The solved-AI problem

A scripted competitor is a fixed function of the game state. It may be a complicated function, but it does not learn, does not remember being undercut, and does not adjust to your reputation. Once you know its inputs and outputs, you have solved it, and a solved competitor is a resource to be farmed rather than a rival to be feared.

Designers have tried three patches. Rubber-banding, where the AI cheats harder as you succeed, feels unfair once noticed. Randomness, where the AI behaves erratically, makes planning meaningless. And genuinely adaptive AI is expensive, fragile and, at the end of the day, still a function you can eventually characterize. The tycoon genre's long history of exponential late-game wealth, where the last third of every campaign is spent watching a number grow, is the visible symptom.

What game theory says about markets with real opponents

John von Neumann and Oskar Morgenstern's Theory of Games and Economic Behavior (1944) founded the study of strategic interaction: situations where the best choice for one participant depends on the choices of the others. John Nash's later work defined the equilibrium concept that bears his name, a set of strategies where no player can improve by changing theirs alone.

Two properties of these systems matter for game design. First, in a competitive market with several participants, the equilibrium is often not where any one participant would like it. Everyone would prefer high prices; each has an incentive to undercut; the result is lower prices for all. Second, and more importantly, equilibria in repeated games with many players are unstable in practice, because participants have incomplete information, different goals and different levels of skill. Real markets do not sit at equilibrium; they circle it.

That circling is the source of enduring challenge. A single-player market converges on a solution. A multiplayer market keeps moving because the participants keep learning about each other.

Price wars and the discipline of the neighbor

In a shared market, price is a message. Undercut your rivals and you gain audience this week; you also teach them that undercutting works, and next week they do it back. The equilibrium of a price war is thin margins for everyone, which is why real businesses develop informal norms against starting them and why a game with real opponents develops the same norms without any rule saying so.

What emerges is something a scripted AI cannot produce: reputation. Players remember who undercut them. Alliances and rivalries form. The optimal price is no longer a formula but a judgment about what the cinema across town will tolerate.

Capacity floods and the tragedy of the blockbuster

The second dynamic is capacity. When a big film arrives, every exhibitor wants it on as many screens as possible. If the market's demand for the film is shared, however, each additional screen anyone adds dilutes the audience for all the others. The individually rational move, book it everywhere, is collectively self-defeating: a market full of half-empty screens showing the same film.

This is a textbook commons problem, and the interesting strategy it produces is restraint. The player who correctly forecasts that everyone else will flood the blockbuster and instead gives it two screens while counter-programming the rest can beat the flooders on retained revenue. The correct play depends entirely on what others do, which is exactly the property that keeps a game alive.

The proof already exists: sports management

If the argument sounds theoretical, look at the one management genre that went multiplayer decades ago and never looked back. Online football management games have run shared leagues since the 1990s. Their economies, transfer markets where clubs bid against each other for a finite pool of players, are exactly the shared-market structure described here, and they have sustained active communities for longer than most video game franchises have existed.

The reason is not that football is more interesting than business. It is that a league gives every decision a rival. Your training regime matters because it is compared to someone else's on match day. Your transfer bid matters because another club wanted the same player. The single-player football managers, however sophisticated, have never held players for as long, because the opponent is eventually solved.

A cinema market is a league with a different scoreboard. The films are the fixtures, the box office table is the standings, and the exhibitor across town is the club you most want to finish above.

Imperfect information and the value of watching

Game theorists distinguish between games of perfect information, where every player sees the whole state, and games of imperfect information, where they do not. Markets are the second kind. A player knows their own bookings and prices, sees aggregate results after the fact, and must infer what rivals did from the gap between forecast and outcome.

This is where the shared market produces a skill that scripted competition cannot: reading the room. A public box office table tells you which films drew and roughly how the audience split, but not exactly who booked what. A ranking tells you who is winning, but not how. The player who studies those signals week over week builds a model of their rivals' habits, and that model is the real competitive edge. It is also perishable, because rivals are doing the same thing back.

Designers can tune how much is visible. Show too little and strategy collapses into guessing; show too much and the game becomes a solvable optimization again. The sweet spot is enough public data to reason from and enough hidden detail to be wrong about.

Bidding for scarce things

Shared markets also make scarcity meaningful. In a single-player game, if a premium screen or a marquee title is available, you can have it. In a shared market, so can everyone else, and the allocation has to be decided somehow. Auctions, relationship tiers, first-come windows: each mechanism turns a purchase into a strategic act with an opportunity cost, because the thing you win is a thing a rival did not.

The economies of large online games show what happens when this is done well. CCP Games famously hired an academic economist to monitor and report on EVE Online's player-driven markets, because those markets exhibited real inflation, real speculation and real crashes driven by player behavior. Edward Castronova's research on virtual economies established that such systems obey recognizable economic laws precisely because the agents in them are people.

Fairness without hand-holding

The obvious objection to shared markets is that experienced players will crush newcomers. It is a real risk, and the single-player genre's answer, difficulty levels and rubber-banding, does not translate. What does translate is structure. Finite demand per film means the leaders cannot absorb the whole market. Fixed costs that scale with size mean a large cinema needs a large audience just to break even. Weekly resets of the film slate mean last week's advantage does not carry over automatically. And honest financing, a startup loan with real repayments, lets a newcomer buy capacity without the game secretly subsidizing them.

None of that guarantees a newcomer will win. It guarantees they can compete on judgment, which is the only kind of fairness a strategy game should promise.

Design consequences

Building a shared-market tycoon game imposes constraints that single-player designers never face.

• Synchronous settlement. Everyone's decisions must be resolved together, at a fixed deadline, against the same market state. A weekly tick is the natural cadence.
• Server authority. If outcomes depend on what rivals did, the client cannot compute its own results. The market must settle in one trusted place.
• Bounded demand. The market's total audience must be finite and shared, or capacity floods have no cost.
• Visible rivals. Players need to see enough of what others did, through rankings and box office tables, to form the beliefs that drive strategy.
• Catch-up without rubber-bands. Because real players have real skill gaps, the design needs honest mechanisms, such as loans and cheaper entry points, that let a newcomer compete without the game secretly favoring them.

None of these is free. All of them are worth it, because they buy the one thing a single-player tycoon game cannot have: an opponent who is still learning.