Strategy Soup How a report gets made

The Long Game Project · The Engine · TLDR

What the engine can and cannot do

A matrix game is a game argued in words, where a referee sets the odds on each move. This engine plays one with opposed AI players, a referee that writes its reasons, and seeded dice, dice whose rolls can be repeated. It can replay a game, and fork it, branching it at one turn. It cannot predict.

Two things come out of this engine. The reports published on this site, where a machine plays the game and a person writes it up. And the tabletop, where your team plays the roles and the engine referees. The reports have shipped. The tabletop is the newer of the two. What each costs is on the commission page.

Two things come out of this engine

The Long Game Project · The Engine

What the engine can and cannot do

Two things come out of this engine: the reports published on this site, where a machine plays the game and a person writes it up, and the tabletop, where your team plays the roles and the engine referees. The reports have shipped. The tabletop is the newer of the two, and the commission page says what each costs. We checked the five things it can do against the code on 18 September 2026. If a report ever claims more than this page allows, this page wins. Tell us through the Long Game Project contact page and we will fix the report.

Two things come out of this engine

Five things it can do

A matrix game: argued in words

An actorClaims a move

A briefed persona, with its own goals, red lines and information.

in writing
The other actorsArgue against it
The refereeSets the odds on each move

No actor grades its own move.

The arbiterThe referee

A language model under its own brief. It plays no role.

One ruling

A targetfor every claimed move
The case for and againstinto the transcript
Its confidencestated with each ruling
The run Replayed Same seed and log: the same run, roll for roll A branch At a chosen beat, one setting changed

One square is one event on the log

Before the branch point: the run as it was

After it: rule-based players and a rule-based referee, logged as new events

Cyan: the compare view marks the first event where the two branches differ

Or rerun historywith different dicewith a shock removed

Before the game runs, a research pass builds

The evidence pack
Real-world claimIts sourceVerdict

Cyan: each claim checked against its source at retrieval

Personas cite itScenario numbers cite itPublished with the report

A reader of a published report gets

The evidence pack
A seams section

Where the expert who signs the report and the machine disagreed.

Who played, and how often

For every report, in one table on Methods.

A run is one path through an argument

The value sits in

The assumptions it drags into the open
The knock-on effects nobody had written down
Your market
Public sources, the research pass
The documents behind the actor
The actor

What its evidence pack leaves out

A thin brief
A thin actor
A well-briefed actor
Now and thenSomething no board would do
The recordclaimrulingrollthe misfire, kept
The reportthe seams section says so
The published reportsA machine plays the game

A person writes it up.

The tabletopYour own people at the table

With the engine as the referee.

The strongest version of this method
The same engine

Each report says how many times it ran

One square is one run

Re-running with only the dice changed tells youWhich outcomes are fragile to the dice
It does not tell youHow likely anything is

A separate question

This pageWhat the engine can and cannot do
The evidence pageWhether the players are any goodOpen ›
On the logthe claimthe objectionsthe rulingthe die facethe outcome
  • Play actors that want different things
  • Rule with written reasons
  • Replay any run and fork it
  • Cite checked evidence
  • Show its working

Cyan: checked against the code on 18 September 2026

  • Predict
  • Know what its evidence pack leaves out
  • Play a persona better than its brief
  • Replace a room full of your own people
  • Tell you how likely anything is
  • Actor-play matrix gameMost published reports
  • Counterfactual fork
  • Market or auction game
  • Red-team single actor
  • Human tabletop with the engine as refereeThe newer of our two products

One beat

One beat of a game, a single turn, in order. Every beat leaves the same five marks on the log: the claim, the objections, the ruling, the die face and the outcome.

Can

Play actors, AI players, that want different things. Rule with written reasons. Replay any run, and fork it at a chosen turn. After a fork, fixed rules play on. Cite checked evidence. Show its working.

Cannot

Predict. Know what its evidence pack, the table of checked facts it plays from, leaves out. Play a persona, a written character, better than its brief. Replace a room full of your own people. Tell you how likely anything is.

Formats

Five. Actor-play matrix game, counterfactual fork, market or auction game, red-team single actor, and a human tabletop with the engine as referee. The first is the format of most published reports. The tabletop is the newer of our two products.

Five things it can do

Run matrix games with opposed actors.

A matrix game is a game argued in words, where a referee sets the odds on each move. Each actor is an AI agent playing a briefed persona, a written character with its own goals, red lines and information. The actors argue against each other’s moves, in writing. No actor grades its own move.

Rule with reasons.

The arbiter is the referee, a language model under its own brief, and it plays no role. It sets a target for every claimed move and writes the case for and against into the transcript. It states its confidence in each ruling. The evidence page shows that AI players overstate their own odds by 14 to 28 points.

