Reference

War-Game Facilitation Script

Part of the War Gaming skill · loaded on demand from SKILL.md

Read this when designing or running a live game, or when writing a desk game and you need the structure the live version would have had. The script assumes a two-day game with three rounds; compress to one day by cutting the team-preparation session and running 45-minute rounds.

Roles

Control lead (the engagement partner). Owns the clock, the adjudication rule and the surprise log. Does not play. Intervenes only to enforce rules, resolve disputes and introduce injects.

Control analysts (two). One runs the model and produces the scoreboard; one records every move sheet and every plenary statement verbatim enough to quote in the debrief.

Home team. Four to six executives with authority over the option under test. The decision owner must be on this team or in the room.

Rival teams. One per key rival, three to five people each. Put the strongest strategic thinkers here, and at least one outsider per team who knows the rival or its industry segment. A rival team led by a junior analyst produces a passive rival and a worthless game.

Market team. Two to four people: real customers under NDA where possible, otherwise the client's customer-insight lead plus a supplier representative. The Market team does not strategise; it reacts. Its output each round is a share allocation and a service score within the adjudication rule.

Two weeks before

  • Send each rival team its briefing pack: setting scenario, its own response profile, its payoff metric, the rules and legal moves, the adjudication rule, the public facts on every other player. Ask each rival team to submit, before the game, a one-page statement of "what we want by the horizon year and what we will not do".
  • Send the home team the option under test and the same rules, but not the rival profiles in full; the home team should discover rival logic through play.
  • Load the model with the setting scenario's parameters and test that a round of moves can be adjudicated in under 40 minutes.
  • Choose injects. Each inject must test a named "what would have to be true" condition of at least one option, and each must come from a signpost or a scenario narrative so the game also tests the scenario set.

Day one

09:00 Opening (30 minutes). Control lead states the focal decision, the setting scenario, the horizon, the payoff metrics for each team, and the adjudication rule. Say plainly: "The home team is expected to lose at least one round. Rival teams are scored on how faithfully they play their rival's likely behaviour, not on how much damage they do." Take questions on the rules only.

09:30 Team preparation (60 minutes). Teams absorb their packs. Control visits each rival team and asks two questions: "What does your rival's CEO want this year?" and "What has your rival done the last three times a competitor moved on its accounts?" If the answers are not specific, the team is not ready; give them the profile again.

10:30 Round 1, deliberation (60 minutes). The home team's option is announced to all teams as public information at the start (this is the sequential-move structure: the client commits, rivals respond). Teams write moves on the move sheet: action, magnitude, timing, rationale in their own voice, expected response from others. Control collects sheets at the deadline; late sheets are treated as "no move".

11:30 Round 1, adjudication (40 minutes). Control analysts apply the rule and the model. The Market team allocates share within the rule and scores service. Control lead writes the first entries in the surprise log: any move that differs from the competitor-response matrix.

12:10 Round 1, plenary (20 minutes). Control publishes the state of the world and the scoreboard. Each team gets two minutes: what it did and why. No debate; questions of clarification only. Home team notes what it did not expect.

12:30 Lunch.

13:30 Round 2 inject (10 minutes). Control reads the inject as a news bulletin dated to the period. Teams may ask factual questions; Control answers only what would be public.

13:40 Round 2, deliberation (60 minutes). As Round 1. Control visits the rival teams and pushes: "Is that what they would actually do, or what would hurt the home team most?" Both errors are corrected.

14:40 Round 2, adjudication (40 minutes). As Round 1. Where two teams' moves conflict (two bidders for the same target, two players adding capacity for the same demand), Control resolves using the published rule; if the rule is silent, Control decides, states the decision and the reasoning aloud, and records it. Do not let adjudication become a negotiation.

15:20 Round 2, plenary (20 minutes).

15:40 Break.

16:00 Desk-game variants (90 minutes, Control and home team only, optional). If the home team is playing one option live and two others are being tested on paper, Control runs the other options through Rounds 1 and 2 using the rival teams' written moves and stated logic, adjusting only where the rival would clearly have acted differently against a different home-team move. Rival teams can be consulted for ten minutes each.

