How I'd wargame an uncrewed warship
If you had one day before a navy commits to several dozen uncrewed, container-carrying warships, what would you need to find out?
A hypothetical one-day seminar wargame for the US Navy's Modular Attack Surface Craft, built entirely from public sources. Not a game that was run: a design for one, and the questions it should force.
In short: The US Navy wants cheap, modular, uncrewed craft that carry weapons and sensors in shipping containers. Before building dozens, a one-day game should test three things: what each size is actually for, what the craft is trusted to do when it loses contact, and whether an armed vessel with no crew calms a crisis or escalates it.
Why it’s worth a game
In 2025 the US Navy folded its Large and Medium USV programmes into one: the Modular Attack Surface Craft (MASC). The Navy wants “non-exquisite” hulls built to commercial standards, payloads in standard containers, and vessels fielded quickly.
This is a good point for an exploratory seminar game. The design is still a set of requirements, so changes are cheap, and you don’t need a high-fidelity model of a ship that doesn’t exist yet. The aim is to find the right questions, not to answer them. Answering them is a job for analytical wargaming and modelling later.
Three craft, three questions
The July 2025 solicitation describes three variants. Each one implies a different answer to what is this for?
Single PayloadOne 20ft container
- Cheap enough to lose
- One job, done persistently
- Carry a mixed load without going home
MASCTwo 40ft containers
- Mix a sensor and a weapon
- Operate with a carrier group or alone
- Be cheap enough to ignore
High-CapacityFour 40ft containers
- Magazine depth
- Several missions in one sortie
- Avoid looking like a warship
The first hypothesis I’d want tested is that these aren’t one vessel in three sizes. The small one may earn its place by freeing up crewed ships for harder work. The large one may earn it by replacing a crewed ship on some tasks. If that’s right, they need different baselines, not one baseline scaled up.
The game I’d run
One day, three phases: agree the operating problem, play the vignettes, then turn observations into requirements. Players rotate through roles that map onto the solicitation’s requirements, so each person looks at the craft from more than one angle.
- 01 Overview White cell
The strategic setting for the scenario.
- 02 Action White cell
A Red move, direction from government, and a plan that uses MASC.
- 03 Reaction Players
Each role works out what the craft must do and carry.
- 04 Counter-action Red Team
Red goes after the craft, its links and its controllers.
- 05 Discussion Everyone
Respond, then name what broke.
- ISR & targeting — Persistent watch
- Counter-ISR — Information operations
- Lost link — Inject: comms cut mid-mission
- Carrier group — Aggregated operations
- Distributed strike — Containerised launchers
What I’d expect the game to surface
These are hypotheses to test, not findings.
-
Lost communications will be the key test. The solicitation asks the craft to keep going on its mission if communications fail. Most players will agree with that in principle. The harder question is what “the mission” means once a weapon is aboard and nobody can see what the craft sees.
-
Persistence will matter more than speed. 2,500 nautical miles at 25 knots describes a transit. For surveillance tasks the real requirement is days on station. Congress’s 720-hour reliability test suggests it has the same concern.
-
“Purpose-built unmanned” and “room for eight” will be in tension. Congress wants proof these aren’t crewed designs. The solicitation lists berths for eight people for 14 days as desirable. A game will show which tasks need people aboard, and how often.
-
Command, not the hull, will be the bottleneck. One off-hull station controlling several craft is efficient until two of them need a decision at the same time.
-
Ambiguity helps and hurts. A container could hold a sensor or a missile. Players will want to exploit that for deception. Red will read it as escalation. Congress has already flagged the risk of miscalculation at sea, and a game is a good way to explore it.
AI shouldn’t smooth away the friction. It should surface the human friction as a key finding.
— James Moorhouse
The same applies to the craft. The most useful results are likely to be about who trusts the vessel to do what, and when, rather than about range or payload.
What a one-day game can’t tell you
- It’s built on public requirements, not classified threat data. It can test logic, not performance.
- One day means one pass at each vignette, with no replays. That’s enough to find questions, not settle them.
- Communications will be assumed to work unless an inject breaks them, which is probably optimistic.
- Seminar play shows how people reason. It doesn’t show how the autonomy will behave.
The take-away
MASC’s biggest open questions are less about the hull and more about trust when the link drops, who decides when several craft need a call at once, and how an armed vessel with no crew is read by the other side. All three are human questions, and a day of exploratory play before construction starts is a cheap way to surface them.