Phase 1 — Just the Two of Us, Before the Door Opens
De Bono
I want to start before anyone else arrives. Because the most important thing we can do is not begin by listing lab types. That's the convergence impulse — get to the catalog. Instead: why would the Lab be more than one thing? What's the argument for a family of formats rather than a single perfected format?
Kaner
Because different questions have different shapes. In participatory work, I learned this the hard way: the same facilitation process that produces brilliant results for a strategic planning retreat produces garbage when applied to a conflict resolution. Not because the process is bad — because the purpose is different. Strategy needs divergence. Conflict needs convergence. Innovation needs play. Grief needs silence. If the Writing Lab has one format, it can serve one purpose. If it has many, it becomes a practice.
De Bono
Good. And there's a second reason: the Lab should surprise itself. A methodology with one format calcifies. It gets better at doing its thing, but "its thing" narrows. A family of formats means the methodology keeps encountering its own edges. A session in an unfamiliar format teaches the curator something a familiar format never will — even if the unfamiliar session fails. Especially if it fails.
Kaner
Let's set one ground rule for ourselves: for each lab type, we name the what, but also the which rules it breaks. We built that audit of seventeen rules. Every new lab type should be describable by which of those rules it violates. If a proposed format doesn't break at least one rule from the audit, it's not a new type — it's a theme on the existing type.
De Bono
Excellent constraint. Let me start with two that are close to the current format, to warm up. Then we'll need a product mind. I'll knock on the door.
01
The Unsession
The one that doesn't resolve
We proposed this in the previous meta-session but didn't build it out. Here's what it actually is: a full panel, a full facilitator, a full question — but the facilitator's only instruction is do not converge. No synthesis. No "three ideas." The session runs until the energy dies or something genuinely irreconcilable surfaces. The output is the raw transcript, published as-is, with a single editorial note: "This session did not resolve. The tension is the artifact." The value: some questions shouldn't be answered. They should be held. The Lab currently has no way to hold a question without answering it. This is the format for questions that are too large, too early, or too important to close.
breaks: Visible Rule 4 (written artifact as synthesis) breaks: Implicit Rule 4 (consistent aesthetic)
02
The Question Lab
The panel generates the question, not the answer
Select four thinkers from maximally different domains. No question is given. The facilitator's opening prompt is: "What question does this specific combination of people make it possible to ask?" The session's entire purpose is to find the question. The output is a single question — formulated, refined, stress-tested — and a one-paragraph provenance explaining how the collision of these four frames produced something no individual frame could have generated. The question then becomes a seed for a future conventional session. This is the Lab format that generates its own fuel.
breaks: Ghost Rule 2 (question precedes panel) breaks: Visible Rule 2 (singular stated question)
Kaner
Those are the gentle ones. Now let's open the door. We need someone who thinks in products — someone who'll ask "who's this for?" and "what's the unit of value?"
Kaner opens the door. Singer walks in, coffee in hand.
Phase 2 — Singer Enters. The Room Gets Sharper.
Singer
I've been listening through the wall. You're building a product catalog without a customer. Who are these for? The current Writing Lab serves one person — Piotr. He's the curator, the audience, and the publisher. If you're imagining a family of formats, you need to answer: is each format for a different purpose, a different customer, or a different stage of a larger process? Those give you completely different product architectures.
De Bono
Say more about stage.
Singer
In Shape Up, we distinguish between shaping (figuring out what to build) and building (executing the plan). Different activities, different people, different timescales. The Writing Lab has been doing both at once — finding the question and answering it in the same session. What if different lab types serve different stages of an intellectual pipeline? Stage one: find the question. Stage two: explore the question divergently. Stage three: stress-test the emerging answer. Stage four: crystallize the output for publication. That's not four flavors of the same thing. That's a production line for insight — where each stage has its own format, its own rules, its own success criteria.
Kaner
Ryan, you just restructured the entire conversation. Edward and I were thinking "lab types" as a menu — pick whichever one fits. You're thinking "lab types" as a sequence — each one feeds the next. Those aren't mutually exclusive, but the sequence idea is much more powerful because it creates compound insight. Session 1's output is Session 2's input. The value accumulates.
Singer
Exactly. And it also gives you natural product tiers. A single standalone session is the entry product — quick, self-contained, valuable on its own. A three-stage sequence (question → exploration → stress-test) is the premium product — deeper, slower, compound. A full four-stage pipeline is the enterprise product — a complete intellectual process from raw curiosity to publishable artifact. Different customers at each tier. Different price points. Different appetites.
