I think a better model is not that some are exponentially smarter than others, but that some folks are operating close to the problem manifold and the rest aren

I think a better model is not that some are exponentially smarter than others, but that some folks are operating close to the problem manifold and the rest aren't. AKA, some people are solving classes of problems in approximately the most efficient way (or relatively more efficient way) and most people are computing solutions in exponentially inferior ways.