All About Game Theory

Introduction to Game Theory

Overview of Utility Theory

• Utility theory examines how rational agents make decisions based on their preferences over outcomes and available action-options.

• Rational agents are characterized by their ability to correlate outcomes with action-options, leading to preferences that guide their choices.

• Chapters 2 and 3 focus on individual optimization, while Chapter 4 introduces interactions between rational agents, emphasizing efficient transactions.

• The concept of Pareto-superior moves is introduced, highlighting scenarios where agents agree on better outcomes without conflict.

• The transition from individual optimization to strategic interaction marks the shift to game theory, where outcomes depend on the actions of multiple agents.

Strategic Rationality

• Strategic rationality involves anticipating the actions of other agents, contrasting with parametric rationality focused solely on nature's uncertainties.

• Decision theory traditionally addresses optimization against nature, while game theory extends this to include interactions with other rational agents.

• The example of Robinson Crusoe illustrates the shift from individual optimization to strategic decision-making upon encountering another agent, Friday.

• Crusoe's initial survival choices are based on environmental conditions, but with Friday, he must consider joint optimization and coordination.

• The theory of games provides a framework for analyzing optimal choices in strategic contexts, developed by John von Neumann and Oskar Morgenstern.

Overview of Game Theory

• Game theory is a mathematical framework for analyzing strategic interactions among rational decision-makers.

• It involves players making decisions that will affect their outcomes based on the choices of others.

• The fundamental assumption is that all players are rational and aim to maximize their utility.

Key Terminology

• Players: Individuals or entities making decisions in the game.

• Strategies: The possible actions available to players.

• Payoffs: The outcomes associated with each combination of strategies, often represented in utility values.

Historical Context and Development of Game Theory

Key Contributors

• John von Neumann (1903-1957) and Oskar Morgenstern (1902-1977) are pivotal figures in the development of game theory, integrating utility functions into strategic analysis.

• Their work was initially titled 'The General Theory of Rational Behavior' but was published as 'Theory of Games and Economic Behavior.'

• Von Neumann's contributions include the mathematical foundations of game theory, while Morgenstern focused on its economic implications.

• The framework they developed applies to various strategic scenarios, including conflict and coordination problems.

• Their analysis of games, such as poker, illustrates the complexities of strategic decision-making.

The Holmes-Moriarty Problem

• The Holmes-Moriarty problem serves as a classic example of strategic decision-making in game theory, highlighting the complexities of anticipating an opponent's moves.

• Sherlock Holmes faces a strategic dilemma against Professor Moriarty, who is equally intelligent and capable of predicting Holmes's actions.

• The problem illustrates the concept of mixed strategies, where agents randomize their actions to remain unpredictable to their opponents.

• Morgenstern noted that resolving such strategic dilemmas requires arbitrary decisions rather than straightforward outsmarting.

• The matrix representation of the problem helps visualize the potential outcomes based on the actions of both players.

Game Theory Framework

Basic Components of Game Theory

• Game theory consists of agents, actions, states of affairs, and outcomes, similar to decision theory but with added complexity due to multiple agents.

• Agents' decisions are influenced by their preferences over outcomes, which are affected by the actions of other agents.

• The strategic interaction among agents leads to a variety of possible outcomes, depending on their choices and the choices of others.

• Understanding these interactions is crucial for analyzing competitive and cooperative scenarios in economics and beyond.

• Game theory provides tools for predicting behavior in strategic situations, making it applicable in various fields, including economics, political science, and biology.

Mixed Strategies in Game Theory

• A mixed strategy involves players randomizing their actions according to a probability distribution, making it difficult for opponents to predict their moves.

• This approach is essential in scenarios where players have conflicting interests, as it allows for optimal decision-making under uncertainty.

• The concept of mixed strategies was highlighted in the Holmes-Moriarty problem, where both players must consider the unpredictability of the other's actions.

