Given people A/B test all kinds of minor things I'm sure someone would have hit on it and then it would spread. That's basically the rocket fuel that drives people really into A/B testing everything rather than just hunches "I could find something which increases metrics you wouldn't otherwise think to try".

In either case, I think the net result is bad news for users, good news for people selling things. And of course the sprinkle of "people just making mistakes/guesses" too, but there's no universe that's not going to be found.