The Intransparency of Political Legitimacy

Transparency and Justificationism

A prominent class of theories of legitimacy rely on the idea of public justification to others:

JUSTIFICATIONISM. An institutional scheme of exercising power is legitimate if and only if it is justifiable to all (reasonable) people.³

What is justifiable to A depends on the actual beliefs of A, subject to some moderate levels of idealization. Thus, justificationism formulates an internalist constraint on legitimacy: political power needs to be acceptable from a wide variety of actual viewpoints.

JUSTIFICATIONISM states a position different from TRANSPARENCY. First, something can be justifiable to someone without it being true; by contrast, only truths can be knowable. Second, something is feasibly knowable as long as there is some way to come to know it. But justifiability is more demanding than this: it requires that, on some appropriate idealization, the other person is more likely to endorse it than not.

Most importantly, however, we treat TRANSPARENCY as a meta-property of theories of legitimacy, while JUSTIFICATIONISM is just one such theory, providing us with an account of when exercising political power is legitimate. Thus, for example, one might think that good theories of legitimacy must be transparent, but reject justificationism. This would entail that the true theory of legitimacy T must be transparent. But T can specify grounds of legitimacy different from public justification — say, aggregate utility — as long as these are knowable. Moreover, the transparency of T does not entail that a T-endorsed scheme of exercising power S is justifiable to you; after all, you might reject T. Thus, S might maximize aggregate utility, and you might be able to know that it does; but you might reject that utility maximization is relevant to legitimacy.

Inversely, without further assumptions, JUSTIFICATIONISM need not satisfy TRANSPARENCY. Intransparent justificationism claims that legitimate power needs to be justifiable to everyone, but that we might sometimes remain in the dark as to when this is the case. This is a logical possibility. However, as I shall now argue, most justificationists have further commitments which strongly push them towards transparency. Let us start with the most prominent form of justificationism, Rawls’s political liberalism.

A central idea throughout Rawls’s work is the notion of publicity. One ingredient in Rawlsian publicity is the idea of common knowledge.⁴ If p is common knowledge, then everyone knows that p, and everyone knows that everyone knows that p, and so forth. Crucial for the late Rawls is the idea that power is justified in terms of public reasons, which are shared reasons which everyone can accept. Thus, if R is a public reason, then R is reasonably acceptable to everyone, and everyone knows that R is reasonably acceptable to everyone, and so forth.

Public reasons, in turn, provide the basis for legitimacy via the liberal principle of legitimacy, which requires that “all citizens […] may reasonably be expected to endorse [the constitutional essentials] in the light of principles and ideals acceptable to their common human reason” (Rawls 1993, 137). Thus, to greatly simplify, if the constitutional essentials C are legitimate, then there need to be public reason(s) R in its support, and it is common knowledge that R is a public reason; thus, everyone knows that everyone knows that R is a public reason. Reasonable people also endorse the liberal principle of legitimacy, such that they accept that C is legitimate only if there are public reasons in its favor. Taking these ideas together, it is likely that Rawls accepts a version of transparency: if C is legitimate, then everyone knows that there are public reasons in its favor, and thus, that it is legitimate. I think similar arguments can be made for a wide range of post-Rawlsian justificationists.⁵ (A formalized version of the argument in this paragraph is offered in the Appendix.)

Other justificationists reject key parts of the Rawlsian view — for example, they reject the assumption that public justification requires reasons that everyone endorses (“convergence” instead of “consensus” liberalism). Nonetheless, these views share the basic normative aims of Rawls’s political liberalism, and thus feel a strong push towards endorsing TRANSPARENCY. The original motivation behind Rawls’s political liberalism is to find a political conception of justice which solves the stability problem: how can people with divergent metaphysical, ethical, and political views live together in society which is stable in the long run and for the right reasons?⁶

Post-Rawlsian justificationists, no matter their differences with Rawls, share this general moral vision. Gerald Gaus, for example, thinks of justificationism as the project of finding rules of social morality that allow peaceful cooperation (Gaus 2011b). Kevin Vallier thinks of the justificationist project as one that solves an “assurance problem,” finding public rules that enable social trust (Vallier 2019). Let us limit this to the issue of legitimacy. The problem, then, is that we need to have rules to assign power to some to organize social affairs amongst people who deeply disagree how power ought to be exercised. It is a natural idea that these rules need to be transparent to play their intended role.⁷

Kevin Vallier comes closest to committing explicitly to a form of TRANSPARENCY. He writes that “[i]f we cannot discern whether a norm or law is publicly justified, then the Public Justification Principle cannot do us much good” (Vallier 2014, 172). Put differently, Vallier treats as a severe objection to justificationism the possibility that we could not know what is justifiable, and thus, what is legitimate. He expresses optimism, however, that we can discern what is publicly justifiable (Vallier 2014, 172–77; 2015, 612–13).