• The use of mixed strategies can lead to Nash equilibria, where no player has an incentive to unilaterally change their strategy given the strategy of others.

• Understanding mixed strategies is vital for analyzing competitive environments and developing effective strategies in real-world applications.

The Holmes-Moriarty Game

Game Representation

• The game is represented in a matrix format where rows correspond to Moriarty's actions and columns to Holmes's actions.

• Each cell in the matrix indicates the payoffs for both players, with the first number for the row player (Moriarty) and the second for the column player (Holmes).

• Example payoff matrix:

Strategic Implications

• The best outcome for Moriarty is to kill Holmes, resulting in a payoff of 100 for Moriarty and -100 for Holmes.

• If Holmes escapes to Dover while Moriarty is in Canterbury, Holmes receives a payoff of 50, while Moriarty gets -50.

• The game illustrates the concept of utility maximization, where both players aim for the best possible outcome.

Types of Games

Zero-Sum Games

• A zero-sum game is one where the total gains and losses among players equal zero; one player's gain is another's loss.

• Poker is a classic example of a zero-sum game, where winnings come directly from other players.

• The strategic form of a zero-sum game can be represented in a payoff matrix, showing the interactions between players.

Maximin and Minimax Solutions

• The maximin strategy for a player is the strategy that maximizes the minimum payoff they can receive.

• The minimax strategy minimizes the maximum loss a player can incur.

• The minimax theorem states that in any finite two-person zero-sum game, there exists at least one minimax solution.

Extensive Form Games

Cake-Cutting Game Example

• In this game, one player (Alf) cuts the cake, and the other player (Betty) chooses a piece first.

• The game can be represented in extensive form, showing the sequential nature of the decisions.

• Alf's utility is represented at terminal nodes, with higher values indicating better outcomes for him.

Strategic Choices in Extensive Form

• Players must consider the actions of their opponents when making decisions, leading to strategic interdependence.

• Betty's choices are influenced by how Alf cuts the cake, demonstrating the concept of rational decision-making in game theory.

• The outcomes are interdependent, meaning that the best choice for one player is often the worst for the other.

Zero-Sum Games and the Cake-Cutting Problem

Terminal Nodes and Utility

• Terminal nodes represent possible outcomes in a game, with associated ordinal utility values for players.

• Alf's utility values range from 1 (worst) to 4 (best), indicating his preferences for different outcomes.

• The game is structured such that Alf and Betty's utilities are inversely related, typical of zero-sum games.

• Example: If Alf receives a utility of 4, Betty receives a utility of 1, demonstrating the competitive nature of their choices.

• The strategic decisions made by Alf and Betty are influenced by their understanding of each other's rationality and preferences.

• The outcome of the game is determined by the strategies chosen by both players, leading to a solution that maximizes utility for Alf.

Strategies in Normal Form

• Normal form representation of the game simplifies the analysis of strategies and outcomes.

• Alf's first move influences Betty's available strategies, as she can condition her response based on his actions.

• Betty has four distinct strategies based on Alf's cutting method, which she will choose to maximize her utility.

• The best response strategies for Betty are contingent on Alf's actions, showcasing the dynamic nature of their interaction.

• Example strategies include 'Always choose big' or 'Choose small when Alf cuts equal, choose big when he cuts big and small.'

• The equilibrium strategy for Alf is to cut almost evenly, leading to a stable outcome.

Visual Representation of the Game

Terminal Nodes: 
- (1) Betty gets big piece (Alf's worst outcome)
- (2) Betty gets slightly bigger piece (Alf's second-best outcome)
- (3) Betty gets slightly smaller piece (Alf's third-best outcome)
- (4) Betty gets smallest piece (Alf's best outcome)

The Prisoner's Dilemma

Overview of the Dilemma

• The Prisoner's Dilemma illustrates a situation where two rational agents face a choice that leads to suboptimal outcomes.

• Alf and Betty are suspects with the option to either confess or remain silent, impacting their prison sentences.

