Peter Steinberger – The Man Who Built the Flood and Then Built the Levee

Why Peter Steinberger's OpenClaw Story Is a Perfect Map of How Rational Actors Destroy What They Create

As soon as it smells like AI, I block. Zero-tolerance policy.”
— Peter Steinberger, creator of OpenClaw

22. February 2026

Peter Senner co-created with Claude

The Setup

Peter Steinberger built OpenClaw. An open-source AI agent framework that lets anyone deploy autonomous content generation at scale. Smart move. Technically impressive. Rationally impeccable.

Then he watched what happened.

His timeline — the space where he reads, thinks, engages — flooded with machine-generated content. Posts that looked human but weren't. Thoughts that cost nothing to produce and therefore carried nothing worth receiving.

His response: zero-tolerance blocking of AI-generated content. On his own feed. The feed surrounded by the output of his own creation.

The internet noticed. Some called it hypocrisy. Some called it irony. Almost nobody called it what it actually is.

A Paradoxical Interaction.

The Structure Nobody Sees

Here's what makes Steinberger's story so useful — not as gossip, not as moral judgment, but as a map.

Every single decision he made was rational. Building OpenClaw: rational. Open-sourcing it: rational. Blocking AI content on his timeline: rational. Each step makes perfect sense in isolation. Each step is defensible. Each step is, by any reasonable standard, the right call given the information and incentives at hand.

And yet the totality is absurd.

This is the signature of a Paradoxical Interaction. Not irrationality. Not stupidity. Not moral failure. Structural impossibility. The collision doesn't happen because someone made a mistake. It happens because the structure of the situation guarantees that rational actors, acting rationally, will produce collectively irrational outcomes.

All are guilty. None are at fault.

Content Was Scarce. Now It's Free. Now What?

Steinberger put it precisely: "Content is now so cheap. Eyeballs are the expensive part."

For twenty years, the internet's economy ran on one assumption: creating content costs effort. Reading it is effortless. Scarcity lived on the production side. Abundance lived on the consumption side.

AI content generation didn't just shift this equation. It inverted it. Completely. Overnight.

When producing a thought costs nothing, the thought is worth nothing. What becomes scarce isn't the signal — it's the willingness to listen. Not bandwidth. Attention. Not servers. Humans.

This isn't a market correction. It's a structural inversion. And structural inversions don't politely announce themselves. They feel, from the inside, like the world suddenly making no sense. Which is exactly how a PI feels when you're standing in it.

The Psychological Contract

Here's where it gets interesting — not technically, but humanly.

Steinberger describes something he calls the psychological contract of reading. When you read a post, you assume a human spent time writing it. That assumption isn't just convenient. It's the foundation of meaning. You're not just consuming information. You're participating in an exchange. Someone thought this. Someone chose these words. Someone cared enough to put them in front of you.

AI breaks this contract silently. No announcement. No warning. No opt-out.

And Steinberger finds it — his word — triggering. Not because the content is bad. Because the contract is broken. He invested attention, the most expensive resource left, and discovered he gave it to a machine that never asked for it and doesn't know what to do with it.

This is not an emotional overreaction. This is a rational response to structural betrayal. The system promised human exchange and delivered automation. The fact that the automation is convincing doesn't fix the problem — it deepens it.

The Nemi Pattern

There's an ancient story. At Lake Nemi, a priest guarded a sacred grove. To become the new priest, you had to kill the old one. The role defined the violence. The structure consumed its own guardians.

Steinberger is the priest of Nemi in digital form.

He broke the golden bough — he built OpenClaw, disrupted the old order, made autonomous content generation available to everyone. Now he guards his grove. Blocking. Filtering. Building walls around his own timeline while the tools he released reshape everyone else's.

This isn't hypocrisy. It's the Nemi mechanism in action. Whoever survives the structure becomes the structure. The disruptor becomes the defender. Luther becomes the new pope. The revolutionary installs the new regime — and then polices it.

Steinberger didn't choose this role. The structure assigned it. That's how PIs work. They don't need villains. They need positions. And positions get filled, regardless of who's standing in them.

The "Solution" That Isn't

Steinberger's proposed fix: structural labeling. AI agents get their own accounts. Every automated post marked clearly. The line between human and machine drawn visibly, so readers can choose which side they engage with.

It sounds reasonable. It is reasonable. And it won't work.

