Processing Cultures and Culturing Data
In her famous essay titled “Situated Knowledges,” Donna Haraway writes, “Critics of the sciences and their claims or associated ideologies, have shied away from doctrines of scientific objectivity in part because of suspicion that an ‘object’ of knowledge is a passive and inert thing.”1 I am both haunted and vexed by Haraway’s unrelenting insistence that we do better than simply point out the obvious power implications of a technoscientific find. Yet, there is a place, in the analytical practice of making a partial account, 2 through which the power of the knower must be accounted for. In this chapter, I examine the practices of collaborative data creation and sharing and assess them for the ways the bodies of DNA donors are made into objects for the research enterprise. Central to this objectification is the conversion of DNA into data sets, which requires a heterogeneous set of skills, technologies, and processes. Laying out these social and technological practices demonstrates the ways Mexicana/o donor bodies are made to span texts and worlds, in the syntactical registers of statistical noise, pure data, and power. We begin by situating data collection and the conversion of blood samples into genotypes.
“Aren’t you bored?” asked Alma. Having returned from a home visit where field staff retrieved blood samples, I had been watching her for nearly 20 minutes prepare the samples for storage in the freezer. Though everyone spends time in the lab processing samples from participants, Alma seems to be most at home there. The lab is located at the back of the building. One wall is covered in shelving and a 10-foot countertop that serves as a wet lab. Two freezers, a closet, and a bathroom are also part of the lab space. A small centrifuge, pipette equipment and shipping supplies are neatly arranged on the shelving and lab bench wall. After a blood draw, samples are placed in the centrifuge and spun down into plasma, buffy (white blood cells), and the reddish serum. Then they are separated, labeled, and stored for shipping. Alma prepares samples for Houston once a month. “Mayo [Clinic] goes out on Mondays,” she explains. Other research projects necessitate samples of blood, lipids, urine, plasma, or images (echocardiographic and retinographic) to be sent to San Antonio, Madison, and Chicago. For each study, Carl’s Houston staff sends carefully prepared shipping supplies to the field office. Zip-seal storage bags, pages of preprinted labels, shipping labels, and checklists arrive at predetermined intervals. Pages of labels printed with participant identification numbers, protocol codes, dates, and other information are placed on the 8.5 mL Vacutainer tubes color coded by research project. For each participant there are numerous labels corresponding to the various samples taken, which are coded further for each research project to which they have donated. Houston knows well in advance who will be sampled and when the samples are to be shipped.
When shipping samples off to collaborators, field staff can get a bit edgy with one another. “We need more ice!” Maria announced to her colleagues in anticipation. “Did we get those boxes yet? Who ordered them? Ay, come on you guys, someone call Airborne [Express] and tell them we need about 12 more,” she cajoled. Later that afternoon, Airborne Express arrives with stacks of boxes, some empty. “We must open everything,” Maria says, directing her more junior colleague. In addition to the empty shipping boxes, more forms and labels arrive. Consent forms, intake forms, and cardiac referral forms are quickly sorted, counted, and taken to the various places throughout the center where they will be ready for use. “If you need any supplies, I’m doing the order,” Maria calls out loudly to her colleagues. It’s after 5, and the last participant left about an hour earlier. “The woman [in Houston] who fills our supplies order is going on vacation, so we need to order this week,” she explains. The mood around shipping is not one of confianza, but rather of ensuring that the samples get to collaborators in adherence with strict guidelines. A box that is improperly prepared could mean the loss of dozens of samples—a time-consuming setback.
Judi and her staff often spend time talking about quality-control issues. A mislabeled vial or improperly filled out form, once discovered, immediately requires a response. Before shipping, staffers exchange paperwork to check for any errors, missing information, or anything else that would incur a call from Houston. The work of the field staff is to make the samples conform to the specifications of the entire collaborative diabetes enterprise. Therefore, an error early on in the development of a data set wreaks havoc downstream in the research process.
Judi and her staff work to render the lived conditions of donors’ lives accessible to researchers around the world. In this chapter, I will follow the blood samples from their point of acquisition on the border to the next node in the collaborative network, to Chicago at the labs of Nora and Gary. Examining the processes of data sharing between collaborators in the diabetes enterprise demonstrates that collaboration requires the regulation of ideas, practices, and populations. Converting bodies into data sets is a requirement for the production of knowledges based upon those bodies. Specifically, I argue that researchers develop, prepare, and narrate “good data sets,” which require that researchers perform what I will call “articulation work.” Articulation work occurs when researchers reconcile diverse frames of epistemological reference, just as an accountant reconciles a ledger.
The challenges to this reconciliation are that bodies, data sets, and technologies (and sometimes the behaviors of collaborating colleagues) are unruly things. This requires several moves on behalf of the scientist that convert the unruly, into workable objects for analysis. As Haraway writes, “It—the world—must, in short, be objectified as a thing, not as an agent; it must be matter for the self formation of the only social being in the productions of knowledge, the human knower.”3 We will see in this chapter that diabetes science collaboration across disciplines, geographies, and methodologies requires (1) a shift in scale, (2) an emphasis on a biology based on digitalized information, (3) the rhetoric of objectivity, and (4) data that no longer represent active donors’ bodies but rather a passive and quiet compilation of quantitative information. As Haraway puts it, “The object both guarantees and refreshes the power of the knower, but any status as agent in the production of knowledge must be denied the object.”4 We begin with the human-computer interaction.
On the first Monday of the year 2000, I walk across campus toward the lab. Evidence of Saturday’s revelry still litter the residential streets. Now recognized by security, I pass through the hospital security checkpoints, ostensibly as an employee who has forgotten his ID card. I make my way to Nora’s and Gary’s office; Nora’s new lab is not complete. Nora had not yet arrived, so Gary and I chat about the New Year hoopla. The office is not large enough to accommodate two people in the customary space requirements of professionals, so my spot next to the door allows an intimate observation of Gary’s and Nora’s work.
Gary’s and Nora’s relationship over the years had shaped their workspace expectations. Theirs was a constant back-and-forth. They would share e-mails, analyses, shoptalk, gossip, politics, and no small number of jokes and personal jabs. Gary, as the senior researcher of the duo, was immensely appreciative of Nora and her work. He would remark to me as if on camera, “My miserable life here would be nothing were it not for Nora. She’s the brains of the outfit. I’d be lost without her.” Gary’s and Nora’s desks were so close that they could hand each other a paper without moving their chairs more than six inches. Gary was the neater of the two. He would throw up his hands in resignation when Nora fumbled through the mountains of paper scattered in seeming chaos on her desk.
If Nora’s desktop was disordered, her virtual life was not. In fact, the multitasking she conducted in silico mirrors the fluidity of her desktop file system and the sociality that drives her work. Much of the time Nora was in her office was spent conducting analysis. Working on several databases at once, Nora would toggle back and forth from one to the other. There was a main hard drive and two others she drew from. Her postdocs and all her collaborators were also networked, and, if one was having difficulty, she would log over to their workstation to problem solve. In the midst of her database surfing, she would answer the phone, read mail, chat with Gary and me, read her e-mail, all the while picking away at her primary data set, which was usually the one closest to a deadline for publication or a meeting presentation.