Transparency and Other Theories of Legitimacy

Justificationists are plausibly not the only theorists pushed towards accepting TRANSPARENCY. Liberals emphasize respect for the reason and autonomy of individuals; thus, Waldron characterizes the demand “that the social order should in principle be capable of explaining itself at the tribunal of each person’s understanding” (1987, 149) as a core liberal idea. The principles according to which political power is exercised are a central part of our social order, and thus it is natural to assume that these principles should be transparent. Similarly, Thom Christiano demands that political institutions satisfy a publicity principle, specifically “that the institutions that satisfy the principles of social justice must be able to display the fact of their justice to ordinary persons” (2008, 51). This notion of public equality also forms the grounds of Christiano’s account of legitimate authority (2008, ch. 6). In a different way, political realists endorse the idea that the political order needs to be justified to its subjects, although the notion of justification here is a non-moralized one. Bernard Williams emphasizes that political structures must “make sense” to its subjects (2005, 10–11).

Across a wide range of non-justificationist views, then, there is some philosophical push towards thinking that good theories of legitimacy are transparent. Admittedly, the question is rarely if ever asked in the precise terms we have phrased it, so only a few interpretative suggestions could be offered that transparency is a widely accepted desideratum in theorising about legitimacy. What if the demand for transparency turned out to be mistaken, however? Crucially, this would have an impact on how convincing we find different theories of legitimacy. Instrumentalist theories of legitimacy, for example, claim that legitimacy should be distributed such that some moral benefit — for example, aggregate welfare or distributive justice — is achieved best (e.g. Arneson 2003). One core suspicion raised against instrumentalism is precisely its intransparency. If it turned out, however, that good theories of legitimacy are not transparent, then an important objection to instrumentalism falls away. Similarly, to the degree that transparency is thought to be an advantage of justificationism and other liberal theories of legitimacy, these advantages are shown to be illusory. Overall, then, rejecting transparency levels the playing field in favor of instrumentalism.⁸

Metaethical Theories

We can also see the relevance of the issue from the vantage point of more general philosophical concerns. Perspectivists claim that what an agent ought to do depends on the agent’s perspective — that is, their beliefs, knowledge, or evidence (e.g. Andrić 2013; Jackson 1991; Kiesewetter 2011, 2016; Lord 2015).⁹ Objectivists deny this claim (e.g. Graham 2010; Srinivasan 2015; Thomson 1990; Weatherson 2019). A related distinction is between internalism, according to which the reasons an agent has are connected to the agent’s psychology, and externalism, which denies that there is any such connection (Lord and Plunkett 2018). Perspectivism and internalism heavily incline you to think that all values satisfy TRANSPARENCY, while objectivism and externalism push one in the opposite direction.

We could approach the issue with a deductivist mind-set. If some general argument in favor of perspectivism or objectivism succeeds, the deductivist claims, then legitimacy — as merely one special sub-realm of normativity — is determined to be transparent (as perspectivism claims) or intransparent (as objectivism claims), accordingly. Perspectivism is often motivated with respect to the importance of action-guidance, for example.¹⁰ The idea is that moral theories are action-guiding; and moreover, the argument continues, action-guidance is only possible under conditions of epistemic access. Similarly, we might conclude, legitimacy is an action-guiding concept, and thus should be epistemically accessible.

I shall avoid the deductivist approach in this paper, however. First, many realists believe that political norms have separate grounds from morality (e.g. Williams 2005) and, thus, that drawing inferences about them from the nature of moral norms is not valid. It is therefore preferable to avoid making any background assumptions in this respect. Secondly, it can be fruitful to adopt the opposite, inductivist approach. The inductivist argues bottom-up: they start by looking at specific areas of normativity and consider whether they satisfy, or do not satisfy, TRANSPARENCY. This careful look then informs more general arguments about whether morality might be transparent. In this way, I take this paper to contribute to the inductivist case for objectivism.

