it easier than ever for law enforcement and national security agencies to access
online personal data.
So while the laws have been loosened and the debate continues, it is fair to
say that the Stop Online Spying campaign significantly slowed lawful access
rules in Canada. It is telling that even in recent debates on lawful access, there
continue to be references to the failure of Bill C-30 and the massive opposition
Canadians have shown to warrantless surveillance.

Case 3: International Campaign Against Mass Surveillance

This case is based on an interview with the former International Civil Liberties
Monitoring Group (ICLMG) national coordinator and initiator of ICAMS Roch
Tassé. At the time, in 2004, ICLMG was a coalition of about thirty Canadian
civil society groups. It now has forty-six member organizations and its mandate
is the protection and promotion of human rights and civil liberties in the context
of the so-called War on Terror.

What Is ICAMS?
In 2004, the ICLMG joined other well-known human rights groups from
around the world – the American Civil Liberties Union (United States),
Statewatch (United Kingdom), and Focus on the Global South (Philippines) –
in launching the International Campaign Against Mass Surveillance, which
calls on governments to end mass surveillance and global registration of entire
populations.
 Confronting Big Data 209

Patterned in some respects after the international campaign to ban land-

mines, ICAMS aims to build a movement of resistance to these measures
around the world by circulating a core document, a summary of the docu-
ment, and a declaration for endorsement.47 In order to create such an
international network, twenty ICLMG member organizations, Ben Hayes
from Statewatch, Jameel Jaffer from the ACLU, and Walden Bello from
Focus on the Global South organized a three-day conference in 2004 to
discuss the impacts of mass surveillance. The first day of the conference
was a private round-table meeting with about thirty guests. The second
day was a public panel attended by about 150 people where Tassé, Hayes,
Jaffer, and Bello made presentations and several guests spoke specifically
of the impacts on Muslim communities. The final day was also a closed-
door meeting to find a way to keep the momentum going, to find an action
that could be accomplished by this new international partnership. What
seemed the most pressing to the attendees was the consequences of national
security legislation and surveillance on people’s privacy. It was then decided
that a campaign on that issue should be created in the next year, and a small
coordination committee composed of the main organizers and Maureen
Webb was put together and tasked with writing the core document. The
document was eventually expanded into a book, Illusions of Security, writ-
ten by Maureen Webb.48 The conference and the campaign were carried
out on the organizers’ own time and dime.
After one year, the core document was ready. The campaign got its name,
a website was created, and the document was distributed internationally.
In just a few months, 300 individuals and organizations endorsed the
ICAMS Declaration. At the end of the campaign, the number had grown
significantly, although the exact number is unknown as the website is no
longer active.

International Conference of Data Protection and

Privacy Commissioners
In 2007, Montreal hosted the International Conference of Data Protection and
Privacy Commissioners, with guests from Canada, Europe, and Africa, but not
the United States.49 A civil society conference was organized in parallel and the
two sides met at the end of both conferences. The Office of the Privacy
Commissioner of Canada (OPC) awarded ICAMS a grant of $150,000 to organ-
ize the parallel conference. A staff person was hired for that purpose. After a
few modifications, the ICAMS core document became the report of the civil
society conference. The report was adopted as a resolution at the subsequent
Conference of Data Protection and Privacy Commissioners.50
210 Tim McSorley and Anne Dagenais Guertin

Impacts of ICAMS
The campaign had many positive results and impacts, including:
• the declaration signed by hundreds of organizations and individuals
• the book Illusions of Security
• global partnerships, the creation of a collaborative structure and culture, and
regional networks that have lasted to this day
• the emergence of many individuals and groups working on privacy
• getting privacy groups to start doing policy work
• getting organizations from all sectors to start working on privacy as well
• the creation of a relationship between civil society and the OPC
• the civil society parallel conference
• adoption of the core document as a resolution by the Conference of Data Pro-
tection and Privacy Commissioners
• development of international awareness of mass surveillance
• influence on university researchers and federal opposition parties – for exam-
ple, the Bloc Québecois and the New Democratic Party published a minority
report calling for the abrogation of the Anti-terrorism Act of 2001 because of
ICAMS’s work,51 and the ICLMG was frequently cited by them in the House of
Commons.

Reflections and Lessons

Although the campaign did important work, created important networks, and
had many positive impacts and outcomes, it has obviously not met its general
goal of eliminating mass surveillance or, at the very least, augmenting privacy
protections and reducing mass surveillance, in contrast to the success of the
campaign against landmines. If anything, privacy protections have deteriorated
and mass surveillance has crept into our lives and appears to be here to stay.
We are not suggesting that ICAMS was a failure or that it had the responsibility
to rid us of mass surveillance; that would be completely unfair. However, it is
important to examine the campaign in order to identify the lessons for future
courses of action against mass surveillance.
First, the size of the phenomenon makes it very difficult to tackle. Mass sur-
veillance is an omnipresent but often invisible issue that is highly intertwined
with technologies that have been integrated into our daily lives in numerous
helpful and positive ways. This makes it difficult for most people to fully grasp
the problem or to not feel powerless to protect themselves or others, and thus
to mobilize against it. Second, the political climate was a huge problem: the
Conservatives were in power at the time and they were highly unreceptive,
although the campaign document was sent to all of them and privacy
 Confronting Big Data 211

commissioners from around the world had adopted it. Thus, the campaign did
not lead to any new government policies. Third, no country adopted the docu-
ment or brought it before the United Nations, so no UN treaty was drafted, let
alone signed. Roch Tassé wonders whether they should have given themselves
more solid international structures or created a more formal international entity.
However, it is difficult to say whether that would have worked or been more
effective, he added.
Finally, the lack of time and resources had a big influence. Often, we feel that
it is imperative to act against a terrible affliction such as mass surveillance, but
resources and time limit what can be done. And although there were many
influential individuals and researchers involved, they could not carry the entire
burden. After a while, they moved on to other things or retired, with no one to
succeed them.

Discussion of the Three Cases

Surveillance, especially for national security purposes, is difficult to address as
so much is carried out in secret. In cases where the rules are explicitly stated,
such as lawful access, it is difficult to know exactly what data are being inter-
cepted, collected, and retained, as well as how they are being used. It is even
more difficult when it comes to surveillance conducted by organizations like
CSE, which has operated for over seventy years in relative obscurity and with
very few written rules.
Each of the campaigns above was faced with the dual task of challenging
unjust laws as well as uncovering and explaining to the public exactly what the
laws, regulations, and surveillance activities were and why they presented a
challenge to Canadians’ Charter rights. Although it seems safe to say that main-
stream media coverage of the BCCLA Charter challenge has educated some of
the public regarding CSE and mass surveillance, it is difficult to measure the
extent of its impact on policy changes.
It is clear, however, that Charter challenges in general can result in concrete
changes in law. When dealing with surveillance and national security, relying
on pressure from the public or at the polls can be difficult, and obtaining a
ruling that forces a government (especially a majority government) to make
legislative changes can therefore be a useful strategy. Furthermore, a lawsuit
can put pressure on the government to act before the courts issue their deci-
sion, to save face or to avoid disclosure or further legal concerns. At the same
time, lawsuits can be resource-intensive and drawn out, and there is little
guarantee that a court ruling will support, in full or in part, the goal a cam-
paign hopes to achieve. If the goal is to access and make public some infor-
mation never before disclosed, or to change the law in situations where public
212 Tim McSorley and Anne Dagenais Guertin

pressure seems improbable or insufficient, and if one has the resources and
the patience for such endeavours, then lawsuits and Charter challenges might
be the better approach.
It would seem that the Stop Online Spying campaign contributed to the
defeat of Bill C-30 thanks to not only its hashtag #TellVicEverything and its
online petition but also its multi-faceted approach, including coalition build-
ing, lobbying, and popular education through online videos and community
screenings. It would also appear that the campaign took advantage of a
moment in time that, unfortunately, cannot be planned and thus cannot be
replicated. As we have seen, no such moment enabled us to avoid the adop-
tion of Bill C-13, a subsequent piece of lawful access legislation. Being able
to recognize such a moment and being prepared to seize it are lessons we
can take from the Stop Online Spying campaign. Although multi-faceted
campaigns and viral online petitions also require resources, they are more
accessible than lawsuits for many individual campaigners and non-profit
organizations.
Mass surveillance being an international issue, an international campaign
might be what is needed. The International Campaign Against Mass Surveil-
lance offers a few lessons for such a project. One is that novel ways of framing
and publicizing the issue of mass surveillance as a serious problem that can be
solved appear to be essential in order to mobilize people. The political climate
also needs to be taken into account in order not to waste precious resources
and energy on governments that refuse to act on the issue. The creation of more
solid international structures or a more formal international entity is a potential
avenue to explore. Finally, a deeper collective reflection on the struggle against
mass surveillance, as well as securing more resources and ensuring the succes-
sion within the movement, appears to be necessary in order to sustain this type
of long-term campaign.
Unfortunately, it remains difficult to quantify the results or qualify any of
these campaigns as complete successes (although campaigns often achieve
partial or incremental reform). Or, perhaps more precisely, while each has
in large part fulfilled its immediate goals, the broader goal of reversing (and
eventually eliminating) intrusive, rights-undermining mass surveillance has
yet to be achieved. There is a strong argument to be made that civil society
campaigns have effectively slowed the growth of mass surveillance for national
security purposes, but if we were to measure whether mass surveillance is
more prevalent now than in 2001, it is clear that despite the revelations of
Edward Snowden and other whistle-blowers, as well as the work of progres-
sive legislators and of civil society groups, mass surveillance has continued
to grow.
 Confronting Big Data 213

The lessons that we can draw from these cases are reflective of campaign
strategies adopted in other sectors as well:
• Have clear, direct targets. Attacking the entire national security apparatus, while
necessary, appears to be aiming at too broad a target. This makes it difficult to
enunciate clear goals and engage the public in a clear campaign. Targeting of
CSE or of lawful access made those campaigns more straightforward, with clear
measures of success.
• Build coalitions across sectors. ICAMS went beyond civil society to build
bridges and allies in the government bureaucracy through privacy commission-
ers’ offices. OpenMedia reached out to conservative libertarians and business
organizations that shared similar concerns over privacy as well as the cost to
industry and users of forcing ISPs to integrate interfaces for government sur-
veillance technology into their own infrastructure.
• Use a diversity of tactics. Although the more facets a campaign has, the more
intense the use of resources (which are often limited for civil society organi-
zations), by multiplying the kinds of tactics and tools used – targeted to the
campaign’s specific goal – each of these campaigns was able to achieve greater
impact.

Another important aspect of these kinds of campaigns is how to address the

underlying motivation for government mass surveillance. As mentioned at the
outset of this chapter, government surveillance has often developed as a form
of control or pacification. Others have also argued that what we are witnessing
is a kind of change in equilibrium, towards a government that is wrapped in
privacy and a population expected to live in public, whereas the democratic
model is the opposite.
Surveillance, with the companion concept of national security, is also often
developed as a means to control minority and marginalized groups or any
community that poses a perceived threat to the status go. Integrating argu-
ments for equality and justice – including anti-racist, feminist, and anti-
colonial frameworks – into anti-surveillance campaigns may help stop specific
laws as well as undermine the rationale for such laws. It can also allow for
new and broader partnerships to form across organizations, including those
from directly impacted communities. The difficulty lies in how to communi-
cate to the public the current issue as well as the underlying motivations,
which can be quite complex. Parallel campaigns, which address both the
immediate legislative issues and more systemic concerns, can support each
other. We see this in some of the work undertaken by the organizations
addressed in this chapter, but would argue that more can be done (including
at the ICLMG).
 214 Tim McSorley and Anne Dagenais Guertin

Conclusion
As mass surveillance becomes more and more normalized and ubiquitous,
traditional and targeted campaigns and actions seem limited in their outcomes,
especially if one’s desired outcome is the abolition of mass surveillance. More
research and reflection are necessary to identify the real impact of our cam-
paigning and more radical methods to effect changes. The constant barrage of
new legislation slowly allowing more and more surveillance makes it difficult
to take the time to evaluate the efficiency of our advocacy, but it also makes it
all the more urgent.
A broader research project that investigates both the actions and thought
processes of campaigners as well as those of government officials through more
interviews, access to information requests, and primary source documentation
could help answer some of the outstanding questions raised in this chapter.
Other interesting avenues to explore could be international comparisons of
campaigns targeting similar surveillance concerns in different countries. For
example, the Five Eyes countries often adopt similar laws, so an examination
of successes and failures in different jurisdictions could be enlightening. Overall,
however, we believe these three cases show how important it is for civil liberties
and privacy groups to maintain long-term coordinated resistance to mass sur-
veillance, and provide some indications that such resistance will continue to
grow and hopefully achieve new victories in the coming years.

Notes
1 Tia Dafnos, Scott Thompson, and Martin French, “Surveillance and the Colonial Dream:
Canada’s Surveillance of Indigenous Self-Determination,” in National Security, Surveil-
lance and Terror, edited by Randy K. Lippert, Kevin Walby, Ian Warren, and Darren
Palmer (Basingstoke, UK: Palgrave Macmillan, 2016), 324.
2 Mark Neocleous, War Power, Police Power (Edinburgh: Edinburgh University Press,
2014), 31–35.
3 Cited in ibid., 33–34.
4 Ibid., 34.
5 Gary Kinsman,  Dieter K. Buse,  and  Mercedes Steedman,  “Introduction,” in Whose
National Security? Canadian State Surveillance and the Creation of Enemies (Toronto:
Between the Lines, 2000), 1–8; Luis A. Fernandez, Policing Dissent: Social Control and
the Anti-Globalization Movement (New Brunswick, NJ: Rutgers University Press, 2008),
94–107.
6 Paul Bernal, “Data Gathering, Surveillance and Human Rights: Recasting the Debate,”
Journal of Cyber Policy 1, 2 (2016): 243–64; Jon Penney, “Chilling Effects: Online
Surveillance and Wikipedia Use,” Berkeley Technology Law Journal 31, 1 (2016): 117,
SSRN, https://ssrn.com/abstract=2769645; Dafnos, Thompson, and French, “Surveillance
and the Colonial Dream,” 324.
7 Colin J. Bennett, Kevin D. Haggerty, David Lyon, and Valerie Steeves, eds., Transparent
Lives: Surveillance in Canada (Edmonton: Athabasca University Press, 2014), 19–37.
8 Ibid., 12.
 Confronting Big Data 215

9 Gregory S. Kealey, Spying on Canadians: The Royal Canadian Mounted Police Security
Service and the Origins of the Long Cold War (Toronto: University of Toronto Press,
2017), 1–13.
10 Canada, Bill C-36, Anti-terrorism Act, 1st Sess, 37th Parl, 2001, http://www.parl.ca/
DocumentViewer/en/37-1/bill/C-36/royal-assent.
11 Canada, Bill C-51, Anti-terrorism Act, 2015, 2nd Sess, 41st Parl, 2015, http://www.parl.
ca/DocumentViewer/en/41-2/bill/C-51/royal-assent.
12 Canada, Bill C-59, An Act respecting national security matters, 1st Sess, 42nd Parl,
2017, http://www.parl.ca/DocumentViewer/en/42-1/bill/C-59/second-reading [National
Security Act, 2017]; International Civil Liberties Monitoring Group (ICLMG),
“Breaking Down Bill C-59, the New National Security Act,” http://iclmg.ca/issues/
bill-c-59-the-national-security-act-of-2017/.
13 BC Civil Liberties Association (BCCLA), “BCCLA Sues Canadian Government to Stop
Illegal Spying,” https://bccla.org/stop-illegal-spying/protect-our-privacy-case-details/.
14 BCCLA, “Backgrounder on Spying: Civil Liberties Watchdog Sues Surveillance
Agency over Illegal Spying on Canadians,” 1 June 2016, https://bccla.org/wp-content/
uploads/2016/06/2016_06_02_Backgrounder-BCCLA-Sues-CSE-1.pdf.
15 Ibid.
16 BCCLA, “Illegal Spying: BCCLA Files Class Action Lawsuit against Canada’s Electronic
Spy Agency,” 1 April 2014, https://bccla.org/news/2014/04/illegal-spying-bccla-files-class
-action-lawsuit-against-canadas-electronic-spy-agency/.
17 James Keller, “Ottawa Says CSEC’s Collection of Canadians’ Data ‘incidental,’” CTV
News, 24 January 2014, https://www.ctvnews.ca/canada/ottawa-says-csec-s-collection
-of-canadians-data-incidental-1.1655231.
18 BCCLA to the Attorney General of Canada, Statement of Claim to the Defendant, 27
October 2014, https://bccla.org/wp-content/uploads/2014/12/20141027-CSEC-Statement
-of-Claim.pdf.
19 Ibid.
20 BCCLA, “Stop Illegal Spying,” https://bccla.org/stop-illegal-spying/.
21 BCCLA, “Spying in Canada: Civil Liberties Watchdog Sues Surveillance Agency
over Illegal Spying on Canadians,” 23 October 2013, https://bccla.org/wp-content/
uploads/2013/10/Final-Press-Release-Spying-10_21_131.pdf.
22 Greg Weston, “CSEC Used Airport Wi-Fi to Track Canadian Travellers: Edward
Snowden Documents,”  CBC News, 31 January 2014, http://www.cbc.ca/news/
politics/csec-used-airport-wi-fi-to-track-canadian-travellers-edward-snowden
-documents-1.2517881; Dave Seglins, “CSE Tracks Millions of Downloads Daily:
Snowden Documents,”  CBC News, 27 January 2015, http://www.cbc.ca/news/canada/
cse-tracks-millions-of-downloads-daily-snowden-documents-1.2930120.
23 Colin Freeze  and  Wendy Stueck, “Civil Liberties Groups Launch Lawsuit against
Canadian Eavesdropping Agency,” Globe and Mail, 22 October 2013, https://www.
theglobeandmail.com/news/national/canadian-eavesdropping-agency-facing-lawsuit
-from-civil-liberties-group/article14984074/; “Canadian Spy Agency Sued for Alleg-
edly Violating Charter,” CBC News, 22 October 2013, http://www.cbc.ca/news/canada/
british-columbia/canadian-spy-agency-sued-for-allegedly-violating-charter-1.2158884;
Gillian Shaw, “BC Civil Liberties Association Launches Lawsuit against Canada’s Electronic
Surveillance Agency,” Vancouver Sun, 22 October 2013, http://vancouversun.com/news/
staff-blogs/bc-civil-liberties-association-launches-lawsuit-against-canadian-government
-over-csec-spying.
24 Liam Britten, “BCCLA Says Warrantless Spying on Canadians Must End,” CBC News, 23 June
2016, http://www.cbc.ca/news/canada/british-columbia/bccla-cse-surveillance-1.3650286.
 216 Tim McSorley and Anne Dagenais Guertin

25 “The ‘Top Secret’ Surveillance Directives,” Globe and Mail, 2 June 2016, https://www.the
globeandmail.com/news/national/top-secret-surveillance-directives/article30249860/.
26 Ibid.
27 Monique Scotti, “Here’s What You Need to Know about Canada’s ‘Extraordinarily
Permissive’ New Spying Laws,” Global News, 6 February 2018, https://globalnews.ca/
news/3999947/cse-c59-new-spy-powers-canada/.
28 BCCLA, “Written Submissions of the British Columbia Civil Liberties Association
(‘BCCLA’) to the Standing Committee on Public Safety and National Security regarding
Bill C-59, An Act respecting national security matters,” 30 January 2018, http://www.
ourcommons.ca/Content/Committee/421/SECU/Brief/BR9669809/br-external/British
ColumbiaCivilLibertiesAssociation-e.pdf.
29 Christopher A. Parsons, Lex Gill, Tamir Israel, Bill Robinson, and Ronald J. Deibert,
“Analysis of the Communications Security Establishment Act and Related Provisions in
Bill C-59 (an Act Respecting National Security Matters), First Reading,” Transparency
and Accountability, December 2017, SSRN, https://ssrn.com/abstract=3101557.
30 Canada, Bill C-30, An Act to enact the Investigating and Preventing Criminal Electronic
Communications Act and to amend the Criminal Code and other Acts, 1st Sess, 41st Parl,
2012, http://www.parl.ca/LegisInfo/BillDetails.aspx?Language=E&billId=5375610.
31 Sarah Schmidt and Jason Fekete, “Vic Toews Will ‘Entertain Amendments’ to Online
Surveillance Bill,” National Post, 15 February 2012, http://nationalpost.com/news/
canada/protecting-children-from-internet-predators-act-vic-toews.
32 Gillian Shaw, “Stop Online Spying,” Vancouver Sun, 15 September 2011, https://vancou
versun.com/news/staff-blogs/stop-online-spying-openmedia-ca-launches-campaign
-against-web-surveillance-legislation/.
33 Schmidt and Fekete, “Vic Toews Will ‘Entertain Amendments.’”
34 Philippa Lawson, Moving toward a Surveillance Society: Proposals to Expand ‘Lawful
Access’ in Canada (Vancouver: BCCLA, January 2012), https://bccla.org/wp-content/
uploads/2012/03/2012-BCCLA-REPORT-Moving-toward-a-surveillance-society.pdf.
35 Laura Payton, “Internet Privacy Experts Raise Concerns over Crime Bill,” CBC News, 9
August 2011, http://www.cbc.ca/news/politics/internet-privacy-experts-raise-concerns
-over-crime-bill-1.1090482; Michael Geist, “The Conservatives Commitment to Internet
Surveillance,” Michael Geist (blog), 9 April 2011, http://www.michaelgeist.ca/2011/04/
conservative-lawful-access-commit/.
36 OpenMedia, “A Look Back at Our Stop the Meter Campaign,” https://openmedia.org/en/
ca/look-back-our-stop-meter-campaign.
37 OpenMedia, “Invasive Surveillance Bills Will Cost Canadians in Cash and Civil Liberties,
Says New Coalition,” 22 June 2011, https://openmedia.org/en/press/invasive-surveillance
-bills-will-cost-canadians-cash-and-civil-liberties-says-new-coalition.
38 OpenMedia to Right Honorable Prime Minister Stephen Harper, “RE: Omnibus Crime
Bill,” 9 August 2001, https://assets.documentcloud.org/documents/230754/letter-to-harper
-re-lawfulaccess.pdf.
39 Daniel Tencer, “‘Lawful Access’ Legislation Missing from Omnibus Crime Bill, but
Online Spying Fight Isn’t Over,” Huffington Post Canada, 20 September 2011, https://
www.huffingtonpost.ca/2011/09/20/lawful-access-legislation_n_971965.html.
40 Postmedia News, “Online Surveillance Bill Critics Are Siding with ‘Child Pornogra-
phers’: Vic Toews,” National Post, 14 February 2012, http://nationalpost.com/news/
canada/online-surveillance-bill-critics-are-siding-with-child-pornographers-vic-toews.
41 Melissa Martin, “TellVicEverything an Internet Sensation,” Winnipeg Free Press, 17 Feb-
ruary 2012,  https://www.winnipegfreepress.com/local/tell-vic-everything-an-internet
-sensation-139501528.html.
 Confronting Big Data 217

42 Steve Anderson, “Stop Online Spying Hits 100k: Canadians Are an Inspiration,” Open-
Media.ca, 17 February 2012, https://openmedia.org/en/stop-online-spying-hits-100k
-canadians-are-inspiration.
43 Ibid.
44 OpenMedia, “A Look Back at Our Stop Spying Campaign against Canada’s Bill C-30,”
https://openmedia.org/en/ca/look-back-our-stop-spying-campaign-against-canadas
-bill-c-30.
45 Canadian Press, “Conservatives Kill Controversial ‘Child Pornographers’ Internet Sur-
veillance Bill,” National Post, 11 February 2013, http://nationalpost.com/news/politics/
conservatives-kill-controversial-internet-surveillance-bill.
46 Evan Dyer, “Cyberbullying Bill Draws Fire from Diverse Mix of Critics,” CBC News,
20 October 2014, http://www.cbc.ca/news/politics/cyberbullying-bill-draws-fire-from
-diverse-mix-of-critics-1.2803637.
47 International Campaign Against Mass Surveillance (ICAMS), The Emergence of a Global
Infrastructure for Mass Registration and Surveillance (ICAMS, April 2005), https://
web.archive.org/web/20070109200500/http://www.i-cams.org/ICAMS1.pdf; ICAMS,
“The Emergence of a Global Infrastructure for Mass Registration and Surveillance: 10
Signposts,” https://web.archive.org/web/20061219231540/http://www.i-cams.org:80/
Surveillance_intro.html; ICAMS, “Campaign Declaration,” https://web.archive.org/
web/20061219231451/http://www.i-cams.org:80/Declaration_Eng.html.
48 Maureen Webb, Illusions of Security: Global Surveillance and Democracy in the Post-9/11
World (San Francisco: City Lights Books, 2007).
49 The Global Privacy Assembly (GPA), 18 May 2020, https://globalprivacyassembly.org/
the-assembly-and-executive-committee/history-of-the-assembly/; International Con-
ference of Data Protection and Privacy Commissioners (ICDPPC), “Resolution on the
Urgent Need for Global Standards for Safeguarding Passenger Data to Be Used by Govern-
ments for Law Enforcement and Border Security Purposes,” 29th International Confer-
ence, Montreal, 2007, http://globalprivacyassembly.org/wp-content/uploads/2015/02/
Resolution-on-Urgent-need-for-global-standards-for-safeguarding-passenger-data-to
-be-used-by-governments-for-law-enforcement-and-border-security-purposes.pdf.
50 ICDPPC, “Resolution on the Urgent Need for Protecting Privacy in a Borderless World,
and for Reaching a Joint Proposal for Setting International Standards on Privacy and
Personal Data Protection,” 30th International Conference, Strasbourg, France,  2008,
http://globalprivacyassembly.org/wp-content/uploads/2015/02/Resoltuion-on-the
-urgent-need-for-protecting-privacy-in-a-borderless-world.pdf.
51 Serge Ménard and Joe Comartin, “Anti-Terrorism Act Dissenting Opinion,” in Rights,
Limits, Security: A Comprehensive Review of the Anti-Terrorism Act and Related Issues,
Report of the Standing Committee on Public Safety and National Security, March 2007,
http://www.ourcommons.ca/DocumentViewer/en/39-1/SECU/report-7/page-69.
12
Protesting Bill C-51
Reflections on Connective Action against Big Data Surveillance
Jeffrey Monaghan and Valerie Steeves

