Reordering Life: Knowledge and Control in the Genomics Revolution

Inside Technology

Edited by Wiebe E. Bijker, W. Bernard Carlson, and Trevor Pinch

A list of the series appears at the back of the book.

Reordering Life

Knowledge and Control in the Genomics Revolution

Stephen Hilgartner

The MIT Press

Cambridge, Massachusetts

London, England

© 2017 Massachusetts Institute of Technology

All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.

This book was set in Stone Sans and Stone Serif by Toppan Best-set Premedia Limited. Printed and bound in the United States of America.

Library of Congress Cataloging-in-Publication Data

Names: Hilgartner, Stephen, author.

Title: Reordering life : knowledge and control in the genomics revolution /

Stephen Hilgartner.

Other titles: Inside technology.

Description: Cambridge, MA : The MIT Press, [2017] | Series: Inside

technology | Includes bibliographical references and index.

Identifiers: LCCN 2016030923 | ISBN 9780262035866 (hardcover : alk. paper)

eISBN 9780262338653

Subjects: | MESH: Human Genome Project. | Databases, Genetic--ethics |

Genetic Research--legislation & jurisprudence | Intellectual Property

Classification: LCC QH447 | NLM QU 450 | DDC 572.8/629--dc23 LC record available at https://lccn.loc.gov/2016030923

ePub Version 1.0

d_r0

To Kate

Table of Contents

Series page

Title page

Copyright page

Dedication

Acknowledgments

List of Abbreviations

1 Introduction

2 Envisioning a Revolution

3 Laboratories of Control

4 Research Programs and Communities

5 Objects of Transformation

6 Regime Change and Interaction

7 Shaping News and Making History

8 Conclusion

Appendix: Fieldwork and the Control of Knowledge

References

Index

Inside Technology

List of Tables

Table 2.1 NRC Committee on Mapping and Sequencing the Human Genome, 1988

Table 2.2 Sizes of Genomes

Table 4.1 Comparison of the HGP and RLS Regimes

Table 5.1 Selected Genomics Startup Companies Founded 1991–1993

Table 7.1 Three Narratives in the Initial News Coverage of the Human Genome Sequencing Competition

Table 7.2 Narrating the Competition

List of Illustrations

Figure 2.1 Four ways genome scientists represent DNA.(a) Double-helical structure showing base pairing and a specific sequence. The strands of the DNA molecule can separate without changing the order of the bases, as shown on the right. (b) A double-stranded DNA sequence displayed as a string of text. The human genome is 140,000,000 times longer than this 22-base-pair sequence. If printed in a linear array (12 letters per inch), the human genome sequence would easily reach from Cambridge, Massachusetts, to Cambridge, England—the cities where practicable DNA-sequencing methods were developed in the 1970s. The sequence shown is from the paper that first described Maxam-Gilbert sequencing in 1977 (Maxam and Gilbert 1977). (c) A representation of a genetic sequence with different regions or features of the sequence indicated using shading. (d) Strands of DNA depicted as lines in a sketch from a laboratory discussion in 1991. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 2.2 How genes are expressed in human cells. This figure and caption appeared in a key NRC report Mapping and Sequencing the Human Genome (1988): How genes are expressed in human cells. Each gene can specify the synthesis of a particular protein. Whether a gene is off or on depends on signals that act on the regulatory region of the gene. When the gene is on, the entire gene is transcribed into a large RNA molecule (primary RNA transcript). This RNA molecule carries the same genetic information as the region of DNA from which it is transcribed because its sequence of nucleotides is determined by complementary nucleotide pairing to the DNA during RNA synthesis. The RNA quickly undergoes a reaction called RNA splicing that removes all of its intron sequences and joins together its coding sequences (its exons). This produces a messenger RNA (mRNA) molecule. The RNA chain is then used to direct the sequence of a protein (translation) according to the genetic code in which every three nucleotides (a codon) specifies one subunit (an amino acid) in the protein chain. Source: NRC 1988, fig. 2-3, p. 18. Reprinted with permission by the National Academy of Sciences. Courtesy of the National Academies Press, Washington, DC.

