Sorting Things Out: Classification and Its Consequences

From AcaWiki
Jump to: navigation, search

Citation: Geoffrey C. Bowker, Susan Leigh Star (2000) Sorting Things Out: Classification and Its Consequences.


Tagged: Sociology (RSS) NatematiasGenerals (RSS), sociology (RSS), labor (RSS), classification (RSS), ethnography (RSS), ScienceAndTechnologyStudies (RSS)


"Each standard and each category valorizes some point of view and silences another. This is not inherently a bad thing--indeed it is inescapable. But it is an ethical choice, and as such it is dangerous--not bad, but dangerous" (5-6)

In Sorting Things Out, Bowker and Star promise to ask three basic questions (9):

  • "What work do classifications and standards do?
  • "Who does that work?"
  • "What happens to the cases that do not fit?"

Overall, they offer an approach to thinking through the political and societal implications of classification schemes at all parts of the classification process, from the people whose lives are inscribed through classification to the people making decisions from those classifications, to the people doing the labor behind the scenes to do the encoding.

Note: this summary only includes some of this book's chapters. You can help improve it by extending the summary to include the rest of the book.Natematias (talk) 15:40, 17 March 2015 (UTC)

In the introduction, Bowker and Star also define two key terms that underpin their work:

  • classification, a "spatial, temporal, or spatio-temporal segmentation of the world," with the following (ideal) properties:
    • "there are consistent, unique classificatory principles in operation" (such as lists sorted by time or hierarchies)
    • "The categories are mutually exclusive" where a thing cannot be in two categories at once: "a rose is a rose, not a rose sometimes and a daisy other times."
    • "The system is complete" and addresses all the things it sets out to describe
  • standards are "any set of agreed rules for the production of (textual or material) objects
    • spans more than one community or practice
    • "are deployed in making things work together over distance and heterogenous metrics"
    • "Legal bodies often enforce standards"
    • "There is no natural law that the best standard shall win"
    • "Standards have significant inertia and can be very difficult and expensive to change"

This book is organized into three sections.

The first part of the book unpacks the idea of "classification systems" in large-scale infrastructures through an intellectual history of the development of medical classification systems leading to the International Classification of Diseases (ICD). To examine this, the authors traced records at the UN and WHO, as well as followed up in personal files held by people defining the process. They pay special attention to moments when definitions are contested for business, religious, or ethical reasons. Throughout, they they focus on unpacking individual decisions or classifications that reveal rich details about the nature of these systems. Finally, the authors imagine how the Internet may be changing the use and role of classification systems, through interviews with students on their use of the Internet for writing papers.

Part two "looks at two cases where the lives of individuals are broken, twisted, and torqued by their encounters with classification systems." In the case of tuberculosis, the authors analyze historical records of patients checked into sanatoria during treatment. They pay close attention to moments when gender and stereotypes influence the details of patients' incarceration and release. In the case of race classification under Apartheid in South Africa, the authors describe "cases of mixed-race people who applied to be reclassified after their initial racial designation by the state," using their stories to "illuminate the underlying architecture of apartheid."

The third part examines "how classification systems organize and are organized by work practice," by following "a group of nursing scientists based at the University of Iowa... to produce a classification of nursing interventions." To collect data, the authors observed the process of developing these standards and unpack the implications of the system created.

Methods: Some Tricks of the Trade in Analyzing Classification

In this chapter, the authors describe methods for studying classification, with the goal of "resistance as a reading of where and how political work is done in the world of classifications and standards, and how such artifacts can be problematized and challenged" (49). The methods used here build on Star's prior work on the ethnography of infrastructure[1]. Throughout the book, they develop the idea of 'torque' between systems of classification and the wider conditions that reveal the existence and nature of those systems.

