Mathematics and politics
Lecture notes, 2/27/03
Announcements
- More homework solutions on the homework archive page.
- Next week: Nadia Masri guest-lectures.
- Current list of escalation projects:
- U.S.-U.S.S.R. Cuban Missile Crisis (10)
- Israel-Palestine (4)
- U.S. Civil War (3)
- U.S.-Colombia (2)
- World War I? (2)
- World War II? (2)
- U.S.-China spy-plane incident (1)
- U.S.-Vietnam (1)
- U.S.-U.K. War of 1812 (1)
- U.S.-U.K. American revolution (1)
- U.S.-Iraq Kuwait invasion (1)
- U.S.-Panama: "Just Cause" (1)
- NATO-Serbia: Kosovo war (1)
- India-Pakistan (1)
- U.S.-Cuba: Bay of Pigs (1)
- Napoleon's reign? (1)
- Way too many Cuban Missile Crises! Some other suggestions:
- U.S.-France ("Quasi-War" of 1798-1800): Negotiations prevented a war
- Iraq-Kuwait (events leading up to August 1990 invasion)
- U.S.-Nicaragua (1980s): Contra war funded secretly, overt war avoided
- U.S.-Iran (1979-1980): Iranian revolution, hostage crisis
- Greece-Turkey (Cyprus, 1967): war avoided
Current assignments:
- Homework assignment on game theory, due on Tuesday, Mar. 4:
Mathematics and Politics, pp. 35--42: 1, 3, 4, 8, 9, 10, 11, 13, 14, 16
- Read Sections 7.1-7.3 of Mathematics and Politics.
Today: More Game Theory and Politics
- Nash equilibrium
- Chicken
- Arms races (Prisoner's Dilemma)
- Nuclear war (Chicken)
Nash Equilibrium
- In a two-by-two ordinal game, a Nash equilibrium is a pair of strategies
for both Row and Column such that:
- If Row changes her strategy and Column keeps her strategy, Row will do worse.
- If Column changes her strategy and Row keeps her strategy, Column will do worse.
- In a Nash equilibrium, neither player has any incentive to unilaterally change her
strategy.
- However, the players might have an incentive to simultaneously change their
strategies.
- Thus a Nash equilibrium is "stable." If two players agree to that outcome, neither has an incentive to cheat.
- Example: Show that in Prisoner's Dilemma, the only Nash equilibrium is (N,N),
where both players testify.
- Notice that if both players switch from (N,N) to (C,C), it's better for both.
But they would not do this sequentially.
- Example: Find all the Nash equilibria in Arbitrary Game, whose matrix is
- More examples:
Generate four random two-by-two ordinal games and find the Nash equilibria, if any.
The game of Chicken
- In Chicken (as seen in "Rebel without a Cause"),
two players drive cars off a cliff.
- The first one to get out is the loser.
- The second one to get out is the winner...
- Unless he doesn't get out, in which case he's the big loser.
- [Video clip]
- We can model this as a two-by-two ordinal game if we make the following assumptions:
- The options available to each player are: get out early (E), or stay in the car (S).
- The rankings for each player are the same.
- e.g. James Dean prefers most: he stays in the car and Buzz gets out early.
(Then he still has time to get out and win.)
- Next James Dean prefers both players getting out early.
(It's a draw.)
- Next James Dean prefers himself getting out early and Buzz staying. (Yes,
he loses, and it's humiliating, but at least he survives.)
- James Dean's last preference is if both players stay in their cars. (Both die.)
- A more common alternative is if both players drive their cars into each other, but
this is equivalent in our model.
- Here is the matrix for Chicken, in this model.
- Question: is there a dominant strategy for either player?
- Question: is there a Nash equilibrium?
- Answers: No and yes. (On blackboard.)
- Nash equilibria are if one player stays and the other escapes.
- Compare to Prisoner's Dilemma, where Nash equilibrium is mutual noncooperation.
- The difference between Chicken and Prisoner's Dilemma:
- Similarity: noncooperating (staying in the car) against a cooperating player is best.
- Similarity: mutual cooperating is second best.
- The difference:
- In Prisoner's Dilemma, mutual noncooperating is third best (both get five years
instead of one player getting ten years).
