THE CONSTITUTIONAL DILEMMA
Part II showed that conditional and reciprocal co-operators meet the primary criterion of procedural rationality: they can be coherently implemented. None the less, two large doubts remain about these players because they are extremely rigid. In order to assure each other of their trustworthiness, they are locked into exceptionless rules. Elster (1986b, p. 120) sees their problem as:
a perennial dilemma of individual behaviour. How is it possible to ensure at the same time that one is bound by rules that protect one from irrational or unethical behaviour—and that these rules do not turn into prisons from which it is not possible to break out even when it would be rational to do so?
This dilemma is familiar from debates over constitutional limits to state action. Citizens seek the assurance of limits over the sovereign power but these limits may prevent the state from adapting to changing conditions. This constitutional dilemma affects our players as well. Rigidly committed players can assure some others but the range of their co-operation is limited by a basic co-ordination problem. For example, selfsame co-operators differing only in the syntax of their principles will fail to co-operate. This suggests that in order to be substantively rational, agents must be more flexible. At a minimum, they should be able to co-operate with a range of agents. Most likely, they will need to be able to adapt their constraints to the needs of their social environment. Now we face the other horn of the dilemma. I can only change my principles if my constraints do not fully bind me. If I am free to change I am also free to exploit your