Figure 2.3 Detecting a specific DNA sequence with a labeled probe. Scientists can exploit the base-pairing property of DNA to test samples for the presence of specific sequences. A radioactively labeled probe containing the sequence of interest is prepared (a). The probe is poured over the DNA to be tested (b). The probe binds only to sequences that match the probe (c). This technique for detecting specific sequences is called hybridization. Fluorescent tags can also be used. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 2.4 Cloning DNA using recombinant DNA technology. A small, circular piece of DNA called a plasmid is used as a vector to transfer a DNA fragment into a bacterial cell. The plasmid, which contains a gene conveying antibiotic resistance, is cut, and the DNA fragment to be cloned is spliced in, as shown in (a). Some bacterial cells take up the plasmid. After exposure to the antibiotic, only cells containing the plasmid survive (b). The cells multiply, producing many copies of the cloned fragment (c). The cells can be grown, stored, and transported, providing an ongoing supply of the cloned DNA. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 2.5 Sanger sequencing. This figure, which describes DNA sequencing by the enzymatic method, appeared in the NRC report Mapping and Sequencing the Human Genome (1988). This method takes advantage of an enzyme that participates in the replication of DNA in living cells. The enzyme (known as DNA polymerase) uses a piece of single-stranded DNA as a template and creates a complementary strand, building up a chain of bases in the correct order. Each new base is attached one by one to the end of the growing chain. Molecular biologists use DNA polymerases to make complementary strands of DNA in vitro. In Sanger sequencing, a polymerase is used to build complementary strands that are artificially interrupted by including an altered nucleotide (dideoxynucleoside triphosphate) that blocks the addition of the next nucleotide in the chain. The result is a family of DNA fragments that vary in length by one nucleotide. These fragments are sorted by length using gel electrophoresis, which makes it possible to read the sequence by inspecting the gel. A single run of such a sequencing gel produces a strip of sequence—or read—that is only about 500 bases in length. Longer sequences can be assembled by combining reads (figure 2.6). Source: NRC 1988, fig. 5-1, p. 63. Reprinted with permission by the National Academy of Sciences. Courtesy of the National Academies Press, Washington, DC.

Figure 2.6 Connecting overlapping reads to form a continuous sequence. Short reads of DNA sequence can be aligned and joined to form larger strips of sequence. The principle is illustrated here using a sequence of English-language characters rather than the four-letter alphabet of DNA. (This example uses a read length of about 12 characters rather than about 500 in DNA sequencing.) First, many copies of an unknown sequence (a), shown in gray to indicate that the sequence is unknown, are broken at random places to form small fragments (b). Sequencing of these fragments then begins, and as each new read is produced, its sequence is compared with the reads already completed. Overlapping reads are joined, creating small contigs (c), shown in black to indicate that their sequences are now visible. As sequencing continues, new reads extend existing contigs (d). Final gaps are eventually filled, revealing the originally unknown sequence in its entirely (e). The DNA sequencing technology at the outset of the HGP produced reads of some 200 to 500 base pairs, typically containing some errors, especially toward the end of the read. The human genome contains many repetitive sequences, which complicate the problem of assembling continuous sequence in the correct order. The sentence shown is a famous line from Watson and Crick’s paper from 1953 describing the structure of DNA. Illustration by Ranjit Singh.

