A/B testing, misinformation, and YouTube

I've been reading that YouTube has just released a feature that lets you A/B test videos. I have opinions on A/B testing, which I've mentioned before. The crux of the problem is that A/B testing presumes either A or B is the right answer, and it optimises for popularity. In YouTube's case, engagement is probably what you want, but it isn't necessarily the best outcome. The other issue is how much the variants can differ. A/B tests are usually used for something small, like the wording on a button, to see which is more compelling. With videos, the variants could differ enormously. That also means it could easily be weaponised into a misinformation campaign: posting multiple videos with different viewpoints, all under the same title. The purpose of a misinformation campaign isn't to convince anybody that a falsehood is true, but to sow enough disbelief that you can't trust anything. This mechanism feels baked into the software as a feature, not a bug. It doesn't really matter to YouTube…