The Borrowed Enemy: Designing Wargames That Resist Generic Adversary Models

Every wargame has an enemy. The question worth asking before your next design session is: whose enemy is it, really?

Most wargame red cells are assembled from the same parts bin. You pull the last threat assessment, you staff the opposing team with subject-matter experts who have spent years studying a specific adversary, and you brief everyone on canonical doctrine. The result is an opponent that behaves like a PowerPoint slide. Players recognize the patterns. They adjust. They win. Nobody learns anything.

Call this the borrowed enemy problem. The adversary you import from an existing intelligence product was built to describe past behavior. It is, by construction, backward-looking. Running that model forward through a novel scenario doesn't surface surprises. It confirms the threat brief.

The Global War Game series at the Naval War College, running in various forms since the early 1980s, has wrestled with this problem longer than most. After-action reviews from the mid-1990s iterations noted that experienced players began to recognize red cell move patterns within the first few hours. Once pattern recognition kicked in, the exercise stopped testing strategic judgment and started testing memory. The game became a quiz on the adversary model rather than a stress-test of the players' own decision-making.

So how do you build an enemy that actually resists capture?

Start by separating adversary intent from adversary behavior. Most threat models fuse these two things. They describe what an adversary does, then infer why, then re-encode the why back into the behavioral description until the two are indistinguishable. Your red cell should receive the intent (political goals, red lines, risk tolerance) and then generate behavior independently, without access to the canonical doctrine your blue team has already read.

This is harder than it sounds. Red cell players default to performing the enemy they know. Give them structured latitude instead. Define a handful of non-negotiable constraints (geography, resource limits, political survival requirements) and then let them solve freely within that space.

graph TD
    A[Adversary Intent Brief] --> B{Red Cell Design Session}
    B --> C[Canonical Doctrine Path]
    B --> D[Constraint-Only Path]
    C --> E(Predictable Moves)
    D --> F(Novel Moves)
    F --> G[Adjudicator Review]
    E --> G
    G --> H[Blue Cell Encounter]

The diagram above sketches two paths through the same starting point. Canonical doctrine produces the borrowed enemy. Constraint-only design forces red cell players to reason from the adversary's actual problem rather than from a cultural memory of their tactics.

A second tool: rotate your red cell's information diet. If everyone on the team has been reading the same threat reports, they will unconsciously reproduce the same adversary. Bring in players who have studied structurally similar adversaries from different domains or eras. A red cell member who has spent years studying insurgent finance in West Africa will model resource denial differently than a conventional-forces specialist. The overlap is where interesting moves live.

Third, and most important: build in explicit model-disruption events. These are adjudicator-controlled injections that force the red cell to abandon whatever approach is working. Historical examples are useful here. In the 1983 Able Archer exercise, the Soviet interpretation of NATO signaling produced responses that NATO planners had not anticipated because Soviet observers were applying a threat model that NATO didn't know existed. The mismatch was almost catastrophic. Your design can replicate this asymmetry deliberately, by giving the red cell access to information about blue cell internal debates that blue cell players believe are confidential.

That last move tends to make players uncomfortable. Good. Discomfort in the exercise room is much cheaper than surprise in the field.

One practical caution: resist the temptation to make your red cell so unconstrained that they stop behaving like any real adversary. Creativity without plausibility produces a different failure mode. Players dismiss implausible red moves as designer error rather than genuine possibility, and the learning evaporates. Every exotic move your red cell makes should be traceable, at least in retrospect, to a coherent adversary logic. Brief that logic after the move, not before.

The borrowed enemy is comfortable. It fills the opposing team slot, it satisfies the pre-game brief, and it lets blue cell players feel like they are engaging a real threat. What it doesn't do is find the things your model missed. Building an adversary who can genuinely surprise your players requires treating the red cell design process as its own separate design problem, with as much rigor as anything else in the scenario. Most wargame programs skip this step entirely. That's the gap worth closing.