In 2023, Bluesky's CTO Paul Frazee was asked what would happen if Bluesky ever turned against its users. His answer: "it would look something like this: bluesky has gone evil. there's a new…
In 2023, Bluesky's CTO Paul Frazee was asked what would happen if Bluesky ever turned against its users. His answer:
"it would look something like this: bluesky has gone evil. there's a new alternative called freesky that people are rushing to. I'm switching to freesky"
That's the same argument people made about Twitter. "If it goes bad, we'll just leave." We know how that played out.
The promise
Bluesky is built on ATProto, an open protocol. The pitch is simple: your data is yours, your identity is yours, and if you don't like what Bluesky is doing, you can take everything and leave. Apps like Tangled (git hosting), Grain (photos), and Leaflet (publishing) all plug into the same protocol. One account, many apps, no lock-in.
It sounds great. But look closer.
Where your data actually lives
When you use any ATProto app, it writes data to your Personal Data Server, or PDS. Your Bluesky posts, your Tangled issues, your Leaflet publications, your Grain photos. All of it goes to the same place.
For almost every user, that place is a server run by Bluesky.
You can self-host a PDS. Almost nobody does. Why would they? Bluesky's PDS works out of the box with every app, zero setup, zero maintenance. Self-hosting means running a server, keeping it online, and gaining nothing in return.
To be fair, migration tools exist. You can move your account to a self-hosted PDS for as little as $5 a month. Bluesky has made this easier over time and even supports moving back. But this only works if you do it before the door closes. If an acquirer disables exports, it doesn't matter that the tools existed yesterday. And we know from every platform transition in history that almost nobody takes proactive steps to protect their data.
The flywheel
Here's the part that worries me.
Every new ATProto app makes this problem worse, not better. Each app tells you "sign in with your Bluesky account", which really means "write more data to Bluesky's servers." The more apps that launch, the more users depend on Bluesky's infrastructure, the less reason anyone has to leave.
The protocol doesn't distribute value across the network. It concentrates it. Developers are building features on top of Bluesky's infrastructure for free, making it more indispensable with every app that ships.
And Bluesky gets to claim the moral high ground the whole time. "We're open! We're decentralized! You can leave whenever you want!" Meanwhile, the switching cost goes up every day.
The chokepoints
It's not just the PDS. Bluesky controls almost every critical layer:
The Relay. All data flows through it. Bluesky runs the dominant one. Whoever controls the relay controls what gets seen, hidden, or deprioritized. Third parties can run their own, but without the users, it doesn't matter.
The AppView. This is what assembles your timeline, threads, and notifications. Bluesky runs the main one. If it goes down or goes hostile, every client that depends on it breaks.
The DID Directory. Your identity on ATProto resolves through a centralized directory run by Bluesky. They've called it a "placeholder" since 2023 and said they plan to decentralize it. There's still no timeline.
At every layer, the answer is "anyone can run their own." At every layer, almost nobody does.
The Gmail problem
Email is an open, federated protocol. Anyone can run a mail server. In practice, running your own mail server is painful and everyone just uses Gmail. The protocol being "open" didn't prevent centralization.
ATProto might be worse. With email, at least each app connects to your own server. With ATProto, each new app adds more data to the same centralized PDS. The open protocol is actually a centralization flywheel.
What happens in an acquisition
Say someone buys Bluesky. They now control:
- The PDS for nearly every user
- The main relay
- The main AppView
- The DID directory that resolves every identity
They could disable data export. They could cut off third-party apps. They could shut down federation. They could insert ads, shadow ban users, deprioritize content.
And the blast radius isn't just Bluesky the social network. It's every app in the ecosystem. Your git issues on Tangled, your posts on Leaflet, your photos on Grain. All stored on infrastructure now controlled by the acquirer.
The protocol says you can leave. But the company that just paid billions for the network has no incentive to let you.
I like Bluesky. I use Bluesky. The team seems to genuinely care.
But every counter-argument to the concerns above rests on the same foundation: technically, users can leave. Technically, you can self-host. Technically, you can run your own relay. The capability exists at every layer. But people don't do these things. They never have with any protocol. Not email, not RSS, not XMPP. The default wins. Always.
And then there's the money. You don't raise $120M at a $700M valuation to run a public utility. Those investors need a return. That return comes from monetizing users, getting acquired, or going public. All three create pressure to consolidate control, not distribute it. A truly decentralized network where users can freely leave is worth less to an acquirer than one where they can't.
