The Long Game Project · How it works · Glossary · TLDR

The ten words

The ten words a first read needs. Every other term is on the full page.

Persona. The diagnostic's main output: a markdown file on how an organisation decides, with its principles, win conditions, red lines and how it reads others.
Strategic fingerprint. The score layer under the persona: an organisation's reading on the 15 Ingredients, with its coherence and its distance from each group.
Cluster. A family of Flavours.
Coherence. How consistent an organisation is across what it says it values, what it does, and how it is built.
Twin briefing. The persona in its system-prompt form, used when an agent plays an actor in a wargame.
Exercised scenario. A question a leadership team is arguing about, turned into a game with rules and played through by the engine.
Seat. One party at the table in an exercised scenario.
Beat. The smallest unit of play in an exercised scenario.
Arbiter. The referee of an exercised scenario, and it holds no seat.
Evidence pack. The sourced claims behind a published report.

Switch to Normal for the full page, or Deep for the working.

The Long Game Project · How it works · Glossary

Glossary

Every term on the site, defined once. When two pages seem to disagree on a word, this page wins.

A

Adaptability Index: A score built from four Ingredients: pace, change posture, dissent handling and risk appetite. It measures how readily the organisation could change its behaviour if conditions changed, which is a different thing from whether it is changing now.
Aggressive Cluster: The three Flavours built to move first and disrupt: The Pirate Ship, The Mercenary and The Insurgent. They differ on loyalty and ideology, and all three treat friction with competitors as the normal state of business.
Arbiter: The referee of an exercised scenario, and it holds no seat. For each move it hears the claim, the arguments for and against it, and the seat's current position. Then it sets the target number the move must clear, consistent with its earlier rulings in that run. Every ruling logs the decision, the reasoning, the assumptions it rests on and a self-reported confidence. It runs in one of four modes, from ruling on its own to assisting a human who rules. Each report states which mode its exercise used.
Authority Shape: Where the power to decide sits: at the top, or out at the front line. Runs from Centralised (the top decides and the rest execute) to Distributed (the people closest to the problem decide).

B

Beat: The smallest unit of play in an exercised scenario. Every beat runs the same five steps: a seat proposes a move as a claim and names the odds it thinks the move deserves, the other seats object in writing, the arbiter sets a target number, a twenty-sided die rolls against that target, and the claim, objections, ruling, die face and outcome go onto the event log.
Branch: A named timeline. Every scenario starts on "main", and clicking Fork at a tick creates a counterfactual branch named fork-1, fork-2 and so on.

C

Change Posture: The organisation's default reading of change: threat, or opportunity. Runs from Resistant (careful, small steps, a preference for the known) to Embracing (change is the normal state and is sought out).
Cluster: A family of Flavours. There are five: conservative, aggressive, culture-led, systems-driven and independent. The diagnostic places you nearest one of them, or two with a hyphen when the call is close.
Coherence: How consistent an organisation is across what it says it values, what it does, and how it is built. High coherence means it walks the talk on several fronts at once, and low coherence shows the gaps between the claim and the practice.
Competitive Stance: How the organisation positions itself against competitors: partner and coexist, or dominate the market. Runs from Cooperative (treats the market as a shared ecosystem) to Aggressive (treats the market as a zero-sum contest).
Consensus Need: Whether a decision comes down as a directive or needs broad agreement first. Runs from Command (authority decides, others execute) to Consensus (nothing proceeds without wide buy-in).
Conservative Cluster: The three Flavours that share a protective instinct: The Fortress, The Heir and The Vault. They differ on knowledge, tradition and risk, and all three put protecting what they have ahead of chasing what they do not.
Counterfactual: A what-if question. The wargame lets you fork a scenario at any tick and run the alternative timeline beside the main one.
Culture-Led Cluster: The three Flavours that treat values, purpose and people as the main strategic asset: The Cathedral, The Gardener and The Missionary. They look different from the outside, and all three run a decision through the culture before they run it through the numbers.

D

Deterministic re-run: A replay of the same game with the same seats and the same menu of moves but different dice. Every roll comes from a seeded generator and is recorded on the event that used it, so any world can be reproduced roll for roll. Seeded policy agents drive the re-runs rather than the language-model seats. The re-runs therefore describe the shape of the game, and say nothing about how a model behaves. Each fork is compared with its own baseline, seed for seed.
Digital Twin: An older name for the persona at work: the persona file loaded over a language model, so the model reasons as that organisation. In a wargame it plays a seat. The chat with a twin on this site was retired in September 2026.
Dissent Handling: How the organisation responds when someone inside challenges it. Runs from Suppressed (challenge is managed out or ignored) to Encouraged (dissent is invited by design as a quality check).

