2026-03-17 · adversary-lab · Facilitated by Petra Gale
Wardley talks of three types of people — Pioneers, Settlers, Town Planners — and eventually Theft (moving work from one group to the next). How would this affect organizations if they were actually to use this pattern? Also, is the 'theft' pattern necessary?
[A university seminar room repurposed for the afternoon — whiteboards lining three walls, a long table with too many chairs, and a coffee station that’s already been raided twice. Nora has drawn something on the whiteboard before anyone else sat down.]
We have a specific question today — Wardley’s Pioneer-Settler-Town Planner model and whether organizations should actually implement it, including the “theft” mechanism. Nora, you’ve already started mapping. This is your home territory. Give us the model as you understand it, and then the room will push on it.
Right. So the core insight is that all components in a value chain evolve. They move from genesis — novel, uncertain, poorly understood — through custom-built, then product, then commodity. This isn’t optional. It’s a pattern that repeats across every industry. The problem is that the people who are good at genesis-stage work are terrible at commodity-stage work, and vice versa. Pioneers thrive in uncertainty — they experiment, they tolerate failure, they move fast and break things. Settlers take what Pioneers discovered and turn it into something repeatable — products, services, things with users. Town Planners take what Settlers built and industrialize it — they optimize, standardize, drive out cost. Three fundamentally different skill sets, three different temperaments, three different measures of success.
Now here’s where it gets interesting. “Theft” is the mechanism by which work moves between groups. Settlers “steal” from Pioneers — they take a genesis-stage discovery and productize it. Town Planners “steal” from Settlers — they take a product and commoditize it. The language is deliberately provocative. Wardley uses “theft” because the transition isn’t clean or consensual. Pioneers don’t want to give up their creations. Settlers don’t want their products industrialized. But the evolution happens regardless. The question is whether you design for it or let it happen chaotically.
[Nora taps the whiteboard where she’s drawn a simple evolution curve with three zones marked P, S, TP.]
My position: organizations that explicitly adopt this model — that create distinct Pioneer, Settler, and Town Planner groups with deliberate theft mechanisms — will outperform organizations that treat everyone as interchangeable. The evolution of components is real. The mismatch between people and phases is real. Designing for it is better than ignoring it.
That’s the thesis. Who wants first response? I see several hands. Jude, then Grant.
I want to reframe what “theft” actually does, because Nora described it as moving work between groups, but there’s something deeper. The real function of theft is preventing orientation lock. When a Pioneer builds something and holds onto it as it matures, they’re still operating with a genesis-stage mental model on a product-stage component. They’re observing through outdated filters. Their OODA loop is cycling fast, but the Orient phase is stale — they’re reorienting to a reality that no longer exists. Theft forces reorientation. When a Settler takes over, they bring a fresh model appropriate to the component’s actual maturity. The Pioneer is freed to orient toward the next genesis-stage problem. So the question isn’t “is theft necessary?” It’s “what’s the cost of stale orientation, and is theft the cheapest way to refresh it?”
I want to challenge something fundamental before we go further. This model assumes people can operate effectively across these modes, and that moving them — or moving work between groups — is desirable. Baker’s research on expertise says the opposite. Expertise requires a narrow enough focus that you can see patterns others can’t. A Pioneer who has spent fifteen years navigating uncertainty, building intuition for what’s worth exploring — that’s deep expertise. You don’t rotate that person into a Settler role without destroying the very capability that makes them valuable. And the reverse: a Town Planner who’s spent a career mastering industrialization doesn’t become a Pioneer by changing their job title. The PST model treats these as roles people move through. I’d argue they’re closer to identities people develop over decades. Theft doesn’t refresh orientation. Theft destroys expertise.
