The atomic unit of normalized deviance — the individually defensible judgment, supported by evidence and precedent, that is rational in isolation and catastrophic in aggregate because each exception is evaluated against the standard the previous exception has already revised.
Vaughan's framework identifies the reasonable exception as the mechanism by which normalized deviance is actually constructed at the level of individual decisions. Each exception is bounded, justified by specific circumstances, and supported by a growing evidence base. The person making the exception can point to data, to precedent, to a rational cost-benefit analysis. The exception is not a failure of judgment; it is an exercise of judgment within an institutional environment where the burden of proof falls asymmetrically on those who wish to stop rather than those who wish to proceed. The AI transition has generated reasonable exceptions at unprecedented pace and scale, each defensible in isolation, cumulatively reshaping what institutions consider adequate oversight.
The Reasonable Exception
In The You On AI Field Guide
The concept's power derives from its resistance to prevention through conventional means. Telling a competent professional that her reasonable judgment is wrong, when the accumulated evidence supports her position and the