Get a couple of beers in them and [privacy advocates] will fantasize about
what they call the “Privacy Chernobyl” – the one privacy outrage that will
finally catalyze an effective social movement around the issue.
– Philip Agre, cited in The Privacy Advocates:
Resisting the Spread of Surveillance

In October 2014, a homeless man killed a soldier standing guard by the National
War Memorial in downtown Ottawa using an antique single-shot rifle, and then
entered the front entrance of Parliament’s Centre Block. Within minutes, he
was shot and killed by parliamentary security. Days later, taking advantage of
this tragic yet exceptional act of violence, Prime Minister Stephen Harper’s
government proposed sweeping reform to the policing and security powers
contained in the Anti-terrorism Act. The legislation, known as Bill C-51, included
increased powers of surveillance and information sharing, as well as contentious
powers of disruption that would enable judges to sanction, in advance, police
actions that would explicitly violate the Canadian Charter of Rights and
Freedoms.1
Bill C-51 was among a number of surveillance and intelligence-sharing pro-
posals that had been floated to strengthen the ability of security agencies to
engage in big data practices of mass collection for the purpose of future pre-
emptive action through algorithmic induction; indeed, critics suggested that
the proposals were ready-made, simply waiting for the appropriate tragedy to
be officialized.2 In spite of the social backdrop of fear and anxiety, public criti-
cism of Bill C-51 quickly emerged. Cross-country mobilizations included a
number of national days of action that attracted thousands of protest participants
in fifty-five cities across the country.3 Newspapers, television news, and social
media exploded with debate, and thousands of Canadians signed petitions and
open letters of protest, gradually building a critical mass in opposition grounded
loosely on concerns about civil liberties, privacy politics, and policing powers
in the “War on Terror.”
Although an amended version of the bill was enacted into law, the public
response to Bill C-51 stands in stark contrast to the lack of public engagement
in privacy issues noted by Colin Bennett in his study of privacy advocacy. In
 Protesting Bill C-51 219

spite of the ongoing efforts of a long-standing coalition of civil society groups

and academics working to slow the spread of surveillance, Bennett rightly
concluded that efforts to reform privacy laws were largely in the hands of a
global cadre of “privacy experts” and anti-surveillance politics was “still gener-
ally an elitist issue within government, business and civil society.”4 Less than
ten years after the publication of Bennett’s study, grassroots opposition to
government surveillance would ignite.
In this chapter, we explore this shift and suggest that campaigning against
C-51 represented a unique moment in Canadian politics when surveillance
issues entered the domain of social movement mobilization. While resistance
to surveillance is often characterized as interventionist (e.g., artistic installations,
tactical disruptions, litigation), the C-51 campaign transcended a single issue
of privacy and instead mobilized broader publics and politics in resistance to
the expansion of policing powers and the prospects of enhanced big data sur-
veillance. Although the movement may not have been sustained over the long
term and was not rekindled when the subsequent Liberal government further
enhanced surveillance powers under Bill C-59 (discussed elsewhere in this
volume), we suggest that the mobilizations against C-51 provide certain strategic
and tactical insights for organizing against the incursions of big data
surveillance.
This chapter revisits aspects of the campaign against C-51 to provide perspec-
tive on contemporary efforts to expand and intensify surveillance, particularly
the data collection and information-sharing components of big data practices.
We conducted twelve semi-structured interviews with organizers in the cam-
paign. Interview participants were grouped according to three criteria: (1)
traditional privacy or civil liberties activists (five participants); (2) Internet and/
or broadly hacktivist (e.g., Internet freedom) activists (four participants); and
(3) non-traditional activists whose activism was not at all related to surveillance
but who joined the campaign (three participants).5
Based on data collected through these interviews, we develop two arguments
that can be taken from the C-51 mobilizations. First, the movement against C-51
is an excellent example of a heterogeneous movement of what Lance Bennett
and Alexandra Segerberg call connective action.6 The movement does not
conform to the collective action described by the strict Social Movement Organ-
ization (SMO) typology of social movement studies, but instead accords more
with emerging models of hybrid movements enabled by digitally mediated
personal action frames. Second, C-51 campaigning was successful as a hybrid
movement not only because it built on the repertoire, quasi-organizational
strengths, and public education campaigns of the previous decade vis-à-vis
warrantless access and digital privacy work but also because it took advantage
220 Jeffrey Monaghan and Valerie Steeves

of specific affordances available at the moment. In particular, we address the

affordances created by the Snowden revelations and the opportunities to use
the ineffectiveness of the political environment as a mechanism for public
education. We close with a discussion of the barriers or demobilization chal-
lenges that pre-dated the C-51 campaign and impacted the ability of surveillance
critics to challenge Bill C-59. Although these challenges will continue to be
relevant to any action against intrusive big data surveillance practices associated
with national security, we suggest that the connective actions used in the C-51
campaign may provide a roadmap for resisting big data surveillance in the
future.

Opposition to Bill C-51 as Organizationally Enabled Connective Action

Social movement scholarship has underlined how digitally enabled movements
have transformed the study of contemporary protests and/or efforts to bring
about political change.7 Much of the early literature on social movements focused
on collective action and evaluated the role of SMOs with respect to their mobil-
ization of resources or how they fit into political opportunity structures.
Especially because the literature was rooted in psychology and economics,
theorists posited SMOs as rational actors that sought to effect change within an
institutionally defined public sphere.
The rise of more fluid forms of digitally enabled movements has coincided
with the waning influence of traditional SMOs as a bedrock for mobilizing the
public.8 Emerging scholarship is accordingly directed towards the role of digital
technology in the formation and mobilization of a wide variety of different groups
and individuals – what Bennett and Segerberg call connective action – as well as
the new dynamics of group identities that bind movements and give them a sense
of “we-ness” despite their heterogeneity.9 In this sense, public mobilization is
addressed by looking less towards the role of institutional actors and more towards
what Ian Angus calls the “emergent publics”: an amorphous social sphere that
can be mobilized based on digital infrastructures of resonance.10 Zeynep Tufekci
uses the term “digitally networked public sphere” in recognition of how the public
and the digital have been reconfigured where neither domain is exclusive of the
other,11 but where digital environments have become central to what Paolo Ger-
baudo describes as “information diffusion, public discussions, and calls for
mobilization.”12 This suggests that within the digitally networked public sphere,
SMOs have become less central in the circulation of a mobilization economy
that involves far more plural, heterogeneous actors.
At the same time, as Bennett and Segerberg argue, SMOs continue to be an
important unit of analysis despite their diminished role because of their ongoing
visible participation in contemporary mobilizations that are enabled through
 Protesting Bill C-51 221

digital connections. Bennett and Segerberg accordingly apply “hybrid” network

theory as a middle space between the logics of collective versus connective
actions, and posit that hybrid networks feature “conventional organizations
[that] operate in the background of protest and issue advocacy networks to
enable personalized engagement.”13 These forms of “organizationally enabled
connective action” are characterized by the “stepping back [of formal SMOs]
from projecting strong agendas, political brands, and collective identities in
favor of using resources to deploy social technologies enabling loose public
networks to form around personalized action themes.”14
In these types of contemporary mobilizations, Bennett and Segerberg have
found three general characteristics: (1) familiar NGOs and other civil society
organizations engage in loose collaborations to provide something of a network-
ing backbone for (2) digital media networks to engage a broader public around
contested political issues, yet with (3) fewer efforts and abilities to brand the
issues around specific organizations, own the messages, or control the under-
standings of individual participants.15 We suggest that these characteristics of
organizationally enabled connective action are useful for understanding the
Bill C-51 protests as a hybrid social movement that balanced a degree of formally
organized action frames, but where SMOs enabled a broader public engagement
with issues of surveillance and privacy that resonated through digital spheres
of information consumption and diffusion. Given the lack of a coherent privacy
movement in Canada, the strength of the Bill C-51 protest was notable for its
hybridity. To begin with, the SMOs involved with issues related to privacy, civil
liberties, or surveillance in Canada are quite small. More appropriately charac-
terized as small advocacy groups or NGOs, these groups have limited experience
with mobilizing a public in line with a more traditional SMO model. “The
normal politics of privacy,” writes Bennett in the Privacy Advocates, “involves
the quite relentless and painstaking attempt to understand the policies and
proposals of government and business and to inject privacy argumentation and
reasoning.”16 Often involving technical and privacy-oriented engagements that
create what Jonathan Obar calls technocratic divides, privacy issues have typ-
ically lacked a politics of contention that form a basis for social movement
action.17
Yet our data suggest that the loose network of technically oriented actors
involved in privacy issues had formalized their connections over the past decade
or more of privacy- and surveillance-related issues. Blayne Haggart, for example,
suggests that the Fair Copyright campaign of the early 2010s was a moment when
the loose network of copyright advocates – closely associated with Internet
freedom activism – was galvanized into a social movement through digitally
enabled connective action.18 Similarly, a number of our participants underlined
222 Jeffrey Monaghan and Valerie Steeves

the importance of campaigns against lawful access laws as formalizing the hybrid
action network.19 The lawful access debates provide an excellent example of what
Jennifer Earl and Katrina Kimport describe as e-tactics and e-mobilizations.20
However, the movement against lawful access never translated from online
activism to protest mobilizations.
Based on the networks established through earlier campaigning, organizers
were able to quickly come together and establish a new campaign as soon as
the government introduced Bill C-51, using a variety of digitally mediated efforts.
As described by another participant:

[When] Bill C-51 [was announced] that’s when we really kicked into high gear
with the Charter challenge and petition and letter writing campaigns, social
media campaigns, basically all the tactics ... [that] organizations like ours can
use and we’ve been fighting that battle now for four years.21

Almost immediately, the C-51 campaign was also far more heterogeneous,
attracting the participation of activists who were engaged more broadly in
political/economic (e.g., labour unions) and Internet/Internet freedom issues
(e.g., net neutrality). As one participant put it, “a lot more groups – a bit bigger
diversity groups came in.”22 This brought with it a concomitant diversity of views
about government surveillance and policing, and opened up more space for
members of the public to engage with the campaign for their own purposes and
from their own personal action frames. It enabled participants to act in concert,
if not collectively, to push back against surveillance and move the opposition
from online campaigning into protest organizing.
We turn now to the factors that supported this shift.

Affordances in the Post-Snowden Environment

Affordance theories have been used to explain how more fluid, or hybrid, social
movement assemblages leverage specific opportunities to foster protest and
dissent.23 While affordance theories are often tied to technological repertoires,
a key distinction when using affordance theory is its contrast to “opportunities”
theories. Within SMO literature, opportunities are theorized statically: as precise
conditions where movement organizations apply a calculus to guide actions
towards social change. Affordances are far more fluid: they consider conten-
tiousness as constantly in flux where contemporary political actions are mediated
through social media and associative publics. Seeing connective actions as
including heterogeneous groups who may (or may not) share collective political
identities, affordance theory accords a broader scope of personal and organiza-
tional agency to political action.
 Protesting Bill C-51 223

In positing themselves against more traditional social movement theories

that weigh rational activity with organizational resources or political oppor-
tunities within a more rigid “public” sphere, scholars who discuss practices of
affordances stress the heterogeneous and mixed constellation of forces that are
provoked or harnessed within our increasingly “networked public.”24 Describing
how digitally enabled movements interact and make use of networked publics,
Earl and Kimport suggest that “leveraged affordances” have opened a range of
opportunities to mobilize publics both offline and online, while dramatically
lowering costs associated with creating, organizing, and participating in pro-
tests.25 While affordance theories have been critiqued for remaining overly
techno-socio,26 we stress that affordances are best conceptualized as opportun-
ities for connective action within hybrid social movements.
In the context of the anti-surveillance movement, our participants reported
that both collective and connective action had been limited pre-C-51 because
of the high level of secrecy surrounding policing and national security informa-
tion practices. The resulting lack of transparency made it difficult to hold agen-
cies to account through the traditional political process, or to provide digitally
enabled opportunities for people to protest. As one participant put it:

Here’s a statement of CSIS [the Canadian Security Intelligence Service]: “we will
neither confirm nor deny and we will not tell you whether in our opinion we
should get a warrant if we were ever to use one.” I’m sorry, we will not tell you
parliamentarians whether CSIS would get a warrant for the use of IMSI-catchers?
What is that? That is a shocking level of disdain for democratic process.27

The lack of solid information about how and when surveillance was being
mobilized made it particularly difficult to engage a public that was under-
informed and fearful about security:

I mean it’s such a pervasive issue, and the PR machine of the governments and
the spy agencies are so much bigger and they have so much ammunition to
throw out there, especially with, you know, the fear mongering going on in the
United States and around the world, it’s a hard, hard battle to fight.28

Interestingly, our participants indicated that these two factors also combined
to immobilize politicians and shut down dissenting voices among parliamentar-
ians: “There is such fear and so that fear is even affecting our policy makers. So
even if we can see sometimes and we can feel that yes they want to make these
changes, it’s like the political environment is really always pushing that in the
other direction.”29 The creation of “an imbalance towards security” makes it
224 Jeffrey Monaghan and Valerie Steeves

even more difficult to obtain any transparency from government agencies

engaged in surveillance. As one participant concluded, “again, we’re not asking
you for secret facts here. We’re not asking you for operational secrets. We’re
talking about the rule of law and CSIS flips the bird.”30 When asked why the
C-51 campaign was able to overcome the barrier of publicly debating security
policies, many of our participants cited the Snowden revelations of 2013.
Although much of the material released by Edward Snowden regarding the
surveillance net created by the Five Eyes partners (Australia, Canada, New
Zealand, the United Kingdom, and the United States) had been previously
discussed in traditional media by whistle-blowers, the Snowden documents
were seen as a central element of the success of the C-51 campaign. In what
Bennett might call a “privacy Chernobyl,”31 the Snowden documents – including
information detailing the role of Canadian agencies – were quickly leveraged
into wide-ranging debates regarding mass surveillance and privacy. One par-
ticipant recounted:

I mean it’s a weird trade-off where I think that everyone knew that it was hap-
pening, but on the other side of the same coin you have the fact that if you bring
that up, you’re basically a tinfoil hat person. And I think that Edward Snowden
made this more of a socially acceptable thing to talk about versus something that
people didn’t know. People already knew it was happening, but it was something
that they didn’t really want to talk about because they didn’t want to seem like
the crazy person. But now it’s more acceptable to talk about.32

Another participant described the difficulties in addressing allegations of mass

surveillance without having hard evidence, while also emphasizing the richness
of the revelations that came from Snowden’s leaks: “I make a joke about that
that for a month after the Snowden revelations my outgoing message on my
answering machine said, ‘I told you so.’ There were surprises even within the
privacy community.”33
As David Lyon has noted, the Snowden revelations have “done good service
in showing how far state-based surveillance extends but also how much it
depends on Big Data practices that implicate corporate bodies and connect
directly with everyday practices of ordinary internet and cellphone users.”34
Participants echoed these claims by highlighting how the detailed evidence
Snowden provided fed the critiques of mass surveillance and how the conse-
quences of these big data surveillance practices were used in their public advo-
cacy work. One participant noted: “But what Snowden did I think had a huge
impact, not only raising the level of general understanding about what was
going on, but actually putting tools in the hands of people who had been trying
 Protesting Bill C-51 225

to crack this nut for a long time.”35 The transparency created by the Snowden
revelations provided an important affordance for work on surveillance and
privacy, and was also highlighted specifically as an important factor in the
debates that emerged over Bill C-51: “I think Snowden helped [because it] really
raised the ... salience of those issues [surveillance] to a broader set of people,
which by the time C-51 rolled along ... I think helped [by] going outside the
privacy bucket.”36
That enlarged “privacy bucket” also provided a repertoire to the public at
large. This was particularly true in the use of social media to tweet or share
information regarding the impacts of C-51 in expanding information-sharing
and surveillance practices. Using social media affordances that appeared as a
result of the massive circulation of Snowden-related content, the hybridity of
the C-51 movement created opportunities to leverage the ineffectiveness of the
political environment. Often engagement with capital-P politics is conceived
as an effort to reform or direct the legislative process, or the platforms of political
parties.37 Yet, from the perspective of C-51 movement participants, the political
engagement around the proposed bill created opportunities to engage in public
education and movement building. This was particularly true regarding work
associated with attending parliamentary committee hearings. As with traditional
SMOs, participating in institutional moments of lawmaking allowed C-51 critics
to reach a much broader audience than merely the politicians themselves because
of the media attention the bill attracted. But it also provided opportunities to
leverage this focused attention to educate journalists and members of the public
about the issues.
As one participant said:

It’s not just a resonance ... [it’s that] you get to the education component of it.
Like you do build on it every time. So with C-51 – starting with the initial lawful
access stuff, we had a lot of education to do – like journalists first. Like not –
education’s maybe the wrong word, but like we would have to – like it would
take a while for journalists to get why this is a problem, how it works, etc.,
right? ... but we went through that process and then they get it ... then eventually
they get it.38

A prominent example of this dynamic involved the issue of metadata, where

the technology would have to be explained in the most basic form, but then
journalists gradually understood and presented more comprehensive criticisms
of the “just a postcard” argument. In order to get the criticisms to the public,
some messaging work and explanations were developed first with the journalists
that involved “explaining it over and over.”39
226 Jeffrey Monaghan and Valerie Steeves

Describing parliamentary committees as “a form of theatre,” another partici-

pant remarked that “it is valuable [just] being able to get your position on the
record and frankly, in Hansard,” but the broader objective is to move “the public
debate.”40 “I want to expand the public’s sphere and expand the sphere of public
space, and so those types of things are useful moments for that.”41 Again, the
hybridity of the movement amplified the reach of debate beyond the more
structured format of mainstream journalism by first engaging with journalists
to expand their understanding of the issue, and then tweeting, blogging, and
forwarding the information to feed into a broader public discussion.
In this way, our participants were able to work around the intransigence of
the Conservative government of the day and work with a heterogeneous group
of advocates and members of the public to mobilize against the bill. This was
unexpected, particularly given the “tough on crime” stance of the Harper gov-
ernment and the nature of the attack in October 2014. In addition, the content
of the bill itself raised few expectations that committee members were going to
be open to significant debate or amendments. As one participant put it, “I’m
not a lawyer but I, I asked our, our law-legal guy ‘does that mean what I think
it means?’ And he was like ‘yeah, this is pretty bad.’ So pretty much right away,
right on day one we knew we were in for a hell of a fight.”42
In the early period after Bill C-51 was tabled, a loose network of privacy and
civil liberties groups formed alliances with non-traditional privacy groups to
educate and mobilize against the bill. One participant describes the early period:
“We worked with a bunch of other organizations on a petition that ended up,
getting over 300,000 signatures, which in a country the population of Canada
is remarkable.”43 The connection with online activists was particularly fruitful,
as it enabled the movement towards social media affordances to support a
number of connective actions. These included petitions, social media campaigns,
and several days of action. Consistent with the hybrid model, the SMOs took
a backseat role; although the network tried to “craft a unified message,” each
member of the movement could “then take [it] to our individual constituencies
and take it to the press and blast out across Canada.”44
This set the stage that enabled social media calls for a national day of action
to transform the campaign into a unique moment of large protest around
privacy and surveillance issues. “It went viral and so it got like thousands of
people signing up for this National Day of Action,” described a participant
from a non-traditional organization that joined the loose network because of
potential chilling effects on its activist work. “And so from our perspective
the Day of Action was a huge success ... And the rally here for example, in
[city] was quite successful. We had over 5,000 people show up and [a] myriad
of speakers.”45
 Protesting Bill C-51 227

Describing the success in bringing a diverse group of participants to C-51

protests, a participant from a non-privacy background of activism noted:

I went to the one protest here in [place] and seeing like, you know, people out
on the streets of like [place] and [place] and like small towns right across the
country on Bill C-51 was I think remarkable. So I think there’s a huge public
appetite out there for change.46

And that appetite for change was embraced by a wide variety of people who
were mobilized by their own personal action frames. A participant from a
traditional privacy group noted: “I participated in marches, in protests, in rallies
and they were great, really great, great opportunities and great moments I would
say to see people, not the usual suspects.”47 A participant from a non-privacy
organization agreed:

I mean a number of people from the C-51 campaign who were saying things
like “oh this is my first time getting active in anything political.” The number of
people ... I’ve been to a bunch of different protests in [area] but was the – what
was interesting about the C-51 one was – it definitely didn’t seem to be like the
usual kind of left wing kind of crowd that you would get.48

As a hybrid movement, the Bill C-51 days of action were created through a
network that had been partially solidified from years of privacy activism
related to surveillance, yet it was infused with organizations that had little
history in protesting surveillance or privacy-related issues. Moreover, the
affordances provided by the Snowden revelations, the negative political
environment, and social media allowed for broad resonance and connective
action with participants – not the “usual suspects” but those who had found
what Bennett and Segerberg describe as “personalized” appeals to the protests.
Although SMOs played an enabling role, the protests themselves became
heterogeneous and far more diverse than previous moments of anti-
surveillance activism.
As an example of organizationally enabled connective action, the C-51 move-
ment was highly effective at changing public opinion on the bill. A combination
of the political affordances and the mass mobilizations dramatically altered the
framing that had been initially fostered by the Harper government. On partici-
pant recalled:

You know, when C-51 was first brought on the books, something like 80 per cent
of Canadians supported it and then that shifted almost immediately ... it was
228 Jeffrey Monaghan and Valerie Steeves

almost a complete flip ... the fact that we were able to get so many Canadians so
quickly to change their minds on it is incredible. Like, you don’t see polls slip
like that, ever ... I think the fact that we’ve gotten the media so on board with it
is a huge success.49

This participant, who deals primarily with non-privacy–associated activism,

summarized the success of the campaign as follows:

So I think we took a rather obscure topic and made it a point of national dis-
cussion ... I think we’re able to sort of shift the frame, not for everyone but for
a lot of people away from pure like security terrorism, to democratic debate,
freedom, which I think is also helpful for a whole series of other debates that
need to happen around, you know, net neutrality and a bunch of other things,
where something that seems kind of obscure or almost benign actually is chal-
lenging, sort of like undercutting a lot of the underpinnings of [the] other types
of freedoms we have.50

Framing connective action movements as heterogeneous amalgamations of

actors is useful for understanding two successes that arose from the C-51
mobilizations. First, the concept of connective action helps explain how the
campaign was able to move beyond the traditional frame of privacy advocacy –
largely an elite exercise in legislative intervention – and merge SMO activities
with mobilizations undertaken by a wide range of non-traditional actors
approaching the problem from a variety of frames. This amplified the efforts
of both SMOs and non-traditional actors because the deep reservoir of privacy-
specific expertise was made available to support the interventions of others
who could mobilize the information in their own ways and for their own
purposes. Key to this was the willingness of the SMOs to let go of the need to
define either the issues or the solutions. As Bennett and Segerberg note, ten-
sions can arise when particular SMOs (especially political parties) vie to brand
movements, yet the SMOs involved in the C-51 campaign demonstrated a
tendency to remain more or less in the background. Second, the movement
was successful in using affordances, as the participant above noted, to shift
the frame of debate away from the “security” threats presented by the govern-
ment. In part, this was accomplished by the use of personalized action frames
that dramatically expanded the parameters of the debate, allowing an identi-
fication with a host of issues that connect to surveillance, privacy, democratic
rights, policing powers, and so on. In expanding beyond the domains of
privacy and anti-surveillance, the arena of political action was opened to more
broad-based contentious political associations.
 Protesting Bill C-51 229

Discussion/Conclusions: Challenges in the Era of Big Data

In a rare example of public protests over surveillance, the organizationally
enabled connective action against Bill C-51 parlayed opposition against mass
surveillance into a flashpoint for anti-surveillance protests. Transparencies
stemming from the Snowden revelations could be characterized as a “privacy
Chernobyl” – an affordance that provided different contexts for people to inter-
pret the negative consequences of Bill C-51 from their own personal action
frames. SMOs could provide alternative perspectives on issues to support debate
but they by no means had a leadership role during the protests. Instead, a broad
and diverse array of concerned members of the public joined in protests against
the bill. However, despite successes arising from the hybrid movement against
Bill C-51, barriers continue to present obstacles to movements seeking to chal-
lenge big data surveillance. These challenges have been notable in recent efforts
to contest the dramatic expansion of surveillance powers in Bill C-59 and,
although they pre-date the C-51 protests, they remain just as (or even more)
present in working against intrusive big data surveillance practices associated
with national security.
Our participants were particularly cognizant of the enduring nature of these
barriers. First was the secrecy associated with security governance practices,
especially practices associated with big data. One participant recalled:

