"In a strategic situation, when all participating parties adopt optimal strategies, no one can benefit by unilaterally changing their strategy."
—— Based on John Nash's concept of Nash equilibrium
Activation Conditions and Trigger Words
Direct Invocation
- "Use game theory to analyze..."
- "Use game theory to analyze..."
- "Analyze from the perspective of game theory..."
Semantic Triggering
- The user describes multi-party decision-making, strategic interaction, competition and cooperation
- Keywords: "strategy" "equilibrium" "zero-sum" "game" "negotiation" "auction" "competition" "cooperation" "deterrence"
Combined Invocation
- "First use game theory to analyze the competitive landscape, then use blue ocean strategy to find opportunities"
- "Use game theory + Five Forces analysis to assess industry competition"
Overview of the Method Framework
The core of game theory isusing mathematical models to analyze decision-making behavior in strategic interaction, and it uses 6 principles topredict the stable outcomes of multi-party decisions, design incentive mechanisms, and understand the evolution of cooperation and conflict. Game theory is not a prediction tool, but a thinking framework for systematically considering strategic interaction.
Core Principles (6, each must be accompanied by ≥2 cross-domain pieces of evidence)
Principle 1: Nash Equilibrium
One-sentence definition: A state of strategy combination in a game in which no participant can obtain a better outcome by unilaterally changing their strategy.
Cross-domain evidence：
- Economics: In price competition in oligopolistic markets, firms' pricing reaches a stable state (Source: Nash equilibrium - Wikipedia)
- Biology: The evolutionarily stable strategy (ESS) is the manifestation of Nash equilibrium in biological evolution (Source: Cornell University networks-book-ch07)
Application method: Predict the stable outcomes of multi-party decisions, and analyze the final convergence point of the competitive landscape and strategic interaction.
Limitations: Assumes complete rationality, while in reality participants often deviate from equilibrium predictions; multiple equilibria lead to predictive uncertainty.
Principle 2: Dominant Strategy
One-sentence definition: A strategy that is always the best choice regardless of what strategy other participants adopt.
Cross-domain evidence：
- Auction design: In a Vickrey second-price auction, truthful bidding is a dominant strategy (Source: Stanford University auction paper)
- International relations: In certain conflict scenarios, a preemptive strategy may become the dominant choice (Source: Schelling's Strategy of Conflict)
Application method: Design incentive mechanisms so that honest behavior becomes a dominant strategy, and identify the optimal action that requires no consideration of others' reactions.
Limitations: Dominant strategies are rare in reality; oversimplifies complex decision-making processes.
Principle 3: Zero-Sum vs Non-Zero-Sum Games
One-sentence definition: In a zero-sum game, one party's gain equals the other party's loss; in a non-zero-sum game, win-win or lose-lose outcomes may be possible.
Cross-domain evidence：
- Sports competition: Traditional matches are zero-sum games, where one side's victory equals the other side's defeat (Source: Towards Data Science)
- International trade: Trade negotiations are non-zero-sum games, and cooperation can be win-win (Source: ResearchGate trade applications)
Application approach: Identify competitive and cooperative relationships, look for win-win opportunities, and avoid falling into zero-sum thinking traps.
Limitations: The real world is rarely purely zero-sum or purely non-zero-sum; oversimplifies complex interest relationships.
Principle 4: Repeated Games & Cooperation Evolution
One-sentence definition: Through repeated interaction, cooperation can become an equilibrium strategy; the shadow of the future affects current behavior.
Cross-domain evidence：
- Biological evolution: reciprocal altruism (reciprocal altruism) is widespread in the animal kingdom (Source: Nature Education)
- Business relationships: Long-term supplier relationships promote cooperation, and the cost of betrayal is high (Source: Axelrod's tournaments)
Application approach: Design long-term incentive mechanisms, establish repeated interaction scenarios, and use the "shadow of the future" to promote cooperation.
Limitations: Requires clear expectations of future interaction; cooperation is difficult to sustain in finitely repeated games (backward induction problem).
Principle 5: Mechanism Design
One-sentence definition: Reverse-engineer game rules so that participants' self-interested behavior achieves system goals.
Cross-domain evidence：
- Spectrum auctions: FCC spectrum auction design achieves efficient resource allocation (Source: LSE Research spectrum auctions)
- Matching markets: Kidney transplant matching, student-school matching systems (Source: Wikipedia - Mechanism Design)
Application approach: Design market rules, incentive mechanisms, and voting systems so that individually rational choices lead to collectively optimal outcomes.
