National Security, Self-rule, and Democratic Action

Abstract

Most discussions of the relationship between liberty and security focus on the idea that enhancing citizens’ security may require imposing constraints on their civil liberties (e.g., freedom of association, of movement, of communication, and so on). This paper explores the question of how measures to enhance security stand vis à vis the idea of political liberty, i.e. the idea of citizens’ collectively directing the power of their state. It distinguishes two models whereby citizens might enact that ideal of self-rule and argues that with respect to issues of national security, the less direct model, which entrusts political agents to make decisions beyond direct democratic input, will often be more appropriate. It argues as well that various practices often seen as fundamentally at odds with the ideal of rule by the people (e.g., government deception, lack of transparency, covert action) are in fact consistent with a reasonable construal of that ideal. It concludes by outlining various criteria that would have to be met for such practices to be morally permissible in democratic states.

Two Models of Self-Rule

Most discussions of the tension between the values of security and liberty see the issue as essentially involving a specific trade-off: how far should citizens compromise their civil liberties (e.g., freedom to move anonymously, to assemble without supervision, to communicate privately, and so on) for increased protection of their personal safety? That way of framing the issue obviously gets something right. But focusing only on that tension ignores the connection between security and a different kind of liberty, viz., the liberty to influence and direct state action, often called political liberty. The locus classicus for the distinction between the two kinds of liberty is Benjamin Constant’s celebrated 1819 lecture in which he distinguishes between “our liberty, [which] has to consist of the peaceful enjoyment of private independence,” and “the liberty the ancients had, consisting in constant active participation in collective power” (Constant 2017: 5). While the ancients were concerned “to share social power among the citizens of a single country,” the aim of us moderns is different: it is “to be secure in our private benefits” (Constant 2017: 7). It’s not surprising, then, that insofar as Constant does register the importance of political liberty in modernity, it is chiefly in terms of its instrumental value in protecting our civil liberty. Nor is it surprising, given the trend Constant discerned in modernity, that contemporary discussion of how to ensure security is almost always cast with an eye to potential costs to our civil liberties.

In this essay I want to explore what strikes me as an underappreciated problem about political action in contemporary democracies that emerges when we consider the value of security in relation not to our civil liberties but to our freedom as political agents. Central to liberal democracy is the idea that state power should reflect the will of the citizens. This principle reflects the enormous value we give to the idea of self-rule, i.e., to the idea that state action should as far as possible constitute an expression of its citizens’ will and agency. But there are, I think, different ways to understand the connection between self-rule and democratic institutions, and in what follows I shall concentrate on two rather different ways of understanding how self-rule can be enacted. On the first model, which we might call the Rousseauian conception, political actors figure ideally as direct, unmediated extensions of citizens’ own agency. On this conception it follows that citizens should shape state action as directly as possible and, accordingly, that state agents should strive for maximal transparency in their actions. On the second, which I call the model of entrusted authority, citizens see themselves as enjoying the benefits of being governed by political actors whom they periodically elect in the expectation that those actors will work on their behalf and for their benefit. I shall construe these models as ideal types.

The model of entrusted authority has struck many (as it did Rousseau) as an abandonment of genuine self-rule. But that dismissal is, I think, too quick. To see why, consider first a parallel connected with the ideal of autonomy. Imagine a liberal regime committed to ensuring that its citizens all have a chance to lead autonomous lives (i.e. lives they choose for themselves, free of coercion, after considering what kind of life they most want to lead from a range of options). In such conditions some will choose a life of religious worship within communities marked by various hierarchies whose members are not encouraged to exercise various traits (critical reflection, revision of their commitments, originality, and so on) that seem connected to the ideal of autonomy. Shall we say that such people are leading autonomous lives or not?

Some part of us, I acknowledge, is inclined to say no, precisely because such persons are not robustly exercising the virtues that so clearly connect to autonomy. But some part of us (or at least: some part of me) is also inclined to say yes, so long as the persons in question have chosen those lives in the knowledge that there are a range of options available to them and through a process that involves some assessment of the value of that way of life as compared with others on offer, in a way that inevitably reflects their own temperament, proclivities, and so on. On one interpretation of autonomy (call it active autonomy), we may conclude that largely obedient religious affiliates are not leading autonomous lives. On the second (call it foundational autonomy), we will conclude they are. How shall we resolve this dispute?

