Government Information Quarterly 33 (2016) 89–104

Contents lists available at ScienceDirect

Government Information Quarterly

journal homepage: www.elsevier.com/locate/govinf

Managing complexity across multiple dimensions of liquid open data:

The case of the Danish Basic Data Program
Thorhildur Jetzek 1
Department of IT Management, Copenhagen Business School, Howitzvej 60, 2000 Frederiksberg, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Current literature on open government data has uncovered a wide range of challenges related to these important
Received 7 April 2015 initiatives. The problems encountered include: insufficient data quality and interoperability, problems regarding
Received in revised form 3 November 2015 governance and motivation, lack of capabilities, and heterogeneous political and ideological agendas. A common
Accepted 4 November 2015
open data infrastructure might resolve some of these problems, however, implementing such an infrastructure is
Available online 24 November 2015
a highly complex task. This longitudinal case study of the Danish Basic Data Program (BDP) is intended to im-
Keywords:
prove our understanding of the challenges related to providing open access to government data through open
Open data data infrastructure. The BDP aims to improve the quality of selected government data, make them more coherent,
Open data infrastructure and improve accessibility through the implementation of a common data distribution platform. The program is
Liquid open data expected to increase government efficiency and stimulate innovation. This case study describes the evolution
System-of-systems governance of the BDP and identifies the main structural elements of an open data infrastructure. Data analysis uncovered
Value generating mechanisms four tensions, which are identified as key challenges of an open data infrastructure implementation. These ten-
sions are presented with four suggested governance strategies that were used in the BDP case. The main contri-
bution of the paper is a process model where the main phases and mechanisms of an open data infrastructure
implementation, use and impacts are identified and explained.
© 2015 Elsevier Inc. All rights reserved.

1. Introduction issues were causing what has been identified by economists as sub-
optimal use of the data resource (see for instance Nilsen, 2010 and
In August 2011, under an increasing need for savings coupled with a Pollock, 2008).
continual pressure to deliver high quality public services, the Danish The value of accessible, interoperable and contextually appropriate
Agency for Digitization published their e-Government strategy for data is becoming apparent in many areas of society. It is estimated
2011–2015. The two main goals of the strategy were: a) to ensure that that the value of more effective use of data resources in the US health
citizens experienced a collaborative public sector and b) to reduce the care sector alone could be USD 300 billion annually (McKinsey &
costs of service delivery (Agency for Digitization, 2011). An additional Company, 2011). Moreover, the value of improved use of data across
important element to the strategy was an increased focus on re-using Europe´s public sector is estimated to be around EUR 250 billion annu-
data across all levels of government. While Denmark enjoys a strong tra- ally (McKinsey & Company, 2011). The growing technical and semantic
dition of maintaining legislative public registers, there was a growing ability to access and merge disparate types of data, as well as the avail-
need for improvement in data quality and data management in order ability of sophisticated data analysis techniques and technologies, offers
for authorities and businesses to harvest the full potential of these the potential to vastly increase the overall value of data. It has been sug-
data, particularly across organizations (Agency for Digitization, 2012). gested that one of the key enablers for increased use of data is the li-
Previous analysis had uncovered problems with both gaps and redun- quidity and openness of data, in the sense that the data are coherent,
dancy in data sets, as well as the lack of a consistent policy for who shareable and published in machine-readable formats (McKinsey &
could utilize the data and for what purpose. Moreover, both public au- Company, 2013). The annual economic value potential of open data
thorities and private businesses were suffering from cumbersome pay- over seven key sectors is estimated by McKinsey and company as
ment procedures and legislation that stood in the way of reuse. These being over 3 trillion USD globally (McKinsey & Company, 2013). How-
ever, an increasing body of research has shown that while the potential
E-mail address: [email protected]. of data-driven value generation is vast, it is still very difficult for most
1
Twitter: @HildaJetzek; LinkedIn: dk.linkedin.com/in/thorhildur. organizations to generate and harness this value in practice. (Conradie

http://dx.doi.org/10.1016/j.giq.2015.11.003
0740-624X/© 2015 Elsevier Inc. All rights reserved.
90 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

& Choenni, 2014; Janssen, Charalabidis, & Zuiderwijk, 2012; Martin, proposed that open data infrastructure initiatives require a central gov-
Foulonneau, Turki, & Ihadjadene, 2014; Zuiderwijk, & Janssen, 2014a). ernance organization willing to provide various support activities
A proposition of this paper is that a lack of a common open data infra- (Hofman & Rajagopal, 2014), this proposition needs more supporting
structure might explain some of these practical problems. evidence.
This paper addresses the question of How can the tensions in a multi- This paper is intended as an input into this emerging body of re-
stakeholder open data infrastructure implementation be addressed through search. I suggest that by engaging in specific (context dependent) im-
governance strategies? This question is approached by examining the plementation strategies, the governance tensions associated with
Danish Basic Data Program (BDP), which was initiated in 2012. The orig- implementation of multi-stakeholder open data infrastructure can be
inal aim of the program was to increase public sector efficiency through alleviated. Moreover, I propose that if the data are to be effectively
better internal use of data. However, the program soon took on a leading and diversely utilized by multiple stakeholders, as opposed to a partic-
role in shaping the open data infrastructure in Denmark. ular group of users using data for specific purposes only; then an impor-
The paper is organized as follows: Firstly, the topic is motivated by tant phase of an open data infrastructure implementation must be
discussing previous research and evidentiary claims, which illustrate transformation of data. A transformation not solely from closed to open
the need for improved understanding of open data infrastructure data, but also from siloed to liquid data. A more detailed discussion of
implementations. Following this, the three different theoretical frame- what this entails is offered in Section 2.1.
works used to analyze the implementation of open data infrastructure
in Denmark are reviewed. Thereafter, the research approach is
discussed, followed by a description of the case and its context. The 2.1. Dimensions and affordances of liquid open data
main phases of the program are summarized in a process model,
which is the main theoretical contribution of the paper. Four gover- Most of the current open data initiatives are driven by governments.
nance tensions and four key implementation strategies are presented Innovators have been shown to value government data due to factors
in the Analysis and discussion section. The paper concludes by such as data quality, reliability of source, and comprehensiveness
discussing the foremost contributions to knowledge and practice, as (Lakomaa & Kallberg, 2013; Jetzek, Avital, & Bjørn-Andersen, 2014). In
well as the limitations of the research approach. well-resourced countries, few other institutions can provide such con-
sistent and all-encompassing data, making open government data a
2. Motivation and frameworks for analysis valuable input to economic activity (Davies, 2013).
From a supplier perspective, the transformation to open data is pro-
The original aim of the BDP was to implement a common data layer posed to be a function of availability, accessibility, format and license
for a limited number of key sets of public data, with the goal of im- (Davies, 2010). From a demand perspective, openness is suggested to
proved efficiency within the public sector (see further discussion in combine unrestricted availability of data with accessibility and technical
Section 4). The central premise was that by giving all users from public interoperability (Tammisto & Lindman, 2011). In practice oriented liter-
and private sectors access to the same data; double entry of data, as well ature, the term open data is interpreted in a variety of fashions, as evi-
as various shadow registers, would be eliminated, effectively raising the denced from the many different working definitions found online. The
quality of the data and stimulating its use. However, as the program Open Knowledge Foundation defines open data as “data that can be
progressed, it evolved into a more generic open data infrastructure pro- freely used, re-used and redistributed by anyone - subject only, at
gram, including a number of additional dimensions necessary for such most, to the requirement to attribute and sharealike” (Open
an infrastructure. Infrastructure in general can be defined as basic phys- Knowledge Foundation (OKF), 2015). However, this definition lacks
ical and organizational structures needed for the operation of a society reference to the technical dimensions of open data. Alternatively,
or enterprise, or the services and facilities necessary for an economy Berners-Lee´s five stars of linked data specify a number of technical
to function (Sullivan & Sheffrin, 2003). Accordingly, I define open data dimensions.2 However, the five stars are not really an open data defini-
infrastructure as the digital infrastructure needed to enable effective data tion, but rather a maturity model that focuses on how to gradually
sharing and use across boundaries. Implementing a data infrastructure transform data into linked data, which indicates a method of publishing
is a complex initiative that amongst other things requires a change in structured data so they can be interlinked and discovered through se-
a number of organizational and societal structures, which are often re- mantic queries.
sistant to change. This important non-technical side of digital infrastruc- An overview of multiple working definitions of open data is present-
ture has recently been conceptualized as soft infrastructure (King & ed in Verhulst, Noveck, Caplan, Brown, and Paz (2014). This overview
Uhlir, 2014). shows that the currently used definitions usually highlight 2-4 dimen-
As open data policies are a recent phenomenon, applicable system- sions that are important for openness of data, in many cases with a spe-
atic research that identifies different policies and outlines their impact cific purpose or ideology in mind. It also shows that no two definitions
has not emerged yet (Barry & Bannister, 2014; Hujiboom & Van den are based on exactly the same dimensions. The Open Data Barometer
Broek, 2011; Zuiderwijk & Janssen, 2014c). Moreover, current research defines “truly open” data as data that are available online, in bulk, and
is inadequate in illustrating the necessary actions for open data publish- under an explicit open license (Davies, 2013). However, in spite of the
ing after development of policy. The relationship between the context of fact that considerable efforts have been exercised to make diverse gov-
open data initiatives (technical, social, organizational and political), var- ernment data available to the public, less than one in ten public datasets
ious elements of open data and data platforms, and the impacts of such reviewed in seventy-seven countries in 2013 could be classified as truly
initiatives, needs to be clarified (Zuiderwijk & Janssen, 2014b). A litera- open according to the Open Data Barometer definition (Davies, 2013;
ture review revealed that information science literature offers insights Höchtl, Davies, Janssen, & Schieferdecker, 2014).
into specific areas such as data quality and use of metadata, and com- To add to this discussion, I propose a construct I call Liquid open data.
puter science literature offers a sizable corpus of research on linked This construct was synthesized from different working definitions of
data principles, data models and platforms (See for instance open data, with an added insight from a review of the open data litera-
Bountouri, Christos Soulikias, & Stratis, 2009; Höchtl & Reichstädter, ture, as well as the analysis of this case study data. The construct consists
2011). However, a void still remains in work that presents a holistic of five main dimensions that offer seven important affordances to po-
view of the implementation and governance challenges faced by organi- tential users. I propose that an open data infrastructure, which enables
zations in the act of publishing open data. Data infrastructure as a con- cross-boundary use of data by multiple stakeholders for a variety of
cept is in general not well represented in the literature, with the
exception of spatial data infrastructures. Moreover, while it has been 2
http://www.w3.org/DesignIssues/LinkedData.html
 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 91

purposes, will consist of all of seven dimensions of liquid open data. The Table 1
extra word liquid is intended to indicate the importance of the technical Five dimensions and seven affordances of liquid open data.

