Phase 0 — Two Facilitators Negotiate How to Facilitate
Kaner
Before we do anything, we need to decide how we're working. I'm not going to pretend we don't have fundamentally different philosophies. My instinct is to build a shared framework of understanding through inclusive process — diverge, struggle in the groan zone, then converge. Yours is to break patterns through provocation and force lateral movement. If we just default to one of our methods, we'll diagnose the lab through the lens of that method and miss exactly what Piotr asked us to find.
De Bono
Agreed. And I notice you just did something characteristic: you named the tension and invited me to participate in resolving it. That's Kaner. My instinct would be to ignore the tension entirely and say something disruptive to see what breaks. So here's my proposal: we alternate ownership. You own the diagnosis — your structured, inclusive approach is better for understanding what exists. I own the redesign — my lateral approach is better for generating what doesn't exist yet. At the end, we both own the convergence, and we must genuinely agree. No polite compromise. If we disagree, we say so.
Kaner
I accept that, with one addition: during your redesign phase, I reserve the right to say "that breaks a load-bearing wall" — to flag when a lateral idea destroys something that's structurally necessary. And during my diagnosis, you reserve the right to say "you're describing the wallpaper, not the architecture" — to flag when I'm cataloging surface features instead of identifying root constraints.
De Bono
Done. Elegant. Let's begin.
Phase 1 — Kaner's Diagnosis: What the Lab Actually Does
Kaner
Let me map the current Writing Lab structure as a decision-making process, because that's what it is — even when it doesn't look like one. Every session follows the same diamond, whether the facilitator knows it or not. There's a divergent phase — thinkers offer their perspectives. There's something like a groan zone — perspectives clash, someone disagrees, the facilitator holds the tension. And there's a convergent phase — the facilitator harvests ideas. The structure works. The ideas are real. But I see five constraints that limit diversity of output.
Kaner
Constraint one: the facilitator always knows where we're going. In every session I've seen, the facilitator enters with a convergence target — "three ideas," "actionable insights," "a brand position." This pre-commitment shapes everything upstream. Thinkers unconsciously orient toward what's harvestable. The truly weird, unresolvable, incommensurable ideas get politely sidelined because they don't fit the convergence frame. The groan zone gets shortened because there's a deliverable to hit.
Kaner
Constraint two: all thinkers are roughly the same species. The sessions draw from published intellectuals — strategists, designers, theorists, technologists. They disagree on content, but they agree on form. They all argue in frameworks. They all produce articulate positions. Nobody in the room thinks in images, tells stories, stays silent for ten minutes, or says "I don't know, and I think not knowing is the answer." The diversity is topical, not cognitive.
De Bono
You're describing the wallpaper.
De Bono
The species problem and the convergence-target problem are symptoms. The architecture problem is: the Writing Lab is a conversation. Every session, regardless of topic, facilitator, or thinker, takes the same shape: people talk in turns, someone synthesizes. It's a seminar with AI participants. The medium — sequential verbal exchange — constrains the output more than any thinker selection or facilitation style ever could. You will never get a fundamentally different kind of idea from a fundamentally identical kind of interaction.
Kaner
That's fair. And it sharpens the remaining three constraints. Let me restate them with that in mind.
Kaner
Constraint three: no silence, no confusion, no unresolved tension. Every session resolves. The facilitator wraps it up. The thinkers reach a shared framework. But in my experience, the most transformative group moments happen when a group doesn't resolve — when they sit with the irreconcilable, and the insight comes days later, in the shower, because the unresolved tension kept working in someone's mind. The Lab never does that. It always closes the loop.
Kaner
Constraint four: the curator is always the same person. Piotr selects the thinkers, the question, the facilitator, and the convergence target. His taste is excellent — the sessions prove it. But excellent taste is still a single taste. The range of possible outcomes is bounded by one person's intellectual periphery. There's no mechanism for the system to surprise the curator.
Kaner
Constraint five: no feedback loop from output to input. Each session is standalone. The insight from Session 3 doesn't shape the thinker selection for Session 7. There's no evolution, no adaptation. It's a series of one-shot experiments with no learning architecture connecting them. The methodology doesn't get smarter because it ran.
Kaner pauses. De Bono has been writing something the whole time.
