Untitled Document

Pols 426 Lecture 5
R. Keohane, After Hegemony

Keohane claims that while there is considerable cooperation in the international system - as the global economy develops - out of this growing interdependence comes increased potential discord via governmental intervention and policy conflict

The central question for Keohane is - How can cooperation be organized in the world political economy when common interests exist?

Keohane takes mutual interests of states as given - does not ask where they arise or how created -
But wants to examine the conditions under which they (mutual interests) lead to cooperation

Keohane begins with the premise that even when common interests exist cooperation often fails

Keohane makes an empirical point - He begins by noting that there is far more cooperation in the international system (trade, financial relations, health, environmental protection …) then can be accounted for by the Realist perspective but he also suggests that those who take an Institutionalist perspective and argue that shared state interests create a demand for international institutions and rules and cooperation run the risk of being naïve about power and conflict and can overestimate the ease with which cooperation can be achieved and maintained in the international system

Institutions for Keohane are recognized patterns of practice around which expectations converge --- and these practice affect state behavior allowing for cooperation - further international regimes are arrangements for policy coordination created by the fact of interdependence - they are made up of rules, norms, principles, and decision-making procedures

Now - Keohane argues that realist and institutionalist predictions about the state of the international system during the twenty years after WW II are similar and "correct" - Both predict cooperation - Realists because of the presence of a political hegemon (The U.S.) and because of the demand for coordination due to interdependence - But as after the mid 1960's and the decline in U.S. hegemony, their predictions should diverge - institutionalist predict more cooperation and realists predict less cooperation.

While Keohane suggests that the Realist approach provides a better fit for the 70s and early 80s, he argues that there is still more cooperation than would be predicted without a hegemon

So the key questions are

1. Under what conditions can independent nations cooperate in the world political economy?
2. Can cooperation take place without a hegemon and if so how?
3. Specifically how do patterns of rule-guided policy coordination emerge, maintain themselves, and decay in world politics

Cooperation and International Regimes

Keohane begins with the assumption that international cooperation is valuable but difficult to ogranize --- in international economic and political relations we cannot rely on the market to work

Keohane argues that "cooperation occurs when actors adjust their behavior to the actual or anticipated preferences of others, through a process of policy coordination

Now what is cooperation -- it is not harmony - a situation where actors would pursue policies without regard for the interests of others to facilitate the attainment of others goals - harmony is a situation where actor interests happen to be in alignment

Cooperation - is a situation where actor policies would hinder the attainment of the goals of others but through a coordination process policies are adjusted so as to obtain cooperation rather than discord if polices were not adjusted -- behavior patterns are altered via coordination to achieve cooperation

International Regimes and Cooperation

Keohane argues that regimes exist and they can affect the likelihood of even "egoistic" nations to cooperate

An international regime is "sets of implicit and explicit principles, norms, rules, and decision-making procedures around which actors' expectations converge in a given area of international relations."

Principles are beliefs of fact causation and rectitude.
Norms are standards of behavior defined in terms of rights and obligations
Rules are specific prescriptions or proscriptions for action.
Decision-making procedures are prevailing practices for making and implementing collective choice.

An example - the international balance of payments regime

Principle - liberalization of trade and payments
Norm - the injunction to states not to manipulate their exchange rates unilaterally for national advantage
Rules - pegged exchange rates and procedures for consultation in the event of change - and later floating exchange rates

The various aspects of regimes - contain injunctions about behavior and they imply obligations though they are not enforceable

Keohane claims that regimes should not be viewed as new elements beyond the state centered international system, but rather as arrangements motivated by self interest - the key

Regimes help coordinate behavior but even so, rational individuals (nations too) who would benefit from cooperation often are unable to do so - they fail to coordinate their actions

To get at this problem Keohane starts by assuming that states are rational egoists and argues that most relations of interest in international politics can be characterized by a combination of mutual dependence and conflict, of partnership and competition --- or in the terms of game theory "mixed-motive games" -- situations where nations can mutually gain from cooperation but also have incentives to not cooperate (unilaterally) and gain more

The Prisoners Dilemma Game
Player B
Cooperate Defect
Player A Cooperate R,R (3,3) S,T (1,4)
Defect T,S (4,1) P,P (2,2)
Payoff ordering T>R>P>S

Public goods and collective action situations

PD Game is a problem because trust to cooperate even if communicated is difficult to establish because commitments are not binding - talk is "cheap" so if individuals or nations were to play the PD game once it is "rational" to defect - and players end up in a mutually undesirable situation (P,P).

