OPR V.16 Evaluating Proposals and the Problem of Indeterminacy

After specifying formal constraints on proposals (V.15), this section focuses on how to evaluate them, especially under conditions of indeterminacy. As with the previous section, I provide a summary of the major arguments and then raise some preliminary questions. I should note that, in various places, I have tried to simplify technical aspects of Gaus’s discussion.


This section begins with the problem of how to rank qualified proposals offered by Members of the Public (MoPs) in the Deliberative Model. One option would be to consider all proposals simultaneously, but Gaus rejects this possibility as unrealistic given our cognitive limitations. He argues instead for pairwise comparisons of proposals to construct ordinal rankings that satisfy modest conditions of social choice, including “asymmetry of preferences, symmetry of indifference, reflexivity of preference, and transitivity of strict preference” (305).

The problem of indeterminacy, to which the section title refers, is that MoPs cannot expect that their evaluative standards will produce sufficient reasons to prefer one proposal over another in pairwise comparisons. A may have an undefeated reason for proposal, P, and an undefeated reason for proposal, Q. In this case, he has no rational basis for choosing between P and QA does not think that P and Q are equally good; he is not indifferent between them (306). Rather, he has no reason for deciding that one is better than the other or that they are of equal value. For A, P and Q are incomparable; or, in Razian terms, incommensurable. Thus, A‘s preference ordering is incomplete, or, as Gaus says, indeterminate. (For the more technical aspects of this discussion, see pp. 305-08.)

This may seem like a Buridanic choice problem, in which, failing to have sufficient reason for choosing either P or Q, A chooses neither. But this type of indeterminacy, or preference incompleteness, can be converted into a type of “indifference relation,” in which we suppose that A prefers P or Q (but not both) to a third option, which involves choosing neither of them. A doesn’t think that either option is preferable to the other, but, for reasons discussed below, we can assume that, unlike Buridan’s Ass, A has sufficient reason to believe that selecting one option from the set of {P, Q} is better than choosing neither. The upshot is that if A’s indeterminate preferences can be converted into an indifference relation, then it may be possible – and I’m simplifying here — to generate a complete ordering of proposals that meets, or at least closely approximates, the social choice conditions stated above (309).

Of course, even if MoPs can generate (or construct) complete orderings of proposals, reasonable pluralism precludes convergence among them. We must expect that MoPs will disagree deeply about which orderings are most justifiable (310).

At this point, Gaus adopts a two-part strategy to deal with reasonable disagreement about proposals. First, he sets a baseline that provides MoPs with strong reason to select at least one qualified proposal. In making pairwise comparisons between proposals, MoPs might argue that neither option is acceptable. But in reasoning this way, they must take into account the opportunity cost of rejecting both options. Before A can reject both P and Q, he must ask what life would be like without any rules of social morality. Putting this in terms of classical social contract theory, A must ask whether he would prefer to live in the state of nature. Of course there is no agreement on how to describe life in the state of nature, but Gaus suggests that the prospects for stable social cooperation would be significantly diminished (313).

The second part of Gaus’s strategy is to narrow the range of proposals in the Deliberative Model. This can be done in two ways. First, MoPs may reject proposals on the grounds that they are not genuinely moral rules; in other words, they fail to satisfy the formal constraints of morality (as described in V.15). Second, MoPs may reject proposals that impose what Gaus calls “excessive costs of moralization” (315). Gaus gives the example of smoking. Suppose A proposes a rule to prohibit smoking. B may reject that rule by arguing that she has a “blameless liberty” to smoke. Gaus says we have a “blameless liberty to φ if and only if neither of us has a duty to φ or a duty to refrain from φ-ing” (316). Thus, if B has a blameless liberty to smoke, then A has no authority, or standing, to complain about B’s smoking. On this view, smoking is a practice that is simply outside the domain of moral authority, such that B can reasonably reject attempts to “moralize” it. More generally, Gaus argues that MoPs may reject proposals which moralize aspects of social life that fall within a “protective perimeter” of blameless liberty (317). This perimeter is comprised of certain “core” rights and liberties (similar to Rawls’s constitutional essentials), which, once secured, may serve as “fixed points” in thinking about more peripheral matters.

Although a defense of the core rights is not taken up until the next chapter, Gaus argues against Wall and others for a general presumption in favor of blameless liberty. If A claims moral authority over B, and if B denies that claim, A and B are not symmetrically situated in terms of their respective burdens of justification. A must show that B has sufficient reason to accept his claim of authority. If he cannot, B remains morally free. She is at liberty to act blamelessly, such that it would be inappropriate for A to have reactive attitudes toward B should she fail to comply with his demands (319-20).

