hahaha fair enough, I'll admit it's hard to quantify, but does hold up from a machine learning perspective. Since I started using these arch types it has become a lot easier to sort people accurately myself to the point where I have a very high success rate identifying a Jew on sight. I deal with them a fair amount in my professional life and have found these to be accurate models, and I am almost never caught off guard by a member of the tribe that I didn't suspect was one.
I'll look into if there is anything extensively done for this, but tragically this is an area of research that will never be publishable, and finding good data is painfully difficult.
(post is archived)