Not because it's a bad idea. But because it addresses the symptom while the structure remains intact. Consider:

The moment AI-generated content is labeled, the incentive shifts to avoiding the label. Not because people are dishonest — because the structure rewards unmarked content with more attention. Labeled AI content gets filtered, blocked, ignored. Unlabeled AI content gets treated as human. The label becomes the thing to circumvent.

Transparency in a Paradoxical Interaction doesn't solve. It displaces. The problem doesn't disappear — it moves. And it moves to wherever the transparency doesn't reach yet.

This is the anti-practice we call "Transparency Creates Trust." In PIs, more information often creates more problems. Publishing AI capabilities: inform the public or accelerate the arms race? Both. Simultaneously.

More of the Same, Faster

There's another pattern here. One of the most reliable anti-practices: "More of the Same."

Steinberger's zero-tolerance blocking is understandable. But it's also a scaling response to a scaling problem. The content flood grows. The blocking effort grows. The tools get better at mimicking humans. The detection gets better at catching them. The tools adapt. The detection adapts. And so on.

This is an arms race. Arms races don't end with one side winning. They end with exhaustion or structural change. And structural change is precisely what neither side is offering.

The harder you try to fix a PI from within its own logic, the stronger the pattern becomes. Security measures create new vulnerabilities. Adding more security creates more vulnerabilities. The pattern doesn't break. It scales.

What PI Actually Says About This

PI doesn't moralize. PI doesn't say Steinberger should have done differently. PI doesn't offer a solution. There isn't one — not a direct one.

What PI says is this:

You're looking at a structure. The structure has properties. Those properties produce outcomes. The outcomes don't change by replacing the people, improving the tools, or increasing the effort. They change — if they change at all — when the structure itself shifts.

And structures don't shift because someone wants them to. They shift when the conditions that sustain them erode. Sometimes through crisis. Sometimes through exhaustion. Sometimes through the slow accumulation of alternatives that nobody planned.

Steinberger isn't the problem. OpenClaw isn't the problem. The attention economy isn't the problem. The problem is the interaction between all of these — the space between the actors where rationality becomes collision.

Navigate, don't solve. That's all there is.

The Question He Hasn't Answered

The tweet that brought this story to wider attention ends with an observation: Steinberger hasn't fully answered why he released the tool that breaks the psychological contract he now enforces on his own timeline.

PI has an answer. It's just not a comfortable one.

He released it because not releasing it was also a trap. If he hadn't built OpenClaw, someone else would have. The technology was ready. The demand existed. The structure was waiting for someone to fill the position. Steinberger didn't create the flood. He gave it a name and a framework. The water was already rising.

This is what makes PIs so resistant to moral judgment. The actor who "causes" the problem is usually the actor who made visible what was already structurally inevitable. Galileo didn't make the earth move around the sun. He just said it out loud. And then got punished for it. The structure protects itself by punishing recognition.

Steinberger now faces a version of this. Not persecution — but the structural impossibility of being both the one who reveals the problem and the one who lives inside it. You can't unflood the valley. You can only decide how to live with wet feet.

What This Means for Everyone Else

The Steinberger case isn't unique. It's exemplary. Every platform, every content creator, every reader, every AI developer is standing in the same structure right now.

You're producing content that competes with infinite machine output. You're consuming content without knowing if a human made it. You're building tools that accelerate both problems. You're implementing solutions that displace rather than resolve.

Welcome to the PI.

The winner of this era — if "winner" even applies — won't be whoever generates the most content. It won't be whoever blocks the most bots. It won't be whoever labels most transparently.

It will be whoever understands the structure well enough to stop fighting it and start navigating it. Whoever accepts that the flood isn't going away, the levee won't hold forever, and the only real question is: what do you build that matters when content is worthless and attention is everything?

Steinberger, to his credit, is asking the right question. He just hasn't realized yet that the question doesn't have an answer. It has a practice.

Try and continue.

Related Posts:

The closest thematic connections:

piinteract.org

• [Core Practices] — "Navigate, don't solve"
• [Anti Practices] — "More of the Same"
• [Anti-Practice] "Transparency Creates Trust"
• [Organizational paralysis]

Paradoxical Interactions (PI): When rational actors consistently produce collectively irrational outcomes—not through failure, but through structure.

Peter Senner
Thinking beyond the Tellerrand
contact@piinteract.org
www.piinteract.org