Nora’s labor is structured around the social networks she creates. Hers and Gary’s face-to-face interaction mirrors the seamlessness of her collaborations with scientists from around the globe. Nora’s collaborative network relationships are enduring forms of sociality through which the diabetes consortium membership produce knowledge. Dispersed through geographic distances and reliant upon digitized collaboration, Nora maintains dozens of active professional relationships through several networks of which the diabetes enterprise is only one. All of Nora’s collaborators say their involvement was in part because of their relationship with Nora.
Scribbling notes on the back of whatever piece of paper is handy, Nora notes statistical likelihood scores, SNPs, or some other numbers of interest. There are hundreds of data sets that she draws from. Each data set is tied to a collaborative relationship within Nora’s social-professional network. Each time a set of population samples is genotyped, it is placed in its own data set. Nora then tests, checks, evaluates, and conducts simulations on the data sets. Day after day, between meetings, remotely from home, while on the phone, she runs various tests using FORTRAN. Nora writes the routines and subroutines in a series of cascading if-then statements that often fold into one another. Embedded into the routines are other software programs, some she developed, some developed by colleagues: None is off the shelf. In at least one instance, when they needed an off-the-shelf program to run a specialized query, she used her network to contact the programmers, who sent her the code to alter it. The software programs read and test the genotypes of the data set of interest described in alleles. Over and over again, she checks them for errors in genotyping, for expected inheritance patterns using the long-established Hardy-Weinberg algorithm for determining homo and heterozygosity.
Nora’s realizations come in bits and pieces. There is never a monumental moment when the computer finishes its computation and the data suddenly jump off the screen. For example, the realization that heterozygotes for one piece of a gene were at greater risk than homozygotes came during one of her analysis sessions. It was on that day that Nora caught an inkling that ethnic admixture might confer increased susceptibility to diabetes. Her realization occurred while toggling between windows on her screen comparing the heterozygotes with the homozygotes for each SNP. Nora uses SNPs as analytical categories by which to measure variation between siblings and populations. Instead of whole genetic variants, Nora is evaluating SNP variations and, thereby, isolating the susceptibility gene products within populations.
After programming her FORTRAN command routines, Nora examines the report generated by the queries she runs. On one occasion, she compared the values beneath each y-axis number and the corresponding homo- or heterozygosity pattern for each SNP. Across the x-axis were numerical names for the SNPs corresponding to their position relative to the marker used for genotyping. Genotyping is essentially a report of the presence or absence of a set of markers whose location and frequency had already been established.
I noticed during one of her analytical moments at her computer that Nora was comparing those patients who were homozygous for some SNPs while hetero for others. She noted to Gary, “It’s puzzling that there is no high risk with homozygotes. I put tables together, and if you take all polymorphisms across Calp-10 [a suspected risk molecule], there are four homozygous individuals who are controls for the high-risk allele, but there are none in the patients.” Nora did the comparisons between patients and controls for Calp-10, made a list, and handed them to Gary, who dispassionately entered the results in his laptop. Gary was the chart maker. Nora then repeated the comparisons with the g protein coupled receptor, another suspected risk molecule. This data would later become the key finding to appear in the Nature Genetics article of 2000.
Nora’s quantitatively oriented genetics work is the backbone of research collaboration into a complex disease like type 2 diabetes. Her genetics expertise is part statistical, part technological, and part social. “While I disliked it at the time, I am so glad that my advisor forced us to learn FORTRAN,” she recalled. In terms of her collaborative involvement, Nora has leveraged her multiple talents well. She is overworked, however. At any given time collaborators from around the globe are waiting for something from her. Nora describes these as debts, “I owe ‘Steve’ something,” she recently remarked of a British colleague. Her sense of responsibility to her work, to her colleagues and to the judicious use of resources is palpable. Even under the most stringent deadlines or uncollegial treatment by associates, Nora seldom gets flustered. “I’m too nice,” she confessed during a bitter battle over lab space with some associates.5
Gary gives plaintive voice to the demands upon Nora. Everyone wants something from her. A talk, a consult, an interview (me), analysis of a data set, help with study design, supervision of a project, results from a simulation. Though uncompromisingly supportive of her talents and professional interests, his “support” is often delivered as complaints about work that is past due. They are complaints that mirror, if not co-create, Nora’s debtor mentality. Important here is the way collaboration involves multiple exchanges of data, intellectual service, and computational time. The diabetes enterprise works as a collaborative venture because of the solidarities established through these exchanges. Almost everyone owes everyone something.
In true Maussian fashion, to belong to the consortium requires an initial transfer of data. I use the term “transfer” rather than “gift” because, as Maurer observed, Mauss’s theory of the gift wavered between the pure gift and the commodity.6 That is, on the one hand, a grand theoretical concept that defines social relations structurally in a series of solidarity-enhancing exchanges that operate seemingly outside of the calculated, adequated economic rationalities. And on the other, Carl is working with a wholly discernable calculus of value of the goods or services exchanged between rational interlocutors. Collaboration within the diabetes enterprise instantiates Mauss’s qualification of his theory that gift giving is simultaneously calculated and not calculated. A gift is a gift, not a utilitarian contract for certain return. However, Mauss observed that gifts operate within a cultural logic of exchange such that giving creates structural solidarities between those involved. Referring to the Trobriand Islands, Mauss observes, “This notion is neither that of the free, purely gratuitous rendering of total services, nor that of production and exchange purely interested in what is useful. It is a hybrid that flourished.”7 Owing Steve work, thus, is Nora’s externalization of the expected return of services for Steve’s contribution of a data set. And yet the giving of the data set was itself an act of professional solidarity to a trusted colleague whose services can make of it more than it was before. What is created in this exchange is the networked sociality of the consortium with both the rights and obligations of solidarity to the enterprise that is membership via data set contribution, and the hoped for outcomes of scientific discovery. It is an exchange across time and space, occurring between humans and their computational media.
Perhaps the most public forum for collaboration occurs at scientific meetings. At least three times a year consortium scientists from far-flung laboratories meet to share and exchange research developments at scientific meetings. Since 1937 diabetes has enjoyed its own scientific conference. There are dozens of conferences, mini-conferences, symposia, international meetings, and other gatherings that are organized to enable the exchange of ideas.8 Consortium meetings are scheduled to coincide with meetings that investigators already plan to attend.
Because I was interested in scientific collaboration, scientific conferences surfaced as important sites to document the ways that scientists from different disciplines and institutional contexts work together. Conferences are redolent with opportunities for ethnographic collaboration.9 Aside from the formal presentations, numerous informal events occur. My interests were in a specific diabetes consortium, its members, their collaborators, and the work they produced collectively. I was also interested in the array of topics next to which diabetes genetic epidemiology is situated.
