Why Do Good Guys Win Wars?
Author: Kristian Helland-Hansen. Tutors: Håvard Hegre (PRIO) and Brynjar Lia (FFI)
This article is an unfinished first draft, which I hope will nevertheless stimulate the reader.
Abstract
The article maintains that norms of civility may affect behaviour in armed conflicts, even if the behaviour is not individually rational, nor rational to the group. Moreover, I argue that such norm-driven behaviour may nevertheless achieve a greater benefit to the group and its individuals than strict rationality could. The reason for this is that norms may “lift the horizon” and make people act in accordance with their long-term interests rather than their short-term interests. The units of analysis are two Algerian Islamist rebel groups.
Introduction
As Henri Dunant witnessed the powder-smoke slowly evaporating over the low hills of Solferino, uncovering thousands of mutilated bodies of men and horses, some still screaming, his mind began resonating with one overpowering thought: There is no way we can achieve a world without war. Therefore wars must be civilized.
This thought was not new. In Medieval Europe, war had indeed developed into an “art.” The ruthlessness and efficiency of the Swiss infantry pikemen in the early Renaissance shocked the nobility of Europe which had for some time engaged in largely ceremonial battles involving a host of chivalrous rites and flying colours – and rather few casualties. But whereas the pikemen are renowned for their efficiency in the battlefield, they are not associated with pillaging and indiscriminate brutality. Before that, limited warfare had been practiced by Saladin, Sun Tzu, Alexander the great, Nebuchadnezzar and King David. Although the conception of non-combatants has varied, and although history is as rife with barbarity as it is with chivalry, there have always been soldiers who regard some potential targets as inviolable. And those soldiers have often been victorious.
In this article I wish to focus on two questions: Firstly, are norms of civility at all likely to affect conduct in war? Secondly: If so, will that automatically make the actor who abides by these norms more vulnerable? The structure of the article consists of an initial theoretical argument in favour of total war. I then turn to what I consider a critical case, namely the Algerian civil war. I compare two rebel groups which have over time varied in their approach to the principle of discrimination, and trace their military success over time with that in mind.
Theory: Why It Pays to Be a Bad Guy.
Why in the world would it pay to pursue limited warfare?... And if the case for limited warfare cannot be credibly cast in strictly instrumental terms, is it then conceivable that any military decision-maker could be persuaded to abide by rules of civility? After all, holding prisoners of war is much more expensive than killing them. Protecting civilians of the enemy opens up the prospect of espionage – and it takes away one major incentive for fighting: the spoils of war.
Initially, there would seem to be two arguments for limited warfare:
1. If soldiers engage in barbaric acts, their morale is corrupted and they lose their self-respect, motivation and perhaps their will to survive.
2. If one’s soldiers are known to treat combatants and non-combatants alike, that takes away the incentive on the enemy’s side to surrender.
In fact, while these arguments may both seem appealing, they are derivable from two opposing schools of thought, which build on diametrically opposite premises.[1]
The first, which I shall refer to as Humanism,[2] postulates that man is essentially a conscientious and peaceful creature. If he or she ever commits a violent act, that is seen as an abnormality - something which needs to be explained. This school is predominant within psychology, medicine, anthropology, and partly in sociology. Depending on the educational background of the analyst, explanations of violence will focus on (electro)chemical reactions in the nervous system, genetics, childhood- or more recent experiences, frustrations, culture, psycho-sociological mechanisms etc.. The common denominator is that man, conceived of as a moral subject, is driven to do what he does, i.e.: an inherently causal/historicist approach.