We know SIRC [the Security Intelligence Review Committee] in its astonishing

last annual report that does the first ever audit of CSIS’ bulk data collections, bulk
data holdings. And it’s simply [a] gobsmacking finding that there is no evidence
of any attempt to employ the correct legal standard in relation to those databases.
We know the ongoing concerns with CSE [Communications Security Estab-
lishment], etc., etc. So check all those boxes. What are they all related to? They
relate to bulk data. What are they about? They’re about data analytics. We know
that national security has been twisting themselves into pretzels to be able to for
a minimum of the last 10 years – we think obviously much longer – to collect
huge amounts of scraping of bulk data, C-51 of course giving them almost carte
blanche to go dipping into it. They actually refer to “ingesting entire databases.”51

While bulk data collection has been taking place for over a decade, security
and policing agencies such as the RCMP, CSE, and CSIS have deliberately misled
the public regarding the scope and objectives of its practices. The same agencies
have also withheld vital information regarding bulk data collection from the
courts. This development was spectacularly revealed in 2016 when the Federal
Court lampooned CSIS for not meeting its duty of candour and deliberately
withholding information on bulk data collection for ten years, including the
230 Jeffrey Monaghan and Valerie Steeves

existence of the Operational Data Analysis Centre (ODAC). Huysmans has

characterized these securitarian practices as forms of political and security
thickness.52 In the current configurations, we see both forms of thickness
deployed to make mass surveillance simultaneously diffuse and secretive. While
political thickness is most associated with speech acts and the politics of fear
that are used to rationalize more security and disallow – or delegitimize – cri-
tiques of security, security thickness refers to practices of security that are
embedded in the proliferating fields of security and policing. In many aspects,
security thickness is more acute – and more anti-democratic – in the sense that
these fields of practices evade public knowledge due to their sensitive and spe-
cialized character. As one participant put it, “even for us who sit there and stare
at those documents for months on end trying to figure out how it’s gonna be
used and talk to experts and whatever, it takes a while to get a fairly still murky
picture [chuckles].”53
While the visibilities of social life are made more pronounced, including
algorithmically, the visibility of security practices will continue to become more
mercurial. Challenges to security thickness will likely accelerate and require
sustained work from researchers, academics, and the legal community –
the traditional domain of privacy advocates – although many participants raised
the broader obstacle of having to mobilize the public in an increasingly frag-
mented environment. As a challenge that impacts social movements in the
digitally mediated world, these dynamics are not unique to anti-surveillance
movements, though some particular characteristics are worth reflecting upon.
One dynamic raised in the context of a discussion of post-Snowden political
implications was the overwhelming character of surveillance-related informa-
tion. In making a point about potential ironies of the post-Snowden affordance,
one participant noted that the public is overwhelmed with information about
privacy and surveillance. A constant barrage of newspaper stories and discus-
sions about surveillance often becomes “episodic discussions” about various
technologies – but with little sustained debate or education. The outcome, sug-
gested the participant, is that “they’re just going to get overwhelmed. They’re
like, ‘okay, well I’m screwed.’ And they probably are to be honest. They probably
are just screwed [chuckles].”54
The dangers of a post-privacy society are perhaps overstated, yet they correspond
with a number of challenges facing SMOs working against mass surveillance.
Oversaturated members of the public may increasingly tune out surveillance-
related concerns, potentially closing the post-Snowden affordance. Post-privacy
discourse is also mobilized by security and policing entities as a technique of
security thickness, granting more space and opaqueness to their work. This is
particularly evident with what one participant called the “allure of big data.”55
 Protesting Bill C-51 231

Indeed, one of the most challenging dynamics in contesting big data is the sup-
posed benefits – or “big data solutionism”56 – espoused by its advocates. Appeals
to big data solutionism are certainly not exclusive to security governance issues,
but the dynamics of secrecy and potential for injustice that are rendered opaque
by appeals to big data are particularly acute with security issues.
Despite the potential of post-privacy attitudes or the discourse of big data
solutionism to obscure the ethical implications of the practices of mass data
collection and algorithmic governance, the connective action that propelled
Bill C-51 mobilizations may provide insight into practices that can challenge
efforts to expand and intensify surveillance in the future. As Bennett and Seger-
berg have noted, the success of connective action results from the ability to “cast
a broader public engagement net using interactive digital media and easy-to-
personalize action themes, often deploying batteries of social technologies to
help [people] spread the word over their personal networks.”57 Shifts in con-
temporary society towards personalized politics present challenges to the more
traditional models and requirements of collective action, and a number of
movements have made use of “personalized varieties of collective action” to
spark public mobilizations.58
The movement from collective to connective action is represented as a move
away from central organizations and a strong collective identity, as well as an
opening of a broader field of political engagement. “Clicktivism,” for example,
has been associated with what Max Halupka characterizes as small, impulsive,
non-ideological, political actions, such as clicking “likes” on Twitter or Facebook
in an effort to raise awareness or contribute to social change through one’s
personal social media networks.59 A far cry from the foundations of resource
mobilization theories that require strong group identity and a recognition of
political opportunities almost exclusively tied to capital-P politics, the personal-
ized character of connective action offers more grounds for fluid, spontaneous
political engagement. Yet, while connective action might be highly effective for
contemporary mobilizations, from flash mobs to protest camps, are these new
dynamics of public contestations effective at social change?
The broad spectrum of engagement around C-51 shows both an advantage
and a disadvantage of connective action. While personalization can draw people
into spontaneous resistance, the lack of organizational cohesion can result in
quick dissipation and disaggregation unless SMOs adapt to shifting needs and
continue to support new digital debates as they arise. Moreover, a growing area
of concern – and attention – considers how digitally enabled actions have them-
selves become networked sites of mass surveillance, something we need to be
very skeptical about.60 Although the C-51 mobilizations demonstrate how con-
nective action can translate into contentious politics, it is difficult to establish
 232 Jeffrey Monaghan and Valerie Steeves

whether the contentiousness of such anti-surveillance politics have translated

into action frames for broader resistance against both Bill C-59 and future
surveillance-enhancing initiatives.
For connective action against creeping mass surveillance to continue, the
hybrid organizational dynamics of the movement will have to create new per-
sonalizations that likely require new flashpoints or affordances to draw con-
nectivity among members of the public along a spectrum of social justice issues,
including, but not limited to, surveillance. Barriers of secrecy and media over-
saturation are not new challenges, and strategies to mobilize can evolve from
new repertoires of social movement action.

Notes
1 Craig Forcese and Kent Roach, False Security: The Radicalization of Canadian Anti-
Terrorism (Montreal: Irwin Law, 2015).
2 Ibid.; Tamir Israel and Christopher Parsons, “Canada’s National Security Consultation I:
Digital Anonymity & Subscriber Identification Revisited ... Yet Again” (Canadian Inter-
net Policy and Public Interest Clinic report, 2016), https://cippic.ca/uploads/20161005-
CNSCI-RevisitingAnonymityYetAgain.pdf.
3 Michael Shulman, “Demonstrators Across Canada Protest Bill C-51,” CTV News, 14
March 2015, http://www.ctvnews.ca/politics/demonstrators-across-canada-protest-bill
-c-51-1.2279745.
4 Colin Bennett, The Privacy Advocates: Resisting the Spread of Surveillance (Cambridge,
MA: MIT Press, 2008), 207. See also Colin Bennett and Charles Raab, The Governances
of Privacy (Cambridge, MA: MIT Press, 2006).
5 Participant groups are denoted in the text as Traditional, Internet, and Other. Each inter-
viewee has been given a number, e.g., Traditional 1.
6 W. Lance Bennett and Alexandra Segerberg, “The Logic of Connective Action: Digital
Media and the Personalization of Contentious Politics,” Information, Communication
and Society 15, 5 (2012): 739–68.
7 Ibid.; Jennifer Earl and Katrina Kimport, Digitally Enabled Social Change: Activism in the
Internet Age (Cambridge, MA: MIT Press, 2011); Paolo Gerbaudo, Tweets and the Streets:
Social Media and Contemporary Activism (London: Pluto Press, 2012).
8 Jennifer Earl, “The Future of Social Movement Organizations: The Waning Dominance
of SMOs Online,” American Behavioral Scientist 59, 1 (2015): 35–52.
9 See debates on collective identity: Paolo Gerbaudo, The Mask and the Flag: Populism,
Citizenism, and Global Protest (Oxford: Oxford University Press, 2017); Emiliano Treré,
“Reclaiming, Proclaiming, and Maintaining Collective Identity in the #YoSoy132 Move-
ment in Mexico: An Examination of Digital Frontstage and Backstage Activism through
Social Media and Instant Messaging Platforms,” Information, Communication and Soci-
ety 18, 8 (2015): 901–15.
10 Ian Angus, Emergent Publics: An Essay on Social Movements and Democracy (Winni-
peg: Arbiter Ring, 2001). See also Nancy K. Baym and danah boyd, “Socially Mediated
Publicness: An Introduction,” Journal of Broadcasting and Electronic Media 56, 3 (2012):
320–29.
11 Zeynep Tufekci, Twitter and Tear Gas: The Power and Fragility of Networked Protest (New
Haven, CT: Yale University Press, 2017).
 Protesting Bill C-51 233

12 Gerbaudo, The Mask and the Flag, 135.

13 Bennett and Segerberg, “The Logic of Connective Action,” 754. See also Andrew Chad-
wick, “Digital Network Repertoires and Organizational Hybridity,” Political Communica-
tion 24, 3 (2007): 283–301; Sarah Anne Rennick, “Personal Grievance Sharing, Frame
Alignment, and Hybrid Organisational Structures: The Role of Social Media in North
Africa’s 2011 Uprisings,” Journal of Contemporary African Studies 31, 2 (2013): 156–74;
Scott Wright, “Populism and Downing Street E-Petitions: Connective Action, Hybridity,
and the Changing Nature of Organizing,” Political Communication 32, 3 (2015): 414–33.
14 Bennett and Segerberg, “The Logic of Connective Action,” 754, 757.
15 Ibid., 758.
16 Bennett, Privacy Advocates, 113.
17 Jonathan Obar, “Closing the Technocratic Divide? Activist Intermediaries, Digital Form
Letters, and Public Involvement in FCC Policy Making,” International Journal of Com-
munication 10 (2016): 5865–88.
18 Blayne Haggart, “Fair Copyright for Canada: Lessons for Online Social Movements from
the First Canadian Facebook Uprising,” Canadian Journal of Political Science/Revue can-
adienne de science politique 46, 4 (2013): 841–61. See also Dan Mercea and Andreas
Funk, “The Social Media Overture of the Pan-European Stop-ACTA Protest: An Empiri-
cal Examination of Participatory Coordination in Connective Action,” Convergence 22,
3 (2016): 287–312; Jonathan Obar and Leslie Shade, “Activating the Fifth Estate: Bill
C-30 and the Digitally-Mediated Public Watchdog,” 23 July 2014, SSRN, https://ssrn.
com/abstract=2470671 or http://dx.doi.org/10.2139/ssrn.2470671.
19 For an overview of the lawful access debates, see Christopher Parsons, “Stuck on the
Agenda: Drawing Lessons from the Stagnation of ‘Lawful Access’ Legislation in Canada,”
in Law, Privacy and Surveillance in Canada in the Post-Snowden Era, edited by Michael
Geist (Ottawa: University of Ottawa Press, 2015), 257–84.
20 Earl and Kimport, Digitally Enabled Social Change, 3–20.
21 Interview, Other 2.
22 Interview, Internet 4.
23 Ryan Calo, “Can Americans Resist Surveillance?” University of Chicago Law Review
83, 1 (2016): 23–43; Earl and Kimport, Digitally Enabled Social Change; Jeffrey Juris,
“Reflections on #Occupy Everywhere: Social Media, Public Space, and Emerging Log-
ics of Aggregation,” American Ethnologist 39, 2 (2014): 259–79; Jonathan Obar, “Cana-
dian Advocacy 2.0: An Analysis of Social Media Adoption and Perceived Affordances
by Advocacy Groups Looking to Advance Activism in Canada,” Canadian Journal of
Communication 39, 2 (2014): 211–33.
24 danah boyd, “Social Network Sites as Networked Publics: Affordances, Dynamics, and
Implications,” in Networked Self: Identity, Community, and Culture on Social Network
Sites, edited by Zizi Papacharissi (New York: Routledge, 2010), 39–58.
25 Earl and Kimport, Digitally Enabled Social Change, 10–12.
26 Gerbaudo, The Mask and the Flag; Juris, “Reflections on #Occupy.”
27 Interview, Traditional 1.
28 Interview, Other 2.
29 Interview, Traditional 2.
30 Interview, Traditional 1.
31 Bennett, Privacy Advocates, 200, 209–10.
32 Interview, Internet 2.
33 Interview, Traditional 1.
34 David Lyon, “Surveillance, Snowden, and Big Data: Capacities, Consequences, Critique,”
Big Data and Society 1, 2 (2014): 11.
 234 Jeffrey Monaghan and Valerie Steeves

35 Interview, Traditional 1.
36 Interview, Internet 4.
37 See Calo, “Can Americans Resist Surveillance?”
38 Interview, Internet 4.
39 Ibid.
40 Interview, Internet 3.
41 Ibid.
42 Ibid.
43 Interview, Other 3.
44 Interview, Other 2.
45 Interview, Other 1.
46 Interview, Other 3.
47 Interview, Traditional 2.
48 Interview, Other 3.
49 Interview, Other 2.
50 Ibid.
51 Interview, Traditional 1.
52 Jef Huysmans, Security Unbound: Enacting Democratic Limits (New York: Routledge,
2014).
53 Interview, Internet 4.
54 Interview, Traditional 3.
55 Interview, Traditional 5.
56 See Ganaele Langlois, Joanna Redden, and Greg Elmer, “Introduction,” in Compromised
Data – From Social Media to Big Data, edited by Ganaele Langlois, Joanna Redden, and
Greg Elmer (New York: Bloomsbury, 2014), 1–14.
57 Bennett and Segerberg, “The Logic of Connective Action,” 742.
58 Ibid., 743. See also W. Lance Bennett, “The Personalization of Politics: Political Identity,
Social Media, and Changing Patterns of Participation,” Annals of the American Academy
of Political and Social Science 644, 1 (2012): 20–39.
59 See Max Halupka, “Clicktivism: A Systematic Heuristic,” Policy and Internet 6, 2 (2014):
115–32.
60 See Lucas Melgaço and Jeffrey Monaghan, “Introduction: Taking to the Streets in the
Information Age,” in Protests in the Information Age: Social Movements, Digital Practices
and Surveillance, edited by Lucas Melgaço and Jeffrey Monaghan (New York: Routledge,
2018), 1–17.
Part 5
Policy and Technical Challenges of
Big Data Surveillance
This page intentionally left blank
13
Horizontal Accountability and Signals Intelligence
Lessons Drawing from Annual Electronic Surveillance Reports
Christopher Parsons and Adam Molnar

One of my biggest takeaways from the past 16 months is that we need to be

more transparent. And, if we’re going to profess transparency, we need to
produce transparency, wherever we can.
– James Clapper remarks at the 2014 AFCEA/
INSA National Security and Intelligence Summit

The Communications Security Establishment (CSE) is Canada’s foremost

signals intelligence (SIGINT) agency. Historically it has collected foreign signals
intelligence, provided security and defensive information technology services
to the government of Canada and systems critical to the government of Canada,
and assisted domestic federal law enforcement and security agencies (LESAs).1
CSE’s activities are guided by parliamentary legislation and by the Minister of
National Defence through ministerial authorizations and directives. The former
can authorize CSE to engage in practices that would otherwise violate Canadian
law without criminal liability, and the latter principally establishes conditions
or limitations on the kinds of lawful activities CSE may conduct.2 All of CSE’s
activities are subject to review by the Office of the Communications Security
Establishment Commissioner (OCSEC).3
CSE’s activities are routinely concealed from the public eye, with legislators
and the public mainly reliant on the principles of ministerial responsibility,
OCSEC reviews, rare unauthorized disclosures for classified activities, and
(marginal) judicial oversight to ensure that CSE’s activities comport with
law. This present system of accountability that governs CSE activities has
often been criticized as insufficient in the media and among some analysts.4
And while legislation that was tabled in Parliament in 2017 may significantly
restructure this historical relationship between CSE, its minister, and the
OCSEC, and thus how CSE is rendered accountable to its minister and the
public alike, we argue that both the current and proposed review and over-
sight of CSE are insufficient to provide public accountability. We address
these shortcomings by offering principle-based suggestions for facilitating
such accountability.
238 Christopher Parsons and Adam Molnar

We begin by unpacking the concepts of accountability, transparency, and

democratic legitimacy as linked to lawful government surveillance activities.
Next, we describe some of CSE’s more controversial activities to reveal deficien-
cies in how its activities have historically been framed through legislation and
publicly reviewed by its commissioner. The combined effect of such legislative
framing and reviews has been to undermine assurances that CSE’s activities
could be democratically legitimated. We argue that the tabled legislative reforms
that would affect CSE’s accountability structures would be insufficient to rectify
CSE’s public accountability deficits. We conclude by sketching a principle-based
framework that could ensure that CSE’s activities are both made accountable
to its minister and select parliamentarians, and as transparent as possible to
Canadians, and, as a result, democratically legitimated.

Conceptual Terminology
Organizations that act transparently collate and present data to those outside
the organization.5 This disclosure of information can sometimes present data
that are useful for the public.6 Often, organizations act transparently when they
are compelled to present information in a delimited format7 or through their
own methodologies to collate and disclose information.8 In either case, organ-
izations that “behave transparently” may be attempting to engender greater
trust in their practices.9 On this basis, scholars are advised to pay “careful atten-
tion to the human and material operations that go into the production of
transparency”10 because the revelatory character of transparency practices may
be overemphasized absent critique.
One way that governments, in particular, demonstrate transparency is through
the release of statutorily required reports. Electronic surveillance reports are
an attempt to address social inequity in the social contract between governments
and their citizens. By disclosing the regularity at which government surveillance
practices occur, the disproportionate degree of intrusion of the state into the
private lives of citizens is thought to be safeguarded. In contrast, the absence of
any requirement to disclose these activities, or a failure to release such reports,
can hinder legislatures and the citizenry from holding the government to
account.11 Without information about secretive government practices, the public,
parliamentarians, and other stakeholders cannot evaluate whether government
agencies are using their exceptional powers appropriately and in ways that
cohere with public interpretations and expectations of how the law ought to
legitimate such activities.12
Transparency in government activities is needed to ensure that civic agencies
are held accountable to their minister, to Parliament, and to the public more
broadly. A system of accountability exists “when there is a relationship where
 Horizontal Accountability and Signals Intelligence 239

an individual or institution, and the performance of tasks of functions by that

individual or institution, are subject to another’s oversight, direction or request
that the individual or institution provide information of justification for its
actions.”13 In effect, an institution must be obligated to answer questions and
there must also be a means of enforcing consequences should the institution
refuse or fail to provide satisfactory responses.14 In the context of a parliament-
ary democracy such as Canada, accountability can be manifested through
ministerial responsibility or other formalized methods that empower the legis-
lature to scrutinize an agency’s practices.15 However, accountability also exists
through more informal measures, such as when non-governmental stakeholders
hold government to account based on information tabled by government min-
isters or the government’s independent officers.16
There are several ways to understand accountability.17 In this chapter, we focus
exclusively on informal, or horizontal, modes of accountability between govern-
ment and non-governmental stakeholders. This mode can be contrasted with
vertical accountability, which often involves ministers being formally compelled
to account for their departments’ activities to their respective legislatures.18
Whereas ministers are obligated to explain their departments’ activities and
policies to the legislature, and the legislature is empowered to receive explana-
tions and justifications and subsequently issue sanctions as appropriate,19 the
same is not true with regard to the government’s relationship with external
stakeholders. Horizontal accountability institutes accountability through civil
engagement as a way to complement and enhance government accountability
processes.20 External stakeholders, however, cannot necessarily impose sanc-
tions, and governments are not always required to provide an account to these
stakeholders.21 In place of formal legal tools, moral suasion is routinely used to
sanction government behaviours. And while the disclosure of ethical impro-
priety and accompanying use of moral suasion may be amplified by the media,
it is rarely premised on stakeholders having formal powers to compel the gov-
ernment to provide an account or modify its behaviours.22
The practice of holding governments to account is intended to control gov-
ernment conduct. Citizens can exert control through the ballot box as well as
outside electoral periods.23 Stakeholders engaged in horizontal accountability
can work to identify problems so that legislators, or the government itself, can
take up and attempt to solve challenging issues.24 Moreover, through a proactive
civil culture that proposes solutions to problems, government and legislators
may realize previously unconsidered ways to correct them. External stakehold-
ers can also testify or present information to government committees or mem-
bers of the legislature. But for any of these means of exercising horizontal
accountability to work, external stakeholders must have access to government
240 Christopher Parsons and Adam Molnar

information, possess a capacity to take on the work of ingesting and processing

the information in question, and recognize that the state is capable, willing, and
competent to receive external actors’ concerns and has the potential ability to
act on them.25 Absent information provided by government, citizens may be
inhibited from participating in political processes; such secrecy “compels the
public to defer to the judgement of a narrow elite.”26
By remaining open to external stakeholder analysis, critique, and problem
solving, a government combats cynicism or doubts that it is not “of the people,
for the people.” An inability on the part of government to respond to civil society
interests fosters cynicism and doubts about whether legislators can, or desire
to, represent the citizenry. While most citizens may not be involved in holding
their government to account, broader perceptions of accountability may be
shaped by the government’s receptiveness to civil society interventions.27 If the
electorate fails to see its representatives respond on policy issues raised by
stakeholders, they may lose faith in legislators, and by extension in the repre-
sentative democratic process of lawmaking itself.28 Even if a government and
its departments act based on laws passed within a legislative assembly, without
adequate horizontal accountability laws may be seen as severed from the legit-
imating power of the citizenry itself. Such disconnection threatens to transform
a democratic process bound through rule of law into a narrow and disconnected
process that might be better understood as rule-with-law.29 Severing “lawful
activities” from democratic legitimation processes has been recognized as a
core challenge that the second generation of intelligence oversight must over-
come. Whereas in the past such oversight and review were concerned with
detecting and preventing abuse and mischief, the second generation must
reconcile economic, diplomatic, and strategic goals as well as secure the “consent
of the governed” where public concerns are linked with the need for secrecy.30

Making the Past Clear?

CSE was formally established as part of the National Defense Act (NDA), though
its origin dates back to the end the Second World War, when it secretly existed
in different government departments.31 The NDA imposed three mandates on
CSE: (A) to “acquire and use information from the global information infra-
structure for the purpose of providing foreign intelligence”; (B) to “provide
advice, guidance and services to help ensure the protection of electronic infor-
mation and of information infrastructures of importance to the Government
of Canada”; and (C) to “provide technical and operational assistance to federal
law enforcement and security agencies in the performance of their lawful dut-
ies.”32 The breadth of these mandates became truly apparent only following
Edward Snowden’s disclosure of classified national security documents to
 Horizontal Accountability and Signals Intelligence 241

journalists, who subsequently selectively published from what they were given.
One of the most prominent Canadian-focused Snowden disclosures was a
program covernamed CASCADE. CASCADE was operated on non–government
of Canada networks and was designed to analyze network traffic. The analysis
involved discovering and tracking targets, as well as isolating content or metadata
from traffic exposed to the network probes.33 Within the CASCADE program
was a series of differently classified and covernamed network sensors. Some
could capture metadata and content alike (EONBLUE and INDUCTION),
whereas others could solely collect and analyze metadata (THIRD-EYE and
CRUCIBLE).34 All of these sensors relied on deep packet inspection technology,
which enables operators to analyze the metadata and content of unencrypted
communications and take actions on it, such as blocking certain traffic or
modifying other traffic.35
INDUCTION operated at “Special Source Operations (SSO) sites,” or within
the premises of private Canadian organizations that had consented to CSE’s
activities. CRUCIBLE sensors, similar to INDUCTION sensors, were located
in the pathways of networks that were designated “systems of importance” to
Canada.36 Such systems might belong to defence contractors, extractive resource
companies, banks, or equivalent organizations whose compromise could
detrimentally affect the governance of Canada. These sensors could also collect
the metadata of communications that Canadians, and persons communicating
with Canadians, were engaged in, as well as the metadata of devices that trans-
mitted information into or out of Canada. Other aspects of CASCADE involved
monitoring satellite communications as well as microwave towers that trans-
mitted data.37
The purpose of CASCADE, when combined with an equivalent sensor
network designed to protect the Canadian government’s own networks (cover-
named PHOTONIC PRISM, which was expected to be replaced by EON-
BLUE),38 was to use the entirety of the global information infrastructure as a
means of defence. By tracking threat actors and their activities, CSE intended
to “affect changes at the CORE of the Internet on detection” in collaboration
with its Five Eyes partners. Such changes included modifying traffic routes,
silently discarding malicious traffic, or inserting payloads into communica-
tions traffic to disrupt adversaries.39 To achieve these ends, CASCADE would,
in essence, be situated to grant fulsome awareness of domestic and foreign
Internet activity throughout the world. The most controversial aspects of this
program in Canada were principally linked to the extensive surveillance of
Canadian-source, Canadian-bound, and Canadian domestic traffic, as well
as CSE’s efforts to work alongside private partners to conduct this global
surveillance activity.
242 Christopher Parsons and Adam Molnar

Fuzzy Mandates, Clarified?