Figure 2.7 Maps at different levels of resolution. The lowest-resolution genome maps depict banding patterns on chromosomes, such as the map of human chromosome 16 shown in (a). As the HGP took shape, the genomics vanguard imagined building maps at very different levels of resolution, with genome sequences (e) conceptualized as maps at the highest level of resolution: one base pair. They also envisioned building contig maps (d), seeking to span whole chromosomes with overlapping clones, the resolution depending on the cloning technology used: for cosmids, roughly 40,000 base pairs; for YACs, from 100,000 to 1,000,000 base pairs or more. Genome scientists also planned to build restriction maps of whole chromosomes (c), with an expected resolution of 1 to 2 million base pairs. The resolution of genetic linkage maps cannot be expressed in terms of base pairs because linkage maps and physical maps measure fundamentally different things. Distances in linkage maps are measured in centimorgans, which indicate the degree of linkage between polymorphic markers or genes (b). On average, one centimorgan corresponds to a physical distance of about 1 million base pairs. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 2.8 Assembling a contig map by identifying overlapping clones. At the outset of the HGP, scientists worked to produce contig maps using several methods to find overlapping clones. This simplified schematic illustrates one method. Many copies of a genome are cut randomly into large fragments (tens or hundreds of thousands of base pairs) (a).These fragments are cloned, and a library of clones is created, some of which overlap. Each clone is systematically tested against a set of markers (M1 to Mn), such as radioactively labeled DNA probes, to find overlaps. A positive test result means that the clone contains the marker. Like the order of the clones, the order of the markers on the chromosome is initially not known. However, test results (b) reveal that some clones share a particular marker, indicating that they overlap. In this example, the first and third clones overlap because they both test positive for marker M2. As testing proceeds, the alignment of overlapping clones creates small contigs that join to form larger contigs as additional test results accumulate. In theory, each chromosome will eventually be spanned by a single contig because a chromosome is a single continuous DNA molecule. A contig is depicted in (c). The order on the chromosome of both the clones (c) and the markers (d) is thus established, although the exact location of neither is determined. The map shown in gray in (e), which displays the contig with overlaps eliminated, cannot be completed using this technique because the extent to which each clone overlaps its neighbors cannot be ascertained. However, the ordered clones (c) and ordered markers (d) can serve as starting points for more detailed mapping or sequencing. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 4.1 Connecting maps using STSs. During the early 1990s, genome mappers began using STSs—unique landmarks detectable by PCR—to link maps at different levels of resolution. The location of these STSs on a variety of physical maps could then be determined. The example given here, from a project to map chromosome 16, shows an STS (d) that was localized using in situ hybridization to a 3 to 4 million base-pair region on the map (a). It was also localized on a 150,000 base-pair YAC (b) and on a cosmid contig (c). In addition, polymorphic STSs could also serve as genetic markers, allowing them to connect genetic linkage maps to physical maps. Source: Adapted from Doggett, Stallings, Hildebrand, et al. 1992, 204–205. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 4.2 Detection of clones using high-density hybridization.(a) A robot moves clones stored in microtiter plates, positioning them in precise locations on filters. (b) A radioactively labeled probe is poured over the filter; it hybridizes with clones containing matching DNA. X-ray film detects the positions where probes hybridized to clones. (c) The coordinates of spots that light up on the filter are used to identify the clones. A sample of the clone is transferred into a tube. (d) The clone is retested against the probe to eliminate any false positives. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 4.3 Script for an RLS transaction. Illustration by Chris Cooley, Cooley Creative, LLC.

Figure 5.1 Creating a library of cDNA clones. Scientists extract total mRNA from cells (a) of the tissue type chosen (e.g., human brain cells). The resulting population of mRNA molecules (b) is treated with the enzyme reverse transcriptase, which forms DNA molecules with sequences that complement those mRNA molecules. The resulting cDNA molecules (c) are then cloned into bacterial cells (d) using recombinant DNA techniques (see figure 2.4). In this way, genome scientists can create libraries of clones containing DNA sequences expressed in the specific tissue type chosen. By beginning with mouse kidney cells, for example, a researcher can create a library of cDNAs made from genes expressed in the mouse kidney. Each cDNA clone may contain only a partial sequence of gene it encodes, not the full-length sequence found in the original genomic DNA.

Figure 5.2 A partial cDNA sequence. This cDNA sequence is a fragment, 308 bases long, of a gene expressed in the human uterus. It appears as the following Genbank entry: EST41904 Human Uterus Homo sapiens cDNA 5- end similar to human tat-binding protein isolog YTA2, mRNA sequence. This entry is linked to a major paper by Craig Venter’s group: Adams, Kerlavage, Fleischmann, et al. 1995.