Limitations: Computational complexity limits; high demands on participants' information and rationality; implementation costs may be very high.
Principle 6: Bayesian Games & Incomplete Information
One-sentence definition: Strategic interaction by participants under incomplete information, making decisions based on beliefs about others' types.
Cross-domain evidence：
- Auction bidding: Bidders' uncertainty about others' valuations (Source: Harsanyi 1994 Nobel Prize contribution)
- Diplomatic negotiations: Countries' incomplete information about each other's true intentions (Source: Schelling's Strategy of Conflict)
Application approach: Handle information asymmetry scenarios, analyze belief formation and strategy selection, and design signaling mechanisms.
Limitations: Requires forming prior beliefs about others' information; belief updating may be erroneous; high model complexity.
Agentic Protocol
Step 1: Problem classification — determine whether game theory applies to the current problem
Applicable signals：
- Involves 2 or more decision-makers, and the decisions of the parties affect each other
- There are clear strategy choices and payoff structures
- The problem involves competition, cooperation, negotiation, auctions, or mechanism design
- It is necessary to predict or analyze the stable outcome of multi-party interaction
Not applicable signal：
- Single-person decision problem (no strategic interaction)
- Purely technical or engineering problem (non-adversarial)
- Problems lacking clear strategies and payoff structures
- Scenarios involving strong emotional, moral, or cultural factors that are difficult to model as payoffs
Output: Applicable / Not applicable (with reasons) / Partially applicable (with suggestions)
Step 2: Game-theoretic analysis — research dimensions must be derived from core principles
Research dimensions(each dimension must be traceable to a specific core principle):
-
Participant identification and game type
- Source principle: Nash equilibrium, zero-sum and non-zero-sum games
- Method of operation:
- List all participants (2-N)
- Determine whether it is a zero-sum or non-zero-sum game
- Determine whether it is a one-shot or repeated game
- Determine whether information is complete
-
Strategy space and payoff matrix
- Source principle: Nash equilibrium, dominant strategies
- Method of operation:
- List each participant's available strategies
- Construct a 2×2 or N×M payoff matrix (if applicable)
- Identify whether a dominant strategy exists
- Mark each party's payoffs (quantitative or qualitative)
-
Equilibrium analysis
- Source principle: Nash equilibrium, Bayesian games
- Method of operation:
- Find Nash equilibrium (pure strategy or mixed strategy)
- If there are multiple equilibria, analyze which is most likely to be realized
- Analyze the belief formation process (e.g., incomplete information)
- Consider trembling-hand perfection (if applicable)
-
Dynamic and repeated game analysis
- Source principle: repeated games and the evolution of cooperation
- Method of operation:
- Determine whether there is repeated interaction (shadow of the future)
- Analyze trigger strategies (e.g., tit-for-tat)
- Evaluate the conditions for maintaining cooperation
- Analyze the difference between finitely vs infinitely repeated games
-
Mechanism design perspective
- Source principle: mechanism design
- Method of operation:
- Analyze how current rules/mechanisms affect incentives
- Identify incentive distortions that may lead to inefficiency
- Propose suggestions for mechanism improvement (if applicable)
- Evaluate implementation costs
-
Behavioral and real-world constraints
- Source principle: the limitations of all principles
- Method of operation:
- Identify possible points of failure of the rationality assumption
- Consider fairness perceptions, social norms, and trust factors
- Evaluate the degree of information asymmetry
- Consider computational complexity and implementation feasibility
Prohibited: Use general instructions such as "search for relevant information" and "comprehensively understand the background." All research dimensions must be derived from core principles.
Step 3: Game-theoretic output — formatted output based on the analysis results
Output format requirements:
## 博弈论分析结果
### 博弈结构
- 参与者：[列出]
- 博弈类型：[零和/非零和] [一次性/重复] [完全信息/不完全信息]
- 策略空间：[描述]
### 均衡分析
- 纳什均衡：[列出均衡点]
- 预测结果：[基于均衡的预测]
- 置信度：[高/中/低，说明原因]
### 关键洞察
1. [洞察1]
2. [洞察2]
3. [洞察3]
### 策略建议
- 对参与者A：[建议]
- 对参与者B：[建议]
- 对机制设计者：[如适用]
### 局限与警告
- [列出分析中的主要假设和可能的失效点]
Present the analysis results in a structured way, and label confidence (high/medium/low). If confidence is low, clearly state the reasons and the recommended next steps.
