I think the theory at the end of the article sounds quite plausible. Namely that blood types are playing rock-paper-scissors against bacteria / viruses. Patho

I think the theory at the end of the article sounds quite plausible. Namely that blood types are playing rock-paper-scissors against bacteria / viruses. Pathogens can learn to target a single blood type if it's predominant (just like you can learn you should play Paper if your opponent usually goes for Rock). So it's better to have a population of mixed blood types, and anytime it starts to get out of whack, disease exerts a force that disproportionately takes the newly popular blood type out of the gene pool, pushing back toward equilibrium. Just like the Nash equilibrium RPS strategy is rand(). How to test this experimentally? I'm no molecular biologist, but maybe let viruses reproduce in an environment where one antigen is present for enough generations for them to mutate to have increased effectiveness attacking it (maybe you can determine this from an increasing viral reproduction rate?), then a testable hypothesis that would confirm the theory is the mutation also caused the viruses to lose efficiency at attacking another blood type's antigen.