The complex system argument is a strawman.

It’s well understood that complex traits are polygenic (no single gene or sets of causal variants) but quite easy to predict such traits when looking at the full genome (I agree the resolution of prediction is still not there — but we are in the infancy of this field).

I recommend learning more about some of the polygenic risk scoring methodologies that are popping up in research: https://www.broadinstitute.org/news/giant-study-finds-rare-i...