The other perspective was coined by Morgenthau (1948) as political realism.[3] Realism postulates that man is an amoral, free and rational, actor, characterised by ability to rank his interests, to choose, and to pursue his own happiness by all means. He is operating in an anarchic world of limited resources, where he can trust no-one. Seeking power is not only logical, it is necessary. It is not violence that riddles realists, but peace, and explanations of it will typically focus on the balance of power (Morgenthau 1993:183ff). Intentional, not causal explanations are characteristic of this perspective. According to the realist school, man will do what he pleases unless deterred. Thus, man as a rational actor would be pro-active, future-oriented, calculating, and communicating through his actions. Conversely, man as a subject would be, re-active, past-oriented, passionate, and not consciously sending messages through his acts. As for the question of how to respond to threats or violence, realists would warn against appeasement, while humanists would argue that it is necessary to somehow address the grievances of the other.
Interestingly, contributions within peace research are rarely explicit about the presumptions they make about the basic units of analysis - individuals. Initially, the quest for “the causes of war” would seem to place the discipline within the humanist tradition. The designs are often quantitative and based on aggregate data, concomitant with a causalist approach. On the other hand, the theories tested and the explanations given very often relate to individuals - or groups which are treated as collectively rational units.
Humanists will have relatively little problem accepting that norms of civility may affect individual behaviour, although they will expect the impact of norms to be reduced under stress, deprivation and pain – common facets of war. Realists-rationalists, on the other hand, would only expect norms to have a moderating effect in so far as it pays for the individual to abide by them. Just like humanists, realists would assume that moderate, civil, behaviour would be less prevalent in situations of acute crisis, albeit by a different causal path, and they would add the probability of being caught redhanded as a crucial control variable. Essentially, my project in this article is to convince realists of the attainability and virtue of civilised warfare. I therefore build the argument below on realist premises and then destroy it by reference to the empirical material. In order to do this, I need to “filter out” counter-arguments in terms of point 2 above, namely those that pertain to the incentives of the enemy to surrender. I do that by focusing on situations in which surrender is irrelevant.
Does it follow from a realist understanding that it always pays to be a bad guy? In so far as you value winning above mutual conciliation and simultaneously value mutual confrontation above surrender, my answer will be yes. Like political realism, game theory works on the presumption of rational unitary actors. The preference structure indicated above is captured in the “game” known as the prisoners’ dilemma. The name of the game refers to a scenario where two arrested criminals are held in separate cells. They have previously agreed on a cover story, and the district attorney has only got enough evidence to give them a light sentence. However, the police offer the following choice:
- Incriminate your comrade, and serve an even lighter sentence, or;
- Do not inform us, and take all the blame if your comrade chooses to tell.
In this situation, there are four possible outcomes:
Table 1: Individual and Collective Payoffs in a Prisoner’s Dilemma Game
Outcomes: / Value of outcome to actor / Sum of valuesPrisoner A / Prisoner B
Both stick to the story / 3 / 3 / 6
Prisoner A tells on prisoner B / 4 / 1 / 5
Prisoner B tells on prisoner A / 1 / 4 / 5
Both tell on each other / 2 / 2 / 4
From the collective point of view, it would be rational if both stuck to the story. However, each prisoner has a double incentive to “cheat”, because whatever the one does, the other would gain from telling. (If both tell on each other, no-one is the sucker). As a result, both prisoners give away evidence without which they could not have been sentenced so hard. Notably, in a Prisoners’ dilemma, knowledge about the other actor’s preference structure will not have a moderating impact on one’s own choice of strategy. If actor A knew, that actor B would co-operate, his incentive to cheat would only be greater. This means, in fact, that given the preference structure above,[4] defiance always pays.
Both game theorists and neo-liberal institutionalists have tried to modify this perverse logic. The various arguments can be summed up in George Tsebelis’ (1990) concept of “nested games.” Essentially, nested games imply that actors are involved in more than one game at a time, and that draws in one game may affect the actor’s position in other games. Since it is the total outcome of the games which is important, the actor may concede or postpone benefits in single games in order to maximize his/her overall output. Games may be linked in time or in space.
