Just like apps and websites implement “Sign in with Apple” and Google couldn’t we build some kind of federated authentication provider? Then everyone creates an account there and fedi apps can implement an easy way to authenticate users. Even non fedi apps could use it. I imagine user interaction between different fediverse platforms would be much easier too.
I guess could run an auth instance. Ideally everyone would run their own, keeping your data safe.
Is there something likes this already? Saw some discussion here but not much else https://socialhub.activitypub.rocks/t/single-sign-on-for-fediverse/712
Nah mate, I don’t think I want to trust some rando identity server with my login, and self hosting just makes them easy targets.
But why? Just use a password manager instead of tying your identity to a Lemmy instance which you do not control.
Having SSO is reliant on having a single trusted server which has your password instead of you maintaining it yourself. This is just an unnecessary risk.
This right here ☝️
I mean a federated authentication server that you can host yourself if you want.
I don’t understand what you mean about using a password manager, you can still do that. Also your identity is tied to a lemmy instance right now anyway.
I don’t understand what you mean about using a password manager
There’s no point in using a password manager in your scenario. It’s redundant.
Also your identity is tied to a lemmy instance right now anyway.
It’s not, actually. But even if it were I wouldn’t create a profile web of accounts that are all linked together by my Lemmy account.
My Lemmy instance only owns this account, not secondary accounts on separate websites
Right well I guess your concerns also apply to apple sso, google and sso in general. I don’t think it’s something you would use but mainstream users might definitely appreciate the feature.
No one should be using any of that. People use that because they don’t understand how password managers work.
It sound like this is just not for you then. Yep password managers are not user friendly. Single tap sign in button is. Is it less secure? Probably. You’d still have the option to login regularly.
After all the fediverse is for social media. It should be easy to use for the masses. It also means you won’t be storing your bank info in a fediverse app so the damage is not that bad if someone hacks you. Instagram and tiktok accounts get hacked all the time btw and it’s not the end of the world.
password managers are not user friendly.
It doesn’t sound like you’ve used a password manager
Single tap sign in button is.
Single tap sign in is available in password managers via passkeys.
Is it less secure?
It is as secure as the service you’re using to sign in. What it’s not is private.
Again, no one should be using this and we certainly should not be encouraging anyone to use it.
Instagram and tiktok accounts get hacked all the time btw and it’s not the end of the world.
For many of those people, it is a giant problem.
I think I understand the self hosted identity server part, and authenticating with it on different sites. But what is the federated element you’re talking about? What would that instance federate, and with whom?
If we’re moving into a single sign-on for several federated accounts, that’s cool. People have been asking for that for ages! But the identity provider itself wouldn’t (need to) be federated for that to work, right?
I’m guessing it would have to federate with whatever platform decides to implement the single sign on feature. Not sure will leave the details to smarter people haha. Hope some day it could be implemented because it would be pretty useful if it can be done in a secure way.
FWIW, you can setup IndieAuth — on its own or as part of the IndieWeb plugin suite for Wordpress. It may not be quite what you’re looking for, but it’s a step in the right direction?
Since I’ve moved to a password manager I find these social logins less useful. Personal opinion.
Yeah, also if the one login gets compromised, oh boy…
Anecdote time. My first e-mail account got hacked. I still had my Steam account attached to it. Now i have a VAC Ban in CS2 because some chinese kid used it for hacking ingame.
Your password manager’s login can also get compromised
So we should make a remote single point of failure, maintained by someone who probably isn’t a security expert or working on it full time?
No, this is unfortunately the opposite of what we should be doing.
Not a single point but multiple points. Anyway I’m not gonna pretend I’m an expert in security! I just think it’s a feature worth exploring.
Also doesn’t this comes against the decentralization principles of the fediverse?
I imagine the auth provider could be decentralised too
I have a hard time wrapping my head around this one. If you “federate” authentication, wouldn’t that just open it up to bad actors?
Well right now Pixelfed has a sign in with mastodon button for example. Admittedly, I don’t know the details but I don’t think anything is stopping me from running my own mastodon instance just to sign up for Pixelfed.
I agree it might be a nightmare to manage tho if everyone has their own instance but that would probably not be the case.
But they are still useful for a lot of people.
I agree but I also thought this could solve things like mentioning a user across platforms for example.
Yes and no.
Decentralized IDs exist, but will almost never be accepted by any large reputable institution.
Why trust every indie site to be 100% truthful, and definitely not full of malicious haXXors?
Just like fediverse works you can federate with the auth providers you want and ban malicious ones.
Lots of lemmy instances chose not to federate with others. I imagine it could work the same way.
Before an instance does something malicious, how do you know it will be malicious?
Even if everyone there running it, & participating is pure of heart, how can you be assured that haXXors won’t simply break in to take advantage of that trust you’ve given them?
Banning bad instances is a reactive stance that only applies after damage has been done. Can you convince the corporate overlords to take that risk? And it only increases as the fediverse gets more popular, and more instances get trusted.
please correct me if i’m wrong on this. lots of people here saying that it’s not practical because we would have to trust tiny instances that may be malicious. however, what if we make user’s identity provable to anyone, simply by the use of logic? suppose we have a way of generating random proof-theorem pairs (for example, the theorems could be something like “the largest proper factor of n is greater than some m, where m and n are some huuuuuge numbers and n is semiprime”, the proofs could be constructive). we let the identity be the theorem and the password be the proof. hence, anyone is able to verify the indentity by the use of a theorem prover like Agda
Congrats you just invented passkeys
why can’t we use passkeys instead of passwords though? is it just a matter of convenience? if so, maybe there is a way to determine a passkey from a password?
That most likely won’t happen in the Fediverse, unless every developer agrees on some common protocol. But it is possible. See the Nostr protocol and DIDs.
Looks pretty cool!
As others have mentioned in the comments, this might not really work because websites/services would have to trust a bunch of tiny, maybe even single-user instances.
I can see a world where sign-in with Fediverse is possible, but only for a select few instances such as .world, .ml, .ee, and a few other highly-moderated servers.
I think you’re right. This might be the biggest problem but it doesn’t seem impossible to solve.
I believe it’s actually possible to sign in with mastodon on Pixelfed. Wouldn’t that work for a single user mastodon instance too?
I believe it’s actually possible to sign in with mastodon on Pixelfed. Wouldn’t that work for a single user mastodon instance too?
You’re right. I’m not sure if it works with single-user instances, but I believe it does. This is the sort of thing that is technically possible — I believe ActivityPods aims to do something about it, too — but I don’t know… I guess federation can be a bit of a safeguard for this, like having a list of flagged instances that don’t allow account creation; requiring certain thresholds of account age or activity to be passed; stuff like that. There’s also the fact that, being social media, no instance wants bots to run wild, so that could, itself, be a check on that sort of thing, and it might not be economically viable to just host an instance strictly for bot-login purposes, so that is just an inherent barrier to wrongdoing.
Meh, maybe it’s more feasible than not.
That looks interesting too, thanks!