The concept of international anarchy is well known: in the absence of a world government, the various states of the world relate to each other under conditions of anarchy. There is no ultimate authority to resolve disputes or enforce law among the governments of the world. Hence any cooperation between states must be cooperation under anarchy, without the intervention of a world government.
While it is commonly recognized that there is anarchy between governments, it is less well known, but equally true, that there is anarchy within governments. That is, just as the agents of different governments are in a state of anarchy vis-à-vis each other, so too the agents of the same government are in a condition of anarchy in their relationships with each other.
For example, under international anarchy, an agent of the Canadian government and an agent of the U.S. government interact without a world government governing their relationship.
Now consider two agents of the U.S. government. In their interactions with each other, they cooperate without some higher government ruling over them. Call this intranational anarchy: the agents of a single government are in a state of anarchy vis-à-vis each other. (See Cuzán’s paper “Do We Ever Really Get Out of Anarchy?“.)
On the classic Hobbesian account, cooperation and social order are not possible under anarchy. Without some ultimate third party authority to force people to respect property rights and fulfill contracts, cooperation is impossible and society breaks down into chaos.
When we observe governments and intranational anarchy around the world, the extreme Hobbesian prediction is clearly wrong. The internal relations of most governments, though anarchic, are fairly peaceful, and politicians do not battle each other in a war of all against all. So the question is: given that the internal relationships of a government are anarchic, how can cooperation and social order within a government be secured?
Can the logic of repeated interactions and reputation be used to explain social order in intranational anarchy? The answer seems to be yes.
The number n of the true rulers of the state, those who control its coercive power, appears to be small enough for repeated interaction to generate cooperation. And in running the government, the rulers are in fact in repeated interactions with each other, so cooperation would be the dominant strategy.
In addition, the nature of the organization of the state can explain cooperation. By cooperating to run the government, the rulers can enrich themselves by taxing the general population; by comparative advantage, this would be their most profitable employment. Hence a sort of honor among thieves brings incentives to cooperate.
One problem with this explanation is drawing the boundary between the rulers and the ruled. How do we identify the individuals that make up n? At what point does a government agent cease to be a ruler and become one of the ruled? A general in the military is definitely one of the rulers. A low-level bureaucrat in the post office is definitely one of the ruled. So the dividing line must be somewhere between the two. It seems that control over the coercive power of the state is the key. The more such control a government agent has, the more they are part of the rulers.
In sum, thinking about how government agents cooperate in intranational anarchy can illuminate our understanding of the state. One question for future research is explaining how the police and the military, as agencies of organized violence, can cooperate with each other.