- In Chicken, mutual noncooperating is worst (both die instead of both humiliated).
- Both games are used to represent political situations in which it might be in both
countries' interests to cooperate, but individual greed makes them noncooperate (for a
worse overall result).
Example: arms races
- Consider nuclear weapons development by India and Pakistan.
- Each has a choice: build nuclear weapons (B) or don't build nuclear weapons (D).
- Neither knows what the other is choosing.
- India's preferences (highest to lowest):
- Have nukes, and Pakistan doesn't have nukes (B,D); India "wins"
- Neither side has nukes (D,D); no threat => peace
- Both sides have nukes (B,B); mutual assured destruction => "peace"
- Not having nukes, with Pakistan having nukes (D,B); India "loses"
- Pakistan's preferences are obviously the same.
- Draw the matrix, with India as Row and Pakistan as Column.
- Notice: this game is exactly the same as Prisoner's Dilemma.
- (Mathematical abstraction makes things easier!)
- So what happens?
- Both India and Pakistan have dominant strategies: noncooperation.
- So both countries will build nuclear weapons.
- So this model predicts the actual outcome.
Example: nuclear war
- Now we change the question.
- Suppose there is a dispute in Kashmir.
- India threatens to go to war if Pakistan doesn't stop funding Kashmiri separatists.
- Pakistan threatens to go to war if India keeps repressing Kashmiris.
- The options for each country are to continue its policy (C) or yield (Y).
- Both sides know that nuclear war will result if neither backs down.
- What are the preferences?
- India prefers continuing its policy while Pakistan yields (victory).
- Next highest is both sides yielding (peace).
- Next highest is Pakistan continuing and India yielding (defeat).
- Last is both sides continuing (nuclear war).
- Pakistan's preferences are the same.
- Draw the matrix:
- This is the same game as Chicken.
- We know there are no dominant strategies.
- But the Nash equilibrium is for one side to continue its policy and for the other side
to yield.
- So crises tend to get defused...?
More examples of game theory in politics
- Consider the current situation, U.S. vs. Iraq.
- U.S. has the option of either going to war (WAR) or not (PEACE).
- Iraq has the option of building weapons of mass destruction (WMD) or not (DISARM).
- Let's suppose: The U.S. prefers war over peace. If there is a war, the U.S. prefers Iraq to have weapons of mass destruction (to provide justification). If there is no war, the U.S. prefers Iraq not to have WMD.
- Let's also suppose: Iraq prefers peace over war. Without the threat of war, Iraq prefers not to have weapons of mass destruction (cheaper that way); but with the threat of war, Iraq does prefer WMD.
- Model this as a two-by-two ordinal game.
- Does either side have a dominant strategy?
- What are the Nash equilibria?
- (On board)
- U.S. has dominant strategy of war. Iraq has no dominant strategy.
- Nash equilibrium is that the U.S. fights a war and Iraq has weapons of mass destruction.
- Consider the Israel-Palestine conflict.
- Let's say Israel can choose to either build settlements (B) or dismantle settlements (D) in the West Bank and Gaza.
- The Palestinians can choose to fight the intifada (I) or negotiate peacefully (N).
- Suppose Israel prefers that the Palestinians negotiate rather than fighting the intifada, but otherwise prefers building settlements over dismantling.
- Suppose the Palestinians prefer that Israel dismantle settlements. If Israel dismantles the settlements, Palestinians would certainly prefer peaceful negotiations. If Israel builds settlements, Palestinians would prefer to fight.
- Model this situation as a two-by-two ordinal game.
- Are there any dominant strategies?
- What are the Nash equilibria?
- (On board)
- Israel has a dominant strategy of building settlements.
- The only Nash equilibrium is that Israel keeps building settlements and the Palestinians continue the intifada.
Larger ordinal games
Consider the following three-by-three ordinal game. Row and Column both have three options:
"Yes," "No," or "Maybe."
|
|
|
|
| (1,6) |
(7,5) |
(5,3) |
| (9,2) |
(3,8) |
(6,9) |
| (2,1) |
(8,7) |
(4,4) |
|
Find the dominant strategies, if any, and find the Nash equilibria, if any.
There are no dominant strategies. The Nash equilbria are at (No, Maybe) and (Maybe, No).