Finite and Infinitely Repeated Games

students, imagine playing the same strategy game with a classmate not just once, but again and again 🎮. In a one-shot game, you make your move and the game ends. In a repeated game, today’s choice can affect tomorrow’s outcome, which changes everything. The big idea in this lesson is that repetition can make cooperation easier to sustain, but only when the future matters enough. By the end, you should be able to distinguish finite from repeated games, explain how repetition changes incentives, and identify when future play matters.

One-shot play versus repeated interaction

A one-shot game is played only once. Each player chooses an action, the payoffs are realized, and the game is over. There is no later round to reward, punish, or build trust. A classic example is a single auction bid or a one-time bargaining decision over the price of a used phone 📱.

A repeated game happens when the same strategic situation is played multiple times. The players may observe what happened before and use that information in later rounds. If a class project group meets every week to divide tasks, that is a repeated interaction. Your choice in one week can shape whether your teammates trust you next week.

The key difference is that in repeated games, the future can influence the present. In a one-shot game, a player usually focuses only on the current payoff. In a repeated game, the player may also care about future payoffs that depend on today’s action.

For example, suppose two firms can either compete aggressively on price or keep prices high. In a one-shot setting, cutting prices may give one firm a quick advantage. But if the firms expect to meet again and again, a price cut today might trigger retaliation later. The possibility of future consequences can change what each firm chooses now.

Finite repeated games: when the end is known

A finite repeated game has a known final round. If a game is played exactly $T$ times, then the last round is round $T$. Because the end is certain, players know that eventually there will be no future rounds left to protect or lose.

This matters a lot. Think about a repeated homework-help arrangement with a classmate that lasts for exactly 5 days. On day 5, there is no future day 6, so a promise like “I’ll help you tomorrow if you help me today” loses force at the end.

In many finite repeated games, the last round looks like a one-shot game. Why? Because in the final round, there is no future punishment or reward after that move. If the game ends at $T$, then the payoff from round $T$ is the last payoff players care about. That means backward thinking can begin. Players reason from the end toward the beginning.

Here is the logic in simple terms:

In the final round, players may have an incentive to act selfishly if that gives the best immediate payoff.
If everyone expects selfish behavior in the final round, then the same logic can spread backward to earlier rounds.
As a result, cooperation can be hard to sustain in a finite game, especially when players are fully rational and know the ending in advance.

This idea is often called backward induction. It means you work backward from the last period to figure out what will happen earlier. If the last round has no future, then threats like “I’ll punish you next time” do not matter at the end, because there is no next time.

A concrete example: two roommates decide each week whether to clean the kitchen or leave it messy. If the arrangement lasts for exactly four weeks, each roommate may think, “Why clean in week 4 if the other person cannot punish me afterward?” If both reason this way, the final week may become messy, and that logic can weaken cooperation in earlier weeks too. 😕

Infinitely repeated games: when the future never fully ends

An infinitely repeated game does not have a known final round. This does not mean the game literally lasts forever in real life, but rather that players do not know when it will stop, or they place enough weight on the future that it feels open-ended.

In this setting, future rounds can matter a great deal. If a player cheats today, other players may punish them in the next round and in many rounds after that. Because the relationship continues, the short-term gain from cheating must be compared with the long-term loss from damaging future cooperation.

A common real-world example is a neighborhood market where the same customers and sellers interact regularly. A seller who overcharges one customer today may lose that customer’s trust forever. If future business is valuable enough, honest behavior today can be a smart strategy.

To see the logic, suppose a player gets a one-time gain of $g$ from defecting today, but loses a future benefit of $f$ in each later round because others stop cooperating. If the future is important enough, the total loss may outweigh the immediate gain. In that case, cooperation can be sustained.

This is why reputation matters. In repeated games, players often care about building a good record. A strong reputation can encourage others to cooperate with you, while a bad reputation can lead to retaliation or exclusion. That is why repeated interaction is so important in business, friendships, and teamwork 🤝.

