Wargaming every COA: from annual event to daily practice

Doctrine is unambiguous. Every course of action developed in the estimate should be wargamed against a thinking adversary before the commander is asked to choose. Wargaming is not a supplementary analytical technique; it is the step that converts a set of plausible-sounding options into a comparison a commander can act on. Remove it and the comparison matrix becomes a record of staff preference dressed in analytical language.

Practice diverges sharply. In most headquarters, wargaming happens properly at the annual exercise, where there is time to prepare a red cell, brief participants on adversary doctrine and run adjudicated turns over several days. During routine planning, it collapses into a ninety-minute discussion of the preferred option. The reason is not indiscipline; it is cost. Credible wargaming requires an opposing team that is prepared, independent and available, and assembling one takes weeks that a live planning cycle never contains.

What gets lost when wargaming is an event

Treating wargaming as an annual event produces three predictable failures. The first is shallowness: one or two courses of action are tested against a single, most-likely adversary response, so the most-dangerous case is acknowledged in a risk paragraph rather than played out. The second is asymmetry: the preferred COA receives all the scrutiny, which paradoxically makes it the option whose weaknesses are best documented and therefore the easiest to attack in the decision brief - while genuinely superior alternatives go untested.

The third failure is skill decay. Wargaming is a staff competence, and competence maintained once a year is competence at the level of a once-a-year activity. Planners who wargame weekly develop an instinct for which assumptions carry weight and which adversary responses are worth exploring. Planners who wargame annually spend the first two days of the exercise relearning the mechanics.

What continuous AI-assisted wargaming changes

An AI adversary agent removes the constraint that made wargaming an event: the availability of a prepared opposing team. A red agent grounded in adversary doctrine and observed behaviour is available at 0200 on a Tuesday, does not tire across the fortieth iteration, and applies the same risk posture consistently unless deliberately reconfigured. It can be set to a cautious profile in one run and an aggressive one in the next, generating most-likely and most-dangerous responses against every blue course of action rather than only the favourite.

Machine-speed adjudication changes the economics of breadth. When exploring a branch costs minutes rather than staff-days, a planner stops rationing exploration. Overnight runs can push every COA through multiple red responses across attrition, logistics, movement and timing, and the morning brief opens with results rather than with a plan to produce results.

The most useful change is subtler. When wargaming becomes routine, its output shifts from a verdict to a sensitivity picture. The interesting finding is rarely “COA 2 scores highest.” It is “COA 2 scores highest, and it stops scoring highest the moment the assumption about rail throughput is relaxed by fifteen percent.” That sentence is worth more to a commander than any ranking, and it can only be produced by running the wargame many times with varied inputs.

Making it a habit, not a capability

The transition is organisational as much as technical. Three practices distinguish headquarters that make this stick. They wargame the estimate rather than the decision brief, so results shape options while options are still malleable. They point the adversary agent at standing contingency plans during quiet periods, treating red-teaming as continuous maintenance rather than crisis activity. And they capture every run, so the record of what was tested becomes organisational memory instead of the recollection of whoever was in the room.

Automation does not replace the human red cell where one exists; it makes the human cell more valuable by handling breadth so that people can concentrate on the imaginative, doctrine- breaking moves an agent will not propose. Nor does it decide anything. The commander decides. The platform rehearses, stress-tests and documents.

SynapseWargame™ exists to make that daily practice affordable. Read about auditable AI in the command post or see how commands put it to work.