• The game is structured such that mutual cooperation (both staying silent) leads to a better outcome than mutual betrayal (both confessing).

• The dilemma arises from the incentives to betray each other, leading to a worse outcome for both.

• Historical context: Developed by Merrill Flood and Melvin Dresher, popularized by Albert Tucker.

• The dilemma serves as a foundational example in game theory, highlighting the conflict between individual rationality and collective benefit.

Payoff Matrix Analysis

• The matrix illustrates the outcomes based on the choices made by Alf and Betty.

• Each cell represents the prison sentences for both players, with lower numbers indicating better outcomes.

• The dominant strategy for both players is to confess, despite the fact that mutual silence would yield a better outcome.

• The equilibrium outcome (10, 10) is Pareto-inferior to the non-equilibrium outcome (2, 2).

• This paradox highlights the tension between individual incentives and collective rationality.

• The game exemplifies how rational decision-making can lead to suboptimal group outcomes.

Dominant Strategies and Equilibrium

• A strictly dominant strategy is the best response regardless of the opponent's choice, while a weakly dominant strategy is no worse than any other response.

• In the Prisoner's Dilemma, both Alf and Betty have confessing as their strictly dominant strategy.

• The equilibrium outcome is reached when both players confess, resulting in a suboptimal outcome for both.

• The concept of dominant strategies is crucial in understanding strategic interactions in game theory.

• The dilemma illustrates the challenges of achieving cooperation in competitive environments.

• The equilibrium outcome is unique in this game, contrasting with many games that have multiple equilibria.

Overview of the Prisoner's Dilemma

Definition and Structure

• The Prisoner's Dilemma (PD) is a standard example in game theory illustrating a situation where two players must choose between cooperation and betrayal (snitching).

• The dilemma arises when both players, acting in their self-interest, choose to betray each other, leading to a worse outcome for both.

• The game is structured such that mutual cooperation yields a better outcome than mutual betrayal, yet rational players often choose betrayal.

• The outcomes can be represented in a payoff matrix, highlighting the conflict between individual rationality and collective benefit.

Payoff Matrix Representation

• The outcomes of the Prisoner's Dilemma can be summarized in a payoff matrix, as shown in Table 5.5.

• The matrix illustrates the preferences of each player (Alf and Betty) based on their choices: Keep Quiet or Confess.

• The Pareto-dominant strategy is to Keep Quiet, but it is not the rational choice when considering the other player's potential actions.

• The Nash equilibrium occurs when both players confess, leading to a suboptimal outcome for both.

Nash Equilibrium Explained

• A Nash equilibrium is reached when no player can benefit by changing their strategy while the other player's strategy remains unchanged.

• In the context of the PD, the strategy {Confess, Confess} is the Nash equilibrium, as both players rationally choose to betray each other.

• John Nash's contribution to game theory established that all finite competitive games have at least one Nash equilibrium.

• The Nash equilibrium serves as a solution concept, guiding rational players towards their best responses.

Utility Functions and Rationality

Ordinal vs. Cardinal Utility

• Ordinal utility ranks preferences without measuring the magnitude of differences between them, while cardinal utility quantifies these differences.

• The transition from ordinal to cardinal utility allows for a more nuanced understanding of player preferences in the PD.

• Table 5.6 presents a general form of the PD using cardinal utility, illustrating how different outcomes can be valued numerically.

• The specific example in Table 5.7 shows how rational players can end up with lower utility than if they had cooperated.

Rationality and Efficiency

• Rationality in game theory often leads to outcomes that are not Pareto-efficient, as seen in the PD.

• The relationship between rationality and efficiency is complex; rational actions do not always lead to optimal collective outcomes.

• The concept of the 'invisible hand' suggests that voluntary interactions can lead to Pareto-efficient results, but this does not hold in the PD.

• The PD exemplifies social dilemmas where individual rationality leads to collective inefficiency, raising concerns for political philosophy.

