Game Theory

How to read this page: This article maps the topic from beginner to expert across six levels � Remembering, Understanding, Applying, Analyzing, Evaluating, and Creating. Scan the headings to see the full scope, then read from wherever your knowledge starts to feel uncertain. Learn more about how BloomWiki works ?

Game Theory is the mathematical study of strategic interaction among rational decision-makers. It provides a formal framework for analyzing situations where the outcome for one individual depends not only on their own choices but also on the choices of others. Developed mid-20th century by mathematicians like John von Neumann and John Nash, game theory has become essential in economics, evolutionary biology, political science, and artificial intelligence. Whether modeling a price war between companies, an arms race between nations, or the evolution of altruism in nature, game theory reveals the deep logic of cooperation and competition.

Remembering

• Game Theory — The study of mathematical models of strategic interaction.
• Player — A decision-maker in a game.
• Strategy — A complete plan of action for a player in a game.
• Payoff — The reward or outcome a player receives at the end of a game.
• Nash Equilibrium — A state in which no player can improve their payoff by unilaterally changing their strategy.
• Zero-Sum Game — A situation in which one person's gain is exactly equal to another's loss.
• Prisoner's Dilemma — A standard example of a game that shows why two rational individuals might not cooperate, even if it appears in their best interest to do so.
• Dominant Strategy — A strategy that is better for a player regardless of what the other players do.
• Sequential Game — A game where players take turns (e.g., Chess).
• Simultaneous Game — A game where players make decisions at the same time (e.g., Rock-Paper-Scissors).
• Information Set — What a player knows about the state of the game and others' moves.
• Mixed Strategy — A strategy where a player chooses their action randomly based on certain probabilities.
• Evolutionary Stable Strategy (ESS) — A strategy that, if adopted by a population, cannot be invaded by a rare alternative strategy.
• Mechanism Design — "Reverse game theory": creating rules/incentives to achieve a desired outcome.

Understanding

Game theory shifts the focus from "what is the best for me?" to "what is the best for me, given what you are likely to do?"

The Nash Equilibrium: The most important concept in the field. John Nash proved that in any game with a finite number of players and strategies, there exists at least one equilibrium. In this state, the players are in a "stalemate" where no one wants to move first. This doesn't mean the outcome is good (as seen in the Prisoner's Dilemma), only that it is stable.

Cooperation vs. Defection: Game theory explains why cooperation is hard but possible.

• One-Shot Games: Often lead to defection (cheating) because there is no penalty.
• Iterated Games: If players interact repeatedly, they can use "Tit-for-Tat" strategies—cooperating initially and then mimicking the other player's previous move. This builds trust and punishes cheating.

Common Archetypes:

• The Stag Hunt: A game of coordination. We both catch the deer if we work together; if one of us chases a rabbit, we both go home with less.
• Chicken: A game of brinkmanship. If neither swerves, we both crash. If one swerves, they are the "chicken."
• Matching Pennies: A zero-sum game with no stable pure strategy, requiring mixed strategies (unpredictability).

Applying

Simulating the Prisoner's Dilemma Tournament: <syntaxhighlight lang="python"> def play_round(strat_a, strat_b, history_a, history_b):

   """
   Returns (payoff_a, payoff_b) for a single round.
   'C' = Cooperate, 'D' = Defect
   """
   move_a = strat_a(history_b)
   move_b = strat_b(history_a)
   
   payoffs = {
       ('C', 'C'): (3, 3), # Reward for mutual cooperation
       ('C', 'D'): (0, 5), # Sucker's payoff / Temptation
       ('D', 'C'): (5, 0), # Temptation / Sucker's payoff
       ('D', 'D'): (1, 1)  # Punishment for mutual defection
   }
   return move_a, move_b, payoffs[(move_a, move_b)]

1. Strategies

def always_defect(opponent_history): return 'D' def tit_for_tat(opponent_history):

   if not opponent_history: return 'C'
   return opponent_history[-1]

1. Tournament

history_a, history_b = [], [] total_a, total_b = 0, 0 for _ in range(5):

   ma, mb, (pa, pb) = play_round(always_defect, tit_for_tat, history_a, history_b)
   history_a.append(ma); history_b.append(mb)
   total_a += pa; total_b += pb

print(f"Total Scores - Defector: {total_a}, Tit-for-Tat: {total_b}")

1. Tit-for-Tat is 'nice' but 'provocable'—it forces cooperation in the long run.

</syntaxhighlight>

Areas of Impact

Biology → Explaining why animals don't always fight to the death (Hawk-Dove game).

Business → Pricing strategies in an oligopoly (Coca-Cola vs. Pepsi).

Diplomacy → Designing nuclear treaties and trade agreements.

Computer Science → Analyzing congestion in networks (selfish routing).

Analyzing

Strategic Situations

Game	Incentive	Outcome
Coordination (Stag Hunt)	Mutual benefit	High trust = High reward
Competition (Zero-Sum)	Direct conflict	One wins, one loses
Dilemma (Prisoner's)	Individual temptation	Mutual loss if untrusting
Brinkmanship (Chicken)	Forcing the other to yield	Extreme risk for both

The Tragedy of the Commons: A multi-player prisoner's dilemma. If every farmer grazes one extra cow on the public land, everyone's profit goes up slightly. But if everyone does it, the land is overgrazed and everyone's cows die. Game theory suggests that without regulation or social norms (shaming), rational individuals will inevitably destroy a shared resource.

Evaluating

Evaluating game models: (1) Predictive power: Do real people actually reach the Nash Equilibrium in a lab? (Often they are more cooperative than theory suggests). (2) Computational complexity: Some equilibria are so hard to calculate that it's unlikely real-world agents could find them (P vs NP in game theory). (3) Assumptions of Rationality: How does the model change if we assume players are emotional or have limited processing power? (4) Incentive Alignment: Does the mechanism actually produce the desired social outcome?

Creating

Future of Strategic Logic: (1) Algorithmic Game Theory: Designing auctions (like Google Ads or 5G spectrum) that are resistant to manipulation. (2) Multi-Agent Reinforcement Learning (MARL): Training AI agents to cooperate or compete in complex environments (e.g., self-driving cars merging in traffic). (3) Blockchain Governance: Using game theory (cryptoeconomics) to ensure that a decentralized network remains secure without a central authority. (4) Global Cooperation: Modeling and solving the "Global Commons" problem of climate change.