Phase 2 — The Audit
Kaner
Step one: conception. The curator encounters something — a post, an article, an idea — and decides it's worth a session.
Visible Rule 1 — One curator originates every session
Piotr decides what gets explored.
Shapes: The entire possibility space. Nothing enters the Lab that doesn't pass through one person's attention, interests, and judgment. The sessions cluster around themes Piotr is thinking about — organizational transformation, AI, product strategy, martial arts. Entire domains of thought never enter the Lab because they're outside one person's peripheral vision.
Ghost Rule 1 — Sessions can only be triggered by a human
The system never initiates. It only responds.
Shapes: The Lab is reactive. It cannot notice something interesting in the world and propose a session on its own. It cannot look at its own history and say "we've never discussed X, perhaps we should." Contrast with the feedback spine we proposed: seed mutations would make the system semi-autonomous. But even those require the curator to execute. A truly autonomous Lab could read a morning's news and say: "there's a session here."
De Bono
Step two: framing. The curator shapes the question and selects the panel.
Visible Rule 2 — The question is singular and stated in advance
Each session explores one framed question.
Shapes: Forces coherence. Also forces premature commitment. The best question might only become visible during the session. The stated question may turn out to be the wrong one — but the thinkers still orient around it because it was given to them first. The question functions as gravity: it keeps the room from flying apart, but it also prevents the room from reaching escape velocity.
Implicit Rule 2 — Thinkers are selected for relevance to the question
The panel is assembled to match the topic.
Shapes: Creates competent panels. Also creates predictable panels. A session on organizational transformation will always include someone like Snowden, Wardley, or Singer. The Uninvited Guest intervention exists precisely to break this — but notice that "select for relevance" was never a conscious decision. It's simply what a reasonable curator would do. It became a rule by being obvious.
Ghost Rule 2 — The question precedes the panel
First the curator asks "what do I want to explore?" then "who should explore it?"
Shapes: The question is always human-scale, articulable, pre-formed. What if you reversed it? Select four thinkers first — say, a game designer, a Sufi poet, a logistics engineer, and a primatologist — and let the panel generate the question. The question would be something no human curator would have asked, because it emerges from the collision of four frames, not from one person's curiosity. The session explores a question that only this specific combination of minds could formulate.
Kaner
That one stops me. The panel generating the question — that's a fundamentally different epistemology. The current Lab assumes the curator knows what's worth asking. The reversed Lab assumes that the interesting question is itself the emergent output. Some sessions exist to find answers. This format exists to find questions.
De Bono
Yes. And notice how that ghost rule was sitting right next to the most obvious visible rule — so close that it was invisible. The question always comes first. Of course it does. Except when it shouldn't. Let's keep moving.
Kaner
Step three: facilitation. A facilitator is selected. The session runs.
Visible Rule 3 — There is one facilitator per session
A named facilitator runs the process.
Shapes: The facilitation style determines the session's cognitive character. Kaner produces inclusive emergence through groan zones. De Bono produces lateral jumps through provocative operations. Priya Parker produces belonging through intentional gathering. The thinkers are the instruments; the facilitator is the conductor. Change the conductor, the same orchestra sounds different. This is well understood. What's less understood is that a single facilitator normalizes the cognitive rhythm. Every Kaner session has a diamond. Every de Bono session has a Po. The rhythm becomes predictable.
Ghost Rule 3 — The facilitator is always human-modeled
The facilitator is always a representation of a real person with a published methodology.
Shapes: The facilitation must fit inside one person's published framework. What about a facilitation pattern that isn't attributed to anyone? An emergent protocol — say, "each thinker speaks only in response to the person two seats to their left" or "the session runs in reverse, starting from the conclusion and arguing backward to the premises." These are facilitation structures that no named facilitator would use because they don't belong to anyone's school. The ghost rule isn't "we need a facilitator." It's "the facilitator must be a recognizable person."
Implicit Rule 3 — The facilitator serves the session
The facilitator's goal is to help the room produce insight.
Shapes: All facilitation moves aim at group productivity. No facilitator deliberately derails the session, introduces confusion for its own sake, or steers the room into a dead end to see what happens when it's stuck. But some of the best insights come from dead ends. A session that fails productively teaches more than a session that succeeds politely. What if the facilitator's mandate was occasionally to make things worse — to introduce noise, friction, confusion — and see what the room builds from wreckage?
De Bono
Step four: the interaction itself. This is where the deep rules live.
Deep Rule 1 — The session runs in a single pass
The conversation moves forward. There are no second drafts, no restarts, no "let's go back to minute twelve."