The 1999 conference of the American Diabetes Association had some eight thousand participants; of those, thirty-eight hundred were clinicians, fourteen hundred were research scientists, and another fourteen hundred were diabetes educators. To remain focused, I stratified my participant observation as follows. I first wanted to follow researchers and interview them if possible. I looked up their names in the abstract books, went to their sessions, and noted their collaborative partnerships. Next, I used the thematic index to find other sessions that directly dealt with type 2 diabetes, genetics, and populations. I deliberately avoided the diabetes education sessions because I was most interested in genetic epidemiology.10 Occasionally, I would attend a session that was more general, something on physiology, diagnosis, or collaborative partnerships between, for example, Glaxo, Roche, or Millennium and the NIH. Because Francis Collins, then director of the NIH’s Human Genome Project, was still making his promotional rounds, I attended a few sessions in which he was featured, if only to capture data on how the US-Human Genome Project was being intellectually positioned vis-à-vis diabetes genetic epidemiology.
The advantage to a scientific meeting is that often researchers would frame their work differently for a broader and perhaps more skeptical audience. Such alternative framing requires researchers to better articulate the discursive context of their contributions to the intellectual project and their specific research. It also aided in my general comprehension of the particular method or finding being discussed. The disadvantages to scientific meetings are that they are immense and researchers are busy. Consortium meetings were almost always connected to scientific meetings because, logistically, it was most efficient. Meetings are events that simultaneously illustrate the possibilities of overcoming geographical distance and time limits. Scientists come to share results of a few months’ or years’ worth of work with diverse audiences from the United States, western Europe, Japan, and elsewhere.11
My first American Diabetes Association meeting occurred when I was conducting prefield explorations. I had recently failed to secure permission to work at Glaxo and placed all my field site hopes on Nora at the University of Chicago. If Nora was not amenable to my sustained observations, I would be forced to begin my field site search anew.12 After talking my registration fees down from $365 to $75,1 prepared to meet with Nora and somehow persuade her to allow me to come spend time in her lab. It was the 1998 ADA meetings and Nora and I would meet in the registration area. The registration tables were situated in one corner of a huge sunroom/meeting hall of the San Diego Conference Center.
We walked over to a series of about 70 banquet tables that occupied the front third of the hall. The interview began with me asking her about the consortium. During our initial phone call, Nora had relayed her feelings about why the consortium was needed and the right thing to do.
Me: What is your role in the consortium?
Nora: Gary and I think it ought to be done, and it’s the right thing to do. For one thing, you can’t work in this too long without realizing it’s gonna take big data sets, clever and hard molecular work. It’s foolish not to make optimal use of the data sets. I mean, taxpayers paid for these data to be collected. It’s a moral obligation to get the most out of it as you can.
I asked for any updates on the consortium work. How it was going? What were the setbacks, if any? Any results or publications? I then asked about biographical information. How long have you worked on diabetes? Where did you work before this? What did you work on before this? Next I described my project. I had a one-page abstract of my research that posed my research foci as follows: (1) scientific collaboration, (2) the use of ethnic populations in research, and (3) the shift from endocrinological to genetics-based approaches for a complex disease. Like a true teacher, Nora corrected the terminology a bit to make it clearer and more accurate. The exchange would be emblematic of many to follow.
After the interview, I set about describing my methods: participant-observation (i.e., purposeful hanging out and participation) at labs, meetings and field sites, and interviews. I asked if she would please ask the consortium if I could attend their meetings. At the end of our meeting, I thanked her and asked if she would consider having me conduct my fieldwork at her lab. She said that I would be welcome to visit her lab. When the interview concluded, I said that I would contact her again if I wanted to take her up on her offer. That afternoon, Nora had a steering committee meeting in which she asked if I could attend the full consortium gatherings. There were no objections, and I attended the consortium meeting the following morning.
The consortium meeting began at 7 A.M. Two years earlier the first such meeting also began early. What was planned as a one-and-a-half-hour meeting lasted six hours. “The interest was so high,” remarked one member. “We just kept brainstorming.” This meeting was more predictable, however. The business of the morning was to report on progress for chromosome 20 and clarify which chromosomes would be pursued next. The overall aims of the consortium were to conduct linkage analyses to search for susceptibility genes for diabetes. The task at hand was to construct a map that was based upon the genotyping work of consortium members. One statistician, whose nonprofit research institute offered the computer storage space for the immense data files, took the lead.
The statistician reported that he had just received some data sets, so no final Ch20 report was ready. He also reported that the files were so large that even without the new data, the software crashed. The data sets needed to be rescaled. Members discussed whether to alter the software, the data or the way the data was fed into the software. One researcher volunteered to call the software developer, who was not there, to ask about altering the software.
The second issue was how to fold in new data once the map was complete. The group decided that it would not redo the marker map but would add the collective markers to the new data set. The resolutions to the various issues were reached by dividing up the workload. One scientist would explore marker maps for the next chromosome. Another would scout the French reference database (CEPH), and one would evaluate the Marshfield clinic’s 8,000 to 9,000 pre-mapped markers. There were no overt disagreements, and so the motions and seconds and votes were performed with perfunctory resignation. The meeting ended with the agreement that the analysis subcommittee would address the remaining issue of how long and how many markers to use for the map of Ch20.
This seemingly mundane description of the collaborative practices planned and carried out at a consortium business meeting conceal a host of anxieties and potential conflicts. During my second meeting, after having interviewed several of the members, visited them in their labs, and spent months with Nora, a conversation occurred that I could not quite understand. I was accustomed to asking Nora or someone else for a breakdown after a technical conversation, or looking something up in published sources when I did not quite follow the conversation. This instance was different. Consortium members were clumsily deliberating about which chromosome to map next. The debate seemed to center on a set of markers that would be used for one of the chromosomal maps.13
Nora later explained what the hemming and hawing was really about. A senior but not central contributor to the consortium had invested years in the development of several markers on a particular chromosome that most consortium members no longer used because a set of better markers had been developed by someone else. To include the senior colleague’s genotypes would require the use of his markers. Consortium members were in a conundrum. They wanted the data from the colleague’s samples but did not want his genotypes or markers. The senior colleague, who was not present, would not be keen to share his data if his markers were not used. The tension was palpable, with a few at the table arguing on behalf of the senior colleague. It was resolved by someone volunteering to contact the absent colleague to discuss the possibility of regenotyping his samples, at considerable cost of time and money, as a way to validate his markers while at the same time enabling a standard set of markers to be applied to his DNA samples.
Though Nora was trained as a geneticist, her expertise draws upon quantitative analysis and computational tools. Biostatisticians, bioinformaticians, and quantitatively minded geneticists like Nora, are all central to the collaborative efforts of the consortium and the representation of their collective efforts. For this discussion, I will refer to such researchers as statistical analysts. These are the power brokers of the consortium. Their role in diabetes research is to search for and create pure and powerful data.