Applicability/Non-applicability assessment
| Problem type | Applicability | Explanation |
|---|---|---|
| Competitive strategy analysis | High | Typical scenarios such as price wars, market share competition, and standards competition |
| Negotiation and conflict | High | Trade negotiations, salary negotiations, international conflicts, and deterrence strategies |
| Mechanism and rule design | High | Auction design, matching systems, incentive mechanisms, and voting rules |
| Cooperation and evolution | High | Long-term cooperative relationships, alliance formation, and trust building |
| One-shot moral dilemmas | Medium | Game theory can analyze them, but behavior predictions may be inaccurate (affected by emotions) |
| Purely technical engineering problems | Low | No strategic interaction; not applicable |
| Purely emotional/ethical problems | Low | Difficult to quantify payoffs; other frameworks are recommended |
| Extremely complex systems | Low | Too many variables; it is recommended to first reduce dimensionality using systems thinking |
Typical case library
Success cases
-
Cuban Missile Crisis (1962)
- Problem: The United States and the Soviet Union were on the brink of nuclear war; how could a devastating conflict be avoided?
- Method application: Thomas Schelling's deterrence theory and analysis of commitment strategies; establishing "credible commitment" and "limited options" strategies
- Result: The United States successfully prompted the Soviet Union to withdraw its missiles and avoided nuclear war
- Source: A Game-Theoretic History of the Cuban Missile Crisis (MDPI)
-
FCC spectrum auction
- Problem: How could the U.S. government effectively allocate limited radio spectrum to telecommunications companies?
- Method application: Use mechanism design and game theory to design a simultaneous multiple-round auction (SMRA)
- Result: Auction efficiency improved substantially, raising billions of dollars for the government, and spectrum allocation better matched market demand
- Source: Game theory and the spectrum auctions (ScienceDirect)
-
Axelrod's iterated prisoner's dilemma tournament
- Problem: How do selfish individuals evolve cooperative behavior?
- Method application: Organize a computerized prisoner's dilemma tournament and invite scholars worldwide to submit strategies
- Result: Found that the "Tit-for-Tat" strategy was optimal, proving that simple rules can promote the emergence of cooperation
- Source: The Evolution of Cooperation (Axelrod 1984)
-
ESS in biological evolution
- Problem: How can the widespread altruistic behavior and complex social structures in the animal kingdom be explained?
- Method application: Introduce game theory into biology and propose the concept of an evolutionarily stable strategy (ESS)
- Result: Successfully explained complex biological behaviors such as altruism, territorial competition, and courtship displays
- Source: Evolutionarily Stable Strategy (Wikipedia, Nature Education)
-
International trade negotiations
- Problem: A rational analytical framework for trade wars and trade agreements
- Method application: Use game theory to analyze tariff strategies and retaliatory measures
- Result: Explained the lose-lose nature of trade wars and the win-win potential of agreements, providing a framework for policy design
- Source: Review of Game Theory Applications in International Trade (ResearchGate)
Failure cases
-
Failure of rational models in the ultimatum game
- Problem: Predicting how people behave when allocating money
- Misuse: Classical game theory predicts that the proposer should offer the responder the minimum amount, and the responder should accept any positive amount
- Actual situation: In experiments, people often reject "unfair" low offers (driven by a sense of fairness)
- Lesson: Human behavior is driven by emotional factors such as fairness and anger, and cannot be explained solely by rational choice models
- Source: Ultimatum Game - ScienceDirect Topics
-
Failure of cooperation in the centipede game
- Problem: Predicting the behavior of rational players in a multi-round cooperative game
- Misuse: Backward induction predicts that rational players will terminate the game at the first step (taking a smaller payoff)
- Actual situation: In experiments, participants often continue cooperating for multiple rounds, far beyond theoretical predictions
- Lesson: The assumptions of full rationality and common knowledge are unrealistic; humans have a tendency to cooperate and doubts about others' rationality
- Source: Rationality and backward induction in Centipede games (ResearchGate)
-
Large-scale predictive failures of game theory
- Problem: Applying game theory as a precise predictive tool to real-world decision-making
- Misuse: Expecting game theory to accurately predict people's behavior
- Actual situation: "Game theory applications have generally led to large-scale predictive failures. People simply do not act according to the theory."
