Continued from part 1.

Then Gaus turns to coordination problems like Stag Hunt / Assurance game (213-15), which (he should add) also involve an independent dimensions of CAPs. It consists in an interaction-situation having more than one equilibrium, at least one of which is not P-optimal, so that arriving at the (or one of the) P-optimal equilibria requires coordination. It is crucial that in many of these cases, a mere convention like stopping at red lights rather than green lights can suffice; nothing deeper need underlie it. Sometimes the natural “salience” of certain phenomena, places, or things does underlie it (red, being associated with blood, is perhaps naturally alarming / arresting).

Gaus makes the familiar point that both coordination problems and social dilemmas illustrate why “injustice social rules” (212) may often be needed to facilitate mutually productive social life (213). To this he adds the plausible point that some imaginable rules for such CAP situations might be worse, from everyone perspective, than no rule at all (the “z” or null rule). “Socially eligible rules” are all those “better than z” from all relevant perspectives (211). Ineligible rules, we might say, are deal-breakers for at least some of the interacter. “All benefits should go to the person with whitest skin, whereas that person should not need to make any contributions to the effort” would clearly be such a deal-breaker, for example.

So far so good. But then Gaus wants to apply his framework of CAPs analysis at the second-order, to imaginable social and/or moral rules themselves. He focuses in on the socially eligible rules (better than z for all) that are also not Pareto-dominated by other eligible rules, labeling these (a bit misleadingly) as “the socially optimal eligible set” (214). In fact he argues against Kurt Baier’s conception of moral social rules as those held (by some or all) to be the best ones to regulate CAP situations (215-17) (Baier, of course, was influenced by Kant’s mistaken belief, following Rousseau general will, that “justifiable to all” would pick out a single candidate policy for each kind of interaction). Instead, Gaus wants to argue that justified rules will be simply those, among the socially “optimal” set on which the parties are “able” to coordinate (224-25). What he means, I think, is to pick out rules among the eligible-and-non-dominated subset that prove to be in equilibrium in a higher-order game of jockeying for different rules – although he sometimes words it as if he means that they emerge, probably via some interpersonal planning, as salient points to overcome a coordination game at this higher-order level itself. Or he could mean that they emerge from Nash Bargaining among the rule-options. These will be rules justified in his radically revised satisficing sense of a social contract.

This is where things go south (no offense to the actual South intended). This is apparent from the impure coordination game he uses as his main example in this section. This is a classic “Battle of the Sexes” game, which I prefer to call “Battle of Friends,” as that is both less sexist and flags the desire to coordinate. Let’s imagine, Gaus implies, that the selection of rules of justice is like a vast Battle of Friends game at the scale of a society of a million people or more. Now this may well be the most realistic way of running his second-order game analysis, because the different eligible-and-non-dominated candidate rules will both (a) facilitate coordination, overcoming the challenge of non-optimal equilibria to which lack of stable expectations can lead, and (b) advantage some people and disadvantage others relative to the division of costs and benefits they could get from their society adopting other eligible rules instead.

Yet then Gaus’s account will be in trouble. For Battle of Friends scenarios involve both coordination problems and that pesky distributive equity dimension: the different eligible rules that conceivably could overcome the coordination challenges will also distribute differently – often very differently among the participants. That some may also be in (near)-equilibrium does nothing to allay resentment that may arise from their perceived unfairness to significant parts of the population. Chicken scenarios also have this dual property of coordination problems plus distributive inequity (harder to overcome in Chicken cases). In both, and in other coordination games that are “impure” because of their distributive equity aspects (such as “Rambo games” – see Katarina Holzinger), the participants need some prior consensus on what DEP to apply. It is clearly not enough that chosen rules are P-optimal (not dominated). That would be true in a second-order ultimatum game to: group A picks the rules, and says to all other groups: now accept these rules or else it is civil war.

The fatal problem, then, is that Gaus’s contractors would need substative DEPs to determine which set of feasible rules to adopt. Nash Bargaining among the possible rules is a way of “solving” distributive equity problems that appeals merely to might making right. The z-“threat point” is the only motive for agreement on the part of those who feel the rules thereby adopted are (distributively) unjust to them. That results in what Rawls called a “mere modus vivendi.” But substantive DEPs for Gaus’s proposed second-order game of rule-choosing cannot themselves be justified in the same way as fair rules to overcome the distributive equity problem in such a game. Are we supposed to imagine that they emerge in a third-order game? The infinite regress is obvious on this horn of the dilemma.

On the other, we are left with the conclusion that fundamental DEPs for choosing rules to govern the basic institutions of society cannot themselves be constructed in a social contract as conceived by Gaus. Put so generally in this way, though, the same applies to other versions of social contract theory as well. That is the fundamental problem that Gaus did not fully diagnose in prior social contract theories. He got close, but in the end, his own proposal falls into the same trap. Nash Bargaining is only acceptable for scenarios in which it has been approved on the basis of more basic principles not “justified” this way. And we cannot “just choose” one eligible batch of rules that cover all relevant situations, because that would assume that randomizing is the right DEP for the higher-order game. Nothing this fundamental could be so arbitrary and produce rules publicly seen as justified by all (or most).

So I think Gaus’ project runs aground here. If convinced by this, I think he would say it says there is no justified way to arrive at a moral social contract; skepticism would be the conclusion. For he thinks he has foreclosed the route that Rawls took. But I’m not so sure. I think cases for different DEPs can be made on susbstantive pre-contractarian grounds based on human experience, intuitions, and fundamental aspects of culture itself.

Think back to Rawls’s idea that the function of justice is to reduce the extent to which life-prospects turn mainly on chance of birth, and other brute luck. This is in effect a negative ur-DEP: it says “don’t chose rules of justice that, as a package, fail dramatically to lower this dependence on brute luck for most people, relative to what is possible with current technology and further advances that just rules can help to promote.” Although Rawls does not say it, this intuition (on which the Original Position is based) has going for it that it captures one of the essential functions of culture. It gives us something substantive to start from in selecting equitable rules for the common life of agents with otherwise quite diverse conceptions of a good life and (among other things). I’m not saying this is enough by itself, but these are the sorts of starting points that could underpin a successful contractarian approach that does not pretend that agreement or coordination can do all the heavy lifting.