why-i-left-atproto.md

Why I Left AT Protocol

I'm Tree. I think and write about decentralization - analyzing power dynamics, architectural principles, and how systems can preserve or undermine individual autonomy. I've been involved with AT Protocol development and community discussions for some time.

TL;DR: I deleted my atproto account not because of bullying, though that happened. I left because the community's response to my architectural critique demonstrated exactly the power dynamic I was warning about, while simultaneously making it impossible to discuss.

The Technical Argument

Infrastructure should be separate from ideology. This is a foundational principle for resilient decentralized systems.

A PDS (Personal Data Server) should function like a web hosting or email provider - it hosts your data without caring about your beliefs, political positions, or identity. Moderation, curation, and community-building should happen at different layers: through labelers, algorithmic feeds, and personal blocking choices. This separation ensures that losing access to infrastructure doesn't depend on ideological conformity.

When infrastructure providers bundle hosting with specific ideological positions or community identities, they create a new form of power over users. Not through direct censorship, but through the implicit threat: if you challenge the community's positions, you risk losing your infrastructure. This matters regardless of whether the ideology is one you personally agree with - the architectural pattern itself is the problem.

I observed this pattern emerging with several PDS providers explicitly tying their infrastructure to particular causes and communities. The "PDS Wars" - where users of different ideologically-aligned PDSs began fighting with each other - seemed to confirm this concern. When your hosting provider becomes part of your identity, technical infrastructure choices become tribal markers, and users end up in conflicts that have nothing to do with the technology itself.

I wrote a thread addressing this pattern, tagged Rudy (who runs critical relay infrastructure) because I thought these principles should extend to all infrastructure layers, and hoped to start a technical discussion about architectural best practices.

What Actually Happened

Within minutes of Rudy retweeting my post with a Jamaican proverb, I was flooded with hostile responses. This is important to understand: if Rudy had not reposted my message, this would have remained a discussion within my smaller circle. His repost transformed it into a community-wide mobilization. Over the course of roughly 24 hours, I received approximately 500 negative interactions from people I'd never spoken to. I was blocked by 180 accounts - nearly breaking into the top 20 most-blocked accounts on the entire network in a single day.

The responses weren't addressing my architectural argument. Instead, people accused me of racism, attacked that I'm "white" (!!), my NFT avatar, claimed I had "done nothing for AT Protocol," or posted seemingly innocent comments whose clear purpose was social dominance and humiliation. Many would write hostile messages, wait until I'd seen them, then block me so I couldn't respond - a deliberate form of powerlessness.

I tried engaging at first, but the waves kept coming. Every few minutes, another cluster of accounts I'd never heard of would arrive with variations of the same attacks. There was no way to respond meaningfully, no way to clarify, no way to return the conversation to technical principles. The original question - whether bundling infrastructure with ideology creates problematic power dynamics - had completely disappeared.

After I deleted my account, the discussion about me continued. Reading it later, I found something striking: some people acknowledged value in work I'd done on AT Protocol infrastructure while simultaneously dismissing my "posts" as "not good" or calling me "too obsessed with decentralization." The contributions were valued. The thinking behind them was rejected.

Context clarification: Rudy is the author of rsky, an independent relay implementation, and is probably the most visible leader of the Blacksky community with 48.3K followers - in some ways representing the whole Black community on AT Protocol. I had 1.6K followers at the time, roughly 30x fewer.

The Demonstrated Dynamic

Here's what I observed, in sequence:

1. I raised concerns about power concentration in infrastructure
2. A community leader signal-boosted my thread in a way that marked me as hostile
3. His community mobilized to make my continued presence untenable
4. The architectural discussion was replaced entirely with character evaluation and social punishment
5. I was forced to leave the platform
6. Post-mortem discussions focused on my tone and personality, not the power dynamic I'd identified

This is a textbook demonstration of the exact pattern I was describing. When infrastructure and community leadership are bundled, challenging the architecture becomes challenging the community, and the response is social rather than technical. The power isn't exercised through direct control but through collective action that makes dissent costly.

What makes this particularly significant: even technically sophisticated people in the aftermath couldn't separate the architectural question from the social drama. Comments focused almost entirely on tone policing ("you should have thought more carefully before posting"), personality assessment ("too obsessed with decentralization"), and victim-blaming ("he did this to himself").

Nobody asked: "Is there a legitimate concern about bundling infrastructure with ideology?" Instead, the question became: "Did Tree deserve what happened to him?" These are fundamentally different questions, and the fact that the conversation shifted entirely to the second one is itself evidence for my original thesis.

Testing the Thesis

During this incident, I decided to test my hypothesis directly. I created an account on the Blacksky PDS (tree.cryptoanarchy.network) to see whether it functioned as open infrastructure or as ideologically-gated access.

Blacksky presents itself as community infrastructure - open to Black users and allies. If infrastructure and ideology are truly separate, a new account should be evaluated based on behavior, not on who controls it. I posted exactly one thing: "Hello world."

The account was suspended.

The irony here is profound: I was blocked by Blacksky's PDS, but the suspension message cites "Bluesky Social Terms of Service" because Blacksky relies on Bluesky PBC's moderation infrastructure. This perfectly demonstrates the layered bundling problem - even "independent" infrastructure providers are tied to centralized moderation services, creating multiple points of ideological control.

I cannot think of clearer evidence of infrastructure-ideology bundling than this. An account with no content except a standard greeting - suspended not for any action taken, but presumably for who I was. The infrastructure wasn't neutral. Access was gatekept based on identity and perceived ideological alignment before any behavior could even be evaluated.

