> He concludes that after all the research that has been done that non-shared environmental differences are unsystematic and unstable which means they are essentially random

No, what it actually means is that the specific environmental influences haven't been identified. “Shared environment” and “unshared environment” are big buckets inferred in “top down" studies from whether people who share some pre-identified candidate influences (family environment, school, etc.) get equal or different effects.

(There's also studies that work bottom-up to try to identify candidate specific environmental factors, but it's a long process of identifying potential ones and then working them into top-down studies where you can tease out the exact influence of each; we've actually identified quite a number of environmental factors, especially in early childhood, that are specific negative influences, and you can, given resources, engineer to avoid them.)