Any proposals rejected by an MoP – either because they fail to satisfy the formal constraints or because they impose costs of moralization – fall into the set of ineligible rules. The remaining proposals, which all MoPs prefer to blameless liberty, constitute what Gaus calls the socially eligible set. As Gaus notes, all of the rules in this set are Pareto superior to the option of having no rule at all (i.e., the state of nature with respect to social morality). The Pareto rule can also be used to narrow this set. If all MoPs prefer one or more proposals to all others, then proposals that are not preferred can be eliminated. The remaining proposals constitute a socially optimal eligible set (OES) (323).

The question now is: what proposals are contained in OES? Gaus notes that there are three possibilities: (i) the set might be empty, (ii) it might contain a single proposal, or (iii) it might contain multiple proposals. The next chapter argues against the null set, and, on grounds of reasonable pluralism, Gaus rejects the possibility of convergence on a unique solution. His view is that OES will yield indeterminacy across a range of undefeated proposals. To borrow a term from Justificatory Liberalism (see sec. 9.4), OES is an example of nested indeterminacy. All MoPs have sufficient reason to select an option within the set, but there is no collective basis for preferring one option to all the others.

Gaus anticipates that some readers may be disappointed with this conclusion. Why go through all the trouble of developing a complex, sophisticated, and, at times, technical model just to reach an indeterminate outcome? Gaus gives three answers. First, OES is useful in excluding proposals that cannot be justified to all MoPs. So, at the very least, we have narrowed the range of rules that might comprise a publicly justifiable social morality. Second, as compared with competing philosophical devices (e.g., the original position), the Deliberative Model’s indeterminacy results from more realistic assumptions about our cognitive and moral capacities and about the extent of reasonable pluralism. Third, the principles used to generate OES are “as uncontroversial as possible” (326). For reasons that I won’t try to summarize here, Gaus argues that application of the Pareto rule to the socially eligible set yields a social choice that satisfies the conditions of Arrow’s theorem (with the exception of transitivity, for which is substituted quasi-transitivity based on the conversion of incomplete preferences to indifference relations, as discussed above). Thus, the Deliberative Model not only captures the idea of Kantian legislation (V.15.1), but it also yields a possibility result under Arrow’s conditions for social choice (327).

In the remainder of this section, Gaus argues that no preference aggregation procedure can resolve the indeterminacy among proposals contained in OES without violating one of Arrow’s conditions. Since MoPs will disagree about which of these conditions is most significant, there is no uncontroversial or publicly justifiable aggregation method for producing a more determinate result (327-29). Gaus similarly rejects bargaining theory because it is too controversial and, perhaps more importantly, because it makes public justification centrally about negotiation and compromise rather than about whether we have good reasons from within our evaluative standards for endorsing the rules of social morality.


There is a lot happening in this section, and those with technical backgrounds may want to address Gaus’s claim that, with some fairly uncontroversial modifications, the Deliberative Model produces results that satisfy Aarrow’s conditions. Here are some additional questions:

  1. How plausible is the null set hypothesis? Gaus has not yet offered his argument against it, but I suspect that this will be a nagging concern. Do we think that all MoPs will have sufficient reason to prefer any qualified proposal to none at all? As Gaus notes, much depends on how we think about life in the state of nature. But given reasonable pluralism, and without some further constraint on evaluative standards beyond mutual intelligibility, can we foreclose the possibility that even a single MoP will decide that it would be better to forego social morality in return for uninhibited pursuit of the values and goods specified by that MoPs evaluative standards? If not, would this amount to a reductio? If the Deliberative Model yields anarchical conclusions, perhaps it ought to be revised to include a greater level of idealization or stronger assumptions about the moral psychology of MoPs.
  2. Another way to frame the first question is in terms of whether the socially eligible (SE) set is Pareto optimal. If any MoP prefers no moral authority to any of the proposals contained within SE, then SE does not meet the Pareto rule. Can this possibility be ruled out a priori, or does it rest on empirical claims about the level of social cooperation that could be achieved without agreement on rules of social morality?
  3. Gaus’s claim that blameless liberty is the default position for justificatory purposes is controversial. Other contributors in the book group have previously addressed this issue, but a further question might be: how important is this claim for the structure of the Deliberative Model? If Gaus conceded the issue, how much would be lost? Wouldn’t he still be able to claim that this procedure “correctly models Kantian legislation but it also meets all of Arrow’s conditions” (327)? If that is the case, does the liberty default put a controversial liberal or libertarian tilt on the model?
  4. Returning to a question I raised on Monday, could a proposal for a tiered morality be part of the socially optimal eligible set? If such a proposal violates the weak publicity condition, then presumably it would be excluded as violating a formal constraint. But assuming weak publicity does not preclude it, does the model contain other resources that might bar asymmetric moral rules?
  5. In this section, Gaus argues that MoPs may often confront indeterminacy with respect to proposals. By this he means that they will have undefeated reasons for at least two competing options, with no sufficient reason to prefer one over the other. As Gaus says, “We have seen that sometimes, perhaps often, the reasons a person has at the appropriate level of warrant are indeterminate; a person may have two competing reasons, neither of which defeats the other … she has no sufficient reason to decide that one option is more choice worthy?” (306) Does this claim mark a shift from an earlier view, in Justificatory Liberalism, according to which individuals often face inconclusiveness in their decision making, but rarely indeterminacy. As Gaus defined inconclusiveness  (see JL, secs. 9.3.3 and 9.4.1, esp. pp. 151-54), a justification is inconclusive when it is undefeated, but not victorious. To be victorious, a justification must (i) survive public challenge, and (ii) meet a higher burden of epistemic proof than mere sufficiency, perhaps something like “beyond a reasonable doubt.” Has Gaus given up this conception of inconclusiveness, or defined inconclusiveness down so that it has the same meaning as indeterminacy? Do both concepts now refer to having undefeated, sufficient reasons for mutually exclusive options? In his discussion of Buridan’s ass, Gaus remarks on the difference between “the concept of indifference and that of indeterminacy (or inconclusiveness)” (306), suggesting that the latter concepts might be the same. If they are, has something been lost in the transition from Justificatory Liberalism? The early (or earlier?) Gaus cautioned against conflating the two concepts and against a presumption of indeterminacy, claiming that our natural, intrapersonal condition was one of inconclusiveness, rather than indeterminacy. Has he modified or abandoned this view?
This entry was posted in Books, Posts, Reading Group. Bookmark the permalink.

One Response to OPR V.16 Evaluating Proposals and the Problem of Indeterminacy

  1. In response to Micah’s question 3, about the blameless liberty default: I think Gaus needs some default, for we can have situations in which no choice is unanimously acceptable. In the simplest case, we have only two options: some rule X, and no rule at all (call this 0). Some people rank the proposals X0, others 0X. Each is reasonably rejectable in favour of the other. Without a default, the procedure doesn’t yield a result.

    I also think that it will be crucial to Gaus’s argument that the default be ‘no rule’, blameless liberty. But if Gaus were to concede that blameless liberty is not the appropriate default, we would need to know what the alternative default would be (equal collective authority?)

    One further comment about Wall’s symmetry thesis. Wall was talking about coercive interference, whereas Gaus is now talking about authority. In the context of the smoking example, Wall’s point would I think be that both threatening people with punishment if they smoke and forcing them to breathe the smoke from one’s cigarettes stand in need of justification, and that we ought not accept the principle that we will legislate against smoking only if no one can reasonably reject the law in question (or: we ought not accept the principle that we will restrict smoking only up to the point at which someone can reasonably think the restriction too expansive).

    Gaus’s criticism of Wall frames the issue in terms of authority:

    There is thus an asymmetry between Alf, who claims that he has moral authority over Betty, and Betty who denies the authority of the rule. Alf’s claim is justified only if Betty has reasons to endorse the authority; Betty’s claim is justified simply if there are no reasons. She does not have to provide a case for rejecting Alf’s claim to authority – she is free and equal. He must provide a case that he has authority over her… [O]nly Alf is asserting any claim to moral authority. To see, as do some, Alf and Betty’s justificatory burdens as symmetric… is to say that Betty is under an equal burden to show Alf that his claim to authority over her is ungrounded as he is to show her that it is well grounded.

    Now, this sounds very plausible. Surely Alf needs to justify his claim to authority over Betty to Betty, and if he can’t do so, then they should default to no-authority: neither recognizes any obligation to obey the other. But here again the distinction between the authority of morality and the authority of persons is important. Alf can simply concede that Betty is under no obligation not to smoke (in a restaurant, say) simply because Alf thinks it wrong that she do so. Alf is claiming that it IS wrong for her to smoke in this context, not that she has an obligation not to smoke simply because Alf thinks it wrong. The question is what the default will be when Alf and Betty try to resolve this moral disagreement politically, and are considering passing a law restricting smoking. Within some limits, Alf and Betty will both be under an obligation to obey the law. It is at this stage that Wall’s symmetry thesis is important; both the state’s coercive interference and the state’s failure to prevent coercive interference stand in need of justification, symmetrically.

Leave a Reply