h is a user on social.coop. You can follow them or interact with them if you have an account anywhere in the fediverse.

@cjd Talking about cjdns. In addition to its stated utility, I was wondering whether it could be useful (and in a way, repurposed) to help reliability of to help provide a kind of ipfs pinning on top of GNUNet that lives on forever, given enough number of nodes voting for the pinned content.

h @h

@cjd Some for-profit services hosting in the corporate cloud are starting to appear defeating the whole point of using , recentralising efforts of decentralisation. Some guys never learn.


So I'm thinking the usefulness of the mdag wouldn't need to be diminished, if corporate fascist pinning can be replaced using the Libre infrastructure parts that are already in place. Useful things like dbpedia and libre blockchains could be pinned there.

@h @cjd I've played with IPFS before, and I've certainly seen there needs to be some way to get your stuff pinned somewhere before the magic of IPFS persistance begins to kick in.

Also given just how large modern harddrives are, and the stopgap nature of the services these companies offer, I see the harms as being relatively minor.

That said I do much prefer FileCoin as a solution to this problem.

@alcinnz @cjd The problems I see with the services these companies offer are not so much cost or capacity, the problem I see is that if these services by small startups become popular, and they just rent resources from Microsoft Azure, Amazon WS, and Google, more than 50% of IPFS resources will end up hosted in one of their data centres, utterly defeating the whole point of p2p tech. A situation worse than the www.

That's why I propose something like + + pinning.

@h @cjd Too true, although at least we'd have a peer-to-peer caching layer if that becomes the case. Something that was tried and failed on HTTP.

But then if it's just a company with a huge (and readily available) harddrive, they can be beneficial. So we do need to discourage the others.

@alcinnz Of course I have exactly nil power to discourage anyone from doing anything. But you probably don't want to discourage me from trying this other pinning system 😃


@h @cjd Ofcourse not, can you explain it some more?

I would probably have to start with a refresher on how GNUNet and cjdns work, I assume you're familiar with them?



I propose that cjdns be used as a routing system helping to securely and reliably map ipfs resources to gnunet resources, and use gnunet as a pinning system, providing the necessary compatible interfaces to ipfs pin mechanisms.


So this imaginary modified cjdns router would present an ipfs-compatible api that behaves exactly like the an ipfs peer that's capable of pinning, and it would map resources existing in gnunet and make them available to the ipfs network.

@h @cjd Ah! An IPFS/GNUnet bridge by way of CJDNS.

Very interesting... Please toot when you have something!

@alcinnz I'm not going to have anything any time soon, swamped with other things. But I thought I would discuss whether it makes sense to @cjd , probably the person who understands cjdns best in the world 😃

@h @cjd That said I do like FileCoin better as it allows pinning services to compete behind the scenes (and apply "coopitition" to do a better job), whereas end-users don't have to see anymore than a single crypto-currency payment.

While I'm not as extreme as many crypto-currency fans, I can be skeptical that federated systems won't just become centralized. But then Mastodon gets away with it thanks to local timelines, and CUPS does even better because physical distance is a factor. to users.