I want to suggest that foundational autonomy is the better interpretation of autonomy. My position hinges on the following claim about autonomy’s value: the reason political communities should protect individual autonomy is that doing so creates conditions most conducive to human well-being. Because different people have different aptitudes, temperaments, and the like, their lives are likely to go best when they choose for themselves how they shall go even where these do not involve the robust active exercise of autonomy. Where those choices have been made in appropriate conditions and honor others’ rights to the same, there are no grounds from the standpoint of the value of autonomy to criticize them. The idea is that the value grounding our commitment to autonomy (i.e., human well-being) should lead us to recognize as equally valid expressions of autonomy the diverse ways in which people choose to lead their own lives. This is why I say that foundational autonomy is the more appropriate account given the reasons underlying our overall commitment to autonomy.

In taking up the status of the Rousseauian conception versus the entrusted authority model as interpretations of the idea of self-rule, I shall pursue an argument structurally similar to that which I have just deployed in exploring autonomy. My goal, however, will not be to show that entrusted authority is in all respects a superior interpretation of self-rule as compared with a Rousseauian conception. Instead, I will aim to show only that it is a plausible conception of the idea of self-rule and that in at least one domain (that of national security) it offers substantial advantages over the Rousseauian conception.¹ Having done that, I shall pursue some implications of my argument with respect to questions surrounding the place of transparency and secrecy within democratic regimes, and I will end by suggesting that there are reasons to see government secrecy not just as compatible with democratic ideals but as a reasonable extension of them.

Before commencing with that argument, two caveats. First, as with the argument on autonomy just sketched, I shall take the promotion of the well-being of persons in any political community as the touchstone by which the relative appeal of the two conceptions of self-rule should be determined.² Apparently competitor accounts (for example, Michael Walzer’s appeal to self-rule as a guarantor of communal self-determination (Walzer 1985)) are, I think, ultimately parasitic on the justification I have just offered, since the best defense of the value of self-determination is that it is profoundly good for individuals to live in a community that expresses their own distinctive practices, norms, and so on. On this approach, the central question is whether the interests of members of political communities are, in at least some contexts, better advanced by construing self-rule through the entrusted authority model than the Rousseauian alternative. The affirmative answer I defend relies on two claims, defended in the following two sections: that the model of entrusted authority is indeed a plausible interpretation of the ideal of self-rule (§ 2), and that in the context of national security it represents a better way of protecting the values underlying that ideal (§ 3).

The second, related point is that despite my use of the phrase “entrusted authority,” what follows is not a comprehensive investigation into the source of state authority with an eye to establishing conditions of legitimacy, and I shall largely bypass foundational questions of the relation between democratic institutions and state legitimacy. I do so in part because if the entrusted authority model is (as I suggest) compatible with the fundamental ideal of self-rule at the heart of democracy, then an argument is available that it too possesses whatever legitimacy democracy confers. But I do so as well because my central concern here is simply a separate issue. If we distinguish the questions (i) what characteristics make political institutions legitimate, and (ii) what characteristics make political institutions likely to be most effective at protecting the basic interests of those who live under them, my concern is chiefly with the second. This may seem to leave unrebutted those who reject accounts of state authority that rely heavily on welfarist considerations, and I grant that I cannot here engage that enormous issue. But in defense of my approach, two points. First, it would be odd in the extreme if an account of democracy’s legitimacy were utterly independent of how well democratic institutions advance human well-being, so in that respect the argument I offer will have at least some bearing on questions of legitimacy. Second, even if the two issues are logically independent (as strict non-welfarists might insist), the question of what political arrangements best advance citizens’ well-being remains of great importance to anyone theorizing the political.³