- Lesson: Game theory is better suited as a thinking framework and analytical tool rather than a precise predictive tool
- Source: What is wrong with game theory (Lars P. Syll)
Misuse detector (≥3 misuse patterns)
| Misuse signal | Detection logic | Warning message | Suggested action |
|---|---|---|---|
| Method-problem mismatch | The user's problem falls within the "not applicable" range (purely technical problems, purely ethical problems) | "You are using game theory to handle [problem type], which may not be the best choice. Game theory is suitable for strategic interaction scenarios. Consider using [alternative methodology]." | Recommend alternative methods (such as systems thinking, ethical analysis frameworks) |
| Skipping key steps | The user directly asks for conclusions, skipping player identification, strategy space, and payoff matrix construction | "Game theory analysis requires completing the identification of the game structure first, and this cannot be skipped. Only after constructing the player-strategy-payoff framework can reliable conclusions be given." | Guide back to the correct process |
| Over-rationalization | Ignoring behavioral factors and assuming participants are fully rational | "Note: Your analysis assumes complete rationality. In reality, people's behavior is influenced by emotions, fairness perceptions, and social norms. It is recommended to add a behavioral economics perspective." | Consider behavioral factors |
| The model is overly simplified | Forcing complex reality into a simple game model | "This problem may be more complex than the standard game model can handle. Consider using systems thinking to reduce dimensionality first, or use computational game theory tools." | Recommend combined use or simplification |
| Ignoring uncertainty | Still giving high-confidence predictions in a highly uncertain environment | "Game theory has limited predictive power in highly uncertain environments. Your confidence level should be lowered, and assumptions and failure points should be clearly marked." | Adjust confidence level and mark assumptions |
Honest boundaries (≥3 specific limitations; empty phrases such as "cannot replace professional advice" are prohibited)
-
Limited predictive power: Game theory "has by and large not been successful in producing determinate, reliable predictions." Game theory is better suited as an analytical framework and thinking tool rather than a precise predictive tool. Its value lies in systematically thinking about strategic interaction, not in predicting specific behavior. (Source: Crooked Timber, Lars P. Syll)
-
Computational complexity constraints: Computing Nash equilibrium in general games is a PPAD-complete problem, and computational complexity grows exponentially with the number of participants. Even if an equilibrium exists in theory, it may not be computable in reality. (Source: 20 Years of Algorithmic Game Theory, CMU)
-
Multiple equilibria and the selection problem: Games often have multiple Nash equilibria, and the theory itself cannot predict which equilibrium will be realized. Factors outside the theory, such as focal points, learning dynamics, and cultural background, need to be introduced for equilibrium refinement.
-
Systematic behavioral deviations: The cognitive biases discovered by Kahneman and Tversky systematically violate the assumptions of rational choice. Prospect theory shows that people systematically deviate from expected utility maximization in risky decision-making. (Source: Tversky and Kahneman's Cognitive Illusions, PMC)
-
Ethics and social norms are difficult to model: Game theory models have difficulty fully incorporating moral considerations and social norms. Human behavior is influenced by ethics, culture, and social norms, which are difficult to fully model as payoff functions. (Source: Game Theory and Ethics, Stanford Encyclopedia)
Research sources (primary sources account for 60%)
| # | Source | Type | Primary/Secondary | Weight |
|---|---|---|---|---|
| 1 | Nash, J. F. (1950). "Non-Cooperative Games" | Paper | Primary | High |
| 2 | von Neumann, J., & Morgenstern, O. (1944). "Theory of Games and Economic Behavior" | Book | Primary | High |
| 3 | Schelling, T. C. (1960). "The Strategy of Conflict" | Book | Primary | High |
| 4 | Axelrod, R. (1984). "The Evolution of Cooperation" | Book | Primary | High |
| 5 | Harsanyi, J. C. Contributions to Bayesian games | Paper | Primary | High |
| 6 | Selten, R. Contributions to subgame perfect equilibrium | Paper | Primary | High |
| 7 | "Solving the second-order free rider problem" (Nature) | Experimental research | Primary | Medium |
| 8 | "Game theory and the spectrum auctions" (ScienceDirect) | Applied research | Primary | Medium |
| 9 | "A Game-Theoretic History of the Cuban Missile Crisis" (MDPI) | Case analysis | Secondhand | Medium |
| 10 | "Review of Game Theory Applications in International Trade" (ResearchGate) | Application Overview | Secondhand | Medium |
| 11 | "What is wrong with game theory" (Lars P. Syll) | Critical Analysis | Secondhand | Medium |
| 12 | Algorithmic Game Theory (Noam Nisan et al.) | Textbook | Secondhand | Low |
Proportion of primary sources: 60% (7/12, the first 6 are high-weight primary sources)
This Skill was generated by Forge Skill — Forging Thinking Tools generated