E

Engine primitives: The building blocks of every wargame scenario. Actor (a player), Resource (something a player spends or builds up), Capability (a flag that gates actions), Action (what a player submits each tick), Tick (one unit of game time), Phase (a step within a tick), Resolver (turns submissions into outcomes) and External event (a scheduled shock from outside the game).
Event-sourced history: The engine stores a log of events and never the game state itself. It computes any state by replaying the log up to a tick, which is why forking, rewinding and exporting an audit trail are cheap.
Evidence Basis: What counts as authority in a decision: pattern-based intuition, or quantified data. Runs from Intuitive (expert judgement and pattern recognition) to Data-Driven (measurement and analysis as the main input).
Evidence pack: The sourced claims behind a published report. Every bracketed marker like [s-003] in the prose points to one claim in the pack. Each claim carries a publisher, a date, a verbatim quote, a retrieval date and a verdict. A linter rejects any sentence with a number in it that cites neither a run event nor a pack claim. Anything without a marker is the author's own, and the reports say so.
Exercised scenario: A question a leadership team is arguing about, turned into a game with rules and played through by the engine. Each party who matters gets a seat and a private brief, play advances in beats, and a named human reads the record and writes the argument. The published reports come from exercised scenarios, and they are scenarios rather than forecasts.

F

Flavour Alignment: How closely an organisation's Ingredient scores sit to one Flavour pattern. High alignment means a clear strategic identity, and low alignment can mean confusion, a transition, or a real hybrid.
Four resolver classes: The four ways the engine can turn submitted moves into outcomes: allocation share, arbiter, matrix payoff and auction. The scenario declares which class each phase uses, and the engine owns how each one computes its result.

G

Growth Model: How the organisation prefers to grow: build the capability inside, or buy it from outside. Runs from Organic (builds slowly from first principles) to Acquisitive (buys capability, market access and talent).

H

Horizon: The time frame the organisation measures success against and allocates money by. Runs from Short-Term (the quarter or the year) to Long-Term (decades of capital allocation and institution building).

I

Independent Cluster: The five Flavours that do not fit the conservative-aggressive or culture-system split: The Swarm, The Laboratory, The Wolf Pack, The Chameleon and The Orchestra. Each runs on its own strategic logic, which is harder to copy than a strategy or a structure.
Interaction Type: How two Flavours get on when they share an environment. There are four types: natural ally, natural adversary, productive tension and indifferent.
IP Openness: Whether the organisation shares its intellectual property or guards it as a moat. Runs from Open (shares IP to build ecosystems and network effects) to Closed (guards IP as a proprietary advantage).

N

Natural Adversary: Two Flavours whose core assumptions clash in ways that cause real friction, well beyond a difference of style. In a partnership or a merger, the pair tends to produce a culture clash that is hard to manage and expensive to fix.
Natural Ally: Two Flavours whose ways of working fit well enough that each makes the other more effective in a shared setting. They need not agree on everything, they just do not waste energy fighting each other.

P

Pace: How fast the organisation decides and acts, given what it knows. Runs from Deliberate (waits for high certainty before it moves) to Rapid (acts on partial information as a matter of strategy).
Persona: The diagnostic's main output: a markdown file on how an organisation decides, with its principles, win conditions, red lines and how it reads others. Load it over any language model and the model reasons like that organisation. It is written from the fingerprint underneath.
Phase: A step within a tick. Most scenarios have one phase per tick, such as allocation, and some have several, such as negotiation, then arbiter resolution, then payoff.
Private brief: What a seat plays from: its goal, its win conditions, its red lines and how it reads everyone else at the table. The seat's Flavour is written into it.
Process Trust: How much the organisation leans on written procedure rather than individual judgement to get work done. Runs from Process-Light (trusts capable people over documents) to Process-Heavy (trusts documents over improvisation).
Productive Tension: Two Flavours different enough to expose each other's blind spots and similar enough to keep working together. These pairings tend to surface more hidden assumptions than natural allies do, and the collaboration is harder work.

R

Reading depth: The three depths every explanatory page and published report can be read at. Easy is a complete crib in its own structure, about 90 seconds. Normal is the full page. Deep adds the annexes: references, mechanics, seams and tables. The control sits beside the theme toggle, and the L key cycles it.
Relationship Signature: A profile of how an organisation habitually responds to eight experiences: uncertainty, conflict, process, success, failure, outsiders, time and identity. These habits change more slowly than strategies do, and they say more about culture than a values statement does.
Risk Appetite: How far the organisation seeks out uncertain bets or steers around them. Runs from Risk-Averse (treats keeping risk low as the strategy) to Risk-Seeking (treats a tolerance for risk as the edge).

S

Scope: Whether the organisation goes deep in one domain or spreads across many adjacent ones. Runs from Focused (deep specialisation with deliberate limits) to Expansive (a wide portfolio, diversified on purpose).
Seams: The section of a published report that says where the machine surprised the author, where the author overrode it, and where the verified evidence cuts against the scenario. Both published reports carry one.
Seat: One party at the table in an exercised scenario. A seat is a character grounded in one of the 16 Flavours rather than a real company, and it plays consistently from its private brief. The arbiter holds no seat.
Second-Order Tension: A strategic dilemma the organisation can manage but never settle, such as exploring against exploiting, or resilience against performance. These tensions come from how the Ingredients interact, and they recur in new forms at every level of the organisation.
Stakeholder Gravity: Who the organisation optimises for: financial return to owners, or a wider purpose. Runs from Shareholder (return to owners is the first filter on a decision) to Mission (purpose and stakeholder impact are the first filters).
Strategic fingerprint: The score layer under the persona: an organisation's reading on the 15 Ingredients, with its coherence and its distance from each group. The persona is written from these numbers, and in day-to-day use you work with the persona rather than the raw fingerprint.
Systems-Driven Cluster: The two Flavours that believe advantage comes from better processes and structures: The Machine and The Architect. They differ in scope and ambition, and both put design ahead of culture as the main lever on performance.