and conceptual dimensions that create the foundation for a technical Dimension Affordance Description
open data infrastructure. The difference between what I conceptualize Strategic Availability Data that are not subject to privacy or national
as openness of data on the one hand, and liquidity of data on the security limitations are open to all by default.
other, is illustrated in Fig. 1. Economic Affordability Data are free or charged for at maximum at
Thus, liquid open data are defined as data that are available online, marginal cost of reproduction.
Legal Reusability Data are published under open licenses.
free-of-charge and under an open access license, conceptually coherent,
Conceptual Interoperability Semantics and syntax of data are clear and data are
and published in useable formats, easily discoverable and accessible. Liquid presented with open data models, standardized
open data can be re-used without discrimination or limitation, linked to metadata and standard identifiers.
other data, and streamed across systems. Table 1 shows the most impor- Technical Usability Users can interact with the data with clarity and
tant dimensions and affordances of liquid open data. ease. Data need to be of high quality (accurate,
complete and timely) and presented using
standardized machine-readable data formats.
Discoverability Data are easily found through search engines or
2.2. Managing complexity — the system of systems approach under central portals or are published using linked
data semantics.
Decision makers within government and industry are currently fac- Accessibility Data are easily, consistently and securely accessible
and downloadable or streamable.
ing societal challenges of increasing reach and complexity. These chal-
lenges range from economic inequality, unemployment and poor
social conditions, to chronic diseases, and climate change. Societal prob-
lems such as these do not pay attention to human-constructed bound-
aries and commonly transcend the boundaries between the public and connecting systems from multiple domains, which cross organizational
the private sectors (Janssen & Estevez, 2013). Accordingly, challenges boundaries, and operate at different levels (Curry, 2012). Table 2 illus-
of this nature can only be effectively addressed by the coordinated ac- trates five characteristics that can be used to identify a SoS (Boardman
tions of many different public and private agencies, operating at several & Sauser, 2008):
different levels (Kettl, Pollit & Svara, 2004). The SoS approach strives to find a balance between allowing the nec-
However, with multiple heterogeneous distributed systems, effec- essary autonomy of specialized units, so that they can fulfill their legal
tive analysis for decision-support quickly becomes unmanageable with- and strategic purposes, and integration on various levels that is neces-
in the silo-context that characterizes many organizations (DeLaurentis sary for a holistic approach. While considered suitable for governing
& Callaway, 2004). These types of societal challenges demand systemat- complex systems, the SoS approach poses significant challenges, includ-
ic thinking at the basic level, creating the necessary infrastructure that ing the need for data and information interoperability, and collaboration
allow data and information to flow across organizational and technical and coordination between independent organizational units. In order to
systems. Systems-of-Systems (SoS) can be defined as a collaborative enable organizations and their systems to interoperate in smart govern-
set of systems where the components are independent, dedicated systems ment environments, we need to consider three levels of interoperabili-
that are separately acquired and integrated to form a single SoS, yet main- ty: technical, conceptual, and pragmatic or organizational
taining a continuous operational existence independent of the collaborative (Hjort-Madsen, 2006; Maheshwari & Janssen, 2014). Table 3 shows
system (Rechtin & Maier, 2009). Adopting a SoS approach allows for these three levels of interoperability.

Fig. 1. Liquid open data

92 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

Table 2 at the societal level, which is intended to explain the causal relation be-
Five characteristics of a system-of-systems. tween openness and liquidity of data on the one hand, and the genera-
Boardman & Sauser (2008).
tion of value on the other. This overarching mechanism is not driven by
Characteristic Description the laws of the markets, but rather by sharing of information over net-
Autonomy Describes the ability of individual systems to make independent works. I call this mechanism the information sharing mechanism and de-
choices while adhering to the common overall goal of the system. fine it as a mechanism that generates societal level value over a network
Belonging Reflected in a shared mission where systems are voluntary of stakeholders through the simultaneous creation, dissemination, and
partners in the overall system.
appropriation of openly shared information. Unlike the essentially bilat-
Connectivity Describes the ability to link with the other systems and thus the
need for interoperability between systems. eral market mechanism, this mechanism is driven by many-to-many in-
Diversity Describes the heterogeneous roles of individual systems. teractions over networks. For this mechanism, the size of the networks
Emergence Highlights the evolutionary development of a SoS. and the interactions themselves are primary sources of value (Bowman,
2015; Viscusi, Castelli, & Batini, 2014).
The framework illustrated in Fig. 2 highlights the two principal
2.3. The value generating mechanisms of open government data mechanisms that facilitate how value is generated through open data:
the information sharing mechanism and the market mechanism. The
Mechanisms are a key construct is many disciplines of science and framework also emphasizes that for each of those mechanisms, value
are defined as frequently occurring and easily recognizable causal pat- generation can happen either through exploitation of current resources,
terns (Elster, 2007). Mechanisms have four main characteristics: First, or through exploration focused on driving change. The first column in
a mechanism is identified by the type of effect or phenomenon it pro- the framework represents the information sharing mechanism. I pro-
duces. Second, a mechanism is an irreducibly causal notion. Third, a pose that value generation through open data can be mediated by infor-
mechanism has a structure, and fourth, mechanisms form a hierarchy mation sharing. In this case, value generation can happen either through
(Hedström & Ylikoski, 2010). In social theory research collective entities transparency or through civic engagement. Value is generated when ac-
and processes unfolding in a social context are understood to typify and cess to meaningful information (transparency) results in less informa-
represent the generation of aggregate effects from the actions of indi- tion asymmetry and better governance, thus reducing the risk of
viduals (Avgerou, 2013). Therefore, a methodological strategy to cope problems such as the principal-agent problem. Alternatively, value is
with theoretical complexity is the acceptance of mechanisms in terms generated when multiple stakeholders collectively contribute to a com-
of collective actors (Avgerou, 2013). At the societal level the market mon cause by sharing and enriching information across boundaries,
mechanism is probably one of the most widely recognized mechanisms thus driving change through civic engagement. The second column in
for value generation. The market mechanism describes how the forces the framework represents the market mechanism. The markets are an
of demand and supply determine relative prices of goods and services, important platform for value generation as they enable us to improve
which facilitates market-based transactions and ultimately determines resource allocation and reduce waste. Sharing and re-use of data can
the way our productive resources are allocated. be used to improve governance and contribute to more efficient process-
The most important driver behind open government data initiatives es and less waste. Alternatively, the markets can be used to diffuse new
largely stems from the expectation that an increased openness of gov- data-driven products and services, driving change through innovation,
ernment data will stimulate the generation of considerable social, eco- subsequently creating new markets and new value.
nomic and environmental value (Janssen et al., 2012; van Veenstra & In the context of this paper this framework is purely used to high-
van den Broek, 2013; Zuiderwijk, Janssen, Choenni, Meijer, & Alibaks, light the archetypical paths through which open data can contribute
2012). The explanatory mechanisms behind this value creation are, to value generation, and how they relate to different political agendas,
however, not clearly articulated in the open data literature. Moreover, driven by different types of government organizations. Further discus-
both the literature and my primary data suggest that market mecha- sion of the individual mechanisms can be found in Jetzek, Avital, and
nisms only explicate how a portion of that value is generated. Thus, I Bjørn-Andersen (2012); Jetzek, Avital, & Bjorn-Andersen, 2013a; Jetzek,
have conceptualized a new overarching value generating mechanism Avital, & Bjørn-Andersen, 2013b).

3. Research setting and approach

Table 3 This research is the result of four years involvement with a Danish
Three levels of interoperability.
open data infrastructure initiative, the Danish Basic Data Program
Level Three levels of Three levels of (BDP). This case study was conducted in the spirit of Engaged Scholar-
interoperability (Gottschalk, interoperability barriers ship (van de Ven, 2007). I was not a direct participant in the program,
2009) (Chen, Doumeingts, &
Vernadat, 2008)
but rather worked from a peripheral space with access granted to all
non-classified internal documents and project events. The purpose of
Technical Interoperability of systems Technological barriers refer
the research was to describe and explain and the resulting approach
which is concerned with the to the incompatibility of
ability of two or more information technologies can be categorized as “co-producing knowledge with collaborators.”
systems or components to such as protocols, encoding, (van de Ven, 2007, p. 27).
exchange data and to use the platforms, or infrastructures. The case study is suggested as a preferred method when (a) how or
data that have been why questions are asked, (b) the investigator has little control over
exchanged.
Conceptual Semantic interoperability Conceptual barriers represent
events, and (c) the focus is on a contemporary phenomenon within a
which has to do with use of the syntactic (format) and real-life context (Yin, 2009). This approach was chosen for the BDP as
common standards semantic differences events were unfolding throughout the case study period, being record-
(interpretation of meaning) ed and analyzed in real-time. Yin (2009) recommends that case study
of exchanged information.
inquiry rely on multiple sources of evidence (e.g. interview, observa-
Pragmatic Organizational Organizational barriers
(organizational) interoperability is concerned concern the organizational tions, documents and archival record) which should converge in a trian-
with the ability of two or incompatibility of definitions gulating fashion. Strong triangulation of data sources is important to
more units to exchange data of responsibility, authority, establish the necessary reliability and validity of a research study (Yin,
and services to operate more and organizational structures. 2009). The case study also benefits from the prior development of the-
effectively together.
oretical propositions to guide data collection and analysis (Yin, 2009),
 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 93

