> How do you propose to deal with unpopular opinions that turn out to be correct with this method?

Multiple tiers of signal (forums open to wider ideas that are, perhaps, more private than YouTube. Additionally, academic forums where people with relevant background can hash things out). I don't think the "everyone can see everything" Facebook / YouTube / Twitter model has been proven to work for difficult and sensitive topics.

YouTube is just not one of the places the messy conversations are safe to have. It's a cat-video host, not a pathology research organization or academic community (nor does it seem it wants to be).

> Also (and related) how do you propose to deal with corruption?

I don't know, but I think there's a burden of proof that the open model prevents corruption (assuming open is what we have now). It's massively vulnerable to propaganda and information distortion based on amount of effort put into amplify signal, not truth of information in signal. People with little background in a technical subject to lean on when exercising their critical thinking are very vulnerable to the notion "Everybody is saying it, so it must be true," and when you couple that to bubble effects I worry we see bad results.

The "wisdom of crowds" was always an experiment. It's possible for the experiment to fail.