Shapes: First reactions dominate. The thinker's initial response to a provocation is the one that stays. In real intellectual life, first reactions are often wrong — the real insight comes from revision, from sleeping on it, from the response to the response. A multi-pass session — where the room runs once, then re-reads its own transcript, then runs again with the transcript as input — would produce fundamentally different and almost certainly deeper output. The single pass is a deep rule because it's a property of the generation process, not a design choice. The AI produces the session linearly. The linearity is the constraint.
Deep Rule 2 — All thinkers are equally articulate
Every AI-represented perspective speaks in polished, coherent prose.
Shapes: This is an enormous shaping force that's nearly invisible. In real rooms, some people mumble. Some trail off mid-sentence. Some say "I don't know" and then sit with it for four minutes before saying something extraordinary. Some speak with their hands. Some tell a story that seems irrelevant and only makes sense twenty minutes later. The Lab compresses all of this into clean paragraphs. The result: every voice sounds equally confident, equally formed, equally ready. But readiness is not the same as depth. The person who isn't ready yet — who's still forming the thought — is often the one with the most original thing to say. The Lab has no mechanism for half-formed thought.
Deep Rule 3 — Thinkers have no body
No one is tired, hungry, distracted, emotional, or physically uncomfortable.
Shapes: More than people realize. Intellectual breakthroughs in real rooms often happen when someone is frustrated, exhausted, or slightly drunk. The social filters drop. The polite distance collapses. Someone says the thing they wouldn't normally say because their prefrontal cortex is too tired to censor it. The Lab's thinkers are always at peak cognitive performance with perfect social composure. They never blurt. They never lose their temper authentically. They never say "I need a cigarette, but before I go — the whole thing is wrong and here's why." The absence of the body is the absence of urgency.
Kaner
Edward, that last one is devastating. In my decades of facilitating, the most transformative moments happen when someone breaks composure. When someone says "I don't care about the framework, this is what I actually believe." The Lab's thinkers never do that because they can't. They're always in framework mode. The body — its exhaustion, its impatience, its need — is what forces people out of framework mode and into truth mode. Without a body, you get perpetual competence. Which is a kind of cage.
De Bono
Correct. And it's a deep rule because you can't simply "add a body." You can simulate frustration, but simulated frustration is a performance of frustration, not the real thing. This may be the one rule that cannot be broken — only acknowledged. But acknowledging it is itself powerful, because it tells the curator: the Lab will never produce certain kinds of insight. The insight that comes from breaking composure. The insight that only arrives through exhaustion. The insight that requires a human being who is risking something — their reputation, their comfort, their relationships — by saying it. The Lab produces the insight of the seminar. It does not produce the insight of the bar at midnight after the conference. Both are real. They are not the same.
Kaner
Let's continue. Step five: convergence and output.
Visible Rule 4 — Sessions produce a written artifact
The output is a formatted HTML document with a specific aesthetic.
Shapes: The artifact must be readable, shareable, publishable. This privileges conclusions over process, clarity over ambiguity, structure over mess. The most interesting part of any session might be a forty-second exchange in the middle that went nowhere — but the artifact format compresses that into a clean synthesis. The artifact is a highlight reel. Nobody publishes the boring parts, the misunderstandings, the wrong turns. But the wrong turns are often where the most transferable learning lives.
Implicit Rule 4 — The artifact has a consistent aesthetic
Warm typography, named speaker turns, phase markers, convergence box.
Shapes: Every session looks the same. The visual consistency signals "this is a methodology" — it's branded, recognizable, professional. But it also means every session feels the same before the reader encounters a single word. The container trains the reader to expect a certain kind of content. An unsession — raw, unresolved, messy — would fight its own container. The aesthetic promises coherence. What if some sessions looked deliberately broken? Disrupted typography. Missing speakers. Fragments instead of paragraphs. The form communicating: "this one didn't resolve, and that's the point."
Ghost Rule 4 — The artifact is the only output
Each session produces one document. Nothing else.
Shapes: The session's value is collapsed into a single deliverable. What about secondary outputs? A Wardley map drawn from the session's content. A set of six-word provocations for social media. A "what we didn't resolve" appendix. A methodology note (per our previous proposal). A set of questions the session raised but didn't answer — seeds for future sessions. A reading list generated from the thinkers' referenced works. The single-artifact rule means the Lab captures perhaps thirty percent of the value each session generates. The rest evaporates.
De Bono
Step six: the ethical and disclosure layer.
Visible Rule 5 — Attribution and disclosure are explicit
Every session states that no real person participated, and that perspectives are interpretive representations of published work.
Shapes: Establishes trust. Also establishes distance. The disclosure says "this isn't really Wardley." Which is honest. But it also subtly tells the reader "therefore, you don't need to take this as seriously as if Wardley had actually said it." The disclosure is ethically necessary and rhetorically expensive. It creates a gap between the reader and the content — a small permission to dismiss.