Philosophical Implications of the Prisoner's Dilemma

Insights from Hobbes

• The PD is often used to illustrate the ideas of Thomas Hobbes regarding human nature and social contracts.

• Hobbes posited that in the absence of common rules, individuals would be in a constant state of conflict, akin to a deadly PD.

• The equality of individuals in their capacity for violence leads to a competitive environment where cooperation is difficult.

• The PD serves as a model for understanding the necessity of moral and political rules to avoid chaos and promote cooperation.

Broader Applications of the PD

• The PD framework can be applied to various real-life situations, including environmental issues, political negotiations, and economic interactions.

• Understanding the PD helps in analyzing how self-interested actions can lead to suboptimal outcomes in society.

• The dilemma highlights the importance of cooperation and trust in achieving better collective results.

• Contemporary philosophers utilize the PD to explore the tension between self-interest and moral obligations in social contexts.

Hobbes's Deadly Prisoner's Dilemma

Overview of the Dilemma

Hobbes's analysis of human nature suggests that in a state of nature, individuals are driven by self-interest, leading to conflict.

• The dilemma is illustrated through the example of two individuals, Alf and Betty, who face choices that can lead to cooperation or conflict.

• If both choose to attack, they enter a state of war, resulting in mutual destruction and loss of societal benefits.

• The consequences of war include a lack of industry, culture, and societal progress, leading to a 'solitary, poor, nasty, brutish and short' life.

• The need for a common moral code is emphasized to avoid the state of war, as individuals cannot rely solely on personal morality.

• Hobbes argues for a powerful government to enforce cooperation and prevent the breakdown of societal order.

The Role of Morality in the Dilemma

• Different moral codes influence individual decisions in the Prisoner's Dilemma, affecting outcomes.

• Individuals with a strict moral code against attacking may become victims, while those with aggressive codes may fare better in conflict.

• The necessity of a common morality is highlighted, as individuals must trust that others will also adhere to cooperative behavior.

• Even with a shared moral code, the temptation to defect remains, as individuals may prioritize personal gain over collective well-being.

• Hobbes's perspective suggests that rational individuals will still defect unless a governing authority enforces cooperation.

• The concept of a 'common power' is introduced as essential for maintaining order and preventing conflict.

Case Studies and Examples

• The example of two ideologically opposed countries illustrates a broader application of the Prisoner's Dilemma.

• Each country believes the other poses a threat, leading to a situation where both prefer peace but may resort to war.

• The dilemma is represented in a table format, showing the outcomes based on their choices to attack or not.

• The analysis of public goods highlights why individuals may not voluntarily contribute, even when they desire the good.

• A multi-person Prisoner's Dilemma is presented, demonstrating how individual rationality leads to under-provision of public goods.

• The need for collective action and cooperation is emphasized to secure public goods.

Coordination Games and Their Implications

Understanding Coordination Games

• Coordination games differ from zero-sum games, as they involve scenarios where players benefit from aligning their strategies.

• The 'Game of Chicken' is introduced as a classic example of anti-coordination, where players must decide whether to swerve or not.

• The outcomes of the Game of Chicken illustrate the risks of mutual defection versus the benefits of cooperation.

• The assurance game is another example, where players must decide whether to build a fence to protect their respective properties.

• The strategic representation of these games highlights the importance of mutual cooperation for optimal outcomes.

• The concept of anti-coordination is explored, showing how individuals may prefer to avoid conflict while achieving their goals.

Practical Applications of Game Theory

• Game theory provides insights into real-world scenarios, such as international relations and public policy.

• The analysis of public goods can inform strategies for funding and providing essential services.

• Understanding the dynamics of coordination games can help in designing effective agreements and treaties.

• The implications of Hobbes's theories can be applied to modern governance and the establishment of laws.

• The study of moral codes and their impact on decision-making can inform ethical frameworks in society.

• Game theory serves as a foundational tool for analyzing strategic interactions in various fields, including economics, political science, and sociology.

