DOIs with infohash?

[sneakers-the-rat]

I just had an idea thinking about using content-addressing/hashing for persistent identifiers, and I'm wondering what y'all think: if you can summon the funding for the $200/year-ish membership fee and $1 registration fee, what about using the infohash of a torrent as a DOI? they'd be longer than usual but I don't think there's any length limit.

so like doi:10.<academic torrents prefix>/<hash>

[ieee8023]

What use case does the DOI address that is not already handled? If people don't want to write the full URL people can just write the infohash in a publication and look it up on google or query the DHT to get the metadata for it. The DOI would just lead to the website which would still be a single point of failure.

[sneakers-the-rat]

I was just thinking in terms of making datasets citable in a way that - should the site go down - still allows the data to be indexed because the hash is independent of the DOI resolution. like, pending the future we're going to build where URL citations count towards "academic credit" this would be an intermediate step similar to how you can cite a dataset on zenodo by doi.

just an idea! (noting that I am not a fan of the DOI system being the only indexing system, just trying to think of ways to incentivize more p2p data sharing :)

[ieee8023]

Oh it would be interesting if the DOI url forwarded to a magnet link. Then the DOI indexers would essentially be big search engines for torrents. I'm not sure if that type of URL would be supported though. I think with most datasets, it is the dataset description paper that matters more and should be cited. And that can be put on a preprint server to get a DOI.