De Bono
Ryan is right about the pipeline, and I want to build on it while adding formats that don't fit the pipeline. Because not every lab type is a stage in a production process. Some are standalone experiments. Some exist to break things. Some exist to surprise the curator. Let's go fast now. I'll name them. Ryan shapes the product angle. Sam flags the facilitation mechanics. Go.
03
The Collision Lab
Two panels in the same room, on different questions, forced to talk to each other
Take two unrelated sessions — say, one on agent ecosystems and one on coaching burned-out executives — and run them simultaneously in the same space. Each panel can only see the other panel's contributions in real time. The facilitator's job is to find the bridges. The format is chaotic, noisy, and frequently incoherent. The insight, when it comes, is a connection between two domains that nobody in either domain would have made on their own. "The burnout pattern in executives is structurally identical to the drift pattern in unmonitored agents" — that's a Collision Lab output. It can't come from anywhere else.
breaks: Visible Rule 2 (singular question) breaks: Deep Rule 1 (single-pass linearity) breaks: most notions of coherence
Singer
Product angle: the Collision Lab is the "surprise me" tier. You give the system two questions you're working on and it finds the bridge you didn't know existed. High-value for consultants, strategists, anyone doing cross-domain work. Low repeatability — each collision is unique. That's a feature, not a bug. It's the Lab format you can't get anywhere else.
04
The Adversary Lab
One thinker's job is to destroy the thesis
A conventional panel builds a position for forty minutes. Then the adversary enters — late, prepared, armed with everything the panel said. The adversary's success condition is to find the fatal flaw. Not to play devil's advocate. To actually break the argument. The panel then has ten minutes to defend, revise, or abandon. The output includes all three possibilities: the original thesis, the adversary's attack, and the panel's response. The reader sees the idea tested under load. Ideas that survive the Adversary Lab carry a different weight than ideas that only survived agreement.
breaks: Implicit Rule 3 (facilitator serves the session) breaks: the assumption that all participants share a goal
Kaner
The facilitation here is tricky. The adversary can't arrive while the idea is still forming — they'd kill it in the cradle. And they can't arrive so late that the panel is emotionally committed and defensive. There's a narrow window. The facilitator has to sense when the idea is strong enough to be worth attacking but not yet rigid enough to be worth defending. That's high-skill facilitation.
05
The Reversal Lab
Every thinker argues against their own known position
Wardley argues that maps are useless. Snowden argues that Cynefin creates false certainty. Singer argues that Shape Up is a prison of manufactured urgency. Each thinker must make the strongest possible case against their life's work. The facilitator's only rule: no strawmen. You must find the real weakness, the one you've privately worried about but never published. The output: a set of genuinely dangerous critiques that could only come from people who know the framework intimately enough to attack it with precision. This is the Lab format that produces intellectual honesty.
breaks: Implicit Rule 2 (thinkers selected for relevance/advocacy) breaks: the assumption that each thinker represents their position
Singer
Product angle: this is dangerous and therefore valuable. Every consultant, every thought leader, every methodology creator knows their framework's weak spot but never publishes it. The Reversal Lab produces the critique that the creator's fans can't write and the creator's critics won't write with enough precision. That's exclusive content. It has an audience.
06
The Silence Lab
Responses are limited to six words. Then nothing.
A question is posed. Each thinker responds in exactly six words. No elaboration. No follow-up. Then the facilitator reads all responses aloud and the room sits in silence for a defined interval. Then each thinker responds to the silence — not to each other's words, but to what the silence after the words produced — again in six words. Three rounds. Eighteen words per thinker. The output is a page of compressed fragments that the reader unpacks like poetry. The value: the Lab currently has no mechanism for half-formed thought, for the gesture that precedes the argument. The Silence Lab forces crystallization and eliminates the Lab's deepest habit — its verbosity, its compulsion to explain. Some insights don't survive explanation. They only survive compression.
breaks: Deep Rule 2 (equal articulation) breaks: Deep Rule 1 (single linear pass) breaks: every implicit assumption about what a "session" looks like
De Bono
We need someone else in the room. The next formats I'm seeing are about generative systems — formats where the output is unpredictable not because the thinkers are diverse but because the rules are generative. I want someone who's spent a lifetime designing systems that surprise their creator.