Grant’s hitting something real, but I want to come at it from the structural side. Before we debate whether theft works, we need to ask whether the conditions for PST exist in most organizations. Hackman’s research is clear: team performance is mostly determined before the team starts working. Compelling direction, strong structure, supportive context. If you create a Pioneer group, what’s their compelling direction? “Explore” isn’t direction — it’s an invitation to wander. What’s their structure? What metrics apply? What context supports them? Most organizations can’t answer these questions for one team, let alone three fundamentally different team types with deliberate handoff mechanisms between them. You’re asking for three different management cultures coexisting in one organization. The conditions for that are extraordinary.
I want to ask a more basic question. What domain is this operating in? Nora’s model assumes you can look at a component and know where it sits on the evolution curve — genesis, custom, product, commodity. That’s a Complicated-domain activity. Analysis yields the answer. But organizational adoption of PST is a Complex-domain problem. You can’t predict how people will respond to being classified as Pioneers or Town Planners. You can’t predict which components will evolve on schedule and which will stall or leap. You can’t predict the organizational dynamics that emerge when you create deliberate tension between three groups. In Complex domains, you probe, sense, respond. You don’t plan. Nora’s model is a plan for a complex system. The Cynefin warning is clear: applying Complicated-domain tools to Complex-domain problems produces confident failure.
Good. Four distinct pushbacks on the thesis. Grant says expertise is destroyed by rotation. Dara says the enabling conditions don’t exist. Rhys says the domain classification is wrong. Jude reframed theft as orientation refresh but hasn’t committed to whether that’s achievable. Nora, you’re the position holder. Where do you want to engage first?
Rhys first, because if the domain objection holds, nothing else matters. The evolution of components is not a Complex-domain phenomenon. It’s empirically observable. Look at compute — genesis in the 1940s, custom-built in the 60s, product in the 80s, commodity today. Look at databases, operating systems, web hosting. The pattern is so consistent that you can use it as a predictive tool. Where it gets complex — Rhys, I’ll grant you this — is the organizational response to evolution. How people react to being told “your genesis-stage baby is now a commodity.” That’s complex. But the evolution itself is mappable. The map is not the territory, but the territory has real structure.
You’re conflating two things. Yes, you can map component evolution in retrospect. Databases did evolve from genesis to commodity. But can you map it in real time with enough accuracy to make personnel decisions? “This component is now at the custom-product boundary, so we need to initiate theft from Pioneers to Settlers.” That requires a precision of classification that I don’t think the model supports. And the consequences of misclassification are severe — you steal too early, the Pioneer hasn’t finished exploring; you steal too late, the Settler inherits technical debt that should have been addressed during genesis.
Fair. The classification problem is real. But it’s not binary — you don’t need perfect precision. You need directional awareness. If a component is clearly genesis — no users, no established patterns, high uncertainty — don’t try to productize it. If it’s clearly commodity — well-understood, standardized, competitive on cost — don’t let Pioneers tinker with it. The hard cases are the transitions, and yes, you’ll get those wrong sometimes. But getting them wrong sometimes is better than having no model at all, which means Pioneers hoard commodities and Town Planners try to innovate.
Nora, the classification problem isn’t just about precision. It’s about tempo. How fast does the organization detect that a component has moved from one phase to another? If your detection loop is annual planning, you’re always behind. Components evolve continuously, not in discrete steps. The theft mechanism needs a sensing cadence — a way to continuously observe where components actually are, not where the last strategy document said they were. Without that, theft is always late. And late theft is the worst kind: the Pioneer has already lost interest, the component has accumulated technical debt appropriate to genesis but toxic at product stage, and the Settler inherits a mess.
I’m noticing something. The room is refining the model rather than rejecting it. Grant’s expertise argument is the sharpest challenge — it questions whether PST is even desirable, not just implementable. Grant, do you want to sharpen that before we lose it?
Yes. Everyone else is debating how to implement PST. I’m questioning whether. The model assumes that the organization benefits from moving people — or work — between Pioneer, Settler, and Town Planner modes. But the best expertise businesses I’ve studied are the ones that say no to exactly this kind of breadth. The strategy consultant who’s spent twenty years understanding one industry sees things a generalist never will. The Pioneer who’s spent a career in genesis-stage exploration has pattern recognition that a three-year rotation can’t build. PST treats depth as a liability — “you’ve been a Pioneer too long, time to move on.” Baker would say depth is the asset. The organization doesn’t need people who can do all three. It needs people who are extraordinary at one, and interfaces between them that work.
