I’ve just finished Matthew Crawford’s great book “Shop Class as Soulcraft,” tripped over a paragraph that I thought relevant to the struggles we’re having dealing with science entirely performed through modelling, and thought I’d share.
Some modern motorcycles have begun to include onboard, computerized self-diagnostic functions, just as cars do. But they haven’t eliminated the kind of judgment mechanics exercise. If we can understand why they haven’t, this will help illuminate further the limitations inherent in the idea of an “intellectual technology,” and the perversities that get laid upon work when those limitations aren’t heeded.
Car manufacturers are supposed to standardize their diagnostics under a protocol called OBD-II (for onboard diagnostics), but as any mechanic will tell you, sometimes the system gives the wrong trouble code. Being off by one digit might give a diagnosis of “System fuel too lean on bank one” (P0171), that is, an air-fuel mixture that is too much air and not enough fuel on the first bank of cylinders, when in fact the problem is “System fuel too rich on bank two” (P0172). An experienced mechanic can tell too lean from too rich by looking at the spark plugs; they will look blanched white in the first case and sooty in the second. Representing states of the world in a merely formal way, as “information” of the sort that can be coded, allows them to be entered into a logical syllogism of the sort that computerized diagnostics can solve. But this is to treat states of the world in isolation from the context in which their meaning arises, so such representations are especially liable to nonsense. To rely entirely on computer diagnostics would put one in the situation of the schoolchild who learns to do square roots on a calculator without understanding the principle. If he commits a keying error while taking the square root of thirty-six and gets an answer of eighteen, it will not strike him that there is anything amiss. For the mechanic, the risk is that someone else committed a keying error.
Computerized diagnostics don’t so much replace the mechanic’s judgment as add another layer to the work, one that requires a different sort of cognitive disposition.