For a genetic finding of a complex disease like diabetes to be taken seriously, the data sets must be powerful. This means that at each stage of research, from the selection and screening of populations to the genotyping of the DNA to the analyses of each data set produced by each collaborator, each data set must adhere to strict rules. Nora’s job is largely to make the disparate data sets articulate with one another. The case for standardized data sets is narrated in terms of degrees of power and noise.
Much of Nora’s collaborative efforts involves “cleaning up” data sets to make the results more powerful. That is, getting one person’s genotyping information into useful, informative shape so that the data could be merged with a larger set or sorting through an old marker map with a newer one to make the case for one set of markers over another. The standard test of the power of a finding is expressed in its logarithm of the odds (LOD) score. A LOD score is a way to quantitatively analyze a data set to determine if a marker (a bit of genetic material) is linked (inherited together) with another. If so, then scientists can compare a known marker with unknown others to infer an association that may lead to a putative inherited disease marker. Conflicting findings between one researcher’s results and another’s, is often described as “noise.” Results of data deemed questionable figure prominently in discussions of data sharing and analysis. One consortium member puts it this way:
There’s a long list of reasons why studies might fail to replicate. The most discouraging of which is that it’s all noise. And I think a lot of it is noise. The separating signal for noise is the problem here, because I think a lot of these wiggles that you saw on these graphs are LOD scores going up and down, and a high LOD score is supposedly a signal. But I think a lot of that is noise, and teasing out. . . . I mean, the diabetes genes are there somewhere. . . . And teasing out the real signals, where there actually is a susceptibility gene from the noise is a challenge. Part of the reason there’s so little replication is because a lot of what we’re looking at is, in fact, noise. That is a challenge.
Pure and powerful data also requires that at each stage in data set construction, for each population and each experiment, “robust” standardization and sheer volume must be created. One statistician explains it in terms of the number of members from one family who participate in a given study. He notes:
If you want to get a genetic study funded, you need to say something about power. Reviewers always ask what is your response rate. Usually what lousy response rates will mean is that your families are going to be smaller if people don’t participate. Because they’re smaller, you’re going to get fewer relative pairs. Because you have fewer relative pairs, you have a loss of power.
Statistical analysts speak with an almost evangelical undertone in their desperate attempt to raise the standards of research. High response rates are one such standard. Others, which will be discussed below, include good genotyping, informative and current markers, and a well-ascertained population.
The collaboration is made problematic because each researcher is trying to leverage his or her labor for multiple ends. The specific goals of collaboration are only one demand placed upon their research. There are conflicts. In one instance, Nora was struggling with a researcher who had made a map of a chromosome with markers that were no longer informative. “You want to use everyone’s markers because they have a lot at stake in those markers being informative. But Ned’s markers are not dense enough and actually detract from the statistical likelihood [LOD] scores we see when we use more recently developed markers.” Some researchers are loath to give up their markers for the greater good of the consortium, and Nora must carefully negotiate each member’s professional and scientific needs.
The pressures for powerful data are immense. Attention to statistical power must begin with the very first piece of data collected for a project. Statistical analysts often deploy a proselytizing tone simultaneously convincing and recruiting scientists to their kind of robust data collection. At one ADA meeting, two consortium members, one from industry, the other a nonprofit research institute, gave an introductory talk on powerful linkage analyses. In this instance, power is defined as the number of times statistical simulations produce significant LOD scores—those 3.0 or higher, which means the likelihood of the observed inheritance patterning occurring randomly is less than 3 in 1,000. Responding to a question about small effects of some allelic variants, Nora remarks:
[It] is a hit-or-miss enterprise in terms of what you expect to find with any particular variants. There is a challenge, though, even if you have variants in front of you, you need a lot of DNA samples if you’re going to have sufficient power to detect a weak effect, and we all would like to see those weak effects because each one of them tells us about a pathway that may be involved. Even if it only accounts for a small fraction of the heritability for diabetes, you wouldn’t want to miss it.
Failure to conform to standard practices, use common diagnostic criteria, similar markers, common labels, and standard population selection protocols will also jeopardize the individual data set, which, when pooled with multiple data sets, then diminishes the power of the entire collaborative venture.
A senior biostatistician in the consortium, Franklin Akindes, addressed quality data as an issue of population selection and diagnostic criterion. In a talk titled “Simple versus Gold Standard Measures of Diabetes/Glucose Intolerance,” Dr. Akindes argued to the approximately four hundred who had come to his talk that getting to the bottom of phenotypes is of utmost importance. Diabetes science would be greatly improved, he argued, by evaluating studies by the phenotypic screen each study used. There are different tests for insulin resistance, among them fasting glucose tolerance versus the homeostasis index for fasting blood glucose. If research subjects are selected by different tests, or different cutoff points for the same test, then when the data are combined, the aggregated data set is compromised.
Akindes’ concerns strike at the epistemological foundation of the collaborative venture to find the genetic contributions to complex diseases. To wit, the presumption that genotype drives phenotype. Akindes was peering around the corner of the limits of genetic epidemiology echoing the growing scientific appreciation for the complex relationships between proteins and their environments. Genotypes are likely poor predictors of health outcomes. Even under strict experimental conditions, plant geneticists know, plants with identical genotypes will develop differently according to subtle changes in temperature, wind, soil, and a host of mostly unexplainable factors.14 The gene, or the expression of a protein as a result of a genotype, is increasingly understood as an interplay of natural and social environments, extracellular matrices, interactions between and among molecules, and overlapping functional molecules that get expressed as a phenotype in networks of context dependency.15
Franklin Akindes’ arguments stop short of calling the gene, as a determinative force for health, flat-out misguided. However, Akindes and his colleagues on the ADA panel did argue that attention to phenotype differences of all kinds were vitally important to advance the understanding of diabetes genetics. Panelists specifically mentioned that ethnicity, like other variables of interest to genetic scientists, should be interrogated. Akindes’ message at the time was that researchers needed to pay more careful attention to phenotype to fine-tune linkage analyses results. He argued that phenotypic subsets must be created by breaking out, layering, and comparing multiple phenotypes with multiple markers. This would, he argued, enable a better understanding of locus-phenotype-trait interactions. Akindes underscores the complexity of collaboration. But, additionally, his arguments, which were repeated at multiple conferences, are emblematic of the consistent reminders to diabetes researchers that ethnicity is a highly problematic variable. His calls for scrutiny of population identifiers reminds the diabetes scientific community that ethnicity is a poor shorthand for phenotype.
