I wrote an article this week about where community actually gets built on open social protocols, taking a closer look at what community shapes have actually been build on activitypub and atproto:
I got some great feedback and comments by and , and figured I'd respond via a short blog, as it deserves more attention than just a microblogging comment
I agree that the architecture of a space matters, but I'm not sure the extent we agree as to how much a protocol really matters, versus what we build with them. Roundabout is built on atproto, and I'm quite happy with the decision. I think our architecture very strongly supports our communities.
Ive been going back and forth on how much protocols constrain social shapes versus how much current shapes are incidental choices by the people building on them. I think protocol code constrains less than people assume, but there are two other layers that matter more. One is protocol design vision, such as the reference implementations and community norms. That's cultural rather than technical, but it's sticky. What Bsky PBC is building does also meaningfully constrain other options by channeling a specific vision of what atproto should look like. The other is user mental models, which I think is the most interesting and most openended for me.
Like how Newsmast has to explain to publishers when they give them their own app that their content ends up on 30k servers they've got no control over. I think that kind of mental model friction ends up being the strongest driver of what community shapes actually get built, more than protocol code itself. Which is why I'm curious about people will perceive Roundabout, like how do you form a coherent mental model of community while also being connected to a global network?
I think we have a similar meta-analysis: structure of AP means identity, moderation, and some visibility are tied to hosting; and that has resulted in social bonds tied to infra. some non-microblogging projects try to get around that. there are some strengths to this, but IMO overall not great
Good point on the labelers, should've included that in the list. That said, governance for how labelers operate can probably deserve and article all on its own. My vibe answer for labelers is that they probably need to be separated out into like four different things:
Mechanical badging (the post 50+ times per day label)
Mechanical moderation (spam detection stuff)
Opinionated curation (this is a good post, this post needs a community label with addtional context)
Moderation (this post broke the rules with harmful behaviour)
Probably all needed for good community formation, but lack of polish as well as the labeling system doing so many different things make it hard to properly talk about.
Bryan's recent article on Community Spaces on this is also excellent:
I think it is important for community spaces not to fall in to the "implicit feudalism" pattern. Governance and control of group accounts should be rich and flexible.
This gets at the core for me, and why I think atproto is super powerful, but also why community formation "hasn't been explored (yet) in the bsky app experience". The implicit feudalism is super important to avoid, and the 'deconstruction' that atproto does (separating data storage and identity and app) is super important step for that. But building it back up into a system that does manage to avoid it is just a hard and novel problem that we haven't really solved yet.
I think there's a lot to be said about how to design this from first principles (the blog post by Bryan is very much worth reading), but also worth pointing out that Ostrom suggests that this sort of community formation usually grows organically. Interesting parallels here with Blacksky, first there was a community, then the technology grew around it as they needed it.