The Intransparency of (In-)Justice

Consider the following case:

PREJUDICE. Smith is a talented high-school student of color from a working-class background. He is accepted into a classist and racist elite college, where he is subjected to pervasive, but often subtle, discrimination. For example, his tutors treat him more critically than his peers and provide him with less support. He finds it difficult to find friends, and he becomes the subject of denigrating treatment from classmates. Social and psychological pressures mount, and Smith’s academic performance suffers. Eventually, he drops out of college. Smith, his tutors, and his peers all form the opinion that he did not have the necessary talent to succeed.¹¹

There are two relevant features of PREJUDICE: (i) Smith was treated unjustly, and (ii) many of the participants, perhaps even Smith himself, were not in a position to know that he was treated unjustly. I take it that (i) is obvious. Smith deserved equal opportunity to succeed in his academic endeavors, but such opportunity was denied to him because of morally irrelevant factors. We can also imagine that Smith’s case is not isolated: there are many others who are mistreated by Smith’s college. Put differently, the injustice is institutionalized and not accidental, so that we can speak of the injustice of an institutional scheme.

To establish (ii), consider PREJUDICE from the point of view of Johnson, the dean of Smith’s college. Johnson, we can imagine, is not openly racially prejudiced. Nonetheless, he might be incapable of recognizing the injustice done to Smith. He does not see that Smith’s tutors treat him more harshly than others; instead, he sees tutors who show “tough love” and a student who fails to live up to the exacting standards of an elite college. Neither does he realize that Smith’s peers viciously denigrate and exclude him; instead, Johnson thinks that “it’s just banter” and that Smith can simply ignore the comments. Johnson, in short, is incapable of seeing the many forces which disadvantage Smith and which together constitute injustice. We can imagine similar stories about Smith’s tutors and peers. Even Smith himself might internalize the judgments others make about him, and come to believe that he did not have the talent necessary to succeed (and thus, was not wronged).

Not much hangs on the details of PREJUDICE. If you dislike the example, structurally analogous examples can be constructed easily. All you need is the general possibility that some injustice is done to some person, and (at least) one person with average mental capacities, owing to prejudice or other forms of bias, cannot feasibly come to know that such injustice is done. We could also build other examples, focusing more explicitly on factual uncertainty. Perhaps, for example, there is accumulative injustice done to some group, but seeing that injustice requires non-trivial statistical work.

The Direct Argument

Let us assume for now that justice has been shown to be intransparent. What follows? After all, justice is not legitimacy, at least conceptually — that much everyone agrees on. Nonetheless, the intransparency of justice helps us with establishing the intransparency of legitimacy. The first argument, offered in this subsection, builds on widespread agreement amongst theorists that extreme injustice defeats legitimacy.¹² If this is true, then we can argue:

THE DIRECT ARGUMENT

• (1) If some scheme of exercising of power is extremely unjust, then it is illegitimate.

• (2) Extreme injustice is sometimes intransparent.

• (3) Thus, sometimes the illegitimacy of some scheme of exercising of power is intransparent.

How might one resist the direct argument? First, one might deny premise (1). However, this reply requires you to reject one of the two highly plausible component subclaims of (1): (1a) if some scheme of exercising power is legitimate, then it is (at least) permissible,¹³ and (1b) it is impermissible to enact extreme injustice. As noted above, every author in the literature is committed to (1a). The term “legitimacy” works as a moral commendation. If it does not even express a permission (or liberty-right) to exercise power, it is no longer clear why it would be a morally desirable property to possess.

Outright rejecting (1b) — to claim that it is permissible to enact extreme injustice — also looks unpromising.¹⁴ More plausibly, one might suggest that it is permissible to enact extreme injustice if the alternative is even greater evil (including greater injustice) (e.g. Rawls 1999, 4). Let us distinguish between necessary injustice — injustice necessary to avoid greater evil — and non-necessary injustice. We can then modify the first two premises of the direct argument accordingly.

• (1*) If some scheme of exercising of power is extremely, non-necessarily unjust, then it is illegitimate.

• (2*) Extreme, non-necessary injustice is sometimes intransparent.