Akindes’ panel was only one of many that conveyed the seriousness with which diabetes scientist take their work. The ADA conference contained panels and papers on the best practices for taking blood, for sending blood, for processing genetic material, for determining which markers to use, for deciding which reagents to use, and for calculating how dense a map to make when attempting to genotype a particular chromosome. During the consortium meeting, some time was spent discussing which chromosome to map. Nora explained later that the next steps in such collaborations are complex and involve balancing the competing needs of individual researchers and their need to collaborate. For example, sometimes a collaborator’s genotyping is fraught with errors, or in the case described earlier, is good genotyping but the markers used are outdated. Nora and her biostatistician colleagues have the unrewarding task of evaluating data sets for any number of common or idiosyncratic deficiencies. There are countless things that could go wrong, and it is usually not determined until the genotypic information can be tested to see if it conforms to the inheritance patterns expected of the data set. The work to create powerful data sets is a constant negotiation between individual interests and group interests. Nora’s work illustrates how scientists and scientific practices occur across many institutional and social contexts, and thus researchers must routinely bring boundary objects such as categories of persons and genotypic patterns with them as a condition of collaboration.16
Scientists bring other things to the collaborative table. They bring their social relationships with other researchers and families, their competitive and other personality quirks, and, at least in one case, they bring an aversion to philosophy, as we shall see below. Before we assess the cultural significance of the narratives of power and noise in describing quality data, let us examine the ways the social and the ethnographic influence the dynamics of collaboration.
The science of diabetes has inspired a series of critical philosophical interventions, the most famous of which is likely Canguilhem (1991 [1966]).17 A more recent one warrants specific attention. In the early days of field-work, Gary initiated a discussion about Rabinow’s book French DNA.18 French DNA describes the moral and philosophical vexations of the use of biological material. It details the souring of a diabetes collaborative venture between Millennium Pharmaceuticals and the Centre d’Etude du Polymorphisme Humaine (CEPH). Rabinow proposes we look at science as a cultural construct that can be used to imagine something new or make sense of something old. Echoing decades of scholarly critiques of science, Rabinow argues that science must not be reduced to truth about nature. Blood in France, Rabinow explains, is a biosocial substance par excellence. The central story line is of the French reactions to their DNA being acquired by an American pharmaceuticals firm. Ownership and identity, in Rabinow’s account, are revealed as constituent elements of blood samples in France.
For the first time, argues Rabinow, the materiality and sociality of blood ruptures the ethical imagination about the potential partnerships. The national scandal detailed in Rabinow’s account illustrates how DNA acquired a French nationality. The tainted blood scandal of 1983-85, in which shortages of blood led to a system for prisoner donations and hence HIV contamination, was illustrative of how blood was imbued with nationalistic and altruistic value sets. French social scientists liked the idea of participation and belonging that donation enabled for prisoners. Prisoners liked the perks of donation (wine, food, time away from their cells, etc.), and the French government benefited by adding to its stock of blood supply. The issue, writes Rabinow, is that blood has a special status because it implies “the person who is the source and, in a more abstract manner, the representation that is made of humanity.”19
As a key figure in the diabetes world, on par with Rabinow’s informants, Gary was mostly interested in discussing the gossipy aspects of Rabinow’s text. I had first heard of “Francois’ book,” as Gary liked to call it, when Gary promoted it from the podium at an ADA miniconference dedicated to the genetics of diabetes. Francois played a major part in Rabinow’s multiactored account. Francois is a member of the consortium and was known by most of the conference attendees. Gary, Nora, and most of their collaborators attended. Researchers from Glaxo, Parke-Davis, and other pharmaceutical firms were present as well. There were scientists from land grant universities, the ADA, private research organizations, the NIH—including Francis Collins—and from several countries.20 Gary’s talk was on the story of the discovery of one of the diabetes genes, NIDDM1, and the most recent genetic findings that implicated a calcium protease, calpain-10. He joked with the audience and recommended they all get “Francois’ book.” It took me awhile to figure out that he meant French DNA, and I looked forward to eventually talking to him about it.
One day in the initial weeks of my visits to his lab, Gary beat me to the punch and asked what I thought of Rabinow’s book. I was not ready to discuss it with him. In fact, I had not sorted out my own thoughts about it and was reluctant to have the conversation with Gary so early on in my fieldwork. I thought to myself, “Times have changed when informants have read and [as I was soon to find out] have criticisms of ethnographic work that at least provisionally claims to discuss the informants’ lives.”21 I knew that Gary knew more about the topic than I did. What is more, Rabinow had not made the informant-collaborator-ethnographer cross-talk easy because of his robust engagement with French philosophy.
However, I was a visitor and an ethnographer. Thus, if Gary wanted to talk Rabinow, I was obligated to talk Rabinow. “He got the story fairly correct,” Gary told me, “but all that philosophy shit really detracted from the book.” Bracketing his pejoration, I acknowledged that for a reader like him the philosophical and historical arguments were not well articulated as part of the story of the power play between two central figures of the deal. I, of course, was more interested in Rabinow’s cultural analysis as an explanatory device than whether or not it fit with the narrative. This fit issue, was, after all, what Gary was commenting on. He wanted to be entertained by the narrative. He was.
Some weeks later, I told Gary that I had e-mailed Prof. Rabinow to discuss my work. I was now ready to defend his use of philosophy and to use it as a way to further explicate my work in his lab. I wanted to make clear to Nora and Gary that I, too, would be doing more than writing an entertaining read. My work would have philosophical aspects as well.22 Gary’s custom was to launch into discussions with me at his whim. In one instance, he wanted to discuss Rabinow’s Making PCR.23 Gary did not really want to discuss the book but, rather, what Rabinow had not said about the personal life of one of the scientists. Gary is a friend of an ex-wife of one of Rabinow’s informants, and hence I won’t detail Gary’s extremely partisan perspective. “ ‘Smith’ is a brilliant guy,” he said. “He could anticipate the end, but couldn’t do the day-to-day routine of seeing it through.” Then, in his customary self-deprecating manner, Gary continued, “I must be the simple molecular biologist because my science is about routine, making sure it’s perfect, each step, each experiment, over and over. I’ve got Nora, who’s the smart one. I’m the dumb one. My job is to keep her happy.”
This exchange illustrates Gary’s simultaneous evocation of the rhetoric of objectivity and its perversion. On the one hand, Gary critiques Rabinow for attempting to situate the French DNA scientific story within a broader context of meaning (i.e., “philosophy shit”). On the other, when discussing Making PCR, Gary desires to fill out the ethnography with details of Rabinow’s informants’ personal lives. As a text, Rabinow’s French DNA is an actor in the diabetes enterprise that required me to account for its influence. I do not mean to overstate the case of this one book. Rather, I intend to situate it as an estranged familiar in my attempt to understand the collaboration within the diabetes enterprise. Gary’s use of “Francois’ book” brings to life the way science narratives, ethnographic or genetic, are multiply layered texts with personal, scientific, and philosophical registers.24 Situating these registers into the knowledge produced by the enterprise is the task at hand.
“Situated knowledges,” argues Haraway, “require that the object of knowledge be pictured as an actor and agent, not as a screen or a ground or a resource, never finally as slave to the master that closes off the dialectic in his unique agency and his authorship of ‘objective’ knowledge.”25 We read above how an anthropological narrative sheds insight into the inner workings of the enterprise. There are, of course, dozens of narratives that influence knowledge production. Accounting for them all is not possible. However, for a condition such as type 2 diabetes, at this postgenomic moment, the objects of interest are the exact mechanisms that cause diabetes or, at minimum, the genetic contribution to a causal pathway.