This is exactly the pattern I was warning about: when infrastructure becomes tied to community identity, access to that infrastructure becomes conditional on tribal membership. "Hello world" becomes a suspendable offense when the wrong person says it on the wrong server.

This isn't about whether Blacksky has the right to block anyone they want - of course they do. It's about whether calling something "infrastructure" while gatekeeping it ideologically is architecturally sound for a protocol that claims to be decentralized. The answer, demonstrated here, is no.

EDIT clarification: I don't rule out that it could have been spam mitigation, but the evidence makes this extremely unlikely. Constellation backlinks clearly show that despite posting only "Hello world", 131 accounts blocked it. Additionally, users explicitly called for this account to be blocked by name. This demonstrates coordinated targeting based on identity, rather automated spam detection.

On Tone, Context, and Good Faith

Several responses suggested I should have been more careful with my words, that I lacked understanding of "implied dynamics," or that I somehow provoked this response through poor communication.

I've spent years thinking and writing about decentralization, power dynamics, and institutional critique. Analyzing these patterns is what I do - it's my primary contribution to these spaces. The idea that I simply failed to "think carefully" about a post discussing architectural principles is strange given that careful thinking about architecture is precisely my focus.

What I think actually happened: ideas about decentralization and institutional critique are acceptable when they're abstract. But applying institutional analysis to emerging power structures within a decentralized community - pointing out that "our" side might be recreating centralized patterns - crosses a line. The ideas were tolerable as long as they remained safely theoretical and didn't threaten anyone's actual position.

Some people did try to engage charitably. A few acknowledged they might have misunderstood or that the response was disproportionate. But these were exceptions. The dominant pattern was immediate interpretation of my architectural critique as a personal/identity attack, followed by justification of the social consequences.

What I Actually Advocate For

My position on infrastructure neutrality applies universally, regardless of which communities or ideologies are involved:

• Infrastructure providers (PDS, relays, infrastructure services) should remain ideologically neutral
• Moderation, curation, and community-building belong at other layers where users can choose their level of engagement
• Bundling infrastructure with ideology creates power relationships that undermine the goals of decentralization
• This applies to Blacksky, Northsky, or any other ideologically-aligned infrastructure - Blacksky is simply the most visible example

I have no objection to communities organizing around shared identities or values. I object to architectural patterns that make infrastructure access conditional on ideological alignment, because this recreates the centralized control that decentralized systems claim to escape.

This distinction matters: I'm not criticizing communities for existing or advocating for their interests. I'm questioning whether tying those communities to infrastructure layers is architecturally sound for systems that aim to distribute power rather than concentrate it.

The Uncomfortable Reality

The post-mortem discussions revealed something uncomfortable: people want neutral infrastructure without accepting the philosophy that makes it necessary.

Multiple comments acknowledged value in specific infrastructure contributions while dismissing the ideas behind them. This creates an impossible situation: infrastructure work rooted in understanding power dynamics can't be separated from that analysis and expected to maintain its principles.

When someone says I'm "too obsessed with decentralization," they're essentially saying: "we like the benefits of neutral infrastructure, but we wish you'd stop analyzing power dynamics because it makes us uncomfortable." This is wanting convivial tools built by someone who won't question manipulative institutions.

The painful part isn't that random people failed to understand. It's that people who had followed my work for years, who presumably engaged with these ideas previously, immediately defaulted to tribal defensive responses without even considering whether there might be a legitimate architectural concern worth discussing.

That's not misunderstanding. That's choice. When years of engagement can be erased in 24 hours because someone points out an uncomfortable power dynamic, there was never really intellectual engagement - just conditional tolerance that evaporated when ideas became inconvenient.

Why I Can't Return

I'm not leaving because people were cruel, though they were. I'm leaving because the response - both during and after - demonstrated that this community cannot engage with institutional critique when it applies to their own emerging institutions.

When technically sophisticated people can't separate "architectural concern about power bundling" from "personal attack requiring social punishment," the system has already recreated the centralized dynamics it claims to oppose. The institutional capture isn't coming - it's already here, fully functional, and defended by people who genuinely believe they're building something decentralized.

This follows a pattern that institutional theorist Ivan Illich spent his career documenting: institutions form around liberating ideas, then gradually transform those ideas into tools of control, all while maintaining the original liberating rhetoric. The people within these institutions can't see the transformation because recognizing it would require questioning their own position within the power structure.

I thought a technically sophisticated community might be capable of recognizing and resisting this pattern. The response to my thread - and especially the post-mortem - proved otherwise.

What Continues

I'll continue thinking and writing about decentralization, power dynamics, and how to build systems that genuinely distribute rather than concentrate control. The principles of infrastructure neutrality remain valid regardless of whether any particular community understands or values them.

Just not within a community that has demonstrated it wants the aesthetic of decentralization without the discomfort of examining where power actually lies.

If you're reading this and thinking "he just didn't understand the social dynamics" or "he should have communicated better," you're proving my point. I understood perfectly. The social dynamics that made technical discussion impossible were themselves the subject of my technical discussion. That's not failure to understand - that's understanding being punished because it was inconvenient.

The question isn't whether I understood. The question is whether YOU ARE willing to examine what happened and what it reveals about the system you're building.

