AT Protocol vs. ActivityPub
AT Protocol vs. ActivityPub
Which is the better protocol overall based on the current information?
You're viewing a single thread.
As far as I know AT isn't actually being used anywhere at the moment. Bluesky has a single server with closed registration and no federation partners. Despite being open source, AT isn't really intended to be an open protocol.
ActivityPub, on the other hand, has a few hundred servers including several dozen large ones all federating nicely together.
What do you mean it's not intended to be an open protocol? There is no other reason for it to exist
Bluesky's protocol implementation is open source but the protocol isn't defined by any standard. Once its open and federating, if that ever happens, anyone who wants to connect will be entirely beholden to the latest published version from bluesky and whatever protocol documentation they provide. They're starting in the middle of the EEE playbook, anyone who wants to join in has to chase them.
There isn't a standard because the protocol/platform doesn't exist yet. "Anyone who wants to connect will be entirely beholden to get the latest published version from Bluesky" is just the definition of a standard. Every standard is maintained by someone. And its also not EEE to make an entirely new system. They are neither embracing nor extending activitypub. They are trying something of their own.
So when Bluesky introduces a new feature or a breaking change in the protocol anyone downstream will find out when it gets pushed, maybe a little ahead of time when it comes in as a pull request. Bluesky goes live with the change immediately, maybe in a public beta channel, maybe straight to prod, depending on their testing setup. Anyone running a bluesky compatible server becomes immediately incompatible until they rush to implement the new changes. The best user experience will be had on first party servers, driving the vast majority of users there.
For a standard defined protocol, like ActivityPub for example, to introduce a change like that it would first go to the standards comittee where it would be discussed publically with stakeholders. The changes would be published and then all parties would begin implementations at a pace that makes sense to them. It's like when you hear about new Wi-Fi versions several years before any devices actually support them. One group doesn't just get to come out with some crazy new change that everyone else has to reverse engineer and then race to keep up.
What Bluesky is doing might be fine and make sense for their model, whatever that may be. I just want to point out that there is a difference and it drastically changes what the future of the service will look like.
Do you have any sources for that? I've seen no indication they don't intend to release the protocol as a standard and that is a pretty big assumption
Nope, I've seen nothing firm about them doing or not doing it, but OP asked about current information and at the moment they are open source and do not have a standards document.
As it stands I wouldn't jump on AT for a project of mine and wouldn't recommend it as a superior protocol on those grounds.
I wouldn't recommend it either, because it doesn't exist yet. But its extremely disingenuous to make claims about how it will work when/if it begins federating in the future and declare it EEE.