Why - Defection is a "dominant" strategy and mutual defection is an "equilibrium solution"

But Keohane offers two "ways" that actors - here nation-states" might avoid the defection outcome of the dilemma

1. He notes that while actor are egoistic (self-interested), they nonetheless are not asocial. They exist within some social situation and in that they have shared experiences, ethical precepts, and expectations of future actions with identifiable individuals - "culture" even in anarchic but regime ridden international relations

The mafia example

2. Repeated Play --- the shadow of the future and Iterated Prisoners Dilemma - can be rational to cooperate - why not individually maximizing to forego long sequences of mutual cooperation (strings of 3,3 payoffs for one or a few temptation payoffs by unilateral defection (4,1) - because these are followed by long strings of mutual defection

When can the collective action problem be solved so as to have cooperation without a hegemon (one that can enforce all others to behave) -

1. when the number of agents is relatively small so all can monitor the behavior of others
2. when agents expect to interact for a long time
3. when agents can identify those that they are playing with
4. when agents can (and will) punish defectors

Hegemonic Cooperation in the Post Cold War Era

Recall - he argues that rational self-interested actors in situations of interdependence will value International Regimes as a way of increasing their ability to make mutually beneficial agreements

The question Keohane asks is how does hegemonic leadership operate -- or how does the hegemon construct international regimes that facilitate the right kind of cooperation from the standpoint of the hegemon itself

1. a necessary condition - sufficient military power to protect the Int pol economy from incursions by hostile powers
2. the hegemon then seeks to persuade others to conform to its vision of world order and to defer to its leadership

vision - international capitalism -free trade, property rights, and democracy (sort of) -- to get support for this vision and U.S. leadership, followers had to see that they received sufficient benefits from accepting or buying into this order and leadership

three key benefits followers received from partnering in the U.S. lead regime

1. a stable international monetary system - to facilitate international trade and payments -- U.S. role was to manage the monetary system
2. Provision of open markets for good - U.S. worked to reduce tariffs and remove discriminatory restrictions
3. Access to oil at stable prices - U.S. and U.S oil companies provided oil to Europe and Japan from the Middle East and from the U.S. itself

Hegemonic leadership by U.S not based simply on dictating terms but based upon providing a set of incentives to participants via formal trade and monetary regimes and a narrow company-based oil regime with occasional independent action

Keohane argues that the strategy worked in the short term - helped European and Japanese economy recovery and growth and global growth and the cohesion of the U.S. military alliance against the Soviet Union but
failed eventually because it did not institutionalize an oil regime that could have warded off rising threats to access at stable (cheap) prices and the regime did not maintain a strong resource base for the exercise of U.S. power

Keohane more generally argues that hegemons and the powerful can postpone adjustment to change and often do so and sometimes refuse to adjust to change until it is forced upon them - Keohane blames success and domestic interests gaining special privileges for not allowing adjustments to change and thus decline

Cooperation without Hegemony - Cooperation without a hegemon is more difficult but does happen so HOW
Material power matters but
Keohane argues that cooperation in mixed motive setting rests on 1. Expectations 2. Transactions costs 3. Uncertainty

Multilateral institutions and regimes must furnish the certainty and confidence that hegemons provided

As hegemony erodes the demand for regimes persists as the supply declines

Keohane argues that the international regimes in trade and money were sufficiently developed to provide for cooperation whereas oil collapsed

Lecture on Wendt -- Three Culture of Anarchy

Anarchy in international relations - two central questions

1. Are there different types of anarchy? --Yes

Three types - three structures where different roles - enemy, rivalry, and friend - dominate the international system

2. Does the international system construct state? -- Yes

The international system not only affects states behavior (ala Realists and Institutionalists) but also their identities and interests -

Wendt suggests a third way of theorizing and understanding International politics

1-realist - top-down where the structure in no way depends on its elements
2 - institutionalist - bottom-up where anarchy is reducible (can be completely explained by) its elements