Fig. 2. Four archetypical value generating mechanisms that indicate different focus areas

and thus I used the frameworks presented in Sections 2.1–2.3 to orga- program leader who provided extensive information about the pro-
nize data collection and data analysis. gram, and its main challenges and opportunities.
The overarching research method is a longitudinal descriptive case The interviews were semi-structured, allowing new patterns and tra-
study, spanning a period of four years. A descriptive case study is used jectories to emerge, while still obtaining the necessary information for the
to outline and describe a phenomenon within the complex real-life con- research. As such, key-questions were prepared beforehand; in most
text in which it occurred (Yin, 2009). To reflect the progress over time I cases addressing the interviewee’s perception of the BDP, what they
chose to apply a process approach to the data analysis. Accordingly, the saw as the main implementation barriers, and what they perceived to
case is structured chronologically, where events are traced over time, be the program’s primary value proposition. In some cases, the interview
and ordered in periods or phases (van der Blonk, 2003). The main guid- questions focused on “historical” aspects to improve understanding of the
ing research question is: How can the tensions in a multi-stakeholder open social, economic, and political context in which this program operated.
data infrastructure implementation be addressed through governance During the data collection process extensive observational notes were
strategies? The next sections describe how data collection and analysis taken, both during interviews, while listening to interviews, and while ob-
were approached. serving events. Overview over dates, number of interviewees in each in-
terview, and their role and organization appears in Appendix A.
3.1. Data collection Additionally, selected project documents were examined, such as
publications and presentations from the organizations that participated
Empirical data was collected between November 2011 and August in the BDP. Other material used to inform the analysis were policy doc-
2015 through engagement with various stakeholders directly and indi- uments, tender documents and meeting notes, and lastly scientific re-
rectly involved in the Danish BDP. Primary data was collected through search papers on open data, interoperability and management of
interviews, participation in various project-related events, such as complex systems.
kick-off meetings for assorted project phases and conferences, as well
as observation in a private sector company, KMD. KMD participated in
the tender for a distribution platform for the basic data and later became 3.2. Data analysis
the primary vendor in the implementation of the technical
infrastructure. Interviews were recorded with participants' permission, although in
A total of 28 interviews with 22 separate interviewees were con- two cases the recording was lost and notes were relied on. In general,
ducted, ranging from just under half an hour to one and a half hours. the interviews provided a deeper understanding of important events, pro-
The main criteria in choosing interviewees was the identification of gram phases, decision points, and tensions related to the implementation
stakeholders that were directly involved in the program and who process. Documentation focused on the experiences and challenges related
were perceived as thought leaders by other participants, as these key in- to managing an open data infrastructure program, with a special emphasis
dividuals were best positioned to influence the program’s direction. The on understanding the project participants' perception of how these data
interviews were conducted face to face in the offices of the respondents, would contribute value to society. Attention was also paid to their percep-
and largely in the respondent’s native tongue of Danish. Key respon- tion of the program itself, the main barriers and enablers, the most
dents were interviewed multiple times to document the progress of dominant tensions, as well as the overall governance of the program.
the BDP during the case study period. The primary informant was the Fig. 3 illustrates the case timeline and time boundaries.
94 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

Data collection began with the publishing of the digitization strategy The third important emerging theme revolved around the gover-
for 2011–2015 in Q3 2011 and ended in Q3 2015. While the first formal nance structure of the program. The Danish BDP gained an overarching
interviews were conducted in August 2012, program events were commitment from seven different ministries in the central government,
attended in November 2011, and the related documentation of these as well as engagement from powerful interest groups at both the re-
events were made available. These data were later used to gain insight gional and municipal level. However, it soon emerged from the data
into the historical context from which this program arose. The context that the program leaders needed strike a fine balance between autono-
itself played a crucial role in explaining many of the events. The analytic my and control. The act of introducing interoperability between hetero-
approach was abductive. Accordingly, the theoretical frameworks used geneous sets of data, owned and used by different organizations under
for analysis, were developed alongside to provide an increasingly accu- different ministries or in different levels of government, called for a
rate documentation of the case. These frameworks or theoretical lenses high level of collaboration. At the same time, individual data registers
(reviewed in the theory Sections 2.1–2.3) were chosen in light of recur- had their own mandates and the program needed to make sure that
ring themes in the data. After the preliminary, ongoing analysis of the the professionalism that had identified each of them was
data during the case timeline, I proceeded in three stages. safeguarded. Moreover, the needs of many diverse users from both
During the first stage, the three biggest themes from the data were public and private sectors must be taken into account when
identified. The first important theme noted was the multidimensionali- implementing an open data infrastructure. Therefore an aim of this
ty and complexity of the open data construct itself, which led to paper is to convey the resulting governance approach that emerged
examining the individual dimensions of liquid open data. Without this from this complex initiative.
being explicitly expressed I perceived a drive to move from organiza- The second stage, after identifying the three themes, involved a
tional silos towards a modular information architecture, highlighting search for relevant frameworks or taxonomies for analysis. As open
the need for more agility and responsiveness in government processes. data is an emerging research field, offering few suitable conceptual
While not only an open data issue, one of the key ingredients in such frameworks, two of the chosen frameworks are a based on my own
architecture are data that are not bound to individual systems but work, while the third was identified from related literature. The first
can flow easily across systems to inform and automate as needed. framework is illustrated in Fig. 2. This framework served as a focusing
However, implementing an open data infrastructure that supports the lens in many of the interviews, and later guided the discovery of impor-
technical and conceptual dimensions of open data (liquidity of data) tant tensions in the program. The second framework is the taxonomy of
introduces more complexity and requires a longer implementation five dimensions and seven affordances of open data, as shown in
period. Table 1. These dimensions were applied relatively late in the analytic
The second theme was derived from the BDP focus on two market- process but are of importance as different project groups focused on dif-
based value generating mechanisms. The first mechanisms is the effi- ferent dimensions of liquid open data, which complicated governance of
ciency mechanism, which reduces waste when data are efficiently used the program further. The third framework applied to data analysis are
across levels of government, thus improving resource utilization. The the characteristics of SoS governance, as mentioned in Section 2.2; as
participants that focused on efficiency as the primary mechanism for the program leaders had themselves identified the program setup as a
value generation, tended to emphasize the technical and conceptual di- system-of-systems.
mensions of liquid open data. The second mechanism is the innovation The final stage involved analyzing the dynamics of the open data
mechanism that generates value through use of data as a resource for infrastructure implementation to develop the process model intro-
new information products and services, thus creating new jobs, new duced in Section 5. Based on all available documentary and interview
companies and even new markets. Participants advocating increased in- material a map with key events and decisions was created. I then
novation as the primary mechanism tended to focus more on the strate- proceeded to identify key issues or tensions and the case narrative
gic, economic, and legal dimensions of liquid open data. The participant´ was constructed (Langley, 1999). The focus during this stage was
s view of the importance of each of the mechanisms, and the subsequent on placing the items in the previously identified taxonomies in the
divergent focus that resulted was the root of a certain amount of tension model and presenting how these different themes were related to
in the program. each other.

Fig. 3. Case timeline.

 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 95

4. Case context and description level reference or master data, used widely for different purposes by
public and private sectors alike. The basic data chosen for first version
This section is intended to give an indication of the economic, social of the BDP were data from the person register, business register and
and political context of the BDP and the progression of events over time. real property register, as well as address data, place names, and maps
Denmark spends relatively more on social protection than the rest of the and geographic data. The common element of these basic data is that al-
EU countries, almost a third (25.2%) of yearly GDP in 2013 (Eurostat, most all other domain specific data within government have a reference
2013). Therein care for the elderly amounted to 14.3% of GDP. 23.6% of to some of these data (events happen in a place, and a person or an orga-
the population was over the age of 60 in 2012, representing a continu- nization is usually involved). The main goal was therefore to make sure
ing shift in demographics (Statistics Denmark, 2013). Administration that this fundamental layer of data was coherent, as this would allow for
costs in the social welfare sector in Denmark are also relatively high, greater consistency of data across many boundaries.
and general government sector employment (excluding public corpora- Accordingly, while the Danish society had gained many benefits
tions) as a percentage of the labor force was 26.8% in 2012, one of the from the basic data registers, the Danish authorities recognized that
highest in OECD (OECD, 2012). Similarly, compensation of public em- there remained opportunities for further benefits. This was reflected
ployees as a percentage of GDP was 19% in 2010 which was the highest in the Danish Digitization Strategy for 2011–2015, which was created
rate of all the OECD countries (OECD, 2012). There are worries that for the entirety of the public sector, including both local and central gov-
without major changes, public finances may reach an almost unsustain- ernment. “We need to be able to bring basic data into play with greater sta-
able level by 2020. These numbers can to a certain degree explain why bility and capacity than we can at the moment. This means finding an
the Danish Ministry of Finance has such a strong focus on improving ef- efficient way to distribute data across the public and private sectors using
ficiency in the public sector. In order to protect the welfare system and a shared infrastructure.” (Agency for Digitization, 2011, p. 38). The BDP
its high level of social security, the government needs to find a way to do was thus established to drive the data related objectives of the strategy.
more with less. The BDP was perceived to offer a particularly promising To ensure cross-boundary collaboration, the BDP included representa-
path to that end, as improved use of data across boundaries was expect- tives from each of the seven ministries, the Danish Regions and from
ed to result in more value for money, new insights and increased KL, an interest group and member authority of the Danish municipalities.
collaboration. The BDP´s main goal was to create an infrastructure that would en-
The next sections describe how the BDP evolved over four years of able more efficient use of basic data across administrations and sectors
observation. A process model of an open data infrastructure program (Horst, Bjerre, Lind, & Hvingel, 2014). Three main challenges were iden-
is thereafter presented in Section 5. tified: (1) While the data from the basic data registers were already
used across levels of governments, they were used differently as a result
4.1. Program identification: from strategy to action of different user needs: “The needs for property information are very dif-
ferent between user groups and that is reflected in how the data are stored
A number of important milestones on the road to more efficient and presented in each register” (interview, product owner, KMD, August
reuse of data in Denmark preceded the BDP. Already in January 2003, 2012). Moreover, the quality of the data was in some cases not sufficient
the rather small Ministry of Housing, Urban and Rural Affairs had for the intended use. Thus, a number of shadow registers with copies
started to freely disseminate data from the dwelling register, the munic- and extensions had emerged over the years. (2) As the registers had
ipal property register and the cadaster, as well as address data. The only all been founded separately during different time periods to fulfill cer-
requirement was that users must themselves pay a distribution fee to tain legal requirements for data registration, they were based on differ-
private sector data resellers (DECA, 2010). “It did not take many years ent, and in some cases outdated, technical platforms. As such it was very
after starting to work with the public data in around 1993 to realize that difficult to combine data from different registers. (3) A variety of differ-
these data could be used for many interesting things. And we felt it was a ent payment models and types of licenses further complicated and lim-
shame, you know, that the data weren´t used more” (interview, project ited the use of basic data by all types of users.
manager in the Ministry of Housing, Urban and Rural Affairs, January
2014). A follow up study on the costs and benefits of free of charge ad- 4.2. Program definition: towards the goose with the golden eggs
dress data was conducted in 2010. This study confirmed the success of
the project, estimating the benefits over a 6 year period as EUR 57 mil- According to the original project plan, the BDP definition phase was
lion while the total costs over the same period were estimated to have to take place in the period from October 2011 until May 2012. In January
been around EUR 2 million (DECA, 2010). However, the 2010 report 2012 the program leaders hosted a large event - the Basic Data Camp. In
noted that there were still problems of “…inconsistent address informa- a traditionally Danish round-table fashion, multiple stakeholders from
tion with resulting duplicate work, errors and uncertainty in other state and various public organizations came together and were given the oppor-
municipal administrative back-end systems.” (DECA, 2010, p. 8). tunity to discuss and collectively define the program. During this meet-
Another prior milestone was the 2010 Open Data Innovation Strate- ing, the tensions between the aim for more efficient public sector, and
gy Initiative (ODIS). The main goal of ODIS was to raise awareness of the the drive for increased innovation were already quite apparent. The pro-
potentials of public sector data, reflecting the renewed global interest in ponents of the innovation agenda wanted to set data free for all users,
open data following president Obama’s 2009 Open Data Directive. How- while the Ministry of Finance had no intention of doing so, as this
ever, while the US initiative was largely focused on the information would most likely require them to replace the income lost to the regis-
sharing mechanisms of transparency and civic engagement, these ters with money from common funds. “Within the program there was a
were not high on the agenda in the Denmark. The Danish authorities pressure building, mainly from the Danish Geodata Agency and the Compa-
were more concerned with the increased demand for public services ny Register, that we should also set the data free.” (interview, program
during the meager years following the Global Financial crises in 2008. leader, Agency for Digitization, November 2013).
“This transparency thing. We believed in it, but it was not going to open In February 2012, a delegation of thirty stakeholders in the BDP par-
any doors in Denmark. Because in Denmark we didn´t really have the feel- ticipated in the Danish–Dutch Key Registers Tour 2012. The objective of
ing that our government wasn´t transparent enough. We never really the visit was to discuss the main building blocks required to establish a
seemed to regard that as a serious issue for Denmark…” (interview, pro- so-called system of key registers (SoS). The Dutch strongly advised their
ject manager, Agency for Digitization, August 2012). Danish friends not to procrastinate the fundamental issues of financing
The Agency for Digitization had classified the information that public and governance of the program (de Vries & Pijpker, 2013). The Danish
authorities register about citizens, companies, property, buildings, ad- program leaders took this challenge head on, and postponed the pro-
dresses, and so forth, as Basic Data. Basic data are defined as core societal gram definition phase for four months. To the surprise of many
96 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