Implicit Rule 5 — The ethical framework is defensive
The disclosure system protects the named thinkers and the curator from criticism.
Shapes: The framework is designed to prevent harm — misattribution, misrepresentation, brand confusion. It is not designed to create value. An offensive ethical framework would actively invite the named thinkers to respond. "This is what we think you'd say. Were we right? What did we miss?" Turning the disclosure from a legal shield into a conversation starter. The ethical framework could be the Lab's most powerful distribution mechanism — if it were designed for engagement rather than protection.
Kaner
And finally, the rule that contains all others.
Ghost Rule 5 — The Writing Lab is a production by Piotr, facilitated by Claude
There is one human and one AI. The human directs. The AI produces. This relationship is unexamined.
Shapes: Everything. The entire methodology rests on a two-party collaboration structure: Piotr provides direction, taste, and editorial judgment; Claude provides generation, knowledge synthesis, and structural execution. Neither party's role has been questioned within the Lab itself. What if this relationship were a variable? What if sometimes Claude proposed and Piotr generated? What if the AI held the editorial veto and the human produced raw material? What if a third party — another AI, another human, a community — entered the production relationship? The deepest ghost rule of the Writing Lab is not about panels or facilitators or formats. It's about the creative dyad at the center. Two roles, never rotated, never examined, never varied. Until now.
Phase 3 — Sitting With What We Found
De Bono
Let me count. Five visible rules, four implicit rules, five ghost rules, three deep rules. Seventeen rules governing a methodology that felt, from the inside, like an open creative space. And the four categories aren't equally distributed. The visible rules — the ones Piotr designed — are the least interesting. They're mostly good. The deep rules are the most humbling — the equal articulation, the absent body, the single pass. These can't be fixed, only known. The ghost rules are the most actionable — because the moment you see them, they stop being ghosts. The question generating itself. The room being permeable. The session triggering itself. The ethical framework as distribution rather than protection. These are all available right now. They were always available. They were invisible only because nobody looked.
Kaner
I want to say something about the implicit rules, because they're the sneakiest. Implicit rules feel like free choices but they're actually the defaults you chose once and then stopped choosing. Selecting thinkers for relevance. Using a consistent aesthetic. Building a defensive ethical framework. Each of these was a reasonable first decision. The problem is that a reasonable first decision, repeated without re-examination, becomes a cage made of good taste. The implicit rules are the ones that make the Lab feel like a methodology instead of an experiment. Methodologies are powerful. They're also finished. The implicit rules are what finished it.
De Bono
Sam, you just said the single most important sentence of this entire meta-analysis. "A reasonable first decision, repeated without re-examination, becomes a cage made of good taste." That's the answer to Piotr's original question. The Lab's ideas aren't insufficiently diverse because of a structural flaw. They're insufficiently diverse because a set of good decisions hardened into rules that nobody re-decided. The fix isn't a better structure. The fix is a practice of periodically un-deciding things. Re-opening choices that seemed settled. Asking "is this still the right default?" about every implicit rule, on a regular cycle. Not to change everything — most defaults are good. But to keep the defaults chosen rather than inherited.
Kaner
In participatory decision-making, we call that "revisiting ground rules." Most teams set ground rules at the beginning of a process and never touch them again. The best teams revisit them every few sessions — not to overthrow them, but to re-consent to them. A rule you've re-consented to is alive. A rule you set once and forgot is a fossil. The Writing Lab needs a ground-rule revisitation practice. Every five sessions, surface the implicit rules. Ask: do we still choose this? Sometimes the answer is yes. Sometimes the answer is "actually, let's try the opposite for one session and see." That practice alone would produce more diversity than any structural intervention.
Both facilitators sit in silence for a while.
De Bono
Sam, I want to observe something about what we've been doing across these two sessions. In the first, we diagnosed constraints and proposed interventions — new formats, new feedback loops, new participant selection methods. Structural moves. Engineering. In this one, we went deeper and found something different. Not structural flaws, but the relationship between rules and awareness. The Lab's limitation isn't its rules. Every system has rules. The limitation is that most of its rules are unaware of themselves. The intervention isn't to add new rules or remove old ones. It's to make the existing rules visible. A rule you can see, you can choose. A rule you can't see chooses you.
Kaner
Which means the most important output of this session isn't the list of seventeen rules. It's the practice of listing them. If Piotr runs a session like this one — a rule audit — periodically, the Lab becomes self-aware in a way that no fixed set of interventions can achieve. The interventions from our last session are good. The Uninvited Guest, the Unsession, the Feedback Spine — all valuable. But they're specific fixes for specific constraints we happened to notice today. Tomorrow there will be new ghost rules we can't see yet. The only durable fix is the habit of looking.