The third way - structure/agent mutual construction - IR as a social process where the structure does construct the elements but that structure has several logics or - cultures - the enemy, rival, friend variants

Bringing the social or cultural components in

Roles not just at the unit based level (i.e. nation states) but as properties of the structure --- the culture of IR is based upon a structure of roles

How roles fit in and work in IR

Role structures have to do with the configuration of subject positions that shared ideas make available to its holders. Wendt claims that subject positions are constituted by representations of self and other as particular kinds of agents related in particular ways. (think of roles in various cultural systems - churches, schools, politics)

He notes in most cultures roles are functionally differentiated BUT he claims anarchy makes it difficult to sustain role asymmetry until violence is mitigated so they argues at the core of each kind of anarchy is one subject position

Enemy - in the Hobbesian culture
Rival - in the Lockean culture
Friend - in the Kantian culture

Each involves a distinct posture between self and other with respect to violence
Enemies - threatening adversaries who observe no limits on violence towards each other
Rivals - competitors who will use violence to advance interests but refrain from killing each other
Friends - friends who do not use violence against each other and who cooperate against security threats

Anarchic systems vary depending upon which of the three roles dominate the system

The Hobbesian culture - enemies

Emnity - long history

Greeks - Persians as barbarians
Crusaders saw the turks and infidels
Europeans treated the Americans as savages
Recent genocides - Holocaust, Cambodia, Bosnia, Rwanda

Implications for foreign policy when enemies posture prevails

1) states will try to conquer or destroy enemies even if they have status quo interests
2) policymakers heavily discount the future and worst case scenarios dominate
3) relative capability is crucial
4) wars will have not limits on violence

high risk aversion, short time horizons, and relative power

A no holds barred power politics world

System characteristics

Lots of war
Unfit actors disappear - empire building and reduction of the number of units
No neutrals - alignment required

The Lockean culture -rivals

Rivals act as if others respects their right to sovereignty and some sense of "property" - not to conquer or dominate but to gain advantage --- Sovereignty as an institution is recognized

Implications for foreign policy when rivalry posture prevails

1) States act in status quo fashion to other's sovereignty
2) Sovereignty makes security less scarce, future matters more, absolute gains matter more than relative loses
3) Relative mil power matters but less so than in Hobbesian world
4) If war occurs, rivals limit violence

System characteristics
1) warfare excepted but constrained - wars do involving killing people but not states; when states are conquered they are restored - WWI, WW2, -- Iraq occupation
2) number of system members remains relatively stable
3) states balance power
4) neutrality and nonalignment are acceptable stances

A Waltzian world -- and Westphalian like -

The two phrases are interesting here

Hobbesian - surviving in a kill or be killed world
Lockian -- getting along (doing well or better) in a live and let live world

The Kantian Culture - Friendship

Friendship is a role structure where states expect each other to observe two simple rules

1) disputes settled non-violently
2) they will fight as a team if security threatened

friends more than allies in the traditional realist sense - links to pluralistic security communities (friends) and collective security (rival) pluralistic security concerns security within a group - the latter security between a group and a rival or other

while war remains a logical possibility, it is not considered a legitimate way of settling disputes

conflict and disputes arise among friends - US - Canada example but
what matters in not relative military capability as in the case of rivalry but
other kinds of power -- discursive, institutional, economic

The structure of anarchy - Hobbesian, Lockean, Kantian - depends on the distribution of ideas or roles

Lecture on Wendt and the Experiments

Wendt argues that there are three types of anarchy - three structures where different roles - enemy, rivalry, and friend - dominate the international system

IR as a social process where the structure does construct the elements but that structure has several logics or - cultures - the enemy, rival, friend variants

Bringing the social or cultural components in

Enemy - in the Hobbesian culture
Rival - in the Lockean culture
Friend - in the Kantian culture

Now the connections to the experiments you participated in is by the very nature of Wendt's arguments limited

1) We cannot make or turn subjects into enemies or for that matter friends - we can make them rivals
2) We cannot have subjects inflict violence on each other and try to eliminate each other so we cannot address some aspects that Wendt refers to that differentiates among the three cultures of anarchy
3) But we can create some environments which put subjects into a structure that more resembles enemies, rivals, and friends