The PBC structure is supposed to be the safeguard. But PBC obligations are vague and untested in court. When $120M in VC money is on one side of the balance, guess which way it tips.
The protocol can't save you from incentives.
Read the original article
Comments
By yellowapple 2026-02-2121:285 reply > At every layer, the answer is "anyone can run their own." At every layer, almost nobody does.
And at every layer except for maybe the PLC directory, there's nothing stopping anyone from fixing that “almost nobody does” problem. The fact that such a thing is even possible, and that it's seamless to move from one to the other, gives ATproto a massive leg-up compared to even other federated systems, let alone its non-federated predecessors.
Yeah they're describing a real problem, but the cause of that problem—a seamless centralized sign-up funded by VC money—is the reason bluesky took off to begin with.
Bsky offers an on-ramp to a more decentralized experience, but most people won't pay the money and experience the friction to move take that ramp. Platforms like Mastodon are entirely decentralized, but that means the friction of decentralizing happens immediately upon sign-up. The people who don't want to self-host PDSes never signed up for Mastodon to begin with.
I try to be skeptical, but I feel like bsky (or something like it) is the best way can do re: bringing decentralization to the masses.
By user3939382 2026-02-234:25 Or you get a Facebook XMPP situation where the federation is a cool feature few use until the platform’s mature enough to say actually no.
By account42 2026-02-2313:49 > Platforms like Mastodon are entirely decentralized
They are not, they're federated and that distinction really matters here. A decentralized platform would be designed to make running your own single user or at least small instance the default but neither ActivityPub nor ATproto do that.
>And at every layer except for maybe the PLC directory, there's nothing stopping anyone from fixing that “almost nobody does” problem.
If there's nothing stopping anyone from fixing a problem, and yet nobody fixes it, then there's something is stopping them.
Might not be a technical impossibility, or a gun in their head. Could be as simple as inertia or addiction.
But saying "the problem is totally solvable" just because there's a solution available, is pretty naive. Solutions have costs themselves, and not all are created equal or equally feasible.
By davidgerard 2026-02-2123:161 reply A relay or appview needs a ton of resources. Blacksky finally created the second ever real-world usable appview instance after 2.5 years.
Also, the open source version of the appview doesn't work at Bluesky scale. You need a proprietary database for sufficient speed.
AT Proto is completely decentralised, except for all the structural and financial points of absolute centralisation.
By verdverm 2026-02-2123:55 The relay is not that bad, the only really bad part is building an index, and most apps on the atmosphere have no need to index bluesky records, so the economics for them look very different.
The work towards permissioned data and group-shared data will make it so apps can choose their own levels of "decentralization" of "federation" on atproto primitives. For example, two diametric options
1. An app that is not open source code, but still does all the same atproto credible exit stuff. Naturally leans into winner-take-all
2. An app that is tied to community, think something like Discord, where most servers don't care about what other servers are doing. Each community could run their own version and only care about their data. This is raspberry pi hostable.
Or people don't think it's a problem!
Maybe there are a ton of people who joined Bluesky because twitter devolved into a room-temperature-IQ right-wing hell hole, not because they cared about federation or whatever.
Everything has trade-offs. Again and again people choose centralized services because they are a better product.
It's exactly that. I have an account on Mastodon that I haven't opened in months. I use Bluesky a couple of times a day. On Mastodon I couldn't find interesting accounts to follow for weeks. On Bluesky I was up and running after an hour thanks to starter packs. Ease of use trumps (what a word!) philosophy for me. And probably most other people too.
BTW I already lost 10 years of posting on Twitter. Did not care for a second. Do people REALLY care about their postings on micro blog sites? It's not like a box of photographs that I would pass to my children on my deathbed...
By black_puppydog 2026-02-228:541 reply Just FYI, starter packs were implemented on Mastodon, or rather, around Mastodon quite a while ago, so maybe give it another shot.
That being said, the nature of Mastodon does still make it more difficult to find interesting accounts.
By oskarw85 2026-02-2217:33 Thanks, that's awesome! I will give it a try because I love idea of federated networks a lot. Just couldn't find my way in the jungle.
By coldtea 2026-02-2214:06 >Or people don't think it's a problem!
Often that's a problem on its own (e.g. climate change)
By neves 2026-02-230:39 This was the reason I went to Bluesky. I understand the political advantages of decentralization, but centralized services work better.
Remember that SMTP, the most decentralized platform that worked, now is centralized in a handful of companies.
By phantomathkg 2026-02-223:13 Fixing the problem requires 2 resources, the knowhow and the money. People need to know how to execute it safely, and people need to have the disposable income to run their PDS.
