A YouTube clone that uses a clone of YouTube's infrastructure is expensive, but what about a distributed p2p YouTube clone?
Obviously it's hard to quantify, as it doesn't exist yet, but I think it's technologically feasible.
That would be more expensive because
- You have a much higher failure rate of the storage media as people say "I'm running out of hard drive space. What should I get rid of?"
- You need to recruit those people to give up a resource that (unlike the spare compute cycles that SETI uses) they are likely currently using.
- You have to convince people to trust you to put arbitrary video content on their hard drives. Therefore, you need to have some process for deciding what video content is objectionable enough that you won't store it.
Not really, because data is replicated between many people. People can delete it and other people would still have it.
> You need to recruit those people to give up a resource that (unlike the spare compute cycles that SETI uses) they are likely currently using.
It's no different than people seeding torrents or people just using ipfs. Simply accessing the system would transparently increase availability. In fact, ipfs is probably suited as-is.
> You have to convince people to trust you to put arbitrary video content on their hard drives. Therefore, you need to have some process for deciding what video content is objectionable enough that you won't store it.
No, data will become more accessible as people would consume it. People only need to understand how the system works, they don't need to trust "me" (whoever you refer as "you" in your post).
As I said, ipfs probably works as-is.
Yes, so you are designing a distributed system with a higher failure rate of the underlying media. That means you need more replication, so you need more people to donate space.
Same reason Bitcoin and its ilk can't scale. Decentralization throws in essentially a log x exp growth rate on bandwidth and storage for every additional peer on the network. Technology can't keep pace, period.
As others said, the Internet Archive may be a good option for these videos. I wouldn't mind writing a system for backing them up to archive.org, but I'm not sure how would I detect them. Marking those videos requires the user to know they should be marked, which just moves the question to how they would know.