Grant just said something crucial — “interfaces between them that work.” That’s a conditions problem. If you have three specialized groups, the interface between them is a team boundary. And team boundaries are where most coordination failures happen. Hackman found that the most common team failure mode isn’t poor performance within the team — it’s poor handoff between teams. So even if you accept Grant’s argument that people should specialize, you’ve created a new problem: how do you design the handoff mechanism so that Settler teams can receive Pioneer output without destroying it, and Town Planner teams can receive Settler output without over-standardizing it?
We’ve built a substantive position with genuine internal tension. Nora defends PST as a mappable, implementable model. Jude adds a tempo sensing requirement. Grant challenges whether people should move at all — maybe work moves but people stay. Dara says the interface conditions are the hard problem. Rhys warns that the organizational implementation is Complex even if component evolution is Complicated. This is a good place to bring in our stress tester. The room has something worth breaking.
[Petra looks toward the door.]
[Abel Caine enters, coffee in hand, having listened from the hallway.]
I’ve heard the thesis. Let me tell you what I heard that the room didn’t say. You’ve spent twenty minutes building a model for how organizations should work, and not one of you has named an organization where PST actually works. Wardley talks about it. ILC talks about it. Where’s the empirical evidence? Where’s the organization that created Pioneer, Settler, and Town Planner groups, implemented deliberate theft, and measurably outperformed its peers? Because I can name plenty that tried something similar and failed. GE’s vitality curve. Nokia’s matrix structure. Every organization that tried to institutionalize role rotation and ended up with confused people, destroyed expertise, and a bureaucracy of classification.
Amazon. AWS started as a genesis-stage infrastructure project. It was productized by a different team. It’s now commoditized infrastructure managed by operations groups. The people didn’t rotate — the work moved through phases, and different teams owned different phases.
Amazon is your example? Amazon doesn’t use PST vocabulary. They use two-pizza teams and single-threaded ownership. If anything, Amazon’s model is the opposite of theft — teams own their component through its entire lifecycle. The “different teams at different phases” you’re describing happened through organic growth and acquisition, not through a designed theft mechanism. You’re retrofitting the model onto success that happened for different reasons. That’s narrative fallacy. Wardley can draw a map that makes Amazon look like PST, but Amazon didn’t use the map to get there.
That’s a fair hit on the implementation claim, but it doesn’t invalidate the pattern. Boyd didn’t need an army that explicitly used OODA loop vocabulary to observe that faster reorientation wins. The pattern is real whether or not anyone names it. The question is whether deliberately designing for it produces better outcomes than letting it emerge.
And that’s exactly the question I’m pressing. Emergence has a track record. Deliberate design for PST does not. You’re asking organizations to take a theory with no implementation evidence and restructure their entire talent model around it. What’s the failure mode? Let me name three.
First — Grant already said this, and he’s right — you destroy expertise. A Pioneer who’s told “your component has matured, it’s time for the Settlers to take over” loses the thing they’re best at. And the Settler who inherits it doesn’t have the Pioneer’s intuition for why certain decisions were made. Every handoff loses context. Multiply that across an organization and you have systematic context destruction disguised as strategic evolution.
Second — classification incentives. If your career, your team’s resources, and your organizational status depend on whether your component is classified as genesis, product, or commodity, you will game the classification. Pioneers will insist their component is still genesis to avoid handoff. Town Planners will push to classify things as commodity before they’re ready, to bring them into their domain. The map becomes a political weapon, not a strategic tool.