Even for tech people in HN, not everyone will have the disposable income to self-hosted every digital life lands on. Somehow, somewhere one may need to use free services paid by VC money.
It's really the defaults that need to be fixed and anyone cannot change those.
I assume the onboarding steers everyone to one PDS provider and the mobile apps only use one appview server.
By verdverm 2026-02-2218:22 Multiple apps are running their own PDS, new users to their app go there by default, but it is app by app.
I believe every bsky mobile app is likely pointed at a single app view, but there's no reason they cannot offer a setting, which now after typing that, maybe I do recall someone saying they did that with their app.
There are dozens of Bluesky apps now, some even have highly requested features. I't part of the beauty of ATProto, anyone can build a client for any one (or more) app views. There are alternative ways, eg. use Slices for queries, write directly to PDS, no app view needed for any app in the atmosphere.
It really feels like folks are trying to find a single whole they can poke and claim some win. Time continues to flow, the atmosphere continues to grow, and if you get tuned in and join us, Oh The Places You'll Go!
By kelvinjps10 2026-02-222:282 reply I feel it enters now in the territory of being more confusing for users and having less adoption. It's tricky to create a decentralized network for the masses.
By verdverm 2026-02-2123:04 I explained in another comment you also replied to. It's not broken, it works as intended, the plan is for improvements and de-risking.
Is there something missing from my answer about what the plan is for the PLC?
By yellowapple 2026-02-2123:09 I don't think “they” have a whole lot to fix. It's more a matter of people needing to fix their own laziness.
I'll be the first to admit I'm guilty of this, too, and still haven't gotten around to moving my main account to a self-hosted PDS (though I've at least taken the steps to backup my CAR and set my own rotation keys, such that if my PDS goes offline or hostile I can still migrate away from it).
Yeah I’m the guy quoted in the opening of the article.
Yes. Be wary of Bluesky. That’s our whole point. Run the infrastructure on your own. Build separate companies.
Most of the complaints here are just about the cost of scale. You are able to fetch the whole network and its history, and that costs time and money. The only structural centralization is PLC, which is being factored into an independent org.
I'd like to encourage anyone who is wary of Bluesky to check out Paul (and Dominic's) back-in-the-day project Secure-Scuttlebot which solved most of the issues that Bluesky suffers from by using content addressable storage and signing key cryptography correctly.
The actual SSB codebase has been kind of broken since 2020, but I have a fork on my own Github that works and comes with a basic client that you can vibe/claw on top of: https://github.com/evbogue/ssbc
I'm happy to supply pub invites to anyone who wants to play around with the old sbot with me as we work towards making social media distributed again.
> Secure-Scuttlebot which solved most of the issues that Bluesky suffers from
I've heard Paul speak about this the other way around, that the experience from SSB informed the design of ATProto. I.e. ATProto solves most of the issues in SSB
For clarity, ATProto is the protocol, Bluesky is one dozens of apps, obv the biggest and most well known outside of the ATmosphere.
Bluesky does solve a lot of SSBs problems. Both projects can learn from each other. The past can become the future and the present inform the past.
This isn't just on Paul, Jay has publically stated that she doesn't believe users (even powerusers) can be trusted with keypairs.
By yellowapple 2026-02-2121:342 reply Jay's unfortunately not wrong about that. Hitting that balance between “so secure even I can't access it anymore” v. “so convenient that cybercriminals can access it, too” is less trivial than a lot of the “just use keypairs” crowd likes to admit — even for those of us with many years of experience working with SSH and PGP keys, let alone people who haven't the slightest idea what a “keypair” even is.
By advael 2026-02-2214:07 Keypairs are fairly easy to use if you're on a reasonable unix-like OS and if you're not then frankly nothing is easy to use. Unfortunately this does mean that your statement is true for the majority of devices people use to access social media
By evbogue 2026-02-220:16 It can be both.
By sydbarrett74 2026-02-2211:12 Kudos, evbogue. Thank you for the hard work you've done to keep this alive.
By bjoli 2026-02-2120:30 Considering how hard it has been, and to some extent still is, to run your own Bluesky instance, the main problem is that it automatically becomes centralised in a way that no open protocol will solve.
If 97% of your users are on one instance it is not a distributed platform. Applying this to mastodon, I am pretty sure most people would consider it a problem if mastodon.social started getting more than 40% of active users (currently at about 15 iirc).
Moving the PLC to an independent org doesn't make it decentralized.
