The Long Game ProjectReports
Exercised Scenario Reports
Decision-rehearsal instruments, not forecasts. Each report is a scenario we have actually exercised - the moves, the breakpoints, and the assumptions that failed first.
Every report opens the same way: this is an exercise built to surface assumptions, not a prediction to be graded. We play a plausible future forward through a structured wargame, publish what broke and where, and keep a changelog as reality agrees or disagrees with us. Where the exercise was wrong is not the embarrassment - it is the finding.
We publish rarely, and heavily.
Published
The reports are published on Strategy Soup, our studio for wargames with AI players. Each opens on a short version and goes deeper from there, down to the record of the game.
- 001The Throat to ChokeWhat AI does to the people who sell thinking, 2026-2030
- 002The Seat PriceWhat AI agents do to software sold by the licence, 2026-2030
- 005The Silence in the DoctrineWho makes the US Treasury admit it is managing long-term interest rates?
- 006Written by the AdversaryHow much can an AI system's own account of itself be allowed to prove?
- 007Measuring the Strategic Stance of AI ModelsDoes an AI model have a stable strategic stance, and does it show in what the model chooses?
- 008The Quiet Spec ChangeIf a senior insider changed what an AI model is trained to want, would the lab's own published plan catch it?
Subscribers get the next report the day it ships.
The private version
The same instrument, pointed at your decision.
These reports are the public form of the method we run privately as the Decision War Room: a counterfactual wargame of the call you are about to make, delivered as a written report and a board-ready readout.
About the Decision War Room →Weighing a decision with an AI agent?
Paste this into Claude, ChatGPT or any agent and it will help you judge which format fits your decision. Only paste into a tool your organisation has approved.