A less quackish way to say the same thing is that a scientific paradigm tells people along which lines to look for answers. Looking at areas the paradigm doesn't recommend are generally not worthwhile, but occasionally you get something important, which doesn't fit in the current paradigm but will eventually help form the new paradigm.
A good heuristic could be "seems like a solid scientist in general, but this niche where he was a top level researcher led him to a split with the main stream" vs "consistently takes anti-mainstream views and has no contributions within the paradigm "
How could anything be positively predicted by gender?
More seriously, selection effects should be much stronger than gender, age and ethnicity, and yet there should still be some correlation in one direction or another, wish there were power and confidence values given
When things are just.. wrong, there's not much reason to be specific. Especially when gwern.net has multiple long-form blog posts discussing these topics
Depends how cheap gene editing gets. If you can mass-produce adenoviruses that improve iq 15 points and health significantly, that's going to be way cheaper than nutrition, forget education and of course opportunity.
Of course, better genes means nutrition, education and opportunity go further
Restrict AI the way you would if it were human