Sequential and Simultaneous Games with

Discrete Strategies

Abdul Quadir
XLRI

12 December, 2019
Reading: Dixit and Skeath, Chapter 3 and 4
Sequential Games with Three Players

I Suppose there are residents living across a street.

I They have been asked to contribute for the decoration
(development of a garden) of the street.
I

I Everybody likes to have the garden.

I The beauty of the garden depend how many are contributing
and contribution is costly.
I For the development and maintenance of the garden at least
two have to contribute.
I How many outcomes are there for each Player?
Sequential Games with Three Players

I There are four outcomes for each Player:

1. 1 does not contribute but two contributes (Pleasant garden
and Player 1 saving her cost)
2. 1 contributes and 2 and 3 or one of them contribute(s)
(pleasant garden but 1 incurs cost of her contribution)
3. 1 does not contribute and only one or neither 2 or 3
contributes (unpleasant part and 1 saves her contribution)
4. 1 contributes but neither of the others does (sparse garden and
1 incurs cost of contribution).
I 1 is the best and 4 is the worst.
I Thus, you can assign the higher payoff to 1 and then the
subsequent outcomes.
I Can you construct the game tree for this game?
Sequential Games with Three Players

I There are four outcomes for each Player:

1. 1 does not contribute but two contributes (Pleasant garden
and Player 1 saving her cost)
2. 1 contributes and 2 and 3 or one of them contribute(s)
(pleasant garden but 1 incurs cost of her contribution)
3. 1 does not contribute and only one or neither 2 or 3
contributes (unpleasant part and 1 saves her contribution)
4. 1 contributes but neither of the others does (sparse garden and
1 incurs cost of contribution).
I 1 is the best and 4 is the worst.
I Thus, you can assign the higher payoff to 1 and then the
subsequent outcomes.
I Can you construct the game tree for this game?
Game Tree with Three Players
Contri. 3,3,3
3
Contribute
2 Don’t 3,3,4

Contribute Contri.
3 3,4,3
Don’t
1 Don’t 1,2,2

Contri. 4,3,3
3
Contribute
Don’t 2 Don’t 2,1,2

Contri. 2,2,1
3
Don’t
Don’t 2,2,2
Example: Strategies

I Let us write C for contribute and D for don’t contribute.

I Player 1’s strategies: {C , D}.
I Player 2’s strategies: {CC,CD,DC,DD}
I Player 3’s strategies: {CCCC, CCCD, CCDD, CCDC, CDDD,
CDDC, CDCC,CDCD, DDDD, DDDC, DDCC, DCCC, DDCD,
DCDD, DCDC,DCCD }
I Use the backward induction to solve the problem.
I Calculate the optimal strategies of the Players.
Example: Strategies

I Let us write C for contribute and D for don’t contribute.

I Player 1’s strategies: {C , D}.
I Player 2’s strategies: {CC,CD,DC,DD}
I Player 3’s strategies: {CCCC, CCCD, CCDD, CCDC, CDDD,
CDDC, CDCC,CDCD, DDDD, DDDC, DDCC, DCCC, DDCD,
DCDD, DCDC,DCCD }
I Use the backward induction to solve the problem.
I Calculate the optimal strategies of the Players.
Example: Strategies

I Let us write C for contribute and D for don’t contribute.

I Player 1’s strategies: {C , D}.
I Player 2’s strategies: {CC,CD,DC,DD}
I Player 3’s strategies: {CCCC, CCCD, CCDD, CCDC, CDDD,
CDDC, CDCC,CDCD, DDDD, DDDC, DDCC, DCCC, DDCD,
DCDD, DCDC,DCCD }
I Use the backward induction to solve the problem.
I Calculate the optimal strategies of the Players.
Example: Strategies

I Let us write C for contribute and D for don’t contribute.

I Player 1’s strategies: {C , D}.
I Player 2’s strategies: {CC,CD,DC,DD}
I Player 3’s strategies: {CCCC, CCCD, CCDD, CCDC, CDDD,
CDDC, CDCC,CDCD, DDDD, DDDC, DDCC, DCCC, DDCD,
DCDD, DCDC,DCCD }
I Use the backward induction to solve the problem.
I Calculate the optimal strategies of the Players.
Optimal Strategy for Player 1
Contri. 3,3,3
3
Contribute
2 Don’t 3,3,4

Contribute Contri.
3 3,4,3
Don’t
1 Don’t 1,2,2

Contri. 4,3,3
3
Contribute
Don’t 2 Don’t 2,1,2

Contri. 2,2,1
3
Don’t
Don’t 2,2,2
Optimal Strategy for Player 2
Contri. 3,3,3
3
Contribute
2 Don’t 3,3,4

