Christine Lemmer-Webber just released an extensive article on Bluesky, and it is certainly worth reading. Christine is the co-author of ActivityPub, and currently working on a next generation protocol for decentralised social networks.
While reading I have some personal thoughts and responses, typing them out while going through the article. I'm organising my response per subheading as Christine uses them in her article
On blogs, search engines, and Google Reader
The dynamic of Google Reader is certainly worthwhile to point out here. Technological architecture of decentralisation is super fun, but Google Reader is a good example that technological decentralisation actually doesnt really matter if the Main Organisation goes down.
This goes for Bluesky as well: even if all parts of ATProto were to be fully decentralised, if Bluesky PBC called it quits tomorrow, I dont think the network would survive, purely on social reasons. People join networks because they expect to be there for a while: if Bluesky got bought tomorrow people would either go to Threads or Mastodon, or whatever else immediately pops up, I strongly doubt that people would stick around for a self-hosted version of the opensource software of Bluesky to appear.
I think the fediverse would also suffer this problem with Mastodon to a lesser extent, although likely a small portion of the network would survive.
Self-hosting resources: ActivityPub and ATProto comparison
The problem with critiquing relay costs is that relays are in active development. The latest development of building a non-archival relay brings down costs very significantly: An estimated ~50EUR/month for current network traffic
https://bsky.app/profile/hosaka.systems/post/3lbjr23nzh22p
This would bring in some new interesting critiques of 'who does run the archival nodes, and pays for that and controls it'.
"Message passing" vs "shared heap" architectures
I think this is a good description that showcases how different the underlying architectures of activitypub and atproto are.
I do question the suggestion that a better architecture would be to have most people run their own small GtS instance, with how chatty ActivityPub is. Would not be fun to have an account with a large number of followers on that network topology, to put it mildly. Nor do I think I could keep my WordPress blog connected to the fediverse if this were to happen, unless there were significant further changes with shared distributions (relays? lol) systems on ActivityPub. Every time I boost my blog I currently already DDoS my website.
"A world of full self-hosting is not possible with Bluesky."
I think our language is not fully evolved yet here. What Bluesky does is make a distinction between selfhosting your data and selfhosting your app. The first one is easy (if youre a techy), the second one much harder. We dont have that language yet for a good distinction in selfhosting, so we end up with this type of statement.
Because once you make the distinction between selfhosting your data and selfhosting your app, the next question is "is having the majority of people selfhost their app something we should even strive towards?"
The costs of decentralizing ATProto
See above on costs.
Also, I think the discussions on relays should probably be even grander in scope, and ask the question of why were actually caring about a decentralised relay? To me this is a matter of governance, and I think there are other cooperative ways that we can run relays that gets us better governance. Aiming to get better governance of critical infrastructure simply by duplicating it many times might not always be the best solution, especially here.
Direct messages are fully centralized
"actually, ActivityPub's architecture is (contrary to many users' expectations since Mastodon is also a Twitter clone and is how most people experience the fediverse), designed for direct communication first and foremost... public communication is clearly supported, but the default and simplest case is direct individual or group messaging"
which makes Mastodons lack of UI/UX for DMs even sadder and funnier
A feature complete Twitter ASAP
amen to decentralisation is a matter of degree! same thing I argued here https://fediversereport.com/bluesky-decentralisation-and-the-distribution-of-power/
Bluesky is centralized, but "credible exit" is a worthy pursuit
Multiple devs have spoken out at this point that they dont like the term federation and that it does not describe well how ATProto functions.
I'd definitely push back against the idea that Bluesky is 'federation-washing' here.
did:plc, the "placeholder" DID
"And even if Bluesky delegates authority to that user to control their identity information in the future, there is still a problem in that Bluesky will always have control over that user's key, and thus their identity future."
You can generate a new recovery key that takes precedence over all other keys. https://whtwnd.com/fei.chicory.blue/entries/How%20to%20get%20a%20recovery%20key
Not gonna argue that this is in any shape of form accessible to the normal person at this point in time lol. But I dont think this statement is true in the absolute sense.
Also, now I'm very curious on how a petname system would actually function atproto, it sounds really interesting.
What should the fediverse do?
"But perhaps that's too ambitious to suggest taking on for either camp. And maybe it doesn't matter insofar as the real lessons of Worse is Better is that both first mover advantage on a quicker and popular solution outpaces the ability to deliver a more correct and robust position, and entrenches the less ideal system. It can be really challenging for a system that is in place to change itself from its present position, which is a bit depressing.
This last paragraph applies to both Bluesky and the fediverse, but again, the fediverse is currently actually decentralized, and from my analysis, if there was willingness to take on the work, the gap of moving towards resolving content addressing and portable identity at least are not so large architecturally. But I'm not sure there's interest or not. Maybe there will be so more now."
I think Christine hits on very important notes in the first paragraph, but from that exact perspective I have serious questions about the claim that the fediverse is currently actually decentralised. From a technological architecture perspective, sure it is, but crucially, meaningfully less so from who actually has power to make decisions. Exactly due to that first mover advantage (hello Mastodon) entranching itself. In the end we dont have a lot of safety tools on fedi because a dictator-for-life said no. So much for having a network that is much more decentralised on a technological level.
Preparing for the organization as a future adversary
"Here is where "credible exit" really is the right term for Bluesky's architectural goals. Rearchitecting towards meaningful decentralization and federation is a massive overhaul of Bluesky's infrastructure, but providing "credible exit" is not. It is my opinion that leaning into "credible exit" is the best thing that Bluesky can do: perhaps a large corporation or two always have to sit at the center of Bluesky, but perhaps also it will be possible for people to leave."
I agree here that credible exit is a good framing for what Bluesky is working on building. I also think that decentralisation is also just an endlessly confusing term at this point and I'm unclear what benefits the term decentralisation even brings. This is going to get significantly messier when we get another non-microblogging AppView that has a serious number of users. And even more complicated when we get another AppView thats sort of microblogging but somewhat different (a text-only microblogging appview?). The conversation around decentralisation is just gonna get so messy.
