Probability - and how you need to reason about independent vs dependent events - is poorly understood generally.

Let's try to make it HN-relevant by asking:

can two things that both AB tested positively in independent one-at-a-time testing possibly backfire if you launch both of them?