People are talking about IPFS a lot lately, and "interplanetary file system" sure sounds amazing, but not being able to delete anything is a concern imo
@gargron Of course you can delete stuff, you just can't necessarily trust others to delete their copies.

Much like OStatus, in fact.

@Gargron I've taken to considering any thought that leaves my brain as read-only forever

For digital data, it's anything that leaves my computer. While that's inconvenient at times, it's the safest way to ensure something I don't want shared forever remains private

@Gargron
Hmmm.
Should it really be INTERplanetary??? That implies to me between planets.

I know we have stuff off planet that is kind of internet connected now, but not much really 😃

Other than that quibble. It does sound rather cool.

@Gargron I can see the non-deletion thing being weird if you had to transmit data dumps out to Mars.

@Gargron That is definitely a concern!

I have worries in the other direction as well: unless content is "pinned", it can eventually be garbage collected by the nodes it is on. And it only gets onto nodes if someone requests it.

So unless you are running a node yourself, it seems like everything rots eventually.

@Gargron reading https://github.com/ipfs/papers/raw/master/ipfs-cap2pfs/ipfs-p2p-file-system.pdf what really frightens me is the guarantee that data can disappear at any time. 😃

@Gargron It's a really cool idea, but it sounds like you'll need to use it for somewhat specific use-cases (storage of published documents, for example).

One thing that remains unclear is whether or not documents stored in #IPFS can in fact have revisions. Maybe you can't delete documents, but it might be possible to update the document itself to be blank, as a type of "soft delete"?

@deadsuperhero @gargron You can delete objects, but you can't force others to delete.

Revisions I think are a feature.

But you can't (again) force people to apply them.
@deadsuperhero @Gargron
Depublication of popular content on the web is hard in general, not only on IPFS. Removal of unpopular content is easy in both systems. Just take it offline and hope that everyone forgets about it. If noone keeps a IPFS hash online, it will vanish.
@Gargron seeing how it's complicated to recover critical files from backup, I'm not concerned with undeletable file.
@Gargron Yeah, that gets archivists happy but not great for privacy. Something like https://upspin.io/ might be better for a social network application.
@Gargron What....like not being able to delete an account?
He he he he.
😛
@Gargron Well, but the same rule that governs IPFS holds for the Internet as a whole, no? You can delete stuff from *your* computer, but if it has been published anywhere once you can never be sure that no one else has a copy and publishes that copy again (although, to be fair, IPFS makes that "publishes that copy again" more seamless than it was before).

@Gargron What worries me more about IPFS is that many people don't seem to realise that your node publishes provider records of any stuff you add - this means that you cannot rely on just keeping the IPFS hash private, everything you add will be announced to the world unless you encrypt!

I think default encryption will be the way forward for IPFS - that would also help against your delete worries a bit - if no one has the decryption key anymore a file is as good as deleted.

@Gargron but that's why its good for services like Wikipedia

@Gargron

It's really the hash name you can't delete, the data might be there. If you think of the name of the file is unique to that file if you change the file then the name or hash changes so the new hash points to the new file and the original hash still points to the original. But unless you pin the data somewhere the hash may point to a data block that has now disappeared. And how sure are you that if you delete something in the cloud that company has actually deleted it?