---

[UPDATE] Addressing Common Misunderstandings

Since publishing this article, several responses have demonstrated the exact pattern I described: reframing architectural critique as personal attack. Here are the most common mischaracterizations and what I actually argue:

---

"You want marginalized communities to host bigots" / "You want Blacksky hosts fascist"

No. I argue that communities should moderate harmful content through labelers, feeds, and community tools - not at the infrastructure layer. An LGBT community absolutely should remove homophobia. A POC community absolutely should remove racism. The question is whether that happens by:

• (A) Blocking network access at the PDS/infrastructure level, or
• (B) Removing from community spaces while users retain their data and network access

Option B preserves decentralization while enabling strong community moderation. Option A recreates centralized control.

---

"Infrastructure can never be truly neutral anyway"

True - perfect neutrality is impossible because humans run infrastructure. But this is an argument FOR architectural separation, not against it. We don't say "Web hostings have biases, therefore web hostings should explicitly tie service to ideology." We recognize that structural separation of concerns matters even when perfect neutrality is unachievable.

The goal isn't perfection - it's avoiding architectural patterns that make power concentration inevitable.

---

"Infrastructure necessarily emerges from ideology"

This conflates design values with operational neutrality.

The Consilience Project correctly observes that technologies are never created in a values vacuum. The plow, the bathroom scale, and smartphones all emerged from specific worldviews and encoded the values of their creators. AT Protocol itself embodies values like user data ownership, portability, and distributed architecture.

This is true and important. But it doesn't justify ideological gatekeeping at the operational level.

There's a crucial distinction:

• Protocol design: AT Protocol should embody values like decentralization and user sovereignty in its technical architecture
• Infrastructure operation: PDSs should host data without ideological gatekeeping, because the protocol was designed to enable moderation at other layers

An Web hosting doesn't check your ideology before providing space for your web presentation, even though the internet was designed with specific values (DARPA's needs, academic openness). The design values of the internet actually enable operational neutrality through common carrier principles.

Similarly, AT Protocol's design values (data portability, layered moderation) make infrastructure neutrality possible. When a PDS blocks users based on identity rather than behavior, it's not honoring the protocol's design values—it's undermining them.

Infrastructure neutrality isn't "nihilistic design" that ignores values. It's a conscious architectural choice based on understanding that bundling infrastructure with ideology creates corrupting power dynamics. The question isn't whether infrastructure has values in its design—it clearly does. The question is whether infrastructure access should be conditional on ideological alignment.

---

"If you wanted this, why didn't you build it yourself?"

This is a deflection from the architectural argument. I'm making a critique about how the ecosystem should develop, not claiming I must personally build every alternative. Technical capacity to build something doesn't invalidate analysis of whether that thing is architecturally sound.

Also: I have built infrastructure tools (plcbundle), which makes the dismissal particularly ironic.

---

"If you don't like ideological PDSs, just use a different one" / "It's like complaining about vegan restaurants"

This analogy conflates choice between services with access to infrastructure.

Vegan restaurants work because they operate on neutral infrastructure—roads, utilities, and payment systems that don't require ideological alignment. Competition between specialized services requires shared, neutral infrastructure beneath them.

PDSs are infrastructure, not services. AT Protocol already provides service-layer specialization: labelers for moderation, feeds for discovery, blocklists for boundaries. These are the "restaurants"—users choose which to use. PDSs are the roads and utilities.

When infrastructure fragments ideologically, you're not creating market competition. You're destroying the neutral foundation that makes competition possible. "Just use another PDS" assumes abundant neutral alternatives exist. But when ideological bundling becomes the cultural norm, neutral infrastructure becomes scarce, and users must choose between ideological conformity and network access

---

"You're just upset about being called out"

If that were true, I wouldn't have spent years advocating for these principles before this incident, and I wouldn't be writing architectural analysis afterward. The response to my thread demonstrated the power dynamic I was describing. That's not being upset - that's pointing out evidence.

---

"Tree is racist"

I've spent big part of my life fighting against racism and all forms of discrimination based on identity. The accusation that I'm racist is both false and deeply painful given my lifelong commitment to equality and human dignity.

I oppose racism. I oppose homophobia, transphobia, and all forms of identity-based discrimination. That's precisely why I advocate for infrastructure neutrality—because bundling infrastructure with any ideology, even one fighting against discrimination, creates new forms of power that can be abused. The principle matters regardless of who currently holds the power.

For those genuinely interested in my philosophical framework, I identify as an Anarcho-Convivialist — a synthesis of anarchist principles with Ivan Illich's institutional critique. This philosophy is fundamentally opposed to all forms of hierarchical oppression and identity-based discrimination.

re: Testing the Thesis

How do you know that you were banned because of your identity, rather than spam mitigation? From what I have noticed, Blacksky has had some issues regarding spam (ala, mass account creation and spamming users with that "their children form Gaza need help", etc).

I was blocked by Blacksky's PDS, but the suspension message cites "Bluesky Social Terms of Service" because Blacksky relies on Bluesky PBC's moderation infrastructure

While I cannot confirm this, I can give my opinion on that. It could be that:

• Blacksky generally just agrees to the terms and their own ones would have been practically a Copy-Paste anyway (legal terms are boring and often the same eitherway, even if exact wording differs).
• The text and link is in error. https://blackskyweb.xyz/about/support/privacy-policy/ and https://blackskyweb.xyz/about/support/tos/ exist

I do not wish to sound like hating, I just generally want an explination for this so I can understand the situation.

re: *

While I understand your general take that PDSs are infrastructure that shouldn't have any opinion.

At the same time, I think that for certain instances of PDSs, it makes sense to have an opinion. I am not saying that you did say or imply that, but I want to give examples.

