Some decisions cannot be evaluated by asking only what one person wants. A school chooses a policy while families adapt. A firm changes price while rivals respond. A country announces a commitment while other states decide whether to believe it. A learner decides how much effort to invest while anticipating how an examiner will reward particular forms of performance.
Game theory studies situations in which the outcome for each decision-maker depends not only on that decision-maker’s own action, but also on the actions, information and expectations of others. It is a framework for strategic interdependence.
Game theory does not say that people are selfish, emotionless or perfectly rational. It provides formal models of strategic situations. The usefulness of a particular game depends on whether its players, actions, information and payoffs represent the real problem closely enough.
All numerical games below are constructed for explanation. They are not descriptions of actual eduKate behaviour, commercial recommendations or claims about any institution.
1. Strategic interaction begins when my best action depends on yours
If my best route to work is independent of what everyone else does, the problem can be solved individually. If congestion changes when thousands choose the same route, my outcome depends on others’ choices. That interdependence creates a game.
The Stanford Encyclopedia of Philosophy’s 2023 treatment of game theory describes the field around interacting choices, preferences, information and equilibrium concepts.
2. A game needs players, actions and payoffs
Players are the decision-makers represented by the model. Strategies specify what they can do. Payoffs represent how each player ranks outcomes.
A payoff is not necessarily money. It can represent utility, cost, probability of success or another ordered consequence. The chosen representation must match the question.
3. Strategic form puts simultaneous choices into a matrix
Consider two fictional study groups deciding whether to Share a common revision room or Hoard it for themselves. If both share, each receives 3 utility points. If one hoards while the other shares, the hoarder gets 4 and the sharer gets 0. If both hoard, each gets 1.
| B: Share | B: Hoard | |
|---|---|---|
| A: Share | 3,3 | 0,4 |
| A: Hoard | 4,0 | 1,1 |
The matrix makes interdependence visible: A’s payoff from Share depends on B’s choice, and vice versa.
4. A dominant strategy is best against every action of the other player
In the constructed game, Hoard gives A 4 instead of 3 if B shares and 1 instead of 0 if B hoards. Hoard is therefore A’s dominant strategy. Symmetry gives B the same conclusion.
The resulting outcome is Hoard-Hoard with payoff 1,1, even though Share-Share gives both 3. Individual strategic incentives can therefore produce a collectively inferior outcome.
5. The Prisoner’s Dilemma is a structure, not a story about prisoners
The familiar Prisoner’s Dilemma represents a conflict between individually dominant choices and a mutually preferable cooperative outcome. The essential object is the payoff ordering, not the narrative attached to it.
When using the label, show why the real incentives actually fit the structure. Many situations described casually as dilemmas do not.
6. Nash equilibrium is mutual best response
A Nash equilibrium is a strategy profile in which no player can improve by changing only their own strategy while the others keep theirs fixed.
In the sharing game, Hoard-Hoard is a Nash equilibrium. Neither side benefits from switching alone.
7. Equilibrium does not mean socially optimal
An equilibrium is a stability concept under unilateral deviation. It is not a moral endorsement, a fairness judgement or proof that the outcome maximises total welfare.
This distinction is essential whenever game theory is used in policy or organisational design.
8. Some games have several equilibria
Imagine two teams choosing between two communication standards. Both benefit most when they coordinate on the same standard, but each prefers a different one. There may be two Nash equilibria, one around each coordinated standard.
The model then predicts a set of stable outcomes, not which one history will select.
9. Coordination creates path dependence
Standards, conventions, language and technology can exhibit coordination effects. Once enough participants use one option, switching becomes costly even if an alternative has desirable features.
This connects game theory to Network Science and Graph Analysis and to the study of standards and interoperability.
10. Mixed strategies represent deliberate randomisation
Some games have no pure-strategy equilibrium. A player may then randomise among actions so that opponents cannot exploit a predictable pattern.
A mixed strategy is a probability distribution over actions. It is not the same as uncertainty about what a player intends.
11. Zero-sum games are only one special class
In a zero-sum game, one player’s gain is exactly another’s loss. Many real interactions are not zero-sum. Trade, shared infrastructure and cooperation can create outcomes in which several players improve together.
Assuming zero-sum structure where joint gains are possible can distort strategy before any calculation begins.
12. Extensive-form games represent sequence and information
When players move at different times, a tree can represent decision nodes, actions and later responses. Sequential structure matters because later players may observe earlier actions.
