Sounds good, but this FedCM seems to be basically a reinvention of Oauth2/OIDC. Even if it brings some minor improvements (credentials storage in the browser or so?), it seems dead on arrival given that there doesn't seem to be a strong dissatisfaction with how OIDC works. Or am I missing something?

[–] erlend_sh@lemmy.world 4 points 5 hours ago (2 children)

What you’re missing is that OIDC is innately centralized and FedCM, in particular thanks to this work, isn’t.

This is all building on or complementing the same underlying OAuth standards, like the CIMD spec that Emelia originally intended for adoption into Mastodon/ActivityPub to set the stage for decentralized OAuth, but it was never brought in. The AT protocol on the other hand adopted it into their decentralized oauth-atproto standard, which is on track to become a protocol-agnostic oauth-dweb standard.

Anyone who cares about decentralized software should be dissatisfied with how OIDC works. If you wanna use your primary fediverse account to log into other fedi apps, this work is for you.

[–] psycotica0@lemmy.ca 2 points 1 hour ago (1 children)

OIDC is innately centralized

Huh, that's not my understanding. I was there when it first came out, and the whole point was to allow you to use any URI of your choice as an authenticator. Let's see what the first line of Wikipedia has to say:

OpenID is an open standard and decentralized authentication protocol

Huh. 🤔

[–] erlend_sh@lemmy.world 1 points 24 minutes ago

See what CIMD solves for. “Innately centralized” was probably a poor choice of words, but OIDC not a good fit for an open social web with decentralized identities and a plethora of small identity providers that cannot be known upfront.

[–] poVoq@slrpnk.net 4 points 5 hours ago

OIDC isn't "innately centralized", thats just how the majority of people use it. And the same will be likely true for FedCM.