Contribute Contri.
3 3,4,3
Don’t
1 Don’t 1,2,2

Contri. 4,3,3
3
Contribute
Don’t 2 Don’t 2,1,2

Contri. 2,2,1
3
Don’t
Don’t 2,2,2
Optimal Strategy for Player 1
Contri. 3,3,3
3
Contribute
2 Don’t 3,3,4

Contri. 3,4,3
3
Don’t
1 Don’t 1,2,2

Contri. 4,3,3
3
Contribute
Don’t Don’t 2,1,2

Contri. 2,2,1
3
Don’t
Don’t 2,2,2
Mover Advantages
I In the above example, observe that Player 1 has an advantage
on others.
I This is known as first-mover advantage.
I However, it is not necessary that first-mover advantage will
exist in every game.
I For instance, suppose two firms which sell the identical
merchandise from a catalogs.
I Thus, the firm which moves second can observe the price of
the first firm’s catalog and undercut its price on all items.
I Hence, second mover will gain advantage over its rival.
I First-mover advantage arises because of commitment such
that force other players to adapt it.
I Second-mover advantage arises from the flexibility that second
mover adapts himself or herself to others’ choices.
Centipede Game

I Choose two players and name them A and B.

I There is a coin of rupee 1 placed on a table.
I Player A can take the coin or pass it to Player B
I If Player A takes the coin, game ends, Player A gets rupee 1
and Player b gets nothing.
I If Player A passes the coin to the Player B, then the
experimenter add one more coin.
I Now Player either takes the coins or passes it to Player A.
I Again if Player B takes the coin, the game ends and if he or
she passes it to Player A, then the experimenter will add one
more coin.
I The game will go on until there are 100 coins.
Game Tree

A Pass B Pass A Pass Pass

0, 0

Take Coin Take Coin Take Coin Take Coin

1, 0 0, 2 3, 0 0, 100
Simultaneous-Move Games: Discrete Strategies
Introduction

I Simultaneous move game arises because

I the players are choosing their actions at the same time.
I they are choosing their actions in isolation.
I Solution concept for these games is Nash equilibrium.
I Almost all automobiles firms and many more more firms make
decisions without knowing what rivals are doing.
I In election voters are casting their votes without knowing
whom others are voting.
I Simultaneous-move games are also known as normal form or
strategic form games.
Game Matrix

I Unlike sequential-move games, you will have one chance to

play the game.
I Therefore, strategies and actions are not distinct in
simultaneous-move games.
I However, the player might randomize over his or her actions
(choose an action with some probability), mixed strategies.
I In this lecture, we will focus on pure strategies.
I A simultaneous-move game comprises on:
I set of players
I set of strategies (actions)
I payoffs
I For discrete strategies the payoffs are depicted with the help
of payoff matrix or game matrix or game table.
Game Matrix

Left Middle Right

Top (2,2) (6,1) (1,1)
High (1,3) (5,5) (9,2)
Low (0,0) (4,2) (8,8)
Bottom (5,6) (4,5) (9,7)

I Two players: Row Player and Column Player

I Rower Player’s strategies: {Top, High, Low, Bottom}
I Column Player’s strategies: {Left, Middle, Right}.
Example 1: Prisoner’s Dilemma

I Suppose N = {1, 2}. These players are prisoners.

I Due to lack of evidence, they have been questioned in
separate rooms and made to confess their crimes.
I Each player has two actions: Confess (C) and don’t confess
(D).
I Thus, the pair of actions are: (C , C ), (C , D), (D, C ), (D, D).
I The police give the following situation to the Players.
I If one confesses and other don’t confess, then confessor will
get lesser punishment (get higher payoff).
I If both confess, they will get lesser punishment.
I If both do not confess, they will not be charged and set free.
I Prisoner 1 will rank the actions as (C , D)  (D, D)  (C , C )
I Similarly Prisoner 2 ranks: (D, C )  (D, D)  (C , C )
Example 1: Prisoner’s Dilemma

I Suppose N = {1, 2}. These players are prisoners.

I Due to lack of evidence, they have been questioned in
separate rooms and made to confess their crimes.
I Each player has two actions: Confess (C) and don’t confess
(D).
I Thus, the pair of actions are: (C , C ), (C , D), (D, C ), (D, D).
I The police give the following situation to the Players.
I If one confesses and other don’t confess, then confessor will
get lesser punishment (get higher payoff).
I If both confess, they will get lesser punishment.
I If both do not confess, they will not be charged and set free.
I Prisoner 1 will rank the actions as (C , D)  (D, D)  (C , C )
I Similarly Prisoner 2 ranks: (D, C )  (D, D)  (C , C )
Example 1: Prisoner’s Dilemma

I Suppose N = {1, 2}. These players are prisoners.

