[loewenskind]

>We observe that intellectual performance, as far as has ever been measured, varies tremendously by individual.

In the ways we have measured, yes. The issue here, I think, is that person A who excels to extreme levels in e.g. math doesn't excel at everything. In fact, the better he is at math the worse he might be in different subjects (Einstein example). So when we measure how well people do aren't we just measuring how good they are at passing our tests?

There does seem to be correlation for people who manage to do well at our system and people who end up successful. I'm just not convinced this is the whole story.