Recently there has been a lot of discourse about ActivityPub and AT Protocol which has been quite dividing and heated.
Uncategorized
67
Beiträge
18
Kommentatoren
0
Aufrufe
-
Recently there has been a lot of discourse about ActivityPub and AT Protocol which has been quite dividing and heated.
Yesterday at the Social Web CG meeting (the group that maintains the ActivityPub and related specifications), I proposed releasing a statement that counters the narrative that one of these protocols must win, when both protocols can co-exist and have a lot to learn from each other.
The statement has been co-signed by various members of both Social Web CG, SocialCG, and the AT Protocol community.
“We do not win by tearing each other down, which only emboldens and empowers those who do not want either protocol to succeed.”
“Arguing between us only emboldens those that seek to derail and destroy efforts to build an open social web.”
You can read the full statement here:
https://writings.thisismissem.social/statement-on-discourse-about-activitypub-and-at-protocol/This was originally in the swicg/general repository, and you can learn about that here:
https://github.com/swicg/general/blob/master/statements/2025-09-05-activitypub-and-atproto-discourse.mdThis is fine. Open protocols are inherently agnostic. The independent efforts on AT Protocol are to be commended, and it may be that AT Protocol has some inherent advantages over ActivityPub. Hopefully this is not interpreted as an attempt to stifle discussion of the current overwhelming dominance of a single US corporation on AT Protocol, making it at this time for all intents a purposes a defacto highly centralized network.
Source: https://arewedecentralizedyet.online/
-
@thisismissem I would add that both protocols support use cases that the other protocol has a hard time addressing. ActivityPub, for example, is much better at point to point communication where no third party overhears what is happening. ATproto, for example, can be used to build “global trending” or a global index much more easily.
I would not be surprised if at the end of they, the open social web would simultaneously end up using both, in a complementary fashion.@j12t@j12t.social yeah I think it would be important to use both. Culturally and technologically there’s aspects of ActivityPub I feel are lacking for important use cases. For example, defederation is huge here. For something like the US government, can you imagine Blue and Red states defederating one another? That’s not healthy nor good. I mentioned that yesterday and someone mentioned ATProto being good use case for that. My point is that you’re right there needs to be both simultaneously
-
@thisismissem Hence my “in practice”. I’m not at all against people trying to do something with AT Protocol, but the proof is in the pudding.
The only thing that I see in my searches for Blacksky is a Peter Thiel project (guessing not the same). It doesn’t mean that small projects are bad, not at all, it means that a project that is obviously named after Bluesky is insignificant in comparison, and everything that mentions AT Protocol will have a relation to Bluesky, and be an excuse to use it.
@ahltorp sorry, what?! (Wasn't aware of that other usage)
Their website: https://www.blackskyweb.xyz/
Source code: https://github.com/blacksky-algorithms/rsky
A podcast about it: https://about.flipboard.com/fediverse/blacksky-rudy-fraser/
-
@thisismissem Hence my “in practice”. I’m not at all against people trying to do something with AT Protocol, but the proof is in the pudding.
The only thing that I see in my searches for Blacksky is a Peter Thiel project (guessing not the same). It doesn’t mean that small projects are bad, not at all, it means that a project that is obviously named after Bluesky is insignificant in comparison, and everything that mentions AT Protocol will have a relation to Bluesky, and be an excuse to use it.
@thisismissem I sort of was in a similar situation (AFS) with a dominant implementation (Transarc, later OpenAFS), and our lesser known implementation (Arla), but Transarc never had anything like the lock-in effects Bluesky has. We were able to make things on a somewhat level playing field and interop just fine.
-
Recently there has been a lot of discourse about ActivityPub and AT Protocol which has been quite dividing and heated.
Yesterday at the Social Web CG meeting (the group that maintains the ActivityPub and related specifications), I proposed releasing a statement that counters the narrative that one of these protocols must win, when both protocols can co-exist and have a lot to learn from each other.
The statement has been co-signed by various members of both Social Web CG, SocialCG, and the AT Protocol community.
“We do not win by tearing each other down, which only emboldens and empowers those who do not want either protocol to succeed.”
“Arguing between us only emboldens those that seek to derail and destroy efforts to build an open social web.”
You can read the full statement here:
https://writings.thisismissem.social/statement-on-discourse-about-activitypub-and-at-protocol/This was originally in the swicg/general repository, and you can learn about that here:
https://github.com/swicg/general/blob/master/statements/2025-09-05-activitypub-and-atproto-discourse.md@thisismissem Exactly. Haven’t these people ever heard of protocol gateways?
The only thing you need is a node that speaks both protocols.