17:30 Close of day one. Control lead reads the surprise log so far, without interpretation.

Day two

08:30 Day-two opening (20 minutes). Control publishes the state of the world at the start of the final period, including any desk-variant results. Reminder of payoff metrics.

08:50 Round 3 inject (10 minutes). The final inject is normally the most severe, because the last round is where teams reveal what they would do when the option's core assumption is broken.

09:00 Round 3, deliberation (60 minutes). Control adds a mid-round inject at 30 minutes if the design calls for one. Teams must submit final-period moves and a one-paragraph "where we stand at the horizon" statement.

10:00 Round 3, adjudication (40 minutes). Final scoreboard produced for all options: live and desk variants.

10:40 Round 3, plenary (20 minutes).

11:00 Break.

11:15 Debrief (120 minutes). Structured in this order; do not skip or reorder.

  1. Out of role (10 minutes). Each rival team states, as themselves, the one thing they learned about the rival by playing it.
  2. Surprise log (20 minutes). Control lead reads each surprise; the room agrees or amends the entry. For each: was it predictable from the response matrix, and if not, what does the matrix now need?
  3. Option by option (40 minutes). For each option: which round hurt it, which rival move hurt it, what the home team would now do differently, and the revised horizon-year payoff. Control analysts update the payoff table live.
  4. Hidden options (15 minutes). What move became available during play that was not in the option set? (Partnerships, pre-emptive contracts, deals with a target on non-price terms.) Record each as a candidate for the adaptive plan.
  5. Signposts (15 minutes). Which injects would the client have seen coming with the signpost table as designed? Add or sharpen signposts accordingly.
  6. Decisions (20 minutes). The decision owner states what changes as a result: option choice, staging, options to buy, triggers. Control records decisions with owners and dates. A game that ends without this step is a seminar.

13:15 Close. Control lead commits to circulating the round record, the surprise log, the updated payoff table and the decision list within three working days.

Adjudication principles

  • Publish the rule before play. Teams should be able to anticipate outcomes; strategy is impossible against a hidden model.
  • Simplicity beats fidelity. A share-response rule with a lag and a cap is enough; do not build a market simulator.
  • No added value, no capture. If a team claims share or margin without a move that changes relative price, service, availability or cost, deny it and say why.
  • Lead times bind. A fulfilment centre takes a year; an acquisition closes two quarters after agreement; a sales force takes two quarters to become productive. Enforce them.
  • Balance sheets bind. A team that breaches its stated debt constraint cannot make the move.
  • Record the reasoning, not just the result, for every ruling. The debrief will need it.

Desk-game instructions

When a live game is not possible, run the same structure alone. Write each team's move sheet in the team's voice, from its response profile and its payoff metric, before looking at the home team's counter. Apply the adjudication rule explicitly and show the arithmetic. Keep the surprise log honestly: if writing the rival's move revealed something the matrix missed, that is a surprise even though no room was involved. Label the output "desk game" in every heading so the client does not mistake it for a live exercise, and recommend a live game before any irreversible commitment above the client's materiality threshold.

Common facilitation failures

FailureSymptomFix
Passive rivalsRival teams "wait and see" every roundControl asks the two readiness questions again; replace the team lead if needed
Hostile rivalsRival teams make moves that would bankrupt the rival to hurt the clientRemind them they are scored on fidelity; enforce balance-sheet constraints
Home team defends the planHome team argues with adjudication instead of adaptingControl lead restates that losing a round is the point; the decision owner sets the tone
Adjudication driftRulings become negotiationsRule published in advance; rulings stated with reasoning and not reopened
Debrief collapses into war storiesNo decisions by the endRun the six-step debrief in order; the decision step has a fixed start time
Game tests the wrong worldInjects unrelated to the decision's assumptionsEvery inject maps to a "what would have to be true" line and a signpost