However, the two revised premises are still highly plausible, and the direct argument remains sound. Note, in particular, that (2*) stands on even firmer ground than (2). Determining whether something has properties A and B is more epistemically demanding than determining whether something merely has property A. Following this observation, if it is intransparent whether some political institution is extremely unjust, then it is even more likely that it is intransparent whether some political institution is extremely unjust and not necessary to avoid greater evil. Thus, relying on (2*) actually strengthens the argument for intransparency.

A second general way to resist the direct argument is to deny premise (2) or its correlate (2*). There are two ways to do so. First, contrary to what I argued, one might suggest that justice generally is transparent. It is hard to see, however, what would motivate this response. Take a simplistic right-libertarian theory of justice, for example, according to which (i) anyone can acquire unowned items for themselves subject to no restrictions, and (ii) any subsequent exchange of items retains justice just in case it is voluntary, subject to no other restrictions.

I chose this theory as an illustration because it is one of the simplest one can imagine. However, even this theory of justice will often make it epistemically inaccessible whether some social arrangement is just, as it requires you to acquire significant knowledge about the history of social interactions. To know whether some traded property justly belongs to you, for example, you would need to trace the history of that property back to its original acquisition and ensure that any subsequent trade has been voluntary. This knowledge will often be intransparent. Once we move to more plausible theories of justice, problems escalate. Almost all libertarians endorse additions, modifications, and exceptions to the simplistic theory which will greatly increase sources of intransparency. Other theories of justice, like a luck-egalitarian theory of justice, will bear their own potential for significant intransparency. In short, any plausible theory of justice will be intransparent.

Alternatively, one could resist premise (2) in a more localized fashion, by arguing that extreme injustice is always transparent, even if justice in general is not. Consider, for example, the view that some institutional scheme is extremely unjust just in case it violates basic human rights. On this basis, one might argue that it is generally transparent whether some institutional scheme violates basic human rights. Thus, even if extreme injustice defeats legitimacy, this does not introduce intransparency into legitimacy.

However, even extreme injustice is sometimes intransparent. Some evidence comes from the historical record. Almost all past societies were morally repugnant in some way — they were based on slavery, colonial exploitation, class hierarchy, sexism, racism, or other injustices. However, some members of these societies with average mental capacities were not in a position to know that their society was unjust. By extension, we can also expect that our current societies are unjust in deep ways, even though we might not be in a position to know how and why.

The Indirect Argument

The second argument for the intransparency of legitimacy is indirect. It rests on the claim that justice and legitimacy play similar roles in our practical thought. To approach the issue, let me offer a plausible diagnosis why justice is intransparent. What is epistemically accessible to people is constrained by their actual social position. Our actual social position, in turn, is determined by deep structural features of our society. Thus, if justice were transparent, then it would be constrained by existing social arrangements. This would make many forms of critique of existing social and political arrangements impossible. However, criticizing existing social arrangements is one of the main functions of the value of justice. Call this the critical function of justice. Because it fulfils this function, it cannot be tied to existing social arrangements, and thus we should not expect it to be transparent.

As an illustration, take a simple case of a capitalist society in which a large majority suffers from a form of false consciousness. That is, most people accept deep-seated rationalizations of the capitalist order as just and rational, whereas it is, in fact, deeply unjust. (You can also switch this around as people holding socialist illusions within an unjust socialism.) The social, cultural, and economic institutions in this society reinforce these false beliefs; for at least some people, it will not be possible to even imagine fundamental criticism of the capitalist system. Thus, if justice had to be transparent, no criticism of injustice could be leveled against capitalism (or in the inverted case, socialism). But I presume that we could.¹⁵

The indirect argument builds on these observations, without claiming that legitimacy is directly connected to justice:

THE INDIRECT ARGUMENT

• (1) Any value which can play a critical function in our moral thought possesses (a degree of) intransparency.

• (2) Political legitimacy can play a critical function in our moral thought.

• (3) Thus, political legitimacy is intransparent.

Note that premise (2) does not claim that political legitimacy only plays a critical function. It might play several functions, a point to which I return in the next section. Nor is the assumption that political legitimacy plays the precise same function as justice. Legitimacy might provide us with one way to criticize political arrangements, justice with another. How might one resist the indirect argument? I suspect the main critical pushback will be against premise (2). I shall consider such claims in the next section. For now, let me provide a positive argument why we should think premise (2) to be true.

