One price a sociologist often has to pay for her/his work becoming widely taught and cited is that it does not take long before people stop considering what you actually wrote but instead start referring to what they think you wrote. Giddens' and Beck's work on risk society, Strauss' grounded theory, and Bourdieu's concept of 'cultural capital' are all examples of the way in which teachers, students and researchers are happy to cite concepts as a shorthand with little consideration of the subtleties of the original. Thus at worst 'risk society' comes to equal 'we live in a dangerous world' and 'grounded theory' comes to mean 'making it up as you go along'. Is Susan Leigh Star's concept of the boundary object in danger of going the same way? This was certainly a concern of Star herself before her untimely death in 2010.
In 'This is Not a Boundary Object: Reflections on the Origin of a Concept' (Science, Technology and Human Values 35(5) 2010) Star offers an account of the development of the boundary object concept and its use and potential misuse since. While not attempting to do justice to the richness of the paper, I think it is helpful to draw out what I think is the nub of Star's argument.
When Star (with Griesemer) first used the concept of a boundary object in 1989 it was to understand "a sort of arrangement that allow different people to work together without consensus." By the time she wrote Sorting Things Out: Classification and its Consequences (with Geof Bowker in 1999) she was interested in the 'scaling up' of boundary objects - how multiple boundary objects and systems of boundary objects grow (or do not grow) to 'boundary infrastructures'. Thus for Star discussion of boundary objects became inseparable from discussion of standards and infrastructures.
Academic discussion of boundary objects tends to emphasise their interpretive flexibility and the capacity/requirement of different users to actively give them different meanings in different settings. But Star argues, interpretive flexibility is only part of the story (otherwise pretty much anything could be considered a boundary object). Two other dimensions need to be considered alongside the mutability of objects. The first of these is that the form of the objects is not random but instead emerges out of information and work process needs and arrangements of the groups that use them. The implication here is that we should be asking why objects are the way they are rather than simply wondering at their mutability. The second dimension is that while there are moments of uncertainty and flexibility in the usage and interpretation of objects this is part of a dynamic process that also includes moments when use is well-structured and tailored to particular settings.
For Star the boundary object concept is most helpful when different users are connected together by a degree of formal organisation and/or infrastructure and actively work together (i.e. something much stronger and concrete than, for example, the notion that a flag can mean different things to different groups in society). Thus the boundary object's flexibility and shared structure "are the stuff of action". Boundary objects 'move up' into standards and boundary infrastructures (of which digitalised information systems are often great examples). Star (with Karen Ruhleder 2006) described the characteristics of such infrastructures as:
- they are embedded in other structures social arrangements and technologies
- they are transparent to use
- they have reach and scope across time/space
- they are learnt as part of a community of practice
- they plug into other infrastructures and tools in a standardised fashion
- they are built on an installed base
- they become visible on breakdown
- they become fixed in increments not globally
The continual (often hidden) work of maintaining such infrastructures entails moments of vagueness and moments of specificity; boundary objects facilitate the movement between those two kinds of moment. Hence categories in digital information systems are both standardised and codified and also incomplete and inconsistent in their use.