The Competence Trap: Designing Wargames That Expose Expert Blind Spots

There's a specific failure mode that shows up in almost every professionally run wargame, and it's the one nobody wants to name: the most qualified people in the room are often the worst at anticipating surprise.

Photo by Ivan S on Pexels.

Not because they're unintelligent. Because they're expert. Their mental models are load-bearing. Challenge them and the whole cognitive structure wobbles, so the brain quietly refuses.

This is the competence trap, and if you design your wargame around subject-matter experts without accounting for it, you've built an expensive mechanism for confirming what those experts already believe.

Why Expertise Narrows the Possibility Space

When the Naval War College ran Global War Game 1979, one of the persistent critiques afterward was that senior naval officers playing the Soviet red cell kept defaulting to Soviet doctrine as they understood it, which is to say, as Americans had modeled it. The truly anomalous Soviet moves, the ones that might have reflected actual Soviet operational culture or leadership pathology, never got played. The experts were too expert in the American version of Soviet thinking.

This isn't unique to that game. Expertise compresses variance. Someone who has spent twenty years modeling a system knows which branches of the decision tree are "unrealistic", and prunes them before play even begins. What gets pruned, reliably, are the branches where the tail events live.

The design question isn't how do we get better experts. It's: how do we build a game that makes expert over-pruning visible and costly?

Three Design Moves Worth Running

Invert the briefing order. In most wargames, subject-matter experts receive detailed background briefings before play. That primes them. Instead, run a short opening turn with no briefing, just a starting scenario state and a decision to make. Record the moves. Then deliver the expert briefing and run the same turn again. The delta between pre-briefing and post-briefing decisions is your first dataset. Where experts changed their behavior after receiving "authoritative" framing, you've identified the priming points. Those are exactly where your scenario's assumptions are doing silent work.

Create a dedicated Naïve Actor role. This is not a junior staffer sitting quietly in the back. It's a formally designed player position, with its own victory conditions, who is explicitly rewarded for moves that violate received doctrine. Brief them separately. Give them access to historical anomalies, actual cases where actors did the "wrong" thing and it worked. The Naïve Actor's job is to play the scenario as if they've never read the expert literature. If the game's adjudication system consistently punishes those moves, that's a signal your adjudication is baking in the same assumptions you're trying to stress-test.

Build a friction log into adjudication. Every time a move gets ruled out-of-bounds, implausible, or too extreme by an umpire or adjudication cell, it gets logged with a brief justification. After the game, the design team reviews that log. Look for clusters, categories of moves that kept getting killed. Those clusters are your institutional reflex map. They tell you what the expert community has collectively decided cannot happen. History has a habit of disagreeing.

The Adjudication Problem

None of this survives contact with a bad adjudication setup. If your umpires are drawn from the same expert community as your players, the friction log will be full of sensible-sounding reasons why anomalous moves were unrealistic. They're not wrong, exactly. They're just optimizing for internal consistency rather than for surfacing the thing the model misses.

One partial fix: bring in outside adjudicators for specific move categories, people expert in adjacent domains rather than the core one. For a wargame modeling port logistics under attack, bring in someone who knows supply-chain collapse from commercial disaster response rather than another naval logistician. They will call different things plausible.

graph TD
    A[Opening Turn - No Briefing] --> B{Record Decisions}
    B --> C[Deliver Expert Briefing]
    C --> D[Replay Same Turn]
    D --> E{Compare Delta}
    E --> F[/Identify Primed Assumptions/]
    E --> G[Naïve Actor Moves]
    G --> H{Adjudication Friction Log}
    H --> F

What You're Actually Building

Designing against the competence trap isn't about distrusting experts. It's about building a game environment where expertise has to compete with alternatives rather than simply defining the space of play.

The goal isn't to prove the experts wrong. It's to find the one scenario branch they've been quietly pruning for twenty years, and ask what happens if you let it grow.

That's where the black swans are nesting.