Replay and fork.

Give the engine the same seed and log and it replays the same run, roll for roll, with every ruling read back as it was made. The seed is a starting number that fixes the dice. We can rerun history with different dice, remove a shock, an outside event, and watch what changes, or branch a finished game at a chosen beat, a single turn, with one setting changed. After the branch point the game continues with rule-based players and a rule-based referee, logged as new events. AI players and the AI referee do not replay a branch yet. The compare view marks the first event where the two branches differ.

Ground play in checked evidence.

Before the game runs, a research pass builds an evidence pack: a table of real-world claims, each one checked against its source at retrieval and given a verdict. Personas and scenario numbers cite it, and the pack is published with the report.

Show its working.

A reader of a published report gets the evidence pack and a seams section that lists where the expert who signs it and the machine disagreed. Methods states, for every report, how much of the game a machine played and how many times it ran.

Five things it cannot do

Predict.

A run is one path through an argument, and dice decide part of it. The value sits in the assumptions it drags into the open and the knock-on effects nobody had written down. Read a run for its ending and you have read it wrong.

Know what its evidence pack leaves out.

The research pass uses public sources. An actor cannot know more about your market than the documents behind the actor.

Play a persona better than its briefing.

The actors are language models in character. A thin brief makes a thin actor. Now and then a well-briefed one still does something no board would do. When that happens, the seams section says so. The misfires stay in the record.

Replace a room full of your own people.

The strongest version of this method puts your own people at the table, with the engine as the referee. That is the tabletop. It is a different product from the published reports, and it runs on the same engine.

Reach statistical significance about the world.

That is the bar a result clears before you can say it is more than chance. Each report says how many times it ran. Report 001 ran once, with a person refereeing. Report 002 ran once with AI players and an AI referee. A fixed-rules version then ran one hundred times, with scripted players and no AI referee, and only the random draws changed. Reports 005 and 006 ran forty-five games each. Re-running the same game with only the dice changed tells you which outcomes are fragile to the dice. It does not tell you how likely anything is.

Whether the players are any good is a separate question. The only page allowed to answer it is the evidence page.

New in the engine: a board

It can now also

Keep a board of who has what, where, and who controls each place, across layers such as sea, opinion, markets and forums. Deny a place to a rival without taking it. Hide an asset until another player’s sensor finds it. Run orders day by day inside a turn. Use up stocks such as fuel, munitions or cash. End a run at an escalation line. Let players make deals, promises and threats before they move.

It still does not

Fight battles. There are no combat tables. A clash is one stated chance, and enough clashes in a short time end the run, recorded as escalated to conflict.

The board is built, tested and in the engine since 22 October 2026, and no published report uses it yet. The first will be a second version of report 011. How it works, on Methods, and the engine changelog.

Five formats

FormatBuilt for
Actor-play matrix game
Several actors, one strategic question, played out over years
Counterfactual fork run
One decision, several branched worlds, compare the endings
Market / auction game
Pricing, allocation, and positional goods, the ones prized because others cannot have them
Red-team single actor
One plan, pressure-tested against an environment that plays to break it
Human tabletop with the engine as referee
Your team plays the roles, the engine referees. The newer of our two products, after the reports

Most reports published so far are actor-play games, written as if from the future looking back. Methods shows which kind of game each one is, and who played it. The beat-by-beat mechanics are in the engine explainer.

The annex · Deep

One beat, step by step

  1. The model arguesA player proposes a claim and states its odds
  2. The model arguesEvery other player argues against, in writing
  3. The model rulesArbiter sets a target from the arguments
  4. No model touches itA die rolls against the target
  5. On the logClaim, objections, ruling, die face and outcome go on the log

One beat, in the order it happens. The model argues and rules. The die, which no model touches, decides.

What one ruling records

  • The decision.
  • The reasoning path.
  • The assumptions the ruling rests on.
  • A self-reported confidence.

The engine never presents a ruling as certain. The evidence page measures a related bias: AI players overstate their own odds by 14 to 28 points.

The arbiter runs in one of four modes

  • Rule at once and log the reasoning.
  • Settle routine calls itself and halt for confirmation on major ones.
  • Propose every ruling for a human to approve.
  • A human rules while the machine assists.

Each report states which mode its exercise ran in.

The engine stores events, not state. To get the state at any point, it replays the log up to there. That is why a what-if is cheap: copy the log to a point, change one thing, replay from there. It is also why you can audit a run beat by beat instead of taking a summary on trust.

One square is one event on the log

The one thing changed

Replayed from there