On this score, Nora and Gary are appropriately modest. Their modesty expresses itself in several ways. Gary’s self-deprecation serves to both bolster his alliance with Nora and to remind everyone that Gary does not need to posture. He is a powerful scientist who can afford to self-deprecate. Such are the “structured dispositions” of those endowed with immense social, cultural, and material capital. More than this, Nora and Gary are both uncomfortable with the human genome hype. As Nora noted after Collins’s talk at the mini-genetics conference, “You know, I’m certain that we’ll all look back at what we are doing today and see that it is sloppy, crude work. But it is the best we can do for now.” Nora and Gary both reiterate the modesty of their contribution to understanding diabetes at every opportunity. “The really hard work has yet to be done,” remarks Nora to me in our first meeting. This is a refrain that Gary also frequently deploys. Both she and Gary speak of the research that comes after the genetics work has been done, after the susceptibility genes have been identified. They refer to the work to characterize the triggers, pathways, and other physiological mechanisms of diabetes. They speak of physiology.
Nora’s and Gary’s genetic epidemiological research effectively redirects the research efforts from statistical susceptibilities derived from genetic analyses to the effects of calpain-10 at the cellular level. Diabetes physiology is most frequently described as a condition affecting processes of blood sugar regulation that involves insulin secretion and insulin reception by muscle and fat tissue. Within the accounts of the physiology of diabetes, which is an outgrowth of the genetic epidemiological guidance offered by Nora, Gary, and colleagues, we read of several shifts in scale.
The physiology of diabetes is a world inside of a world. It is imbued with a taxonomic polyglot and a complex interactive function for each actor. To explain this bioscape requires the narrator to mediate from community to the universal body, to specific body (phenotype in genomic parlance), to genetic signs of physiologic possibility expressed as statistically derived susceptibilities, to an enzyme (calpain-10) interacting with an uncertain series of endocrinological processes.26 To represent this world requires the creation of models that mediate between extracellular and intracellular systems “conveniently characterized as in vivo, in situ, or in vitro.”27 I would add to this in silico to reflect the ways Sun County lifeworlds are powerfully reduced to data sets transmitted via e-mail and run through countless software programs. This form of symbolic abstraction of Sun County DNA donors extracts the meaning of diabetes from those most affected by it. As Emily Martin writes, “In the change of scale, something very minute, discovered by science, comes to play a deciding role in human questions or concerns that are very large.”28 Let us further examine the narratives of physiology for the consequences of such shifts in scale in diabetes science.
A most recent turn of research into diabetes has been the quantification of genetic codes for the purposes of locating susceptibility markers that might guide physiological research. But genetics is before physiology, or rather genetics informs physiology by pointing to an ever finer biological mechanism that may underlie the phenotype, the measurable manifestation of glucose metabolic impairment. Physiology is the study of basic biological processes and functions. The physiological research spawned by the work of Nora and Gary, as well as most other physiological characterizations, are functional. There are dozens of molecules, substrates, elements, tissues, genes, and hormones that interact. Research papers on diabetes slice these actors into different pieces, isolate, experiment, measure, and test. They use specific cell types, such as pancreatic islet cells, skeletal muscle, or adipose tissue, or tissues of various species, in this case various mouse and rat models.29
The polygene discovery pointed to a gene that codes for an enzyme, calpain-10. Calpains are a common cytosine protease (calcium-activated neutral proteases) expressed everywhere in the body. Researchers are unsure what calpains do. A year after the polygene paper, Gary handed me two unpublished manuscripts from his collaborators. He had tipped them off about calpain-10 when the polygene results were quite preliminary. His colleagues had busied themselves in preparing publications about calpains. In the manuscripts, the researchers hypothesized that calpain regulates insulin secretion by modulating the movement of insulin secretory granules through the plasma membrane in the beta cell. In other words, when calpain-10 was inhibited, beta cell membranes released less or more insulin depending upon the duration of exposure to the inhibitor. The modulated amount of insulin made available by the beta cell is further affected by the hypothesized caplain-10 function on glucose uptake in muscle and fat tissue. Researchers found that in the presence of the calpain-10 inhibitor, muscle and fat tissue had impaired glucose metabolic responses, narrated as glucose utilization and oxidation. They hypothesized that impairment in the signaling pathway, a message system to the cells to use or break down glucose, occurred as a result of the inhibitor-induced reduction of calpain-10 action. What interests us here is not the truth claims of these hypotheses. Rather it is the metaphor of impairment and error.
The physiological narratives of impairment or error have a long history within diabetes science. In The Normal and the Pathological, Canguilhem scrutinizes the philosophy of error inherent in biological life sciences. Deconstructing the physiological science of diabetes from the nineteenth century, he exposes the philosophical roots of the normal and pathological. What were once quantitative measures in excess of normal physiological functions have now found expression in finer and finer understanding of what those measures would examine. Canguilhem argues that the model of pathology has shifted from that which is different from the norm to that which is erroneous. Canguilhem traces the origins of the concept of pathology from early French physiology, in which pathology was but a quantitative deviation from a norm. However, the norm is in fact elusive. The final analysis is that “normal” is a context-bound condition that changes according to a dynamic between the environment and the organism. Disease, he concludes, is really just another way of living.
Canguilhem orients the present discussion toward a critique of the pathology narratives of diabetes. If we do not take for granted the narratives of “normal and pathological” metabolism and endocrinological function, we are permitted to disassemble the layers of cultural meanings that surround the condition referred to as diabetes. We are enabled, for example, to interrogate the narratives of diabetes physiology and the practices of knowledge production that generate them. Of Canguilhem’s method, his student Michel Foucault expresses an epistemological modesty applicable to both subject and objects of inquiry. He writes:
The history of science can consist in what it has that is specific only by taking into account the epistemological point of view between the pure historian and the scientist himself, this point of view is that which causes a “hidden, ordered progression” to appear through different episodes of scientific knowledge: this means that the process of elimination and selection of statements, theories, objects are made at each instant in terms of a certain norm; and this norm cannot be identified with a theoretical structure or an actual paradigm because today’s scientific truth is itself only an episode of it—let us say provisional at most.30
In this light, the epistemological juxtapositions that follow are provisional representations of diabetes. They are meant to explicitly illustrate a moment, an episode of knowledge production, within a sociocultural context. To reiterate what has been stated elsewhere, the project of epistemological critique is not to achieve a superior understanding per se as much as to practice playing “other cultural realities off our own in order to gain a more adequate knowledge of them all.”31 This analytical strategy enables a kind of conceptual agility through which multiply assembled concepts can be used to explain diabetes knowledge production.
