This excerpt is from the same book as the previous Axelrod piece, but is offered here as an empirical application of the theoretical concepts offered in this section on markets. Along with the ants excerpt it's parallel to the Cohen & Vandello and Willis pieces in previous sections.
Robert Axelrod (born 1943) is a professor of political science and public policy (his actual title is "Professor of Human Understanding") at the University of Michigan. His work on human cooperation, complexity, and emergence makes use of game theory, mathematics, and agent-based modeling.
1. What's the basic real world scenario here
2. What are the research questions here?
- How could live and let live get started?
- How is it sustained?
- Why did it break down at end of war?
- Why did it happen in this war context but not others? (186.8)
3a. What is the source material for the article? What is it based on?
Axelrod used extensive archival research by sociologist Tony Ashworth (187.1). Diaries, letters, reminiscences, mostly English but also some French and some German.
3b. Who are the sides in the PD game here? What are the "moves"?
Two opposing military units, battalions of about 1000 men. They can "shoot to kill" or they can shoot to avoid causing harm (187.3)
4. What are the outcomes?
The components of the outcomes are to stay alive or die, and to gain or lose advantage for future fighting. More specifically, one can (187):
- stay alive and gain advantage
- stay alive but not gain advantage
- die and not gain advantage
- die and lose advantage
5. Arrange these payoffs and moves in a standard game matrix/table.
No Shoot Shoot No Shoot Live w/o gain, Live w/o gain Dead & lose advantage, Live & gain advantage Shoot Live & gain advantage, Dead & lose advantage Dead w/o gain, Dead w/o gain
No Shoot Shoot No Shoot R, R S, T Shoot T, S P, P
R = Live w/o gain
S = Dead & lose advantage
T = Live & gain advantage
P = Dead w/o gain
6. Show that these outcomes meet the ordering criteria for PD (see 178-179).
Live w gain > Live w/o gain (T > R)
Dead w/o gain > Dead & lose advantage (P > S)
Live w/o gain > Dead w/o gain (R > P)
Therefore : T > R > P > S
The second ordering criteria is that players cannot just alternate treating one another poorly (mutual cooperation has to trump alternating exploitation 179.1)). Here that would mean "live and let live" has to be better than "we kill you one day and gain advantage and you kill us the next day and gain advantage."
The expected payoff of alternating is the "average" of "die and lose advantage" (S) and "live and gain advantage" (T). How does that compare to live without gain?
The average of live and die should be less than live. The average of gain advantage and lose advantage is equal to no gain.
Therefore R > (T + S) / 2 (See 187.5)
7. What important elements of social/organizational context made this into a "two person game"?
Members of a battalion knew one another by sight, the unit was big enough that the enemy "knew who did it" if they were aggressive, and small enough for informal control to exist among members. Battalions faced one another across a few hundred yards and so were mutually identifiable as "teams."(187-188)
8. In what sense were the players involved in "repeated play"?
The pace of the war was such that battalions could end up facing one another for weeks at a time (188.3)
9. Early on, there were sometimes "direct truces" — what are these? And what happened to them?
Shouts, signals, flags would mark times (meals) or places (latrines) that would be off limits to snipers. The generals found out about it and suppressed these techniques. On the allied side, at least, it was partly because the global strategy was one of attrition — even if they lost the same number of soldiers as the Germans, the Germans would run out of soldiers first. This strategy was less attractive on the local level (188-189)
10. What were some ways that reciprocity could spread?
- Time: restraint exercised during certain hours could be extended
- Type: One kind of restrain could lead to experiments with other types.
- Space: What worked in one area could spread to other areas (189.5).
11. What were some of the elements of behavior used to sustain cooperation?
- Demonstration of sniper prowess showed that the lack of killing was not due to incompetence ("there really is a 'let live' thing going on here!")
- Response to violations (defections) was often severe — suggest high value placed on cooperation ("don't mess with this!!") (189.7)
12. How was the cooperation sustained in the face of changes in personnel?
"Socialization" during overlap (190.3)
13. What is the difference between artillery and infantry (190.5)?
Artillery refers to big guns (cannons, etc.) that are typically BEHIND the front lines. Infantry refers to the guys on foot, in the trenches. The point here is that it's the artillery that suffer the most when the fighting is fierce. They are the ones getting shot and blown up.
14. So, what's the relationship between artillery and infantry? Why does the organizational/structural relationship threaten to undermine the cooperative outcome?