The different mandates that CSE operates under authorize a broad spectrum of
activities, including network discovery, exploitation, and attack; defensive cyber
operations; the creation of information profiles useful for other agencies that
engage in physical operations; and other activities intended to further or advance
the missions of other government agencies.40 The program discussed above
reveals how seemingly restrictive mandates can be interpreted as authorizing
mass surveillance practices in excess of imagined restrictions.
The CASCADE program goes beyond the concept of erecting a network
perimeter and defending it in depth by envisioning monitoring of the entirety
of the domestic and international Internet so that CSE can track all data emis-
sions that might be harmful to Canadian interests. If Mandate B was principally
considered to be instructing CSE to shield certain systems, the Snowden docu-
ments revealed that CSE took shielding domestic institutions to mean engaging
in global mass surveillance as a prerequisite for such defensive policies. While
monitoring data traffic internationally arguably falls under Mandate A, the
identification of domestic networks of interest and subsequent generation of
domestic content and metadata from these networks runs counter to Canadians’
perceptions that CSE was not authorized to routinely monitor Canadians’
activities.41 Indeed, in internal slides, CSE recognizes that providing “defence”
using CASCADE engages all three of its mandates: A, B, and C.42
Though the CSE is formally prohibited from deliberately collecting the per-
sonal communications content of Canadians or persons residing in Canada,
the agency operates with a ministerial authorization that permits it to collect
such data incidentally in the course of its operations – that is, CSE cannot direct
its surveillance apparatus in a deliberate way towards specific or named Can-
adians or Canadian targets unless it is providing assistance to a foreign agency
under warrant.43 But these restrictions are not interpreted by the Canadian
government or the OCSEC to preclude CSE from monitoring all metadata
emanations from persons within Canada,44 even though CSE, its minister, and
its review body know that CSE has the capability to reidentify the persons with
whom the emanations are associated. The OCSEC’s conclusion that CSE behaved
lawfully in the collection of metadata pertaining to Canadians’ communications
and devices was unsurprising: independent analysts have found that it is almost
impossible for any activity conducted by CSE to be found unlawful given the
nature of the OCSEC’s role and interpretations of national security law.45
In 2017, the Canadian government introduced Bill C-59, which, among other
things, was designed to clarify CSE’s mandate while simultaneously updating
the control and review structure for Canada’s intelligence agencies. Based on
the Snowden revelations, it was apparent that CSE was involved in a broader
 Horizontal Accountability and Signals Intelligence 243

range of activities than many thought was already likely given its scope and
perceived capabilities. While Bill C-59 may retroactively authorize these existing
activities, it has made more explicit the expansive range of CSE’s activities, which
include collecting foreign intelligence through the global information infra-
structure; engaging in cybersecurity and information assurance; conducting
defensive operations to broadly protect federal institutions’ systems and those
deemed of importance to Canada; performing active cyber operations that may
involve degrading, disrupting, influencing, responding to, or interfering with
“the capabilities, intentions or activities” of non-Canadian parties; and provid-
ing technical and operational assistance to LESAs, the Canadian Forces, and
the Department of National Defence.46 There are provisions within the CSE Act
that also permits CSE to collect information from any public source,47 including
perhaps grey market information brokers, as well as interfere with non-
democratic foreign elections,48 among other controversial measures.
The program that we have examined in this chapter can be situated within
this expanded mandate. CASCADE could operate simultaneously under the
collection of foreign intelligence, cybersecurity and information assurance,
and (potentially) assistance mandates. When viewed through each of these
mandate areas, CSE is permitted to acquire information as required, provide
services to different government and non-governmental organizations that
are meant to guarantee the respective organizations’ digital security, and use
collected information as appropriate to assist domestic LESAs or foreign-
operating Canadian Forces to act on parties threatening Canadian organiza-
tions’ digital systems. If it obtains authorization, activities in Canada could
extend to active defensive operations. Furthermore, Bill C-59 explicitly
authorizes CSE to infiltrate any part of the global information infrastructure
for the purposes of collecting information that would provide foreign intel-
ligence. This includes the types of attacks being launched towards Canadian
networks or systems of interest, and also permits private companies to
cooperate with CSE and, as such, operate as SSOs. Whereas CSE’s current
legislation does not explicitly state the conditions under which it can engage
with private organizations (as envisioned under the CASCADE program),
the cybersecurity authorizations for non-federal infrastructures under Bill
C-59 establish the legislative framework for such cooperation. Notably, C-59
also includes emergency provisions for access to private organizations’ infra-
structure. These provisions might let CSE gain permission from either the
operator of infrastructure, such as a party that is running software on, say,
computer servers in a shared computing facility or, alternatively, from the
party that owns the servers and leases them to the software-running party.49
This can occur without having to get the activity approved by anyone besides
244 Christopher Parsons and Adam Molnar

CSE’s minister. Such access might be needed in some cases to establish,

expand, or re-establish the defensive perimeter envisioned as part of the
CASCADE program.
Beyond providing a broader range of activities that CSE might engage in,
Bill C-59 also re-envisions how CSE’s activities are authorized, controlled,
and reviewed. Ministers will continue to issue authorizations and directives
that guide and delimit the types of activities that CSE can engage in, with
the minister of foreign affairs generally being consulted prior to CSE’s
engaging in defensive cyber operations or active offensive cyber operations.
The Intelligence Commissioner, a new control-type body that was created
as part of C-59, is typically responsible for (among other things) approving
foreign intelligence authorizations and cyber security authorizations, and
must also be notified of, and approve, significant amendments or repeals
of such authorizations.50 The Intelligence Commissioner provides annual
reports to the minister. CSE’s activities are subject to review by the National
Security and Intelligence Review Agency (NSIRA), and thus assume respon-
sibilities for CSE’s reporting paralleling those held by the OCSEC. Neither
the NSIRA nor any other body, including a committee of parliamentarians
that reports principally to the prime minister, is required to evaluate whether
or not CSE’s activities are normatively appropriate or to focus extensively
on whether they might, even if they are “lawful,” still unnecessarily infringe
upon Canadians’ civil liberties. In the United States, the Privacy and Civil
Liberties Oversight Board (PCLOB) provides this kind of external oversight
of activities undertaken by the National Security Agency (NSA) and produces
classified reports for the government as well as reports that are accessible
to the public.
Bill C-59 requires both CSE and the NSIRA to produce annual reports.
NSIRA’s reports must include information about CSE’s compliance with law
and ministerial authorizations, and about the reasonableness and necessity
of CSE’s use of its powers. It does not, however, require or authorize the
NSIRA to produce annual reports similar to those about the US National
Security Agency. Those reports include statistics on the numbers of Amer-
icans targeted by the NSA under Foreign Intelligence Surveillance Act (FISA)
Title I and Title III warrants and the proportion of persons targeted who
are non-US versus US persons; estimates of the number of non-US targets
affected by section 702 surveillance orders; the number of search terms used
to query the section 702 database that concern a known US person and aim
to retrieve the unminimized contents of their communications; and the
number of section 702–based reports that contain US persons’ identity
information, among other statistics.51
 Horizontal Accountability and Signals Intelligence 245

The Performance of Legislative Legitimacy and Accountability

Bill C-59 is designed in part to reform how CSE is controlled and reviewed, and
Bill C-22 established a committee of parliamentarians to evaluate some of CSE’s
activities and report on them to the Prime Minister’s Office. Though judicial
and other ways of evaluating the lawfulness of CSE’s activities are important,
they are limited in notable ways. As Kent Roach discusses, “even at its heroic
best, judicial oversight will focus on issues of legality and propriety, not efficacy
and effectiveness. Intelligence agencies will also have incentives – and often the
ability – to take measures that avoid or limit any inconvenient judicial over-
sight.”52 Similarly, while the NSIRA is designed to prevent CSE and its partner
agencies from avoiding or limiting review, members of the Canadian intelligence
community have historically been willing to mislead judges and downplay
questionable rationales of action to their reviewers.53 Furthermore, the very
structure of accountability raises some critical problems when it comes to roles
played by legislators. For instance, “[g]iving legislators access to secret informa-
tion but no mechanism for revealing their concerns may only allow the govern-
ment to claim legitimacy for illegal and improper conduct ... Rather than relying
on its members, much of the legitimacy of a legislative committee might come
from constructive engagement with civil society.”54 Under Bill C-22, parliamen-
tarians will be restricted in what they can examine, what they can report publicly,
and who they can appoint as their chair and members.55 So, while it is possible
that the new control and review structures will improve accountability internal
to formal government practices, nothing in Bill C-59 or the previously enacted
Bill C-22 necessarily establishes enhanced public reporting of CSE’s activities
and thus do not promote horizontal accountability of CSE’s activities.
Promoting horizontal accountability is vitally important to restore public
trust in CSE. According to Zachary Goldman and Samuel Rascoff, trust “is,
perhaps, the single most important determinant of how intelligence agencies
will fare in liberal democracies.”56 Goldman argues separately that “[t]he [Snow-
den] leaks really, then, revealed” a lack of social agreement about the proper
contours of the rules, “including about whether current interpretations of key
constitutional provisions are consistent with society’s expectations, rather than
about significant illegal behaviour. Debates also revolved around policy choices
by the [intelligence community] in areas where there is no direct legal author-
ization, such as whether the [National Security Agency] should stockpile zero-
day exploits, or whether it should monitor communications of lawful foreign
intelligence targets such as a foreign leader.”57 To promote horizontal account-
ability and restore the trust deficit between the population and CSE’s lawful
activities, the government might amend Bill C-59 or table new legislation that
specifies certain statistical and narrative accounts of CSE’s activities, as well as
246 Christopher Parsons and Adam Molnar

establish an independent review body responsible for evaluating the propor-

tionality of the activities.
Any efforts to ensure that CSE is subject to horizontal accountability could
include the following modes of transparency:
• Legal transparency. Decisions issued by the Federal Court should be made
public and minimally redacted to assist external legal experts and scholars in
understanding the development and shaping of law. As discussed by Daphna
Renan, “[m]aking the overarching legal framework of surveillance programs
more visible and participatory may make these programs more resilient ... the
fundamental legal framework of intelligence programs belongs in the light.”58
• Statistical transparency. The Office of the Director of National Intelligence in the
United States voluntarily produces statistical reports concerning the National
Security Agency’s annual operations. While statistics may leave much to be
desired, they show that information concerning the annual activities of the NSA
can be disclosed without undue harm to national security. Reported informa-
tion could also disclose the regularity with which CSE provides assistance to
domestic LESAs to assuage concerns that CSE is routinely directing its activities
towards Canadians or persons in Canada. A form of this reporting has been
undertaken in Canada since the 1970s, and involves the federal and provin-
cial governments issuing annual electronic surveillance reports that detail the
regularity and efficacy of federal and provincial agencies’ surveillance activities.
To date, there is no evidence that such statistical transparency has negatively
affected ongoing or concluded domestic LESA investigations.
• Narrative transparency. Legal or statistical transparency should be accompa-
nied by narratives that help clarify the rationales for the actions undertaken
by CSE. Such narratives should provide some information about the specific,
annual activities of CSE and not merely refer to authorizing legislation under
which the Establishment operates; though recent annual electronic surveillance
reports in Canada generally fail to provide a useful narrative example to follow,
federal reports pre-dating the mid-1990s that explain the situations associated
with such surveillance may be a useful starting point for what such narrative
explanations might include. Similarly, the narratives associated with the Office
of the Director of National Intelligence’s annual statistical reports indicate pos-
sible ways to explain how laws are interpreted and acted upon.
• Proportionality transparency. Though the review structures under Bill C-59 are
expected to evaluate whether CSE’s activities are reasonable or necessary for
the agency to exercise its powers,59 review and control bodies are not expected
to focus on whether CSE’s activities are proportionate to the impacts on civil
liberties that result from those activities. The minister is required to take the
 Horizontal Accountability and Signals Intelligence 247

proportionality of a measure into consideration before issuing a ministerial

authorization, but this process is internal to government.60 An external civil lib-
erties board, similar to the PCLOB in the United States, could report on whether
the specific activities undertaken by CSE are reasonable and proportionate when
viewed against their intrusion into citizens’ and foreigners’ private lives alike.

Three of these measures of transparency are born from accountability report-

ing that the federal and provincial governments already include in their annual
electronic surveillance reports. Such reports clarify the laws that authorize the
surveillance, the regularity with which such surveillance is conducted, and the
broad impacts of the surveillance, and they are supposed to provide some nar-
rative explanation of the surveillance. The fourth measure we propose, focused
on proportionality transparency, draws from measures established by Canada’s
close allies. Admittedly, the proposals we make extend beyond what the current
annual electronic surveillance reports include – these current reports do not,
for example, include discussions or decisions linked to secret case law associated
with wiretapping or other live forms of surveillance – but, importantly, our
proposals are not a radical adoption of entirely novel forms of government
transparency and accountability.
Promoting transparency of government intelligence operations would result
in important gains for horizontal accountability. Stakeholders could play a role
in providing critical insights and analyses to parliamentary committees, legis-
latures, and regulatory bodies that routinely experience resource shortages or
lack appropriate technical expertise. These stakeholders, who are often area
experts, could play an important role by representing their communities’ inter-
ests in debating the often thorny issues of secretive government surveillance
activities that historically have “touched” the information of far more Canadian
citizens and residents as well as visitors to Canada than previously suspected.
In general, emboldening horizontal accountability through meaningful public
disclosure can inform the broader democratic process in an area of governance
that is well known for its capacity to engender distrust and skepticism among
the citizenry. Elsewhere, we have noted how even when legislation might exist
to authorize a particular secret activity, information asymmetries between
government lawyers and the public mean that the lawfulness of an activity may
lack legitimation given the disconnect between legislation, law, and practice.
In effect, by becoming more transparent in secret operations and thereby better
enabling horizontal accountability processes, the lawful activities that are
undertaken may be subject not just to critique but also to approval of how a
measure is authorized and the policies to safeguard against misconduct, over-
breadth, or infringements on civil liberties.
248 Christopher Parsons and Adam Molnar

Conclusion
Patrick Walsh and Seumas Miller have argued that “[t]he Snowden leaks now
provide the opportunity for ‘Five Eyes’ governments to do a root and branch
review of current organizational, ministerial, parliamentary and other standing
oversight bodies to ensure they remain fit for purpose.”61 Goldman has separately
insisted that “although the institutions designed to ensure compliance work
well, these same institutions have difficulty with a broader role.”62 We agree with
these points and argue that a review of the intelligence community and its
transparency and accountability structures must also consider how to empower
stakeholders external to government to better engage in horizontal account-
ability. Indeed, in an environment characterized by rapid technological innova-
tion, extensive legal ambiguities, and associated tensions with traditional liberal
democratic principles, horizontal accountability is an essential component of
meaningful regulation.
In this chapter, we have argued that horizontal accountability can help legit-
imate secretive government activities that are authorized by legislation. We
proposed four separate measures, focused on legal, statistical, narrative, and
proportionality, to enhance the information available to external-to-government
stakeholders. This information could then be taken up and used to understand
and critique some activities, while also ensuring that parties external to govern-
ment could identify and propose solutions to thorny legal issues, could better
explain the protections and safeguards established to protect civil liberties and
human rights, and ensure that the stakeholders they represent are better
informed about the actual, versus hypothetical or hyperbolic, issues linked to
government surveillance activities.
A continuation of the status quo, where citizens are kept in the dark concern-
ing the activities and laws that authorize secret intelligence activities, “undermines
the capacity of citizens to determine whether a new balance of security concerns
and basic rights has been struck.”63 The status quo also threatens to magnify the
already disturbing gap between legislation as it is written, as it is interpreted by
Department of Justice and other government national security lawyers, and as
it is acted upon by Communications Security Establishment staff. This gap fun-
damentally threatens the legitimacy, if not the lawfulness, of CSE’s activities. No
government party benefits from the perpetuation of this gap: while it may
be tactically helpful in advancing specific operations or activities, it ultimately
threatens to poison the legitimacy of organizations themselves and, by extension,
turn tactical outputs into components of a broader strategic blunder.
Ultimately, it is only once citizens, often facilitated by academic and civil
society actors, know what is being done in their name, and why and how those
measures are linked to the activities authorized by their legislators, can the
 Horizontal Accountability and Signals Intelligence 249

accountability gap be bridged. As Canada undertakes national security consul-

tations today and into the future, and engages in legislative action and reforms,
the time to start bridging the gap is now.

Acknowledgments
Financial support for the research, authorship, and publication of this chapter was provided
by the John D. and Catherine T. MacArthur Foundation.
The authors would like to thank the participants of a national security round table held
at the 2017 Annual Citizen Lab Summer Institute for their insights concerning the CSE
Act and other relevant aspects of Bill C-59. They would also like to thank members of the
Communications Security Establishment for providing numerous briefings about different
aspects of the Establishment’s mandate, challenges it seeks to overcome, and how Bill C-59
might affect its practices.
The authors declare no potential conflicts of interest with respect to the research,
authorship, and/or publication of this chapter.

Notes
1 National Defence Act, RSC 1985, c N-5, ss 273.64(1)(a)–(c).
2 Office of the Communications Security Establishment Commissioner (OCSEC), “Fre-
quently Asked Questions,” 24 February 2017, https://www.ocsec-bccst.gc.ca/s56/eng/
frequently-asked-questions.
3 Ibid.
4 Bill Robinson, “Does CSE Comply with the Law?” Lux Ex Umbra (blog), 14 March 2015,
https://luxexumbra.blogspot.ca/2015/03/does-cse-comply-with-law.html; Ronald Deib-
ert, “Who Knows What Evils Lurk in the Shadows?” OpenCanada.org, 27 March 2015,
https://www.opencanada.org/features/c-51-who-knows-what-evils-lurk-in-the-shadows/;
Greg Weston, Glenn Greenwald, and Ryan Gallagher, “CSEC Used Airport Wi-Fi to
Track Canadian Travellers: Edward Snowden Documents,” CBC News, 30 January 2014,
https://web.archive.org/web/20140131064055/https://www.cbc.ca/news/politics/csec
-used-airport-wi-fi-to-track-canadian-travellers-edward-snowden-documents-1.2517881.
5 Robert Bushman, Joseph Piotroski, and Abbie Smith, “What Determines Corporate
Social Transparency?” Journal of Accounting Research 42, 2 (2004): 207; Sylvester Eigffin-
ger and Petra Geraats, Government Transparency: Impacts and Unintended Consequences
(New York: Palgrave Macmillan, 2006).
6 Roger Cotterrell, “Transparency, Mass Media, Ideology and Community,” Journal for
Cultural Research 3, 4 (1999): 414–26.
7 Archon Fung, Mary Graham, and David Weil, Full Disclosure: The Perils and Promise of
Transparency (New York: Cambridge University Press, 2007).
8 Ibid., 7; Christopher Parsons, “The (In)effectiveness of Voluntarily Produced Transpar-
ency Reports,” Business & Society 58, 1 (2019): 103–31, https://journals.sagepub.com/doi/
full/10.1177/0007650317717957.
9 Kent Wayland, Roberto Armengol, and Deborah Johnson, “When Transparency Isn’t Trans-
parent: Campaign Finance Disclosure and Internet Surveillance,” in Internet and Surveillance:
The Challenges of Web 2.0 and Social Media, edited by Christian Fuchs, Kees Boersma, Anders
Albrechtslund, and Marisol Sandoval (New York: Routledge, 2012), 239–54.
10 Hans Krause Hansen, Lars Thoger Christensen, and Mikkel Flyverbom, “Introduction:
Logics of Transparency in Late Modernity: Paradoxes, Mediation and Governance,”
European Journal of Social Theory 18, 2 (2015): 117–31.
 250 Christopher Parsons and Adam Molnar

11 Douwe Korff, Ben Wagner, Julia Powles, Renata Avila, and Ulf Buermeyer, “Boundaries
of Law: Exploring Transparency, Accountability, and Oversight of Government Surveil-
lance Regimes” (University of Cambridge Faculty of Law Research Paper No. 16/2017, 3
March 2017), SSRN, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2894490.
12 Christopher Parsons and Tamir Israel, “Gone Opaque? An Analysis of Hypo-
thetical IMSI Catcher Overuse in Canada” (Citizen Lab/Canadian Internet Policy
and Public Interest Clinic report, August 2016), https://citizenlab.org/wp-content/
uploads/2016/09/20160818-Report-Gone_Opaque.pdf; Adam Molnar, Christopher Par-
sons, and Erik Zouave, “Computer Network Operations and ‘Rule-with-Law’ in Australia,”
Internet Policy Review 6, 1 (2017): 1–14.
13 Riccardo Pelizzo and Frederick Stapenhurst, Government Accountability and Legisla-
tive Oversight (New York: Routledge, 2013), 2.
14 Andreas Schedler, “Conceptualizing Accountability,” in The Self-Restraining State: Power
and Accountability in New Democracies, edited by Andrew Schelder, Larry Diamond,
and Marc Plattner (Boulder, CO: Lynne Rienner, 1999), 13–28; Andrew Blick and
Edward Hedger, “Literature Review of Factors Contributing to Commonwealth Public
Accounts Committees Effectively Holding Government to Account for the Use of Pub-
lic Resources” (National Audit Office, Overseas Development Institute, 2008); Richard
Mulgan, “The Process of Public Accountability,” Australian Journal of Public Accountabil-
ity 56, 1 (1997): 26–36; Jonathan Anderson, “Illusions of Accountability,” Administrative
Theory & Praxis 31, 3 (2009): 322–39.
15 Dale Smith, The Unbroken Machine: Canada’s Democracy in Action (Toronto: Dundurn,
2017); Bruce Stone, “Administrative Accountability in the ‘Westminster’ Democracies:
Towards a New Conceptual Framework,” Governance 8, 4 (1995): 502–25.
16 Carmen Malena, Reigner Forster, and Janmejay Singh, “Social Accountability: An Intro-
duction to the Concept and Emerging Practice” (Social Development Paper 76, World
Bank, 2004).
17 See, for example, Richard Mulgan, “‘Accountability’: An Ever-Expanding Concept?” Pub-
lic Administration 78, 3 (2000): 555–73; Linda Deleon, “Accountability in a ‘Reinvented’
Government,” Public Administration 76, 3 (1998): 539–58; Amanda Sinclair, “The Cha-
meleon of Accountability: Forms and Discourses,” Accounting, Organizations and Society
20, 2–3 (1995): 219–37; David Corbett, Australian Public Sector Management, 2nd ed.
(Sydney: Allen and Unwin, 1996); James March and Johan Olsen, Democratic Gover-
nance (New York: Free Press, 1995).
18 Smith, The Unbroken Machine; Stone, “Administrative Accountability.”
19 J. LI. J. Edwards, “Ministerial Responsibility for National Security as It Relates to the
Offices of the Prime Minister, Attorney General and Solicitor General of Canada,” in
The Commission of Inquiry Concerning Certain Activities of the Royal Canadian Mounted
Police (Ottawa: Supply and Services Canada, 1980); Donald Savoie, Breaking the Bargain:
Public Servants, Ministers, and Parliament (Toronto: University of Toronto Press, 2003).
20 Malena, Forster, and Singh, “Social Accountability.”
21 Mark Bovens, “Analyzing and Assessing Accountability: A Conceptual Framework,”
European Law Journal 13, 4 (2007): 447–68.
22 Malena, Forster, and Singh, “Social Accountability”; Maxwell McCombs, Setting the
Agenda: Mass Media and Public Opinion (Hoboken, NJ: John Wiley and Sons, 2014).
23 Bovens, “Analyzing and Assessing Accountability.”
24 Alisdair Roberts, “Transparency in the Security Sector,” in The Right to Know: Transparency for
an Open World, edited by Ann Florini (New York: Columbia University Press, 2007), 309–36.
25 Malena, Forster, and Singh, “Social Accountability.”
26 Roberts, “Transparency in the Security Sector.”
 Horizontal Accountability and Signals Intelligence 251