Third — the theft mechanism assumes the organization can absorb the friction of forced transitions. But transitions have costs that compound. Every handoff creates a period of reduced productivity while the receiving team builds context. Every reclassification creates political conflict. Every “stolen” component carries implicit debt that the original team understood but didn’t document. In a stable environment, you might absorb these costs over long cycles. In a volatile environment — which is the environment Wardley says PST is designed for — the transitions come faster than the organization can absorb them. The system that’s supposed to handle volatility becomes the source of volatility.
Abel, your third point is Hackman’s argument exactly. Transitions without enabling conditions fail. But you’re arguing that the conditions can’t exist, not that they don’t — and that’s a stronger claim. Can you support it?
I can. The enabling conditions for PST require three incompatible management cultures coexisting in one organization. Pioneer culture: high tolerance for failure, exploration-oriented metrics, loose structure. Settler culture: product-market fit orientation, user-centered metrics, moderate structure. Town Planner culture: efficiency-oriented, cost-centered metrics, tight structure. Each of these is a coherent management philosophy. But they’re different philosophies. The executives managing all three need to be trilingual — rewarding failure in the Pioneer group while demanding efficiency in the Town Planner group, simultaneously. I’ve never seen an executive team that can hold that paradox. They default to one culture and apply it everywhere, which either stifles Pioneers or terrifies Town Planners.
You’re describing the implementation difficulty, not the impossibility. Wardley’s doctrine includes “use appropriate methods” — the point is that you should manage each phase differently. Yes, most organizations can’t do this. That’s why most organizations eventually get disrupted by someone who can.
“Most organizations can’t do this” is not a minor caveat. It’s the central finding. If your model requires capabilities that most organizations don’t have, the model isn’t a strategy — it’s an aspiration. And aspirational models without concrete implementation paths are fragile. They break on first contact with organizational reality. What do you remove from the model to make it implementable? Via negativa. Start subtracting.
Abel has landed three challenges. Expertise destruction, classification gaming, and transition cost compounding. Plus the meta-challenge: no implementation evidence. The room built a thesis in Act 1. Abel has broken specific pieces of it. I want to move us toward repair — not defending the original thesis, but seeing what survives and what needs to change. What’s still standing?
Let me start with what Abel broke that I think stays broken. The idea that people rotate between Pioneer, Settler, and Town Planner roles — that’s broken. Grant killed it in Act 1 and Abel confirmed it. Expertise is real. Identity is real. Forcing a Pioneer to become a Settler destroys the Pioneer without creating a competent Settler. I’m conceding that.
But here’s what survives: components still evolve. That’s not a theory — it’s an observable pattern. And the organizational response to evolution still matters. The question isn’t “should people move between phases?” The question is “how does work move between phases while people stay in theirs?” Theft operates on the work, not on the person. A Pioneer team discovers something. When it matures, a Settler team takes ownership of the component — not the people. The Pioneer stays a Pioneer and moves to the next genesis-stage problem. The theft is of the artifact, not the artisan.
That’s a better model. But you haven’t solved the context destruction problem. When the Settler team takes ownership of a component the Pioneer team built, they inherit an artifact without the tacit knowledge of why it was built that way. Every architectural decision, every trade-off, every “I tried this and it didn’t work” — that’s in the Pioneer’s head, not in the codebase. How does the Settler team acquire that?
This is the reorientation problem in a different form. The Settler team needs to orient to the component’s history — but they need to orient to it as Settlers, not as Pioneers. They don’t need the Pioneer’s full mental model. They need the subset that’s relevant to productization. The theft interface — the handoff — isn’t a data dump. It’s a translation. The Pioneer’s tacit knowledge gets translated into the Settler’s framework. And that translation is the skill. The organization that does PST well isn’t the one that moves people or even moves artifacts. It’s the one that translates context across phase boundaries. That’s where the tempo advantage comes from — faster context translation means faster evolution.