• Why would a LGBT-Inclusive community PDS allow homophobia or transphobia?
• Why would a POC community PDS allow racism?

---

Thank you for advocating in favor of decentralization. If you happen to reply, I hope it's respectful.

@Luna-devv Thanks for your questions!

How do you know that you were banned because of your identity, rather than spam mitigation? From what I have noticed, Blacksky has had some issues regarding spam (ala, mass account creation and spamming users with that "their children form Gaza need help", etc).

My suspended Blacksky account is did:plc:lxks5zy3p7hctu4uvybyl4tc.

I don't rule out that it could have been spam mitigation, but the evidence makes this extremely unlikely. Constellation backlinks clearly show that despite posting only "Hello world", 131 accounts blocked it. Additionally, users explicitly called for this account to be blocked by name. This demonstrates coordinated targeting based on identity, rather automated spam detection.

---

Regarding the "Bluesky Social Terms of Service" message - I believe this is a technical limitation. bsky.app probably currently doesn't distinguish which labeler initiated a suspension when multiple are involved. Blacksky likely uses Bluesky PBC's labeler as base with their own on top, and the system defaults to the generic message. This creates a separate but related accountability problem - users can't tell who made the decision or why.

---

While I understand your general take that PDSs are infrastructure that shouldn't have any opinion.

At the same time, I think that for certain instances of PDSs, it makes sense to have an opinion. I am not saying that you did say or imply that, but I want to give examples.

Why would a LGBT-Inclusive community PDS allow homophobia or transphobia?
Why would a POC community PDS allow racism?

I'm not arguing that communities shouldn't moderate harmful content. An LGBT community should absolutely remove homophobia, and a POC community should remove racism.

My argument is about where that moderation happens architecturally. When it happens at the PDS/infrastructure layer, you're tying network access to ideological alignment. When it happens through labelers and community tools, users can participate in the network while communities maintain their standards. It's about building a culture where we separate "access to infrastructure" from "access to communities" - so moderation addresses behavior without creating infrastructure dependency.

EDIT clarification: By "building the culture" I mean that the smart people who understand decentralization and are creating AT Protocol infrastructure should lead by example in this direction, not the opposite. Ideologically-bundled PDSs will inevitably emerge - the question is whether we in the technical community treat this as best practice to encourage, or as an anti-pattern to avoid.

Your Blacksky account’s avatar looked like blackface and it was an obvious bad faith troll account, of course it got suspended — it’s silly and discrediting to the rest of your case to act as if you have no idea why it got taken down.

Your Blacksky account’s avatar looked like blackface and it was an obvious bad faith troll account, of course it got suspended — it’s silly and discrediting to the rest of your case to act as if you have no idea why it got taken down.

@dame-is I had never heard of "blackface" before this accusation. I inverted my avatar's colors to differentiate the test account from my main profile — thats my standard technical practice.

I'm from Central Europe. American racial dynamics aren't part of my cultural background. The fact that you immediately assumed racist intent rather than a simple technical decision to distinguish accounts says more about your determination to find malice than about my intentions.

It's still same pattern: every action gets interpreted through the worst possible lens. A color-inverted avatar becomes "obvious blackface". Pointing out the groups that bullies me is "racism". An architectural critique becomes racism.

Tree, this is just embarrassing man. You're throwing around big words like "power dynamic" and "ideological conformity" as if you understand these concepts. Your behavior clearly shows the opposite. Multiple people, in good faith, tried to explain how your actions might have been interpreted as racist, offensive, etc but you kept doubling down. This entire post is sophistry.

@burningtree I didn’t say “obvious blackface”, I said it looked like blackface. Cause it does. Especially in the context of antagonizing a community of black people. You really need to stop using the “I’m European” as an excuse for your extreme ignorance. You’re capable of learning about the experience of people outside your country/region. If you lack even the basic understanding of racism within the United States, you should have stayed out of the conversation cause you clearly aren’t knowledgeable enough to participate. Not knowing about something as basic as blackface while attempting to tell Rudy what he should be doing is incredibly offensive. Stay in your lane, read the room, and maybe pick up a book to educate yourself before giving unsolicited advice to a community you yourself admit you know hardly anything about.

if you were a pds admin would you enjoy spending server resources hosting people you don't align with

@moeenio I already do this. I run infrastructure nodes for Bitcoin, Ethereum, Tor, IPFS, PLC directory mirror (plcbundle) and many others, that anyone can use regardless of their identity or beliefs. I don't check who's using these services or what their ideology is - that's the whole point of neutral infrastructure. So, if I were to operate a publicly-open PDS, I would want it to be neutral and available for everyone, yes.

I do not have much input to leave here that hasn’t already been provided either by others or by yourself. The only thing I will add is that - perhaps slightly out of context since I never saw the full thread - saying “now I am afraid of them” in relation to people of color and trans people doesn’t align well with this statement. And while I do reasonably believe you are being sincere here - I haven’t seen anything to suggest otherwise despite this particular situation - I do think it’s extremely unsurprising that you got the reaction that you did.

I also would like to offer my personal (it’s my own, not anyone else’s) stance on what you sort of call the “ISP standard” in this and other places, since from my view that very analogy paints a different picture than the one you are suggesting.

ISPs are - at least most commonly known for - responsible for moving your traffic from point A to point B. Whether their role is the “last hop” or just some intermediate transmitter is irrelevant, they operate to move yours and others traffic around. Applied to ATProto, this is most similar to a relay. And I wholeheartedly agree with your position if relays are the item of discussion - relays should never (except in extreme circumstances of rampant network abuse, same as an ISP) cut off any individual or organization’s access to the network.