T

Talent Philosophy: Whether the organisation invests in its people for the long run or treats talent as a resource to deploy. Runs from Loyal (invests in tenure, development and deep institutional knowledge) to Mercenary (treats people as interchangeable and optimises for current performance).
The Architect: A systems-driven Flavour that builds for long-term scale through deliberate structural design. It thinks in platforms and ecosystems, guards its IP and prefers to design systems that others run. Strong where platform advantages compound over time.
The Cathedral: A culture-led Flavour that builds for permanence and quality. It measures progress in decades, treats craft as the advantage and guards institutional knowledge, so it is strong wherever quality earns a lasting premium.
The Chameleon: An independent Flavour defined by adaptability on purpose. It changes fast, reads context closely and treats a flexible identity as the advantage. Strong in fast-shifting markets, and it can struggle to say what it stands for over time.
The Fortress: A conservative Flavour built to protect what it already has. It moves slowly, avoids unproven bets and treats stability as the strategy. Strong in settled markets, exposed when a market breaks with its past.
The Gardener: A culture-led Flavour that grows through patient relationships and deep investment in people. It prefers development that emerges over outcomes that are engineered, treats trust as a long-term asset, and leans to cooperation over competition.
The Heir: A conservative Flavour built around an institutional legacy. It treats inherited knowledge and tradition as assets, puts continuity before transformation and moves slowly to protect what earlier generations built. Strong in trust-based markets, exposed to first-principles disruption.
The Insurgent: An aggressive Flavour that sets out to disrupt the established order. It moves fast, makes big bets, treats incumbency as a liability and frames its purpose as changing the rules. Strong at the start of a market shift, and it can struggle to become the incumbent.
The Laboratory: An independent Flavour that treats the organisation as a machine for testing hypotheses. Data is the authority, dissent is invited by design and failure is treated as signal. Strong in complex, uncertain domains where evidence compounds over time.
The Machine: A systems-driven Flavour built for scale and consistency. Process is the advantage, variance is managed out and reliable execution is what it sells. Strong in stable, high-volume settings, brittle when conditions change.
The Mercenary: An aggressive Flavour built around performance and results. Talent is a resource to deploy, loyalty depends on outcomes and commercial return is the one clear goal, which builds a culture that repels anyone not motivated by money or winning.
The Missionary: A culture-led Flavour driven by a stated purpose that reaches past commercial return. It attracts true believers, decides through a values filter and builds deep loyalty, and the same conviction can harden into strategic rigidity.
The Orchestra: An independent Flavour that delivers complex, coordinated outcomes through disciplined ensemble work. It values collective excellence over individual virtuosity, invests in the machinery of collaboration and treats long relationships as a performance multiplier.
The Pirate Ship: An aggressive Flavour built for speed and opportunism. It decides on partial information, treats process as drag and does well in chaos, which makes it dangerous in a stable, regulated market.
The Swarm: An independent Flavour that runs on distributed intelligence. It pushes decisions to the edges, runs many experiments at once, and holds together through shared principles rather than central direction.
The Vault: A conservative Flavour built to protect and compound proprietary knowledge and assets. Very low risk appetite, heavy reliance on process and deep scepticism of outside influence. Strong wherever knowing more than the other side is the main advantage.
The Wolf Pack: An independent Flavour built around elite individual performers. It moves fast, rewards producers over managers and is loyal to the outcome rather than the institution. Strong where results are all that count, exposed to coordination failures at scale.
Three measurement registers: The three kinds of question in the diagnostic, one of each per Ingredient. Preference asks for a stated position on a slider. Forced choice asks what you do when you cannot do both. Values asks what the organisation believes it stands for.
Tick: One unit of game time. Atlas Regional runs six quarterly ticks (Q1 to Q6), and other scenarios use monthly, weekly or event-driven ticks. Tick zero (T+0) is the starting state before anything has been played.
Twin briefing: The persona in its system-prompt form, used when an agent plays an actor in a wargame. It is the same document as the persona, under its older name. The diagnostic produces it, and you can also write or edit one by hand in Studio.

The annex · Deep

Which page is the authority on each term.

Terms	Authority
Persona, Twin briefing, Strategic fingerprint, Coherence	The diagnostic
Cluster, and the sixteen Flavours	The Flavours
Exercised scenario, Seat, Private brief, Beat, Arbiter, Evidence pack, Seams, Deterministic re-run	The engine
Reading depth	This page
Anything about accuracy	The evidence page

A definition on this page and a description on its authority page should say the same thing. When they do not, the authority page is the one we maintain, and the glossary entry is the one to fix.