Porter, Theodore M. 1995. Trust in Numbers: The Pursuit of Objectivity in Science and Public Life. Princeton: Princeton University Press.
Disinterested? Self-delusional? Both?
Theodore Porter describes two kinds of objectivity. The very approach of sorting kinds of objectivity would suggest objectivity is more than the positivist view, that it is the way to describe actual truth. Porter calls the positivist view mechanical objectivity, yet he questions if such objectivity really exists. He points out how theoretical (mathematical) reasoning is more easily shared, but has “no relation to any actual world” (p.14). Scientific knowledge for example depends on a correct approach in an effort to tame human subjectivity (p.21), yet the very act of taking the correct approach in itself requires definition as to available approaches and which would be correct. For Porter, truth is negotiated by a community of disciplinary specialists (p. 12), meaning it is really disciplinary objectivity.
Expanding on Porter, Funda Ustek-Spilda acknowledges his distinction of mechanical and disciplinary objectivity and how they seem to melt into a single negotiated objectivity. Ustek-Spilda gives specific examples in the form of how countries in Europe count asylum-seekers and refugees using a seemingly universal set of rules for counting people. The trouble is there is some ambiguity in the description of who to count and how, so local administrators have to decide on rules and processes to incorporate the definitions using the context of their specific country. The result is a mix-match of approaches and something less than universal data-sets.
The process of localized decisions makes the bureaucrats not just implementers, but effectively policy-makers in their aggregation of approaches. She notes several important STS-related concepts that support Porter’s argument about the predominance of disciplinary objectivity. Statisticians become performative in their approach in that they enact the social world in efforts to describe or represent it (p. 295). She describes how statistical methods are really like Latour’s idea of ‘sociotechnical’ in that the resulting data are neither completely technically nor socially produced, but rather are a product of both. Yet, there is power in the numbers that are eventually normalized and aggregated. She describes uncertainty absorption in that the discretion (socially influenced decisions) behind the numbers fade into the background as the numbers are accepted as truth. The numbers, though, are really a result of statistical rule-making based on interpretation, adaption, and application of abstract standards, guidelines and concepts (p. 295). Interesting how in his article, Warwick Anderson seems to advocate for that sort of approach in modeling using numbers seeking to "open up a space for greater ecological, sociological, and cultural complexity in the biopolitics of modelling" (p. 167).
Given all this, can we even put any real trust in numbers? Perhaps with an understanding of all the caveats we can say some number represents some special set of circumstances, defined in a very specific way, under very specific conditions. Seems like so long as we consider numerical representation as a sort of ‘good-enough’ data to make a relatively informed decision there may be some reason to trust what numbers are telling us. At the same time, how likely is it for those who actually make decisions to understand and accept all that qualification language? It’s much easier to just ‘go with what the data is telling us’ and absolve oneself of personal responsibility. In this way there is at least an appearance (and maybe a self-delusion) of disinterested objective decisions.
Disinterested? Self-delusional? Both?
Theodore Porter describes two kinds of objectivity. The very approach of sorting kinds of objectivity would suggest objectivity is more than the positivist view, that it is the way to describe actual truth. Porter calls the positivist view mechanical objectivity, yet he questions if such objectivity really exists. He points out how theoretical (mathematical) reasoning is more easily shared, but has “no relation to any actual world” (p.14). Scientific knowledge for example depends on a correct approach in an effort to tame human subjectivity (p.21), yet the very act of taking the correct approach in itself requires definition as to available approaches and which would be correct. For Porter, truth is negotiated by a community of disciplinary specialists (p. 12), meaning it is really disciplinary objectivity.
Expanding on Porter, Funda Ustek-Spilda acknowledges his distinction of mechanical and disciplinary objectivity and how they seem to melt into a single negotiated objectivity. Ustek-Spilda gives specific examples in the form of how countries in Europe count asylum-seekers and refugees using a seemingly universal set of rules for counting people. The trouble is there is some ambiguity in the description of who to count and how, so local administrators have to decide on rules and processes to incorporate the definitions using the context of their specific country. The result is a mix-match of approaches and something less than universal data-sets.
The process of localized decisions makes the bureaucrats not just implementers, but effectively policy-makers in their aggregation of approaches. She notes several important STS-related concepts that support Porter’s argument about the predominance of disciplinary objectivity. Statisticians become performative in their approach in that they enact the social world in efforts to describe or represent it (p. 295). She describes how statistical methods are really like Latour’s idea of ‘sociotechnical’ in that the resulting data are neither completely technically nor socially produced, but rather are a product of both. Yet, there is power in the numbers that are eventually normalized and aggregated. She describes uncertainty absorption in that the discretion (socially influenced decisions) behind the numbers fade into the background as the numbers are accepted as truth. The numbers, though, are really a result of statistical rule-making based on interpretation, adaption, and application of abstract standards, guidelines and concepts (p. 295). Interesting how in his article, Warwick Anderson seems to advocate for that sort of approach in modeling using numbers seeking to "open up a space for greater ecological, sociological, and cultural complexity in the biopolitics of modelling" (p. 167).
Given all this, can we even put any real trust in numbers? Perhaps with an understanding of all the caveats we can say some number represents some special set of circumstances, defined in a very specific way, under very specific conditions. Seems like so long as we consider numerical representation as a sort of ‘good-enough’ data to make a relatively informed decision there may be some reason to trust what numbers are telling us. At the same time, how likely is it for those who actually make decisions to understand and accept all that qualification language? It’s much easier to just ‘go with what the data is telling us’ and absolve oneself of personal responsibility. In this way there is at least an appearance (and maybe a self-delusion) of disinterested objective decisions.
| anderson.pdf |
| ustek-spilda.pdf |
RSS Feed