> Statistically, crowdsourcing does better at converging to the actual answer.

Statistician here, and this is a good example of a relatively meaningless statistic, to be honest. Crowdsourcing statistically tends to be more accurate than just asking one person, in the average case, for what should be mathematically obvious reasons.

But the “average case” isn’t applicable to literally every situation. I would posit that when we start to talk about areas of expertise that require a PhD to even begin to be taken seriously for your opinion, crowdsourcing from unverified users starts to become a whole lot more biased.