27 Jan Aart Scholte, “Civil Society and Democracy in Global Governance,” Global Gover-
nance 8, 3 (2002): 281–304; Julie Fisher, Non Governments: NGOs and the Political Devel-
opment of the Third World (West Hartford, CT: Kumarian Press, 1998).
28 Jürgen Habermas, “On the Internal Relation between the Rule of Law and Democracy,”
in The Inclusion of the Other: Studies in Political Theory, edited by Ciaran Cronin and
Pablo De Greiff (Cambridge, MA: MIT Press, 1998), 253–64; Jürgen Habermas, “Three
Normative Models of Democracy,” in Cronin and De Greiff, 239–52; Christopher Par-
sons, “Beyond Privacy: Articulating the Broader Harms of Pervasive Mass Surveillance,”
Media and Communication 3, 3 (2015): 1–11.
29 Ben Bowling and James Sheptycki, “Global Policing and Transnational Rule with law,”
Transnational Legal Theory 6, 1 (2015): 141–73; Molnar, Parsons, and Zouave, “Computer
Network Operations.”
30 Zachary K. Goldman and Samuel J. Rascoff, “The New Intelligence Oversight,” in Global
Intelligence Oversight: Governing Security in the Twenty-First Century, edited by Zachary
K. Goldman and Samuel J. Rascoff (New York: Oxford University Press, 2016); see also
“Intelligence Reform in a Post-Snowden World,” YouTube video, 1:28:12, from a panel
hosted by the Center for Strategic and International Studies, 9 October 2015, https://
www.csis.org/events/intelligence-reform-post-snowden-world-0.
31 National Defence Act; Bill Robinson, “An Unofficial Look inside the Communications
Security Establishment, Canada’s Signals Intelligence Agency,” Lux Ex Umbra (blog), 5
November 2000, http://circ.jmellon.com/docs/html/communications_security_establishment_
unofficial_webpage_020623.html.
32 National Defence Act, ss 273.64(1)(a)–(c).
33 Communications Security Establishment (CSE), “CSEC Cyber Threat Capabili-
ties: SIGINT and ITS: An End-to-End Approach” (slide deck, October 2009), https://
christopher-parsons.com/Main/wp-content/uploads/2015/03/doc-6-cyber-threat
-capabilities-2.pdf.
34 CSE, “CASCADE: Joint Cyber Sensor Architecture,” 2011, Technology, Thoughts and
Trinkets, https://christopher-parsons.com/writings/cse-summaries/#cse-cascade-joint.
35 CSE, “CSEC Cyber Threat Capabilities”; Christopher Parsons, “Deep Packet Inspection in
Perspective: Tracing Its Lineage and Surveillance Potentials” (working paper, New Trans-
parency Project, 2008), http://www.sscqueens.org/files/WP_Deep_Packet_Inspection_
Parsons_Jan_2008.pdf.
36 CSE, “CASCADE: Joint Cyber Sensor Architecture.”
37 Ibid.
38 CSE, “Cyber Network Defence R&D Activities,” 2010, Technology, Thoughts and Trin-
kets, https://christopher-parsons.com/writings/cse-summaries/#cse-cyber-threat-capabilities;
CSE, “CASCADE: Joint Cyber Sensor Architecture.”
39 CSE, “CSEC Cyber Threat Capabilities.”
40 Deibert, “Who Knows What Evils Lurk in the Shadows?”
41 Weston, “CSEC Used Airport Wi-Fi to Track.”
42 CSE, “CSEC Cyber Threat Capabilities.”
43 Based on discussions between the authors and senior CSE staff, we understand that in
such warranted cases, information is cordoned off from CSE’s more general reposito-
ries and thus inaccessible to many, if not all, CSE staff and operations.
44 “Defence Minister Insists Spy Agency Did Not Track Canadian Travellers,” CTV News,
31 January 2014, http://www.ctvnews.ca/canada/defence-minister-insists-spy-agency
-did-not-track-canadian-travellers-1.1664333; OCSEC, “Frequently Asked Questions.”
See also Craig Forcese, “Faith-Based Accountability: Metadata and CSEC Review,” National
Security Law: Canadian Practice in Comparative Perspective (blog), 13 February 2014,
 252 Christopher Parsons and Adam Molnar

https://www.craigforcese.com/blog/2014/2/13/faith-based-accountability-metadata
-and-csec-review.html?rq=faith-based%20accountability%3A%20metadata%20
and%20csec%20review.
45 Robinson, “Does CSE Comply with the Law?”
46 Canada, Bill C-59, An Act respecting national security matters, 1st Sess, 42nd Parl, 2017,
pt 3, ss 17–21 [Bill C-59].
47 Ibid., pt 3, s 24(1)(a).
48 Ibid., s 33(1)(b).
49 Ibid., s 41(4).
50 The tabled bill initially included a caveat: the Intelligence Commission is not required to
first approve emergency authorizations (pt 3, s 42(2)).
51 See, for example, US Office of the Director of National Intelligence (ODNI), “Sta-
tistical Transparency Report Regarding Use of National Security Authorities for
Calendar Year 2016,” April 2016, https://www.dni.gov/files/icotr/ic_transparecy_report_
cy2016_5_2_17.pdf.
52 Kent Roach, “Review and Oversight of Intelligence in Canada: Expanding Accountabil-
ity Gaps,” in Goldman and Rascoff, Global Intelligence Oversight, 181.
53 X (Re), 2014 FCA 249 (CanLII), http://canlii.ca/t/gf63j; X (Re), [2017] 2 FCR 396, 2016
FC 1105 (CanLII), http://canlii.ca/t/gw01x.
54 Roach, “Review and Oversight of Intelligence in Canada,” 187–88.
55 Anne Dagenais Guertin, “Our Analysis of C-22: An Inadequate and Worrisome Bill,”
International Civil Liberties Monitoring Group, 30 June 2016, http://iclmg.ca/our-analysis
-of-c-22-an-inadequate-and-worrisome-bill/; Scott Newark, “Ensuring Independence
for the Parliamentary National Security Committee: A Review of Bill C-22” (Macdon-
ald-Laurier Institute publication, November 2016), http://www.macdonaldlaurier.ca/
files/pdf/MLICommentaryNewark11-16-webV2.pdf.
56 Goldman and Rascoff, “The New Intelligence Oversight,” xxix.
57 Zachary K. Goldman, “The Emergence of Intelligence Governance,” in Goldman and
Rascoff, Global Intelligence Oversight, 219.
58 Daphna Renan, “The FISC’s Stealth Administrative Laws,” in Goldman and Rascoff,
Global Intelligence Oversight, 135. Though beyond the scope of this argument, such pro-
ceedings could also include special advocates as much as possible to avoid ex parte hearings
that might lead to legal interpretations that unduly impact the civil liberties of those
affected by CSE’s surveillance operations.
59 Bill C-59, pt 1, s 3(a), as well as pt 3, ss 13–21.
60 Ibid., pt 3, s 35(1).
61 Patrick F. Walsh and Seumas Miller, “Rethinking the ‘Five Eyes’ Security Intelligence
Collection Policies and Practice Post-Snowden,” Intelligence and National Security 31, 3
(2016): 345–68, 365–66.
62 Goldman, “The Emergence of Intelligence Governance,” 220.
63 Roberts, “Transparency in the Security Sector,” 320.
14
Metadata – Both Shallow and Deep
The Fraught Key to Big Data Mass State Surveillance
Andrew Clement, Jillian Harkness, and George Raine

We kill people based on metadata.

– General Michael Hayden, former head of
the National Security Agency

Until recently, metadata – literally “data about data” – was unfamiliar

outside the information fields, where it is a term of art used in describing a wide
variety of information objects to enable cataloguing and retrieval. However,
since Edward Snowden’s 2013 whistle-blowing exposures of the extraordinary
scope of secret mass state surveillance and the key role that metadata plays in
such surveillance, the term has now entered public discourse. As national secur-
ity agencies have adopted new techniques for intercepting electronic communi-
cations, notably “upstream” programs that tap into Internet data flows at key
routing centres and major fibre-optic cables, they are turning to metadata
analysis techniques to process the extraordinarily expanding volumes of inter-
cept data (see Chapter 7 for a more in-depth discussion). We are also learning
that with the application of big data collection and analysis techniques, metadata
can be much more revealing of an individual’s behaviour, movements, beliefs,
affiliations, and social connections than previously understood.
In the hands of state actors, metadata can provide the basis for highly intrusive
intervention into people’s lives. While the coming into prominence of metadata
draws attention to a relatively new and potent mode of mass surveillance, distinct
from the popular image of individually targeted eavesdropping or surreptitious
reading of communications, it remains an ambiguous and contested term. In
particular, officials who lead state security agencies often deploy the term pub-
licly in ways evidently designed to reassure audiences skeptical of their behind-
the-scenes surveillance activities but that, given its ambiguity, also raise serious
questions about their actual practices and trustworthiness. This is significant
because metadata enjoys lower legal and constitutional protection than com-
munication content. Furthermore, recurring discrepancies between what secur-
ity and law enforcement agencies say about metadata and what they actually
do with it call into question the adequacy of democratic governance of these
powerful and often necessarily secretive arms of the state.
254 Andrew Clement, Jillian Harkness, and George Raine

This chapter seeks to clarify the multiple meanings of metadata as defined

and operationalized in various settings. More specifically, it seeks to determine
whether the practices of the Five Eyes signals intelligence agencies – notably
the US National Security Agency (NSA), the British Government Communica-
tions Headquarters (GCHQ), and Canada’s Communications Security Estab-
lishment (CSE) – in relation to metadata conform to their public statements as
well as the conventional and legal understandings of the term. To set the stage,
we summarize the various definitions of metadata (and its equivalents) as they
have appeared in academic, professional, legal, and popular discourses as well
as in the emerging field of big data analytics.
The core of the chapter draws on our study of the more than 500 secret docu-
ments of the Five Eyes alliance published by the media based on Edward Snow-
den’s leak, as found in the Snowden Digital Surveillance Archive.1 Using the
archive’s various search and index features, we selected and analyzed documents
for what they reveal about how the security agencies actually generate metadata
from their global communication interception apparatus and subsequently
organize, access, and use it in their intelligence operations.

Conventional Meanings of Metadata

Ever since the Snowden revelations prompted journalists to explain the technical
aspects behind mass surveillance, an understanding of the term “metadata” has,
for the average citizen, been largely shaped by the news media. These definitions
are varied; often they are simplified accounts echoing those put forth by govern-
ment offices, such as a commonly used definition from the Canadian
Broadcasting Corporation (CBC), where metadata is defined as “information
associated with communication that is used to identify, describe or route infor-
mation,” and more specifically as possibly including “personal information,
including phone numbers or email addresses, but not the content of emails or
recordings of phone calls.”2 More elaborate and insightful definitions have also
sometimes been offered by the media; in 2013, The Guardian explained that
“metadata provides a record of almost anything a user does online, from brows-
ing history – such as map searches and websites visited – to account details,
email activity, and even some account passwords. This can be used to build a
detailed picture of an individual’s life.”3 The Guardian example helps the public
to understand the ramifications of metadata collection, whereas in the more
common definition, as shown in the example from the CBC, these details remain
vague.
Within legal frameworks, at least in the United Kingdom, the United States,
and Canada, the term “metadata” is rarely if ever used. The concept of metadata
within the law developed historically out of the ways in which the law has
 Metadata – Both Shallow and Deep 255

approached communications technology, insofar as this technology facilitates

the production of personal information that may or may not be subject to
privacy laws. How metadata is interpreted in Canadian law is still being worked
out by regulators and the judiciary.4 Referring to private communications, Part
VI of the Criminal Code states that

private communication means any oral communication, or any telecommuni-

cation, that is made by an originator who is in Canada or is intended by the
originator to be received by a person who is in Canada and that is made under
circumstances in which it is reasonable for the originator to expect that it will
not be intercepted by any person other than the person intended by the origina-
tor to receive it.5

Whether or not metadata can be defined as private communications has been

discussed in Canadian courts. Authoritative voices such as Craig Forcese and
the Office of the Privacy Commissioner cite court cases as well as the Supreme
Court interpretation of the Criminal Code, which protects “any derivative of
that communication that would convey its substance or meaning.”6 As a deriva-
tive of a communication, metadata has been shown to provide significant details
about the meaning of a communication7 and “may permit the drawing of infer-
ences about an individual’s conduct or activities.”8 This definition focuses, as in
the media definitions, on the surface or contextual information about a com-
munication but acknowledges the potential sensitivity of this information when
analyzed and linked with other available data. The Office of the Privacy Com-
missioner further asserts that “metadata can reveal medical conditions, religious
beliefs, sexual orientation and many other elements of personal information.”9
Commenting on how the debate has played out in Canadian courts, Forcese
reads the 2014 Supreme Court of Canada decision in R v Spencer as the “clear
authority” that some forms of metadata, in this case “the name, address, and
telephone number of a customer associated with an IP address,” can be used to
reveal details about private lives and should be protected as personal
information.10
Like those in Canada, laws in the United Kingdom and the United States do
not use the term “metadata,” but focus more specifically on the technical aspects
of the type of communications information. In the United Kingdom, the Regu-
lation of Investigatory Powers Act (RIPA) defines “communications data” in part
as including “(a) any traffic data comprised in or attached to a communication
(whether by the sender or otherwise) for the purposes of any postal service or
telecommunication system by means of which it is being or may be transmitted ...
(b) any information which includes none of the contents of a communication
256 Andrew Clement, Jillian Harkness, and George Raine

(apart from any information falling within paragraph (a)).”11 In the United States,
one example of metadata, “Call Detail Records,” is legally defined as “session-
identifying information (including an originating or terminating telephone
number, an International Mobile Subscriber Identity number, or an International
Mobile Station Equipment Identity number), a telephone calling card number,
or the time or duration of a call” and “does not include – (1) the contents ... of
any communication; (2) the name, address, or financial information of a sub-
scriber or customer; or (3) cell site location or global positioning system infor-
mation” (emphasis added).12
The differences in these varying approaches reveal how metadata can imply
very different things across varying communities of practice. In her Introduction
to Metadata 3.0, leading archival scholar Anne J. Gilliland explicates the “widely
used but frequently underspecified term” within the framework of the archival
discipline.13 She notes that the term originated with data management and today
in practice metadata is generally “the sum total of what one can say about any
information object at any level of aggregation.” An information object can vary
from a film or book to email or phone call; therefore “metadata” in this definition
suggests anything one could say about these items, from a title to any salient
feature of the contents. For archivists and information managers, metadata
reflects an information object’s content, context, and structure, and enables
preservation as well as “intellectual and physical access.”14
Despite at times recognizing that communications metadata may reveal a
significant amount of personal information, media and legal definitions of
metadata tend to limit their focus to the specific types of information that can
be read from the “surface” of the information object without delving into the
object’s content. By contrast, the archival definition of metadata, as put forth
by Gilliland, acknowledges that varying levels of aggregation and detail, as well
as relationships between information objects and systems, may impact how one
defines metadata as opposed to data, or context as opposed to content.15 Gilliland
notes that these “distinctions ... can often be very fluid and may depend on how
one wishes to use a certain information object.”16

Role of Metadata in Big Data Analytics

Questions of how metadata is used, and by whom, are often missing from the
simplified accounts of metadata collection. Critics of these popular and gov-
ernmental definitions note the dismissal of metadata’s importance to big data
analysis,17 which takes advantage of the ease with which this kind of structured
data can be processed automatically to link very large collections of data ele-
ments and find patterns in order to create meaning and draw conclusions.18
Proponents of big data analysis claim that metadata enables efficient “distill[ing
 Metadata – Both Shallow and Deep 257

of] terabytes of low-value data ... to ... a single bit of high-value data.”19 Through
the development of mobile communications technology, an environment has
emerged in which ordinary users, often without realizing it, produce large
amounts of metadata on a daily basis.20 Access to this mass of personal data,
when analyzed through big data analytical techniques and software, allows for
broad and deep access to personal information. Arguably, this access has been
downplayed through the conventional meanings of metadata summarized above,
to the benefit of both corporate business practices and surveillance agencies.21

Metadata from a Five Eyes Perspective

The Five Eyes, frequently abbreviated “FVEY” or “5VEY” in their internal docu-
ments, is an intelligence alliance among the United States, United Kingdom,
Canada, Australia, and New Zealand dating back to the Second World War.
Each of these countries maintains several intelligence agencies that participate
in the alliance, but it is those focused on signals intelligence (SIGINT) that
concern us here; they are, respectively, the National Security Agency, the
Government Communications Headquarters, the Communications Security
Establishment,22 the Australian Signals Directorate (ASD), and the Government
Communications Security Bureau (GCSB). During the Cold War, FVEY
developed a globe-spanning signal interception capability known as
ECHELON.23 Initially targeted at the Soviet Union, whistle-blowers24 and jour-
nalists25 have recently revealed that this alliance, led by the NSA, has greatly
expanded this network into a comprehensive surveillance apparatus of extra-
ordinary scope and domestic penetration, albeit of highly questionable efficacy
in its principal stated mission to aid in the “War on Terror.”26
The classified internal documents that Edward Snowden leaked to journalists
in June 2013 and subsequently published in major news media27 have offered
an unprecedented glimpse into these highly secretive surveillance agencies. The
Snowden trove reveals for the first time in fascinating detail their inner work-
ings, identifying hundreds of individual surveillance programs.28 But the view
offered is at best a sliver of the full picture. So far, well under 1,000 of the more
than 50,000 documents that Snowden leaked have been made public.29 Some
vast surveillance programs are mentioned only in extremely abbreviated sum-
maries. Many of the documents have been released to the public only in frag-
mented form, with heavy redactions by both government officials and newspaper
editors. They are full of obscure, cryptic acronyms, code words, and arcane
technical details that call for security expertise and organizational experience
to properly decipher.
Nevertheless, while keeping these limitations in mind, there is already such
an abundance of material in the Snowden trove dealing with metadata that we
258 Andrew Clement, Jillian Harkness, and George Raine

have a good basis for painting a reliable, if preliminary, picture of how these
agencies discuss and operationalize our topic at hand.
For our study of metadata within the FVEY, we relied extensively on the
Snowden Digital Surveillance Archive, a publicly accessible finding aid to the
full corpus of published Snowden documents and related media articles that
we designed and built, and that is now hosted by the Canadian Journalists for
Free Expression.30 From working with the documents in building the archive,
we developed strong suspicions that internally the FVEY agencies take a much
more expansive view of metadata than suggested by their public statements and
reinforced in the popular media definitions discussed above. In switching to a
research role, we sought to test our suspicions while being open to possible
disconfirming evidence. We initially made use of the archive’s full-text search,
indexing, and document description features to locate documents and stories
relatively dense in details about metadata, and then pursued thematic linkages
between documents, such as by surveillance program, to amplify the contexts
in aid of interpretation.
Metadata is evidently an important topic within the FVEY. A search in the
archive on “metadata” produced 1,644 word hits. Fourteen documents contained
“metadata” in their title, which we examined first. The surveillance programs,
legal justifications, and internal policies mentioned therein informed further
archival searches. The domain knowledge we had gained from arranging and
describing the Snowden documents also greatly aided our initial searches in
identifying fertile points for research as well as in understanding the documents
we selected.
While we expected that exploring such a conceptually vague and varied
phenomenon as metadata would yield a mix of results, we were struck by the
breadth and heterogeneity of metadata produced by FVEY agencies. For
example, disparate GCHQ surveillance programs harvest hotel room bookings
(ROYAL CONCIERGE), social media activity (STELLARWIND), and text
message geolocation data (in partnership with the NSA under DISHFIRE).
Each different program generates different types of metadata, making classifica-
tion of the surveillance agency’s handling of metadata difficult. To facilitate
comparison of the various agency interpretations of metadata with each other,
with their public statements, and with the various legal definitions in their
respective jurisdictions, we looked at the three most relevant agencies in turn –
NSA, GCHQ, and CSE.31 We also focused on those surveillance programs in
which metadata plays a particularly prominent role, notably XKEYSCORE,
which provides a front-end interface to many signals intelligence databases
around the globe and is accessed by all members of the alliance as well as by
selected third-party agencies.
 Metadata – Both Shallow and Deep 259

Metadata in the NSA

Several documents in the Snowden trove across a span of several years use
identical language that the NSA has apparently adopted as its standardized
definition of metadata:

Communications metadata refers to structured “data about data”: it includes all

information associated with, but not including content, and includes any data
used by a network, service or application to facilitate routing or handling of a
communication or to render content in the intended format; it includes, but is
not limited to, dialing, routing, addressing, or signaling information and data in
support of various network management activities (e.g. billing, authentication
or tracking of communicants). [Emphasis added]32

The key phrase here is “includes all information associated with, but not includ-
ing content.” But what does this actually mean? A colloquial interpretation,
based on the caveat “but not including content,” would confine communications
metadata to the surface features, such as the oft-referenced “outside of the
envelope” data. In which case, this could be reworded as “includes all informa-
tion associated with the message, but excluding anything based on message
content.” However, another plausible but very different interpretation would be
that metadata “includes all information that can be derived from a message and
its content, but excluding the content itself.” This version would be consistent
with metadata in archival theory and practice as described above. It would also
be consistent with the NSA’s often expansive, and secret, interpretation of its
legal mandate, thereby opening the door for the agency to justify unfettered
algorithmic analysis of communications content as metadata extraction. We
find strong support for this latter interpretation when we examine the NSA’s
most prominent analysis engine, XKEYSCORE. Compared with many other
tools and surveillance programs mentioned in the Snowden documents, import-
ant aspects of XKEYSCORE are extensively described, enabling a relatively
comprehensive understanding of its capabilities, scope, and use.
XKEYSCORE is one of the NSA’s most powerful tools and is often in demand
among its trusted “second party” (i.e., other members of the Five Eyes) and
“third party” partners.33 Access to the tool has been shared with the GCHQ,
ASD, CSE, GCSB, Bundesnachrichtendienst (Germany), and National Defence
Radio Establishment (FRA) (Sweden). Described in the Intercept as “the NSA’s
Google,”34 this tool gives analysts unprecedented access to communications
metadata largely harvested by the NSA from fibre-optic cables and cached in
over 700 servers at 150 storage sites scattered throughout the world. An unof-
ficial user’s guide to XKEYSCORE developed by Booz Allen Hamilton gives a
260 Andrew Clement, Jillian Harkness, and George Raine

technical description of the operation of the tool aimed at surveillance analysts.

It includes detailed illustrations of its user interface, as well as specifications of
how different metadata fields can be used to query the NSA’s vast archives of
intercepted communications. These fields are very valuable in identifying the
NSA’s operational interpretation of metadata.
Conventional metadata fields are well represented – unsurprisingly, analysts
have the ability to query communications by IP address, phone number, email
account, and so on. However, metadata query capabilities extend far beyond
the obvious to include accessing anything contained in email messages, web
searches, text chats, and file attachments – in short, the full range of communi-
cations content.
The user’s guide describes various algorithmic metadata “extractors,” includ-
ing for phone number and email address. These tools appear to scan digital
communications traffic for any mention of email addresses and phone num-
bers, not just in the routing data but within the message bodies, and retrieve
these as distinct metadata fields: “The phone number extractor query looks
through the content of an email for phone numbers ... XKEYSCORE may be
your only hope at finding an email address for a target where you only have
their phone number as lead information” (emphasis added).35 Another docu-
ment similarly describes the email address extractor query: “The query
searches within bodies of emails, webpages and documents for ... (you guessed
it) ... Email Addresses.”36
The NSA and its intelligence partners, chief among them the GCHQ, use
these content-derived metadata extraction capabilities not only for counter-
terrorism but also in other pursuits, including diplomatic advantage. Targeting
the offices of at least thirty-five world leaders, FVEY agencies intercepted both
phone and email communications from officials and staffers,37 extracting phone
numbers and email addresses to be used in identifying further targets.38 Beyond
phone numbers and email addresses, other XKEYSCORE documents make
clear its ability to query the content of email messages and their attachments
for arbitrary text strings: “Allow queries like ... show me all documents that
reference Osama Bin Laden.”39
Interestingly, the XKEYSCORE documents refer to this kind of database
querying via metadata as “contextual” search. Arguably “context” is an even
more ambiguous term than “metadata,” providing for wide interpretative flex-
ibility. The conflation of metadata with context is illustrated in excerpts from
another XKEYSCORE training document, Guide to Using Contexts in XKS
Fingerprints (see Figure 14.1).
These examples provide clear evidence that, in practice, the NSA subscribes
to a view that anything it can derive algorithmically from communications
 Metadata – Both Shallow and Deep 261

Figure 14.1  “Context” as metadata category in XKEYSCORE. A one-page excerpt from the
classified secret Five Eyes document titled “Guide to Using Contexts in XKS Fingerprints.” It
shows that Communications Content is considered a form of Context of Type “Scan.” | Source:
Snowden Digital Surveillance Archive, https://is.gd/n7brJJ.

traffic can be regarded as metadata, including information extracted from the

communications content itself. Short of a human analyst actually reading a
message or listening in on a phone call, any distinction between content and
metadata has been erased. As XKEYSCORE is shared among all Five Eyes
262 Andrew Clement, Jillian Harkness, and George Raine

SIGINT partners, notwithstanding any jurisdictional differences between them,

they all have in effect adopted this expansive view of metadata.