The Idea of Entrusted Authority

The case for seeing entrusted authority as a plausible interpretation of self-rule parallels the autonomy analogy, in two ways. First, imagine a person who has control over some area of her life but decides that in this area someone else (a close friend, trusted mentor, etc.) is much better positioned to make decisions that will be in her interest. She then chooses to follow decisions in that area made by that other person. This deference is conditional and lasts only as long as the person wishes, which will likely be only so long as she finds that her life goes better under that arrangement. So long as it is up to her whether to follow the other’s decisions, then allowing that other person’s decisions to be authoritative for her is how she chooses to rule her life.⁴ Similarly in the collective case: should a community judge that it wishes to entrust certain decisions to persons best situated to address them (for reasons explored in § 3 below), then doing so is how that community enacts self-rule in its own case. As before, the argument for seeing this as a case of self-rule derives from the reasons justifying the commitment to self-rule in the first place, i.e. the understanding that a community’s interests are likely to be best advanced when it decides for itself how collective problems shall be faced.

The second parallel with autonomy that strengthens the case for seeing entrusted authority as a form of self-rule is that just as there is no guarantee that a person will exercise their autonomy well, so is there no guarantee that a community will hit upon a form of self-rule that best advances the interests of its members. But for entrusted authority to stand as a plausible conception of self-rule we needn’t establish that in exercising it a community always advances its citizens’ interests. We need only show that with regard to at least some cases there is good reason to think it will do so.

There is an obvious objection to my argument so far. Some may insist that a community’s right to self-rule is independent of any connection self-rule makes to human well-being. (Indeed, the argument I’m advancing might appear vulnerable to a devastating reductio: if a community decides to advance its interests by living under a benevolent despot, on my account this qualifies as a form of self-rule!⁵) If, as the objection alleges, the value of self-rule does not hinge on the connection between self-rule and the advancement of citizens’ interests, then the argument I just offered—which relies on the idea that in certain instances the model of entrusted authority might best advance the interests of a community—would gain no traction.

Now I have already made clear that I believe the best defense of communal self-rule is grounded on its connection to well-being. But it’s crucial to see that the case for entrusted authority does not directly hinge on that claim. It does not because however the ideal of self-rule is vindicated, it is still the case that self-rule confers on a community the right to structure its political arrangements so as to best advance its citizens’ well-being if it wishes to. To see the point, consider a parallel objection against my earlier discussion that grounds autonomy not through its connection to well-being but in some other way. One could grant that alternative grounding and still allow that a person may exercise autonomy in ways they think will advance their well-being. Similarly, even if the ideal of self-rule is not itself grounded in a claim about what would advances the interests of a community, that ideal would permit a community to choose political arrangements that they feel best advances their members’ interests. So if a community (1) decides that it wishes to adopt political arrangements that best advance the interests of its members, and (2) judges that the model of entrusted authority offers the best hope of advancing its interests members, then the model of entrusted authority would be vindicated even if we don’t ground self-rule directly on well-being.

Two further reasons for seeing entrusted authority as a plausible interpretation of self-rule are found by examining more closely how political rule operates in all existing democracies of any substantial size and complexity. The first concerns the general relation of authorization which holds between citizens and their elected officials. Authorization can take two forms, explicit and tacit.⁶ In the former, I state explicitly what I authorize you to do. In the latter, I authorize a goal without specifying the steps necessary for you to do what I have explicitly authorized. For example, in explicitly authorizing a mechanic to fix a problem with my engine, I tacitly authorize him to unlock my car, pop the hood, turn the ignition, and so on. It is obvious that a great deal of state action is authorized by citizens only in the sense of tacit authorization. This point will be especially important when we consider to what degree political agents may exercise independent judgment in their decision-making, but already it shows that a good deal of looseness is built into the idea of political agents’ enacting citizens’ will.

The other reason bolstering entrusted authority as a form of self of self-rule is that within democracies citizens author their state’s actions largely in indirect and attenuated ways. They designate persons who then determine what the government will do and who do so in part by appointing others to address a range of issues where citizens’ will can only be guessed at. Other than periodically selecting government actors, there are few avenues for ensuring that the government is in fact enacting the will of its citizens.⁷ If such institutions are consistent with the ideal of self-rule animating liberal democracy, this suggests that the ideal can accommodate substantial departures from the Rousseauian conception.