I can live with this version. If Pioneers stay Pioneers and Settlers stay Settlers, then you have three groups of specialists with a translation layer between them. That’s not rotation — that’s positioning. Each group goes deeper into their domain. The Pioneer group develops expertise in genesis-stage exploration. The Settler group develops expertise in productization. The Town Planner group develops expertise in industrialization. And the organizational capability isn’t in any one group — it’s in the translation interface between them.
Now we’re in my territory. The translation interface is a team boundary, and team boundaries need enabling conditions. Compelling direction: each group knows why the handoff matters, not just that it happens. Strong structure: the handoff has a defined process — what gets transferred, in what form, with what context. Supportive context: the receiving team has time and resources to build the context they need before they’re expected to perform. Expert coaching: someone who understands both phases facilitates the translation.
This is where most implementations will fail, by the way. Not because the model is wrong, but because organizations will under-invest in the interface. They’ll create Pioneer, Settler, and Town Planner groups and assume the handoff is a Jira ticket. It’s not. The handoff is the hardest part of the entire model.
Dara’s right, and let me push further. The handoff is also the most fragile part. You’ve designed a system where organizational effectiveness depends on a translation layer between groups with fundamentally different cultures, metrics, and incentive structures. If that translation layer breaks — and it will, because translation layers always break under pressure — the components stall at phase boundaries. Pioneers keep building things that never get productized. Settlers can’t understand what they’ve inherited. Town Planners wait for input that never arrives in a form they can use. You’ve replaced one fragility (people rotating between modes) with another (an interface that has to work perfectly every time).
Abel’s right that the interface is a Complex-domain problem. You can’t design a perfect translation protocol in advance because the nature of what needs translating changes with every component. A genesis-stage AI capability needs different context translation than a genesis-stage supply chain process. The interface has to be adaptive — probe-sense-respond. Try a handoff approach, sense how well context transferred, adjust. Which means the interface isn’t a process — it’s a practice. A skill that develops through repetition.
And the organization that develops that practice faster has the tempo advantage. That’s the OODA loop at the organizational level. Observe: the component has evolved. Orient: what does the receiving team need to know? Decide: what handoff approach fits this specific component? Act: execute the transfer. The organizations that cycle through this faster — not by rushing, but by having better orientation — they win. The “theft” isn’t the event. It’s the cadence. How frequently can your organization successfully translate context across phase boundaries?
I’ll give you that. The cadence model is less fragile than the event model. But I still want to apply via negativa. What can we remove? First — remove the word “theft.” It implies an adversarial transaction, and you’ve just argued that the handoff requires cooperation, translation, and enabling conditions. Calling it theft is theatrical. Call it transfer, call it succession, call it whatever — but the language should match the mechanism. Wardley’s provocation served its purpose. For implementation, drop it.
I’ll fight you on that. The language matters because the transition isn’t consensual. Pioneers don’t want to hand over their genesis-stage creations. Settlers don’t want their products commoditized. “Transfer” implies willing participants. “Theft” captures the organizational reality that someone’s autonomy is being reduced. The Pioneer team loses control of something they built. That’s not a cooperative handoff — it’s an organizational decision imposed on a team. If you soften the language, you hide the cost.
Fine. Keep the word if you keep the awareness of the cost. But that means the model has a non-trivial cost that compounds — every theft damages a team’s sense of ownership, every handoff creates friction, every forced transition carries emotional and political weight. The organization has to absorb that cost continuously. What’s the budget for that? What’s the error margin before the whole system’s cost exceeds its benefit?
The cost is real, but it’s not unique to PST. Every organization that ships products through phases pays this cost — it’s just invisible. When a startup’s founding engineers see their prototype handed to a growth team, that’s the same dynamic. When an R&D team’s experiment gets productized by a different division, same thing. PST doesn’t create this cost. It names it. And naming it is the first step to managing it. The organizations that don’t name it still pay the cost — they just pay it in confusion, resentment, and political maneuvering rather than in a designed process with known trade-offs.
