@evan @julian@fietkau.social @julian@activitypub.space @smallcircles re: the Undo Accept thing vs Reject thing, using a Reject is prone to issues with out-of-order deliveries. if you send a Follow and get back an Accept and a Reject with the same published datetime, then are you currently a follower? what if the Reject came first but then you receive a later Accept? it makes it incredibly hard to reason about the distributed state machine, and leads to terrible desyncs when one side thinks you follow them and the other doesn't.
Discussion
Loading...
@deadsuperhero I don’t necessarily think we need the big players. Especially where those projects have stated their intent to stick to micro, or photo -blogging, with limited interop.
@deadsuperhero I don’t necessarily think we need the big players. Especially where those projects have stated their intent to stick to micro, or photo -blogging, with limited interop.
@deadsuperhero imho, building out c2s will allow new projects to bootstrap the front end, and focus on new application domains, or support multiple activity/object types.
Here’s the thing: Mastodon already has a really good API. There’s a whole ecosystem of clients around it, to the point that many other Fediverse implementations adopted it, so that they can use the apps.
I don’t think this is a bad thing in and of itself. But, if we want projects like Mastodon to support it, the value proposition has to provide things that the Mastodon API does not.
I think a killer feature to focus on would be identity management.
"Mastodon already has a really good API".
Not really. It's inconsistent, incompletely documented, stringly typed, has a poor backward/forward compatibility story, and no clean way for other servers to extend the API without risking breakage.
E.g., https://github.com/pachli/pachli-android/issues/2043
All of these are fixable, but Mastodon GmbH has no incentive to fix them.
Yes. Mastodon has always given their own product decisions precedence over healthy evolution of the ecosystem as a whole. And despite many people being very frustrated about that, I think this is perfectly valid decision. After all choosing to implement an open standard should not come with the obligation to maintain/evolve that standard. It is only smart to do so, and Mastodon did this with an eye on their own product development.
Imho it is really the broader dev ecosystem that is at fault in letting the fedi be taken hostage by past Mastodon decisions, making them the post-facto #interoperability leader. As for Mastodon API I'd argue that its users are not on the fediverse. They are on Mastodon.
Identity management may be killer feature, but only when first a sound #ActivityPub foundation is in place. AS/AP isn't as-yet robust enough to be the future of social networking. I'd say the extensibility mechanism is killer feature, and having SDK's and devtools for that.
Right now extensibility of #ActivityPub shapes up as custom app-by-app app-centric development where individual devs just pragmatically throw new stuff on the wire, and when their app gains any popularity or other apps to integrate in a similarish application, things are bolted onto that in random ways. That whole story really constitutes a Big Ball of Mud anti-pattern that only introduces protocol decay, tech debt, and whack-a-mole programming, that is very hard to get rid of once there exists an installed base.
The reason that we do things that way is very understandable. It works in a grassroots environment where indivualist devs find it very hard and not valuable to collaborate at scale in what amounts to a kind of design-by-consensus process. But it comes at a high cost, where interoperability is basically out the door and any app has to be shaped as a pretzel and adopt all the quirks introduced by predecessors in a particular app domain to fit itself on the wire.
It does not need to be that way. I am quite happy after all (after being initially frustrated) by how #ATProto has disrupted things, and opened the eyes of devs in the #ActivityPub ecosystem that we must act or lose out (stay niche, which may be fine too) to the Atmoshpere and how it enables devs to focus on service and product delivery instead of low-level wire plumbing and continuous breakages.
ATProto also shows the way that we can now follow on the #fediverse to catch up again: cocreate a similar robust basis for people to build on. #Bluesky had the advantage of a greenfield start and dedicated team unburdened by past decisions. And they build this whole Lexicon system and ways to introspect functionality.
We can do that too, solve the #LinkedData conundrum, and create an extensibility mechanism that allows devs to focus on service modeling. The more introspection this mechanism allows for, the less design-by-consensus is required, easing expansion to new domains.
@smallcircles @deadsuperhero I've quietly been borrowing ideas from AT Protocol for ActivityPub:
- declaring moderation actors
- supporting content labels (first party abd third party)
- looking at how custom feeds, DIDs and other tech may translate across
I was looking at #ForgeFed which is a very sizable #ActivityPub extension (constituting the "Code forge" app domain in app-centric view, but arguably "Software development" top-level business domain in a service-oriented fedi).
The way that things are modeled here adheres more to the actor model where there's a Factory actor, which in turn creates resource actors that expose various sub-domains. For instance for the management of Issues and PR's there's a TicketTracker actor to obtain via a Factory actor on a forge instance. Though I'm not sure whether I'd modeled that in similar fashion, it is a fascinating direction where we focus much more on good protocol extension design.
All in all AS/AP offers a very granular foundation that allows for very interesting architectures, if only we dare explore them and do not dogmatically stick to some engrained notion how "social media" ought to be. I see #SocialMedia as but a small subset of #SocialNetworking.