Here it must be acknowledged, however, that probing questions have been raised about how far such elections are in fact consistent with the ideal of people ruling themselves, with some expressing the concern that those who win election are likely to differ from those who have elected them in ways that call into question whether self-rule genuinely obtains.⁸ But even if that general claim concerning the dynamics of elections is granted (I shall return to it in a moment), it does not follow that it vitiates self-rule within such communities. After all, any large-scale political community requires both that some designated group of citizens discharge central tasks of governance and that there be procedures for designating such persons. Designation by lot (i.e. through chance) offers one such approach, allowing citizens freely to elect specific persons another.⁹ And while selection by lot eliminates the possibility of electing persons deemed superior to ordinary citizens in some respect, it is not at all obvious that this enhances self-rule. Instead of imagining that self-rule requires similarity between ordinary citizens and those who occupy political office, we can just as well see it as requiring that citizens themselves determine who shall occupy such office. Indeed, if we assume (as is surely the case) that current citizens of liberal democracies would overwhelmingly prefer that holders of political office result from citizens’ conscious selection rather than from the workings of chance, that fact must have moral salience in understanding what self-rule involves. Why think self-rule is better advanced by giving luck a greater place in the selection of those who hold office than by granting citizens collectively greater agency in that process?

It may seem, however, that this response on behalf of the entrusted authority model invites an objection over the degree to which the model is consistent with the commitment to political equality that liberal democracies embrace. Richard Bellamy is surely right when he asserts that liberal democracies are committed to what he calls “the equal political status of all citizens,” i.e. the idea that all citizens exercise equal political power and that in the political domain none is superior to any other (Bellamy 2019: 7). In permitting the allocation of particular tasks to elected government agents beyond citizens’ direct input, some may object, entrusted authority valorizes the distinction in political status between ruler and ruled in a fashion that undermines that ideal of political equality.

This concern can be pressed in two ways. First, it might be argued that in allocating authority to only some citizens the entrusted authority model implies that some are superior to others and so violates the principle of equal political status. But this is much too quick. As Bernard Manin observes, invoking a point from Carl Schmitt, “democracy does not necessarily exclude a functional differentiation between rulers and ruled.” Rather, it excludes relations embodying what Schmitt calls “a qualitative difference, … [the idea] that the rulers are in some way qualitatively better than the ruled.”¹⁰ The key question, then, is whether entrusted authority incorporates the recognition of such “qualitative differences” (a phrase Manin glosses as referring to “the superior nature” of those in office) in a way that threatens political equality. The answer hinges entirely on how such differences and judgments of superiority function within a political system. If the claim is that some people’s natural superiority entitles them to rule over others, that would indeed be highly problematic from the standpoint of political equality. But such a reading deeply misconstrues the entrusted authority account, according to which whatever authority is possessed by political agents derives from (i) the fact that citizens have chosen to confer that authority on them, coupled with (ii) the fact that some urgent task that citizens endorse can be carried out only if some persons are entrusted with that authority (this second claim I defend in § 3 below). It is true that citizens will want to have such positions occupied by persons they think especially likely to discharge those tasks well; in that respect the occupants of such positions may reflect citizens’ judgments of quality, superiority, and the like. But the authority attached to those positions derives not from any natural superiority, but from the fact that citizens have selected them through a process modeling political equality.¹¹

The other way of pressing the worry over political equality targets the fact that entrusted authority seems not just to tolerate but to be predicated on government agents’ exercising what Manin calls “a degree of independence from the wishes of the electorate,” and this may seem again to describe a relationship between political agents and ordinary citizens at odds with political equality (Manin 1997: 6). But this concern is defused once we examine more closely the notion of a political agent’s acting independently of the electorate’s will. On a strong interpretation, such independence implies an agent’s preparedness to act in ways that deviate from the electorate’s wishes, i.e. to enact policies and pursue ends their constituents do not endorse. That kind of strong independence is just what Edmund Burke championed in his famous 1774 speech to the Bristol electors when he endorsed the idea that representatives should determine through their own “reason and judgment” what ends ought to be pursued, rather than work to enact citizens’ will. On a weak interpretation, in contrast, the idea that government agents act independently of the electorate’s wishes implies only that they may act on matters where citizens have not directly expressed their will and judgment, though still in a manner constrained (as the strong interpretation is not) by the will of those ordinary citizens. On the strong account, in which government agents substitute their own will for that of the people, the ideal of political equality is severely endangered; on the weak account, where such agents work to enact the people’s will, it is not. This response assumes, of course, that there are some areas of governmental decision-making where there are good reasons why citizens will not have more directly expressed their will. The next section defends that assumption.