I don't seem any claim in GP's comment that it would make it decentralized. It does seem, by looking across your comments in this thread, that
(1) You feel very strongly about what decentralized means w.r.t. social media, bluesky, and the PLC
(2) ATProto accepts that it's not planned to be as decentralized as some want, and that it is currently centralized with secondary validators.
(3) No answer or plan for the PLC is going to satisfy you. Nor is any argument you make going to change the plans for identity in ATProto for the foreseeable future.
This is all fine, people can have different perspectives and work/play in different ecosystems, no one is right or wrong. This is precisely why there are multiple protocols out there and bridges between them.
May I then ask why you keep making comments to the same effect aas those you made in the post and multiple times here ~12h ago?
By southerntofu 2026-02-2210:211 reply > This is precisely why there are multiple protocols out there and bridges between them.
Yes, that's great! What's not great is Bluesky attempting a hostile takeover on federated and decentralized social networks. It's been advertised from day 1 as an alternative to centralized silos and it's a lie. [0]
To be fair, projects like Blacksky try to decentralize it (except the identity server, as it's impossible??), and there's now a vibrant developer community around ATProto. That doesn't make the centralization and false marketing claims any less problematic.
Develop the protocol you want. Don't lure my friends into it by pretending it's something that it's not.
"lure" comes off as real biased, what reasons do your friends give?
> What's not great is Bluesky attempting a hostile takeover on federated and decentralized social networks.
Can you clarify this? Am I unaware of some active, intentional campaign?
By southerntofu 2026-03-037:36 > what reasons do your friends give?
In my (arguably not very representative) circles, unlike the big Facebook->Instagram migration, which was motivated by "i don't like that it's run by a Silicon Valley tech-bro neofascist, but that's where everyone's at", the Twitter->Bluesky migration was motivated by "finally an alternative that's not centralized so it can't be bought and controlled by american neonazis".
> Can you clarify this?
Well Bluesky's number 1 selling point was always decentralization. Just looking at a few past articles from the wikipedia page's sources:
"Twitter CEO Jack Dorsey announces a new research team, called Bluesky, to create a set of open and decentralized technical standards for social media platforms" (CNBC)
"Bluesky now operates as a “decentralized” social media platform, which means users can create their own servers on which they can store data and set their own rules" (Forbes)
To be fair, they did fit some of the bill which is now why we're complained that Bluesky is not 100% federated/decentralized. And they did improve compared to AP in terms of nomadic identity and letting users know everything on the platform is public (unlike Mastodon where people had a false sense of security).
The whole premise of a free social media protocol is that it is resistant to hostile takeovers. All issues stem from this.
1. I absolutely feel very strongly about decentralization. If there is a part of the stack that isn't it opens up the whole project to the kind of issues I'm talking about in the blog post.
2. Then it is not made to be resistant to the above problems
3. Actually, this is where you are wrong! If atproto implemented a more robust, decentralized default identity system I would be a very happy camper.
I make comments because I care about the subject, obviously. I use Bluesky a lot and I don't want it to end up like Twitter.
By verdverm 2026-02-2217:46 There is not one right answer and being hostile towards atproto doesn't earn you friends or support for your ideas.
I recommend adjusting how you argue for your position, especially the tone. If you want to pay it forward, and repair some the damage to your ability to make proposals within the atmosphere, I would further recommend you write an alternative version with a leading apology and take down the antagonistic version you have published.
Does the existence of did:web make it decentralized? You don't have to use the centralized identity provider at all. And if you own a domain why would you?
When reading any essay about the perils & merits of Bluesky's architecture, save yourself some time by searching for "Blacksky" in the post. If they don't address Blacksky, more than likely the author's understanding of the space has major gaps.
(Blacksky is the/one of the furthest along in building competing versions of each part of the AT proto stack.)
I know very well what it is, it doesn’t change anything in the grand scheme of things. I wish it did!
By runako 2026-02-211:21 Re-reading my reply, it is worded more harshly than I intended. My apologies.
I do think it's a critical omission to not address the main player(s?) who are working on key parts of this, and where they may yet run into problems.
But how is that 'decentralized' which was the entire point of Bluesky and the AT protocol to begin with? We're just back to running centralized services. Without decentralization this is just XMPP with extra steps. You might as well just run something like Movim and save yourself the hassle.
There's "decentralized" in the sense that every device runs the whole stack. In an analogy to another protocol, this would be like running SMTP and IMAP on your phone and laptop.
Then there's "decentralized" in the sense that the protocols that govern are open and anyone can plug in without permission. This is how email works in practice. Most people do not choose to run their own email servers, but they nonetheless benefit from the fact that people who are interested can do so and provide email service.