-
@ahltorp sorry, what?! (Wasn't aware of that other usage)
Their website: https://www.blackskyweb.xyz/
Source code: https://github.com/blacksky-algorithms/rsky
A podcast about it: https://about.flipboard.com/fediverse/blacksky-rudy-fraser/
@thisismissem Search for Blacksky on English Wikipedia. In the first 20 hits, only one is not about the Peter Thiel company, and that is about a Japanese race horse.
-
@j12t@j12t.social yeah I think it would be important to use both. Culturally and technologically there’s aspects of ActivityPub I feel are lacking for important use cases. For example, defederation is huge here. For something like the US government, can you imagine Blue and Red states defederating one another? That’s not healthy nor good. I mentioned that yesterday and someone mentioned ATProto being good use case for that. My point is that you’re right there needs to be both simultaneously
@damon I can imagine much worse things than blue states and red states defederating their social media platforms … but I get your point!
-
This is fine. Open protocols are inherently agnostic. The independent efforts on AT Protocol are to be commended, and it may be that AT Protocol has some inherent advantages over ActivityPub. Hopefully this is not interpreted as an attempt to stifle discussion of the current overwhelming dominance of a single US corporation on AT Protocol, making it at this time for all intents a purposes a defacto highly centralized network.
Source: https://arewedecentralizedyet.online/
@mastodonmigration this erases all the hard work of the Blacksky team, along with all the other independent applications that exist like tangled.sh, smokesignal.events, bridgy fed, etc.
Yes, majority of PDS's are currently on Bluesky's PDS servers, however, that's not the full picture, and over time that picture will change.
Additionally, if we look back at ActivityPub adoption, that was originally quite centralized with Mastodon in many ways, and so many building in the ecosystem try to aim for compatibility with Mastodon.
So really, it's just a matter of time and age accounting for the differences.
-
@thisismissem I sort of was in a similar situation (AFS) with a dominant implementation (Transarc, later OpenAFS), and our lesser known implementation (Arla), but Transarc never had anything like the lock-in effects Bluesky has. We were able to make things on a somewhat level playing field and interop just fine.
@ahltorp not sure why you're mentioning multiple completely unrelated projects/companies that aren't even in the social web space.
Bluesky doesn't have lock-in effects, arguably ActivityPub as widely implemented today has more. There are third-party implementation in multiple other languages, for instance Blacksky (blackskyweb.xyz) which is a fairly complete implementation in Rust
-
@thisismissem Search for Blacksky on English Wikipedia. In the first 20 hits, only one is not about the Peter Thiel company, and that is about a Japanese race horse.
@ahltorp well, snyway, now you have the links, you can educate yourself on how much non-Bluesky PBC work is happening
-
@mastodonmigration this erases all the hard work of the Blacksky team, along with all the other independent applications that exist like tangled.sh, smokesignal.events, bridgy fed, etc.
Yes, majority of PDS's are currently on Bluesky's PDS servers, however, that's not the full picture, and over time that picture will change.
Additionally, if we look back at ActivityPub adoption, that was originally quite centralized with Mastodon in many ways, and so many building in the ecosystem try to aim for compatibility with Mastodon.
So really, it's just a matter of time and age accounting for the differences.
Not erasing Blacksky's work at all. It is to be highly commended and holds enormous promise for spearheading real independent instances on AT Protocol.
Hope you are right and AT Protocol is on a real path to statistically relevant decentralization.
But, to say that discussion of the present reality is not warranted, only serves to undermine these efforts. The objective can only be understood in relation to a factual assessment of the current state of the network.
-
@ahltorp not sure why you're mentioning multiple completely unrelated projects/companies that aren't even in the social web space.
Bluesky doesn't have lock-in effects, arguably ActivityPub as widely implemented today has more. There are third-party implementation in multiple other languages, for instance Blacksky (blackskyweb.xyz) which is a fairly complete implementation in Rust
@thisismissem You are perfectly free to ignore my unrelated examples, I’m just providing my personal context for this.
The Bluesky relay is lock-in, since they require considerable resources to replicate if you want to interop with Bluesky. What else is the point of the $30M freeourfeeds campaign? Why raise $30M to break the lock-in if there is no lock-in?
-
Not erasing Blacksky's work at all. It is to be highly commended and holds enormous promise for spearheading real independent instances on AT Protocol.
Hope you are right and AT Protocol is on a real path to statistically relevant decentralization.
But, to say that discussion of the present reality is not warranted, only serves to undermine these efforts. The objective can only be understood in relation to a factual assessment of the current state of the network.
@mastodonmigration right, but you've been given factual information that shows that not all of the network is centralised and that there's many efforts outside of Bluesky PBC, yet you keep going on about it.