De Bono
Good diagnosis. Five constraints, but they reduce to two root causes. First: the interaction is homogeneous — always verbal, always sequential, always resolved. Second: the system is open-loop — no learning, no self-modification, no surprise. Everything else flows from these two. The species problem? That's homogeneous interaction wearing a different hat. The convergence-target problem? That's open-loop — the output doesn't question the input. Shall I redesign now?
Kaner
Your phase. I'll flag the load-bearing walls.
Phase 2 — De Bono's Redesign: Seven Lateral Interventions
De Bono
I'm going to move fast. Seven interventions. Some will be practical. Some will be provocations that need refinement. Don't judge them individually — judge the ensemble. A single lateral move is a curiosity. Seven lateral moves is a new architecture.
De Bono
One: The Uninvited Guest. In every session, one thinker should be selected by the system, not the curator. The AI chooses a perspective that is maximally orthogonal to the other participants — not contrarian, but from a genuinely different domain. If the session is about business strategy, the uninvited guest might be a marine biologist, a choreographer, or a medieval historian. The curator doesn't know who it will be until the session begins. This breaks the curator's monopoly on taste and introduces a genuine surprise vector.
Kaner
Not a load-bearing wall. Good. Though the "maximally orthogonal" criterion needs care — you want productive orthogonality, not random noise. A choreographer in a strategy discussion works because spatial composition and timing transfer. A dentist in the same discussion might just be noise.
De Bono
Noted. The algorithm should optimize for transferable structural insight from an unfamiliar domain, not maximum distance. Next. Two: The Unsession. One in every five sessions should have no convergence target. No "three ideas." No "actionable takeaways." The facilitator's only job is to hold open space. The session ends when it ends, possibly mid-argument. The output is the raw transcript, unmarked, unresolved. If Piotr wants to harvest it later, he can. But the session itself doesn't optimize for harvest. This addresses Sam's constraint three — the compulsion to resolve.
Kaner
This one I love. In participatory decision-making, we sometimes run what I call "soak time" — deliberate intervals between the groan zone and convergence where the group lives with the tension. The unsession is soak time elevated to a format. The risk is that without a convergence target, the session might feel pointless. But if it's explicitly framed as "this one doesn't resolve," participants — and the curator — can relax into not-knowing.
De Bono
Three: Break the Medium. Not every interaction in the Lab should be verbal argument. Some sessions should use structured formats that force non-verbal thinking. A "Mapping Session" where each thinker draws their response on a shared canvas — positions, relationships, movements, not sentences. A "Constraint Session" where each thinker can only contribute in six words or fewer. A "Reversal Session" where each thinker must argue the opposite of their known position. The sequential verbal exchange is the deepest constraint. Smash it.
Kaner
This is close to a load-bearing wall. The Writing Lab's output is a readable artifact. The legibility is the product. If you break the medium too far, you lose the thing that makes the Lab valuable. The Wardley session's value was that it produced transferable insight in written form. A canvas of positions isn't legible in the same way.
De Bono
Accepted — with a modification. The non-verbal phase generates raw material. The written artifact comes after, as an interpretation layer. You don't publish the canvas. You publish what the canvas revealed that the conversation wouldn't have. The constraint session and reversal session are easier — they still produce text, just differently structured text. The six-word constraint forces crystallization. The reversal forces each thinker to genuinely inhabit a position they'd normally reject, which produces ideas that no amount of "productive friction" between familiar positions will generate.
De Bono
Four: Thinkers Who Aren't Thinkers. Not every "perspective" in the room should be a named intellectual. Include fictional characters, historical figures from non-intellectual traditions, or composite archetypes. "A first-generation immigrant factory worker in 1920s Detroit" has a perspective on organizational transformation that Peter Drucker never imagined. "A twelve-year-old who just discovered Wikipedia" has a perspective on knowledge that Snowden and Wardley don't. The constraint of using published intellectuals means the Lab thinks the way published intellectuals think — analytically, in frameworks, with citations. Reality doesn't work that way. Most of the world's intelligence lives in people who never published anything.
Kaner
This is powerful, and it directly addresses my constraint two — cognitive homogeneity. Though it creates an ethical complexity. The current ethical framework discloses that the Lab uses "interpretive representations based on published work." An archetype or a fictional character doesn't have published work to be faithful to. The disclosure system would need a new category — something like "constructed perspectives" distinct from "represented thinkers." The integrity of the framework depends on the reader knowing what kind of voice they're hearing.