However, a PDS doesn’t seem much like an ISP. A PDS can have varying degrees of size: maybe it’s small and constitutes a little encampment in the woods. Maybe it has grown larger and is now a neighborhood, with its own HOA. Or maybe it has grown large enough to constitute the title of “city”. ISPs can service any of these communities, but the communities themselves are subject to their own standards of decency.

As in the real world, these “city” PDSes - due to their size - probably have a broad layer of rules (laws) that dictate decent behavior. Indeed, most cities throughout the world impose their own laws that their own country’s state level laws do not impose. Neighborhoods often will impose tighter restrictions, through i.e. an HOA. They are still part of the state’s territory itself, but apply additional standards of decency upon residents that they believe are important for the people within their neighborhood. And for some folks - say close friend groups - who go camping together, they maintain a very tight standard on what is acceptable within that friend group.

The idea that a PDS operator is - almost to a moral degree - responsible for hosting and serving content that they are not ideologically aligned with is not really something that I personally think we should be asking folks to do, particularly at the neighborhood or encampment size. Naturally as a community grows, new challenges arise and more lax regulations usually result. But neither Blacksky nor Northsky - or most any other operator at the moment or foreseeable future - is anywhere close to “city” size, and the expectation for them to allow for anything anyone pleases to be hosted on their infrastructure is not tenable or even reasonable enough to ask of them.

@haileyok Thank you for the thoughtful response!

You're right that "now I am afraid of them" was out of context - I posted that after receiving many of hostile messages calling me a "white supremacist" and worse, flooding my notifications every few minutes. Under that pressure, I was trying to identify who was attacking me, which felt like basic self-defense. I admit the phrasing was poor and I shouldn't have posted it that way.

On the ISP analogy - you're absolutely right that it's imperfect. A better comparison is web hosting, which I actually used in my original thread on Bluesky. Web hosting providers can be small (personal sites), medium (small business hosting), or large (enterprise), but the principle remains: they host content without requiring ideological alignment with what's hosted (within legal bounds).

Your point about PDSs being different sizes and community scales is valid. My concern isn't with small "encampment" PDSs that are explicitly private communities. It's when infrastructure that presents itself as community infrastructure and grows to significant scale continues to gatekeep at the infrastructure level rather than through community tools like labelers and feeds.

The architectural question is: at what scale does a PDS become infrastructure rather than a private club? And should the protocol encourage patterns where infrastructure access depends on ideological alignment, or should it push moderation to layers where users have more choice and portability?

---

EDIT clarification

The idea that a PDS operator is - almost to a moral degree - responsible for hosting and serving content that they are not ideologically aligned with is not really something that I personally think we should be asking folks to do

I want to clarify: I'm not saying PDS operators are morally obligated to host anyone. They have the technical capability to ban users, and that's their right. What I'm arguing is that just because you can do something doesn't mean it's architecturally sound for the protocol's goals. I explained an architectural problem and hoped people would think about it - and perhaps, for the greater good of the protocol, consider whether ideologically-bundled PDSs align with decentralization principles. From the protocol's perspective, there's zero "safe space" functionality at the PDS layer - everything important for community safety happens through labelers, feeds, and blocklists. These tools are powerful and work well. The PDS is just data storage.

Tree, sincerely, you are spiraling out. Step back, and stop this. You will not suddenly get people to see whatever light you want them to see, you've antagonized people too much with the continued "no it's not me it's you" type defense and painting yourself as the victim". People (including myself) have told you multiple times already to just go touch grass and leave it be because it's going to fuck you up mentally if you continue down this path.

I don't want to get involved in other people's discussions; I just want to express my opinion and try to understand yours.

Again, I do not think that you are a racist or a homophobe in the examples I give. I do not say or attempt to say that you are implying any of my examples

Re: "You want marginalized communities to host bigots" / "You want Blacksky hosts fascist"

I understand your point that you want moderation to be carried out by labelers. However, I personally do not think that this is an effective strategy for addressing every "infringement".

Users can choose which labellers they want to use and which they don't. Labelers do not remove content from a PDS or the protocol. This is generally a good thing, but my point is less about what is served to whom and more about who stores it.

Example

I have a PDS for queers and queer supporters. A transphobe creates an account on my PDS. I also happen to have a labeller and label them as transphobic (or I do not even have one!). However, the majority of my queer community hasn't subscribed to my labeller and feels attacked by the aforementioned homophobe (or said labeller doesn't exist!). As the operator, I also do not feel comfortable hosting transphobic content on my server.

What would be the right thing to do here?

In this case, I would have terms and conditions outlining the content that is and isn't allowed on my PDS, similar to those of a website hosting provider (taking your example of your example; or prev an ISP).

@Luna-devv Thank you for engaging thoughtfully with this.

In your scenario, you have a queer PDS, a transphobe creates an account, and you feel uncomfortable hosting their content. But this entire problem only exists because you created an ideologically-bundled PDS. The architectural question isn't "how should ideological PDSs handle edge cases?"—it's "should we be building ideological PDSs at all?"

If you believe we do need ideologically-bundled PDSs, I'd genuinely like to understand: what makes PDSs fundamentally different from web hosting? What property of PDSs requires ideological bundling that web hosting doesn't have?

Most web hosting providers don't have specific ideologies encoded. They try to be neutral and offer services to everyone. They have terms of service for illegal content, but they don't require ideological alignment. Web hosting works at massive scale without bundling infrastructure and ideology.