We can start from the observation, sometimes offered, that the difference between justice and legitimacy is between which social order is desirable and who is permitted, and how, to impose it (Pettit 2012, ch. 3; Larmore 2020, 44). However, exercising power will normally have a deep and powerful impact on our lives — our welfare, our autonomy, our relations to each other, and so forth. It is natural that existing patterns of power can, for this reason, be assessed in a critical manner. Radical feminists will object, for example, that we live in a patriarchy. This is not (primarily) some concern over the desirability of some social order, but a concrete concern about how men shape the lives of women.

Nothing hangs on whether you think the feminist critique is correct. What matters is that it is intelligible as a charge of illegitimacy, and as a charge that does not take itself to be constrained by existing arrangements of power. The same is true of anarchist, Marxist, conservative, and other critiques of the existing political order. Thus, even if we accept a clean distinction between the “desirability” and “permissible imposition” of a social order,¹⁶ there is no reason to think that the former is subject to critical assessments while the latter is not.

The Settling Function of Legitimacy

Fabienne Peter presents an argument for the transparency of legitimacy which she calls the “accessibility objection” (Peter 2020, 4–6). She argues that normative properties can play two functions in our practical thought, an “orientation function” and a “settling function.” The orientation function is “world-regarding”: “geared toward what truly is the case — what the world is like.” By contrast, the settling function is “agent-regarding,” “geared toward what we should believe and how we should act.”¹⁷ Peter’s main argument, in a simplified interpretation,¹⁸ can be captured in the following syllogism.

ACCESSIBILITY OBJECTION

• (1) Political legitimacy (exclusively) plays a settling function — it settles how we should act.

• (2) For some value to play a settling function, it needs to be transparent.

• (3) Thus, political legitimacy is transparent.

I have two main objections to this argument. First, Peter repeatedly speaks of “settling what we should do,” but what do we mean by this phrase? Here are some natural candidates:

• (A) If it is settled what we should do, then what we should do will not foreseeably change.

• (B) If it is settled what we should do, then engaging in further practical deliberation is pointless.

• (C) If it is settled what we should do, then we know what we should do.

• (D) If it is settled what we should do, then our balance of reasons favors a unique course of action.

• (E) If it is settled what we should do, then we have agreed on what should be done or what we will do.

However, the argument is dubious on many of these interpretations. Against (A), it is not clear why legitimacy needs to provide us with unchanging assessments of what we should do. The legitimacy of a particular policy or even an entire government can alternate with changing conditions. The same is true of interpretation (B). Calling some institution legitimate is not a conversation-stopper, nor is it clear why it should be. Interpretation (C) begs the question, because if you know that p, then you can feasibly come to know that p, on the maxim that the actual is possible — but that is what we are trying to establish.

Interpretation (D) brings us back to debates in meta-ethics mentioned earlier. Imagine that legitimacy is tied to authority, such that legitimate institutions provide us with exclusionary and content-independent reasons to pursue some course of action (Raz 1986). Thus, we can restate the two premises of the Accessibility Objection as follows.

• (1*) Political legitimacy gives us (exclusionary, content-independent) reasons for what we should do.

• (2*) You have a reason to φ only if it is epistemically accessible to you that you should φ.

Let us set aside that some authors reject (1*).¹⁹ This argument, through (2*), rests on a general form of reasons internalism. On this interpretation, then, Peter’s accessibility objection is revealed as a deductivist argument, the force of which relies on internalism or some related meta-ethical commitment.²⁰ This does not straight-out beg the question — there is still a small step from reasons internalism in general to internalism about legitimacy in particular. Still, it would be misleading to think that this argument has much dialectical purchase against someone who is skeptical about the transparency of legitimacy, as such a sceptic is likely to deny reasons internalism. If we wish to avoid deductivist arguments altogether, then this strategy is also an obvious non-starter.

The second objection is that we should reject premise (1): political legitimacy does not exclusively play a settling function. Instead, it is a multi-functional concept that plays various roles in our practical thought. At least one of those functions is to allow critical assessments of the political order.²¹ There is value in finding ways to live with those who disagree with us. But it is prima facie implausible that cooperation trumps all other concerns. The example of emancipatory social movements is again instructive. A movement which tried to reform the racist and sexist nature of elite colleges is likely to find instrumental, and perhaps even inherent, value in accommodating itself to the defenders of those colleges, because we value the social bonds with them. At some point, however, the importance of accommodation runs out; in such cases, the function of the language of legitimacy is no longer to transcend our disagreements with others, but to reject their errors.