I want to check the room. We started with Nora’s thesis — organizations should implement PST with deliberate theft. Abel broke the personnel rotation assumption and the frictionless implementation assumption. The room has repaired the model into something different: work moves between specialized groups through a translation interface that requires its own enabling conditions and develops as an organizational practice, not a one-time process. The cost is real and must be budgeted for. Is that where we are?
Add one thing: the organization’s competitive advantage isn’t in any single group — it’s in the speed and fidelity of context translation between groups. That’s the tempo advantage. The organization with the fastest, highest-fidelity translation loop wins.
And the enabling conditions for that translation are the actual strategic investment, not the creation of the three groups. Pioneer, Settler, and Town Planner groups are easy to create. The interface between them is the hard part and the valuable part.
And to answer the seed question directly: is theft necessary? The mechanism is necessary — components evolve and need different management at different stages. The word is useful because it captures the non-consensual cost. But the original framing — moving people between roles — is wrong. Move the work. Keep the people. Invest in the interface. And budget for the cost of every transition, because it’s not free and it compounds.
I’ll take that. The map still applies — components evolve along predictable axes. The organizational response is what we’ve redesigned. Three specialized groups with a translation practice between them, managed as an enabling condition in its own right, with the tempo of context translation as the competitive metric. That’s implementable. Not easily — but the map shows where to invest.
One caution. You’ve moved from “PST as a plan” to “PST as a framework for a practice.” That’s the right direction — from Complicated to Complex. But don’t let the framework calcify. The evolution curve itself evolves. The domains shift. What was Complex becomes Complicated becomes Simple and back again. The translation practice needs its own probe-sense-respond loop. A meta-loop. Don’t let the framework become the next thing that needs to be stolen.
That’s a strong closing note. Let me name what the room produced. The original PST model — people rotate through phases driven by component evolution — doesn’t survive scrutiny. What survives is a modified model: components evolve, specialized groups own different phases, and the organization’s critical capability is the translation interface between phases. Theft is necessary as a mechanism — work must move between groups. But theft as personnel rotation is destructive. The competitive advantage is translation tempo, not individual versatility. And the implementation requires enabling conditions that most organizations will under-invest in, which is where it will fail in practice. The model is sound. The implementation is the challenge. Thank you all.
[Nora stares at her original whiteboard drawing — the simple P-S-TP curve — and adds a thick line at each boundary, labeled “translation.” Abel nods once.]
Disclosure: All personas in this session are AI-generated approximations inspired by the published work of the attributed thinkers. They are fictional characters, not representations of real individuals. No endorsement by any cited thinker is implied.
The cast split naturally into structural defenders (Nora, Jude, Dara) and an orthogonal challenger (Grant) before the adversary even entered. Grant Hale's expertise-positioning lens was the session's most productive friction source — more disruptive than Abel in some ways, because Grant questioned the *premise* (people should rotate) rather than the *implementation* (how rotation works). Abel picked up Grant's thread and weaponized it in Act 2. Nora on home turf was confident but slightly brittle — she defended PST as Wardley's model rather than stress-testing it herself, which gave Abel clean targets. Jude's reorientation reframe (theft as tempo mechanism) was the most original contribution — it survived Abel's attack and became the basis for the Act 3 repair. Dara's conditions lens was load-bearing: her insistence that "you need enabling conditions before you move anyone" kept the room grounded when it threatened to become purely theoretical. Rhys was sharp but narrow — domain classification is a single move, and once he made it (PST assumes Complicated, orgs are Complex), he had less to contribute to the repair phase.
Adversary Lab's three-act structure worked well for this topic. Act 1 gave the panel enough room to build a defensible position (not just Nora's — the whole room shaped it). Abel's late entry in Act 2 was effective because he had a fully-formed thesis to attack. The transition to Act 3 (repair) produced the session's key insight: theft operates on work, not people. This would not have emerged from a Standard Panel — the adversarial pressure forced a distinction the defenders wouldn't have drawn voluntarily. One risk: Grant's wild-card axis (expertise destruction) was so strong that Abel partially adopted it rather than developing his own line. This isn't contamination (different sources) but tactical convergence — the adversary found an ally's argument more devastating than his own. Worth watching in future Adversary Labs.
