That already exists. Peers find each other through a distributed hash table which can be bootstrapped from a variety of sources.

I would say the problem is discoverability and actual deployment.

For suddenly popular files it can be a way to donate bandwidth most of all, because then there might suddenly be a lot of peers. For the vast majority of files however, there won't be any other peers and they have to be web seeded by Archive.org either way.

Then there's the discoverability problem. Ultimately you need something like a magnet link to connect to a swarm.