Peter herself repeatedly writes that legitimacy “primarily” plays a settling function (2020, 5), which leaves the door open for legitimacy also playing other functions. This changes the argument somewhat — the crucial premise must then actually be the following,

• (2**) For some value to fulfil several functions, but primarily a settling function, it needs to be transparent.

But it is far from clear what speaks in favor of (2**). Consider an analogy. Grades in university play several functions, amongst them (i) to express the merit of academic work submitted, and (ii) to instruct, motivate, guide, etc. students, in a way that improves their overall learning. These aims can conflict: bad grades might sometimes be deserved, but be extremely discouraging to some students; some students who achieve effortless “A”s might profit from some harsher grading to prod them to achieve their full potential.

If grades merely should realize one of these two functions, then we would be confident in predicting further desirable features of grades. For example, if grades should merely reflect merit, then we could perhaps predict that anonymous grading is desirable. However, if grades primarily but not exclusively were tied to merit, then this is no longer true: it will at least sometimes be desirable to not grade anonymously, but to take into account other factors.

Similarly, if legitimacy exclusively played a settling function, then we might feel confident to predict its transparency. (We could still question how much transparency is really necessary for “settling,” but that is a separate issue.) However, if legitimacy plays other functions too, like a critical function, then we can no longer predict its transparency. In other words, we should reject (2**).

Meta-Coordination Arguments

Interpretation (E) moves us closer to the idea that legitimacy plays some practical function in coordinating our behavior. Perhaps the role of legitimacy in our practical thought is that we can find agreement in beliefs or coordination in our behavior. We can summarize these various claims through the idea that legitimacy plays a “meta-coordination” function, an idea which goes back to Allen Buchanan and is now endorsed by a variety of authors.²²

The main idea is simple but needs to be introduced carefully. Individuals can give or withhold their moral support to certain social and political institutions. Moral support must be distinguished from habitual acquiescence in, toleration of, or prudential support for, some institution. If you give your moral support to an institution, then you think that it is morally support-worthy in some sense. This entails a set of beliefs and attitudes, as well as dispositions to act in certain ways. For example, you are more willing to bear burdens for institutions you morally support, you will show deference, and at least respect, towards such institutions, you will normatively expect others to adapt similar attitudes, and so forth. Note that it is a purely descriptive question which institutions enjoy moral support: people can morally support evil institutions.

A second empirical claim is that widespread moral support provides institutions with greater institutional capacity than institutions without such support. A morally supported institution can pursue more ambitious aims and is likely to achieve its aims with less resistance, other things being equal. The next significant claim in the meta-coordination argument is that the moral support of a group needs to be coordinated on one set of governing institutions. If there were two or more institutions which enjoyed the moral support of the same group, then each of the competing institutions would not be able to achieve its aims, or it could do so only extremely inefficiently.

These are empirical claims which do not establish anything normative yet. But we are not far from completing the argument. The next step is to assume that legitimate governments are also effective: they have the capacity to achieve their aims in a somewhat reliable way. Note that, while a normative commitment, this is a rather minimal one, and compatible with a variety of first-order views concerning what legitimizes government. The last puzzle piece is the idea that coordinated moral support requires transparency. You could not give your genuine moral support to some institutional scheme S if you could not see for yourself, in the light of your own beliefs, that S is morally support-worthy.

Putting all these ideas together, we can make

THE META-COORDINATION ARGUMENT

• (1) Any legitimate institutional scheme of exercising power must be able to operate effectively.

• (2) To be able to operate effectively, any institutional scheme of exercising power must rely on coordinated moral support.

• (3) Coordinated moral support is only possible on the basis of a transparent conception of legitimacy.

  Thus, a legitimate institutional scheme of exercising power requires a conception of legitimacy which is transparent.

The three premises of the meta-coordination argument, however, are doubtful — individually, but even more so collectively.