Remember the Hobbesian culture -- enemies

The Hobbesian culture - enemies

Wendt claims subjects have

high risk aversion, short time horizons, and relative power

Hobbesian - surviving in a kill or be killed world

The Lockean culture -rivals

Sovereignty makes security less scarce, future matters more, absolute gains matter more than relative loses

Lockian -- getting along (doing well or better) in a live and let live world

The Kantian Culture - Friendship

Friendship is a role structure where states expect each other to observe the following behavior

they will fight as a team if security threatened

friends concern with security within a group - the latter security between a group and a rival or other

The Experiments

Three aspects were manipulated

1. Communication
2. The structure of the conflict
3. Objectives of the subjects that I will link to different cultures

We know from Keohane and a vast literature in game theory that

When players can communicate that this helps players overcome fears, help coordination of actions, overcome mistakes and thus makes cooperation more likely to occur in games of conflict and coordination than when they do not. Think about communication, eve if it is cheap talk, as providing useful information among the players and as providing more of a social setting

So regardless of the structure of the conflict situation, communication ought to make cooperation more likely than when there is no communication.

The likelihood of cooperation is highly dependent upon the structure of the conflict situation. Some structures are far more "nasty" than others - So among the game structures we talked about we can rank them by how difficult they are for players to achieve a cooperative outcome

Most difficult Deadlock then Chicken, PD, Stag, and Assurance

So we would expect subjects in our experiment to cooperate more when the structure of the situation was Stag than Chicken

Finally ow do we try to represent different role relations - enemy, rival, friend

I used objective of the game as a surrogate -- I cannot make the subjects enemies

In one experiment the objective was "Have the group (sum of all players) gain more points than the other three groups" Here I tried to have the subjects think that they were working together to maximize points in a competition against the other three groups - So the idea was to instill a we/they sense - our groups against the other - thus engendering a "friends" like culture in the experimental group of 6

In a second experiment the objective was to "Maximize your total points" Here the effort was to create a sense of maximizing absolute gains --- do the best you can without respect to others - a rivals like culture

In a third experiment the objective was to "Have more points than any other player in your group" here the effort was to create a sense of maximizing relative gains - a do better than the other players (win) relationship --- an enemies culture

So, Given Wendt's argument we would anticipate that subjects would cooperate most in the friends culture, next in the rivals culture, and the least in the enemies culture

So this sets up the following experimental design with expected results with respect to the level of cooperation we would anticipate across the various experiments

Three conditions - Culture - Friends (F), Rivals (R), Enemies (E)
Structure of conflict - Chicken (CH), Stag (ST)
Communication - Talking (T), No Talking (NT)

Experiment Group 1 Group 2 Group 3 Group 4
1 F, CH, NT F,CH,T F,ST,NT F,ST,T
2 R,ST,T R,ST,NT R,CH,T R,CH,NT
3 E,CH,T E,CH,NT E,ST,T E,ST,NT

Two big Groups (A and B) X four groups (1,2,3,4), X three experiments X 21 rounds -- 192 rounds of decision - Each round has 6 players making decisions with five other players or 30 decisions per round or 5760 decisions in total

To Determine whether the level of cooperation varies across the three conditions what I did was calculate the total number of cooperative choices and defection choices made by all 6 players per experiment

So for example For Group A, Group1, Experiment 1 with was a "case" of Friend, No talking, Chicken across the 7 rounds of the game the players made 125 cooperative choices and 85 defection choices for a rate of 59% cooperation

We can then sum these cooperation rates across the various groups and conditions to see if there is a difference in levels of cooperation across conditions

So - Does Culture matter?

Culture Just Group A Just Group B Groups A and B Combined
Friends 86.25% coop 95.5% coop 90.9 % coop
Rivals 81.75 % 89.6% 85.7%
Enemies 64.4% 69.5% 66.9%

These results demonstrate that culture (or objective) has a strong effect on levels of cooperation

A 24% difference in levels of cooperation between friends and enemies
19% more cooperation between rivals than enemies
5% more cooperation between friends than rivals

Does Communication Matter?

Communication Just group A Just Group B Group A and B combined
Talk 79% coop 85.5% 82.5%
No talk 76% 83.8% 79.9%

These results suggest that there is really not that much difference between the two conditions - Talking and No Talking - Communication helps players cooperate marginally -- 2.5% more cooperation with communication

Does the Structure of Conflict Matter?

