Surprise comes from the gap between the plan and the habit

Strategy is how an organisation would like to decide. Behaviour is how it decides on a Tuesday afternoon with the phone ringing. The gap between the two is where surprise comes from.

Strategy Soup measures the behaviour and writes it down as a persona: one file precise enough to teach an AI to reason like that organisation. Then it gives you places to put the persona under pressure. A conversation, a wargame, a scenario. None of it predicts what happens next. It makes your assumptions visible enough to argue with, and the argument is where the cracks show.

Four parts, and the flow runs one way

• The diagnostic reads you. You answer 54 questions in your browser, free, in about 18 minutes. The answers stay in your browser session unless you save the result to the public gallery. It scores 15 Ingredients, the traits it measures: pace, risk appetite, horizon and twelve more. Then it matches the profile to the nearest of 16 Flavours. Take it here.
• The 16 Flavours name the patterns. The Fortress, The Pirate Ship, The Cathedral and thirteen others, in five groups: Conservative, Aggressive, Culture-Led, Systems-Driven and Independent. Each is a recognisable strategic character with its own principles, blind spots and red lines. They are mirrors.
• The wargame plays personas against each other. Up to eight personas in one scenario, each moving in character, round by round. Yours, a competitor profile built from public information, or a starter Flavour.
• The engine runs the published reports. Our reports come from exercised scenarios. Each seat plays from a private brief. An arbiter (the referee that holds no seat) sets the odds on every move, seeded dice decide, and each report says how many times its game ran. Dr Dan Epstein writes the argument. The engine explainer explains the machinery and what it is not.

The flow runs one way. The diagnostic measures, the measurement becomes a persona, and the persona plays. Everything downstream inherits the quality of the first step, so the first step is the one we tested hardest.

What it does not do

It does not forecast. The wargame knows nothing about next year that you do not. It is not a personality test and not a maturity score, because no Flavour is the good one. And it is not a black box you take on faith.

The evidence page says what we tested, what held, what broke, and what we still cannot say. The short version: the personas are distinct enough to tell apart, and they change what agents decide. Whether they are accurate models of real organisations is an open question, and nobody should claim otherwise, including us.

Where to start

• See it first. Read a report in easy mode. The Throat to Choke takes 90 seconds. Then browse the Flavours and find yours.
• Check it before you trust it. Read the engine explainer for the machinery, then the evidence page for what we can and cannot claim.
• Use it. Take the diagnostic, then put the persona to work in a wargame.
• Do the due diligence. The four technical studies sit under the evidence page, starting with the first study, which asks whether a blinded reader can name the character from play alone. Pre-registration documents and run data are available on request through the Long Game Project contact page.

Lost on a word? The glossary defines every term in one place.