I really like your comment about theories that “patch” themselves when something unexpected occurs – it’s always an interesting issue, at what moment an empirical observation comes to figure as a refutation, and at what points we’re willing to disregard observations, assuming we’ll resolve them another time…

Speaking of creative destruction, I just got to read a draft of a paper on Schumpeter that seeks to align him with gropings toward non-linear theories of change. It says some interesting things about the cause/effect prejudices built into deterministic analytical schemas (which disappear once multiple variables and feedback loops are introduced, along with the fantasy of prediction).

It also addresses ‘unintended consequences’ by pointing out that they’re symptoms of bad theory, specifically attempts to linearize causal inputs and effective outputs in terms of some restricted dimensionalization of dynamic state spaces. Staying in linearity requires theories to be ‘patched’ each time an unexpected outcome occurs (in my own work I argue this is exactly what marxists have tended to do with proletarian action), but it would be better to scrap the linear model altogether and produce dynamically richer theories.

So ‘unintended consequences’ is a joke to tell on folks who are trying to get camels through the eyes of needles. The nasty truth of more adequate non-linear, multivariable, dynamic theories is that they do not offer predictive certainty or definite accounts of exact outputs from inputs. Of course, in practice neither do linear models, but at least they promise such certainty in some eventual perfect future where all the ‘exogenous’ variables have been cleaned out.