Bluesky, which uses it, has been opened to federation now, and the standard basically just looks better than ActivityPub. Has anyone heard about a project to make a Lemmy-style “link aggregator” service on it?
Bluesky, which uses it, has been opened to federation now, and the standard basically just looks better than ActivityPub. Has anyone heard about a project to make a Lemmy-style “link aggregator” service on it?
ha, no… bluesky is not open to federation. they control the only router and do not allow connectivity to routers not controlled by them.
there isnt a single non-bluesky controlled instance that can federate natively with bluesky.
bluesky is just twitter with a little more user-controllable data sourcing. not that theres anything wrong with that, but its certainly not a part of any federation.
e. suggested reading: https://dustycloud.org/blog/how-decentralized-is-bluesky/
I believe the “free our feeds” people are working to change this though.
And they are either in for one of the following:
There are important features that ATPro has that activity pub doesn’t. I’d prefer activity pub be the winner but they really need to improve some things. Namely, identity. Bluesky identity is more portable.
Identity was already solved with Zot aka Nomad, which is part of the Fediverse and easier to implement than ATProto.
It’s mostly been ignored because it’s just not that important to people, apparently.
We still don’t need ATProto for that. ActivityPods solves that.
ActivityPub itself is built around the principle that the server owns your identity: the best you can do is abandon an identity (i.e, your actor URL) and tell everyone else (via the
MoveActivity) that you are adopting a new identity.The move activity ain’t a great solution. We need federated identity or else ux will continue to lag. When I want to move servers, I can set the move activity but there’s no guarantee my followers will subscribe to the new account. It’s bad ux. Mass adoption is not going to happen with that kind of flow.
Activity Pods is cool bit not implemented on mastodon.
What I am saying is that the ActivityPub protocol is inherently built towards a server-centric system, where identities are owned by the server. Go read the spec: even the “Client-to-Server” specification assumes that the server owns the keys and dictates that the client (i.e, users) must do everything through the API provided by the server (i.e, the client’s outbox).
Anything that is built with a design where the client owns the keys may even be able to interoperate with ActivityPub, but is not ActivityPub.
It’s the other way around. We shouldn’t be looking for “Mastodon on ActivityPods”, but “ActivityPods applications that can talk with Mastodon servers”, and those do exist.
It’s ridiculous they were asking for $30m to do something that ActivityPub already does. Wasted money that could have gone anywhere else
good luck to them. the router piece is incredibly top heavy and not designed for horizontal scaling.
Yep. What do you think the chances are you could write something that does the job of the router and app view, but in a totally off-standard, more point-to-point way?
In the meanwhile, it’s just a matter of bridging, I guess.
the protocol itself creates a barrier to entry preventing other organizations. it puts all the eggs in one basket.
It’s a good blog post, thanks. I made a quick summery elsewhere in the thread.
It’s really unfortunate that we’ve ended up with two populated protocols for federation, both of which have a major flaw. In our case, it’s no established support for moving accounts. In theirs, its a component that’s so bulky the federatability is questionable (and no federated DMs).
What major flaw do you believe ActivityPub has?
This is such a well written piece, it’s closer to a serious article than a blogpost
I find it bizarre and plain wrong to imply that blog posts can’t be serious articles.
agreed. the follow up is just as good.