That's a good question. As I understand this paper (it's come up on HN before), Plomin takes the consensus of the literature† as establishing 50% heritability via twin studies (accounting for "half the variance in intelligence").
The project then is to use modern genetics to replicate that number and, as importantly, to identify the genes that contribute to that variance. Of course, he's unable to do so: this is the famous "Missing Heritability Problem", that large-scale genetic surveys, as methods have improved and deconfounding is applied, have only managed to associate specific genes with something like 15% of the variance.
In this paper (again, as I understand it) Plomin takes 10% of the total variance as established by GWAS, and then expects much of the rest of the "missing" heritability to be accounted for by rare variants that current GWAS methods can't pick up. But the math for that probably can't work (this paper is from 2018, though, and everything I've read about rare variant stuff has been in the last few years).
† I don't think he believes this, though; I think he's an 80%-er.
The project then is to use modern genetics to replicate that number and, as importantly, to identify the genes that contribute to that variance. Of course, he's unable to do so: this is the famous "Missing Heritability Problem", that large-scale genetic surveys, as methods have improved and deconfounding is applied, have only managed to associate specific genes with something like 15% of the variance.
In this paper (again, as I understand it) Plomin takes 10% of the total variance as established by GWAS, and then expects much of the rest of the "missing" heritability to be accounted for by rare variants that current GWAS methods can't pick up. But the math for that probably can't work (this paper is from 2018, though, and everything I've read about rare variant stuff has been in the last few years).
† I don't think he believes this, though; I think he's an 80%-er.