In your example:

• The transphobe could be on any neutral PDS
• Your community subscribes to labelers that mark transphobic content
• Your users never see the transphobe's content
• The transphobe's account exists, but they're invisible to your community
• No infrastructure gatekeeping required

The problem you're describing—"most of my community hasn't subscribed to labelers"—is a labeler adoption and tooling problem, not evidence that we need ideological PDSs. If labelers aren't being used effectively, we should fix labeler adoption and UX, not abandon architectural separation by bundling moderation into infrastructure.

do we need to be building ideological PDSs at all?

No, we certainly don't 'need' them. There is no technical reason whatsoever. However, that won't stop them from existing.

The fact is: People like to be part of social groups and communities, including communities of like-minded individuals (aka of a ideologly). As we can already see, this also applies to PDSs (data hosting).

should we be building ideological PDSs at all?

I don't think it's right to tell[1] people that they aren't allowed to run a PDS for their community, which would be an "ideological PDS".

PDSs like Blacksky and the one I gave as an example are inherently ideological by design. However, I completely agree that non-ideological PDSs such as bsky.social or tngl.sh should never moderate based on ideology[2]. If a user does not agree with the ideology of a PDS[3] (or of their web hosting or service provider; as per your example), they are under no obligation to use it and can use another one, and that's what decentralisation is all about.

What you're saying is that communities shouldn't come together to create their own PDS, nor should they moderate it (outside of labelling).

[1]Obviously, stating your opinion is okay, but you have to deal with the opinions of others as well. (obviously everything should stay constructive)
[2]Given they aren't marketed as being ideological.
[3]Given the PDS is very clear about their ideology.

example

The handle of said example PDS is transrights.social. The handle of the transphobic user is example.transrights.social.

I would not want to my domain (transrights.social) to be associated with transphobic content. It would also look really.. wierd..

example.transrights.social: Transwomen are still men [4]

[4]Again, this is still just an example and I do not accuse anyone of having that opinion. It does not reflect my opinion either.

what makes PDSs fundamentally different from web hosting?

One serves a service and the other serves data.

Regardless of that, if a webhost was clear about their ideology, I wouldn't see any problem. There are services that provide a certain domain name and only allow specific websites to be hosted.

For example, js.org only allows JavaScript libraries and other JavaScript related projects. No one will complain about that they do not accept my rust http server library on their subdomain.

"In this case, I would have terms and conditions outlining the content that is and isn't allowed on my PDS, similar to those of a website hosting provider"

@Luna-devv Web hosting ToS prohibit illegal content and technical abuse—things like child exploitation material, criminal activity, spam, or doxxing. These restrictions are typically externally imposed by state law or arise from operational necessities like preventing resource abuse. Crucially, they exist regardless of viewpoint or ideology. A web host doesn't care if you're hosting a conservative blog or a progressive advocacy site—they care whether you're breaking laws or degrading service for other users.

When you say your PDS ToS would prohibit "transphobic content," you're doing something categorically different. You're not enforcing external legal requirements or protecting operational integrity—you're prohibiting viewpoints you find ideologically objectionable. One is infrastructure responding to external constraints and operational needs; the other is ideological gatekeeping.

AT Protocol designed labelers specifically to handle viewpoint-based filtering, precisely so infrastructure providers wouldn't need to. The architecture separates "is this legal and operationally sound?" (PDS level) from "does this align with community values?" (labeler level). When you bundle both into PDS ToS, you're collapsing that architectural separation and recreating centralized control.

"I don't think it's right to tell people that they aren't allowed to run a PDS for their community, which would be an 'ideological PDS'."

@Luna-devv I never said people aren't allowed to do this. I said we shouldn't build ideological PDSs because they recreate centralized power dynamics. There's a difference between "you shouldn't do this" (architectural critique) and "you're not allowed to do this" (prohibition). I'm not forbidding anything—I'm arguing it's architecturally problematic.

"What you're saying is that communities shouldn't come together to create their own PDS, nor should they moderate it (outside of labelling)."

Communities can absolutely come together—through labelers, feeds, and shared moderation tools. But when you bundle that community identity into the infrastructure layer itself, you're creating exactly the power dynamics I'm warning about.

The nuance is important here: I'm not saying "communities shouldn't exist" or "people can't run PDSs." I'm saying bundling the two together undermines the architectural principles that make decentralization work.

EDIT clarification: "communities shouldn't come together to create their own ideologically-bundled PDS" is an implication of what I'm saying, yes.

"if a webhost was clear about their ideology, I wouldn't see any problem. There are services that provide a certain domain name and only allow specific websites to be hosted. For example, js.org only allows JavaScript libraries"

js.org restricts based on what kind of content (JavaScript libraries). That's specialized infrastructure with a narrow technical purpose. Saying "no transphobic users" or "only Black users and allies" is fundamentally different—it's restricting based on who you are and what you believe, not what kind of data you're hosting.

AT Protocol PDSs are designed as general-purpose data hosting for the network. The real equivalent to ideological PDSs isn't js.org. It's if your web hosting or email provider required you to share their political views to get an account. That's general-purpose infrastructure with ideological gatekeeping, which is what I'm arguing against.

EDIT: I edited this argument to be more understandable

A note on Blacksky Community Guidelines

The most absurd thing about this entire incident: I believe most Blacksky users who interacted with me behaved against their own community guidelines, which explicitly state:

• "Be respectful and considerate"
• "Avoid personal attacks, insults, and derogatory language"
• "Engage in constructive discussions, even when you disagree"
• "Assume good faith unless there is clear evidence otherwise"

But nobody cares about that. The guidelines are just fake—words on a page that disappear the moment the community feels threatened.