First, some pressure needs to be put on the empirical premises (2) and (3). Against (2), it seems implausible to think that effective governments need everyone’s moral support to fulfil their function. Note that the baseline against which we compare is not the absence of support, but merely the absence of moral support. Governments might be able to operate effectively even if they rely, in the majority, on support on the basis of individual self-interest, or acquiescence, or fear of sanction, or mere habit. It might be that these alternative grounds are morally inferior in some way — perhaps governments ought to seek genuinely moral support from their subjects. But insofar as premise (2) expresses a descriptive claim, this is irrelevant.

Problems arise specifically if we consider premises (1) and (2) in combination. The main issue is how we should understand the demand for effectiveness. We can understand effectiveness in a thin sense, as the mere ability of a government to achieve certain aims somewhat reliably. If we understand effectiveness minimally in this sense, then premise (2) seems dubious, in the way I just criticized. After all, there are many types of institutions which are minimally effective without enjoying much moral support (cf. Klosko 2015).

On the other hand, we could read premise (1) in a thicker, more demanding sense. We might claim, for example, that a legitimate institution needs to realize effective governance in the sense of ensuring social stability. This idea can be understood in different ways, but it might entail, for example, that subjects to an institution see it as more than a modus vivendi (Rawls 1999). If understood in such a thicker way, then it becomes more plausible that moral support is needed for effective government — that is, premise (2) gains plausibility.

On the flipside, however, it becomes less plausible why effectiveness, understood expansively now, should always trump other moral concerns (cf. Wendt 2019, 51–54). Consider the view of a feminist politician who believes that radical reforms are necessary to abolish patriarchy. Such a feminist might accept a minimalist effectiveness constraint — a feminist government, after all, would need to be capable of implementing its intended reforms. But they are likely to reject a more substantive effectiveness demand. They might argue, for example:

It would be nice if we could achieve the abolition of patriarchy in a way that respects social stability, understood in some thick sense, and which garners everyone’s moral support. Alas, in practice this is impossible. The aim of abolishing patriarchy is so important that some loss of social stability is acceptable.

Note that our imaginary feminist does not deny that stability is of some value; they merely deny that it is of primary importance. Sometimes power should be used, all things considered, for divisive and controversial purposes. But then premise (1) is wrong: some legitimate governments do not guarantee stability.

Degrees of Transparency

A first objection rests on the observation that there are degrees of transparency.²³ Some value can be transparent to a greater or smaller number of people, in a greater or smaller number of cases. One might accept the argument that legitimacy is sometimes intransparent, but replace it with a demand like the following:

MAXIMIZE TRANSPARENCY. In as many cases as possible, as many people as possible²⁴ should be in a position to know what a theory of legitimacy demands.

There are several problems with this response, however. First, the new position no longer allows certain dialectical moves. Peter, for example, objects to “factualist” theories of legitimacy because they entail some intransparency. MAXIMIZE TRANSPARENCY no longer allows that objection, at least not in an easy fashion.

Second, there is a big philosophical step from “transparency is required” to “some intransparency is fine.” Roughly speaking, in doing so we make the step from a moral constraint (“don’t kill”) to a minimizable moral aim (“kill as few as necessary”).²⁵ This suggests, too, that our background motivation for transparency has changed from a deontological story to some axiological one. This is not something to be done lightly and, without further justification, it seems ad hoc.

Third, MAXIMIZE TRANSPARENCY also looks prima facie implausible. MAXIMIZE TRANSPARENCY amounts to the claim that transparency should be our overriding priority in theoretical choice. Note, for example, that this would favor highly simple accounts of legitimacy over even moderately complex ones. It would favor, for example, the simplistic right-libertarianism discussed in section 4 over more complex versions. It is implausible that we should choose between theories of legitimacy primarily on this basis. Transparency might be a valuable property in choosing between theories of legitimacy but only when balanced against other concerns.

Relaxed Feasibility

Alternatively, one might observe that there are degrees of accessibility. Remember that accessibility is determined relative to some notion of practical feasibility. If we relax what is feasible, then there is a wider degree of propositions that one can feasibly come to know. To return to the example used previously, on a relaxed notion of feasibility, quantum mechanics is epistemically accessible to me, because I could come to understand it, although it would require serious amounts of time and study. We might, then, advocate

RELAXED FEASIBILITY. Moral demands should be transparent to people, relative to a relaxed notion of what they could feasibly come to know.

