So IPFS is basically some kind of automatic behind-the-scene torrenting? I see how that solves the "dead server" problem, but how does that solves the "content longevity" problem?
If you can specify "I've got 1gb spare to replicate other people's assets, as long as each of them replicates mine in return" then your incentive to replicate other people's sites is more redundancy and bandwidth for your own.
If you also prioritise files that have the least replicas across the entire network, and allow servers to re-distribute the content they are replicating, it would surely become quite hard for a file to be lost forever.
This makes more sense to me than having each site pay for their own data to be replicated, because those who don't pay don't get their content replicated, which results in missing content when the server eventually disappears. That is just as bad as HTTP.
How would this stand up to an attack of uploading random data in large quantities?
I have no idea how hard this would be to implement. I'm new to ipfs and crypto.
Example: I'd love to use one of these projects as an additional backup layer for my NAS, but the technical complexity necessary to set it all up makes it.. feel very fragile. Definitely not set and forget.
My hope is that in 10 years something cool comes out of one of these projects, but we're definitely not there yet.