involved, the group announced in October 2012 that all data from the discourse is rooted in a market-like logic, where concepts such as effi-
included key registers would be freely available to anyone, for any pur- ciency and value-for-money in public sector institutions are central
pose, excluding the data in the person register (Digitaliseringsstyrelsen, building blocks (Henriksen, 2006). One aspect of this type of gover-
2012).3,4 nance is that it applies pressure on public sector organizations to sell
The focus of the BDP had suddenly shifted heavily towards open data to generate revenue. “The payment models were in a way contradic-
data, promoting the ideological agenda of innovation and growth. The ting the intentions of more reuse of data” (interview, project manager,
brochure ‘Good Basic Data for Everyone — A Driver for Growth and Effi- Ministry of Housing, Urban and Rural Affairs, January 2014). It was vir-
ciency’ was published in October 2012, and marked the end of the pro- tually impossible for individual organizations to open their own
gram definition phase. The publication stated the following five datasets under this paradigm without some kind of compensation
principles: from the Ministry of Finance. However, some of the BDP participants
consistently advocated the open and free public data agenda. “They con-
1) In order to ensure re-use of data and to prevent double registration tinued to push the agenda and then we turned it back upon them and said:
and shadow registers, map data, cadastral maps, central business “OK, tell us the costs. We don't want to join this from an ideological perspec-
register data, and company data will be financed by the government tive; we want to know the costs and you have to document the benefits”
and released to the public and the private sectors, as is already the (interview, program leader, Agency for Digitization, November 2013).
case with address and real property data [Strategic dimension of Liq- In order to make open and free data a reality, the groups responsible
uid open data]. By releasing this basic data, public authorities and pri- for each sub-program were required to write a business case where they
vate businesses alike will be able to use it freely [Economic dimension identified which public organizations would benefit from higher quality
of Liquid open data], for commercial as well as for non-commercial data and could thus make more efficient use of them. Afterwards, the
purposes provided, of course, such use is lawful [Legal dimension of program leaders effectively transferred money from these organizations
Liquid open data]. to the registers that were responsible for collecting and delivering data.
2) In order to enhance the quality and coherence of data [Technical and This involved a great deal of restructuring and the movement of respon-
conceptual dimensions of Liquid open data] the registers of map data, sibility and authority between organizations, across ministries, and
real property data, address data, as well as business registers, will be levels of government. As an indicator of the complexity of the task, the
expanded to include other necessary data. As a result, a number of BDP group needed to alter around ten different laws in order to make
existing registers will become redundant and therefore can be these changes possible. The Ministry of Finance taking leadership of
phased out. this project was considered one of the key success factors, according
3) In order to make it possible to link data, efforts will be made to en- to those interviewed. The interviewees also highlighted the important
sure that all data conforms to the same conceptual and technical re- role of the program leader in making these changes possible. He exhib-
quirements [Technical and conceptual dimensions of Liquid open data]. ited both political acumen and leadership in finding a way to make open
4) In order to improve the distribution of common public-sector data, a data acceptable to the Ministry of Finance.
common technical infrastructure is to be established providing for The business case for the entirety of the program was calculated up-
stable and efficient distribution of data; a data distributor [Technical front, and all monetary transfers were made to main stakeholders be-
dimension of Liquid open data]. fore the project’s commencement. The total investment was budgeted
5) In order to ensure efficient, effective and coordinated development as EUR 125 million (Krieger-Røyen, 2014). Around 20% of this amount
and use of basic data, a cross-institutional basic-data committee is was to be spent on the implementation of the data distribution platform
to be established [SoS governance]. (technical infrastructure). Around 40% was allocated for reimbursement
of lost income to participating registers.5 20% of the budget was to be
The Danes, in a very short time, had overtaken their colleagues in the spent on improving data quality, and around 20% was intended to
Netherlands, creating a program that would address the required ele- cover other project related costs. On the benefits side, the simplification
ments for an open data infrastructure, simultaneously ensuring efficient and modernization of the IT architecture, and the introduction of a co-
internal use of basic data, while opening these data up for external use. herent and interoperable data model was expected to reduce the need
for double entry and shadow registers and reduce administrative costs
4.3. Free data does not come for free overall. These benefits were estimated to cover around 10% of the im-
plementation costs. Moreover, reduced costs from data maintenance
One core issue with making government data open is the fact that was expected to cover about 10% of the initial costs. Finally, it was esti-
the beneficiaries of open data are seldom the same as those who render mated that efficiency gains from simplified, and more carefully consid-
the costs of collecting the data (Fioretti, 2011; Martin et al., 2014). Many ered work processes in the public sector would cover around 80% of the
of the organizations that collected basic data in Denmark were not only costs. After the year 2016, when the program was to be fully imple-
accustomed to selling these data to fund their data collection efforts (or mented, the net annual benefits to the public sector were estimated at
a part of them), but required by law to do so. As previously stated, the EUR 35 million while the annual benefits to the private sector were es-
Ministry of Finance had originally no plans to “spend money” on subsi- timated at EUR 65 million (Agency for Digitization, 2012).
dies to these organizations, and the financing of free basic data After having overcome this important hurdle, the morale in the pro-
remained an issue in the early phases of the program’s definition. More- gram was high and there was excitement that open and free basic data
over, the intended data quality improvements and the common infra- for use across public and private sectors would finally become a reality.
structure were very costly. “Money needs to be redistributed. It is “The fact that we managed to make data free really brought trust to the
expensive, and difficult, to make these changes.” (interview, subprogram leadership, showing that we were able to move large sums of money
manager, Agency for Digitization, November 2014). around and really restructure how data collection and dissemination was
Denmark had for many years followed new public management financed.” (interview, program leader, Agency for Digitization, March
governance policies that aim to modernize and render the public sector 2014).
more efficient (Janssen & Estevez, 2013). The new public management

3
The financing of the free data is discussed in Section 4.3.
4
Some of the data in the person register are subject to privacy limitations and the rel-
5
evant sub-program group is currently defining how these data can be disseminated via a http://www.slideshare.net/EUDataForum/edf2014-nicolas-lemcke-horst-
common data distribution platform to individual public organizations without ambassador-of-the-danish-basic-data-programme-agency-for-digitisation-ministry-of-
compromising privacy. finance-of-denmark-danish-basic-data.
 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 97

4.4. Managing complex systems — the system-of-systems approach subprograms. This was especially true for two of the subprograms that
were already waiting for the delivery of the technical infrastructure. In
The BDP implementation was officially launched in October 2012. the summer of 2013 it became evident that the time plan for the plat-
The program consisted of seven sub-programs, and by the end of form needed to be delayed for a period of one year. Amongst other is-
2013, these subprograms were managing a total of thirty-eight individ- sues, there was a growing realization that the security setup required
ual projects. “That is huge and that is vastly complicated. And of course to enable sensitive internal data from the person register to be dissem-
things go wrong. It is kind of like constant firefighting.” (interview, pro- inated via the same platform as other less sensitive and open data
gram leader, Agency for Digitization, November 2013). Each of the would be extremely complex. “What has collided is that we have some
sub-programs had their own steering committee, but a cross-program subprograms that go across all the others but then we have the ministries
project management organization called the Basic Data Secretariat [that are responsible for different subprograms] with their own finances
was allocated to take care of daily operations and cross-program coordi- and resources, responsible to their own laws, having ownership over their
nation. The secretariat did not have any decision power and acted rather own registers. And when something goes a little bit wrong in the programs
as a supporting organization. A board of directors consisting of high- that go across [the other subprograms], you have to stop and look and say
level officials from the largest stakeholder groups (agencies, ministries that while this part seems to have collapsed, the only thing we know is that
and interest organization) was responsible for the political agenda. Six we are held responsible for our own finances and resources. So this cross-
out of seven subprograms were responsible for re-organizing a specific over of responsibilities is hard.” (interview, project manager, Ministry of
subset of the overall data model while the seventh subprogram was Housing, Urban and Rural Affairs, January 2014).
assigned the task of designing the architecture of a new common tech- At this point it was decided to get an external review of the program to
nical platform for data distribution, as well as creating tender docu- find out what changes in governance could improve coordination be-
ments for the platform. During the first months of the program the tween individual subprograms. The review report suggested that project
groups worked more or less independently. management and coordination across subprograms needed to be
Morale remained high and participants stayed true to the original strengthened, which created the need for new competencies, and an in-
program vision [SoS characteristic Belonging] and held the conviction creased number of personnel devoted to program management. “It is
that the program had a sound foundation and the potential to deliver not possible to reach our goals with consensus. We must strengthen gover-
a great deal of value to society. As time passed, however, certain gover- nance, else we might end up with a program that doesn´t deliver.” (inter-
nance challenges started to appear. While Denmark has a long history of view, data ambassador, Agency for Digitization, September 2014). It was
collaborative government, the public sector is still largely decentralized decided to change the program organization (Fig. 4) in order to ensure
and public authorities experience a good deal of autonomy [SoS charac- clear and uniform decision channels and well-defined roles and responsi-
teristic Autonomy] (Kettl, Pollit & Svara, 2004). This type of structure bilities [SoS characteristic Emergence]. “Even though we have been doing
calls for a multi-level, multi-authority co-ordination and networking what we ought to think of as programs, we have not had the tools or the un-
[SoS characteristic Connectivity] which demands specific skills from derstanding of how to control a program of projects. So we are developing this
leaders (Kettl, Pollit & Svara, 2004). In the case of the BDP, individual understanding and these tools now and the BDP has actually been a pilot pro-
subprograms and projects under the overall program management gram used by our Agency to develop these tools to use across government.”
depended on each other; if one subprogram did not deliver on time, an- (interview, program leader, Agency for Digitization, April 2013).
other subprogram could endure delays. As the basic data secretariat was As each subprogram has their own budget, steering committee and
given authority only to coordinate, it lacked the ability to handle issues goals [SoS Feature Diversity], the overall program management devel-
that arose across subprograms. The result was the referral of these cross- oped into a structured informal coordination, where political, and
boundary technical issues to the board of directors for resolution. other cross boundary issues were discussed. By strengthening this coor-
In early 2013, delays in the seventh subprogram (design of the tech- dination through different stakeholder engagement forums, and
nical architecture) had started to cause inconveniences for the other through standardized procedures and deliveries, the program leaders