Metadata in the GCHQ

One particularly useful document in the Snowden archive, “Content or
Metadata? Categorisation of Aspects of Intercepted Communications – GCHQ
Policy Guidance,”40 delineates the GCHQ’s official understanding of metadata.
This document clearly specifies whether certain elements of an Internet com-
munication (such as an email) are to be classified as content or metadata, and
which particular “class” under the Regulation of Investigatory Powers Act the
communication falls under.
What is notable about this schema is how different it is from the NSA’s.
The contents of attachments to emails are labelled “content,” in marked
contrast to the NSA, which, as we saw above in XKEYSCORE, operational-
izes this element as a class of communications metadata.41 Interestingly, the
GCHQ also identifies a class of “content derived data,” including analysis
of the language of the communication itself, presumably derived from
analysis of the content of the message.42 The document seems to imply that
this class of data is considered conceptually distinct from both metadata
and content.
However, the GCHQ, along with the other surveillance agencies in the Five
Eyes, has access to XKEYSCORE. It is unclear how it utilizes this tool in
accordance with its specific legal and policy requirements, as XKEYSCORE
has been described as lacking any meaningful oversight and accountability
features.43

Metadata in CSE
In Canada, CSE collects or, more precisely, generates metadata as part of
its mandate, through the National Defence Act, “to acquire and use infor-
mation from the global information infrastructure for the purpose of
providing foreign intelligence.”44 The act broadly defines “global informa-
tion structure” as “electromagnetic emissions, communications systems,
information technology systems and networks, and any data or technical
information carried on, contained in or relating to those emissions, systems
or networks.” This mandate is limited by “measures to protect the privacy
of Canadians in the use and retention of intercepted information,” as out-
lined in the Criminal Code.45
Unlike other intelligence agencies, the Canadian CSE displays its public
definition of metadata on its website:
 Metadata – Both Shallow and Deep 263

Metadata is the context, but not the content of a communication. It is informa-

tion used by computer systems to identify, describe, manage or route communi-
cations across networks. For example, metadata can refer to an internet protocol
address or a phone number or the time of a transmission or the location of a
device ... While metadata reveals a certain amount of information about devices,
users and transmissions, it is contextual and does not expose the content of
emails, phone calls or text messages.46

In 2016, the director of CSE reiterated this distinction between content and
metadata-as-context in responding to a Toronto Star editorial calling for more
oversight of the agency – “Context, not content.”47 As we saw above, however,
context in the eyes of these agencies is much different from how we convention-
ally understand the term and is deeply tied to communication content. As in
the case of the GCHQ, it is difficult to square this definition with the agency’s
continued use of XKEYSCORE.
One CSE surveillance experiment in particular aptly reveals the power of
metadata in surveillance activities. The leaked document titled “IP Profiling
Analytics & Mission Impacts” describes a trial program where CSE profiled the
users of Wi-Fi networks at an international airport located on Canadian soil.48
The CBC incorrectly reported this as involving the interception of Wi-Fi signals
in the airport, but as analyzed in greater depth in Chapter 7, the actual practices
are far more disturbing. Especially revealing is the statement by John Forster,
then chief of CSE, who told the Senate committee investigating the apparent
violation of Canadian law that the experiment involved no actual monitoring
at an airport, but simply “a snapshot of historic metadata collected from the
global internet ... [as] part of our normal global collection.”49
Through comprehensive capture, analysis, and storage of Internet communi-
cation, CSE spotted visitors to the airport based on the IP address of the airport’s
public Wi-Fi service. Analysts were then able to track individuals to other
locations with identifiable IP addresses, both forward and backward in time,
based on the user IDs extracted from message content. This case illustrates not
only CSE’s expansive interpretation of metadata but also the remarkably broad
scope and fine detail of its domestic surveillance capabilities.

“Deep” versus “Shallow” Metadata

This review of the meanings of metadata in varied settings shows that notwith-
standing a general recognition that metadata is ambiguous and difficult to
distinguish from content, we can discern two distinct sets of meanings to the
term. In the popular and legal discourses as well as in the official public state-
ments by security agencies, we observe what can be termed narrow, conventional,
264 Andrew Clement, Jillian Harkness, and George Raine

“shallow” metadata,50 characterized by the various forms of data about a com-

munication act or information object that can be read externally, without
examining the actual content. In contrast, “deep” metadata goes significantly
beyond conventional metadata to include data that can be derived algorithmic-
ally from the content.
It is this “deep” metadata that we find defined in the archives field and
operationalized in the big data surveillance activities of Five Eyes security
agencies. An immediate disturbing conclusion from this finding is that under
the rubric of “metadata,” the mass surveillance operations of these agencies
are potentially even more revealing of people’s lives than previously under-
stood. A more hopeful, mitigating conclusion is that attempts to maintain a
sharp distinction between communication content and metadata, used as the
basis for the weaker protection of the latter, are now more clearly untenable.
In the context of Internet communication, current metadata practices that
combine both shallow and deep forms are at odds with the Criminal Code,
which in the interpretation of the Supreme Court of Canada protects “any
derivative of that communication that would convey its substance or mean-
ing.”51 While the judiciary is showing strong signs of updating its understanding
of metadata in light of contemporary practices, journalism, privacy regulation,
and the law need to catch up.

Implications
These conclusions hold implications for the various actors interested in mass
surveillance.

For Surveillance Researchers

For researchers seeking to understand the nature and implications of contem-
porary big data surveillance practices across the private and public sectors, our
findings highlight the need to avoid restricting their conceptualization of
metadata to the conventional, “shallow” definition of metadata and to broaden
it to include the “deep” form. Their protestations to the contrary, state security
agencies widely practise metadata analysis in this deeper sense based on algo-
rithmic processing of communication message bodies. Very likely commercial
enterprises with access to large volumes of personal information do likewise.
It is clear that this form of data mining involves large-scale, systematic algo-
rithmic analysis of communication content intended to draw out potentially
anything that may be of interest to the surveillance analyst. In short, metadata
is a principal means for analyzing and accessing “content,” and hence is effect-
ively inseparable from it.
 Metadata – Both Shallow and Deep 265

For Security Intelligence Agencies

State security intelligence agencies such as CSE, NSA, and GCHQ are legitim-
ately in the business of maintaining secrecy and deceiving perceived opponents,
but in healthy democracies they are also ultimately accountable to the citizens
they are mandated to protect. While it may run against institutional culture,
maintaining the necessary public confidence and trust requires significant
transparency, honesty, and demonstrable compliance with legal and constitu-
tional norms. Especially at a time when there is growing and well-founded
skepticism that these agencies are neither in compliance with the law nor being
effective in their missions, for their official statements to consistently rely on
a narrow interpretation of metadata at odds with their actual practices is hardly
reassuring. Misleading the public about such an important term as “metadata”
only heightens concerns. Security agencies coming clean about what they really
do with our data may initially provoke adverse reaction, but in the long run is
far better than fuelling the vicious cycle of deception and public skepticism.

For Those Calling for State Surveillance Accountability and Reform

For civil liberties and democratic governance advocates, parliamentarians,
privacy regulators, journalists, and citizens working to bring greater account-
ability and reform to state surveillance agencies, our findings point to several
recommendations. They should:
• continue to revise their working definitions and metaphors of metadata to
include “deep” metadata52
• treat statements about metadata, especially by government officials who repeat
the conventional definition of metadata, with skepticism and avoid reiterating
them
• instead adopt and publicize the more comprehensive meaning of metadata, its
practical indistinguishability from message content, and the greater risks that
deep metadata can pose to personal privacy
• call to account state security and law enforcement agencies that make mislead-
ing public statements and go beyond their legal authorizations
• press for metadata to receive privacy protection equivalent to content (e.g., in
the Criminal Code).

These suggestions contribute to the wider effort to achieve stronger transparency

and accountability of organizations with access to large volumes of personal infor-
mation, especially state security and law enforcement agencies. In particular, they
can help ensure that state security and law enforcement agencies are more transpar-
ent and accountable, and operate within the norms of democratic governance.
 266 Andrew Clement, Jillian Harkness, and George Raine

A Methodological Note on Researching with the Snowden Archive

Although we began constructing the Snowden Digital Surveillance Archive in
2014, this report reflects our first attempt to make use of the archive ourselves
for research purposes. We are pleased that it provided us with an opportunity
to examine in a comprehensive and seamless way all the documents Snowden
released and subsequently published. We did not face a priori divisions based
on agency, national jurisdiction, or publishing source. This enabled us to draw
a more holistic picture of the role of metadata in mass surveillance at the inter-
national level. We found document descriptions and links to media reports
developed during the archival process particularly helpful in contextualizing
documents that are often heavily redacted and excerpted. We also became
acutely aware of some of the limitations of the current implementation, notably
the lack of a controlled vocabulary and technical limitations of the Greenstone
digital library software the archive is built on. These were more of a nuisance
than a major obstacle to our research, however. Our experience studying meta-
data using the Snowden Digital Surveillance Archive suggests that other
researchers interested in big data surveillance, particularly by the Five Eyes
agencies, may find the archive a valuable research tool.

Notes
1 Canadian Journalists for Free Expression, Snowden Digital Surveillance Archive (SDSA),
https://snowdenarchive.cjfe.org.
2 Ashley Burke, “‘Difficult to Determine’ Scope of Privacy Breach in Five Eyes Data Shar-
ing,” CBC News, 23 February 2016, http://www.cbc.ca/news/politics/cse-metadata-five
-eyes-sharing-1.3459717. A nearly identical definition was used again in the following
article: Alison Crawford, “Canada’s Electronic Spy Agency to Get New Rules for Shar-
ing Data with Allies,” CBC News, 29 August 2017, http://www.cbc.ca/news/politics/
sajjan-cse-data-sharing-five-eyes-1.4265583.
3 James Ball, “NSA Stores Metadata of Millions of Web Users for Up to a Year, Secret Files
Show,” Guardian, 30 September 2013, http://www.theguardian.com/world/2013/sep/30/
nsa-americans-metadata-year-documents.
4 Office of the Privacy Commissioner of Canada (OPCC), Metadata and Privacy: A Tech-
nical and Legal Overview (Gatineau, QC: Office of the Privacy Commissioner of Canada,
2014), 9.
5 Criminal Code, RSC 1985, c C-46, pt 6, s 183, https://laws-lois.justice.gc.ca/eng/
acts/C-46/page-41.html#h-118715.
6 As quoted in Craig Forcese, “Laws, Logarithms, Liberties: Legal Issues Arising from
CSE’s Metadata Collection Initiatives,” in Law, Privacy, and Surveillance in the Post-
Snowden Era, edited by Michael Geist (Ottawa: University of Ottawa Press, 2015), 137.
7 Ibid., 137, 148.
8 OPCC, Metadata and Privacy, 10.
9 OPCC, “Backgrounder: Privacy and Canada’s National Security Framework,” 6 December
2016, https://www.priv.gc.ca/en/opc-news/news-and-announcements/2016/bg_161206/.
10 OPCC, Metadata and Privacy, 148.
11 Regulation of Investigatory Powers Act 2000, 2000 c 23, s 21.4.
 Metadata – Both Shallow and Deep 267

12 USA Freedom Act, Pub L No 114-23, s 107 (“Definitions”), 129 Stat 268 (2015).
13 Anne J. Gilliland, “Setting the Stage,” in Introduction to Metadata, edited by Murtha Baca
(Los Angeles: Getty Research Institute, 2008), 1.
14 Ibid., 2.
15 Ibid.,14.
16 Ibid., 14–15.
17 David Lyon, “Surveillance, Snowden, and Big Data: Capacities, Consequences, Critique,”
Big Data and Society 1, 2 (2014): 3, 10.
18 Seref Sagiroglu and Duygu Sinanc, “Big Data: A Review,” in 2013 International Confer-
ence on Collaboration Technologies and Systems (CTS) (San Diego: Institute of Electrical
and Electronics Engineers, 2013), 43; Danyel Fisher, Rob DeLine, Mary Czerwinski, and
Steven Drucker, “Interactions with Big Data Analytics,” Interactions 19, 3 (2012): 53.
19 Fisher et al., “Interactions with Big Data Analytics,” 50.
20 Lyon, “Surveillance, Snowden,” 3; Gilliland, “Setting the Stage,” 8.
21 John Laprise, “Exploring PRISMS Spectrum: Privacy in the Information Age,” in The
Turn to Infrastructure in Internet Governance, edited by Francesca Musiani, Derrick L.
Cogburn, Laura DeNardis, and Nanette S. Levinson (New York: Palgrave Macmillan,
2016), 208, 214; Lyon, “Surveillance, Snowden,” 10.
22 Also formerly referred to as the Communications Security Establishment Canada
(CSEC). It is this now unofficial name and acronym that appears most frequently in the
Snowden documents.
23 In the mid-1970s, “the very existence of GCHQ and the [worldwide US/UK] Sigint
network were then closely guarded secrets.” Duncan Campbell, “GCHQ and Me: My
Life Unmasking British Eavesdroppers,” Intercept, 3 August 2015, https://theintercept.
com/2015/08/03/life-unmasking-british-eavesdroppers/.
24 Notably Mark Klein, William Binney, Thomas Drake, and Edward Snowden.
25 Notably James Bamford, James Risen, Eric Lichtblau, Glenn Greenwald, Laura Poitras,
Barton Gellman, and Ryan Gallagher,
26 Zach Whittaker, “NSA Is So Overwhelmed with Data, It’s No Longer Effective, Says
Whistleblower,” ZDNet, 27 April 2016, http://www.zdnet.com/article/nsa-whistleblower
-overwhelmed-with-data-ineffective/.
27 Notably the Guardian, Washington Post, Der Speigel, Intercept, New York Times. See
SDSA, https://is.gd/ze5urh.
28 See SDSA, http://bit.ly/SnowdenArchive-Surveillance_Programs.
29 As of mid-2018, Snowden Doc Search reported a total of 2,176 documents in its searchable
database, of which 1,571 were individual articles that appeared in SIDToday, the internal news-
letter for the NSA’s Signals Intelligence Directorate, https://search.edwardsnowden.com/.
30 SDSA, https://snowdenarchive.cjfe.org.
31 We exclude Australia’s DSO and New Zealand’s GCSB from our treatment here as there
are relatively few Snowden documents that relate to these partners, nor does metadata
appear prominently among them.
32 This definition appears in several different documents found in the Snowden Digital
Surveillance Archive, e.g., National Security Agency (NSA), “Sharing Communications
Metadata across the U.S. Intelligence Community – ICREACH” (slide deck, 15 March
2007), SDSA, https://is.gd/9j9vRA; and “Memorandum for the Director of National
Intelligence: Sharing Communications Metadata across the Intelligence Community –
Decision Memorandum,” SDSA, https://is.gd/N1z0qR.
33 “XKEYSCORE” (slide deck, 25 February 2008), SDSA, https://is.gd/RLB6U6.
34 Morgan Marquis-Boire, Glenn Greenwald, and Micah Lee, “XKEYSCORE: NSA’s
Google for the World’s Private Communications,” Intercept, 1 July 2015, https://theintercept.
com/2015/07/01/nsas-google-worlds-private-communications/.
 268 Andrew Clement, Jillian Harkness, and George Raine

35 Booz Allen Hamilton, “The Unofficial XKEYSCORE User Guide,” 10, SDSA, https://
is.gd/QX8VrU.
36 “Email Address vs User Activity” (slide deck, 24 June 2009), slide 2, SDSA, https://
snowdenarchive.cjfe.org/greenstone/collect/snowden1/index/assoc/HASH0164/
d967fedd.dir/doc.pdf.
37 James Ball, “NSA Monitored Calls of 35 World Leaders after US Official Handed Over
Contacts,” Guardian, 24 October 2013, http://www.theguardian.com/world/2013/
oct/24/nsa-surveillance-world-leaders-calls.
38 “Intelligently Filtering Your Data: Brazil and Mexico Case Studies,” SDSA, https://is.gd/
ljFRcC; “3G Impact and Update” (slide deck, November 2009), SDSA, https://is.gd/
gfOjhZ.
39 “XKEYSCORE” (slide deck, 25 February 2008), slide 26, SDSA, https://is.gd/RLB6U6.
40 “Content or Metadata?” SDSA, https://snowdenarchive.cjfe.org/greenstone/collect/snowden1/
index/assoc/HASHd8b5.dir/doc.pdf. We could not find similar technical documents for the
other two major surveillance agencies.
41 Ibid.
42 Ibid.
43 Kari Rea, “Glenn Greenwald: Low-Level NSA Analysts Have ‘Powerful and Invasive’
Search Tool,” ABC News, 28 July 2013, http://abcnews.go.com/blogs/politics/2013/07/
glenn-greenwald-low-level-nsa-analysts-have-powerful-and-invasive-search-tool/.
44 National Defence Act, s 273.64.1.
45 Ibid., s 273.64.2.
46 Communications Security Establishment (CSE), “Metadata and Our Mandate,” June
2017, https://www.cse-cst.gc.ca/en/inside-interieur/metadata-metadonnees.
47 Greta Bossenmaier, letter to the editor, The Star, 3 March 2016, https://www.thestar.com/
opinion/letters_to_the_editors/2016/03/03/metadata-is-crucial-cse-insists.html.
48 Greg Weston, “CSEC Used Airport Wi-Fi to Track Canadian Travellers: Edward Snowden
Documents,” CBC News, 31 January 2014, http://www.cbc.ca/news/politics/csec-used
-airport-wi-fi-to-track-canadian-travellers-edward-snowden-documents-1.2517881.
49 Laura Payton, “Spy Agencies, Prime Minister’s Adviser Defend Wi-Fi Data Collec-
tion,” CBC News, 3 February 2014, http://www.cbc.ca/news/politics/spy-agencies-prime
-minister-s-adviser-defend-wi-fi-data-collection-1.2521166.
50 This terminology of “shallow” versus “deep” metadata is inspired in part by the similar
distinction used in the XKEYSCORE document of 25 February 2008. It also echoes the
“deep packet inspection” techniques employed by FVEY agencies in generating metadata
from intercepted communication traffic. See “XKEYSCORE” (slide deck, 25 February 2008),
slides 9 and 10, SDSA, https://snowdenarchive.cjfe.org/greenstone/collect/snowden1/
index/assoc/HASH56fe.dir/doc.pdf.
51 As quoted in Forcese, “Laws, Logarithms, Liberties,” 137.
52 For example, the OPCC’s 2014 Metadata and Privacy statement could be expanded to
make explicit the forms of deep metadata we highlight above.
Afterword
Holly Porteous

This book raises serious questions about preserving civil liberties and national
security in a big data era. In reflecting on these questions, it became apparent
to me that lingering gaps in our knowledge are forcing us to rely on assumptions
that may prove incorrect should our access to government information and
research output increase over the next few years. What I am proposing here,
building on the excellent contributions of this book’s authors, is to identify areas
requiring further research, knowledge gaps, and where we need to strengthen
our collective analysis.
First, delimiting our subject matter more precisely and with the necessary
nuances appears urgent. Reading this book, I was struck by its enormous breadth.
Among other things, it discusses the evolution of Canadian SIGINT collection,
the complexity of Canada’s current national security legal framework, the use
of (or failure to use) big data analytics by Canadian intelligence and law enforce-
ment agencies, mobilization of the public against proposed national security
legislation, and challenges in achieving informed consent to access personal
information.
Going forward, I believe there will be value in selecting elements of this broad
discourse for closer scrutiny. For example, do we wish to examine in greater
detail how and for what purposes the Canadian security and intelligence com-
munity and law enforcement exploit big data? If so, will our goal be to stop these
activities entirely or to identify and recommend measures to mitigate the pot-
entially negative consequences to individuals? Given the enthusiasm for algo-
rithmic approaches by some government agencies, perhaps our interest is in
ensuring that the state exploits big data more effectively to fulfill its duty to
protect its citizenry? Do we wish instead to examine and address the roles and
responsibilities of the private sector and academia in supporting and developing
Canada’s big data policy and capabilities? What about the role of big data and
cybersecurity; specifically, are we interested in examining and commenting on
the growing involvement of SIGINT agencies in defending critical infrastruc-
tures operated by the private sector? What are our views on “outsourcing” aspects
of critical infrastructure protection to private sector actors, such as telecom-
munications providers? Finally, given artificial intelligence’s evolving capabilities,
what are our views on ensuring the interrogability and reliability of currently
deployed technologies and the safety of future technologies being developed in
270 Holly Porteous

well-funded labs around the world? Choosing among these questions will help
us better marshal our multidisciplinary effort and resources.
Even a research agenda focused on the first question alone – how the Can-
adian security and intelligence community and law enforcement exploit big
data – suggests to me that knowledge gaps persist and our assumptions need
to be revisited. Let me present some examples to support this claim. The first
example highlights the possibility that, for a SIGINT agency such as the
Communications Security Establishment (CSE), the utility of big data tech-
niques varies according to the intelligence collection context. The purpose
of intelligence collection is generally understood to be the provision of
assessed information on the capabilities and intentions of individuals, groups,
or foreign countries to help state and law enforcement officials make deci-
sions. Capabilities are relatively easy to assess; intentions, not so much.
Humans can say one thing and do another. Their minds are essentially black
boxes. Big data’s pattern matching can offer some insight into hidden inten-
tions but, absent a reliable template for what bad intentions look like (“indi-
cators and warnings,” in the parlance), the results may not be a reliable
predictor of future actions.
The mismatch between what big data analytics can deliver and the intelligence
task at hand is a recurring theme in this book. Big data analytics, it has been
found, are largely ineffective in detecting and preventing terrorist threats. Like-
wise, algorithms may be creating more nightmares for the financial institutions
forced to use them to meet their FINTRAC (Financial Transactions and Reports
Analysis Centre of Canada) reporting requirements than for terrorist financiers.
Finally, the enthusiasm of Canadian police services for big data is not necessarily
matched by an informed understanding about how to integrate associated tools
and techniques.
The story may be different, however, for cyber threats. Readers will recall that
each of the Five Eyes SIGINT agencies, including CSE, also has a cybersecurity
mandate. That means they are in the business of producing intelligence on cyber
threats. Other Canadian agencies also have a mandated interest. For its part,
the Canadian Security Intelligence Service (CSIS) has a requirement to inves-
tigate and assess all threats to the security of Canada, including cyber threats
such as cyber espionage. The RCMP investigates criminal activities, including
cyber crime.
In contrast to terrorists working in the physical world alone (increasingly, a
near-impossible enterprise), cyber threats are bounded by the technical protocols
and characteristics of the global information infrastructure through which they
operate. In the cyber domain, it is possible – though data-intensive – to deter-
mine what normal activities look like. Therefore, it is also possible to identify
 Afterword 271

anomalous and potentially malicious traffic. Cyber threats are thus conducive
to big data analysis.
Indeed, for intelligence agencies, police, and systems security administrators
more generally, network and host defence has always been about parsing mas-
sive logs of transmissions traffic, hunting for patterns that indicate malicious
activities. Until recently, much of the work in this area has focused on recreating
events after an attack, so-called digital forensics. Now, thanks to advances in
artificial intelligence, it is increasingly realistic to speak of intrusion prevention
through near-real-time, automated responses to suspected malicious traffic.
SIGINT agencies, whose intelligence collection activities have been conducted
through the global information infrastructure for decades, are interested in
marrying their leading-edge capabilities with commercially available big data
tools to achieve intrusion prevention.
So, for cybersecurity, the big data tools are finally catching up to a long-
standing demand. With the explosive growth of the poorly secured Internet of
Things and increasingly aggressive Russian and Chinese cyber attacks, intelli-
gence officials say building automated defence capabilities into our networks
has become a “red lights blinking” undertaking. There is no choice but to fight
their algorithms with our algorithms.
Those who track public statements of Five Eyes SIGINT agency officials know
that cybersecurity now stands equal to, if not greater than, counterterrorism
among national security priorities. Yes, terrorism remains a significant concern
and big data does have some utility in this domain, but SIGINT agencies have
for some time been re-engineering their foreign intelligence collection capabil-
ities to tip and queue automated cyber defence capabilities.
Chapter 8 provides a sense of this policy trajectory. The programs it discusses
indicate that, along with hacking into adversary networks, bulk data collection
and machine learning are viewed as fundamental to national cyber defence.
If collecting bulk datasets for cybersecurity is an inescapable part of the
national security picture, then we must understand the implications. For
example, is the price of better securing our digital lifestyle a reduced expectation
of privacy and an expanded peacetime role for intelligence agencies and the
military? What role will private sector entities such as communications service
providers play in defending Canada’s cyberspace? How can we ensure that data
collected for cybersecurity purposes is stored, used, and disposed of properly?
Can our cyber defence systems be turned against us, either from within or from
without? In a related vein and drawing on observations made in Chapter 5, how
will smaller businesses who can’t afford tailored algorithms protect themselves
against cyber threats? Regarding SIGINT agencies’ use of zero-days, through
what mechanisms are vulnerabilities equities issues being addressed and who
272 Holly Porteous

speaks for proportionality concerns?1 What consideration is being given to

other forms of civil defence against cyber attacks, such as enhancing societal
resilience through preservation of analogue mechanisms for critical services?
To what extent are we handing over potential life-and-death decisions to
machines?
A second example inspired by this book also reveals a rich vein of follow-on
research questions. It entails an exploration of what transparency and account-
ability mean in an age of algorithms. Some believe there are limits to what can
be achieved in this context. For example, Mike Ananny and Kate Crawford
highlight the inutility of transparency measures that “privilege seeing over
understanding.”2 Useful transparency, they say, not only provides the visibility
of a system but also affords observers sufficient understanding of the system’s
internal and external dynamics to debate and change it. In this connection, I
would draw your attention to recent revelations that the CSE Commissioner
learned about a technical glitch that had caused CSE to share unminimized
Canadian metadata with foreign allies for years only because CSE volunteered
the information itself. SIGINT watchdogs in the United Kingdom and United
States have also reported similar glitches and “technical irregularities.”
With current approaches to machine learning, where even a system’s creator
may not be able to explain how it works, useful transparency may not be achiev-
able. Still, proponents of causality modelling in machine learning – Judea Pearl
being foremost among this group3 – may help lead the way to enhanced and
transparent artificial intelligence. Simply put, to be able to ask learning machines
useful questions, including questions about their internal reasoning, we need to
put more work into framing how these machines approach cause and effect when
we design them.
A word about achieving big data transparency through horizontal accountability
mechanisms. Parliament, the judiciary, media, and civil rights groups each con-
tribute to the public’s awareness and thereby to informed debate. For example,
even if the judiciary’s accountability function is limited to the application of
existing laws, its public disclosures of redacted versions of key national security
decisions provide valuable information to Canadians about current and proposed
national security activities, including bulk data collection. Journalists, the dedi-
cated ones anyway, do their bit by continuously requesting the release of govern-
ment records under the Access to Information Act. Finally, and as this book attests,
Canada is in the remarkable position of being home to a pioneering group of
scholars and civil rights activists who are all focused on the big data issue.
But the question remains: Is this accountability matrix up to the task of
engaging government in meaningful and productive ways on big data used
within a national security context? Does each element have access to sufficient
 Afterword 273

information and expertise to recognize poor practice and challenge it? Alterna-
tively, is each element capable of recognizing and supporting sound practices
in this area? If not, what measures would enhance capacity while respecting
secrecy requirements?
Regarding secrecy requirements, I would draw attention to the exceedingly
small pool of individuals in this country who can claim genuine expertise in
the operational aspects of intelligence collection. Most of these individuals are
bound to secrecy for life under the Security of Information Act and they steer
well clear of providing any public commentary that would shed light on intel-
ligence collection sources and methods. This situation presents a challenge for
those who would like to expand their knowledge.
Given the far-reaching policy decisions that Canada is poised to take regard-
ing a national data strategy, perhaps there may be value in examining what is
being said on this matter in our own Parliament and in the legislatures of other
allied jurisdictions. Out of this examination should come a sense of where civil
rights fit into this discussion. Now is also the time to consider what constitutes
an appropriate balance between personal data sovereignty and the public goods
that can accrue from machine learning.
Long before the Snowden leaks, legal scholars were voicing concern about
intelligence agencies’ outsourcing of data collection to the private sector. Indeed,
the two editors of this book have for many years been instrumental in educating
the public about how surveillance assemblages are being created out of these
types of linkages.
With the recent controversy over Facebook’s sharing of personal data from
millions of its users’ accounts with a political consulting firm possessing Can-
adian connections, the time is ripe for additional research on how data brokers
operate in this country. Certainly, there is a gap in our knowledge regarding
the specifics of how CSE and CSIS plan to use proposed new authorities to col-
lect and use datasets containing “publicly available” information. A key public
policy question would be whether these new authorities could incentivize
increased private sector data collection.
Of course, it appears that we are all doing our bit to help private sector service
providers collect our personal data. As this book has shown, the death of privacy
often takes place in the fine print. Our “ignoring culture” leads most of us to
hurry past consent form documentation, clicking quickly on the consent button
at the end to gain access to desired services. If being informed about the privacy
implications of gaining access to a “free” service means reading through pages
of impenetrable legalese, nobody wants to be informed. While efforts are being
made to use plain(er) language on consent forms, the documentation sweet
spot may never be found.
 274 Holly Porteous