Figure 7.1 Clone-by-clone and whole-genome shotgun (WGS) strategies. The clone-by-clone strategy (left), also known as the hierarchical shotgun strategy, begins with the construction of a set of overlapping BAC clones. Each of these clones is separately shotgun sequenced—that is, shattered into small pieces, sequenced, and assembled into contiguous sequence. The WGS strategy (right) begins by shattering the entire genome into small pieces, sequencing them, and assembling all of them at once into contiguous sequence. This strategy skips the steps needed to map the BACs, but the assembly process is more complicated. The assembly step in the clone-by-clone approach benefits from compartmentalized assembly. A typical BAC contains about 200,000 base pairs—or about 10481_e007_001.jpg of the human genome. Assembling each BAC thus entails working with many fewer fragments than must be assembled in WGS strategy. Each BAC also has a known location on the genome, providing additional information that the clone-by-clone strategy can use to assemble and anchor contigs to the genome. Source: Waterston, Lander, and Sulston 2002.

Figure 7.2 Repeated sequences produce ambiguities. DNA sequences that occur multiple times in a genome can produce ambiguities when sequencing. The basic principle is illustrated in greatly simplified form using a sequence of English-language characters. (a) Two strips of text from different parts of the same genome appear in 1 and 2. They are shown in gray to indicate that the sequences are initially unknown. Both of these sequences contain a repeat: the 14-character text •enjoy•eating•. In this example, the maximum read length is 12 characters, which is shorter than the repeat. As a result, when these sequences are shattered into small fragments (b) and read and assembled into contigs (c), none of the reads extends all of the way through the repeat. For this reason, the assembly process will yield multiple configurations (d), only one of which is correct. Much more complicated ambiguities result from repeats that recur many times in a genome. Producing highly accurate, finished DNA sequences thus may require undertaking additional experiments and analyses to resolve ambiguities and obtain accurate assemblies. Illustration by Ranjit Singh.

Acknowledgments

I could not have undertaken a project of this scope without the assistance and support of many people. I am especially indebted to my interlocutors in the world of genomics. Although the strictures of confidentiality prevent me from thanking them by name here, I greatly appreciate the generosity with which they welcomed me into their laboratories and meetings; gave freely of their precious time; shared their knowledge, opinions, hopes, and frustrations; tolerated intrusive questions; referred me to their colleagues; and assisted me in many other ways, large and small. Interacting with them was a pleasure and a privilege, and their excitement and enthusiasm was often contagious. I will always remain extremely grateful to them for so willingly sharing their worlds with me.

A number of colleagues and friends carefully read part or all of the manuscript, including Angie Boyce, Peter Dear, Michael Dennis, Jenny Gavacs, Rob Hagendijk, Jonathan Luskin, Ilil Navah-Benjamin, Nicole Nelson, Shobita Parthasarathy, Trevor Pinch, Judith Reppy, Kasia Tolwinski, and several anonymous reviewers. Their astute comments and criticisms improved the manuscript considerably.

Over the years many people contributed to this project through helpful conversations, challenging questions, and other forms of assistance, encouragement, and advice. I particularly thank Richard Bell, Carin Berkowitz, Charles Bosk, Sherry Brandt-Rauf, Alberto Cambrosio, Adele Clarke, Mehmet Ekinci, Ulrike Felt, Joan Fujimura, Jean-Paul Gaudillière, Tom Gieryn, Dave Guston, Linda Hogle, Tony Holtzman, Ben Hurlbut, Pierre-Benoit Joly, Peter Keating, Timothy Lenoir, Sabina Leonelli, Bruce Lewenstein, Javier Lezaun, Susan Lindee, Michael Lynch, Clark Miller, Yoshio Nukaga, Alain Pottage, Barbara Prainsack, Jenny Reardon, Hans-Jörg Rheinberger, Dirk Stemerding, Kaushik Sunder Rajan, Mariachiara Tallacchini, and Peter Weingart. It was a pleasure to work on the diagrams with Chris Cooley and Ranjit Singh, and I thank Chip Aquadro for his invaluable comments on several of them.