I Due to lack of evidence, they have been questioned in
separate rooms and made to confess their crimes.
I Each player has two actions: Confess (C) and don’t confess
(D).
I Thus, the pair of actions are: (C , C ), (C , D), (D, C ), (D, D).
I The police give the following situation to the Players.
I If one confesses and other don’t confess, then confessor will
get lesser punishment (get higher payoff).
I If both confess, they will get lesser punishment.
I If both do not confess, they will not be charged and set free.
I Prisoner 1 will rank the actions as (C , D)  (D, D)  (C , C )
I Similarly Prisoner 2 ranks: (D, C )  (D, D)  (C , C )
Example 1: Prisoner’s Dilemma

I Suppose N = {1, 2}. These players are prisoners.

I Due to lack of evidence, they have been questioned in
separate rooms and made to confess their crimes.
I Each player has two actions: Confess (C) and don’t confess
(D).
I Thus, the pair of actions are: (C , C ), (C , D), (D, C ), (D, D).
I The police give the following situation to the Players.
I If one confesses and other don’t confess, then confessor will
get lesser punishment (get higher payoff).
I If both confess, they will get lesser punishment.
I If both do not confess, they will not be charged and set free.
I Prisoner 1 will rank the actions as (C , D)  (D, D)  (C , C )
I Similarly Prisoner 2 ranks: (D, C )  (D, D)  (C , C )
Prisoner’s Dilemma

I If one of them confesses, but the other one does not confess,
then the confessed player gets a payoff of 3 but the player
who does not confess gets a payoff of 0
I If both of them do not confess (D), then they can achieve
higher payoffs of 2 each.
I If both confess (C), then they each achieve a payoff of 1.
I What are the strategies in this game?
I The payoff matrix can be written as:
2
C D
C 1, 1 3, 0
1
D 0, 3 2, 2
Prisoner’s Dilemma

I If one of them confesses, but the other one does not confess,
then the confessed player gets a payoff of 3 but the player
who does not confess gets a payoff of 0
I If both of them do not confess (D), then they can achieve
higher payoffs of 2 each.
I If both confess (C), then they each achieve a payoff of 1.
I What are the strategies in this game?
I The payoff matrix can be written as:
2
C D
C 1, 1 3, 0
1
D 0, 3 2, 2
Dominance

I Observe that Prisoner 1 will get higher payoff by playing C

whatever Player 2 plays.
I This is known as dominant strategy.
I A dominant strategy is one which always gives higher payoff
whatever your opponent does.
I It could be weak or strong.
I Similarly, C is also dominant strategy for Prisoner 2.
I Therefore, (C , C ) is a dominant strategy equilibrium.
I This is called dilemma because (D, D) is a better outcome.
Dominance

I Observe that Prisoner 1 will get higher payoff by playing C

whatever Player 2 plays.
I This is known as dominant strategy.
I A dominant strategy is one which always gives higher payoff
whatever your opponent does.
I It could be weak or strong.
I Similarly, C is also dominant strategy for Prisoner 2.
I Therefore, (C , C ) is a dominant strategy equilibrium.
I This is called dilemma because (D, D) is a better outcome.
Price Competition

I There are many games where cooperation is going to be a

superior outcome but players does not play that.
I Examples include price war, arms-race, dispute settlements
with or without lawyers etc.
I Suppose there are two stores are being opened by two retailers.
I They would like to price a toy: they price it either high or low.
2
High Low
High 20, 20 5, 30
1
Low 30, 5 10, 10
Price Competition

I There are many games where cooperation is going to be a

superior outcome but players does not play that.
I Examples include price war, arms-race, dispute settlements
with or without lawyers etc.
I Suppose there are two stores are being opened by two retailers.
I They would like to price a toy: they price it either high or low.
2
High Low
High 20, 20 5, 30
1
Low 30, 5 10, 10
Dominated Strategy

I Let us analyze the following game:

L C R
T (2,2) (6,1) (1,1)
M (1,3) (5,5) (9,2)
B (0,0) (4,2) (8,8)
I is there any dominant strategy is this game? No.
I Observe the payoff of Player 1 if she plays B is 0, 4, 8 and if
she plays M is 1, 5, 9.
I Thus, M strictly dominates B.
I Thus, B is a strictly dominated strategy by M for Player 1.
I A strategy is called strictly dominated if there exists another
strategy which provides higher payoff than other strategy
independent of other players strategy.
Dominated Strategy

I Let us analyze the following game:

L C R
T (2,2) (6,1) (1,1)
M (1,3) (5,5) (9,2)
B (0,0) (4,2) (8,8)
I is there any dominant strategy is this game? No.
I Observe the payoff of Player 1 if she plays B is 0, 4, 8 and if
she plays M is 1, 5, 9.
I Thus, M strictly dominates B.
I Thus, B is a strictly dominated strategy by M for Player 1.
I A strategy is called strictly dominated if there exists another
strategy which provides higher payoff than other strategy
independent of other players strategy.
Rationality and Common Knowledge of Rationality

I Rational player will never play a strictly dominated strategy.

