![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
Abstract
This article considers challenges to getting information systems to recognize “objects” in other information systems. We explore a tension between commitment to standardization and the constitutive and situated social fact requirements of human comprehension and work through an ethnographic study of an information system design team meeting.2 Facilitating interoperability between systems is an important challenge currently facing design teams. Our study elucidates—in the design team's own words—problems that design teams confront with “object” and “concept” certainty, practical “use” and “language”: with defining what they call a “What” when data objects must cross boundaries. The refrain “What is the ‘What’?” punctuates the meeting. Although human use and comprehension frame their concerns, they treat meaning, practical use, and language as “technical” and “philosophical” issues. The social issues they acknowledge are confined to “governance.” Given their reliance on performed social objects, including “role” and “identity”, at key points in their discussion, however, and the importance of “language” and “concepts” to their concerns, we point out social dimensions of their task, suggesting that a broader understanding of the importance of constitutive practices and their situated character in achieving information objects—and social objects more generally—could change the team's perception of their options.
Notes
1. This paper is a version of MITRE Technical Report: Document Number MTR 10-2594.
2. We use “constitutive” as a technical term, distinguishing rules/practices that are constitutive of objects, from rules that are merely descriptive (or accountable). Although it is usually misunderstood, the distinction is essential to Durkheim's approach to modernity (1893/1933) and has been elaborated in J. Rawls (1955), Garfinkel (1963), and Searle (Citation1964).
3. “Information system” indicates a set of processes by which information is created, stored, communicated, managed, transformed, and destroyed. These processes include human activities as well as—and sometimes in lieu of—those performed by information technology. An information system is used for, and is an artifact of, knowledge management. Unless otherwise indicated, “design” means the design of an information system, including, in particular, the design of any data schema or models intrinsic to such a system. Design is one of many systems engineering activities.
4. Garfinkel's phrase “each next first time” indicates that while “each” needs to be “recognizably” the same to count as an instance of the “same” social object and each must be achieved all over again each “next” time, all achieved instances of social objects would be in some sense different from others—and thus also “first” times. The process could thus not be reduced to a matter of routine—or habit.
5. The CVE Team we studied prior to this observation adopted an approach that went farther than TrackIt in stripping off as much semantics as possible. The objective was to maximize the “indexicality” of data objects so they would be meaningful to a large number of diverse stakeholders (see the section “Governance Versus Architecture” and Rawls et al. 2009).
6. There are various ways of constructing and referring to these data models, and these terms are scattered through the discussion. Objects or concepts in data models are defined in terms of "attributes", which correspond to the idea of essential features in classical category theory (Jacob 2004). The term "taxonomy" typically refers to a model in which objects are related to each other hierarchically, typically enforcing a subset relationship among objects. "Schema" typically refers to a data model in which objects can be related to each other in arbitrary, nonhierarchical ways. "Ontology" has many different meanings within information system design, but typically refers to a schema that has been represented in formal knowledge representation language to allow computers to perform some form of "automated reasoning" (Mann and Brooks 2010; Obrst 2010). While information system designers may draw important distinctions between these models, they sometimes use the terms interchangeably in reference to data models. We use the general term “data model” unless there is a clear need to be more specific.
7. See Methodology section for an explanation of line numbers from field notes. Participants in the discussion are identified by pseudonyms (e.g., “Tom” and “Gary”).
8. To satisfy MITRE requirements, some words and names were later changed to preserve project anonymity.
9. “Accountably” as used here means both what they are held accountable for being engaged in by higher ups and what they would say (i.e., give as an account of their activities) if asked about the work by peers. An example of accountability to peers versus higher ups from the notes occurs at lines 269–273, when George explains (269) why they reduced the number of attributes from 11 to 8. It was (272) “to cut off the people who were trying to wrap us around the xml axle.” He explains that (273) “They were counting the number of schemas.”
10. The team follows a common engineering practice of blurring the distinction between IT system architecture, “information architecture,” and “data architecture” (“how data is stored, managed, and used in a system”: Lewis et al. Citation2001). While the term “information architecture” has multiple definitions, in using the term “architecture” the team seems to be oriented toward “the structural design of an information space to facilitate task completion and intuitive access to content” (Rosenfeld 2002).
11. Engineers generally use this term to refer to IT governance (“the culture, organization, policy and practices that provide for IT management and control”: ITGI 2005) or to data governance (“a quality control discipline for adding new rigor and discipline to the process of managing, using, improving and protecting organizational information”: IBM 2007).
12. The problem of object constancy surfaced in early Greek philosophy as the problem of identity. Data modelers attempt to use classification and definition to resolve the ambiguity. However, for a sufficiently rich set of objects, each attempt at rigid classification creates further additional defining attributes, which inevitably produce more exceptions. Despite this problem, it is a persistent belief among team members that objects are defined by attributes.
13. An identifier system is list of identifiers, together with the syntax (and, if applicable, high-level semantics) for defining list members and the organizational processes by which the list is created and maintained. Identifiers can encode classificatory information (VIN, Dewey Decimal) or ordering information (room numbers, street addresses), or can be entirely nominal (license plates). In all cases, an ongoing process is sustained to produce new identifiers. The section “Identifier Systems vs Defined Objects” discusses identifier systems. For information on types of identifier systems see Mann and Brooks (2010).
14. See Obrst (2003; 2010) for discussion.
15. Each word can mean many things, a property known as indexicality. But in particular sequences of talk, “use” conventions and conversational devices of turn-taking and preference order allow words to convey precise meanings. For instance, after a question, the word “yes” will usually be treated as an answer. However, in other sequential contexts, it can function as a question (when a homeowner calls “yes” on hearing a noise). Impending disagreement is usually “marked” with word forms such as “yes, but” that indicate a disagreement is coming up. It is a more diplomatic way of disagreeing and has been found to protect communication from problems that can arise when disagreements are stated without such markers (Schegloff 2007). In such cases, “yes” and “but” function together as a marker, a property of sequential order (like a TrackIt linkage) that constitutes its conceptual parameters.
16. In addition to the pioneering work of Sacks (1992; 1995), and the groundbreaking turn-taking paper (Sacks, Schegloff, and Jefferson 1974), Schegloff and Jefferson are important sources. Schegloff (2007) published an introduction to the field. Anita Pomerantz contributed essential early work on assessments (Pomerantz 1975) and went on to do important work in communication. Alene Terasaki contributed the first work on “pre-sequences” (1982/2004).