Right so my point is, with no smooth optimization path towards X in a given species the question "why doesn't species just X" has a readily available answer - because biased random walks are fickle and the pathway likely convoluted. Whereas when a smooth optimization path is available it becomes reasonable to ask "then why isn't species doing X already"? Assuming a long enough timescale then there must be some selection pressure preventing it.

I agree that the ice age is a reasonable hypothesis. If we were exposed to repeated boom-bust cycles of extended duration (ie multi-generational) then it's at least plausible that we gained some mechanism dedicated to preventing us from adapting in that direction.

Still it's an entirely reasonable question that needs to be asked when considering the potential for serious side effects. If the hypothesis is that this is actively prevented for historical reasons then we should want to see the precise molecular mechanism hunted down in order to validate that. Keep in mind that an alternative hypothesis would take the general form that ramping it up causes serious deleterious side effects in the long term.

It's quite like Chesterton's fence. There must be a reason so we would do well to understand approximately what that might be before tinkering.