Instead of private (darknet) bittorrenting I'd rather have a solution for all those files with few seeders who are half dead, in the cloud. It's easy to torrent the latest episode. Everyone is in on that. But what about the long tail? We need a different scheme there.
My idea would be to lump together all the files in a huge torrent. That would inevitably attract many peers, so the problem would be solved if users did selective downloading, as the whole would be too large for any one of them.
The system would need to use some disk space and bandwidth from each peer to host some shards of the whole. Not all the data in the mega-torrent would be downloaded, just some sensible section of it, like, say, 5GB. Between 1000 users, we could have 1TB of files. With a million, we could have a huge library. This way we could have enough seeds for any part of the whole. Of course, it would need a way to add new stuff on the fly and balance the replicas.
tl;dr - I'd rather have a solution for rare torrents. I'm worried for all the content that is not "hot enough". We could amass a huge library of rare stuff, in time.
There are other P2P networks that can give you files based on a hash of their content, for example ED2K and Kademlia. I think even Gnutella supports hash-based retrieval.
BitTorrent clients now have distributed hash tables accessible; it would be nice if magnet URLs could just point to a hash of the file, instead of to a specific torrent.
Bonus: Even ED2K links point to a tree of hashes, so you can verify the data while it is being downloaded.
My idea would be to lump together all the files in a huge torrent. That would inevitably attract many peers, so the problem would be solved if users did selective downloading, as the whole would be too large for any one of them.
The system would need to use some disk space and bandwidth from each peer to host some shards of the whole. Not all the data in the mega-torrent would be downloaded, just some sensible section of it, like, say, 5GB. Between 1000 users, we could have 1TB of files. With a million, we could have a huge library. This way we could have enough seeds for any part of the whole. Of course, it would need a way to add new stuff on the fly and balance the replicas.
tl;dr - I'd rather have a solution for rare torrents. I'm worried for all the content that is not "hot enough". We could amass a huge library of rare stuff, in time.