The authors start with a working definition of infrastructure (alongside several others, including [2]):

  • "A historical process of development of many tools, arranged for a wide variety of users, and made to work in concert"
  • "A practical match among routines of work practice, technology, and wider scale organizational and technical resources"
  • "A rich set of negotiated compromises ranging from epistemology to data entry that are both available and transparent to communities of users"
  • "A negotiated order in which all of the above, recursively, can function together."

as well as having the qualities of (238):

  • Embeddedness... into other structures"
  • Transparency, as in "ready to hand"
  • Having a reach or scope, reusable across instances and contexts
  • It is learned as part of membership and "associated with communities of practice"
  • Multifunctionality, supporting a wide set of agendas.

The chapter offers four "methodological themes" for considering infrastructure, that work like "an infrastructural inversion" in a "struggle against the tendency of infrastructure to disappear (except when breaking down)."[3] These four themes all share a common practice of "inversion," finding "examples of counterintuitive, often humorous struggles with constraints and conventions in the crafting of classifications (36).

  • Ubiquity - research on classification will "step back from this complexity and think about the issue of ubiquity rather than try to trace the myriad connections in any one case" (38). One way to do this is to look at "residual categories" such as "other."
  • Materiality and Texture - since classifications "have material force in the world" (39). To observe this, one trick is to "question every apparently natural easiness in the world around us and look for the work involved in making it easy" (39). Watching a programmer write code, we might look at the keyboard, the programming manuals, etc, and then contextualize those objects and that work by taking "quite literally the kinds of metaphors that people use when describing their experience" (40)
  • The past as indeterminate - "revisions... mean the introduction of new voices-- many possible kinds of interpretation... the indeterminacy of the past implies recovering multivocality; it also means understanding how standard narratives that appear universal have been constructed" (41). To study this, we might examine history, looking at the categories used at the time, and putting them in conversation with categories from other periods, such as our own.
  • Practical politics of classifying and standardizing - the process of "arriving at categories and standards, and, along the way, deciding what will be visible or invisible within the system" (44). To examine this, researchers could observe the negotiations or records of these negotiations.

Across this work, the authors argue that "both words and deeds are valid kinds of account;" by focusing on what people say they're doing, we might miss their actual actions, while focusing too closely on actions could risk taking on the limitations of the actors' perspectives.

Classification and Large-Scale Infrastructures

The Kindness of Strangers: Kinds and Politics in Classification Systems

The ICD as Information Infrastructure

Classification, Coding, and Coordination

In this chapter, the authors set out to describe the work of coordination represented by classification systems, since "in many ways software is frozen organizational and policy discourse" and the ICD classification they're studying is "an important infrastructural component of medical and epidemiological software."

To begin, the authors consider the ICD system as a list, situating it in the history of human use of lists, where

  • "there is a permanent tension between attempts at universal standardization of lists and the local circumstances of those using them"
  • "this tension should not and cannot be resolved by imposed standardization because the problem is recursive"
  • "ad hoc responses to standardized lists can... be mined for their rich information about local circumstances"
  • the ICD is "the sort of object that must satisfy members of communities or organizations with conflicting requirements" (139)

The authors offer descriptions of the kinds of conflicts arising from the ICD, from differences in how the data is collected to how it is coded, to how it is processed. For example, countries with confidential records of causes of death tended to have higher counts of stigmatized diseases. Coding issues might be related to differences in how countries weight multiple causes of death, or the priority they give to diseases that are lower priorities. Law enforcement, religious groups, pharmaceuticals, insurance companies, and public statisticians all faced very real implications from the details of the classification and how it was coded.

