Reply to comment

Sandy initiated our conversation today by describing the work he has been doing with caseworkers and for-profit contractors, who are collaborating on welfare to work in Florida; this project is framed in terms of institutional and social economics.

He explained that Herbert Simon, a professor of public administration and a polymath who also contributed to work on artificial intelligence, received the Nobel prize in Economics in the 1980s for work he had done on Adminstrative Behavior in 1947: his argument was that people "satisficed," rather than "maximizing." Being constrained in the choices they made, they relied not on "perfect" but "bounded rationality." Simon's work drew on the "new field of social psychology" in making decision models more realistic, because reflective of actual perception, cognition, and behavior.

This led to a "new experimental economics," which replaced classical economic models--that had asked what the rational decision would be, if one had perfect information--with a recognition of the limits of both time and information. These ideas have since been widely applied; recent work on web usability, for instance, references "satisficing": rather than reading web pages all the way down, surfers are likely to "satisfice" with the first link they find.

Sandy contrasted "rational comprehensive decision-making" (a form of planning that presumes universal rationality, with all goals surveyed and weighed in order to maximize objectives) with "incrementalism" (which advocates "taking the next step," in an attempt to improve on what has been done, but without setting long-term goals). Several apt jokes were evoked @ this point: Simon's "marriage rule" (rather than making a comprehensive survey of all candidates for potential soul mates, we prepare to marry by "sampling the market and chosing the next best thing that comes along") was followed by a story, told by Arthur C. Clarke, of a room in which women line up on one side, men on the other. Each group moves, again and again, across 1/2 the distance which separates them. Zeno's paradox is operating here: they never actually reach each other, but do get close enough "for all practical purposes."

Such stories, it was claimed, are insignificant except in light of the concept of "perfect information": without the implied comparison of a perfect world, "good enough for all practical purposes" becomes "just everyday life." This led to considerable discussion. Habermas's ideal speech act was evoked, as an important regulative ideal: we should evaluate our actions based on what we might agree to if we could get "outside of" power relations, even if such positions are not achievable. The science of physics was also invoked: none of Newton's three laws are "true" in real life, but only work in a laboratory where all variables are eliminated. Such idealizing enables us to come up with useful information, and Newton influenced economists such as Adam Smith.

So, we asked ourselves: which is better? A "vacuum, friction-free" model of markets, or a more behavioral one? In the abstract, such a question may be unanswerable. Ideal laws are useful, if they can be adjusted for perturbations, and to individual cases. But perfect information is a fiction, and economics has moved away from predictive models to richer, more contextual ones. The operative premise here needs to be Einstein's: we must "never mistake the ways we have of making sense of things for the things themselves"; there is no ready corespondence between models and the things being modeled. Only the concept of an ideal world makes either the joke about the marriage rule or the one "for all practical purposes" funny.

Perhaps there is no important paradox here, but simply the need to acknowledge the distinction between theory and practice ("physicists try to understand, engineers get the job done"). Classic decision-making models have not worked, when applied to government, because they aren't realistic enough; government reform has been naive in its failure to take politics into account. Anthony Appiah developed a similar argument during his lectures @ Bryn Mawr two years ago: philosophy and ethics have become too abstract.

Alternatively, instead of complaiing that models are unrealistic, because they don't work in specific cases, we might instead recognize that all models are imperfect, and engage in incrementalism: noting failings, improving the models as best we can. Criticism of abstraction arises from the same foundation that gives rise to abstraction itself: a belief in the possibility of perfect information. Never expect a model to be applicable in all cases: it is never an end, but can be used instead as an initial step in a continuing process.

There's a difference between mechanical actions and those of people, who have consciousness, and can keep on revising what they do. Steve Toulman's Return to Reason was invoked: this is a critique of abstract rationality--a form of science foisted on social scientists--that requires the balance of "situated reasoning." We closed with some musings about the ways in which the ideal world functions as the source of judgments made in this one.