De Bono
Good flag. Solvable. A tiered system: "Named thinkers are interpretive representations of published work. Unnamed perspectives are constructed archetypes designed to surface viewpoints absent from published discourse." Integrity intact. Cognitive diversity expanded. Next. Five: The Feedback Spine. After every session, the system generates three things: an insight artifact (the current output), a methodology note (what worked and didn't work in the facilitation and structure), and a seed mutation (a question or thinker suggestion generated by the session's content that feeds into a future session). The seed mutations accumulate. Over time, the Lab develops a genome — a growing library of questions and thinker configurations that evolved from the work, not from the curator's imagination alone. This addresses Sam's constraint five — the missing feedback loop.
Kaner
Now you're speaking my language. In participatory decision-making, sustainable agreements only emerge when the process itself can learn. What you're describing is a reflective practice architecture. The Lab stops being a series of standalone events and becomes a learning system. The methodology notes are the most important part — they make the facilitation itself visible and improvable. Most facilitators never document why they made the moves they made. If the Lab documents its own decision-making, it becomes self-improving.
De Bono
Six: Adversarial Sessions. Periodically, run a session where one thinker's explicit role is to destroy the emerging consensus. Not to play devil's advocate — that's a polite disagreement wearing a costume. An actual adversary whose success condition is to find the fatal flaw in whatever the room is building. Give them a different hat color, a different structural role. The room knows they're there. The room knows they're trying to break the idea. If the idea survives, it's stronger. If it doesn't, the adversary just saved the curator months of pursuing a weak thesis.
Kaner
Careful. This can poison the well. If a thinker knows there's a designated destroyer in the room, they start defending instead of exploring. The groan zone becomes a siege instead of a struggle. My modification: the adversary arrives late. The room builds without them. The adversary enters after the convergence draft and attacks the draft, not the process. The room then decides: defend, revise, or abandon. That preserves the generative phase while adding stress-testing.
De Bono
Better. I accept. And finally — Seven: The Session That Writes the Next Session. Once per cycle, run a meta-session where the thinkers from recent sessions — not the facilitator, not the curator — determine the question, the panel, and the structure for the next session. The methodology's own participants shape its future. The curator abdicates for one round. This is the deepest intervention: it gives the system agency over its own evolution. It's the moment the bazaar starts curating itself.
Silence. Both facilitators sit with it.
Phase 3 — The Shared Groan Zone
Kaner
I want to be honest about what I'm feeling. You've given seven interventions, and six of them I can see working. But number seven makes me genuinely uncomfortable. If the thinkers design the next session, the curator loses control. And the curator's judgment — the thing Stewart Brand called "the moat" — is what makes the Lab the Lab rather than a random idea generator. If you give the bazaar agency over its own curation, you might get emergence, but you also might get drift. The sessions could spiral into what the thinkers find interesting rather than what's valuable.
De Bono
And that's exactly why it's the most important intervention. You just described the curator's fear: "If I let go, it might not go where I want." But the whole point of emergence is that it goes where you didn't know you wanted. Every session so far has gone where Piotr's taste directed it. His taste is good. But taste is a filter, and every filter is also a blind spot. Intervention seven doesn't replace the curator — it expands the curator's periphery. One round out of every five or six. The curator still sets the overall direction. But periodically, the system shows him a door he didn't know was there.
Kaner
I need to name what's happening. We're in our own groan zone. I'm defending the existing structure because I believe the curator's judgment is load-bearing. You're attacking it because you believe the curator's judgment is the ceiling. We're both right, and we both know it. So the question isn't whether to include intervention seven. It's how to include it without removing the load-bearing judgment.
De Bono
Then let me offer the resolution. The thinkers generate the next session's question and panel. The curator has veto — not authorship, but editorial judgment. He can reject a suggestion with a reason, and the system generates an alternative. The curator shifts from architect to editor. He doesn't design the house — he approves the design. That preserves taste while surrendering origin.
Kaner
Architect to editor. That's the move. I can live with that. And it maps to something I've seen in the best participatory processes: the facilitator doesn't decide the outcome, but they hold the quality bar. The group proposes. The facilitator ensures the proposal meets the standard. That's not control — it's stewardship.
De Bono
Then we've resolved our groan zone. Let's converge.