The Loyal Opposition: Designing Wargame Red Teams That Don't Secretly Want to Lose

Most red teams lose on purpose, and nobody admits it.

Photo by Saifee Art on Pexels.

Not through deliberate sabotage. The failure is subtler than that. Red Cell players are usually drawn from the same institution running the exercise. They share professional norms with Blue Cell. They understand, implicitly, that the purpose of the event is Blue's development, not Red's victory. So they probe without biting. They simulate pressure without applying it. And Blue walks away having confirmed its existing playbook, which was probably the point all along.

This is one of the most reliable failure modes in exercise design, and it has a name in the red-teaming literature: role capture. The adversary becomes a training aid rather than an adversary.

Designing your way out of it is harder than it looks.

Why Red Cells Go Soft

The social pressure is enormous. Red Cell players often rotate back into Blue-side jobs afterward. Embarrassing the organization's leadership, even inside a simulation, carries real professional risk. Add to that the common practice of pre-briefing Red Cell on Blue's intended response (ostensibly for realism), and you've handed the adversary a map to their own defeat.

There's also a design problem hiding inside the incentive problem: most wargame scoring systems measure Blue performance, not Red performance. Red has no explicit win condition. Without one, players optimize for plausibility over aggression.

The RAND Corporation documented a version of this in their post-exercise analyses of multiple Cold War-era NATO command-post exercises. Red would consistently pull punches on logistics interdiction because disrupting Blue's command infrastructure made the exercise unplayable, and unplayable exercises got cancelled. Red learned to be disruptive enough to be realistic, not enough to be dangerous.

Design Intervention 1: Give Red an Explicit Victory Condition

This sounds obvious. It isn't, in practice.

A victory condition isn't "inflict maximum casualties", that's a default, and defaults get gamed toward the path of least resistance. A real Red victory condition names a specific state of the world that is strategically meaningful and that Blue is structurally unprepared to prevent.

For a logistics-focused exercise, Red's win might be: delay Blue's operational consolidation by 72 hours without triggering escalation protocols. That constraint forces Red to be surgical, which is actually more realistic than brute-force attrition, and it gives Red players something to plan toward.

Write the victory condition before you design the scenario. If you can't articulate what Red winning looks like, your Red Cell will invent a definition during the exercise, and it will be the comfortable one.

Design Intervention 2: Separate Red Cell from Blue Cell Socially and Informationally

Red and Blue should not share a planning room, a facilitator, or pre-exercise briefings. This is operationally obvious and logistically inconvenient, which is why it gets skipped.

When USSOCOM ran its internal red-team development program in the mid-2000s, one of the core principles was information asymmetry by design: Red received only what a realistic adversary intelligence picture would provide, not the clean, consolidated Blue order of battle. This forced Red to plan under uncertainty and stopped them from reverse-engineering Blue's expected response.

You can implement a version of this even in small tabletop exercises. Give Red a separate pre-read packet. Run their planning session in a different room. Don't let them watch the Blue Cell brief.

Design Intervention 3: Recruit Outsiders for Red Cell

The most reliable fix is structural: staff Red Cell with people who have no professional stake in Blue's success. Subject matter experts from outside the organization, academic specialists, contractors without ongoing relationships, anyone whose career doesn't depend on the exercise going smoothly for leadership.

This is uncomfortable. Outsiders ask rude questions. They don't know which assumptions are sacred. That's exactly the point.

Here's a rough decision flow for staffing a Red Cell that will actually threaten your assumptions:

graph TD
    A[Define Red Victory Condition] --> B{Is Red staffed internally?}
    B -->|Yes| C[Add at least one external SME with no org loyalty]
    B -->|No| D[Brief Red separately, no shared planning space]
    C --> D
    D --> E{Does Red have access to Blue's full order of battle?}
    E -->|Yes| F[Restrict to adversary-realistic intelligence only]
    E -->|No| G[Run Red planning session independently]
    F --> G
    G --> H[Score Red performance explicitly post-exercise]

The Thing Nobody Wants to Hear

If your red team never wins, not once, across multiple iterations, your exercise is broken. Not because Blue is genuinely that good, but because something in the design is protecting Blue from the result.

The whole point of red-teaming is to find the scenario where Blue's playbook fails before that scenario finds you in the real world. A Red Cell that loses gracefully every time isn't an adversary. It's a sparring partner who pulls every punch.

Design for the uncomfortable outcome. Build in the conditions that let Red actually win. Then watch what Blue does when the comfortable ending isn't available.

That's the exercise worth running.