Assurance Games and Their Implications

Overview of Assurance Games

• Assurance games are strategic interactions where players benefit from cooperating, but individual incentives may lead to non-cooperation.

• The classic example involves two players deciding whether to build a fence to protect their resources, leading to mutual benefits if both cooperate.

• The payoff matrix illustrates the outcomes based on their decisions, with the best outcome being mutual cooperation.

The Game of Chicken

• The game of Chicken models situations where two players face a conflict and must choose between cooperation and defection.

• In the context of the movie 'Rebel Without a Cause', the game illustrates the tension between two drivers heading towards each other, where swerving signifies 'chickening out'.

• The Nash equilibrium in Chicken occurs when one player swerves while the other does not, leading to a strategic advantage for the non-swirling player.

Case Study: The Cuban Missile Crisis

• The Cuban Missile Crisis serves as a historical example of a Chicken game, where the U.S. and USSR faced off over missile installations in Cuba.

• The U.S. had to decide between an air strike or a blockade, with the potential for nuclear war looming if miscalculated.

• The outcome was a blockade that led to the USSR withdrawing missiles, illustrating a non-equilibrium solution in the Chicken game framework.

Limitations of Game Theory

• Game theory's application to real-world scenarios is sensitive to how options are framed and perceived by decision-makers.

• Some games may lack Nash equilibria when only ordinal utilities are considered, complicating predictions of outcomes.

• Real-world interactions often occur under imperfect information, leading to unpredictable results in one-shot games.

The Stag Hunt: Cooperation and Individual Incentives

Rousseau's Stag Hunt

• The Stag Hunt illustrates the tension between cooperation and individual self-interest, as described by Jean-Jacques Rousseau.

• In this game, two hunters can either cooperate to hunt a stag or pursue a hare individually, with the stag providing greater rewards if both cooperate.

• The game emphasizes the importance of trust and mutual cooperation for achieving the best outcomes.

Payoff Matrix of the Stag Hunt

• The payoff matrix for the Stag Hunt can be represented as follows:

• The best outcome (4, 4) occurs when both players cooperate to hunt the stag, while pursuing a hare alone yields lower payoffs.

Implications of the Stag Hunt

• The Stag Hunt highlights the challenges of achieving cooperation in situations where individual incentives may lead to suboptimal outcomes.

• It serves as a metaphor for social cooperation, where individuals must weigh personal gain against collective benefits.

• The game illustrates the necessity of communication and trust in fostering cooperative behavior among individuals.

Stag Hunt Game Overview

Key Concepts of Stag Hunt

• The Stag Hunt illustrates the importance of cooperation in achieving optimal outcomes. Both players must work together to hunt a stag for maximum benefit, while hunting a hare can be done individually but yields less reward.

• The game presents two Nash equilibria: (Stag, Stag) and (Hare, Hare). The former is Pareto-superior, meaning both players are better off than in the latter equilibrium.

• The dilemma arises from the uncertainty of the other player's choice; each player must consider the other's potential actions when deciding their own strategy.

• The Stag Hunt is less paradoxical than the Prisoner's Dilemma, focusing on cooperation versus individualism rather than betrayal.

• Communication between players can enhance the likelihood of achieving the Pareto-optimal outcome by establishing common intentions.

• The game highlights the significance of common conjectures in rational decision-making, where players must believe the other will cooperate to achieve the best outcome.

Payoff Matrix Analysis

• The payoff matrix for the Stag Hunt is crucial for understanding the outcomes of different strategies. The payoffs are structured as follows:

• The matrix shows that mutual cooperation (Hunt Stag) yields the highest payoff for both players, while mutual defection (Hunt Hare) results in a lower payoff.

• The analysis of the matrix reveals the strategic interdependence of the players' choices, emphasizing the need for trust and communication.

• The concept of dominated strategies is explored, where neither player has a strictly better option, leading to multiple equilibria.