With RELAXED FEASIBILITY one might argue, for example, that we are always in a position to know that something is extremely unjust, thus avoiding the direct argument.

This raises a question we have not discussed: What is the correct notion of feasibility to apply? Ultimately, I suspect, our notion of what is feasible is determined relative to normative concerns — that is, concerns which lead us to think moral theories should be transparent in the first place. This is not a matter we have addressed, but we can make some plausible suggestions on the basis of the arguments we have considered. The meta-coordination argument, for example, suggests that our notion of feasibility cannot be too relaxed. To engender moral support, the support-worthiness of political power must be reasonably accessible to people as they are. Similar remarks apply to the settling function we discussed. It is hard to see how accessible knowledge of quantum mechanics, under a relaxed notion of feasibility, can provide me with much action-guidance, or settle any collective issue between us.

RELAXED FEASIBILITY, then, seems incompatible with two of the main arguments for transparency that we have considered. I suspect that the same will turn out to be true for other normative motivations for transparency, but establishing this suspicion goes beyond the limits of this paper (but see Enoch 2005).

Restricted Audiences

Alternatively, one might suggest that only certain people need to be in a position to know what legitimacy requires. One might propose, for example,

RESTRICTED AUDIENCE. Transparency must only be achieved with respect to a certain audience — in particular, morally reasonable people.

RESTRICTED AUDIENCE is similar to an argumentative move made by many justificationists, who claim that political institutions only need to be justifiable to reasonable people. This restriction has attracted much debate, and I think RESTRICTED AUDIENCE suffers from roughly the same issues authors in that literature have pointed out (e.g. Raz 1990; Enoch 2015), so I shall keep discussion brief. A general problem with RESTRICTED AUDIENCE is that it needs to introduce the notion of “reasonable” in a natural, non-gerrymandered way. If, for example, we want to argue that liberal policies are legitimate, then it is question-begging to assume that liberalism is epistemically accessible to all reasonable people. On the other hand, if we operate with an ordinary-language conception of “reasonable,” then it seems that restricting our audience does not achieve much; for example, all the participants in PREJUDICE are arguably reasonable in an ordinary meaning of the term. Thus, we can formulate a general dilemma: either the move to restrict the audience is question-begging, or it fails to resolve the direct and indirect arguments for intransparency.

Note also that RESTRICTED AUDIENCE stands in tension with several other claims about legitimacy that we have considered — for example, the claim that legitimacy’s main function is to allow practical coordination between people with divergent views. But the more severely we restrict our audience — and thus, the group to which claims of legitimacy are epistemically accessible — the less it will be capable of working as a coordinating value across the whole society. Instead, it will merely express the internal morality of a particular group.

Forms of Legitimacy

I have operated with the ecumenical assumption that legitimacy is the right to rule, and at least the liberty-right of an institution to coerce. But one might think that this stacks the deck against the defender of transparency. The best way to press the objection is to highlight that, if legitimacy was more than a liberty-right to coerce, then it would entail duties and liabilities. For example, if the ruler has a claim-right to rule, then this corresponds with a duty on behalf of the ruled, perhaps a duty to obey. One might argue that there is a much greater demand that the existence of such a duty should be epistemically accessible to us when compared to the mere absence of a claim (i.e. the correlate of a liberty-right).

Focusing on these more expansive notions of legitimacy makes the issue of intransparency more vivid, but I do not think it requires any fundamental changes in the argument. First, we should note that the objection runs the danger of begging the question. The objection seems to rely on some implicit background commitment — for example, that one can only have a duty to ϕ if it is epistemically accessible that one has such a duty. But this is simply a version of perspectivism, which the critic of transparency is likely to deny. Either way, adjudicating whether such a generalized form of perspectivism is true is beyond the scope of this paper.

On the contrary, it seems entirely possible to be subject to someone else’s authority without knowing that one is. Consider once again the case of a radical feminist government dismantling patriarchy. We can imagine that such a government has a power-right to make law and a claim-right to obedience, even if the existence of these rights is not epistemically accessible to some. Such a government, then, would have the ability to create new duties and rights for sexists who deny its very right to exist. My suggestion is not that this is the normal situation of rulers vis-à-vis ruled, or that it is not regrettable in important ways. But it is intelligible, possible, and in some situations real. If so, then legitimacy is intransparent, even if understood to be a quite expansive right to rule.