The strategic question becomes not only “What should I do?” but “How will others interpret and respond to what I do?”
13. Backward induction reasons from the end of a finite game
If later choices are rational and observable, a player can reason backward from final decisions to earlier ones. This resembles dynamic programming: future continuation values affect current action.
The connection is conceptual rather than identity. Game theory includes multiple decision-makers with potentially conflicting objectives.
14. Credible threats must be optimal when the time arrives
A player may threaten a costly response to deter another action. If carrying out that threat would be irrational once the situation occurs, the threat may lack credibility.
Subgame-perfect equilibrium refines Nash equilibrium by requiring strategic consistency in every relevant continuation game.
15. Commitment changes the game by changing future options
A contract, public rule, precommitted budget or technical lock can remove later choices. By reducing future flexibility, commitment can sometimes improve the credibility of current promises.
This is a striking strategic insight: having fewer future options can sometimes strengthen present bargaining power.
16. Information can be private, public or asymmetric
One player may know their own cost, quality or intention while another does not. Asymmetric information changes incentives because actions can reveal information as well as produce direct outcomes.
Models of signalling and screening study how information moves through strategic behaviour.
17. Signalling works only when signals have strategic meaning
A signal is useful when different underlying types have different incentives or costs for producing it. A statement that every type can make equally cheaply may carry little separating information.
This principle applies broadly to credentials, warranties, costly commitments and reputation, but each real application requires its own evidence.
18. Repeated games can support cooperation
When players expect to interact again, today’s defection can affect tomorrow’s treatment. The shadow of the future can therefore change incentives relative to a one-shot game.
Repeated interaction can support cooperation, retaliation, forgiveness or persistent conflict depending on payoffs, observability and expectations.
19. Reputation is a state variable in repeated interaction
If future partners condition their behaviour on past conduct, reputation links one game to the next. A short-term gain can create a long-term strategic cost.
This is another bridge to dynamic programming and stochastic processes: strategic histories create states that affect future decisions.
20. Mechanism design reverses the direction of analysis
Ordinary game theory asks what outcomes follow from a given set of rules. Mechanism design asks how to choose rules so that strategic behaviour produces desirable outcomes.
Auctions, matching systems and allocation rules are classic examples. The mechanism must be evaluated for incentives as well as stated intentions.
21. Incentive compatibility means truth or compliance is strategically sensible
A rule can fail if participants benefit from misreporting information or gaming the metric. Strong design asks what behaviour the rule rewards, not only what behaviour its author hopes for.
This is especially important in education and organisations: measurement systems can change the behaviour they measure.
22. Evolutionary game theory relaxes some rational-choice assumptions
Strategies can spread because they perform well relative to alternatives rather than because players consciously solve an optimisation problem. Evolutionary models examine population dynamics and stability.
This does not make the model biological by default; the same mathematics can describe imitation and adaptation in social systems.
23. Behavioural evidence can contradict simple equilibrium assumptions
People may care about fairness, reciprocity, identity and bounded attention. They may make mistakes or misunderstand the game. A formal equilibrium remains a conditional model, not a universal psychological law.
Empirical validation matters whenever game theory is used to predict real behaviour.
24. Games can change when the rules become known
Publishing a metric, ranking rule or enforcement policy can change strategic behaviour. The model is then part of the causal environment rather than a passive description.
This connects to Models and Simulations: deployed models can become performative.
25. A practical strategic-analysis checklist
- Who are the actual decision-makers?
- Which actions are genuinely available?
- What does each player know, and when?
- How does each player rank outcomes?
- Which actions are best responses?
- Are there dominant strategies or multiple equilibria?
- What changes if the game is repeated?
- Which promises or threats are credible?
- Can participants manipulate the rule or metric?
- Does empirical behaviour resemble the assumed solution concept?
26. The deepest lesson
Strategy begins when another person’s response enters your own decision. Game theory makes that interdependence explicit.
Its best use is not to turn every human relationship into a contest. It is to reveal where incentives, information and expectations change what rational action means.
Sources and further reading
- Stanford Encyclopedia of Philosophy — Game Theory, substantive revision 2023.
- Nobel Prize — 1994 Economic Sciences Prize for pioneering analysis of equilibria in non-cooperative games.
- Nobel Prize — 2007 Economic Sciences Prize for mechanism design theory.
Continue through eduKate: Decision Theory and Expected Utility → Dynamic Programming and Bellman Equations → Network Science and Graph Analysis → Research Collections Directory.