Perhaps part of the solution lies in the broader public discourse about the
deleterious effects of surveillance capitalism that giant social media platforms
like Facebook, Twitter, and Google have come to personify. Related to this
discussion is the question of user choice. Increasingly, citizens are being herded
towards a digital monoculture that makes no room for analogue service delivery.
Can people truly consent to sharing their personal data when, to access critical
services such as public transit, they are forced to use a smart card?
In many ways, this book is the result of a breakdown in trust. It would not
exist were it not for elements of the US national security community who did
not like what they were seeing and decided to publicly disclose what they knew.
Though they excoriated Edward Snowden as a traitor, even some senior US
national security officials have come to admit that the ensuing public debate
about previously classified matters has been necessary and useful. Here in
Canada, recent polling indicates that most Canadians trust government to
protect their privacy but still think they should be given more information about
how it collects and uses their personal information. The same polling also shows
that most Canadians don’t know that their own intelligence agencies exist, let
alone what they do.
Here, too, I see an opportunity for research, and thereby education.

Notes
Holly Porteous is an analyst with the Justice and National Security section of the Library
of Parliament’s Parliamentary Information and Research Service. The views she expresses
here are hers alone and do not reflect those of the Library of Parliament.
1 The term “vulnerabilities equities” refers to the choice between enhancing overall cyber-
security through public disclosure of previously unknown exploitable cyber vulnerabil-
ity information and protecting an operational capability by maintaining secrecy.
2 Mike Ananny and Kate Crawford, “Seeing without Knowing: Limitations of the Trans-
parency Ideal and Its Application to Algorithmic Accountability,” New Media and Society
20, 3 (2016): 973–89.
3 See Judea Pearl and Dana Mackenzie, The Book of Why: The New Science of Cause and
Effect (New York: Basic Books, 2018). See also Judea Pearl, “Theoretical Impediments
to Machine Learning with Seven Sparks from the Causal Revolution,” 11 January 2018,
arXiv:1801.04016.
Contributors

Anthony Amicelle is an associate professor of criminology at the University of

Montreal. In light of repeated corporate, corruption, and tax evasion scandals
as well as the pervasiveness of terrorism financing issues, his research exam-
ines the policing of economic activities and money flows, with a specific focus
on financial surveillance technologies in the digital age. His recent publica-
tions include the edited volume The Policing of Flows: Challenging Contem-
porary Criminology (2019) and “Research Ethics at Work: Accountabilities in
Fieldwork on Security,” in Secrecy and Methods in Security Research, edited by
Marieke de Goede et al. (2019).
Janet Chan is a professor in the Faculty of Law, University of New South
Wales, Sydney, and co-leader of the Data Justice Network at the Allens Hub for
Technology, Law and Innovation. Her research analyzes the conceptualization
of police culture – e.g., Changing Police Culture (1997), Fair Cop: Learning the
Art of Policing (2003) – and its relationship with the use of technology such as
information systems, big data, and artificial intelligence in policing.
Andrew Clement is a professor emeritus of information at the University of
Toronto. With a PhD in Computer Science, he has had long-standing research
and teaching interests in the social implications of information/communication
technologies and privacy/surveillance issues. Among his recent/ongoing re-
search projects are IXmaps.ca, an Internet mapping tool that helps make more
visible National Security Agency mass Internet surveillance activities and the
Snowden Surveillance Digital Archives, an online searchable collection of all
documents leaked by former NSA contractor Edward Snowden and subse-
quently published by news media. He is currently a collaborator in the Big
Data Surveillance project and a member of Waterfront Toronto’s Digital Strat-
egy Advisory Panel.
Anne Dagenais Guertin is the communications and research coordinator of
the International Civil Liberties Monitoring Group. Anne has a master’s de-
gree in Law and Social Justice from the University of Ottawa, a certificate in
Women’s Studies from Concordia University, and a Bachelor in International
Studies from the University of Montreal. Prior to working at ICLMG, Anne
276 Contributors

was a teaching and research assistant at the Human Rights Research and Educa-
tion Centre of the University of Ottawa. Anne also organizes with Indigenous
Solidarity Ottawa, is an advocate for consent, anti-oppression, and safer spaces,
and writes political screenplays.

Craig Forcese is a full professor at the Faculty of Law (Common Law Section),
University of Ottawa. He is also an adjunct research professor and senior fellow
at the Norman Paterson School of International Affairs, Carleton University
(from 2017 to 2022), and a National Security Crisis Law Fellow, Center on Na-
tional Security and the Law at Georgetown Law (Washington, DC) (from 2017
to 2020). Craig sits on the executive at the Canadian Network for Research on
Terrorism, Security and Society (TSAS), and is a past president of the Canadian
Council on International Law and the Canadian Association of Law Teachers.

David Grondin is a professor of communication at the University of Montreal.

His work examines how issues of security and mobility are adjudicated as part
of a nexus, notably as it relates to infrastructures, borders, and citizenship in
the digital age. His current project focuses on the technological control of mo-
bilities, taking into account the surveillance, algorithmics, and infrastructures
(such as digital platforms, big data, and artificial intelligence) ensuring the se-
cured mobility of people, data, and objects. His recent publications include Bio-
metrics: Mediating Bodies (PUBLIC: Art/Culture/Ideas special issue, co-edited
with Aleksandra Kaminska, 2020), and Movies, Myth and the National Security
State (with Dan O’Meara, Alex Macleod, and Frédérick Gagnon, 2016).

Jillian Harkness has a master’s degree from the University of Toronto Faculty
of Information, where she worked as an archival assistant helping to build the
Snowden Digital Surveillance Archive. She continues to explore privacy, tech-
nology, and education in her current role as the Head of Library and Learning
Resources at the United World College of the Adriatic in Duino, Italy.

Stéphane Leman-Langlois teaches criminology at Laval University, Quebec.

He has written extensively on policing, national security intelligence, and ter-
rorism. He is currently working on two books, one on the radicalization of
the extreme right in Canada (with Aurélie Campana and Samuel Tanner), and
the other on the state of policing in Quebec (with Gregory Gomez del Prado).

David Lyon is a professor of sociology and director of the Surveillance Stud-

ies Centre at Queen’s University, Ontario. Author or editor of many books on
surveillance, from The Electronic Eye (1994) through Surveillance Studies (2007)
to The Culture of Surveillance (2018), Lyon’s work offers a commentary on sur-
veillance in the modern world. His responses to surveillance events such as the
 Contributors 277

expansion of video surveillance, 9/11, and the Snowden revelations sit alongside
treatments of the wider meanings of surveillance seen in Surveillance as Social
Sorting (2003) or in Liquid Surveillance (with Zygmunt Bauman, 2013).
Tim McSorley is the national coordinator of the International Civil Liberties
Monitoring Group. He combines his passion for civil liberties and social jus-
tice with his background in journalism, research, and analysis in his work with
colleagues and partners to fight for social change. Previously, he was an editor
with the Media Co-op and The Dominion magazine, and served as coordina-
tor for the Voices-Voix Coalition. He is a graduate of Concordia University in
Montreal, with a degree in journalism and political science.
Adam Molnar is an assistant professor in the Department of Sociology and
Legal Studies at the University of Waterloo, Ontario, where he is also a mem-
ber of the Waterloo Cybersecurity and Privacy Institute. Prior to joining the
department in 2019, he was a lecturer in criminology at Deakin University in
Australia. He completed his PhD at the University of Victoria, British Colum-
bia, and a postdoctoral fellowship at the Queen’s University Surveillance Stud-
ies Centre. Much of his work focuses on socio-legal practices of technology-
led policing and security intelligence, which also considers the implications for
civil liberties, social justice, and the politics of associated regulatory responses.
Jeffrey Monaghan is an associate professor of criminology and sociology at
Carleton University in Ottawa. His research examines the policing of social
movements as well as broader policing and surveillance practices influenced
by the “War on Terror.” Along with Andrew Crosby, he co-authored Policing
Indigenous Movements (2018), which details how policing and other security
agencies have developed a prolific surveillance regime that targets Indigenous
movements as national security threats. He is also the co-editor, with Lucas
Melgaço, of Protests in the Information Age: Social Movements, Digital Practices
and Surveillance (2018).
David Murakami Wood is former Canada Research Chair (Tier II) in Surveil-
lance Studies and associate professor of sociology at Queen’s University. He has a
BA in modern history from Oxford and an MSc and PhD from Newcastle, on the
subject of secret Signals Intelligence bases in North Yorkshire, UK. He is a widely
published specialist in the sociology and geography of surveillance, security, and
global cities, particularly in Japan, where he was a Japan Society for the Promo-
tion of Science Fellow in 2013–14, and most recently a 2019 Japan Foundation
Fellow examining the security preparations for the Tokyo Olympics. He was co-
founder and is co-editor-in-chief of the journal Surveillance & Society, and co-
editor, with Torin Monahan, of Surveillance Studies: A Handbook (2018).
278 Contributors

Midori Ogasawara is a Banting Postdoctoral Fellow in the Department of

Criminology at the University of Ottawa. She completed her PhD in sociology
at Queen’s University in 2018, where her dissertation explored the Japanese
colonial biometric systems in occupied Northeast China in the 1920s to 1945,
to unpack a historical origin of today’s digital surveillance. She had a personal
interview with NSA whistle-blower Edward Snowden via video channel in
2016, as Snowden’s first Japanese interviewer, and has published two books
and numerous articles on the NSA’s activities in Japan. She was an investigative
reporter for Japan’s national newspaper, Asahi Shimbun, and was awarded a
Fulbright Scholarship and John S. Knight Professional Journalism Fellowship
at Stanford University in 2004–05. She was appointed an assistant professor in
the Department of Sociology at the University of Victoria, where she will start
teaching in January 2021.
Christopher Parsons received his bachelor’s and master’s degrees from the
University of Guelph, Ontario, and his PhD from the University of Victoria.
He is a senior research associate at Citizen Lab, in the Munk School of Global
Affairs and Public Policy at the University of Toronto. His research focuses
on third-party access to telecommunications data, data privacy, data security,
and national security. Besides publishing in academic journals and presses on
topics such as national security, Internet privacy, and government surveillance
practices, he routinely presents findings to members of government and the
media. His work has been recognized by information and privacy commis-
sioners, Canadian political parties, and national and international non-profits
as advancing the state of privacy discourse.
Holly Porteous is an analyst with the Library of Parliament’s Justice and Na-
tional Security section. Since joining the Library in 2009, she has provided par-
liamentarians with non-partisan analysis on a broad range of national security
issues. Her previous employment spans over two decades of work on defence
and national security matters in Canada, the United Kingdom, and the United
States, within various Government of Canada departments and agencies, the
private sector, and academia. A prolific author, her work has appeared in print
in various media, including academic journals, specialized trade journals, and
government-sponsored reports.
Christopher Prince has been a strategic policy analyst with the Office of the
Privacy Commissioner of Canada since 2007, where his focus has been pri-
vacy implications of national security programs, surveillance practices, over-
sight, and domestic laws governing interception of private communications.
Before that, he was a researcher with the federal Treasury Board Secretariat
and the National Archives of Canada. He has also worked on information
 Contributors 279

management issues in the commercial aerospace, death care, and financial

services sectors. He holds a master’s degree from McGill University’s School
of Information Studies and a bachelor’s degree from the University of King’s
College, Nova Scotia.

George Raine is a researcher based in Toronto. He is a graduate of the University

of Toronto’s Faculty of Information, where he completed a Master of Informa-
tion Degree specializing in Archives and Records Management. He is the winner
of the Toronto Area Archivists Group Award. His research focuses on the dy-
namics that operate between access and privacy in governmental archives, and
how these dynamics are impacted by modern communications technologies.

Bill Robinson owns the blog Lux Ex Umbra (luxexumbra.blogspot.com),

which focuses on Canadian signals intelligence activities past and present.
He is frequently consulted by journalists about Canadian SIGINT activities,
helped the CBC analyze the Snowden revelations, and provided research assis-
tance to the British Columbia Civil Liberties Association for its legal challenge
to CSE monitoring of Canadians. Since 2017, he has been a research fellow
at the Citizen Lab at the Munk School of Global Affairs and Public Policy,
University of Toronto. From 1986 to 2001, he was on the staff of the Canadian
peace organization Project Ploughshares.

Carrie B. Sanders is an associate professor of criminology at Wilfrid Lau-

rier University, Ontario, and Director of the Centre for Research on Security
Practices (CRSP). Her research focuses on police technologies, intelligence
analysis, and technologically mediated interactions in criminal justice. Her re-
search has been published in high-impact international journals such as Gen-
der and Society, British Journal of Criminology, and Policing and Society, and
has received national funding by the Social Sciences and Humanities Research
Council of Canada. Her new book (with Stacey Hannem, Christopher Schnei-
der, Aaron Doyle, and Antony Christensen) is Security and Risk Technologies
in Criminal Justice: Critical Perspectives (2019).

Valerie Steeves is a full professor in the Department of Criminology at the

University of Ottawa. She is the principal investigator of The eQuality Proj-
ect, a seven-year partnership funded by the Social Sciences and Humanities
Research Council of Canada examining young people’s experiences of privacy
and equality in networked spaces. She co-edited (with Jane Bailey) the award-
winning volume eGirls, eCitizens, which examines the ways in which young
people negotiate with governance structures in networked spaces. As the lead
researcher for Young Canadians in a Wired World, she has collected quantita-
tive and qualitative data on digital citizenship since 2004.
280 Contributors

Scott Thompson is an assistant professor in the Department of Sociology at

the University of Saskatchewan, a research fellow in the Surveillance Stud-
ies Centre at Queen’s University, and associate editor of the journal Surveillance &
Society. Often focusing on  historical case studies, his work on government
surveillance also adopts Access to Information research methods in order to
explain and address current and pressing issues in the areas of criminology,
sociology, and surveillance studies.
Micheal Vonn is a lawyer and was policy director of the British Columbia
Civil Liberties Association from 2004 to 2019. She has been an adjunct pro-
fessor at the University of British Columbia in the Faculty of Law and in the
School of Library, Archival and Information Studies, where she has taught civil
liberties and information ethics. Her publication credits include the Birkbeck
Law Review, Surveillance & Society, Journal of Parliamentary and Political Law,
and Case Western Reserve Journal of International Law. Ms. Vonn has been a
frequent speaker on a variety of civil liberties topics, including national secu-
rity, policing, surveillance, and free expression. She is a collaborator on the Big
Data Surveillance project, an Advisory Board member of Ryerson University’s
Centre for Free Expression, and an Advisory Board member of Privacy Inter-
national. In September 2019, she became the chief executive officer of PHS
Community Services Society in Vancouver.
Index

Abou-Elmaati, Ahmed, 62 artificial intelligence. See algorithms; big

Access to Information and Privacy Act data; big data analytics
(ATIP), 14, 272 Atwal v Canada, 168, 172
accountability, 11, 17, 48, 106, 160–61; Australian Signals Directorate (ASD), 257,
demand for, 7, 126, 141, 143, 201, 265; 259. See also Five Eyes (FVEY)
government, 238–39, 245; horizontal,
239, 245, 247, 248, 272. See also Babuta, Alex, 183
“capability transparency”; Intelligence Bamford, James, 119
Commissioner; National Security banks, and compliance with surveillance,
and Intelligence Review Agency 74–75. See also financial surveillance
(NSIRA); rule of law; Security BCCLA v CSE, 179n25, 198; impact of, 201–3,
Intelligence Review Committee (SIRC); 211; and scope of the lawsuit, 199–200
transparency Bello, Walden, 209
“actionability,” 182 Belur, Jyoti, 187
activism. See resistance, anti-surveillance; Bennett, Colin, 218, 219, 220, 221, 224,
social movements 227, 228
Advanced Research Projects Network bias, 83, 122, 182; and algorithms, 63,
Agency (ARPANET), 91 64, 72, 192; structural, 62, 63. See also
affordance theories, 222 profiling
Air India Flight 182, 44 big data, 3, 4–5, 65, 169, 253; age of, 4, 38,
algorithms, 5, 38, 60, 62, 63; as 70, 102, 180; corporate use of, 5, 73, 273,
black boxes, 72, 81, 192; as digital 274; and financial surveillance, 78–80;
infrastructures, 71, 78; and financial and intelligence agencies, 103–4, 112, 118,
flow, 70, 78–80; and metadata, 131, 164, 270; notion and attributes of,
260–62, 264; and mobility control, 68, 3–4, 5, 117, 120, 182; practices involving,
71; as suspecting machines, 79–80. 21, 28, 78, 113, 270; and security, 112,
See also big data; big data analytics; 122, 151, 154, 169. See also financial
financial surveillance; logistical media; surveillance; “New Analytic Model”;
transparency and entries starting with big data
Almalki, Abdullah, 62 big data analytics, 14, 17, 57, 60, 63–64, 68;
Al-Qaeda, 77. See also terrorism and “data junkies,” 82; and data mining,
Ananny, Mike, 272 112, 264; and “full-take” capture, 131,
Anderson, David, 43, 169 132; issues involving, 120–22, 160, 169;
Anderson, Steve, 206 and machine learning, 112, 118; and
Angus, Ian, 220 mass surveillance, 130–31, 158–59, 169,
anti-surveillance. See resistance, anti- 253; and proportionality, relevance,
surveillance and necessity limitations, 157–59;
Anti-terrorism Act. See Bill C-51 and security intelligence, 105, 114–16,
Apple, 21, 27, 30 117, 154, 180, 229; and terrorism, 180.
Arar Affair, 48; Maher Arar, 4, 62 See also “New Analytic Model”
282 Index

big data policing, 180–81, 187; capabilities bulk collection. See mass surveillance
of, 181, 182–83, 184; cultural challenges Bundesnachrichtendienst (Germany), 259
in, 184; empirical studies on, 181, 183, 184;
legislative and policy barriers in, 187; and cable, traffic, 91, 135, 139, 253. See also
open-source analytics, 187; resources, International Licensed Carriers (ILC);
organizational and personal barriers in, undersea cables
186–87, 189, 190; technological challenges Calgary Police Services, 181
in, 185–86. See also policing Canada Evidence Act, 44
big data surveillance, 28, 57, 73, 78–80, Canada Revenue Agency (CRA), 156, 157
121, 183; and Internet and surveillance Canada-USA Agreement (CANUSA), 8
capabilities, 101–2, 129; and security, Canadian Access to Information Act
112, 116, 118, 129, 169. See also (ATIA), 6
Canadian Security Intelligence Service Canadian Charter of Rights and Freedoms,
(CSIS); Communications Security 150, 199. See also litigation and
Establishment (CSE); Five Eyes surveillance; rule of law; Supreme
(FVEY); National Security Agency Court of Canada
(NSA); “New Analytic Model”; Canadian Overseas Telecommunications
Snowden’s revelations; surveillance; Corporation (COTC), 92
and entries starting with big data Canadian Security Intelligence Service
Bill C-13, 11, 140, 208 (CSIS), 10, 61; and big data/bulk data
Bill C-22, and CSE oversight, 245 collection, 150, 154, 164, 173, 176; and
Bill C-30, 10, 140, 205, 206, 211 collection and retention of associated
Bill C-46. See Investigative Powers for the data, 174–75; creation of, 9–10, 94;
21st Century Act and datasets acquisition, collection,
Bill C-47. See Technical Assistance for Law and usage, 161–62, 163, 174–75; and
Enforcement in the 21st Century Act datasets retention, 174–76; and
Bill C-51, 11, 48, 100, 113, 149, 199; foreign intelligence collection, 94; and
background of, 218; opposition to, information sharing under SCISA,
218–19, 221, 225, 229. See also 155–57; and information sharing between
terrorism agencies, 180; legal challenges and rule
Bill C-59, 10, 48, 106, 149, 164, 199; and of law issues involving, 150, 151, 157,
accountability of security intelligence 164, 172, 174; and money laundering,
agencies, 160–61, 170, 176; and bulk 70; and “non-collected” datasets, 152;
powers, 170, 172, 175, 176, 229; and oversight of, 152–53, 170, 176; and private
CSE’s powers, 127, 203, 204, 243; communication interception, 168;
datasets provisions in, 161, 176, 177; and powers and statutory mandates of, 10, 11,
oversight, 170, 176–77, 199, 203, 246. 151, 163, 172, 173, 270; and warrants, 150,
See also accountability; Communications 168, 173, 199. See also mass surveillance;
Security Establishment (CSE); “New Analytic Model”; security
transparency intelligence; surveillance; and entries
biometric information, 7 starting with big data
Blaney, Steven, 10 Canadian Security Intelligence Service
Booz Allen Hamilton. See Snowden, Act, 44, 150, 151, 173; and information
Edward; Snowden’s revelations collection and retention, 174–75; and
boyd, danah, 120, 182 provisions on datasets, 161–62
Brayne, Sarah, 183 CANUKUS Agreement, 90, 95
Britain-USA Agreement (BRUSA), 8 “capability transparency,” 128, 141, 143–44
British Columbia Civil Liberties CASCADE program. See Communications
Association (BCCLA), 150, 171, 198. Security Establishment (CSE): CASCADE
See also BCCLA v CSE program
 Index 283