To address these conflicts, the ICD included a variety of strategies:

  • "distributed residual categories" e.g. "other" fields. "Their advantage is that they can signal uncertainty at the level of data collection or interpretation under conditions where forcing a more precise designation could give a false impression of positive data" (150).
  • heterogenous lists that simply appear to be a single ICD standard, all varying by the part of the body addressed, the origin of the disease, the tests involved in diagnosis, and ethical-political concerns.
  • parallel different lists. Rather than create a single universal list, the ICD has published guidelines for carrying out processes to "fork" and modify the list for different purposes, requiring an audit trail for connecting a given list to the main ICD.
  • complementary localization. In some cases, list-making is avoided to prevent incompatibilities in classifications between countries from preventing those countries from understanding their own health issues.
  • convergent bureaucracy, e.g. training people how to use the classification system, and potentially spreading the associated bureaucracy systems through that training.
  • Computerization, where computer memory systems and algorithms are used to resolved differences
  • Standardized forms across settings

The authors share policy implications from these observations:

  • "it is unrealistic and counterproductive to try to destroy all uncertainty and ambiguity in these sorts of infrastructural tools." (157)
  • "no such tool can be defined once and for all. They are always the products of continuing negotiation and change."

Classification and biography, Or System and Suffering

Of Tuberculosis and Trajectories

"With tuberculosis, the body is constantly in motion and the disease is constantly in motion... The disease may be localized or spread throughout the body. The state or general condition of the body and of the person's life both enter into the treatment regime, which may take months and historical has often taken years, sometimes a lifetime" (166)

This chapter focuses on the classification of tuberculosis, which has resisted classification at many levels (the authors describe it as having an "amodern nature: culture, nature, discourse, and infrastructure") (173). To untangle this, the authors draw from the "body-biography-trajectory" theory[4], an analysis of studies of tuberculosis sanatoria and hospitals, approaches from the study of chronic illness, and research on AIDS[5].

  • Classifications tend to omit time, despite many of the things we classify changing over time.
  • Classifications tend to focus at a single scale or unit of analysis, but is it the right one? Although Tuberculosis has declined, there is an ongoing argument over whether the decline has resulted from better treatments, isolation of carriers, or natural selection?
  • When classifications include time, they often place people on a trajectory of treatment, but Tuberculosis is unpredictable in this way
  • The causes of tuberculosis are varied, as are the places it attacks

Notably, the committee classifying tuberculosis judged it impossible to classify, settling for "a well-considered compromise of the views of outstanding clinicians" (Diagnosis 1955 qted on 175). Recorders of data have been faced with the problem of converting this system to a single record,

This uncertainty adds considerable strain to the experience of patients, and the classification of their condition becomes a major point of interest and contention. Bowker and Star offer a reading of a Thomas Mann's novel The Magic Mountain[6] and an ethnography by Julius Roth[7] of tuberculosis hospitals to describe the effect of the struggle to classify on patients' sense of time, their sense of their future, and their perception of these metrics. These uncertainties become contested across the sanatorium, with one patient submerging her thermometer in tea to raise her temperature and maintain her life in the hospital. The doctors devise a "silent sister", a thermometer with no markings, to prevent her from influencing the metric. The authors also describe the "house-of-mirrors effect" that occurs when patient and doctor categories disagree, and privacy requirements prevent the doctor from making comparisons to others (184).

The authors offer a "model of the TB landscape" in conversation with other models:

  • the Body-Biography trajectory method by Strauss and Corbin, one approach to TB is to imagine the body and biography of a person in a "body-biography chain" of "intertwined trajectories" between "sick-well" and "able-disabled" within wider structures(186)
  • multiple classification systems themselves, as "the rich topography of body and biography intercalates with a bureaucratic infrastructural typology" that are negotiated in each case (191)
  • the degree to which time is involved, from the institutional medical knowledge, which tends to focus less on time, and knowledge closer to the patient, where "time seeps into the classification systems that get used" (192)

The Case of Race Classification and Reclassification Under Apartheid

As classification systems become embedded in bureaucracy and grow more invisible, their politics become more entrenched. In this chapter, Bowker and Star consider the case of race classification under Apartheid as a form of banal evil[8]. The chapter opens with background on Apartheid-era laws in South Africa, including the Population Registration act and Group Areas Act, which controlled every part of life on the basis of race. This system classified people into four basic groups, with "natives" or "Bantu" people subdivided into subgroups, each with associated homelands. Interracial sexuality banned and those bans enforced. Identity cards and pass books were required to move around, and anyone lacking a pass risked arrest.