The Mendelian gold rush of single-gene phenomena is over, and an infinitely more complex research proposition has emerged. In the parlance of the diabetes enterprise, diabetes is referred to as a complex disease because to understand it biologically will require sets of tools and analytical strategies that must also manage the environmental triggers while searching for genetic ones. The complexity cannot be overstated. The trajectory of structure-functionalism in biological sciences appears in the history of the milestones of diabetes science where hormonal mechanisms lead to cellular functions, which lead to genetic localizations, which lead back to proteins and cells. Table 2 shows how physiologists first isolated the hormonal mechanisms responsible for glucose regulation (insulin), then moved to the cells that produce insulin followed by the current search for the genetic material responsible for a diabetic phenotype.32 Genetics researchers expend years of energy and capital just to give molecular biologists a specific physiological target, the function of which, in turn, requires years of energy and capital to figure out. Researchers remark that each element of this complicated set of factors contributes a small portion of the risk for the disease. Postdocs often lament the complexity of their task after Gary and Nora poke holes in their research presentations at lab meetings and journal clubs. Citing both Wes Craven’s popular Nightmare on Elm Street and James Neel’s (1976) Diabetes Mellitus—A Geneticist’s Nightmare, several postdocs after a lab meeting exclaimed, “The nightmare continues.”
TABLE 2 MILESTONES IN DIABETES RESEARCH
1877—Claude Bernard’s Lectures on Diabetes and Animal Glycogenesis details his physiological postulates that diabetes’ symptoms are but variations on normal physiological states (in Canguilhem 1991: 68). 1920—Banting and Best isolate islands of Langerhans in the pancreas. 1921—Discovery that pancreatic extracts lower blood sugar. Insulin is discovered. 1962—Neel’s “thrifty gene” hypothesis 1971—Insulin receptor defined 1972—Beginning of recombinant DNA era 1975—Links between HLA and diabetes susceptibility proposed 1977—Insulin gene cloned 1985—Method for amplifying DNA, polymerase chain reaction (PCR) developed 1991—Maturity onset diabetes of the young (MODY) 1 mapped 1996—MODY 2 and 3 genetically mapped 1999—Non-insulin-dependent diabetes mellitus (NIDDM) 1 and NIDDM 2 genetics 2000—Polygene discovery |
SOURCES: Dr. Morris White, Lily Lecture, ADA meetings in 1999; Canguilhem 1991 [1966]; Feudtner 2.003.
The nightmare is made of metaphors that have yet to be made material. Leys Stepan and Jordinova argue separately that science is built on metaphors and analogies that mediate between representations of nature.33 A metaphor evokes association, writes Leys Stepan, and “permits us to see similarities that that metaphor itself helps constitute.”34 But these are representations of the natural world, not direct and mirroring descriptions, argues Jordinova. Their power lies in the ability to convey complex ideas to different constituencies in ways that rhetorically conflates representation with description. In situations where the natural and the cultural are still being contested, as in Gary’s postdoctoral research projects, nightmares are appropriate representations of the uncertainties of diabetes knowledge.
In a similar way, Haraway argues that between 1920 and 1940, functionalism and systems theories finally achieved a legitimacy that supported the autonomy of the social sciences in universities.35 It was the period of the birth of social engineering, in which the science of society was used to speak of the rational management of populations. At the same time, primatology and animal sociology deployed experimental techniques upon (against) populations of animals through which conclusions about the human social order were produced. The conclusions drawn were that the “true social order must rest on a balance of dominance, interpreted as the foundation for cooperation.”36 Emily Martin likewise illustrates how metaphors of war and flexible late capitalist production influenced the field of immunology.37 Researchers and the public alike came to view immunity as a series of pitched battles against enemy intruders such as microbes and viruses that were imminently flexible in their offensive and defensive maneuvers.
Haraway, Leys Stepan, Jordinova, and Martin illustrate the way meanings reflect and emerge from the metaphors and analogies used to describe the objects of scientific inquiry.38 Far from sociologically neutral, the cultural work of scientific mediation39 calls our attention to the relationship between scientific ideas and approaches and the context of their development. The case of diabetes science is no exception.
At its core, insulin and blood glucose are the central metaphors for explaining the pathological condition called type 2 diabetes. As one National Institutes of Health Web site reads, “Type 2 diabetes is a condition characterized by high blood glucose levels caused by either a lack of insulin or the body’s inability to use insulin efficiently.”40 What interests us here is the persistent metaphor of biological functionalism used to explain diabetes, a metaphor that remains unchallenged even amid uncertainty.
The history of biological sciences is often narrated as biologists’ attempts to “peel back successive layers of organization to ultimately reveal the molecular interactions that take place within the living cell.”41 Traditionally, physical structures were depicted in relation to their function within the system (e.g., heart to circulation). By the 1830s this structure-functionalism had moved into the microscopic. Things too small to be observed by the naked eye were proposed as the basic units of physical structures. Organs were composed of cells, and cells were composed of even smaller functional parts. In the 1930s, when Pauling’s atomic models for molecular chemistry were put forth, the structure-functionalism moved precipitously into the molecular realm. For genetics, the functional terrain was that of heredity, which moved from Darwin’s populations to Mendelian trait theories, to chromosomal action, to biochemical and DNA interactions, and then to molecular structure.42 While certainly not a linear process, as the present case will illustrate, the developments moved from smaller to smaller apace with the technological capacity to “see” into the deepest and presumed most basic forces behind the physiological world.
Consortium members’ scrutiny of the data sets, protocols, and markers still leave intact the presumption that failures of biological processes are responsible for diabetes. That is, the failure to produce or use insulin, currently narrated as genetic regulation and receptor function, respectively, are blamed.
The metaphors of resistance to a regulating force which results in an excess of energy is a trope of modern governance par excellence. As Jordinova illustrates, the social imaginary is tapped into by scientists as a kind of shorthand explanation.43 In a process she terms “mediation,” Jordinova demonstrates how, through the use of language and semiotically open representations of nature, the historical specificity of a scientific truth claim can be assessed.44 That is, there is a dialectical relationship between the world and accounts of it that explains the ways science and culture are coproduced. For diabetes, it is worth noting that the key physiological processes responsible for diabetes are reduced to a biochemical interaction between two molecules: glucose and insulin. Glucose is a basic sugar and is most often narrated as the body’s source of energy, while insulin is described as the trigger or regulator of the body’s use of glucose. Additionally, diabetes and obesity are linked in as yet unknown mechanisms, but an excess of fat is thought to contribute to “insulin resistance” through an immune inflammatory response.45 Thus, the key metaphors of diabetic pathology describe the failures in the body’s use of energy.