The artillery don't directly suffer consequences of escalation of fighting. If they "do their job" the "truce" breaks down and the infantry suffer. So, the tendency was for the infantry to treat the artillery guys very well and to socialize them into non-action. Their agreement would be "we won't lob any bombs UNLESS YOU want us to" which was a way of saying "we will not threaten the balance, but if there needs to be a response to a violation of the truce, we'll let 'em have it" (190.6).
15. In our reading of Weber on organizations, one of his interesting observations was the constant challenge of getting subordinates to obey orders and to know whether or not they were obeying orders.(The theme of how organizational supervisors can know whether subordinates are conducting themselves "on behalf of the organization" (that is, doing what they are paid to do) is a constant one in the social sciences. Erving Goffman has a famous description in a section called "The Recalcitrant Self" in the essay "The Underlife of a Public Institution" (in Asylums). In economics we talk about the "principal-agent problem." The core of all of these are divergent interests and asymmetric information.) Recount Axelrod's description of the challenges the generals faced trying to control these "truces."
High command could ascertain whether orders were being followed during big battles, but between battles they could not really monitor what was happening. And "[t]he soldiers became expert at defeating the monitoring system…" (190.8)
16. What finally destroyed the live and let live system?
The "raid" (190.9). Small groups would be sent to kill or capture enemy in their trenches. Since there was direct evidence of raid, no truce possible. And since raids could happen at any time, everyone had to be constantly vigilant and ready to fight back.
17. "Damping" is mentioned on both 190.2 and 191.3. What is damping?
Damping is a term borrowed from physics. It refers to the tendency of a system to slow its response to disturbances rather than to allow disturbances to amplify. Here it refers to factors that prevented a retaliation from motivating an even bigger counter retaliation and so on in a spiral of revenge ("an uncontrolled echo of mutual recriminations" (190.2)). During the live and let live phase, if A breaks the truce and B launches a nasty response, A understands what motivated the response and just backs off. Later, when raids were being used, the generals could control the retaliation so it was just "raids as usual" and so the other side could not use a harsh retaliation raid to say "look, this is what will happen if you raid us, so knock it off!"
18. How does this quote, "Without realizing exactly what they were doing, the high command effectively ended the live-and-let-live system by preventing their battalions from exercising their own strategies of cooperation based on reciprocity" (191.6) support Hayek's claims about centralized decision making?
19. At 191.8-9 Axelrod says that the evolution of cooperation we see in this case was not about "blind mutation" and did not involve "survival of the fittest." What does he mean and why is this important?
Blind mutation refers to random changes without planning or intention. That's not what happened here — the soldiers came up with their behavioral variants by thinking about what was going on. Survival of the fittest would mean that entities that engaged in suboptimal behaviors would not survive. Here the "entities" are the battalions and the behaviors are either the cooperative ones or not. But even units that don't do live-and-let-live lose more soldiers, each one is replaced and so the unit itself survives. (192.2)
20. In what sense did "ethics" emerge?
Players in a game in which mutual restraint has emerged experience sense of value in the rule and in the other's trust that one intends to follow the rule. And raids generated an ethic of revenge. "[B]oth cooperation and defection were self-reinforcing" (192.7).
"…[N]ot only did preference affect behavior and outcomes, but behavior and outcomes also affected preferences" (192.8).
21. In what sense does ritual emerge?
Last problem was 0480 (HELP)
Q236. In a prisoner’s dilemma game, the rational thing for both players is to defect. This makes mutual defection an equilibrium, though it is not a preferred one (the collective would be better off with another outcome). In other words, in a single game of prisoner’s dilemma, cooperation is “impossible.” But cooperation does happen in the world.
Demonstrate your understanding of Axelrod's ideas by describing the mechanism and conditions under which this can happen without assuming anything “social” about the agents.
Explain how this works and how it adds to or modifies Smith’s and Hayek’s story about how markets can be a source of social order.
Q425. What research questions did the "live and let live" phenomenon from WWI raise for Robert Axelrod?
- 0069 Coordination, define
- 0070 Coordination and cooperation, examples
- 0196 Prisoner's Dilemma
- 0197 Prisoner's Dilemma
- 0198 The "tit-for-tat" strategy
- 0199 Game Theory: Lame Ducks
- 0200 Game Theory: Equilibrium
- 0210 Hayek: "Cosmos and Taxis"
- 0213 Hayek: "Cosmos and Taxis"
- 0217 Axlerod: Cooperation
- 0218 Axelrod: "tit-for-tat"
- 0233 Hobbes, from "Leviathan"
- 0234 Hayek "Cosmos & Taxis"
- 0236 Axelrod: "The Evolution of Cooperation"
- 0238 Fehr and Gintis