Fig. 4. New organizational diagram of the BDP.

98 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

addressed the governance related problems that had threatened to tool that ensures conceptual interoperability. The data model is also
upset the strongly collaborative atmosphere of the program. The new openly available which makes it easier for third party users to under-
organizational diagram is shown in Fig. 4. stand and work with the data. Besides the organizational challenges,
implementation of the conceptual dimension demands data-modeling
4.5. Implementing the open data infrastructure skills that are unfortunately in short supply in both public and private
sectors. “What we have as a goal is to create a complete data model with
An increased awareness of the potential value of open data was ev- a complete table of contents with a list and a diagram that explains what
ident over the case study period. This awareness created a pressure all these data elements are, and how to they relate to each other.”… “This
from both public and private sectors to increase the availability of knowledge lies on the boundary between technology and business.” (inter-
open data. Among other initiatives, a collective of various industry orga- view, project manager, Agency for Digitization, November 2014).
nizations created a formal document with a call for more availability of The design and implementation of the technical distribution plat-
open government data.6 However, due to a shortage of resources and form has perhaps been the single most challenging aspect of the BDP.
skills, the BDP did not have the means to include more types of data in The platform is key in a common data infrastructure as it provides the
the program. “…things are difficult now. I think we have uncovered a foundation on which many of the technical dimensions rely. To ensure
new set of problems. Some capabilities are lacking in our national [IT] infra- usability, all data distributed via this platform will be machine readable,
structure architecture and this lack is creating big problems for some of the provided in open standard formats and of improved quality, both in
subprograms.” (interview, program leader, Agency for Digitization, No- terms of timeliness, completeness and accuracy. In regards to accessibil-
vember 2013). ity, the platform guarantees (based on the tender material) high uptime
In January 2013, the economic and legal dimensions of liquid open (99%) and stability of service, as well as open API’s, web services and
data (affordability and re-usability) were successfully implemented for bulk-downloads. Finally, discoverability will also be improved, as the
geographic data (Geodata Agency) and data from the business register data will be available through a single point of access, making use of
(Danish Business Authority). Making these data more affordable has re- standardized metadata.
sulted in exponential growth in data downloads. Interestingly, despite
the fact that the agencies responsible for these datasets were mostly in-
terested in the use of them for external innovation purposes, a lot of in- 4.6. User engagement, use of data and realization of benefits
ternal barriers were actually removed in the process of making data free
for all, making it easier for authorities to use each other´s data and there- The BDP offers the opportunity for public and private organizations
by creating internal process efficiencies. This showcases the potential of to use high quality, open government data across boundaries. The Dan-
open data to generate synergies between the different value generating ish private sector is already a major user, and appreciator, of these au-
mechanisms featured in Fig. 2. “It is crystal clear, that it is a huge feat of thoritative data, as can be seen in a recent publication from some of
the [Basic Data] Program to get the authorities to work together. And the the largest industry organizations in Denmark.7 However, there is still
key to that is that there are no longer economic barriers between us.” (in- work to be done regarding presentation of the data: “The private sector
terview, subprogram leader, Danish Business Authority, November is already today closely linked to the public sector´s systems in many
2014). areas. Continuous access to public data is thus a critical prerequisite for a
Prior to the BDP, the municipalities could not publish their own geo- wide range of private actors. As public sector systems are developed, stan-
graphic data (for instance maps) openly on their websites. In fact there dards are amended and interfaces are continuously adjusted. Viewed in iso-
was a court order prohibiting one municipality from openly disseminat- lation, this occurs for good reasons, but it can be challenging [for the private
ing their own geographic data, as by doing this the municipality was en- sector users], as even small changes can have serious consequences, causing
gaged in unlawful competition with commercial data providers, who [private sector] systems using these data to stop.”
had purchased the geographic data and generated income by re- The BDP has taken steps to enhance this relationship with the pri-
selling them. Due to this legal barrier, many municipalities had chosen vate sector. Relatively early in the program, a data ambassador was
to use available geographic data from open sources such as the Open hired. His responsibility was to serve as a communication channel be-
Street Map (OSM), rather than the authoritative data from the Geodata tween the BDP and potential private sector users, ensuring that relevant
Agency. To overcome this barrier, the BDP changed the law, stating that information was shared bi-directionally. In seems that the establish-
from January 1st 2013 the municipalities could freely and openly dis- ment of this role has succeeded in creating more awareness and engage-
seminate the geographic data. The three major data releases (the geo- ment. The data ambassador created a group for the program on the
graphic, address and business register data) of the BDP all use a social media site LinkedIn which has 921 members and active participa-
common standardized license based on the creative Commons- tion. As an example, a status announcing the publication of the new data
licenses CC0 and CC-BY, for improved reusability. model got 54 likes, many of which came from private sector users of
The conceptual dimension of liquid open data focuses on semantic data who are closely following the development of the BDG. The
and syntactic interoperability of data and is of great importance to the group is used for discussing other sets of open data as well. For example,
BDP as interoperability is a key to many of the project´s projected effi- when Copenhagen Municipality announced they were opening the city´
ciency gains. However, interoperability is challenging to achieve be- s 3D-model data they received 312 likes, largely from private sector
cause of the need for coordination and centralized standardization users. In the BDP governance restructuring of 2014, the role of data am-
efforts and therefore a new subprogram was created in late 2014 to ad- bassador was expanded to include a collaboration forum between pri-
dress this. Subprogram number eight was given responsibility for de- vate sector representatives from chosen sectors representing large
signing the overall data architecture, including common data users of public data (finance, telecommunications and utilities) and pro-
modeling rules. To strengthen collaboration between the eighth subpro- gram participants. Unfortunately, this forum has not been very active
gram and the other seven, an architectural forum was created at the thus far. While the willingness for cross-sector collaboration exhibited
program organization and coordination level. A general set of data through open forums and open documentation seems to be appreciated
modeling rules is currently being developed. These rules standardize by private sector stakeholders, there is not much evidence for the exis-
how individual registers are to model their data and are perceived as a tence of formal engagement mechanisms.

6 7
http://www.fsr.dk/~/media/Files/Presse%20og%20nyheder/Nyheder/2015/ http://www.fsr.dk/~/media/Files/Presse%20og%20nyheder/Nyheder/2015/
Initiativkatalog%20til%20den%20f%C3%A6llesoffentlige%20digitaliseringsstrategi- Initiativkatalog%20til%20den%20f%C3%A6llesoffentlige%20digitaliseringsstrategi-
final%20Copy.ashx. final%20Copy.ashx.
 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 99

As discussed earlier, the Ministry of Finance´s vision for the BDP was survey in 2016 and compare the results to the baseline. The agency
first and foremost to increase efficient use of a fixed set of basic (master wants to measure only the effect of free geographic data, correcting for
or reference) data within the public sector. Because of this agenda of general market trends that have moved towards increased use of geo-
greater efficiency, the BDP´s members needed to justify their expendi- graphic data for a number of years. This is, however, difficult to achieve
ture on opening data through the creation of business cases. However, beyond the first layer of users, as the open licenses allow for all kinds of
despite the initial success of creating and promoting the overall business redistribution and reuse. Despite mandatory user registration, the ad-
case, documenting realized the benefits remains a challenge. As the vent of open data has impacted the direct contact the agency has with
complexity of the chosen approach was not fully realized from the users. “Here in 2012 we had 800 users, we knew all the users…. We
start, the program´s resources came under increasing pressure as the knew them pretty well because we were sending them a bill.” “[The situa-
implementation progressed. “We are facing a shortage of key resources tion before January 1st is] exemplified with the bonsai tree. But in 2016
and competences.” (interview, program leader, Agency for Digitization, we expect it to look more like a rainforest…This is going to be really difficult
April 2013). Due to these pressures and resulting delays, the realiza- to measure.” (interview, department manager, Geodata Agency, January
tion of the benefits that were the main pillars of the business case, 2014). However, what the Geodata Agency does already know is that
such as closing down shadow registers and improving process effi- the total number of users of geographic data increased tenfold in the
ciency, were compromised. This creates a potential issue for the pub- first year after the data were set free, and the growth of external users
lic sector beneficiaries who have participated in financing the continues.
program but will not realize the benefits until at a later date. Despite
the delays, an attempt has been made to improve how each subpro-
gram documents realized benefits in order to measure if efficiency 5. A process model of open data supply and value generation
goals have been met. A standard template dictates how qualitative
and quantitative benefits should be documented. This bottom-up ap- The process model illustrated in Fig. 5 has been developed to sum-
proach makes it much easier for the program leaders to keep track of marize the main phases and mechanisms identified from the data as
benefits that otherwise would be very difficult to identify. Moreover, well as to integrate prior theoretical frameworks. This model explains
this standardization of benefit calculation method implies a prag- the progress from the creation of an open data infrastructure strategy,
matic interoperability coordination mechanism. towards implementation, use of data and the resulting impact, concep-
Alternatively to the Agency of Digitization´s focus on internal gains tualized as four distinct phases. Additionally, different classes of under-
from open data, the Business Authority and Geodata Agency have fo- lying mechanisms are identified in the orange boxes. While the
cused more on external use of data and the innovation value generating progression from an open data infrastructure strategy to impact mea-
mechanism. The Geodata Agency, which is the owner of nautical, topo- surement is a complex process, including many different stakeholders
graphic and cadastral data, created a baseline measurement for use of and recursive elements, I illustrate it as a linear process with a cyclical
geographic data before it was made available free of charge under an feed-back loop and omit stakeholders from the model, for the benefit
open license January 1st 2013. Their ambition is to preform another of conceptual clarity.