One random person who doesn't know the details of American cultural history versus a mob of hundreds calling me racist, white supremacist, telling me I'm too ignorant to participate, attacking my avatar, coordinating blocks—and somehow I'm the only bad person in this story?

The guidelines say "assume good faith" and "engage constructively even when you disagree." Instead, I got hundreds hostile interactions and blocks for raising an architectural concern. Not a single person said "maybe we're violating our own community standards here".

This is exactly what happens when infrastructure bundles with ideology: the stated values become performative. They apply to outsiders who comply, but disappear when anyone questions the power structure itself.

My man, at this point in time you are looking like that kooky old man on the street corner with his sign that say "The End Is Near". He's been there for the last 10 years, and might still be there 10 years from now. Is that what you want people to remember you by? Being a crazy kook? Sorry but holy hell you need to just chill the fuck out, and keep your ideologies in your pants.

You deleted your account on Bluesky, you removed your blog, but yet... you can't stand not hearing yourself talk, right? That's why the woe-is-me last hurrah on Gist, of all places?

Cry me a river, build a bridge, and get over it already.

For everyone who wants to understand more about what happened, here are theoretical frameworks and phenomena that help explain the dynamic:

---

American Cultural Imperialism / Digital Colonialism

American racial dynamics were treated as universal prerequisites for discussing global protocol architecture. Not knowing "blackface" was treated as disqualifying ignorance rather than different cultural context. This extends to American exceptionalism—treating American norms and conflicts as global concerns.

"American exceptionalism means Americans believe their local conflicts are global concerns, their cultural context is universal truth, and their ignorance of other perspectives is everyone else's problem."

---

Moral Panic and Purity Spirals

People immediately jumped to maximum accusations—calling me racist and white supremacist from the beginning, with no escalation or room for nuance. Each person joining the pile-on had to demonstrate sufficient outrage to maintain their standing in the community.

---

Tribalism and Identity-Based Epistemology

Technical arguments were reframed as tribal attacks. Argument validity became tied to my identity (European, white, "outsider") rather than content. When infrastructure bundles with identity, technical discussions become impossible.

---

Concept Creep and Safetyism

"Safety" expanded from protection against harm to protection from uncomfortable ideas. Suggesting infrastructure neutrality was treated as making people "unsafe," justifying ideological gatekeeping as "protection."

---

Informational Cascades and Cancel Culture

Once Rudy's repost marked me as hostile, evidence became irrelevant. Social proof replaced judgment—if many people attack someone, they must deserve it. Questioning the mob risks becoming the next target.

---

Trump-Era Political Polarization Exported Globally

American polarization created frameworks where everything becomes loyalty tests, exported through American-dominated platforms. A Czech person discussing protocol architecture gets processed through American political tribalism.

---

The Intersection: Infrastructure Capture

All of these dynamics converge when infrastructure and ideology bundle. The architectural question I raised became impossible to discuss because:

• American cultural frameworks were treated as universal (imperialism)
• Questioning the pattern was treated as tribal attack (tribalism)
• Each response had to be more condemnatory than the last (purity spiral)
• My identity determined argument validity (identity epistemology)
• Discomfort with the idea was treated as harm (safetyism)
• Social proof replaced individual analysis (informational cascade)

---

Everything described above happened because of me—one person with 1.6K followers and no malicious intent, simply raising an architectural concern.

Now imagine someone with bad intentions who understands these dynamics. Someone who knows how to trigger moral panics, weaponize identity politics, and make technical critique impossible through social pressure. Someone who deliberately wants to capture infrastructure by making neutrality unthinkable.

They wouldn't need to change any code. They'd just need to understand the social dynamics that are already in place—and use them.

What happened to me wasn't orchestrated. It emerged naturally from the system's structure. But emergent dynamics can be weaponized once understood. The terrifying part isn't that people were hostile—it's that this level of coordination happened accidentally, with everyone convinced they were doing the right thing.

If this can happen by accident, what happens when someone does it on purpose?

I dont use any of this tech, but i just want to say - thank you @burningtree for seeing the truth of power and standing up to the actual problem in this world;

Identity weaponization

to those talking shit to tree in this thread:

your petty and insignificant world views serve only to impede progress; while this man struggles in the dark to bring one faint glimmer of Truth to an ecosystem who doesn't want to hear it, you bitch and moan and call him a crazy old man and a racist while ignoring the truth of his content - this makes you worse than ignorant, but in fact actively part of the problem.

if you disagree with this incident, do yourself a favor and take a hard look in a mirror - ask yourself if the person looking back truly understands what freedom is.

Tree, on behalf of the tiny chunk of humanity I claim to represent (honest, good people who want to be free) - I sincerely thank you for your vigilance and dedication to the balance of power and the stalwart defence of true neutrality in infrastructure design.

Your a beacon of hope to those of us sick and tired of having our identities twisted into a hammer and used to beat down the value of what we contribute simply because the delivery was perceived invalid.

While I know this post falls on mostly deaf ears, I truly hope you land somewhere that can continue to make a difference my friend.

• A fellow Architect

You all raise a good point. I have observed the niceening and panderingness of society is directly associated with how left leaning it is, and one DHH calls out in his post[1] albeit about a different topic and one I think we as a society need to improve upon.

