The game before the company.

01 · a plan is not a game

There is a difference between a decision and a game, and most venture failures I have studied live in that difference. A decision is a choice against nature: you pick, the world pays out, and the world does not care what you picked. A game is a choice against players: the payout of your move depends on their moves, and they are choosing with the same self-interest you are. Weather is a decision. Entering a market is a game, because the incumbent gets a vote on your margins.¹

The organizing idea is the equilibrium, formalized by John Nash in 1950: a set of strategies is in equilibrium when no player can do better by changing their move, given what everyone else is doing.² Said plainly, a venture design is only sound if it still works after everyone else responds in their own interest. A plan that requires competitors to nap, customers to overpay, or partners to act against their incentives is not a strategy. It is a wish with a spreadsheet attached.

02 · five concepts that earn their keep

The literature is vast, but venture design mostly draws on five results. Best response: before committing to a move, write down each player’s most profitable reply to it, because that reply is the world your venture will actually live in. Commitment: Thomas Schelling showed that visibly limiting your own options can change what opponents do, which is why sunk investments, signed partners, and public promises are strategic instruments rather than mere expenses.³ Avinash Dixit extended this to market entry: incumbents fight or accommodate depending on payoffs that your design can change in advance.⁴

Added value: Adam Brandenburger and Harborne Stuart proved a result every founder should have framed on the wall, that in a competitive game you can capture at most what the game would lose without you.⁵ Not what you create, what would be lost. Coordination: markets with network effects are games where everyone wants to join only if others do, and such markets tip rather than split, which is why two-sided ventures subsidize one side to start the cascade.^6,7 And mechanism design, the field’s reverse gear: instead of analyzing a game you are given, you design the rules so that self-interested players produce the outcome you want.⁸ Every equity split, pricing page, and marketplace policy is a mechanism, whether or not it was designed as one.

03 · how I use it when designing a venture

The discipline starts before the product does. The first deliverable in any venture design I run is a player map, not a feature list: who gains, who loses, who can retaliate, who must cooperate, and what each of them would most profitably do the day you launch. It is remarkable how many venture ideas die honorably at this stage, and how cheap that death is compared to the funded version of the same discovery two years later.

Then come three design moves, in order. The added-value test asks what the game would lose without you, because that number, not your revenue projection, is the ceiling on what you can capture; if adequate substitutes exist, the ceiling is near zero no matter how good the product feels. Commitment design decides which options to visibly close: partners signed before launch, capacity built in advance, exclusivities accepted. In the Foundry, recruiting three to five design partners before a spinout exists is not just distribution, it is a commitment device that changes the incumbent’s arithmetic, as the figure above shows. Mechanism design writes the rules so the venture holds together without supervision. The Foundry’s equity structure is a worked example: the founder holds the largest stake so the company is fundable, the client parent holds a minority without veto so it cannot suffocate what it seeded, and my stake sits on top of cash fees rather than replacing them so my incentive to do excellent work never competes with my incentive to hold equity. Each rule exists because some spinout died where it was missing.

The last move is the red team, and this is where my agents earn their keep, applied exactly the way Field Note 01 describes: the science lives in the agent’s standing instructions, its worked examples, and its evaluations. A venture-design agent trained this way runs the best-response analysis relentlessly, player by player, and refuses to score a plan viable while any player has a profitable deviation that kills it.

venture_design_rules:
  player_map_first: >
    No product work until every player is listed with their
    payoffs and their most profitable response to our launch.
    A plan that survives only if a player naps is rejected.
  added_value_test: >
    State what the game loses without us, in dollars. That is
    the capture ceiling. If substitutes put it near zero, the
    venture is redesigned for scarcity or killed at the screen.
  mechanism_audit: >
    Every rule we control (equity, pricing, exclusivity, data
    rights) is restated as an incentive: who does this rule pay,
    to do what? Rules that pay someone to hurt the venture fail.

04 · what it returns

The return is mostly ventures that never get built, and I mean that as the benefit. The expensive way to learn that an incumbent will cut prices, that a platform will clone you, or that your equity split makes the founder ungovernable is to fund the venture and watch. The cheap way is an afternoon with a payoff matrix. When a design does pass, it passes differently: it launches with commitments already made, a capture ceiling already measured, and rules that pay every player to keep the venture alive. In the Foundry, this analysis is part of the screen itself, sitting beside the profit math from the build: the problem tells us whether a venture is possible, and the game tells us whether it is defensible.

There is also a pricing dividend, and it connects this note to the first one. Neuroeconomics tells you how a buyer’s brain will read your price; game theory tells you how your competitors will answer it. A price that survives both readings, the psychological one and the strategic one, is rare, and it is usually not the price the spreadsheet suggested.

05 · the honest limits

Three cautions, as always. Game theory assumes rational, self-interested players, and fifty years of behavioral evidence, including everything in Field Note 01, says real players deviate in predictable ways: they reject unfair deals that would profit them, they overweight losses, and they anchor on histories the model ignores, so I run every game twice, once with rational players and once with human ones, and take the design that survives both. Equilibrium analysis is a snapshot, while markets move, technologies shift, and yesterday’s commitment can become tomorrow’s anchor chain, so the player map gets revisited at every stage gate rather than framed and forgotten. And the numbers in any payoff matrix are estimates dressed in precision; the discipline is in being forced to write them down and defend them, not in the decimals. A payoff matrix you cannot defend line by line is a story, not a system, and the whole point of this series is to build systems.