Learning Envision / Reading The Next Prediction

Slower and smaller, but never nothing

The sensible default is to expect less than announced, on a longer timeline, without concluding that change does not occur.

The default earns its place empirically rather than temperamentally. Applied backwards to the record, expecting a fraction of the announced effect over a multiple of the announced timeline would have produced better estimates than either the promotional claims or their flat rejection. It would have anticipated the administrative gains, the slow spread through compatibility rather than merit, the long imitation phase, and the eventual normalisation of capabilities that once seemed remarkable. That is a reasonable amount of predictive work for a rule that fits in a sentence.

The rule needs a guard, because the fraction is not zero and treating it as zero is its own failure. Institutions that dismissed each wave entirely ended up genuinely behind, and the cost fell on their learners rather than on the people who made the decision. The productive stance is to take small positions rather than none: to keep capability available, to allow local experimentation without mandating it, to protect the people doing the experimenting from carrying the cost alone, and to review at intervals long enough to be meaningful.

Perhaps the most useful discipline is separating the question of whether something is impressive from the question of whether it will change practice. The two come apart routinely. Genuinely remarkable capabilities have left practice largely untouched, and unremarkable ones have quietly reshaped how institutions run. Judging the second question by the first is the error underneath most of the record considered here. It is an easy error to make, because impressiveness is immediately visible and consequence is not, and it will continue to be made for exactly that reason.