@clacke Yeah I get that, here it's a file on my desktop that gets filtered by a script, there's stuff from nearly 10 years ago in there so I quite trust it. Meanwhile I don't trust a phone for even keeping contacts.
@iceloops@clacke@sun yeah, I hopped to the Friendica side of things with friendica.eu / social.hacktivis.me / snarl.de at some point though so I could connect with the folks at diaspora without yet-another-account (and then diaspora got irrelevant).
@roboneko@dielan But Tor has yet to have a petname system, right?
And even with a petname system… How do I easily give say an onion address to someone when AFK? Or do you just assume everyone is terminally online? (increasingly true to be honest)
@dielan Regardless of whatever Mastodon branding crap. Censorship (as in abusive admins) on fedi is annoying but it's a much lesser version of what's on Twitter and it fixes itself with people moving to better instances. Moving to a better twitter instance? You can't, or that's called moving to fedi.
@dielan Registrar/TLD censorship or even just instability has existed for decades prior though, it's why most seasoned ~admins prefer OG net/com/org/… or ccTLDs of their own region.
@dielan And to me, fedi without using DNS wouldn't have worked more than say Tox did, because human-meaningful names are quite critical. And there has yet to be a true alternative to DNS with adoption beyond early-adopters.
@dielan Yeah, I think nostr handwaves away a lot of problems unique to it's style and we're going to see them get bigger plus discover new ones as it gains adoption.
Like who could have acknowledged stuff like fediblock back in GnuSocial era? Specially before servers like smugloli.
Or that we could solve the account migration problem either with public-key cryptography (friendica-hubzilla-zap lineage) or just forward and back references (mastodon/pleroma/…). Both having pretty massive pro and cons but none being really the best.
@roboneko@dielan Public keys: - Weaken over time together with the algorithms and their parameters (Like dsa1024-sha1 OpenPGP keys are probably way too weak to be significant) - Game over if you leak your key, but also if you loose it. Much more dangerous than say a password or an OAuth token. - You need a very good cryptographic implementation and a good source of entropy. (see: logjam, debian openssl, …) I wouldn't trust a browser to provide this.
That said I could find Erlang BEAM VM style of OS be interesting but uuh… that has processes, lol. And if you would want it at OS-level you will want a different kind of hardware protection, which pretty much doesn't exists outside of laboratories.