Though one could debate about the "hardness" of the upper (or lower) bound. Gaussian RVs have some "softness" there, though of course the double exponential fal

Though one could debate about the "hardness" of the upper (or lower) bound. Gaussian RVs have some "softness" there, though of course the double exponential falls off rather quickly. (To which extent intelligence is well described by a Gaussian at the extremes is yet another can of worms.)