Bluesky is the second kind of decentralized.
>Bluesky is the second kind of decentralized.
But why do we keep getting articles trying to convince us that it needs to be the first kind?
By verdverm 2026-02-2218:35 There are some people in other networks who feel very strongly their answer is the right answer to the great question of decentralization [insert south park atheist otters]. I think they are in part frustrated that ATProto (not the "right" answer) has attracted the users and developer. The meanness and lack of curiosity certainly provide the undertones to justify this interpretation.
By kevinak 2026-02-227:37 Because centralization matters. It is what stops a hostile agent from ruining things. There is no real win in being "semi-decentralized".
By runako 2026-02-2215:59 Purists. There are some people who run email for their personal domains on Raspberry Pi machines sitting in their homes. Maybe they want everyone to live the same way?
Personally, I think it's better that there is choice. I do not want to run my own social media site any more than I wanted to run an IRC server.
By verdverm 2026-02-2120:34 > the entire point of Bluesky and the AT protocol
is really to find a good enough middle ground that has competitive enough UX to get people off of the fully centralized, locked in social media providers. In the broader context, ATProto to me means user choice and provenance, which ATProto does better than any other protocol. See all the parts beyond just data hosting, where the entire distributed system is plug-n-play. [1]
ATProto not being purist, preferring pragmatism, is what attracts me over alternatives like AP and Nostr.
[1] https://atproto.com/articles/atproto-for-distsys-engineers
Does it require people change defaults? If so then 99% will never use it.
A system or protocol is whatever the easiest user journey is. Anything outside of that will never be seen by many users unless there is some value to be gained by going there. And that value has to be something gained now, not a hypothetical like insurance against future closing of the network. People don’t like to buy insurance.
I think these are reasons that Mastodon and Nostr aren't ever going to have a critical mass of users, remaining a niche thing for people who care about the hypotheticals (which is fine). Imho, BlueSky is the only distributed social media project that has a chance of meeting users where there are with usable search, realtime discoverability, and other consequences of centralizing event-busses.
People wine about BlueSky being too centralized, but the fact is that this type of infrastructure isn't self-hostable. You can do social-media over email a la Mastodon (which admittedly is pretty great), but most people will trade that for a walled garden.
The big problem is that all this AT infra is pretty much charity, which doesn't feel sustainable. I wish it could be funded more like public libraries than ad tech.
By verdverm 2026-02-2121:14 For some context
25G < PLC postgres < 100G, depending if you want to keep all the spam operations (> 50%) and/or add extra indexes for a handle autocomplete service (like me, takes it over 100GB with everything)
Repo data (records) is in the double digit TB range (low end, without any indexing, just raw)
Blobs are in the Petabyte range.
I aim to find out current and accurate details soon.
By snapetom 2026-02-220:28 I agree 100%
Bluesky works because people are told "Go to Bluesky" and they hide the federation. When you're told go to Mastodon and pick mastodon.social or any of the hundreds of other servers, you've lost. For some reason, the federation fans never understood this. I remember an interview with Diaspora's developers and they couldn't stop talking about how people can run their own servers.
Dude.
I have two friends who left Twitter for Bluesky. One's an HR rep and the other is a business analyst for warehouses. Does anyone think a selling point for them was that they can run their own Bluesky infrastructure?
By atherton94027 2026-02-2120:382 reply I mean it's a repo with 1 very active contributor (https://github.com/blacksky-algorithms/rsky/graphs/contribut...), I get that they decided to skip on that
By runako 2026-02-225:37 (There are multiple repositories owned by that organization, reachable by one click from OP's link.)
By atherton94027 2026-02-2121:071 reply Sorry I'm not sure I understand your point
By carb 2026-02-2122:35 Sorry, meant say that Blacksky is much more important than the metrics you point to, with more detail on that wiki.
They're the first alternative full stack, the first alternative AppView, and that is something that the author should have mentioned. However, it weakens the argument so they left it out.
"Number of contributors" has never meant impact. You wouldn't dismiss openssl or curl, ya know?
By knowtheory 2026-02-2123:562 reply Anybody can sign up for Blacksky.
By snigsnog 2026-02-224:28 [dead]
By RandomNickname 2026-02-220:102 reply [flagged]
By blackqueeriroh 2026-02-235:29 gasp Afrofuturism! How dangerous. Black people imagining the future, what a scary scary thing!
By snigsnog 2026-02-230:02 [dead]