As for games linked in time, it would refer to situations in which the actors expect several encounters, where one’s draw in this game will frame the other’s expectations for future rounds. Theoretically, it is quite possible that moderation in this game would pay, if it could induce the other to apply a similar approach in later games. Game theory allows for discounting future, potential, gains against immediate ones. It is common to operate with a discount factor, δ, which is usually treated as equally regarded by the parties, i.e.: both actors weigh immediate vs. future gains similarly. The discount factor is multiplied with the potential payoffs of later rounds, allowing for comparison between what could be gained now, and what could be gained later, if moderation is chosen now. Conventionally,
T stands for temptation, i.e.: the payoff of unilateral defection (D-C)
R stands for reward, i.e.: the payoff of bilateral co-operation (C-C)
P stands for punishment, i.e.: the payoff of bilateral defection (D-D)
S stands for sucker’s treat, i.e.: the payoff of unilateral co-operation (C-D)
Thus, for two rounds, in which A defects in both, and in which B co-operates in the first, and retaliates for A’s defection by playing D in the second, the payoffs would (from the vantage point of immediately before round 1) be:
Table 5: Payoffs to Actors A and B over 2 Rounds in Which A Cheats in the First Round and B Retaliates
Actor / Round 1 / Round 2A / T / δ P
B / S / δ P
The discounted payoffs of infinite rounds can be calculated (Morrow 1994:265;316f). It can, for instance, be shown that the payoff of infinite games of full co-operation would be
R+ δR + δ2R +δ3R +…. = R/(1- δ), if 0 < δ < 1.[5]
Actors may use threats of various types of retaliation in order to discipline their opponents. Morrow (1994:264f) refers to the threat of all-time defection as the grim trigger, and the threat of punishing each round of defection with defection in the next round as tit for tat. Naturally, the grim trigger has the largest potential for deterrence, but, once initiated, there is no going back to a co-operative framework, unless at the expense of a severe credibility loss. Morrow (1994:265) shows that defection will be deterred by the grim trigger if,
δ > (T – R) / (T – P)
and by tit for tat if,
δ > (T – R) / (R – S)
It follows from this, that, the chances of co-operation increase if,
- δ increases, i.e.: if the value actors place on future payoffs increases;
- T is reduced, i.e.: if less can be achieved through unilateral defection/defiance;
- R increases, i.e.: if more can be gained through co-operation/conciliation;
- P decreases, i.e.: if the state of mutual defiance becomes harsher;
- S decreases, i.e.: if the cost of restoring co-operation becomes less bearable.
How does all this relate to the question of total or limited warfare? Mearsheimer (1995:11) argues that the extreme risks involved in international politics, will force actors to adopt a myopic outlook, and place greater priority on the present rather than the future. This argument holds even stronger for manifest war, and, it would seem, no less for intra-state wars than for inter-state ones.
Is it possible to think of situations in which civility would pay in the sense that it would increase the chances of long-term benefits? Clearly, if the war can be conceived of as a battle for the hearts and minds of the populace, it might pay in the long run to nurture a sense of sympathy by behaving moderately. On the other hand, one could argue with Hobbes (1904), that the faculty to rely on is fear. According to this logic, excesses today would enhance the payoffs in later rounds. It is probably correct to say that the conventional view today is that the whip gives stronger incentives than the carrot.[6]
Limited warfare in this context would seem to imply a rough understanding of who is the “enemy’s civilians” and who is “our civilians.” Let us for the sake of illustration presume that civilians are those who do not take up arms against the enemy, and that they therefore are the least belligerent. Let us furthermore presume that in any conflict, there are two sides, which are each divided between hawks and doves. The hawks are those who wish to maintain the conception of two monolithic and irreconcilable sides, while the doves are the pragmatists who are torn between their identification with their hawks on the one hand and their desire to end the conflict quickly on the other. The doves have two possible alliances: They can realign themselves with their hawks, or they can form a new identity through a compromise with the doves of the other side. The hawks on the other hand have only one possible alliance – that with their own doves.