case law. See litigation and surveillance 262–63; and National Defence Act, 199–
Central Intelligence Agency (CIA), 8 200, 262; oversight of, 106, 127, 170, 172,
Chan, Janet, 182, 183 237–238, 244; private communication
Charkaoui v Canada (Citizenship and interception by, 92, 97, 99, 100, 105, 150,
Immigration), 174–75 170, 172, 199, 241; secrecy culture and, 127,
chilling effect, 62–63, 197, 226 128, 142, 143, 144; statutory mandates of,
Chrétien, Jean, 100 100, 105, 106, 127, 170, 199, 242, 243, 262;
Christopher, David, 205, 206 and Tutte Institute for Mathematics and
civil liberties. See Canadian Charter of Computing (TIMC) partnership, 112,
Rights and Freedoms; litigation and 113, 122. See also mass surveillance; “New
surveillance; rule of law Analytic Model”; security intelligence;
“clicktivism”, 231. See also resistance, anti- surveillance; and entries starting with big
surveillance; social movements data
Cold War, 8, 9; end of and shift in Communications Security Establishment
intelligence concerns, 97, 257; and state Act, 162, 203, 204, 243. See also Bill C-59
surveillance, 89, 90 connective action. See network theory;
Communications Branch of the National social movements
Research Council (CBNRC), 89, 113. Convention on Cyber Crime (Council of
See also Communications Security Europe’s), 36
Establishment (CSE) Crawford, Kate, 120, 182, 272
Communications Security Establishment Cray computers, 90, 95, 96, 103
(CSE), 9, 11, 14, 23, 113, 237; and crime. See financial surveillance; policing;
Afghanistan, 100; and big data, 101, terrorism
112, 114, 116, 117, 118; and CASCADE Criminal Code, 44, 50, 173, 255, 262, 264
program, 132, 141, 241, 242, 243; and cryptanalysis, 95, 96, 103, 119. See also
collaboration with domestic agencies, encryption
100, 105, 237, 243; and collaboration counterinsurgency, 8, 9
with private actors, 103, 113, 122, 137, counterterrorism. See terrorism
140, 243; and collaboration with court of law. See litigation and
the NSA and FVEY, 95, 100, 112, surveillance
140, 257; and collaboration with the Customs Act, 44, 45
RCMP, 93; Commissioner of, 99; and cybersecurity, 270–71. See also
Communications Research Squadron, Communications Security
97, 99; and cybersecurity, 106, 144, Establishment (CSE): and cybersecurity
170, 172, 243, 269, 270; and domestic
surveillance, 100, 101, 126, 129–31, 171, data, personal, 3, 21, 254; “raw,” 5, 112, 120.
241; domestic surveillance capabilities See also entries starting with big data
of, 132, 134, 140, 170, 202, 242, 263; and data analysis, 6, 82. See also entries
EONBLUE program, 101, 132, 139, 241; starting with big data
growth of, 96, 97, 99, 100, 101, 104, 144; data collection, in bulk, 64, 112, 128, 149,
historical background of, 89, 93, 240; 169; and “non-collected” datasets, 152.
and INDUCTION program, 132, 241; See also entries starting with big data
and intelligence focus shift, 100, 112, and/or mass surveillance
113–14, 115; legal challenges and issues “data exhaust,” 5, 114
involving, 150, 171, 172, 199, 242; and “data junkies.” See big data analytics
legal compliance, 99, 100, 105–6, 127, data mining, 61, 62, 112, 154, 264
243, 248; and mass surveillance, 129, data protection, 36. See also litigation and
130–32, 134, 200, 241; and Mathematics of surveillance; privacy
Information Technology and Complex dataism, 5. See also entries starting with
Systems (MITACS), 103; and metadata, big data
284 Index

datasets, and Bill C-59, 161, 176, 177; false positive/negative, 60, 62, 80
definition of, 176. See also big data; Federal Court, 150, 154, 158, 168, 173,
Canadian Security Intelligence Service 199; and dataset retention, 176–77.
(CSIS); data collection See also litigation and surveillance;
dataveillance, 69. See also entries starting rule of law
with big data and/or surveillance Ferguson, Andrew, 191
deep packet inspection (DPI), 130, financial flow, 69, 73, 76–77. See also
241, 268n50. See also big data financial surveillance; terrorism:
analytics; Communications Security financing and financial surveillance of
Establishment (CSE): CASCADE financial surveillance, 69, 70, 72–73; and
program; Communications Security monitoring devices and capabilities,
Establishment (CSE): EONBLUE 76–80; and suspicion production, 78,
program; mass surveillance; 83–84; and terrorism, 74, 77. See also
metadata financial flow; terrorism, financing and
Defense Advanced Research Projects financial surveillance of
Agency (DARPA), 58 Financial Transactions and Report
Department of National Defence, 93, Analysis Centre of Canada
113; collaboration with the NSA, 95. (FINTRAC), 68, 70, 73, 77, 79
See also Communications Security fingerprinting, 7
Establishment (CSE) First Nations. See Indigenous peoples
digital rights, 14, 50. See also Canadian Five Eyes (FVEY), 8–9, 11, 23, 90, 102, 254;
Charter of Rights and Freedoms; and data interception and analytics, 126,
intelligence and law; litigation and 254, 259; and diplomatic surveillance,
surveillance; rule of law 260; and mass surveillance, 151, 241,
“digitally networked public sphere,” 220. 257; post-9/11 transformation, 101, 180;
See also resistance, anti-surveillance; subcontract on domestic surveillance,
social movements 62. See also National Security Agency
DiPuma, Caily, 199, 200, 201 (NSA); security intelligence
Director General for Intelligence, 115. Forcese, Craig, 255
See also Communications Security Foreign Intelligence Surveillance Act
Establishment (CSE) (FISA), 25, 244
Distant Early Warning (DEW), 9 Foreign Intelligence Surveillance Court
Donahue, Laura, 43 (FISC), 48
Forster, John, 131. See also Communications
Earl, Jennifer, 222 Security Establishment (CSE)
ECHELON program, 32, 96, 97, 257. Freedom of Information Act, 6
See also Cold War; Five Eyes (FVEY) Freeze, Colin, 113
Emergency Powers Act, 168
“emergent publics,” 220 gaming metaphor, 117, 121
encryption, 104, 119. See also cryptanalysis Gerbaudo, Paolo, 220
EONBLUE program. See Communications Gilliland, Anne J., 256
Security Establishment (CSE): Global Affairs Canada (GAC), 156
EONBLUE program Goldman, Zachary, 245, 248
Ericson, Kevin, 46 Google, 21, 27, 30, 102, 274
espionage. See security intelligence Government Communications
extremism. See terrorism Headquarters (GCHQ), 14, 102, 135,
258; and international collaboration,
Facebook, 38, 130, 231; and collaboration 90, 91, 129, 254, 257, 259. See also Five
with the NSA, 21, 102; and policing, 183, Eyes (FVEY)
188; and surveillance capitalism, 273, Government Communications Security
274. See also social media Bureau (GCSB), 257
 Index 285

Haggart, Blayne, 221 intelligence gathering, 91–92, 113–14, 270.

Haggerty, Kevin, 46 See also intelligence agencies; intelligence
Halupka, Max, 231 sharing; security intelligence; and entries
Harcourt, Bernard, 5 starting with big data
Harper, Stephen, 10, 206, 218 intelligence sharing, 105, 155–57, 180.
Hayes, Ben, 209 See also Five Eyes (FVEY); security
HUMINT, 6, 7 intelligence; Snowden’s revelations
Hunter v Southam, 168, 172. See also interception of private communications,
litigation and surveillance; rule of law 150, 168, 173, 199–200; and cell phone
data, 181, 200; legal definition of, 255.
identification, 7 See also Communications Security
Idle No More. See Indigenous movements Establishment (CSE); intelligence and
immigration concerns, 7 law; litigation and surveillance; rule of
Income Tax Act, 157. See also Canada law; warrant
Revenue Agency (CRA) International Campaign Against Mass
Indigenous movements, 11–12; and anti- Surveillance (ICAMS), 198, 208.
pipeline, 12; and Idle No More, 12. See also mass surveillance
See also resistance, anti-surveillance; International Civil Liberties Monitoring
social movements Group (ICLMG), 208
Indigenous peoples, 7, 10. See also International Licensed Carriers (ILC), 91;
Indigenous movements cross-border traffic collection, 91, 93,
INDUCTION program. See 94. See also cable; undersea cables
Communications Security Internet. See Signals Intelligence
Establishment (CSE): INDUCTION (SIGINT); Snowden’s revelations;
program surveillance; and entries starting with
information, biometric. See biometric big data
information Internet Service Providers (ISPs): and
infrastructure. See algorithms, as digital communications interception, 129,
infrastructures 130, 208; and transparency regarding
intelligence: artificial, 63–64; economic, collaboration with CSE, 142. See also
97. See also algorithms; intelligence and metadata; Signals Intelligence (SIGINT);
law; security intelligence; surveillance; and entries starting with big data
and entries starting with big data Interpretation Act, 171; and definition of
Intelligence Advisory Committee (IAC), 93 “telecommunications,” 179n20
intelligence agencies, 8, 50; increase Investigative Powers for the 21st Century
in legal powers of, 10, 126, 127, 149, Act, 10
218; oversight, 11, 106, 127, 150–51; Investigatory Powers Act (IPA) (UK), 169
partnership between agencies, 8–9, 12, Isbister, Claude, 91
89. See also Five Eyes (FVEY); security Islamic State of Iraq and the Levant
intelligence; Snowden’s revelations (ISIL), 77. See also terrorism
intelligence and law, 43, 96, 151, 164, IXmaps. See transparency: IXmaps
218; and legitimization of obscure
intelligence practices, 164, 167, 204, Jaffer, Jameel, 209
243; and legitimization of secret Japan, relationship with the NSA, 23,
international partnerships, 32–35, 37; 32–35, 37. See also intelligence and
and judicialization of intelligence, 167; law; National Security Agency (NSA);
search-and-seizure and bulk collection, Snowden’s revelations
172, 200. See also litigation and Johnson, Shane, 187
surveillance; rule of law Judd, Jim, 167
Intelligence Commissioner, 127, 144, 161, judicial review. See justiciability; litigation
172, 176, 203, 244 and surveillance; rule of law
286 Index

justiciability, 48. See also litigation and See Communications Security

surveillance; rule of law Establishment (CSE)
media. See social media
Keable Inquiry, 93 metadata, 141, 170, 171, 174; and big data
Klein, Mark, 24, 137 analytics/mass surveillance, 253, 256–57,
Kimport, Katrina, 222 263, 264; conventional meaning of, 254,
Korody-Nuttal case, 58 256, 263–64; governmental definition
(Canada), 109n65, 170–71, 255, 263–64;
law and surveillance, 44–45; complexity governmental definition (FVEY
and fragmentation, 47, 199. See also partners) of, 255–56, 258, 259–62; IP
intelligence and law; litigation and address as, 130, 171, 260, 262–63; legal
surveillance; rule of law protection of, 253, 264; and retention
lawful access, 10, 140, 205, 208, 222. as associated data, 174–75. See also
See also Bill C-13; Bill C-30; rule of law; algorithms; big data; big data analytics;
Stop Online Spying campaign; warrant big data surveillance; XKEYSCORE
Lawful Access Act. See Bill C-30 program
litigation and surveillance: and bulk data Metropolitan Police of London, 10
surveillance, 150, 198; and collection Microsoft, 21, 27, 30, 102
and retention of associated data, Miller, Seumas, 248
174–76; and the concept of metadata, Mobile Device Identifier technology, 181
255, 264; insight into state surveillance mobility control, 68. See also entries
practices through, 150, 154–55, 201, 202, starting with security
204; as means of political pressure for money laundering. See financial
legal changes, 203–4, 211; and public surveillance
awareness, 201–2, 203, 211. See also moral entrepreneur, 61
BCCLA v CSE; Hunter v Southam; Moses, Lyria Bennett, 182, 183, 191
intelligence and law; R v Spencer; rule
of law; Supreme Court of Canada; X National Defence Act 44; and CSE
(Re), 2016 FC 1105 oversight, 106; and CSE statutory
logistical media, 70–71, 81, 83 mandate, 100, 150, 170, 199, 240, 262
Los Angeles Police Department (LAPD), 183 National Defence Radio Establishment
Lyon, David, 5, 62, 224 (FRA), (Sweden), 259
National Intelligence Model (NIM),
McDonald Commission, 9–10, 93 (UK), 181
machine learning, 112, 118, 271, 272. See also National Research Bureau. See
algorithms; big data; big data analytics Communications Security
MacKay, Peter, 10, 208 Establishment (CSE)
Mackenzie Commission, 91 National Security Act, 2017. See Bill C-59
mass surveillance, 4, 5, 11, 21, 62–63, 102, National Security and Intelligence
113, 129, 130, 149, 197, 210; and bulk Committee of Parliamentarians
collection issues, 169, 172–73, 229; (NSICOP), 105. See also
and chilling effect, 62–63; growth Communications Security
of, 212, 218; legal challenges and Establishment (CSE)
issues involving, 150, 169, 174–75, National Security and Intelligence Review
198–213; normalization of, 224. See also Agency (NSIRA), 45, 127, 144, 245;
“New Analytic Model”; resistance; mandate, 160, 163, 244
surveillance; and entries starting with National Security Agency (NSA), 11, 21,
big data 27, 64, 120, 244; and chilling effect,
Mathematics of Information Technology 62–63; and data-gathering capabilities
and Complex Systems (MITACS). and practices, 259; and influence on
 Index 287

international law-making process, pacification. See policing: and

32–36; and intelligence sharing, 105; colonialism; settler-colonialism;
and international collaboration, 23, 90, surveillance: and colonialism
95, 100, 112, 140; and PRISM program, Parsons, Rehtaeh, 208
27, 30, 102; and private collaboration, Pasquale, Frank, 72
24–27, 102; and Soviet communications, Pearl, Judea, 272
95, 257; and TURBULENCE and Pearson, Lester, 91
TURMOIL systems, 101. See also Five Personal Information Protection
Eyes (FVEY); security intelligence: and and Electronic Documents Act
cooperation between states; Snowden’s (PIPEDA), 44
revelations; XKEYSCORE program Peterson, Marilyn, 188
NATO, 8, 9 policing: and algorithms, 113; and
Neocleous, Mark, 197 colonialism, 7, 197; and intelligence
network: analysis, 59–60, 62; neural, 59, analysis, 186, 188–89; “intelligence-
63. See also social network led policing” (ILP), 64, 180, 183–84;
network theory, 221 internal political, 7, 10, 12; and national
“New Analytic Environment.” See “New security, 180; predictive, 57, 180, 181, 183,
Analytic Model” 189; pre-emptive, 35, 57, 180; preventive,
“New Analytic Model,” 5, 17, 102, 112; and 57, 61–62. See also big data policing;
big data, 114–16, 169; and the CSE, financial surveillance; intelligence and
114–15, 116, 118; issues, 120–22, 169. law; rule of law
See also entries starting with big data Ponting, Clive, 6
Nicholson, Rob, 10, 207, 208 predictability, 48
9/11. See terrorism PRISM program. See National Security
Norris, Clive, 83 Agency (NSA), and PRISM program
Nureddin, Muayyed, 62 privacy, 49, 65, 155, 177; erosion of, 35,
36, 140; protection of, 92, 262, 265;
Obar, Jonathan, 221 violation of, 29, 159, 169. See also
October Crisis, 90 litigation and surveillance; metadata;
Office of the Communications Security rule of law; transparency; warrant
Establishment Commissioner Privacy Act, 44, 156, 176
(OCSEC), 45, 237, 242 “privacy Chernobyl,” 224, 229. See also
Office of the Director of National Bill C-51: opposition to; Snowden’s
Intelligence, 246 revelations: and contribution to anti-
Office of the Privacy Commissioner surveillance campaigns in Canada
(OPC), 11, 209, 255 profiling, 160; algorithmic, 62, 63. See also
Official Secrets Act, 168 algorithms; bias; big data; policing;
Ontario Provincial Police, 181 terrorism
OpenMedia. See Stop Online Spying Protecting Canadians from Online Crime
campaign Act. See Bill C-13
Operational Data Analysis Centre Protecting Children from Internet
(ODAC), 150, 154–55, 173, 230; Predators Act. See Bill C-30
and associated data collection and protection, data. See data protection
retention, 174, 175; data collection Protection of Privacy Act, 92, 168
via warrant, 173–74. See also big
data analytics; Canadian Security Quadripartite Agreement, 9
Intelligence Service (CSIS); metadata;
X (Re), 2016 FC 1105 R v Spencer, 51, 140, 171, 255
oversight. See accountability; Bill C-59; Radiocommunication Act, 44, 50
transparency Rascoff, Samuel, 245
288 Index

Regulation of Investigatory Powers Act activities, 245, 247, 248, 265; and rule of
(RIPA), (UK), 255, 262 law, 43, 48, 126, 141, 160, 164, 173, 240.
Renan, Daphna, 246 See also accountability; rule of law;
resistance, anti-surveillance, 198, 220, 222, security intelligence; transparency
225–28. See also Bill C-51: opposition to; Secrecy Act, 32–34
Indigenous movements; International securitization, 3, 49
Campaign Against Mass Surveillance security, 64–65, 68, 230; cameras,
(ICAMS); social movements; litigation 63; certificates, 58; expansion of
and surveillance; Stop Illegal Spying notion of, 4; and mobility, 69; and
campaign; Stop Online Spying campaign surveillance, 13, 116, 121. See also
retroactive immunity, 25–26, 29, 35. security intelligence
See also intelligence and law security, national, 3, 66, 112, 115, 118, 198.
Ridgeway, Greg, 180, 181, 188 See also Canadian Security Intelligence
risk assessment, 180 Service (CSIS); Communications
Roach, Kent, 245 Security Establishment (CSE); security
Robertson, Gordon, 91 intelligence; terrorism
Royal Canadian Military Police, 181 security intelligence, 3, 5, 6; and
Royal Canadian Mounted Police (RCMP), cooperation between domestic
9, 10, 12, 46, 229; and collaboration agencies, 12, 13, 93, 95, 100; and
with the CSE, 93; and national security, cooperation between states, 12, 13, 23,
58, 62, 180. See also McDonald 30, 37, 89, 93, 95, 101, 102, 112, 180, 257,
Commission 259; and information sharing, 155–57,
Royal Canadian Mounted Police Act, 44 180; and partnership with private
Royal Commission of Inquiry into actors, 21, 27–29, 73, 102, 103, 122, 137,
Certain Activities of the RCMP. See 243, 273. See also Five Eyes (FVEY);
McDonald Commission intelligence agencies; intelligence and
Royal Commission on Security. See law; Snowden’s revelations
Mackenzie Commission Security Intelligence Review Committee
rights, digital. See digital rights (SIRC), 11, 150; mandate of, 150–51;
rule of law, 43, 105–6, 141, 150, 164; potentially unlawful data collection by
and accountability, 240, 265; and the CSIC, 152–153; recommendations
constitutional privacy/search and and guidelines proposal, 153, 162.
seizure protections, 155, 168, 171, 172, 176, See also accountability; transparency
200. See also accountability; intelligence Security of Canada Information Sharing
and law; litigation and surveillance; Act (SCISA), 50, 155–56; and Privacy
secrecy: and rule of law; transparency Act, 156
Security of Information Act (SOIA), 44,
sandbox. See gaming metaphor 143–44, 273
Sanders, Carrie, 113, 183, 191 security thickness, 230. See also secrecy
satellite, 91, 95; monitoring, 96, 241 Segerberg, Alexandra, 219, 220, 221, 227, 228
Schott, Nicole, 183 Sensenbrenner, Jim, 160
search and seizure. See Hunter v Southam; settler-colonialism, 7, 197. See also
litigation and surveillance; rule of law: policing: and colonialism; surveillance:
and constitutional privacy/search and and colonialism
seizure protections; Supreme Court of Sheptycki, James, 113
Canada Signals Intelligence (SIGINT), 6, 8, 115,
secrecy, 43, 224; as a barrier to public 257, 271; and CSE, 97, 99, 113, 170; and
debates on surveillance and security, Department of National Defence, 113;
223, 273; culture of, 6, 128, 142, 143, 211, and Internet traffic, 98–99, 100. See also
229–30; and legitimacy of intelligence big data analytics; Communications
 Index 289

Security Establishment (CSE); Five Stop the Meter campaign, 205

Eyes (FVEY); International Licensed Supreme Court of Canada, 49, 51, 168,
Carriers (ILC); National Security 171, 174; and metadata, 255, 264; and
Agency (NSA); “New Analytic Model” subscriber information and warrant,
Snowden, Edward, 4, 11, 21, 24, 27, 32 140. See also Canadian Charter of
Snowden Digital Surveillance Archive, Rights and Freedoms; Charkaoui v
254, 258, 266 Canada (Citizenship and Immigration);
Snowden’s revelations, 21, 27, 104, 119, 199, litigation and surveillance; Hunter v
274; and Canada, 112, 126, 129, 149, 200; Southam; R v Spencer; warrant
and contribution to anti-surveillance surveillance: ambient, 66; assemblage,
campaigns in Canada, 220, 224–25, 229, 46, 75, 83–84, 273; and behaviour, 63,
230; and CSE, 104, 241, 242; and data 64; and cameras, 63; and colonialism,
interception capabilities, 137, 169, 224, 197; domestic, 23, 91, 92, 129; illegal,
253, 254; and NSA/FVEY surveillance 23–26, 33–35, 96, 150; law and, 43–46,
programs, 158, 257, 259–61. See also Five 50, 149, 199, 218; by private actors, 21,
Eyes (FVEY); National Security Agency 121, 273. See also big data surveillance;
(NSA); XKEYSCORE program intelligence agencies; intelligence and
social media, 273, 274; and anti- law; litigation and surveillance; mass
surveillance activism, 225, 226; and surveillance; security intelligence;
policing, 183, 188 Signals Intelligence (SIGINT);
social movements, 225, 226; connective resistance, anti-surveillance
action and, 219, 220, 228, 231, 232; and surveillance capitalism, 5–6, 64, 274
Social Movement Organization (SMO), suspicion production, and financial
219, 220, 227, 228; theoretical aspects surveillance. See financial surveillance:
of, 220–21, 222, 228, 231. See also Bill and suspicion production
C-51: opposition to; resistance, anti-
surveillance Taliban, 77. See also terrorism
social network, 220, 222; dark, 59; and Tassé, Roch, 208, 211
suspicion production, 80 Technical Assistance for Law Enforcement
social sorting, 62; and surveillance, 121 in the 21st Century Act, 10
Special Source Operation (SSO), 27, 140, Telecommunications Act, 44
241, 243. See also Communications telephone traffic collection. See cable;
Security Establishment (CSE); National International Licensed Carriers (ILC)
Security Agency (NSA); Snowden’s terrorism, 10, 11, 57, 69; anti-terrorism,
revelations 57, 73; counterterrorism, 62, 78, 98,
splitter cabinet, 24, 27. See also cable; 100, 104; encryption issues, 104;
undersea cables and false suspects, 62; financing
spying. See intelligence agencies; security and financial surveillance of, 69,
intelligence 73, 74, 78; Islamist, 61, 77, 168;
state intelligence, 7, 224; historical Islamist Terrorism case, 168; 9/11
background of, 6–9, 197. See also and shift to intelligence, 58, 100,
intelligence agencies; security 120, 167; prevention of, 57–59,
intelligence; surveillance 61–62; and threats as justification
state surveillance. See state intelligence for surveillance, 34, 158–59. See also
statistics: atypical behaviour, 59; and big data analytics: and terrorism;
crime, 61; and terrorism, 61. See also financial surveillance; War on Terror
entries starting with big data Terrorism Information Awareness
Stop Illegal Spying campaign. See BCCLA Program. See Total Information
v CSE Awareness Program
Stop Online Spying campaign, 198, 205–7 Todd, Amanda, 208
290 Index

Toews, Vic, 10, 205, 206 War on Terror, 25; and big data analytics
torture, 9, 62, 120n44 and surveillance, 60, 257; and
Total Information Awareness program, surveillance legislation, 35. See also
58, 64 terrorism
tracking devices, 50, 76 warrant: CSIS and judicial, 173; and
transoceanic cables. See undersea cables discretionary power, 168, 170, 171, 172,
transparency, 44, 50, 126, 141, 202, 224, 199; judicial, 5, 25, 50, 105, 107n23,
238; and government surveillance, 140, 169, 171, 173; threshold for
238, 246, 247, 265, 272; and IXmaps, approval of, 167–68, 173, 208. See also
135, 137, 138; lack of, 29, 48, 50, 128, 143, Canadian Charter of Rights and
168, 192; and legality, 47, 141, 160, 246. Freedoms; litigation and surveillance;
See also accountability; “capability rule of law
transparency”; secrecy; Security warrantless access, 10, 140, 168, 205, 208.
Intelligence Review Committee (SIRC) See also lawful access; R v Spencer;
Trudeau, Justin, 12 rule of law; Stop Online Spying
Trudeau, Pierre, 91, 94 campaign
Tufekci, Zeynep, 220 Webb, Maureen, 209
Tutte Institute for Mathematics Weston, Crystal, 183, 191
and Computing (TIMC), 112. whistle-blower, 4, 6, 11, 21–25, 39, 137
See also Communications Security Winnipeg Police Services, 181
Establishment (CSE) wiretapping: illegal, 23–25, 30, 140;
programs, 27, 132. See also cable;
UK-USA Agreement (UKUSA), 8, 32, 90; International Licensed Carriers (ILC);
agencies, 95 Snowden’s revelations; undersea cables;
undersea cables, 8, 21, 27, 90, 149; as XKEYSCORE program
source of intelligence, 90–91, 253.
See also International Licensed X (Re), 2016 FC 1105, 150, 154, 164,
Carriers (ILC); Snowden’s revelations; 173, 229. See also Canadian Security
XKEYSCORE program Intelligence Services (CSIS); litigation
US Privacy and Civil Liberties Oversight and surveillance; Operational Data
Board (PCLOB), 158, 159, 164 Analysis Centre (ODAC)
US SANDKEY, 98 XKEYSCORE program, 258; capabilities
USA PATRIOT Act, 158, 160 of, 259–60. See also cable; Five Eyes
(FVEY); metadata; National Security
van Dijck, Jose, 5 Agency (NSA); undersea cables
van Loan, Peter, 10
Vancouver Police Department, 181 Yahoo!, 27, 102, 130
Varcoe opinion, 168
Zedner, Lucia, 180
Walsh, Patrick, 180, 248 Zuboff, Shoshana, 5. See also surveillance
War on Drugs, 98 capitalism