How much has diminishing returns set in for various forms of plant and animal breeding (oil content, milk yield, meat yield, feed-to-meat conversion efficiency, growth rate)? Or for very long-term selection experiments like the E. coli experiment? Yes, maybe there’s some upper limit to human biology where diminishing returns sets in sharply, but given that there are thousands of variants influencing intelligence, no one has ever come close to having all of the good ones. (Imagine we could only manufacture +7SD humans instead of +10SD. Would that make much of a difference to how wrong people like you were?)

This doctor on talk radio a number of years ago suggested that this has been going on for some time or has been contemplated for some time. Breed two types of humans. One very smart and the other not so smart but docile enough to be controllable.