Rationality and Common Knowledge

• Rationality in the Stag Hunt is contingent upon the players' beliefs about each other's actions. Each player must consider the other's rationality when making decisions.

• The concept of rationalizable strategies is introduced, where players eliminate dominated strategies to simplify the decision-making process.

• The need for common knowledge is emphasized; both players must believe that the other will choose to hunt stag for it to be a rational choice.

• The cyclical reasoning of beliefs is highlighted: each player's decision to hunt stag depends on their belief that the other will also choose stag, creating a loop of uncertainty.

• The role of imperfect rationality is discussed, suggesting that short-sightedness can lead to suboptimal outcomes, such as both players choosing to hunt hare.

Assurance Game and Communication

Assurance Game Dynamics

• The Assurance Game is a variation of the Stag Hunt, where the payoffs differ, allowing for potential deception and miscommunication.

• The payoff matrix for the Assurance Game is structured as follows:

• In this game, while both players prefer to cooperate, the second-best option for one player may lead to strategic manipulation.

• Communication can be less effective in establishing common conjectures, as players may have incentives to mislead each other about their intentions.

• The Assurance Game illustrates the complexities of trust and strategy in cooperative scenarios, where the potential for deception exists.

Implications for Political Theory

• The Battle of the Sexes game is introduced as a related concept, emphasizing the need for coordination despite differing preferences.

• In this game, both players prefer to spend time together but have conflicting desires for activities, leading to multiple Nash equilibria.

• The payoff matrix for the Battle of the Sexes is structured as follows:

• The game highlights the importance of coordination in politics, where differing preferences must be reconciled for collective action.

• Jeremy Waldron's perspective on politics as an impure coordination game is discussed, emphasizing the necessity of settling on a common policy despite disagreements.

Understanding Politics as an Impure Coordination Game

Definition of Impure Coordination

• Politics is described as an impure coordination game where individuals desire to coordinate but disagree on the best coordination point.

• An impure coordination problem arises when at least one individual prefers a non-coordinated outcome over a coordinated one, indicating that unilateral defection can lead to better outcomes for some.

• Example: Alf, a Millian liberal, believes there should be no laws on self-regarding acts, which complicates the coordination on legislation regarding pornography.

Government as an Impure Coordination Problem

• While individuals agree on the necessity of government, they disagree on the type of government that is best, making it an impure coordination problem.

• The concept suggests that not all forms of government are seen as better than no government; oppressive regimes may be viewed as worse than anarchy.

• The challenge lies in identifying a set of governments that everyone agrees is superior to the absence of government, which can then be evaluated for selection.

Dominance and Rationalizability in Game Theory

Nash Equilibria and Dominated Strategies

• Nash equilibria are found by eliminating dominated strategies, which are strategies that are always worse than another strategy regardless of the opponent's actions.

• A strategy S₁ strictly dominates S₂ if it performs better against all of the opponent's strategies, while weak dominance means it performs at least as well and sometimes better.

• Example: In a game with Alf and Betty, if Alf's strategy S₁ is always better than S₂ against Betty's strategies, S₂ can be eliminated.

Example of Strategy Elimination

• The game illustrated in tables shows how to eliminate dominated strategies step by step, focusing on Betty's strategies first.

• In the first stage, we compare Betty's payoffs for Left and Center, finding that Center dominates Left in certain scenarios, but not universally.

• In the second stage, we find that Right dominates Left, allowing us to eliminate Left from consideration, simplifying the game.

Tables and Illustrations

Game Payoff Tables

• The tables illustrate the payoffs for each strategy combination, highlighting which strategies can be eliminated based on dominance.

Conclusion and Implications

Summary of Key Concepts

• Politics involves complex coordination problems where individual preferences can lead to disagreements on governance.

• Understanding dominated strategies is crucial for analyzing strategic interactions in political and economic contexts.

• The elimination of dominated strategies simplifies decision-making processes in games, leading to clearer insights into potential equilibria.