[grumbel]

What I'd like to see is a global repository of just pure metadata, file hashes, descriptions, thumbnails, everything you can get away with under the current law. That way you could organize all the data freely and publicly, while you could leave the retrieval to other, potential illegal, parties (torrent, IPFS, random websites, ...).

But due to having all the metadata, one of those parties going down wouldn't be the end of the world, you could just wait until somebody else reuploads it and retrieve it from there.

The main problem after all isn't storing the data, storage is cheap these days, but that the act of mirroring is so damn ugly and brittle. URLs don't last because they encode the storage location, not the content and that's something one could fix with such a metadata database.