National Security, Secrecy, and Deception

Having argued that the model of entrusted authority stands as a form of self-rule, I turn now to showing why that model might, with respect to the question of national security, be superior to the Rousseauian conception. I begin by distinguishing three kinds of arguments that might be made on behalf of either model.

The first appeals simply to a community’s preference for one over the other. Given that each model is understood as an interpretation of self-rule, satisfying this condition is probably a necessary condition for the acceptability of either model, and perhaps a sufficient condition as well. But I will have little to say about this sort of argument here, for two reasons. First, rational citizens want their preferences to reflect the balance of reasons for and against each model, so it makes sense to concentrate directly on the reasons favoring one model or another. Second, I am not at all sure how to identify citizens’ preferences here. Should we look at how much time they spend being politically engaged—how much they vote, are active on specific issues, donate their time to political campaigns, and so on? Even if we could measure such engagement, what would that reveal? Perhaps higher levels of engagement show a preference for the Rousseauian conception, lower levels a preference for entrusted authority. But how could we be sure that in communities with lower engagement citizens don’t also prefer the Rousseauian model but have simply given up on robust participation as a result of disillusionment, fear, feelings of disempowerment, and so on? Why think higher levels reflect a preference for the Rousseauian model and not simply a desire to elect politicians who will govern well and allow citizens to pursue the private interests associated with their civil liberties? Appealing to citizens’ preferences thus does not seem promising as a way to adjudicate between the two models.

A second kind of argument appeals directly to the values at the heart of each model. Since the Rousseauian model gives greater weight to the value of political liberty, to assess its claim we need some account of why political liberty is so valuable. I see two main ways of making that case.¹²

First, it might be that participating in collective governance is an especially worthwhile activity for all people to pursue. This line of thought faces serious challenges. To begin with, many contemporary theorists resist invoking judgments about intrinsically valuable forms of life, seeing them as an arrogation of moral authority at odds with a commitment to equal respect. And even if political engagement is intrinsically valuable, so are a range of other goods (knowledge, friendship and sociability, artistic achievement, etc.) whose pursuit might be diminished by the vigilant exercise of political liberty.¹³ This first approach would thus have to show that a life involving the exercise of political liberty is somehow better than a life built around those other goods. Aristotle notoriously seemed unable to make his mind up whether such a case could be made, and in the modern age the prospects for success seem even less promising. Nor could the Rousseauian invoke the more modest claim that political liberty is simply one viable means of human flourishing, because the model of entrusted authority affords citizens various ways to exercise political liberty.¹⁴ I am doubtful, then, that the Rousseauian conception can be vindicated by appeal to the intrinsic value of political liberty in a flourishing life.

The other way of explaining the great value of political liberty appeals to the civic republicanism that Quentin Skinner sees running through figures like Machiavelli and Harrington (Skinner 1991). The idea here, invoked by Constant as well, is that the adequate exercise of civil liberty can be assured only where citizens collectively attend to the res publica in ways that involve the exercise of political liberty. Political liberty has enormous value not in itself, but because only by exercising it can citizens ensure their civil liberty.

In Skinner’s reconstruction of Machiavelli, this argument for political liberty rested heavily on the idea that actively defending the nation was the responsibility of every citizen. In many modern states this role has been professionalized and delegated to a group of citizens with distinct responsibility to protect national security. But the argument from civic republicanism can be broadly reconstructed, by appealing to the substantial role of the state and its functions today. The idea here is that as the extent and office of the state has increased—as the state has become a massively powerful agent whose reach extends in manifold ways to the ongoing activity of ordinary life—the need to ensure its proper functioning takes an importance comparable to Machiavelli’s concern with defense of the republic. If not rising to a matter of life and death, we nonetheless need state actors to discharge their duties effectively, fairly, free of corruption, and so on. Citizens might try to ensure this either by themselves collectively discharging those duties in the proper spirit à la the Rousseauian conception, or by exercising active vigilance over the decisions made by their designated actors in some version of the entrusted authority model. If we assume that citizens are moved by the appropriate concern and motivation, it may be that the Rousseauian conception, in offering a more direct form of control, is to that degree preferable, all things being equal.

