Yeah, the inherent problems of video hosting are real and highly-devolved decentralized hosting (e.g. IPFS) is nowhere near a satisfactory solution in a world where most end user devices sleep much of the time and need to conserve battery power, but these are separate from the reasons that Google finds humane moderation too difficult to even credibly attempt—the reasons video hosting is necessarily hard, aren't the reasons YouTube moderation is hard. I do agree that discovery would be a problem to solve, but I don't think it's insurmountable. After all, I hear running a search engine can be profitable...

I think at least separating hosting and the not-necessarily-connected role of curation and promotion (plus, maybe, separating distribution from hosting) would go a long way to solving a lot of problems. That'd basically require regulation for it to actually happen, though, because there's too much value in capturing that entire vertical, effectively "dumping" on parts of that potential-market to feed network effects and build a moat around for whichever part of it (the ad-laden curation and promotion interface, in YouTube's case) you're making money on.

This is how a lot of Google's—among others'—properties work. It's frustrating because dumping "free" products for the purpose of marketshare-capture in adjacent markets stifles not just interest in other commercial efforts with different funding models, but also FOSS or truly-free hobbyist efforts. I think this kind of thing is also why we basically don't develop new open Internet protocols anymore, or if we do, they don't take off, even when the need is there—they compete with free but deliberately closed and non-interoperable services funded by ads or propped up by other wildly profitable services, so are DOA even if you can convince anyone that trying to develop them in such an environment is worth their time in the first place.