My personal views about IPFS - Distributed Permanent Web

Post date: Apr 20, 2015 2:38:57 PM

A few links to begin with:

• IPFS official
• IPFS PDF
• IPFS README.md

IPFS is like running Git which stores all of the objects (files, directories, commits, etc) into a BitTorrent swarm. The files are accessed using file hash so files are automatically integrity verified using cryptographic hash even when using insecure networks. As a bonus it's possible to browse this network using HTTP (Just like Freenet) but even better, you can mount it as a file system. Of course this means that there has to be available seeds for the files. So it's not hosting or storage solution. Of course popular files will be seeded by multiple people, making those quick and reliable to access. But as we know from Edonkey 2000 (ED2K) and other peer to peer (p2p) networks, files go and die after a certain period. Some times even surprisingly quickly. Of course files remain available as long as clients are still sharing the files. So it doesn't matter if the peer / original seed or content publisher goes offline.

I think it could be interesting to implement a discussion board like LclBd or Social network like Facebook, Twitter, Instagram or 4chan using IPFS. At least delivering all static data would be easy using it and it would require much less server resources. Completely another question is if there would be enough peers to support data structures or if server would need to continuously "reseed" p2p network for data. But that's exactly what happens with BitTorrent when ever there is stuff which is mostly downloaded and rarely seeded. It's possible that there are other seeds than the "original official source" but it's also possible that there aren't. This is one of the reasons why there is a support for HTTP seeds. I'm sure this offers interesting possibilities. There has been CDN networks like this already.

This is exactly what kind of direction I assumed that the The Pirate Bay would have been doing when they announced that they're doing state of the art stuff. But alas, they just published some reverse proxy junk, which they claimed to be latest cloud stuff, yawn.

It's really worth of noting, unless you already really acknowledge it that data added to IPFS is only stored locally until another interested party requests it from you. IPFS itself (the protocol) provides no guarantees of redundancy or remote storage. So the original claims about "permanent web" are pure bogus making the idea sound like important and good.