Fig. 5. Process model of data-driven value generation.

100 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

The first phase of the process model is focused on strategy and plan- Table 4
ning — identifying the open data infrastructure program through Implementation of liquid open data in the BDP case.

(open) data strategy and (open) data policies, followed by a definition Dimension Focus of the BDP Status
and planning phase for the program itself. The second phase describes Strategic: The BDP only focuses on the The program definition phase
the implementation, focusing on transforming data across various di- Availability few but important datasets included a decision over
mensions of liquid open data: Making data open in the legal and eco- classified as basic data. Thus I which data should be
nomic sense, cleaning up data and improving data quality and conclude that during the last included in the BDP. However,
few years, Denmark has had there is ongoing discussion
coherence, and finally implementing a technical infrastructure for dis-
less focus on general avail- regarding what additional
semination of data. For successful completion of these two phases, pro- ability of central government datasets should be classified
gram governance must facilitate collaboration across boundaries for data than other countries as basic data in a potential
multi-stakeholder involvement. Different governance strategies for ad- where it has been declared BDP 2.0, as well as how to
dressing the four most important tensions in the BDP are discussed in- that all non-person identifi- integrate other more domain
able government data should specific government data to
depth in Section 6. be open by default. the basic data models.
The third phase represents the actual use of the data for various pur- Economic: Section 4.3 describes how the The funding model was
poses and the required engagement mechanisms. With the opening of affordability BDP managed to make the developed in the program
data, its use moves beyond control of the data custodians (Davies, basic data free of charge definition phase which
despite financial austerity in removed “economic barriers”
2010). Accordingly, data custodians need to engage with the data
the public sector. Key factor isand greatly simplified internal
users if they want to understand how the data are being utilized, and the funding model and the data sharing as well as
gain the insight that allows them to measure impact and improve the success in showing a positive stimulating external use.
implementation of various dimensions of liquid open data. business case. However, some existing data
Finally, the fourth phase represents the impacts of data use. The im- intermediaries objected to
this (disruptive)
pacts include improved public-sector efficiency and/or effectiveness,
development.
which will drive more value for money and improved level of service Legal: reusability A common license was The license was developed in
in the public sector. Alternatively, an impact from releasing liquid developed based on the the program definition phase.
open data can be data-driven innovation, some of which will hopefully creative Commons-licenses Change of the licensing
CC0 and CC-BY. However, this structure required changes in
address difficult societal challenges. The implementation of an open
license is currently only laws but removed barriers to
data infrastructure can result in increased transparency which improves applied to data that are part of reuse of data across the public
the “vision” of both the public sector itself and citizens, and thus reduces this particular program. sector as well as for private
information asymmetry. Finally, using open data to reach and activate sector.
more citizens can result in a higher level of civic engagement and Technical: The BDP has had a very strong The need for improved quality
usability focus on the usability of data and standardized
more informed decision making.
dimension, especially on representation was identified
To support the implementation of an open data infrastructure and improving data quality. The already in the program
the subsequent use of the data for value generation purposes, society focus on data quality was identification phase.
needs to possess certain IT capabilities. Most countries are currently fac- strongest in the address- and Improving data quality is
property data subprograms, perceived as one of the key
ing a shortage of skills in the area of data management, data modeling
potentially as these data had enablers for more efficient use
and data manipulation, analytics and visualization. Denmark is no ex- already been made available of data.
ception from these trends, as was evident from the case data. Moreover, free of charge prior to the
allowing multiple stakeholders to stream large amounts of data across BDP.
boundaries calls for a high level of network connectivity. In fact, global Technical: The new data modeling rules This dimension has evolved
discoverability are utilizing standards which through the implementation
IP traffic has increased fivefold over the past five years, and is expected
make it possible to apply phase alongside the
to increase threefold over the next five years. Overall, IP traffic is expect- linked data principles later on. development of the platform
ed to grow at a compound annual growth rate of 23 percent from 2014 The data distributing platform architecture and data
to 2019.8 The business environment needs to be supportive for entre- will provide one-stop access modeling rules. It is expected
to the basic data and use of that discoverability of basic
preneurs, and new types of business models are needed to drive value
metadata makes web search data will be greatly improved
generation from liquid open data. Finally, society needs political support easier. when the technical platform
for dissemination and use of data, and the regulatory infrastructure is fully implemented.
must not only support the dissemination of open data, but also offer Technical: A strong focus on accessibility The tender process for the
clear rules and regulations around protection of individual privacy and accessibility through a high-performance data distributing platform
platform was evident in the influenced this dimension.
use of personal data.
groups that focus on efficient, The first datasets (Geodata)
internal use of basic data. The will be published through the
groups that focus more on platform in fall 2015.
6. Analysis and discussion innovation seem to be
somewhat less interested in
“We are reworking our entire data infrastructure from the bottom up. this dimension.
Conceptual: According to the five stars of The modeling rules are still
And nobody else is doing that.” (interview program leader,
interoperability basic data,a basic data are under development and not
Digitaliseringsstyrelsen, November 2013). semantically coherent and all of the data have been
Because of the BDP, Denmark now has a data infrastructure model modelled accordingly to the modelled accordingly. The
that defines how basic data are organized, how different sets of basic basic data modeling rules. The current version is 1.0.0 but
data are interconnected, as well as the political and legal infrastructure design of these rules has for version 1.1.0 is under
the most part been driven by development.a
that guides reuse of data (Horst et al., 2014). I conclude that the BDP is in the central program
leadership function, while all
of the subprograms recognize
the importance of coherent
basic data.
a
http://www.digst.dk/Loesninger-og-infrastruktur/Grunddata/Delprogrammerne/
8
http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual- Faelles-datamodel/Hvad-er-Grunddatadatamodellen.
networking-index-vni/VNI_Hyperconnectivity_WP.html
 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 101

fact implementing all of the five dimensions of liquid open data, includ- stagnating or sluggish income since the financial crisis of 2008. While
ing both the technical and the “softer” dimensions, but to a varying de- the need for financial austerity is a fact for most public sectors, the im-
gree. Table 4 shows a summary of the results. minent danger is that the structural changes necessary for stimulating
In an infrastructure project of this range and complexity there are use of open government data will not happen. Too little investment in
bound to be challenges. The interviews and observations revealed a ten- infrastructure will limit the growth potential of open data, which
sion between the ideologies of efficiency and innovation. Efficiency will again negatively impact the willingness to invest in data
gains happen mainly through better internal use of data in government infrastructure.
information systems. This type of use demands quite an advanced high- The BDPs strong focus on internal gains, which are predicted to cover
performance, high-uptime technical solution that provides internal sys- the required investments with a surplus, has to a certain degree allevi-
tems with real time data of high quality and coherence. However, for ated this chicken-and-egg type of paradox. Accordingly, the first imple-
those purely focusing on stimulating innovation, it would be sufficient mentation strategy identified is to guarantee long-term funding upfront
to provide data updated daily (at most), through APIs or web services. with a transparent funding model that has been agreed upon by all
“The fundamental idea behind the BDP, to create authoritative registers stakeholders. Not only will this create short-term gains while enabling
from which data can be used by everyone, both authorities to use in their the program members to make the necessary long-term changes to
processes but of course also the private sector, is a good one. To set data data collection and publication, it also removes uncertainties over the
free has also been a huge success, made it possible to freely share data sustainability of open data initiatives that often times limits third
where there were previously thousands of barriers, both internally between party use of open data.
public authorities and between authorities and businesses. So two very
good principles there.” “It is also very, very important when you want to 6.2. Complexity: depth vs. comprehensiveness
link information across [registers] to have a common data model. Great.”
“However, there are many unknowns and risks in this data distributing Every open data initiative has to balance their own ambitions for
platform concept.” (interview, program leader, Danish Business Author- open data with the level of funding they receive. While most concur
ity, November 2014). that dissemination of high quality liquid open data is a good idea, it is
One argument is that the strong focus on the technical and concep- very difficult to achieve in reality. The approach chosen by the BDP
tual dimensions of liquid open data has caused the delays in the pro- was to focus on a limited number of key datasets, which can be
gram as these dimensions introduce too large a degree of complexity. linked to most other types of data, and to develop in-depth data
This level of conceptual and technical interoperability called for a modeling principles for these. The data modeling principles are
much higher level of organizational alignment than would have other- based on EU Inspire standards for geospatial data, to ensure compat-
wise been needed, and introduced multiple governance challenges. ibility with data from other countries whenever possible. Moreover,
Some stakeholders worry that the complexity and the resulting delays the rules for each basic data unit are supposed to be general enough
have been detrimental to the use of the data for innovation purposes. to be reused by other public sector organizations in their efforts to
Those that follow the innovation agenda would prefer a faster conver- publish liquid open data. This allows for continual improvements
sion from planning to actual use of data. External users have also to interoperability, without including an overwhelming amount of
expressed their wish for more availability of open data. However, data sets during the first round of modeling and publishing.
since January 1st 2013, use of open data for entrepreneurial activities Accordingly, the second suggested implementation strategy is to
has increased in Denmark. Based on quantitative data from the Open start small, preferably with data that can be identified as master or ref-
data Barometer as well as my own work, I estimate that entrepreneurial erence data, and use these to develop data modeling rules based on
use of open data in Denmark has increased from 5 to 7 on a scale of 10 open standards. Most other data can be linked to such key datasets as
between 2013 and 2015. events or activities must happen somewhere (geolocation, address,
To answer the research question: How can the tensions in a multi- property), usually with some stakeholders involved (people and/or
stakeholder open data infrastructure implementation be addressed through organizations).
governance strategies? I have identified four tensions in the BDP and
offer my perspective on how they were addressed in the program. 6.3. Governance structure: autonomy vs. control
Such tensions might be inevitable in a complex infrastructure program
but are in many cases not apparent at first glance — and might first ap- The BDP can be seen as indicative of an ideological transformation
pear after years of observation and analysis. Thus, I propose that there is of public governance in Denmark, away from new public manage-
significant practical value in having an early understanding of these po- ment towards what has been termed lean government (Henriksen,
tential tensions, as well as the suggested governance strategies, al- 2006; Janssen & Estevez, 2013). Lean government is characterized
though these will need to be considered within each initiative´s by the need to address societal problems without increasing the
individual context. size of the public sector, thus requiring collaboration across bound-
aries (Janssen & Estevez, 2013). Interoperability is the foundation
6.1. Duration: short term horizons vs. long term investment for such collaboration, including a common understanding of the
basic elements of information that enable seamless functioning of
Although it has been implied by open data enthusiasts that govern- fundamental societal services. It has been also been proposed that
ments should only worry about setting data free and let the market take information services are more likely to achieve success if they are de-
care of the rest, the experiences so far have shown that the reality for signed toward autonomy (modular design) rather than authoritarian
government organizations is much more complicated (Zuiderwijk & top-down control, because of the ability to adapt more quickly in re-
Janssen, 2014b). Both public and private users want high quality, reliable, sponse to dynamic changes (Montealegre, Hovorka, & Germonprez,
liquid open data that they can trust will continue to be delivered without 2014).
major changes, irrespective of political landscapes and external trends. Open government data do not originate from a single organiza-
If stakeholders do not trust the quality and sustainability of the current- tion and therefore can be difficult to publish in a coherent manner.
ly provided data, they will hesitate to create mission critical products, However, while each data custodian is collecting data for their own
services or processes that depend on these data. This has created a (regulatory) purposes, these data can undeniably be of much use to
‘chicken-and-egg’ type of problem for many open government data ini- other organizations. The data custodians must have enough autono-
tiatives, which face the reality of scarce resources simultaneously to an my to fulfill their individual roles, while contributing towards a com-
ever-increasing demand for public services, all balanced against mon goal of liquid open data, which means using a set of common
102 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

