I see this is downvoted. Actually, if a blockchain used proof of actual useful storage or computation, that'd be great. (instead of wasting energy by computing "useless" hashes)

It's probably downvoted because replying 'blockchain' to a question about how Archive.org stores their data is not even a useful answer.

There may well be fascinating ways one can use blockchain for archives of this sort, but the grandparent comment was essentially useless noise.

Proof of storage is used in siacoin, but not to secure the network. I haven't seen a scheme where proof of storage can be used to secure the network.

In terms of making sure the work that is done is useful it's fairly hard to do that because if you let someone dictate what work should be done then you potentially allow them to do the work upfront and a 51% attack becomes pretty easy.

"you potentially allow them to do the work upfront and a 51% attack becomes pretty easy."

I do wonder if there are potential consensus algorithms that rely upon game theory, the threat of being booted off the network, and the need for some investment before they're "trusted" by other peers. The basic idea is that yes, you could 51%-attack the network with a lot less than 51% computing power, but it would never be profitable because if you're discovered you're booted off into your own little partition where nobody plays with you, and the value you get from being part of the network in the future is greater than the value you would get from rewriting history into something only you believe.

51% attacks are possible now - roughly 75% of Bitcoin hashing power is located in China, so if the Chinese government decides that they want to shut it down, they send the military after the largest Bitcoin miners, say "Hash these transactions or else", and take control of the chain. They haven't found it profitable to do so yet, though - what would they gain from it?

The thing is that reputation isn't built into the protocol. Everyone is basically anonymous strangers.

When space-intensive computations are used for consensus, it's usually called proof of space. There are a few projects doing it already.

https://en.wikipedia.org/wiki/Proof-of-space

Proof of space and proof of storage are two different things.

I suppose that would be kind of interesting. Although if it's conceivable that someone could just store all the useful information then you can't just request a random piece of information as a proof of 'work', so you'd need something clever.

Anyway the comment was probably downvoted because it consisted of a single buzzword.

Blockchain actually could be used to provide data integrity in a distributed environment with known bad actors (presuming that they fall under the 50% threshold), but that is not applicable in this case since archive data is not stored in a distributive fashion by clients (as far as I know). Having worked on a production application that uses blockchain, my take away was it cleverly provides distributed ledger transparency, but at a price that makes it very inefficient as a data store.