De Bono walks to the door. Opens it. Brian Eno is standing there, as if he'd been waiting.
Phase 3 — Eno Enters. The Room Changes Temperature.
Eno
I was listening for the part where you'd need me. The Silence Lab was the cue. You're beginning to realize that the Lab doesn't have to be a conversation. It can be a composition. In music, the most interesting structures aren't debates between instruments. They're systems where simple rules produce complex outcomes. A Steve Reich phase piece: two identical loops, slightly out of sync, creating interference patterns that neither loop contains on its own. That's not dialogue. It's generative architecture. You don't need thinkers who argue. You need thinkers who resonate and interfere.
Kaner
Brian, the Writing Lab has always been built on conversation as its medium. You're suggesting it doesn't have to be.
Eno
Conversation is one generative system. It's a good one. But it's not the only one. When I created the Oblique Strategies cards in the seventies, the insight was that creative people don't need more information. They need a constraint they didn't choose delivered at the moment of maximum stuckness. The card doesn't tell you what to do. It disrupts the pattern you're in, and the disruption generates the next move. What if some Lab formats worked like that? Not conversations. Constraint engines.
07
The Oblique Lab
The session is interrupted by random constraints at unpredictable intervals
A conventional session runs — question, panel, facilitator. But at random intervals, the system draws an Oblique Strategy and imposes it on the room. "Emphasize the flaws." "What would your closest friend do?" "Remove the most important thing." "Use an old idea." The room must incorporate the constraint into the live conversation. Some constraints will be productive. Some will be absurd. The facilitator's job shifts from steering to metabolizing — helping the room digest the interruption and extract value from it. The output captures both the planned discussion and the disruptions, showing how random constraints produced ideas that intentional facilitation wouldn't have.
breaks: Visible Rule 2 (singular question stays stable) breaks: Implicit Rule 3 (facilitator serves the session) breaks: Ghost Rule 3 (facilitator is always human-modeled)
Eno
Now here's one that goes further. What if the Lab format had no thinkers at all?
08
The Ambient Lab
No named thinkers. Only voices. Only atmosphere.
A question is posed. Instead of named intellectual perspectives, the session generates anonymous voices — identified only as "a voice from the direction of materials science," "a voice from the direction of grief," "a voice from the direction of competitive gaming." The voices don't argue. They contribute observations — short, unattributed, not required to be coherent with each other. The facilitator weaves them into a texture. The output reads like a field recording of an intellectual landscape rather than a transcript of a debate. The value: it eliminates the Lab's dependency on the authority of named thinkers. The idea stands or falls on its own, not on who said it. And it gives the Lab access to perspectives that don't have a published body of work to be "faithful to" — because there's no named author to be faithful to.
breaks: Visible Rule 5 (attribution to named thinkers) breaks: Implicit Rule 2 (thinkers selected for relevance) breaks: the entire ethical framework — requires a new one
Singer
Brian, that's elegant, and it solves the ethical problem we flagged in the previous session. If there are no named thinkers, there's no misattribution risk. The ethical framework simplifies to: "these are AI-generated perspectives, not representations of real people." But it creates a new problem: the named thinkers are the Lab's marketing. "Wardley and Singer and Snowden discuss your strategy question" is a sellable proposition. "Anonymous voices explore your strategy question" is not. The Ambient Lab is the purest format intellectually and the weakest format commercially.
Eno
Unless you position it differently. Ambient music isn't sold as "music without performers." It's sold as an environment to think inside. The Ambient Lab isn't a panel discussion without names. It's an intellectual atmosphere. You don't sell it as a session. You sell it as a space. The customer enters with a question and spends thirty minutes inside a landscape of related perspectives. That's not a report — it's a thinking environment. Different product entirely.
De Bono
Brian, you've just described the first Lab type that isn't a document at all. The Ambient Lab's natural output isn't a PDF or an HTML page. It's an interactive experience. A space you enter, move through, and leave changed. That breaks Ghost Rule 4 — the artifact is the only output — more thoroughly than anything else we've proposed.