The faculty and graduate students of Cornell’s Department of Science & Technology Studies created a stimulating intellectual environment, and the department’s dedicated staff—especially our administrative manager Deb Van Galder—has earned my heartfelt appreciation. At MIT Press, I thank my editor, Katie Helke, for her enthusiasm and assistance, Virginia Crossman for her editorial advice, and Annie Barva, who copyedited the manuscript. Chris Hesselbein prepared the index.

Three people stand out for special thanks: David Rothman for his enthusiastic support of this research during its early stages; the late Dorothy Nelkin for helping me launch a career in social studies of science; and Sheila Jasanoff for her inspiring intellectual vision and many forms of guidance over the years.

No written acknowledgment can adequately express the depth of my gratitude to my family. I thank my mother for her love and support, and for her loving presence in the lives of my children. Mel, her late husband, is sorely missed. My father’s enthusiasm for science was an important influence during my early years. My children, Nathan, Kevin, and Erin, continually inspire me, and I have known no greater joy than being their dad. And then there is Kate. The joining of our lives many years ago was a wonderful and transformative change. I cannot thank her enough for her insightful comments on my work, for her wisdom and support, and most of all for being the love of my life.

* * *

The research reported here is based on work supported in part by grants from the National Institutes of Health (NCHGR ELSI Program) and the National Science Foundation (Grants No. 0083414 and 035200). The findings and interpretations expressed in this book do not necessarily reflect the views of the National Institutes of Health or the National Science Foundation.

List of Abbreviations

1

Introduction

Understanding the dynamics of scientific and social change is an urgent challenge for the twenty-first-century social sciences. Consider the genomics revolution. In the mid-1980s, a scientific vanguard emerged with the explicit aim of revolutionizing the biological sciences and medicine. These scientists sought to make genomes—the totality of an organism’s inherited DNA—into tractable objects of analysis, finding all the genes once and for all and creating a new biology for the twenty-first century.¹ Their vision of transforming the sciences of life, manifested most dramatically in calls for a Human Genome Project, soon captured imaginations, winning a $3 billion commitment from the US Congress. By June 2000, when US president Bill Clinton and UK prime minister Tony Blair held a joint news conference to announce that a first survey of the entire sequence of the human genome had been completed, genomic knowledge and technology had become indispensable to biological research and to the biotechnology and pharmaceutical industries.² The rise of genomics, most observers agreed, portended even greater change to come, not only in the new life sciences but also in the lives of ordinary citizens.

During the 1990s, biologists and others increasingly integrated genomic perspectives into ways of doing and funding research, producing significant change both within and beyond the scientific community. New factory-style laboratories took shape, greatly accelerating data production but fitting poorly with established ways to organize work and careers in molecular biology. The difficulty of working on a genomic scale made translaboratory cooperation more desirable, leading to conflicting visions of how best to orchestrate it. The quantity of data in the genome databases used by the scientific community continued to grow exponentially, destabilizing the existing systems for collecting and circulating data and challenging long-standing practices of scientific publication. Scientists and entrepreneurs reimagined genomic information as a form of capital, inspiring the formation of a new wave of companies and provoking intense debate about the ownership of the human genome. The prospect of a revolution in medicine inspired hope but also raised ethical, legal, and social questions. Such changes did not emerge from an orderly scientific debate or a straightforward policy-making process but through dynamic—and often contentious—negotiations among many actors in many institutional locations.

This book examines the genomics revolution, treating it as an empirical site for investigating transformative scientific and social change. My study concentrates on what is often seen as the scientific and symbolic centerpiece of genomics: the Human Genome Project (HGP), an international effort that ranks among the most celebrated scientific achievements of the late twentieth century. Bringing genomics into being during the HGP was simultaneously an epistemic, material, and political process, and I examine how problems of governance arose in concert with new genomic knowledge and technology. Although some of these problems were addressed through official policy, others were settled in ad hoc forums and informal encounters. Controversial issues emerged at many levels, ranging from the individual laboratory to the field of genome research, to the wider worlds of science, industry, and society. Scientists, along with many other actors, contested a variety of questions: What forms of accountability should govern the production and use of genomic knowledge? Who should own genomic knowledge, and what should ownership mean? Under what conditions should researchers publish or withhold data? How should translaboratory collaborations be structured? And how should researchers interact with the news media and respond to societal concerns?