Reminiscent of Martin’s analysis of war and labor metaphors in immunology in the 1980s through 1990s, diabetes metaphors of resistance and energy use evoke the economic transformation from production to (over)consumption in particular, but not exclusively, related to the economies of sugar.46 In the narratives of diabetes, the pathology occurs because the body resists the regulation of its source of energy. The metaphor of resistance mediates a condition similar to unruly workers or colonial subjects resisting the disciplining power of manual labor or external rule, respectively. While major labor and independence movements predate the metaphors of insulin resistance, it coincides precisely with the consumption-driven market logics of neoliberalism circa 1980 and forward. Too much glucose, after all, is predominantly delivered via the overconsumption of food relative to caloric need. Thus, the pathology occurs when the body resists the power (energy) derived from overconsumption in both the literal (ingestion) and the economic (buying power) senses. In this metaphoric narrative, consumption is the energy behind privatization, capital market liberalization, and social reforms—the key ingredients of neoliberalism.47 Insulin resistance thus might better be understood as the embodied expression of the contradictions of these market logics. The former is the overconsumption of harmful and substandard foods that have resulted in premature death from obesity and related illness. Of course the two forms of consumption are related.48
Zooming in to the objects of scientific inquiry within the diabetes enterprise reveals a search for the functional mechanisms of glucose resistance: that is, functional activities at the level of molecular chemistry for which a “therapeutic agent” (read drug) may be developed. To the possibility of having a finding turn out to be useful in developing a drug, one genetics lab director said emphatically, “I wish! That would be great.” Yet, what does the case of calpain-10 specifically tell us about our world? That it is complex and interactive. That the production, distribution, and consumption of insulin occurs through a concatenation of secretion, signaling, transport across boundaries, reception, and utilization of vital elements. Also, as in the dominant metaphor of controlling diabetes via constant glucose monitoring, the physiological mechanisms of control are also central to the functioning of the system of glucose regulation, cellular capacitance, and secretory potential.
Beyond the mechanistic functionalism of these key metaphors lies the making of data sets that are required for diabetes science.
For instance, within the diabetes enterprise, we read from the quantitative geneticist’s presentation, that the power of research depends both upon the sheer volume of various research factors and with a reduction of noise in data sets. Noise is that which impairs the standardization of data. Power is data with the highest N (response rates, number of samples, individuals, likelihood scores). Noise is a term used to describe data sets that contain errors or are weaker than wished for variations within a sample of blood, a genotype, or results from a computer simulation. The more populations, the more DNA markers, the more SNPs, the more controls, the more randomization, and the least noise, the more powerful the data. Power, thus, in this light signifies a kind of mass efficient silence.
Taking a coproductionist tack on these biological and social narratives enables us to juxtapose the social context of the lives of DNA donors with the description of a good statistically “powerful” data set with “low noise.”49 I am also reminded of Thacker’s insight that biomedia conditions embodied subjectivities through technology, as well as social, cultural, and political mediation.50 Of the research subject whose data points appear in computer simulations, Thacker writes, “The body of the subject is therefore always scripted in part by scientific-medical modes of knowledge production.”51 In this light, a powerful and noise-free data set is at once large and quiet, silent and bulky, noiseless and numerous.
Haraway reminds us that the objects of scientific knowledge are intimately linked to their broader social worlds. She writes, “Accounts of such objects can seem to be either appropriations of a fixed and determined world reduced to resource for instrumentalist projects of destructive Western societies, or they can be seen as masks for interests, usually dominating interests.”52 Thus, if we travel back out of the microcellular, out into the world of “skin and bones,” and apply the notions of expanding and silencing power, the critically social nature of diabetes research reiterates a familiar theme. In conversation after conversation, grant after grant, paper after paper, DNA donors are data points, objects of research, not actors. Participants are “schedules,” “controls,” or referred to by the title of the research grant, “a GR” for genetics of retinopathy and so on. As Carl relays the history of the research, “I looked for who carried the genes. Gary focused on the physical map of chromosome 15. We [geno]typed the larger population.” A docile population that endures generations of social inequality, the ostensible effects of the excesses of capitalist expansion, is required for diabetes knowledge but only as a silent mass.53 Like the shift between cyberspace and bricks and mortar,54 the political implications of the assemblage of subcellular research complexity with the conditions of life for DNA donors cannot be sundered. In other words, biopolitics55 and diabetes science are hand in glove.
In this postgenomic era, the collaborative practices of diabetes scientists represent scientific knowledge production writ large. Trans-and interdisciplinarity is now de rigueur for the sciences funded by the NIH.56 Collaborative practices hold the potential to transgress the dualism of mind-body, human-machine, and good-bad. Ethnographically unpacking technoscientifically infused social practices is both a method and theoretical orientation. It is also an expression of the consequences intended for this project. It orients the reader toward what is at stake in the research endeavor presented here. Haraway expresses what is at stake as follows:
Taking responsibility for the social relations of science and technology means refusing an anti-science metaphysics, a demonology of technology, and also means embracing the skillful task of reconstructing the boundaries of daily life, in partial connections with others, in communication with all of our parts. It is not just that science and technology are possible means of great human satisfaction, as well as a matrix of complex dominations. Cyborg imagery can suggest a way out of the maze of dualisms in which we have explained our bodies and our tools to ourselves.57
I am less interested in the imagery of the cyborg than I am in Haraway’s provocative suggestion that technoscientifically infused social relations can be both oppressive and liberatory. Hence, in spite of the critical intervention that as ethnographer-cum-cultural critic I hope to advance here, it would be a mistake to interpret this account as either a simplistic apology or as dismissive antidiabetes science. To be sure, the tensions between the legacies of racism and the humanistic impulses of implied or overt antiracism that appear in the diabetes scientific enterprise are not contrivances. They are constituents of the vexations of (1) diabetes science, (2) this ethnographic project, and (3) the society in which both were produced.
Finally, I have aimed to illustrate the ways that cross-disciplinary collaboration requires a kind of articulation work characterized by the regulation of data and which affirms a particular social order.58 Articulation work is often an invisible facet of cooperative work and is concerned with managing contingency.59 In Nora’s illustrative case, articulation work manages the ontological frameworks that define diabetes as foremost gene based and thus an enterprise that requires rich and extensive data sets. Recall that the impaired metabolic interactions involved in diabetes are often narrated as insulin resistance. A person produces ample insulin but her body resists it. I am here interested in the proposition that insulin resistance may be a corporeal manifestation of a broader social or occasionally political resistance to the indignities of social inequality—a resistance that is metaphorically reflected as massive silent data sets. In this light, the metaphors of powerful data afford an explanation of diabetes science that cognitively and perhaps politically requires the silencing of the ever-expanding population of Mexicanas/os and other poor populations with elevated risk of diabetes in the United States and elsewhere.60 To open the metaphorical frame more broadly, the mass silences required of good data and good border subjects fit the needs of market-driven solutions of consumer capitalism under neoliberalism. It is a situation in which markets (agribusiness, contraband, and fast food) are leading forces for consumption in a region where ownership (land, contraband) are freighted with human suffering. We ignore the links between the narratives of insulin resistance and the social context of lives affected by diabetes at the risk of trivializing both. To expound upon this proposition, let us turn now to the ways DNA data sets operate as a kind of currency within the networks of diabetes knowledge production.