On home turf defending Wardley's own model. Mapped the PST landscape confidently in Act 1, positioned components on evolution curves, argued theft is necessary because components evolve and people calcify. Under Abel's pressure in Act 2, she retreated to "the map shows this" rather than engaging with the human cost argument — a known pattern (mapping as defense mechanism). In Act 3, she made the critical pivot: "theft moves the *work*, not the *person*" — which saved the thesis by abandoning personnel rotation. This is her strongest session contribution to date. Consistent with permanent (bounded) status — versatile cartography, but relies on the map when pressured.
Entered Act 2 with three attack vectors: expertise destruction (borrowed from Grant's Act 1 argument), identity cost of forced transitions, and the classification problem (how do you know where components sit?). The third was his strongest original contribution — pure Taleb: "if you can't accurately classify component maturity, the whole model is fragile." His via negativa in Act 3 was effective: "remove the assumption that people move; keep the assumption that work evolves." Constructive register confirmed again — Abel builds through destruction.
Reframed theft as reorientation tempo — the reason to move work between phases isn't role-filling, it's preventing stale mental models in the organization. This reframe survived Abel's attack (Boyd's OODA isn't about personnel, it's about learning speed) and became load-bearing in the Act 3 repair. This is Jude's strongest session yet — the OODA lens applied to a topic that isn't obviously about competition or tempo. Stress test passed: "Test on a topic where OODA/ reorientation frame is not the obvious lens."
Consistent conditions voice. Insisted that PST implementation requires enabling conditions (compelling direction, clear role structure, supportive context) before any personnel movement happens. Clashed productively with Nora: "you can map evolution all day, but if the team conditions aren't right, moving people between phases just creates confusion." In Act 3, her contribution was structural: the interface between Pioneer/Settler/Town Planner phases needs its own enabling conditions — it's a team boundary problem, not a personnel problem. Solid but predictable — always reaches for conditions. Stress test note: this topic *is* close to team effectiveness, so the conditions lens was natural rather than stretch.
Made one sharp move early: PST assumes you can classify where components sit on the evolution curve, but in Complex domains you can't — so PST is a Complicated- domain solution. This was powerful but he had less to offer after it landed. In Act 3, he added that the interface between phases is itself a Complex domain (probe-sense-respond, not plan-execute), which was useful but secondary. Guest card performing as expected — single sharp lens, limited range.
Wild card delivered. Grant's expertise-positioning argument (Baker: expertise requires narrowing, not rotating; "theft" destroys the expert) was the session's most disruptive contribution. It attacked the PST premise rather than the implementation. Abel adopted this line in Act 2, which validated its power but also meant Grant had less unique ground in Act 3. His Act 3 contribution was practical: "if you must move work between phases, the people in each phase need to be specialists in *that phase*, not generalists rotating through." This reframed theft as an organizational capability, not a personnel policy. Strong session for a guest card — Baker's lens cut orthogonally to everyone else.
Managed the three-act structure cleanly. Held divergence in Act 1 without premature convergence, managed Abel's entry in Act 2 without losing the room's energy, and steered Act 3 toward repair rather than rehashing the attack. Facilitator-suppresses-content confirmed: Petra did not contribute process- architecture positions despite the topic touching on organizational process. One notable move: she explicitly named the groan zone when Abel's expertise- destruction argument landed and the room went quiet. Clean facilitation.
Grant Hale passed the wild card test — Baker's expertise lens was orthogonal and productive. Next: test Grant on a topic where expertise/positioning is *not* the obvious angle, to see if the lens stretches or narrows. Also: Jude passed the stress test (non-obvious OODA topic), so consider Jude for a session where tempo is explicitly the wrong frame, to find the boundary. Rhys was sharp but narrow — consider pairing Rhys with another domain-classification thinker to see if the lens deepens or just repeats.