Examining how such questions were contested and resolved allows me to investigate how new forms of knowledge and new modes of control took shape as genomics emerged. From this point of view, the genomics revolution entailed not only change in knowledge and technology but also change in knowledge-control regimes—structures that allocate entitlements and burdens pertaining to knowledge. I will elaborate this concept more fully below. For now, suffice it to say that in a manner broadly analogous to the law, knowledge-control regimes constitute modes of control that apply to specific actors, entities, and jurisdictions—although they vary in the extent to which they are formally codified. In the biological sciences, computing and information technology, and many other sociotechnical domains, the processes that reconfigure knowledge-control regimes are an important aspect of the politics of contemporary societies.

This study is an empirical one, and I collected data using ethnographic and interview methods. Participant observation and interviewing took place in laboratories, scientific meetings, government agencies, biotechnology companies, and forums addressing ethical, legal, and social issues. Most of my fieldwork was conducted in the United States, and my account focuses on that country, but I also did some field research and interviews in France, Germany, and the United Kingdom. Because my goal was to investigate change rather than to characterize a relatively stable area of research, it was necessary for me to follow the action over time. The field research underlying this book began in 1988, two years before the formal launch of the genome project, and ended (with the exception of some follow-up fieldwork) in 2003, the year when the HGP was officially completed.³ I introduce the field of genomics and the HGP in chapter 2, but my first order of business is to outline the theoretical framework and research methods underlying this study.

Transformative Scientific Change

To analyze the dynamics of change in knowledge and governance during the HGP, this book focuses on the contestation and reallocation of control during the production of genomic knowledge and technology. Put otherwise, I use this revolutionary moment as a convenient place to examine how modes of control were reconfigured as knowledge changed (and vice versa). As new forms of knowledge took shape, what adjustments were made in the lawlike regimes that governed biological research? How—in the laboratory and beyond—did actors attempt to use, resist, or change extant regimes and practices? How did they seek to establish and justify (re)allocations of rights, duties, privileges, and powers? Which regimes and practices proved durable, which changed, and why? And what implications did the process have for the distribution of authority, wealth, power, and the capacity to shape the future?

The existing literature on transformative scientific change offers some useful starting points for addressing such questions. Consider three famous examples, beginning with Thomas S. Kuhn’s The Structure of Scientific Revolutions. For Kuhn, scientific revolutions overturn the extant paradigms that govern the research process, transforming scientists’ conceptual, theoretical, instrumental, and methodological commitments. Knowledge does not simply accumulate but is fundamentally reconstructed to fit the new worldview.⁴ What constitutes a worthwhile research question, what counts as adequate evidence, and which scientific skills matter may all be redefined. Revolutions thus reorder both what is known and the community of knowers. In this sense, Kuhn treats knowledge and social order as forming a single system that revolution transforms—an approach I build on here. Yet the social order that he describes is internal to the scientific community, set apart from the wider society.⁵ Kuhn’s revolutions occur within science, and he is relatively uninterested in how change in scientific paradigms connects to broader social change.⁶ After a revolution brings a new paradigm into being, he writes, scientists may inhabit a new world, but outside the laboratory everyday affairs usually continue much as they did before.⁷

