The Witness Problem: Designing Wargames That Capture What Players Know But Never Say

Every wargame leaks. Information falls through the gaps between what players know, what they say, and what the adjudication system actually captures. Most designers treat this as noise. It isn't. That leakage is often where the most operationally relevant insight lives.

Photo by Matias Luge on Pexels.

Call it the Witness Problem: the people in the room carry knowledge that never surfaces because nothing in the game's design demands it.

Senior practitioners rarely articulate their priors. Ask a retired submarine commander why a particular Soviet-era maneuver would have failed, and you'll get a shrug and "it just doesn't work that way." Press harder and you might get two sentences of genuine insight that no doctrine manual contains. Run a wargame with that person at the table and watch how often those two sentences get spoken aloud. Almost never. The game structure gives them no reason to externalize what they know tacitly.

This isn't a people problem. It's a design problem.

Why Standard Move-Countermove Formats Suppress Tacit Knowledge

Conventional two-sided wargames (Red vs. Blue, with an adjudication cell) reward speed and decisiveness. Players who pause to explain why they're making a move slow the game down and often feel like they're teaching rather than playing. The social dynamics push toward action. The game loop punishes reflection.

The result: you run a six-hour exercise, fill three whiteboards with decisions, and walk out with almost no record of the reasoning that produced those decisions. You have outcomes. You don't have the model.

The RAND Corporation's early political-military games in the 1950s and 60s partly suffered from this. Participants made sophisticated moves, but debriefs frequently captured conclusions rather than the mental models driving them. Herman Kahn's own post-game analyses leaned heavily on his personal read of what players meant, which introduced a separate distortion. The tacit knowledge problem was baked into the format from the start.

Three Design Interventions That Actually Help

Here's how to build against this failure mode.

Force spoken justification before the move registers. Require each team to verbally state their decision rationale to a recorder before the umpire will accept the move. Not a written form, which players treat like paperwork. A spoken statement, captured on audio or transcribed live. This creates a small friction that pays large dividends: players have to surface the assumption underneath the action. Two or three of those assumptions, per game, per player, will surprise the design team. Budget for the transcription cost. It's worth it.

Run a pre-move prediction market inside each team. Before the opposing team's move is revealed, ask each player individually to forecast what the other side will do and assign a confidence level. Write these down. Compare them to what actually happens. The gaps between a player's confident wrong prediction and reality are a direct map to their blind spots. You've turned the game into a tacit-knowledge extraction engine running in parallel with the scenario.

Design "explain the anomaly" inject points. Seed the scenario with one or two events that experienced players will recognize as odd but that no formal doctrine covers. A shipping route that historically carries more traffic than it should. A diplomatic signal that doesn't match the stated posture. When players encounter these, pause the game and ask the table: what's going on here? The explanations offered in those five minutes are frequently more analytically valuable than anything in the formal move record. Flag them. Return to them in the debrief.

graph TD
    A[Player Decision] --> B{Verbal Justification Required}
    B --> C[Assumption Captured by Recorder]
    B --> D[Move Accepted by Umpire]
    D --> E{Anomaly Inject Point}
    E --> F[Table Discussion]
    F --> G[Tacit Knowledge Surfaced]
    C --> G

The Debrief Is Where You Either Win or Lose This

All three interventions fail if the debrief doesn't treat tacit knowledge as a primary output. Most wargame debriefs organize around chronology: what happened in Turn 1, Turn 2, and so on. That format protects the game designers (it's easy to run) and loses most of the insight.

Reorganize the debrief around the assumptions that surfaced. List them on a board. Ask whether each assumption was shared across teams or held by only one side. Ask which ones, if wrong, would flip the game outcome. You're building a live map of the model that the players were running inside their heads.

That map is the real deliverable. The scenario was just the instrument that extracted it.

Designers who treat wargames as decision records miss this entirely. The game isn't an archive. It's a context for making knowledgeable people say things they'd never say in a meeting room. Your job is to build the conditions where that happens, then catch what comes out.