standards. In order to achieve a balance between those two compet- As a practice oriented contribution, I propose that the liquid open
ing demands, the BDP chose a governance approach that seems to fit data construct, the framework of four value generating mechanisms,
the characteristics of a system-of-systems (SoS) governance. This and the process model can be used as tools for governments interested
style of governance is well suited for coordinating between loosely in understanding how different dimensions of open data influence use,
coupled participants, but demands a high level of network gover- its subsequent impact, and any possible value generation. Moreover, I
nance skills (Provan & Kenis, 2008). propose that due to the features of liquid open data, we must adopt a
Accordingly, the third implementation strategy offered concerns multi-stakeholder perspective for all open data infrastructure
the adoption of a SoS approach, which has helped the BDP to steer implementations. Such multi-stakeholder constellations have the char-
away from the inherent silo structures that are characteristic of acteristics of a SoS and should be governed accordingly. Finally, the four
many governmental hierarchies, and achieve coordination without identified tensions, and the governance strategies employed by the BDP
losing autonomy. group, might prove useful for leaders of other data infrastructure
initiatives.
A limitation of the case study approach in general is that each case of
6.4. Ambidexterity: exploration vs. exploitation
open data implementation is highly dependent on a number of contex-
tual conditions. I have endeavored to report on the context of this par-
Sailing unchartered waters always brings challenges and unsur-
ticular case to make it easier to compare and contrast to other similar
prisingly, the BDP has met its share. The main disparity within the
initiatives. As such, an effort has been made to include contextual ele-
program could largely be traced to the divergent view of the most
ments in the process model. It is important to state that key strategies
important impacts held by the stakeholders. However, where there
are highly dependent on the specific context and goals of such an
are tensions, synergies are also possible. The BDP case study indi-
initiative.
cates that the tensions between different focus areas and related
I posit that the Danish case can nonetheless provide considerable in-
value generating mechanisms and impacts, might be resolved by
sight into the characteristics of open data infrastructure initiatives, as
making data liquid and open over as many of the dimensions as
well as an important glance into the possible future of government
possible.
data management. The BDP members are laying the foundation for a
Data that are open and liquid over the conceptual and technical di-
significant change in public sector data management. Right from the
mensions are akin to infrastructural resources. Infrastructural resources
start, there has been strong motivation and belief in a common cause,
are defined as shared means to many ends, which satisfy the following
which has survived over the course of the program, despite multiple
three criteria: 1) they are non-rivalrous, 2) social demand is driven pri-
challenges. The overarching belief is that when data become liquid
marily by downstream productive activities, 3) the resource can be used
and open they offer the opportunity to generate a great deal of social
as an input for a wide range of purposes (general purpose criteria)
and economic value for public and private sectors alike. I propose that
(Frischmann, 2012). Additionally, the strategic, legal, and economic
this motivational factor has played a large role in the program’s success,
dimensions of liquid open data imply that open data also have the fea-
shifting the focus towards the eventual value generation, an important
tures of a public good. The BDP members have experienced unexpected
focus area that has been suggested to be lacking in other such initiatives
synergies between different value generating mechanisms. These syn-
(Zuiderwijk and Janssen, 2014b). Finally, my estimate is that true to
ergies are most likely a result of the unique features of liquid open
their Danish origin, the BDP members have taken the first steps towards
data, as economic theory has long recognized the potential for positive
a future where individual data elements function as Lego bricks that can
externalities offered by infrastructures and public goods. This indicates
be put together in plethora of ways, offering the potential for creating
that there is indeed potential for serendipitous value generation if
modular, adaptive and agile systems. However, only time can tell if
data become open and liquid, while at the same time this finding em-
this vision will indeed become reality for the open data infrastructure
phasizes the impact measurement problem that governments must
in Denmark.
address.
Thorhildur Jetzek holds a M.Sc. in Economics from University of
Thus the fourth, and final, implementation strategy is to focus
Iceland and a Ph.D. in Information Technology Management from Co-
on all of the dimensions of liquid open data, as these are consid-
penhagen Business School. She is currently an postdoctoral researcher
ered to be an infrastructural resource with public good features,
at the Department of IT Management, Copenhagen Business School. Pre-
capable of providing synergies between different value generat-
viously, Thorhildur was a director of Business Development for the Nor-
ing mechanisms and stimulating unexpected future value
dic IT company Advania Norden. Her current research focuses mainly on
generation.
how societies and organizations can facilitate value generation through
open and big data. Throughout her career, Thorhildur has strived to
7. Contributions, limitations and conclusions understand, explain and improve how public and private stakeholders
use information and information technologies to generate value for
This study provides two main contributions to theory. The first society.
contribution is the multi-dimensional framework for an infrastruc-
tural construct called liquid open data. The development of the con-
struct occurred largely through a synthesis of the rich data of the BDP Acknowledgements
group´s experiences, the extant literature and secondary data
sources. The construct is intended as a contribution to conceptual I would like to thank the reviewers and senior editor for their in-
clarity in an emerging area of research. Moreover, my hope is that sightful comments and constructive feedback. My sincerest appreci-
the construct will make it easier to communicate the important fea- ation goes to all the members of the Basic Data Program for providing
tures of liquid open data, and improve understanding of how we can me with valuable insight into their important initiative. Special
stimulate value generation through open data. The second contribu- thanks go to program leader Jens Krieger-Røyen for his never-
tion of this longitudinal case study is a process model illustrating the ending support and for providing access to all the documents and
main phases of an open data infrastructure implementation. The events. I also want thank all the other informants I had the pleasure
model is intended to show the interplay between different types of of meeting during these four years of research, especially my
mechanisms (highlighted in orange in Fig. 5) and different events colleagues at KMD for sharing their knowledge and expertise on
that occur over the course of an open data infrastructure program, the subject matter. For providing financial support, I would like to
and beyond. thank KMD and Innovations Fonden.
 T. Jetzek / Government Information Quarterly 33 (2016) 89–104 103