1 more replies (not shown)
Bluesky absolutely got it right with ATproto: you can make any kind of client, not just microblogging, and it will seamlessly work with your Bluesky identity. Everything you post goes into your personal storage, and the clients that know how to interpret special data types are able to reach into your PDS and the timelines you’re following to present that stuff.
ActivityPub API needs to follow a similar story.
Why do I care so much about this? A few reasons.
One, I don’t want to have a dozen different accounts across separate types of applications on the same network. I really hate that, it’s messy and does nothing to unify my identity across all the spaces I’m active on.
Two, a seamless login across the entire network could be very powerful for discovery. Instead of having to find people to follow on Pixelfed and PeerTube, the folks I’m already connected to would already be there, right when I sign in.
Three, we could develop a new generation of rich clients that all do really different things, but all tie back to a singular identity. “Sign In With ActivityPub” could work for the entire network, and it wouldn’t have to be a hack using Mastodon’s API.
@deadsuperhero I don’t want a unified experience across services. I use Mastodon, Loops, Pixelfed etc for completely different things. I follow different people. Different topics (hashtags).
My point being, there has to be a layer to separate the services somewhere. Right now it’s at the account level, which honestly is fine for me because hey I’m used to signing in separately to Twitter, Instagram, etc.
But if accounts are unified, at what level do you separate the services?
@deadsuperhero Is your idea that you have one unified account with one list of follows? Because then for each service I’d want to select a subset of follows from whom I’d like to see posts.
For example, imagine I used Friendica or another Facebook-analogous service, and Mastodon. If I had a unified Fediverse account I’d only want to follow some of my friends and family on Mastodon, and I’d want to see almost zero of the Mastodon people on Friendica.
The unified account covers both cases. The current system does not.
With a unified route you can still have multiple accounts. If you prefer to keep things separate — you can.
If the unified account is not supported, not you'd prefer that, there's no way unfortunately.
Finally, I think there are some serious improvements we can bring to fill the gaps that the current spec is missing. What if we had a standard endpoint for notifications?
Suppose we also developed a standard way of doing timelines as well, and used it as a springboard for custom feeds that could work with any app, any client, and any server?
There’s a lot of useful stuff we could do here to make the whole thing extremely compelling for anyone building on this network.
@deadsuperhero@social.wedistribute.org thanks for posting.
I think what makes it hard for existing implementations to support it is that if you're not AS-native (that is, consuming ActivityStreams activities directly without converting to some normalized form for your software), then it's possibly a big lift to "level up" your code to do it that way.
Like Mastodon, NodeBB ingests the activity, extracts the juicy bits, and discards the rest. It makes it hard to later on recall an activity because they were ephemeral (at least in our chain of custody).
So an entire new middleware layer needs to be built for NodeBB to catalog and store (and later, recall) these activities for ActivityPub API support.
Correct me if I'm wrong @evan@cosocial.ca!
@julian @general @evan Yeah, this is a really good point. It’s not currently super straightforward to implement, and as you’ve rightly pointed out, there’s a bunch of extra things that need to be built to support such a thing.
I almost wonder whether there’s a “minimum viable implementation”, like something that just handles authentication and low-level posting to the outbox / fetching from the inbox that might work as a starting point?
Would love to hear what @evan thinks about this.
I am heart to heart with you on this, friend.
@deadsuperhero so, here's my best bet. I can be wrong!
1. Get some servers to implement the API well.
2. Get some must-have clients that run on those servers. This shows the value of the API.
3. Our leading servers shift to supporting it.
That may work; I don't know. It's my best bet right now!
I want to note that WordPress is working on the API!
@django I would object to FedBOX having "signaled interest or intenent".
FedBOX, and the GoActivityPub library that's been built upon, simply "support" client to server and have been since it's inception 7-8 years ago.
@evan Yeah, I mostly agree with this. It’s just that the buy-in is a little bit of a chicken and egg problem. You need servers to adopt it, but you need a compelling first mover. Bonfire, maybe?
The spec definitely needs love, too. I think one of the harder things is building a timeline out of inbox activities. I feel like maybe a future version of the API could specify timelines somehow, whether it’s an endpoint or some kind of basic query? Maybe there’s even a way to implement alternative timelines at that level?
These are all just guesses on my part, but I feel like this could be a gateway to universal custom feeds.
Does the inbox have to map to a timeline, specifically? Mastodon called this out as being difficult to do because you would have to real-time parse the inbox every time you wanted to load the timeline.
Of course one could always reduce the inbox into a single timeline and serve that instead, but then we're braching out with our own proprietary APIs again.
Is that ok?
@julian @deadsuperhero We actually have a meeting tomorrow if either of you wants to come.
https://www.w3.org/events/meetings/3432cf4c-a9fe-4f72-8de5-fa6809b57767/20260219T110000/
2 more replies (not shown)
@julian @general @evan Again, this is sort of why I’m advocating for supporting timelines as a concept in the ActivityPub API. Instead of repeatedly parsing the inbox, we could do exactly what you’re saying with some kind of representation of a timeline. Even if it’s just plain old algorithmic time-sort.
3+ more replies (not shown)