I But can we ignore the strictly dominated strategy of our
opponent when analyzing a game?
I The answer depends not only that I am a rational player but
also on the knowledge that our opponent is also rational.
I Let us consider the following example to understand this
point:
L C R
T (2,2) (6,1) (1,1)
M (1,3) (5,5) (9,2)
B (0,0) (4,2) (8,8)
Rationality and Common Knowledge of Rationality

I Rational player will never play a strictly dominated strategy.

I But can we ignore the strictly dominated strategy of our
opponent when analyzing a game?
I The answer depends not only that I am a rational player but
also on the knowledge that our opponent is also rational.
I Let us consider the following example to understand this
point:
L C R
T (2,2) (6,1) (1,1)
M (1,3) (5,5) (9,2)
B (0,0) (4,2) (8,8)
Iterated Elimination of Strictly Dominated Strategies
I Note that strategy B of Player 1 is strictly dominated by his
strategy M.
I If Player 1 is rational, he will not play strategy B.
I If Player 2 does not know that Player 1 is rational, then he
cannot eliminate B from the support of his belief.
I Suppose Player 2 knows that Player 1 is rational.
I Then he can conclude that Player 1 will not play B.
I Now, strategy R is strictly dominated by strategy L.
I Since Player 2 is rational, so he will not play R.
I Now, if Player 1 knows that Player 2 is rational and
Player 2 knows that Player 1 is rational, the Player 1 will
not play M.
I Because it is strictly dominated by T .
I Similarly, Player 2 will not play C because it is strictly
dominated by L.
Iterated Elimination of Strictly Dominated Strategies
I Note that strategy B of Player 1 is strictly dominated by his
strategy M.
I If Player 1 is rational, he will not play strategy B.
I If Player 2 does not know that Player 1 is rational, then he
cannot eliminate B from the support of his belief.
I Suppose Player 2 knows that Player 1 is rational.
I Then he can conclude that Player 1 will not play B.
I Now, strategy R is strictly dominated by strategy L.
I Since Player 2 is rational, so he will not play R.
I Now, if Player 1 knows that Player 2 is rational and
Player 2 knows that Player 1 is rational, the Player 1 will
not play M.
I Because it is strictly dominated by T .
I Similarly, Player 2 will not play C because it is strictly
dominated by L.
Iterated Elimination of Strictly Dominated Strategies

I Thus, (T , L) is the only strategy profile that survives the

elimination.
I This process is known as iterated elimination of dominated
strategies.
I We will not deal with this topic in detail.
I We assume common knowledge of rationality (CKR) for
using iterated elimination of strictly dominated strategies.
I This means that a fact is common knowledge among
players of a game if for a finite chain of players i1 , . . . , ik the
following holds:
Player i1 knows that Player i2 knows that Player 13 knows
that . . . Player ik is knows the fact.
Example

I Consider the following game:

L M R
T (1,0) (1,2) (0,1)
B (0,3) (0,1) (2,0)
I Iterated elimination of strictly dominated strategies will give a
unique prediction, i.e. (T , M).
I In some games this procedure will lead to the solution of the
game.
I But if there is no strictly dominated strategy in a game, then
we cannot apply this procedure.
I The order of elimination of strictly dominated strategies does
not matter.
Example

I Consider the following game:

L M R
T (1,0) (1,2) (0,1)
B (0,3) (0,1) (2,0)
I Iterated elimination of strictly dominated strategies will give a
unique prediction, i.e. (T , M).
I In some games this procedure will lead to the solution of the
game.
I But if there is no strictly dominated strategy in a game, then
we cannot apply this procedure.
I The order of elimination of strictly dominated strategies does
not matter.
Nash Equilibrium

I Let us consider the following game:

L C R
T (0,6) (6,0) (4,3)
M (6,0) (0,6) (4,3)
B (3,3) (3,3) (5,5)
I There is no dominated strategy in this game.
I How can we predict the outcome of this game?
I We require a concept of stability to solve this sort of games.
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).
Nash Equilibrium
Consider the following thought process:
I If Player 2 knows that Player 1 will choose T , he will choose
L (his best response to T ).
I If Player 1 knows that Player 2 will choose L, he will choose
M (his best response to L).
I If Player 2 knows that Player 1 will choose M, he will choose
C (his best response to M).
I If Player 1 knows that Player 2 will choose C , he will choose
T (his best response to C ).
I If Player 2 knows that Player 1 will choose B, he will choose
R (his best reply to B).
I If Player 1 knows that Player 2 will choose R, he will choose
B (his best reply to R).
I Suppose Player 2 plays R, then B is the best strategy for
Player 1.
I Thus, (B, R) satisfies a stability property.
I In other words, no player has a profitable deviation if the
players choose (B, R).