Game Structure Just Group A Just Group B Group A and B Combined
Chicken 71% Coop 83.1% 77.1%
Stag 84% 86.5% 85.25%

As anticipated there is substantially more cooperation in Stag games - 8% more on average - than in Chicken Games -- a significant difference but perhaps not as much as we might expect

So - culture (or its surrogate player Objective) matters considerably -

Player cooperate much more when they are working as a team (friends) against others and least when they are trying to maximize relative gain - as close as we can get to enemies

Being able to communicate - creating a more socially rich environment helps some

Cooperation is more likely when the structure of the game is more conducive to cooperation - Stag rather than Chicken

For Fun

How did your group "rate" in terms of cooperation

Group Group A Group B
1 70% coop **** 95% ****
2 75% 83%
3 72% 82%
4 92% 79%

The real outliers -- Group A - Group 1 59% coop Friends treatment -- really low
Group A -- Group 3 60% coop Rivals treatment low
Group B -- Groups 2 82% coop Friends treatment low
GroupB -- Group 1 87% coop Enemies treatment - really

Some preliminaries for the iterated prisoners dilemma game simulation (SimSociety)

Recall the PD game
Player B
Player A Coop Defect
Coop c,c c,d
Defect d,c d,d

Where 1. c,c is mutual cooperation
2. c,d A cooperates and B defects
3. d,c A defects and B cooperates
4. d,d both defect - mutual defection

Now we can characterize a set of IPD game strategies by indicating the probabilities that a player will cooperate given each of the 4 outcomes (noted above) of the game from the last or prior iteration and denoting for the very first iteration whether the player will choose to cooperate or defect so lets denote

1. P(1) as the probability of cooperating on the current iteration given that c,c was the outcome from the last or prior iteration
2. P(2) as the probability of cooperating on the current iteration given that c,d was the outcome from the last or prior iteration
3. P(3) as the probability of cooperating on the current iteration given that d,c was the outcome from the last or prior iteration
4. P(4) as the probability of cooperating on the current iteration given that d,d was the outcome from the last or prior iteration

Thus as strategy can be defined as

[ P(1), P(2), P(3), P(4)]

with the addition of a cooperate or defect choice stipulation on the first iteration

So a strategy that always cooperates (All-c) can be represented as - cooperate on the first round and
[1,1,1,1]

A strategy that always defects (All-D) can be represented as - defect on the first round and
[0,0,0,0]

Some classic strategies

Tit-for-Tat (TFT) cooperate on the first round and then choose whatever the opponent chose in the previous round
[1,0,1,0]

Pavlov - win stay - lose change - if you like the result from the last round continue doing what you are doing - if you do not then change
Cooperate on the first round and then
[1,0,0,1]

Grim - cooperate on the first round and then cooperate as long as the opponent cooperates - If the opponent defects then defect from that point forward

Generous TFT cooperate the first time and with 100% prob reward prior cooperation by the opponent - either c,c or d,c and with some probability cooperate after you instigated defection c,d and if there is mutual defection d,d
[1, r,1,r] where r is between 0.0 and 1.0

Mean TFT - defect on the first round - always punish prior defection and sometimes but not always reward prior cooperation
[q,0,q,0] where q is between 0.0 and 1.0

Some important characteristics of strategies

We can characterize strategies as

Nice (N) if they are never the first to intentionally defect
Nasty (A) if they are never the first to cooperate
Retaliatory (R) if they immediately defect after an unprovoked defection
Forgiving (F) if they have the propensity to cooperate after the other player has defected
Exploiting (E) if they intentionally defect while the opposition cooperates

All-C -- N,F
All-D - A,E
TFT - N,R,F
Pavlov - N,E,R
Grim - N,R
GTFT - R,F,E
MTFT - R,E

What kinds of strategies must agents use to survive and succeed in a world of anarchic self interested agents

How does cooperation develop in such a world - Why? What strategies seems to make this work and why?

What kinds of strategies seem to be able to maintain cooperation?