Why repetition changes incentives

Repetition changes incentives because actions have two effects instead of one:

Immediate effect: what you gain or lose right now.
Future effect: how your action changes what happens later.

In a one-shot game, only the immediate effect matters. In a repeated game, the future effect can be just as important.

This is easiest to understand with a trust example. Suppose two students are sharing notes. If both share honestly, both benefit. If one student grabs the notes without sharing back, that student may get a quick advantage. But if the other student notices and stops sharing next time, the cheater loses future help. The chance of losing future benefits can make honest sharing the better choice overall.

Repeated games can support cooperation through several mechanisms:

Reward: cooperating today leads to cooperation from others later.
Punishment: cheating today leads to retaliation later.
Reputation: others learn about your behavior and adjust their choices.
Trust building: consistent cooperation makes future coordination easier.

These mechanisms work best when players expect enough future interaction. If people know they will never meet again, future punishments are weak. If they expect many future meetings, the future becomes a powerful incentive.

When does future play matter?

Future play matters when the value of later rounds is large enough compared with today’s payoff. In economics, this idea is often described using discounting. A player may care less about payoffs far in the future than payoffs today. If future payoffs are heavily discounted, then long-term rewards may not be enough to stop cheating.

The important idea is not the exact formula, but the comparison: does the long-run loss from breaking cooperation exceed the short-run gain from cheating? If yes, cooperation may survive. If no, players may defect.

Future play matters more when:

the relationship is long-lasting,
players expect many future rounds,
future benefits are valuable,
punishments for cheating are credible,
and reputation is important.

Future play matters less when:

the game is almost over,
players know they will not meet again,
future benefits are small,
or the chance of being punished is low.

A simple example is borrowing a charger from a friend. If you know you will see your friend every day, returning it on time matters because you want future trust. If you know it is the last time you will ever interact, the future incentive is much weaker.

Comparing the two cases with a bargaining example

Bargaining is a good way to see the difference between one-shot and repeated games. Suppose two friends must divide a pizza 🍕. In a one-shot bargaining game, each friend may push hard for the biggest slice because there is no tomorrow to worry about.

Now imagine the same two friends must share food every week. If one friend takes too much today, the other may remember it next week and respond by taking more later or refusing to cooperate. Over time, both friends may do better by being fair, because fairness helps preserve the relationship.

In a finite repeated bargaining game, cooperation may start early but weaken near the end because everyone knows the final round is approaching. In an infinitely repeated bargaining setting, fairness can remain stable if both sides care enough about the future and believe the other side will keep playing along.

This helps explain why long-term partnerships often involve patience and compromise. People are not only dividing today’s gains; they are also protecting tomorrow’s opportunities.

Conclusion

Finite and infinitely repeated games show how time changes strategy. In a one-shot game, players focus on the current round. In a finite repeated game, the known ending can weaken cooperation, especially near the last round. In an infinitely repeated game, the future can support cooperation because today’s choice affects tomorrow’s opportunities.

students, the main lesson is that future play matters when players expect to interact again and care enough about what happens later. That is why trust, reputation, and credible punishment are so important in repeated games. Understanding this helps explain behavior in friendships, markets, workplaces, and negotiations.

Study Notes

A one-shot game is played once, so only the current payoff matters.
A repeated game is played over multiple rounds, so past actions can affect future outcomes.
A finite repeated game has a known last round.
In a finite game, the last round can behave like a one-shot game because there is no future afterward.
Backward induction is reasoning from the final round backward to predict behavior earlier in the game.
An infinitely repeated game has no known final round, so future consequences can support cooperation.
Repetition changes incentives because each action has both an immediate effect and a future effect.
Cooperation can be sustained by rewards, punishments, reputation, and trust.
Future play matters most when players expect many future interactions and value later payoffs enough.
Future play matters less when the relationship is ending or when future rewards are small.