Ludwik Fleck’s book The Genesis and Development of a Scientific Fact, published in 1935, presents another important analysis of transformative scientific change. Key aspects of Fleck’s account anticipated The Structure of Scientific Revolutions, offering what is ultimately a richer sociology of knowledge.⁸ Fleck’s concept of a thought style—a closed, stylized system that constrains thinking and the interpretation of evidence—resembles Kuhn’s paradigms. His concept of a thought collective composed of individuals who share a thought style also encompasses what Kuhn would call a scientific community, but Fleck’s concept is broader.⁹ All individuals belong to many thought collectives, which are not associated exclusively with science but also with commerce, the military, the arts, politics, religion, fashion, and so forth. In each domain, some thought collectives are relatively specialized, whereas others are less so. These thought collectives are tied together through multiple intersections and interrelations … both in space and time. Moreover, scientific ideas are embedded in society; culturally available proto-ideas promote their emergence and acceptance. Fleck also argues that new knowledge does not simply diffuse unchanged. It inevitably undergoes a stylized remodeling as it moves from the esoteric vanguard of a scientific field to increasingly exoteric thought collectives that incorporate it into their ideas and practices.¹⁰ Fleck thus describes a dynamic process in which, as Ilana Löwy explains, expert knowledge is influenced by popular knowledge, and then influences it in turn.¹¹ Transformative scientific change unfolds through interactions within and among thought collectives that are thoroughly embedded in modern societies.¹²

Finally, consider Bruno Latour’s study The Pasteurization of France. Latour rejects the idea that the revolution attributed to Louis Pasteur can be analyzed by any approach that takes for granted an a priori distinction between science and society.¹³ The world is composed of hybrids of nature and culture, so understanding this revolution requires avoiding all distinctions between the content and context of science, between the natural and the social, even between human and nonhuman agency. Rather than attributing social change to scientific change or vice versa, Latour investigates the reconfiguration of networks of associations connecting heterogeneous actors while rejecting such distinctions as human/nonhuman or natural/social. His method involves tracing the process through which new actors—such as Pasteur’s microbes—are constituted as entities, circulated in texts, and built into material practices in medicine and hygiene. For Latour, this process of network building constitutes new scientific knowledge and new social groups, but neither the knowledge nor the groups cause change; they are both the outcome of the reconfiguration of networks. Latour thus develops a metaphysical and methodological perspective on transformative change that treats nature and society as being constituted through a single process of reconfiguring networks.¹⁴

Each of these perspectives provides important insights about the nature of transformative scientific change. Yet neither Kuhn nor Fleck devote sustained attention to questions about the changes in the allocation of entitlements and burdens or questions of governance. Kuhn focuses narrowly on a scientific community that is strangely detached from the rest of society. Fleck’s examination of intersecting thought collectives addresses this problem, but his project—to provide a historically and socially situated account of epistemic change—is not concerned with the distribution of power or the machinery of governance. Latour focuses on the emergence of new forms of power and offers a deep view of how knowledge and social order change together. His important insights about hybridity and heterogeneity, and his attention to the role of material entities in establishing order, are particularly helpful to my analysis. But his emphasis on metaphysical matters directs inquiry away from the critical and normative questions of greatest concern to people interested in how extant allocations of authority, wealth, and power are reordered during periods of transformative change in knowledge.¹⁵

The framework that Sheila Jasanoff terms interactional co-production offers a more promising theoretical point of departure for addressing such questions.¹⁶ Jasanoff’s approach builds on Steven Shapin and Simon Schaffer’s Leviathan and the Air-Pump, extending its theoretical reach using comparative analysis of law and politics.¹⁷ She proceeds from three main assumptions: (1) that ways of knowing the world and ways of ordering social relations are deeply interconnected, forming a single mode of ordering knowledge and society; (2) that established orders change through processes in which knowledge and social relations are mutually adjusted; and (3) that analyzing these processes requires being symmetrically attentive to changes in knowledge and governance. Her theoretical framework emphasizes such questions as how institutions, discourses, identities, constitutions, and imaginaries shape modes of decision making and guide public reason in specific societies.¹⁸ Jasanoff’s analytic perspective is well suited to exploring such politically significant questions as: During periods of transformative scientific change, how do reconfigurations of knowledge and lawlike modes of control take shape? And with what effects on the allocation of power, wealth, and the capacity to participate in processes that influence the future of sociotechnical change?

