There are a lot of good things to say about Gerald Gaus’ book, The Tyranny of the Ideal (Princeton University Press, 2016). It is a difficult work because it operates mostly at a meta-theoretical level, focusing on properties and problems of “ideal” theories of justice in general – although there is quite a bit of commentary on Sen’s theory and Rawls’s approach in Political Liberalism and after. Still, it contains may insights on these topics, and especially epistemic difficulties in discerning what ideal justice actually requires. But I will not focus on many of the good points here, simply in the interests of space.

A particularly striking strength of the work is the way that it raises serious problems for moral contractarianism beyond those which I believe are already implied by the finding in Arrow and social choice theory that there is no one uniquely privileged “general will” we can construct simply from individual preferences. Gaus puts a real dent in Rawls’s program in Political Liberalism. His summation in ch. V, like the section 4 of ch. IV on imperfect coordination, was very helpful. But I think the main problem that this work actually reveals for contractarianism is not what Gaus says – or he has not followed it completely through – as I will explain below.

I can work towards an explanation of this lacuna by starting with another problem related to Gaus’s own positive proposal, which is a collection of different “open societies” each pursuing a “local optima” in the landscape of justice according to substantial overlap among each society’s pursuers of justice on the most relevant goods and institutions in terms of which their debates should be framed. The conversation across these open societies would, Gaus hopes, make it easier to see blind spots in each local optimum and perhaps allow each to see a bit farther towards an ideal for justice lying in a direction that may not be exactly where any of these societies are pointing is their pursuit of their perceived local optimum. For Gaus defends the key thesis that sociopolitical justice is (at least) two dimensional, which is why Sen’s “climbing approach” alone will be insufficient.

The problem is that Gaus follows Ryan Muldoon in interpreting such an “interspectival justice” as “a rational bargain or compromise” (175), or what Rawls called a mere modus vivendi, because he thinks this is the best we can get. But Gaus seems not to understand that rules arrived at that way cannot command the respect needed to make them moral rules rather than mere conventions or artifices justified by nothing more than the bare need for coordination. As David Potts put it in an email exchange on Gaus, my critique is that moral rules need to have a sort of absolute status that they cannot have if we treat them as negotiating points. Rules of property, legal obligation, fundamental rights, and the like, have to command an allegiance strong enough for people to follow them against their interests, to be willing to enforce them with serious punishments, and so forth. But this is incompatible with their being rules that we accepted merely for the benefits of living in a community.

Throughout the book, Gaus often implicitly minimizes this distinction, while still recognizing it here and there (e.g. on 212). My contrary view does not require that justice has no function or technical roles (183), but only that its coordination function is derivative or dependent on it have a meaning not justified by that function alone (without which it could not play this role).

This leads to the deeper problem, which I believe Gaus inherits from game-theoretic conceptions of collective action that were most prominent from the 1980s up to roughly 2010 – or at least, among more philosophical and moral-psychological receptions and deployments of such theories of collective action problems (CAPs). The deficiency is that, while these theories eventually became clearer about social dilemmas and coordination problems as two distinct aspects or dimensions of CAPs, they have been slower to recognize what I will call distributive equity problems as a third, distinct dimension – and to appreciate the full significance of this aspect of CAPs [see my discussion of Gauthier and Joseph Heath in “Why Habermas Needs Distributive Equity Principles: Heath’s Critique, Game Theory, and Collective Action Problems,” Constellations 26 no.2 (2019): 268-285]. For example, I believe this is generally true of Elinor Ostrom’s important work, to whom Gaus refers.

Gaus refers to social dilemmas, including (of course) prisoner’s dilemmas, a couple times in this book (see esp. 179). As he notes in response to James Buchanan, social dilemmas do not require that the agents are only selfishly motivated (180). However, as Gauthier stressed in Morals by Agreement, they do require (a) that the agents involved are acting on non-tuitive preferences (not referring to interpersonal comparisons), and not on strong evaluations of putative objective goods that cannot be adequately measured by consumer demand functions. More generally, Gaus does not clarify what is distinctive about social dilemma-ness as opposed to the other two dimensions of CAPs. And that is that the only Pareto-optimal outcome(s) available to the interacters, as acting on non-tuitive bare preferences, are not in equilibrium; although at least one such P-optimal outcome that is feasible via cooperation (social dilemmas are not zero-sum games).

Gaus then returns to ultimatum games to show that rules about fair distribution among different kinds of interacting parties (friends, total strangers) have to underpin property rights (201-5). But Gaus does to highlight that what normally solves (or allows interacters to overcome) Ultimatum games are familiar distributive equity principles (DEPs), which will take conventional forms usually as informal expectations, such as taking turns, lining up (first-come first served), equal division, random division, or division according to contribution in the production of the social gain from interaction – or according to other forms of merit (several classes of objective value-judgments come directly into these sorts of standards).

Nor does Gaus clarify that distributive equity is a dimension of CAPs that is constituted by a situation of interaction having more than one P-optimal that is feasible for the parties. Ultimatum is a non-simultaneous mover example of this dimension in its pure form, without the social dilemma and coordination dimensions. In Ultimatum, all of the P-optimal outcomes lie along the Pareto-frontier, which consists in all of the possible splits of the social gain (or “payoff”), e.g. with a pot of $1000 at stake, $100 for player A and $900 for player B (who moves first), or $200 for A and $800 for B, or $700-$300, etc. all the way down to $999-$1 – which could perhaps occur if B were maximally altruistic (but that is a kind of tuitive preference). Exactly the same points are all in equilibrium, and there are no other equilibria that can be reached by interacters pursuing non-tuitive (simple) bare preferences for more money.

And yet sometimes the party with the second-mover disadvantage still decides to reject the first-moving player’s “offer,” in which case both get zero – no social gain to divide up. This means that they sometimes end up under the Pareto-frontier in a P-sub-optimal outcome that, amazingly, is also not in equilibrium. Who would be that irrational though? Isn’t getting $10 out of $1000 better than getting nothing? No, not to almost all human beings, because we care about the inequity of such a split, unless there are special reasons for it in particular cases.

This means that distributive equity concerns are a sub-class of tuitive motives; and the various modes of fairness they embody is also one sub-class of objective value convictions too (both of these also bear on the division of labor in cooperative efforts, as well as on the division of benefits). Those tuitive and objective aspects – which, note, are more general than objective judgments involved in many sorts of merit (in the narrow sense of earned deserts) – help to explain why the first two generations of game theories did not recognize them as a distinct dimension. But for Gaus, these omissions the otherwise-interesting case that he makes for a fairly libertarian Principle of Natural Liberty with a presumption that a type of action is legitimate enough unless it arguably falls afoul of specific prohibitions alleged to be justified in the community (187-98). That is unfortunate, as this is one of the more specific normative claims about justice that he explores in the book.

The next post will explain the problem that Gaus gets into when he turns to coordination games in hopes of using them as a clue towards a way of reconceiving the social contract for Open Societies.