His Ink Cloud:
Estimates are calculated, and it's all inferential. Pinpointing specific genes doesn't escape the fact that you're still just estimating, it doesn't free you from the problem of induction.
He held to the "Gotta Find the Genes (TM)" standard, and is now trying to say that he agreed that you could make heritability estimates without reference to specific genetic structures (he didn't, this is a lie), by claiming that what he was really doing was saying you couldn't come up with a "calculated quantity" without reference to specific genetic structures.
This is completely bogus. An estimate IS a "calculated quantity", and estimating the effects of specific genetic structures faces all the same problems of induction and inference that estimating the effects of a whole genome does; PLUS the confounding effects of other genes.
It makes no sense. In both cases you are merely inferring causation from correlation, i.e. you are doing calculations, controlling for confounders as best you can, to make estimates.
A heritability estimate without reference to specific genes is probably going to be MORE accurate than one that tries to estimate heritability based on the inferred effect of specific genetic structures. Because an aggregate estimate, a whole-genome estimate, BY DEFINITION already encapsulates all of the effects of other confounders that go uncalculated in the estimate derived from the effects of specific genetic structures.