Appendix A. Overview over interviews Avgerou, C. (2013). Explaining trust in IT-mediated elections: A case study of e-voting in
Brazil. Journal of the Association for Information Systems, 14(8), 399–419.
Barry, E., & Bannister, F. (2014). Barriers to open data release: A view from the top.
Information Polity, 19(1), 129–152.
Boardman, J., & Sauser, B. (2008). Systems thinking: Coping with 21st century problems. CRC
Press.
Dates # # Role Organizations Time Bountouri, L. P., Christos Soulikias, V., & Stratis, M. (2009). Metadata interoperability in
interviewees interviews public sector information. Journal of Information Science, 35(2), 204–231.
Bowman, C. (2015). The role of technology in the creation and capture of value.
August 2012 2 1 Program Agency for 1:15:00
Technology Analysis & Strategic Management, 27(2), 237–248.
manager Digitization Chen, D., Doumeingts, G., & Vernadat, F. (2008). Architectures for enterprise integra-
Program tion and interoperability: Past, present and future. Computers in Industry, 59(7),
leader 647–659.
April 2013 1 4 Program Agency for 0:44:22 Conradie, P., & Choenni, S. (2014). On the barriers for local governments releasing open
November manager Digitization 0:58:48 data. Government Information Quarterly, 31(1), S10–S17.
2013 Program 0:29:20 Curry, E. (2012). System of systems interoperability using a linked dataspace. In the Pro-
July 2014 leader 0:36:15 ceedings of the 7th International Conference on System of Systems Engineering
January (SoSE), Genoa, Italy, 101-106.
2015 Davies, T. (2010). Open data, democracy and public sector reform: A look at OGD use from
August 2012 1 1 Product KMD 0:30:54 data.gov.uk. Available at http://tinyurl.com/7joks46 [Accessed December 2012].
owner Davies, T. (2013). Open data barometer: 2013 global report. Available at http://www.
opendataresearch.org/dl/odb2013/Open-Data-Barometer-2013-Global-Report.pdf.
August 2012 1 1 Project Agency for 1:11:44
DECA. (2010). The value of the Danish address data: Social benefits from the 2002 agree-
manager Digitization
ment on procuring address data etc. free of charge. Available at http://www.adresse-
August 2012 1 3 Product KMD 1:00:03 info.dk/Portals/2/Benefit/Value_Assessment_Danish_Address_Data_UK_2010-07-
September owner Geodata 1:16:16 07b.pdf [Accessed July 2012].
2014 Project Agency 0:57:35 DeLaurentis, D., & Callaway, R. K. (2004). A system-of-systems perspective for public pol-
November Manager icy decisions. Review of Policy Research, 21(6), 829–837.
2014 de Vries, M., & Pijpker, U. (2013). The Danish Dash: A short story unravelling the Danish
September 2 1 Product KMD magic of shaping a System of Key registers in less than six months. Available www.
2012 owners pblq.nl/media/331260/danish_dash__final__1_.pdf.
September 1 1 Technical Agency for 0:54:12 Elster, J. (2007). Explaining social behavior. Cambridge: Cambridge University Press.
2012 architect Digitization Eurostat (2013). General government expenditure on social protection and health. Avail-
September 1 2 Product KMD 49:01 able http://ec.europa.eu/eurostat/statistics-explained/index.php/General_
2012 owner Ministry of 58:56 government_expenditure_on_social_protection_and_health#Social_protection:_the_
most_important_function_of_government_expenditure.
January Project Housing,
Fioretti, M. (2011). Open data: Emerging trends, issues and best practices. Available at
2014 manager Urban and
http://www.lem.sssup.it/WPLem/odos/odos_2.html [accessed July 2012].
Rural Affairs
Frischmann, B. M. (2012). Infrastructure: The social value of shared resources. Oxford: Ox-
October 1 1 External 1:25:56 ford University Press.
2012 consultant Gottschalk, P. (2009). Maturity levels for interoperability in digital government.
November 1 1 Project KOMBIT 0:51:14 Government Information Quarterly, 26(1), 75–81.
2012 manager Hedström, P., & Ylikoski, P. (2010). Causal mechanisms in the social sciences. Annual
January 1 2 Project Geodata Review of Sociology, 36, 49–67.
2013 manager Agency 1:25:40 Henriksen, H. Z. (2006). Fad or investment in the future: An analysis of the demand of
January e-services in Danish municipalities. Electronic Journal of e-Government, 4(2),
2014 19–26.
February 1 1 Consultant Ministry of 1:27:07 Hjort-Madsen, K. (2006). Enterprise architecture implementation and management: A
2013 Housing, case study on interoperability. In the Proceedings of the 39th Annual Hawaii Interna-
Urban and tional Conference on System Sciences HICCS, Vol. 4, 71c-71c.
Hofman, W., & Rajagopal, M. (2014). A technical framework for data sharing. Journal of
Rural Affairs
Theoretical and Applied Electronic Commerce Research, 9(3), 45–58.
May 2013 1 3 Basic data Agency for
Horst, N. L., Bjerre, S., Lind, M., & Hvingel, L. (2014). Grunddataprogrammet-en dansk
February ambassador Digitization 1:06:39 infrastrukturmodel for offenlige data. Geoforum Perspektiv, 13(24).
2014 0:55:32 Hujiboom, N., & Van den Broek, T. (2011). Open data: An international comparison of strat-
September egies. European Journal of e Practice.
2014 Höchtl, J., Davies, T., Janssen, M. F. W. H. A., & Schieferdecker, I. (2014). Open data: Grow-
October 2 1 Project Copenhagen 0:38:26 ing up and getting specific. JeDEM: eJournal of eDemocracy and Open Government,
2013 manager City 6(1), i–iii.
Programmer Höchtl, J., & Reichstädter, P. (2011). Linked open data — A means for public sector infor-
January 1 1 Project Ministry of 0:53:06 mation management. Lecture Notes in Computer Science, 6866(2011), 330–343.
2014 Manager Housing, Janssen, M., Charalabidis, Y., & Zuiderwijk, A. (2012). Benefits, adoption barriers and
Urban and myths of open data and open government. Information Systems Management, 29(4),
Rural Affairs 258–268.
Janssen, M., & Estevez, E. (2013). Lean government and platform-based governance —
February 2 1 App Geodata 0:38:11
Doing more with less. Government Information Quarterly, 30(1), S1–S8.
2014 developers Agency
Jetzek, T., Avital, M., & Bjørn-Andersen, N. (2012). The value of open government data: A
September 1 1 Data user KMD 0:41:01 strategic analysis framework. Paper presented at the 2012 pre-ICIS e-government SIG
2014 workshop. USA: Orlando.
November 1 1 Project Agency for 0:59:46 Jetzek T., Avital, M., & Bjorn-Andersen, N. (2013a). The generative mechanisms of open
2014 manager digitization government data. In the Proceedings of the 21st European Conference on Information
November 1 1 Project Danish 1:05:20 Systems (ECIS) 2013, Utrecht, The Netherlands.
2014 leader Business Jetzek, T., Avital, M., & Bjørn-Andersen, N. (2013b). Generating value from open govern-
Authority ment data. Proceedings of the 34th International Conference on Information Systems
Total 22 28 (ICIS 2013), Milan, Italy (pp. 1–20).
Jetzek, T., Avital, M., & Bjørn-Andersen, N. (2014). Data-driven innovation through open
government data. Journal of Theoretical and Applied Electronic Commerce Research,
9(2), 100–120.
References Kettl, D. F., Pollit, C., & Svara, J.H. (2004). Towards a Danish concept of public governance:
An international perspective. Report to the Danish Forum for Top Executive Manage-
Agency for Digitization (2011). eGovernment strategy 2011–2015: The digital path to fu- ment. Available http://www.publicgovernance.dk/docs/0408260903.pdf.
ture welfare. Available http://www.digst.dk/Servicemenu/English/Policy-and- King, J. L., & Uhlir, P. F. (2014). Soft infrastructure challenges to scientific knowledge dis-
Strategy/eGOV-strategy. covery. Communications of the ACM, 57(9), 35–37.
Agency for Digitization (2012). Good basic data for everyone — A driver for growth and Krieger-Røyen, J. (2014, June). The Danish BDP and the governance perspective Available
efficiency. Available http://www.digst.dk/Home/Servicemenu/English/Digitisation/ http://inspire.ec.europa.eu/events/conferences/inspire_2014/pdfs/plenaries/Grunddata_
Basic%20Data. INSPIRE_JRO5.pdf
104 T. Jetzek / Government Information Quarterly 33 (2016) 89–104

Lakomaa, E., & Kallberg, A. (2013). Open data as a foundation for innovation: The enabling Sullivan, A., & Sheffrin, S. M. (2003). Economics: Principles in action. Upper Saddle River,
effect of free public sector information for entrepreneurs. IEEE Access, 1(2013), New Jersey: Pearson Prentice Hall.
558–563. Tammisto, Y., & Lindman, J. (2011). Definition of open data services in software business.
Langley, A. (1999). Strategies for theorizing from process data. Academy of Management In M. A. Cusumano, B. Iyer, & N. Venkatraman (Eds.), Paper presented at the third In-
Review, 24(4), 691–710. ternational Conference on Software Business (ICSOB 2012), Cambridge, MA, USA. 114.
Martin, S., Foulonneau, M., Turki, S., & Ihadjadene, M. (2014). Risk analysis to overcome (pp. 297–303). Software Business Lecture Notes in Business Information Processing.
barriers to open data. Electronic Journal of e-Government, 11(2), 348–359. Van de Ven, A. H. (2007). Engaged scholarship — A guide for organizational and social re-
Maheshwari, D., & Janssen, M. (2014). Reconceptualizing measuring, benchmarking for search. Oxford: Oxford University Press.
improving interoperability in smart ecosystems: The effect of ubiquitous data and van der Blonk, H. (2003). Writing case studies in information systems research. Journal of
crowdsourcing. Government Information Quarterly, 31(1), S84–S92. Information Technology, 18(1), 45–52.
McKinsey & Company (2011). Big data: The next frontier for innovation, competition, and van Veenstra, A. F., & van den Broek, T. A. (2013). Opening moves — Drivers, enablers and
productivity. McKinsey Global Institute. barriers of open data in a semi-public organization. Proceedings of the 12th IFIP WG 8.5
McKinsey & Company (2013). Open data: Unlocking innovation and performance with liq- International Conference, EGOV 2013, Koblenz, Germany (pp. 50–61).
uid information. McKinsey Center for Government and McKinsey Business Technolo- Verhulst, S., Noveck, B. S., Caplan, R., Brown, K., & Paz, C. (2014). The open data era in
gy Office: McKinsey Global Institute. health and social care. The Governance Lab. [online]. Available: http://images.
Montealegre, R., Hovorka, D., & Germonprez, M. (2014). A coevolutionary view of infor- thegovlab.org/wordpress/wp-content/uploads/2014/06/nhs-full-report.pdf.
mation services development: Lessons from the US National Oceanic and Atmospher- Viscusi, G., Castelli, M., & Batini, C. (2014). Assessing social value in open data initiatives:
ic Administration. Journal of the Association for Information Systems, 15(9), 577–613. A framework. Future Internet, 6(3), 498–517.
Nilsen, K. (2010). Economic theory as it applies to Public Sector Information. Annual Yin, R. K. (2009). Case study research: Design and methods (4th ed.). Los Angeles: Sage.
Review of Information Science and Technology, 44(1), 419–489. Zuiderwijk, A., Janssen, M., Choenni, S., Meijer, R., & Alibaks, R. S. (2012). Socio-technical
OECD (2012). Human Resources Management Country Profiles Denmark. Available http:// impediments of open data. Electronic Journal of e-Government, 10(2), 156–172.
www.oecd.org/gov/pem/OECD%20HRM%20Profile%20-%20Denmark.pdf. Zuiderwijk, A., & Janssen, M. (2014a). Barriers and development directions for the publi-
Open Knowledge Foundation (OKF) (2015). Available at http://www.opendefinition.org/ cation and usage of open data: A socio-technical view. In M. Gascó-Hernández (Ed.),
and http://opengovernmentdata.org. Open government, opportunities and challenges for public governance (pp. 115–136).
Pollock, R. (2008). Economics of PSI. Cambridge: University of Cambridge Available at New York: Springer.
http://rufuspollock.org/economics/papers/economics_of_psi.pdf [accessed August Zuiderwijk, A., & Janssen, M. (2014b). Open data policies, their implementation and im-
2012]. pact: A framework for comparison. Government Information Quarterly, 31(1), 17–29.
Provan, K. G., & Kenis, P. N. (2008). Modes of network governance: Structure, manage- Zuiderwijk, A., & Janssen, M. (2014c). The negative effects of open government data-
ment, and effectiveness. Journal of Public Administration Research and Theory, 18(2), investigating the dark side of open data. Proceedings of the 15th Annual International
229–252. Conference on Digital Government Research (pp. 147–152).
Rechtin, E., & Maier, M. W. (2009). The art of systems architecting (3rd edition). CRC Press.
Statistics Denmark (2013). Denmark in figures, 2013 Available http://www.dst.dk/pukora/
epub/upload/17953/dkinfigures.pdf.