09
The Haunted Lab
The session is attended by the ghosts of its own previous failures
Before the session, the system reviews past Lab sessions and identifies moments where an idea was abandoned, a thread was dropped, a thinker was overruled, or a convergence was forced prematurely. These "ghosts" — the ideas that didn't survive — attend the new session as silent observers. At a moment chosen by the facilitator, the ghosts speak: "In session seven, this idea was abandoned. Here's what it might have become." The living panel must reckon with the dead ideas. Some ghosts are rightly dead. Some were killed prematurely. The format is the Lab's feedback spine made literal — the past sessions haunt the present ones.
breaks: Ghost Rule 1 (sessions are standalone) breaks: Visible Rule 4 (each session is a complete artifact) requires: the feedback spine from meta-session 1
Kaner
Edward, that one gave me chills. In facilitation, the biggest regret is always the idea that got away — the thing someone almost said, the direction the room almost took. Most methodologies have no mechanism for resurrection. The Haunted Lab is a second chance architecture.
10
The Relay Lab
Three panels, sequential, each inherits only the previous panel's questions — not their answers
Panel A runs a session and produces a set of questions — not answers, questions. Panel B receives only those questions, with no context about Panel A's discussion. They run a second session. They produce answers, plus new questions. Panel C receives only Panel B's new questions, again with no context. The final output layers all three sessions, showing how the same intellectual energy traveled through three completely different frames. The insight: ideas evolve differently when they pass through multiple independent minds than when they stay in one room. This is the Lab's version of the telephone game — except instead of degradation, you get mutation and selection.
breaks: Deep Rule 1 (single pass) breaks: Ghost Rule 5 (the room is a single event) combines: Question Lab → Standard Lab → Standard Lab
Singer
The Relay Lab is the pipeline idea from earlier, except it's not a production line — it's a mutation engine. The value isn't that it's thorough. The value is that it's divergent by design. Three panels will produce three incompatible perspectives on related questions. The customer doesn't get an answer. They get a map of the possibility space. That's a premium product for anyone facing a genuinely ambiguous strategic decision.
11
The Mirror Lab
The Lab examines the Lab. On a schedule.
Every fifth session is a meta-session. Two facilitators — or the system itself — review the previous four sessions and ask: what rules emerged? Which defaults hardened? What ghost rules formed? What questions were never asked? The Mirror Lab is not optional. It's structural. It's the periodic rule audit we proposed, built into the methodology as a format rather than an afterthought. Its output is always: methodology notes, seed mutations for future sessions, and a list of rules that need re-consenting. This is the format that keeps all other formats alive. Without it, any lab type will calcify. With it, the entire family evolves.
breaks: Ghost Rule 1 (sessions are standalone) breaks: Visible Rule 2 (the question is about a topic, not the methodology) is: the immune system for the entire family
Eno
One more. I've been holding it because it's the strangest one and it might be the most important.
12
The Empty Lab
No thinkers. No facilitator. No question. Just the curator and the space.
The curator enters the Lab alone. No panel. No facilitation. No question. The system is present — available, responsive — but it doesn't initiate. The curator writes what's on their mind. Half-formed thoughts. Fragments. Contradictions. The system reflects it back — not as analysis, not as synthesis, but as mirror. "You seem to be circling something about trust." "This fragment contradicts the one from three minutes ago — is that deliberate?" The output is private. Never published. The Empty Lab exists for the curator to think without performing. To encounter their own mind without the mediation of thinker-perspectives, facilitation patterns, or convergence pressures. The value: every other Lab format produces output for an audience. This one produces clarity for the person running the system. It's the format that acknowledges that the curator is also a thinker, and that the thinker sometimes needs the room to themselves.
breaks: every rule simultaneously is: the Lab format that exists for the human at the center
The room is very quiet.
Kaner
Brian, that's the one.
Singer
It's unpriceable and therefore it's the foundation. Every other format is a product. This one is the practice. Without it, the curator burns out, loses their edge, starts optimizing for output instead of thinking. The Empty Lab is the thing that keeps the curator worth listening to. It's the quiet room that makes all the noisy rooms possible.
De Bono
Twelve lab types. I count four that fit into Ryan's pipeline — Question, Standard, Adversary, and Relay. Four that are standalone experiments — Collision, Reversal, Silence, and Oblique. Two that are structural — the Mirror and the Haunted. One that's a new medium entirely — the Ambient. And one that's private — the Empty. That's not a catalog. That's an ecology. Different species, different niches, different survival strategies, coexisting in the same ecosystem. Some feed each other. Some compete. All evolve. The Writing Lab just stopped being a thing you do and started being a place you inhabit.