A guide to doing research on cooperation in Simsociety

We are interested in three types of questions and maybe others

1) Under what circumstances does cooperation arise in SimSociety
2) Under what circumstances is cooperation maintained in SimSociety
3) What strategies are successful in generating and maintaining cooperation and why?

In effect you will use SimSociety to run a set of experiments. Therefore you will want to establish some "lab" conditions.

I suggest the following basic research structure

1) Run the basic experiments with 60 agents (you may want to vary this later for fun)

2) Leave all setting (except the agent strategies) at their default settings - that is leave them alone - -again you may want to play with them later and yo an ask me about ones you find interesting

3) I suggest running your basic simulations for 20,000 iterations - this will not take very long

4) To address the question of if and how cooperation can be generated in an un-cooperative, anarchic world do the following experiment

Do a minimum of five runs of each of the following mixes

a) 50 All-D and 10 All-C
b) 50 All-D and 10 Grim
c) 50 All-D and 10 TFT
d) 50 All-D and 10 Pavlov
n based on your results and analysis how much to you have to adjust the "strategy mix" to consistently generate cooperation for each of the four cooperative type strategies

5) To address the question of if and how cooperation can be maintained do the following experiment

Again do a minimum of five runs of each of the following strategy mixes

a) 50 All-C and 10 All-D
b) 50 Grim and 10 All-D
c) 50 TFT and 10 All-D
d) 50 Pavlov and 10 All-D

Based on your results how sensitive are these results to increases and decreases in the number of All-D agents

As a variation you might see what happens with a more nasty, greedy exploiter -- Mean TFT vs the four

What features of strategies appears to make them successful or not in maintaining cooperation (that is what kinds of characteristics do they have) and how does this relate to Keohane's arguments about generating cooperation in mixed motive situation

Pols 426 Lecture 7 R. Keohane and After Hegemony and The Simulation SimSociety

Keohane argues that cooperation can be achieved and maintained by the presence of international regimes; principles, norms and rules guiding nation-state (agent) behavior. The various strategies of the iterated prisoners dilemma can be viewed as rule-based behavior and that norms or institutions of bilateral interaction may be represented as homogeneous systems of this rule based behavior. For an institution to exist in a population, it is not that all actor strategies are identical but rather that some behavioral characteristics are common to all members. Under these circumstances, these behaviors become norms and are institutionalized.

Many of the different kinds of structural situations of conflict and cooperation that nation-states find themselves in can be captured by the following set of (2X2) games: PD, Chicken, Stag, Assurance, Deadlock.

Prisoners Dilemma

The PD is a situation where agents' decisions are motivated by fear, greed, and a lack of trust; motives commonly attributed to groups acting on behalf of nation-states by scholars of international relations and repeated PD games have been applied to such phenomena as arms races and arms control arrangement.
Chicken

International crises are often characterized as Chicken games though typically single shot affairs. Bargaining about trade agreements with threats of sanctions and the possibility of trade wars are often characterized as repeated play games of Chicken.

Stag

Groups of states that impose economic embargos or sanctions find themselves in something like a stag situation. If all adhere to the sanctions, then the chances of making the state or states facing the sanctions act in proscribed ways are high. But some may find it in their interest to break the economic embargo and trade with the embargoed state while all others do not.

Assurance

International agreements such as the Convention on Rights of the Child or treaties such as Ottawa Treaty to Ban Landmines, Kyoto Accords on Global Warming, and the Nuclear Non-proliferation Treaty appear to be structured like assurance games.

Deadlock

Before the outbreak of the Gulf War in 1990, the United States was bent on protecting its oil supplies, forcing Iraq from Kuwait, and maintaining the balance of power in the Middle East. On the other hand, Iraq was apparently determined to maintain its occupation of Kuwait even in the face of what seemed to be a likely military defeat. Both sides preferred to "win" but both sides also preferred armed conflict to negotiation or settlement; an example of deadlock.

