Most wargames are rehearsals. Two teams sit down, the controllers hand out a scenario someone wrote three months ago, and over the next eight hours everyone proves what they already believed coming in the door. The scenario was vetted. The injects were sequenced. The afteraction report writes itself before the lunch break.

Detailed close-up of miniature toy soldiers arranged in a battle formation, showcasing military uniforms and weapons. Photo by Kaique Rocha on Pexels.

That is not the game we're interested in here.

The game we're interested in is the one you cannot rehearse — the one where the central event isn't on the menu, where the most important move comes from a player no one credentialed, where the scoring rubric breaks because the thing that happened wasn't in the rubric. Black swan events do not arrive in formats that wargame designers find convenient. They arrive sideways. They arrive through the supply chain you didn't model, the actor you classified as irrelevant, the second-order effect that the simulation engine averaged away because the variance looked like noise.

Designing for those events is a different craft than designing for the rehearsal kind. It requires giving up the comforts of a clean rubric. It requires injecting genuine anomalies — not just "high-impact, low-probability" injects pulled from an existing taxonomy, but events whose category the players have to construct on the fly. It requires letting the game break, and treating the break as the data.

This site is a notebook on that craft. Posts will look at specific historical wargames (named, dated, sourced), failure modes in current simulation methodology, mechanisms for injecting genuine anomalies into otherwise-structured exercises, and the design tradeoffs between fidelity, surprise, and tractability. Some posts will be about computational models, some about tabletop exercises, some about red-team operations, some about prediction markets and calibration. The throughline is design — how would you actually build this, and what would you do differently next time.

If you build wargames, run simulations, or design exercises that try to take tail events seriously: this is for you. Pull up a chair.

The Game You Can't Rehearse