We could talk about the centralisation of fediverse software implementations, too, because that doesn't necessarily look great either, for example Mastodon accounts for over 70% of the monthly active users within the ActivityPub ecosystem.
(source: https://fedidb.com/software?vi=list&st=active / https://fedidb.com/ )
Many moderators and server operators are really at the mercy of whatever Mastodon does or doesn't want to ship. Is that decentralisation?
We can agree to disagree.
-
@thisismissem You are perfectly free to ignore my unrelated examples, I’m just providing my personal context for this.
The Bluesky relay is lock-in, since they require considerable resources to replicate if you want to interop with Bluesky. What else is the point of the $30M freeourfeeds campaign? Why raise $30M to break the lock-in if there is no lock-in?
@ahltorp no they don't, it's possible to run a relay for like $30 / month now. PDS's are much cheaper than that to run, and can run on like $5 infrastructure.
You can also move all your data should your PDS shutdown or go rogue, with the Fediverse today, you can only really move your relationships, not your posts, though efforts on that are underway.
-
Recently there has been a lot of discourse about ActivityPub and AT Protocol which has been quite dividing and heated.
Yesterday at the Social Web CG meeting (the group that maintains the ActivityPub and related specifications), I proposed releasing a statement that counters the narrative that one of these protocols must win, when both protocols can co-exist and have a lot to learn from each other.
The statement has been co-signed by various members of both Social Web CG, SocialCG, and the AT Protocol community.
“We do not win by tearing each other down, which only emboldens and empowers those who do not want either protocol to succeed.”
“Arguing between us only emboldens those that seek to derail and destroy efforts to build an open social web.”
You can read the full statement here:
https://writings.thisismissem.social/statement-on-discourse-about-activitypub-and-at-protocol/This was originally in the swicg/general repository, and you can learn about that here:
https://github.com/swicg/general/blob/master/statements/2025-09-05-activitypub-and-atproto-discourse.md@thisismissem I signed this document, as folks can see. My main motivation for doing so is to call for shared efforts to protect emerging, noncorporate social media from being destroyed through state regulations. Currently, that means age verification laws, but of course there have been other proposed or enacted laws that threaten the emergence of alternative social media.
1/2
-
This is fine. Open protocols are inherently agnostic. The independent efforts on AT Protocol are to be commended, and it may be that AT Protocol has some inherent advantages over ActivityPub. Hopefully this is not interpreted as an attempt to stifle discussion of the current overwhelming dominance of a single US corporation on AT Protocol, making it at this time for all intents a purposes a defacto highly centralized network.
Source: https://arewedecentralizedyet.online/
@mastodonmigration Apologies for butting in, but I think https://atp.fyi/network does a better job at showing how decentralized Bluesky/ATProto really is, compared to this site you shared, which, as it explains, only takes PDSs into account.
-
@mastodonmigration right, but you've been given factual information that shows that not all of the network is centralised and that there's many efforts outside of Bluesky PBC, yet you keep going on about it.
We could talk about the centralisation of fediverse software implementations, too, because that doesn't necessarily look great either, for example Mastodon accounts for over 70% of the monthly active users within the ActivityPub ecosystem.
(source: https://fedidb.com/software?vi=list&st=active / https://fedidb.com/ )
Many moderators and server operators are really at the mercy of whatever Mastodon does or doesn't want to ship. Is that decentralisation?
We can agree to disagree.
The issue is the degree of centralization because that dictates the power of the dominant player to assert control. This issue, as you point out, is also a concern, to a lesser, but still very significant extent, for the ActivityPub Fediverse.
As proponents of open distributed systems we should be concerned about concentrations of technology, power and the potential to assert outsized influence wherever they occur in open networks.
-
@mastodonmigration Apologies for butting in, but I think https://atp.fyi/network does a better job at showing how decentralized Bluesky/ATProto really is, compared to this site you shared, which, as it explains, only takes PDSs into account.
Appreciate the link. These kinds of ground truth analytics are important for framing the discussion and establishing objectives for the future.
-
@damon I can imagine much worse things than blue states and red states defederating their social media platforms … but I get your point!
@j12t of course but that’s not great at all. We are colonies not states and if ActivityPub was dominant it would be a much larger issue than you are considering -
@mastodonmigration Apologies for butting in, but I think https://atp.fyi/network does a better job at showing how decentralized Bluesky/ATProto really is, compared to this site you shared, which, as it explains, only takes PDSs into account.
@stefan that visualization isn't particularly great at showing how (de)centralized it is though.
Things are not to scale in it: Single user PDS is as much as 1/50th the area of a Bluesky Corporate PDS with almost 400,000 users.
