When Russia annexed Crimea in March, American policymakers were taken by surprise. They shouldn’t have been, argued the political theorist John J Mearsheimer in a New York Times op-ed. After all: ‘Mr Putin’s behaviour is motivated by the same geopolitical considerations that influence all great powers, including the United States.’

Mearsheimer is one of the leading exponents of offensive realism, the theory that international politics has always been, and will presumably remain, ‘a ruthless and dangerous business’. In the absence of a world government that could protect the weak from the strong, all states seek as much power as possible: there is no better way to ensure their own survival. So says the offensive realist.

There are, of course, other ideas in the marketplace of international relations theory (which might itself explain the general confusion over Russia’s recent activity). Liberal theories, for example, tend to downplay the quest for strategic advantage, focusing instead on such internal characteristics of states as their form of government. But, strange as it might seem, such distinctions do not make much difference on the international stage. Democracies and dictatorships appear similarly jealous of power.

 So offensive realism enjoys better empirical support than other theories of international relations. At the same time, it is clear that more is at stake in international politics than naked geopolitical calculus. One limitation common to most realist theories is their assumption that states act as purely rational agents, coldly calculating the course of action that would yield the highest material advantage. In fact, state policy is often influenced by seemingly irrational considerations. No truly rational utility-maximiser could take something such as ‘national honour’ seriously, yet states frequently do.

An injection of evolutionary thinking might help to explain why. Imagine a livestock herder – a traditional Kazakh nomad or an American cattleman on the Western frontier – who lives in a stateless, anarchic society. His wealth is movable and therefore vulnerable to theft. Since there are no police and no courts, he must rely on his own efforts to protect himself, just as states must rely on themselves to ensure continued survival. In such a situation, one strategy is to maintain a reputation for extreme toughness: ‘If you mess with me, you’ll regret it.’ Potential rustlers are deterred because they know that the owner will go all-out to punish them for any transgression.

Now, on a realist view such as Mearsheimer’s, such retribution would seem irrational. It yields no immediate gain and entails significant costs. If one does it oneself, there is the risk of injury or death. If one outsources the work, a bounty must be paid. But in spite of these liabilities, the punishment strategy turns out to be the one that wins in the long run. Herdsmen who do not cultivate a tough reputation become ‘men without honour’. Eventually they lose all their herds and become extinct (indeed, that possibility is what makes this genuinely an evolutionary process, although the relevant adaptation is probably more cultural than genetic). ‘Honour’ means that your commitment to punish a thief is credible. You cannot be dissuaded by danger and you cannot be bought off. If you succumb to either temptation, you lose your credibility, and with it, the capacity to deter robbers.

The problem is that rustlers are also under pressure to cultivate tough reputations: they have to intimidate the herders and deter punishment. So we end up in a coevolutionary arms race in which everybody becomes increasingly tough. The end result is a spiral of violence in which all parties run a high risk of extermination. An apparently sensible strategy leads, in short order, to suicidal madness. This is hard to understand within the rationalist framework of offensive realism. From an evolutionary point of view, on the other hand, it seems inevitable.

When they want to investigate conflict, game theorists have long turned to the classic ‘hawk-dove’ game. ‘Doves’ are individuals who never fight. If attacked, they run away. ‘Hawks’, on the other hand, are always ready for violence and will attack anybody who has something that they want. In a country populated by meek doves, the hawk strategy does very well. But as hawks become more numerous at the expense of doves, they spend more and more time fighting and killing each other.

There is, however, a simple modification of the hawk strategy that is superior to both hawks and doves: playing ‘bourgeois’. First, you declare a resource item – a herd, a piece of cropland – as your private property (hence the ‘bourgeois’ designation). Then you signal that you are willing to defend it no matter what it takes. Again, this is not rational in the narrow sense. You must be willing to escalate conflicts to the point where your life is at stake, even though your life is worth incomparably more than the disputed property. But again, in evolutionary terms, the strategy is a winner. While the hawks overreach, getting embroiled in self-destructive conflict, the bourgeois steadily divide the spoils among themselves, fighting only to defend their property against hawks. In the long run, the bourgeois always replace the hawks.

What does this mean for the seeming irrationality of states? Well, typically, they fight over territory. Land supports a population, which provides the state with taxes and army recruits. It can also have strategic value, if it allows the state to project power or control a choke point. And, of course, states are essentially territorial entities: without land, they are nothing...