Except that's not true. We can actually test for sobriety. And by doing drug testing instead, we're actually using a worse signal in place of a better signal!

See e.g., http://www.predictivesafety.com/news/2017/2/6/the-advantages...

> Perhaps you are too young or live in too modern a world to understand how dangerous rare events are. Constructive paranoia has helped select evolutionary survivors for a very long time.

I'm not really sure if there's more than a vague analogy connecting Diamond's hypotheses to drug testing policy. And I don't find that exceptionally vague analogy to be compelling evidence. (And I'm one of those people who doesn't even buy a lot of Diamond's work anyways; in fact, I think I'm in good company on this one among the sober, grown-up anthropologists out there...)

My weathered experience tells me to prefer hard data over vague analogies to pop science writers who are themselves known more for their well-written vague analogies than their well-researched hard science :-)