This book addresses these and similar questions by examining the coproduction of genomic knowledge (including technology) and the regimes and practices of control during the period of the HGP. In contrast to Jasanoff, whose research usually takes such sites as regulatory agencies, courts, and ethics commissions as empirical starting points, this study takes a specific scientific community as its empirical focus, following the activities of the genomics community in the laboratory and many other sites. As the vision of a new genomic knowledge was fleshed out, how did existing social orders integrate genomics into their practices? What changes took place in the lawlike regimes through which control was secured, contested, allocated, resisted, justified, and reconfigured? What possibilities were realized, which roads were not taken, and how can we account for these outcomes?

Three forms of control are particularly relevant to my investigation:

Control over objects. One way that control is allocated among agents is by managing the social, spatial, and temporal distribution of knowledge objects—entities or things that contain or constitute knowledge. Knowledge objects include not only scientific results, data, and information but also preliminary findings and plans, inscriptions and biomaterials, instruments and skilled personnel, techniques and software, as well as rumor, speculation, and scuttlebutt. An inclusive definition is needed because a wide range of materials, inscriptions, and devices are involved in the production of knowledge and control.¹⁹ In examining genomics, I pay close attention to the distribution and contestation of control over knowledge objects and examine how change in knowledge objects is implicated in changes in control.

Control over jurisdictions. Control is also allocated by constituting jurisdictions—bounded spaces (physical, sociopolitical, and discursive) onto which agents and their capacities are mapped.²⁰ Because actors often contest jurisdictions, particularly during periods of change, the rhetorical moves that Thomas F. Gieryn calls boundary work are often central to struggles over control.²¹ For example, classifying a contentious question as a matter of science or a matter of policy may have a strong bearing on who is entitled to decide the issue and how.²² In examining genomics, I pay close attention to the (re)structuring of jurisdictions and to jurisdictional contests and boundary work.²³

Control of relationships. Taking a relational view of control, I systematically investigate regimes and practices that allocate entitlements and burdens among agents. Like the legal relationships identified by Wesley Newcomb Hohfeld,²⁴ control relationships structure the terms of interaction among agents. If a control relationship grants one agent a specific entitlement (e.g., to prohibit trespassing on her property), it exists with respect to another agent (or the class of similarly situated agents) who bears a correlative burden (e.g., not to enter without permission). This study examines how control relationships formatted the roles and conduct of agents, how actors sometimes resisted control, and how control relationships changed during the action in the rise of genomics. For example, I analyze how science policy makers attempted to introduce new forms of accountability to control genome research and I consider ways that researchers adapted to—and sometimes resisted—such controls.

These three forms of control—of objects, of jurisdictions, and of relationships—do not operate separately; they are best thought of as aspects of a single, dynamic process through which specific configurations of knowledge and control are made, reproduced, and changed. I investigate this process in a variety of sites, including face-to-face interactions among scientists, exchanges among laboratories, policies to guide research programs, collaborations among companies and between companies and scientists, and attempts to shape the interpretive narratives about genomics displayed to publics.

Knowledge-Control Regimes

To analyze the rise of genomics, this book employs a theoretical framework centered on what I call knowledge-control regimes. The term regime is used widely in the social sciences in a variety of contexts. Most usage shares the idea that a regime imposes order on a domain or activity, typically through some combination of formal rules, informal norms, material means, and discursive framings.²⁵ As a first step in introducing the concept of a knowledge-control regime, let me offer some examples: military classification, trade secrecy, the regulation of confidential human-subject information, and the legal instruments that formally codify intellectual property.²⁶ The concept, however, is not limited to regimes that restrict the flow of knowledge objects; it also encompasses those intended to accomplish the reverse, including the regime of authorship and publication in scientific journals and such open-source regimes as the General Public License, Creative Commons, and the BioBricks Initiative.²⁷ Genomics companies’ business models—which constitute particular modes of control over knowledge—also qualify as knowledge-control regimes. So do the procedures that science advisory committees use to regulate access to information while they prepare written reports.²⁸ A moral economy of shared expectations about resource exchange also falls within this category, as do the public-relations practices that corporations and governments use to shape news coverage.²⁹ Even