Remember some important characteristics of strategies

We can characterize strategies as
Nice (N) if they are never the first to intentionally defect
Nasty (A) if they are never the first to cooperate
Retaliatory (R) if they immediately defect after an unprovoked defection
Forgiving (F) if they have the propensity to cooperate after the other player has defected
Exploiting (E) if they intentionally defect while the opposition cooperates
All-C -- N,F
All-D - A,E
TFT - N,R,F
Pavlov - N,E,R
Grim - N,R
GTFT - R,F,E
MTFT - R,E

Assessing Levels of Cooperation across the Various (2X2) Games

To determine whether expected variations in cooperation for the five (2X2) game structures noted earlier occur, thirty simulations of the agent-based model were run for each of the five game structures. Outcomes of the simulation runs fall into one of four possible categories: stable cooperation (SC), no cooperation (NC), punctuated equilibrium (PE), and stable cooperation at the .5 level. As noted earlier, a stable cooperative outcome occurs when a high level of cooperation (the average cooperation rate among agents is over 95%) is achieved at some point in the simulation run and is maintained until the end of the simulation. A non-cooperative outcome occurs when the average cooperation rate quickly declines and stays at less than 5% for the entire simulation run. A simulation run is categorized as an instance of punctuated equilibrium if, after stable cooperation is achieved, it is followed by periodic (one or more) massive dips to near universal defection, A simulation run is categorized as a case of stable cooperation at the .5 level when the average cooperation among agents reaches 50% and is maintained for the duration of the simulation run.

The simulation results reported in Table 1, not surprisingly, generally conform to expectations. First, stable cooperation fails to evolve in Deadlock. Second, stable cooperation is significantly more likely to evolve in Assurance and Stag games than in PD or Chicken games. Third, while high levels of cooperation are achieved with considerable frequency in PD games (combining the stable cooperation and punctuated equilibrium categories) and Chicken games (combining the stable cooperation, punctuated equilibrium and cooperation at the .5 level categories), they are far more fragile than in Stag and Assurance games and usually collapse. Lastly, the switching (DC and CD) .5 cooperation level equilibrium does occur with considerable frequency (53%) in Chicken games.

Most scholars of international relations would not be surprised by these long-term simulation results. However, it is a bit surprising that these cooperative "regimes" are so stable since agents have some incentives to defect. Once cooperation is achieved it does not collapse, a result that most scholars would be skeptical of even in settings where incentives to cooperate are very high. In more high-risk contexts such as PD and Chicken, where the incentives to defect are high, cooperation is still achieved at fairly high rates for both PD (87%) and Chicken (90%). However, it is very fragile and collapses 57% of the time in PD settings and 83% of the time in Chicken settings.

In the simulations, fifty exploitive All-Defect (All-D) agents and ten cooperative agents employing three types of cooperative strategies-TFT [1, 0, 1, 0], Grim [1, 0, 0, 0], and All-Cooperate (All-C) [1, 1, 1, 1]-are randomly distributed on the grid of the agent-based model. An agent employing the All-D [0, 0, 0, 0] strategy defects whenever it plays another agent for the first time. If it interacted with an agent in the previous round, it always defects the next time regardless of the prior joint outcome. An agent employing the All-C [1, 1, 1, 1] strategy cooperates the first time it interacts with another agent and always cooperates the next time it interacts with that agent regardless of the prior joint outcome. As noted earlier, an agent employing the TFT [1, 0, 1, 0] strategy cooperates the first time it interacts with another agent. However, it cooperates the next time only if the other agent cooperated the last time [following a (CC) or (DC) outcome]. It will defect the next time if the other agent defected during the prior interaction [following a (CD) or (DD) outcome]. An agent employing the Grim [1, 0, 0, 0] strategy cooperates the first time it interacts with another agent. It cooperates for all subsequent interactions with that agent as long as the agent cooperates. Once the other agent defects just once, it will always defect whenever it interacts with that agent again.

To determine whether differences in the ability of cooperative agents to invade a large group of exploitive agents exist, five simulations of each strategy mix (i.e., fifty All-D and ten TFT) were run for each of the five different game structures. Each simulation was run for 200,000 iterations and these results are reported in Table 2. The outcome of each simulation run is placed into one of three possible categories: SC, NC, PE. These outcomes are defined in the same fashion as earlier.