Classifying race proved difficult, and by 1985, South Africa had amassed 3,000 pages of law defining race. In this context, both a scientific and a "street" definition of race: the official census-based race assignment with an appeals process and the daily judgments of "police, and tram drivers to judges." A person's classification was dependent on the classification of parents (typically giving people the "lower" classification), and if you were a woman, the classification of your husband. The authors tell the story of Vic Wilkinson, who was reclassified five times as changing circumstances required adjustments for work and family. The criteria of reclassification changed over time, starting with "appearance" and "general acceptance" rather than ancestry (208). Bowker and Star point out that like the case of TB, this classification put people in limbo for years, where doing things like attending school or getting married would be taken as evidence towards the classification.

The chapter briefly describes technologies of classification, which were used alongside folk theories of race, from appearance, living conditions, the softness of earlobes, the capacity of one's hair to hold a pencil, and color scales, with the burden of proof on the person being reclassified. It also briefly describes the practice of passing[9], focusing especially on a process in some schools for holding hearings on the race eligibility of individual students, which often worked in favor of the student. At the level of institution and administration, classification was negotiated on the basis of implications for the school and implications for the official, related to how other parents or other officials might respond (215).

The chapter also describes a series of individual cases that "torque" the classification system. One neighborhood circumvented the "general acceptance" clause by signing an affidavit that their neighbor was white. One group of Afrikaans-speaking light-skinned people were classified as bantus and forced to learn indigenous African language in school. A famous boxer was reclassified just before a major fight. Afrikaner children with an adrenal gland condition were treated as nonwhite.

Finally, the authors link the consequences of classification under Apartheid with the classification of race, ethnicity, and Native American status in the U.S..

The authors argue that "the advantaged are those whose place in a set of classification systems is a powerful one, and for whom powerful sets of classifications of knowledge appear natural. For these people the infrastructures that together support and construct their identities operate particularly smoothly (though never fully so). For others, the fitting process of being able to use the infrastructures takes a terrible toll" (255)

Classification and Work Practice

What a Difference a Name Makes: the Classification of Nursing Work

In the first of three chapters focused on the classification of nursing work, Bowker and Star shift from classifications of "natural" to "social" kinds, by looking at the classification of "units of nursing work" (229). The authors argue that although the public tends to debate "artifacts" or outcomes of work processes, that these work processes, which are less open to public debate, are no less important. In this chapter, the focus is on the creation of those classifications. It is organized around three core challenges for creating a work classification scheme:

  • Comparability across individuals or sites
  • Visibility of things to be observed and assessed
  • Control over the execution of the work, to varying degrees based on the nature of the classification
"from the point of view of design, the creation of a perfect classification scheme ideally preserves common-sense control, enhances comparability in the right places, and makes visible what is wrongly invisible, leaving justly invisible discretionary judgment [....] In the real world, these areas trade off against each other." (232)

To reveal these qualities of work classification, the authors describe the definitions of "hope installation" and "humor" in the Nursing Interventions Classification(NIC) system as units of nursing work. The classification includes detailed breakdowns of the process of humor. The classification also includes a table from the NIC including 13 sub-actions for offering spiritual support to patients; both sections include at least three citations of books or scholarly articles on the topic.

How did humor and spiritual support come to be described in such ways? Firstly, an influential group of nursing teachers argued that nursing could not be scientifically studied without standard language and classification, developing three interrelated, complementary classification schemes to measure nursing interventions in scientific studies. Secondly, classification was developed to support and defend the professionalization of nursing. Finally, classification was seen as a way to represent nursing work in hospital computer systems rather than risk becoming further marginalized.

