1. a polygenic score is not 'the impact of common variants'; that's the SNP heritability, such as the GCTA or LDSC. It's not 1.6%, it's closer to 30%. You edited out the part of the abstract which correctly specifies this: "the single-nucleotide polymorphism heritability for the extreme IQ trait was 0.33 (0.02)"

2. that is not the best predictor of intelligence; it is not now (that would be Allegrini et al 2018), and it was not then either - you have to go all the way back to like 2014 for 1.6% to be SOTA and I'm puzzled why you cite Zabaneh at all, which is one of the most obscure IQ GWASes... The actual SOTA is 11% or higher.

3. your claims about power are obviously wrong, as Allegrini et al 2018 uses far less than 'tens of millions' and still greatly outperforms it. (Which has something to do with those rare beneficial variants not existing which they were looking for...)

So, we can predict 11%, not 1.6%, and common variants have a large impact, 3x that, and rare variants account for even more, easily 50% including what you might call semi-rare variants (https://www.biorxiv.org/content/10.1101/106203v2).