Overall, across the five games and three types of cooperative invaders (All-C, TFT, and Grim) there is a 63% chance that cooperative agents were able to successfully invade and essentially eliminate all exploitive All-D agents. Simulation outcomes categorized as either SC or PE indicate that high levels of cooperation were achieved and that cooperative agents successfully invaded. Cooperative agents successfully invaded 100% of the time in Assurance games, 80% in Stag games, 67% in both PD and Chicken, and 0% in Deadlock. Given the payoff structure of Deadlock, it is hardly surprising that cooperative agents (Grim, TFT, and All-C) have no success in invading All-D and creating cooperation. Nor is it surprising that invasion is more successful when agents employ Grim and TFT strategies because, unlike the All-C strategy, these strategies punish defection.

What kinds of strategies must agents use to survive and succeed in a world of anarchic self interested agents?

First, cooperation in an anarchic, self-help world where there is both opportunities for conflict and cooperation and where there is long term interaction (shadow of the future) is quite likely to occur. But it is not guaranteed to either happen at all and if achieved to last; it can collapse -- So in Keohane's terms cooperation without a hegemon is certainly possible.

Second, the transition from near universal defection to near universal cooperation is always characterized by two features; 1) networks of cooperative agents form and 2) Grim strategies seem to be most successful

Grim is a rather odd strategy to act as the catalyst or transitional strategy to more cooperative worlds populated by Generous TFT or Pavlov. Like TFT, Grim is nice and
not nasty. But, unlike TFT, it is not very forgiving. It appears to succeed in invading the world of agents that are essentially almost pure All-D because it is provocable, retaliatory, nice, and only slightly forgiving (DC probabilities range from .05 to .3). It is the characteristic of being basically unforgiving that appears to allow Grim to build cooperative networks that can withstand invasion and fragmentation by All-D. When networks are small and fragile, agents that employ strategies that are too forgiving are dangerously susceptible to invasion, constituting weak links in the network.

The emergence of small, relatively stable nodes or networks of agents characterize every transition from uncooperative to cooperative worlds. The networks of cooperative agents form what could be considered to be primitive social networks or the basic foundations of "communities." The interactions within these communities are almost universally cooperative but, when defection occurs, it is immediately met with reprisal with little chance for contrition for DC or CD results. As the nodes become large enough so that cooperative neighbors surround some agents, those surrounded agents receive multiple cooperative payoffs per turn.

Cooperation requires a primitive form of organization and community and a very tough stance toward non-cooperative behavior - Retaliation and non-forgiveness are important features - International regimes, in order to survive need some "mass" and organization and they must punish free riders and potential exploiters - All must punish exploiters in place of the hegemon and
forgiveness at least initially cannot be tolerated without a hegemon.

Third, high levels of cooperation characterized by large clusters of cooperative agents may collapse. When cooperation collapses, it is usually because the networks of cooperative agents evolve into networks made up of agents with strategies that are "too nice." The typical pattern that leads to collapse is the upward drift in the probability of cooperating following a c,d outcome. In these cases, average c,d probabilities drift up to over .8. These agents, having a strategy mix such as [.99, .85,
.10, .07], can easily be taken advantage of allowing nasty strategies to emerge and succeed. The generation of too many nice strategies makes successful exploitation too easy leading to the collapse of cooperation. We can liken this situation to the "classic predicament of societies going 'soft' --- Nice agents like and succeed in a very cooperative world where almost all interaction is cooperative and they "forget" over time that they must punish defection --- particularly when they cooperated (c,d) but also after mutual defection (d,d) -- Pavlov does well in cooperative environments - indeed by being exploitative (defect after d,c), but because it does not punish mutual defection, it is itself subject to exploitation --- Regimes that have a norm of reciprocity and also one where defections are consistently sanctioned even when it is costly to the particular agent that must sanction are much more likely to last - maintain cooperation

Non-cooperative worlds can be invaded by nice agents but they must be willing to retaliate. Forgiveness is ok (TFT) but greed coupled with openness to being exploited (Pavlov) is dangerous because it can prevent successful invasion and/or allow nasty agents to collapse an attained cooperative world.

Summary: The results indicated that norms of niceness, and retaliation must be institutionalized if cooperation is to emerge and be maintained. Norms of nastiness do not generate cooperation - agents (nation-states) must be willing to initiate cooperation. Cooperation cannot be generated or maintained by institutions characterized by exploitation unless they also are nice. Cooperation cannot be generated or maintained by institutions characterized by niceness and forgiveness unless they are also coupled with retaliation.