To explore the process of classifying nursing, the authors observe all minutes of the NIC team, considered all their publications, conducted eighteen open-ended, in-depth interviews with people involved and affected, observed NIC team meetings.

The chapter briefly compares the creation of a classification system to qualities of information infrastructures (as defined above), describing the NIC as "actively developing infrastructure" connected to decisionmaking systems, accounting systems, professionalization, research, and cultural values that are continually under revision. The authors also link this work to "notions of accounting and quantifying as forms of social order" (240) [10].

Comparability is a move from the particular towards the universal, and it serves the purposes of curriculum development, professional training, accounting, and research across sites and even disciplines. One creator of the system noted that administrators were devaluing nursing work because no one could explain what nurses do: "the only thing that they know is that they can't work without us" (242). Critics of the system saw it as "outmoded and inflexible"; one critic proposed a system that would use natural language processing to develop a taxonomy from the everyday language of nurses (243). The need for comparability is the main critique of the machine classification approach.

Classifications influence the visibility of different kinds of work. The NIC group originally focused on direct care to patients, but they found that ignoring indirect care (scheduling patients, managing discharge from the hospital) and administrative activities(managing hospital staff) might render those activities invisible and uncompensated. Further tension arose around biases in the classification towards the strengths of its designers in physiological care. In one case, "leech therapy" was omitted despite its widespread use, out of fear that the classification would not be taken seriously in some contexts (248). Other common practices were omitted because they had not been sufficiently considered in scholarly literature.

Bowker and Star describe a tension of control between a level of detail useful for newcomers and detail considered insulting to experienced nurses (249). Citing a large number of studies (including [11]), they identify another trade-off: while classification fo work offers potential benefits to funding gained by making nursing work visible, at the cost of surveillance or social control at work.

The chapter ends by situating the nursing classification program to wider literature on professionalizagion, including Abbott's The System of Professions[12].

Organizational Forgetting, Nursing Knowledge, and Classification

The Theory and Practice of Classifications

Categorical Work and Boundary Infrastructures: Enriching Theories of Classification

Why Classifications Matter

Theoretical and practical relevance:

This book offers a thorough, revealing and well-grounded exploration of the process and politics of classification, as well as some a diverse palette of research methods for studying them, with examples.

Notable References

  1. Star, S. L. (1999). The ethnography of infrastructure. American behavioral scientist, 43(3), 377-391.
  2. Star, Susan Lee, Karen Rohleder. 1996. Steps Toward an Ecology of Infrastructure: Design and Access for Large Information Spaces]. Information Systems Research vol 7 no 1, March 1996
  3. Bowker, G. C., Baker, K., Millerand, F., & Ribes, D. (2010). Toward information infrastructure studies: Ways of knowing in a networked environment. In International handbook of internet research (pp. 97-117). Springer Netherlands.
  4. Corbin, J. M., & Strauss, A. (1988). Unending work and care: Managing chronic illness at home. Jossey-Bass.
  5. Epstein, S. (1996). Impure science: AIDS, activism, and the politics of knowledge (Vol. 7). Univ of California Press.
  6. Mann, Thomas. 1924 The Magic Mountain. S. Fischer Verlag
  7. Julius A. Roth. Timetables: Structuring the Passage of Time in Hospital Treatment and Other Careers (Indianapolis: Bobbs-Merrill Company, 1963)
  8. Arendt, Hannah. 1963. Eichmann in Jerusalem: A Report on the Banality of Evil Penguin Books.
  9. Watson, G. (1970). Passing for White: a study of racial assimilation in a South African school (Vol. 102). Tavistock Publications.
  10. Rose, N. (1991). Governing by numbers: Figuring out democracy. Accounting, organizations and society, 16(7), 673-692.
  11. Wagner, I. (1993). Women's voice: The case of nursing information systems. AI & society, 7(4), 295-310.
  12. Abbott, A. (2014). The system of professions: An essay on the division of expert labor. University of Chicago Press.