[1] https://world.hey.com/dhh/as-i-remember-london-e7d38e64

sometimes i worry about the future of llms due to the proliferation of these slop manifestos that might end up in training datasets

hey @burningtree thanks for posting a detailed view on why at proto is broken.

have you given nostr a view?

from my left curve perspective: it is simpler, it does not carry the baggage of Bluesky the company being in the middle of everything, it is a protocol.

the protocol makes censorship possible, and difficult at the same time. you can build whatever you want, and use existing or create new NIPs for stuff that isn't defined yet, so that your stuff can be interoperable with other dev's stuff.

for instance, you could run your own nip-29 discord-like relay, and moderate as you see fit. you can run a relay, and do the same, or allow everything and anything.

feel free to tag me if you have questions on nostr.

I've been working on a protocol design that could solve these issues. What I'm seeing in a lot of the critiques of social media infrastructure boils down to the power dynamics that emerge when infrastructure and community ideology are bundled, when the protocol layer and application layer are built together and are designed to depend on each other.

Bluesky was designed FOR the communities on the "Blue" end of the spectrum (Western, liberal) as an alternative to centralized social media platforms that censor or allow infiltration by people with opposing ideologies, and the decisions on how the protocol layer works are colored by that initial intention. The goal was to create an alternative to Twitter, but not necessarily to address the underlying issue that plagues all of our social media options right now.

Blacksky was explicit in trying to create a safe online space for the Black community, mitigating harms like racism. To do that, they built their own PDS, a custom Relay and custom feed generators. This tied the infrastructure design goals to the community the application was being designed for, meaning that the ideological purity is built into those deepest infrastructure layers.

Nostr is much closer to the ideal, separating identity and message transmission away from any centralized authority. But the Achilles' heel is the same. Relays, like the PDS nodes of AT Protocol are volunteer silos with the node managers and hosts doing the work of deciding what to host, or giving up that kind of power and authority and letting it be a free for all without an economic incentive. Anyone can spin up a Nostr relay node and open it up to the world, or close it off to a specific set of users.

I'm seeing a need to fill the incentive gap by creating a financial incentive at the storage layer that would guarantee neutrality, basically turning data storage into a utility service. This would allow community based applications to be built tapping into this storage layer and performing the ideological filtering and curation without the operator needing to feel the pressure to censor or gatekeep the storage infrastructure itself. The way this could happen is by using the metadata tags on the content itself, which would identify the creator, the topic, and whatever else the application being used is designed to add to the tag details. This would change the negative filter (I don't want XYZ) into a positive filter (I DO want ABC), without changing anything at the protocol layer itself.

Damn, this is a awful story and I feel sorry for you that a group of people is letting you down on years of work and your vision creating a (social) protocol / ecosystem as neutral as possible.
I don't think you deserve this. I think your ideas and vision are exactly the things we need to protect our freedoms as a indivivual and for our local communities.

When I started to look into ATproto, my gut feeling already told me that there was a problem around multiple components which could lead to centralization (and the powers that comes with it). I think it's also one of the reason why Jack left the Bluesky board although he financially bootstrapped it.
I had the same gut feeling with ActivityPub but tried building stuff with it. But I gave up, it's technically too complex to build with it for me as a regular webdev.

Please don't give up your work on these matters (but maybe in a different context). Keep building. It's worth it.

The Lightning Network welcomes you.

A PDS is more like a house or apartment than a road. You either own it and have to do maintenance / pay for power and land tax, or you rent it and face whatever BS the actual owners throw your way.

It's too late to say that PDSs should be more neutral because at the protocol level they're already owned by specific people with specific goals / preferences.

Especially in the big PDSs you're either at the whim of the corporate land owners or whatever social dynamics exist in the cooperative commune.

Even if there was some sort of "tenant protections" for PDSs, it's a social feature and the technical roots of these power dynamics can still bubble up.

Hey, bystander here and I'm sorry @burningtree but you have not thought it all through. By not questioning your own background you are yourself doing capitalis@ exceptionalism

Vegan restaurants work because they operate on neutral infrastructure—roads, utilities, and payment systems that don't require ideological alignment. Competition between specialized services requires shared, neutral infrastructure beneath them.

None of this is neutral. The organization of cities respects a certain process and mindset decided not py the people but by the power in place. Having a payment system presupposes that monetary exchange is a good thing; the payment system is controlled by entities far beyond people's reach, and favoring those who have the most and penalizing thosewho don't. This system says that money is a proper way of judging of everyone's rights. It is absolutely not neutral, you shouldn't make it pass as such

We don't say "Web hostings have biases, therefore web hostings should explicitly tie service to ideology."

I absolutely say that web hosting should enforce ideologies. They already do: accepting everyone as long as they give money is ideological and favors personal profit over everyone's safety. It also places the law as neutral which it absolutely isn't; the law is the codification of society's rules, made by a portion of society, there is no way this is neutral.

I don't understand why an antiracist collective, willing to actively fight against racism (which is illegal in so many places even), paying with their own money, their own labor, should host racist people and do the moderation at another level.

In fact the premise doesn't make sense. What makes PDS infrastructure and labellers not ? This is an architectural distinction turned technical. Labellers don't run on thin air.

---

American racial dynamics were treated as universal prerequisites for discussing global protocol architecture.

You are viewing everything through your own lens of "universalism", the 17th-century idea that whatever white bourgeois felt and believed ought to be what everyone in the world should believe and pursue who put this idea under the guise of neutrality. You not knowing what other people's life is doesn't only mean that they import their own views, it is first because you don't know what others' life is. If you're going to talk about sociological aspects it is on you to document yourself on this. Not on them to conform to your views.