The one-two punch of Goodhart's law and the McNamara fallacy bites hard here, though. We only measure things we can easily quantify, and then we optimize around those measurements, destroying their efficacy as second-order indicators of the things we couldn't figure out how to measure.

For example is the purpose of educating children about history A) to give them a systemic, if somewhat cursory, understanding of the evolution of society, in order allow them to contextualize the present? or B) ensure they know a list of a hundred or so people and events that happened?

It is much easier to quantify the latter.