There's also the fact that people spend a lot of time, effort, and money making these videos. How would they get paid by a decentralized system?

I want to decentralize too, absolutely. But I'm at a loss as to how we'd accomplish it in a way that ultimately benefits society instead of making an inferior product.

Payment is easy if the publisher has a machine online to accept micropayments through something like the lightning network. Maybe there's a way you could make micropayments work through an email-style system or through a delegation system.

I don't think these problems are 10 year problems, I think a good team with a determined attitude could design it in one or two years (coding aside).

Making a direct YouTube replacement is probably not the best use of time though. Instead improving the building blocks or creating a system where at least the data itself is decentralized is probably better. The decentralized replacement for YouTube likely won't be a clone, just as YouTube is not really a clone of traditional TV, yet still has replaced it in many ways.

Is something like Patreon tied to youtube? Could you fund someone on vimeo or your own website with it?

We used to have web rings. We used to follow people that posted interesting stuff and weren't too annoying, either manually or by rss.

No reason it can't be done afain I think.

How would they get paid by a decentralized system?

By editing in a sponsor message in the film? By adding a few pngs with links to sponsors?

I think that's perhaps one of the most fascinating lessons of the last 1-1.5 decades. As far as anyone can tell, there was nothing to stop the initially near-fully decentralized web from flourishing --- just as there was nothing to stop increasing centralization from taking hold. No "bigly" tax incentives / subsidies for big IT conglomerates to build up and solidify strongholds etc. Just "let's see how things settle over time", and this is how. (UNLESS we're still in some kind of early railroad-like macro (not micro ala 99-01) "bubble" which inflates slowly largely unnoticed over more than a decade.) At least this whole development has been a boon to FOSS as a slight silver-lining (despite all the transient rubbish that also floats to the top intermittently --- same as ever though).

Anyway, if this centralization evolved near-organically, surely any more decentralized future will also have to evolve mostly-organically? Likely through multiple painful series of "clouds raining down" or some such..

More generally, the combination of incorporation and contract law also strongly encourages centralization: because negotiations are by mutual assent, those with better negotiating positions extract more of the added value from such agreements than those with weaker negotiating positions, and incorporation trivially allows bundling strengths with almost no downside.

Then there's the way the legal system is structured, which is extremely complicated: "red tape". That's a generalized barrier to entry; a kind of cost that doesn't scale with size. Given that legal costs are high, those that can't afford to take risks are going to be outcompeted by larger entities that can afford such gambles, assuming the merit is at least somewhat unclear. In extreme cases, even if a small company (or individual) wins a court case, they may nevertheless be the net loser because cases take long and cost of lot of attention and money - the opportunity cost of engaging is higher for small entities than it is for large entities (at least in general).

Some of those effects are probably intrinsic to any system of organization, but some of them are likely a historical quirk. I'm not sure which are which, though ;-).

I don't see why a video recommendation engine is inherently harder than a web index (which, as hard as it may have been, clearly was possible). You could probably monetize it in exactly the same way.

Video wasn't centralized before Youtube, but since they began offering free hosting for everyone (at astronical cost! they were burning through their money in record time before Google picked it up, and they have been burning money ever since) and video bandwith is probably by far the most expensive thing on your web hosting bill that's where everyone went. If anything, perhaps there should be rules regarding hwo much money you are allowed to spend undercutting your competition at a loss, but that's a polical decision. It can not be solved by technology.

I guess really the term I am looking for is 'trustless'. I don't want to have to trust Google to do the right thing, I just want it to be manifest that there is no manipulation, and no single point of failure.

Try out a project I am working on named biddle for your distribution needs: https://github.com/prettydiff/biddle

biddle fights censorship by allowing you to put your publications anywhere. There is no central repository, and since you are required to specify your publication point you just need to find file hosting. This means private publications are free (aside from your hosting expenses) if you publish on an internal network.

As it were I'm working on a decentralized file hosting / data storage platform called Sia. We are probably not quite ready to integrate, as Sia does not yet support allowing others to download your files, but it's in our roadmap.

IPFS would probably also work here, given that the files are public and likely to be downloaded many times. It works like bittorrent in the sense that distribution is unpaid and not guaranteed, but as long as you have enough seeders the file can stay alive.