But all things are not equal, and this matters greatly when we consider the third kind of argument that might be offered to adjudicate between the two models. This argument takes its cue from Elizabeth Anscombe’s suggestion that we should understand state authority by appealing to a task that needs to be completed and the conditions necessary to discharging that task.¹⁵ Adapting Anscombe’s insight, I want to suggest that we can adjudicate between the Rousseauian and entrusted authority models by showing that one better advances some important goal that can reasonably be seen as the task of the state. That fact would constitute a pro tanto reason to favor one conception over another, and where the goal is very important would constitute a very strong reason indeed.¹⁶

This line of argument requires specifying the important goals that state action advances. Some of these will reasonably reflect the distinct choices of political communities: some may pride themselves on a robust system of public education, others may emphasize protecting cherished natural beauty, still others may celebrate their system of extensive national health care, and so on. These seem to me goals to which states may reasonably give differing priority, reflecting the natural variation in what people care about. But along with these discretionary goals, there exist goals to which all states are committed—goals which states as such must care about, and which can be seen as the proper aim of all states.

I suggest we see national security as one such goal, taking national security to comprise at least the following: integrity of physical borders, personal safety of citizens against attack, and maintenance of civil society against substantial disruption. To be sure, nations take different steps in protecting national security, with some devoting proportionally far more of their resources to so-called “defense spending” than others. But those differences reflect a range of differentiating factors—the perceived level of threat, the overall militarism of any national culture, the degree to which states seek to exercise force around the world, and so on—and do not undermine the idea that preserving national security is a central goal for all states.

Right here, as my discussion turns to a particular goal of state action, it is important to see that there is no reason to think we must endorse either the Rousseauian or entrusted authority model as an all-or-nothing account of how state power should operate, tout court. One model may be more appropriate in some domains, the other in others.¹⁷ Our question here is whether there are reasons to endorse entrusted authority with respect to the goal of ensuring natural security.

In answering this question, note first how much we accept something like the entrusted authority model in times of war, how far we tolerate state actors in such conditions making a range of decisions beyond the direct oversight of ordinary citizens. We accept that such actors and their designates will make decisions about troop preparation, weapons deployment, strategic planning, and the like. In times of peace as well we largely tolerate the existence of covert intelligence and security agencies whose activities are not directly accountable to citizens. Various factors explain such acquiescence. Partly we want effective decisions made in a relatively speedy time frame that can’t be achieved if all citizens must be consulted. Also, we want decisions made by people with distinct expertise.¹⁸ But a third factor, especially important to our discussion here, is the need for decisions to be made that by their nature must be hidden from the public. This built-in requirement of secrecy, and the steps attendant to secrecy, constitutes an especially strong reason favoring the entrusted authority model in matters of national security. I say this because I take it as axiomatic that the Rousseauian model entails that state decisions should, in Glen Newey’s phrasing, “be maximally public … [and] maximally open to public scrutiny” (Newey 1997: 93). After all, for citizens directly to exercise state power they must evaluate the reasons behind various decisions and know what actions issue from those evaluations. If meeting the challenges of national security requires substantial secrecy, this would bolster the case for entrusted authority.

The chief consideration recommending secrecy in this context is the existence of enemies who wish to harm a community and whose efforts would be substantially aided if they knew more about what their target nations were doing to combat those threats and what information those nation possessed about the agents of threat. Where these conditions obtain, maximal transparency diminishes the ability of state actors to combat those threats and so ensure the security of their citizens. Martin Hollis neatly summarizes the case: “there are evil men about who do not wish the good of the people and who do not keep faith; … they can be thwarted only by marshaling the apparatus of legitimacy against them; … the apparatus has to be used dishonestly” (Hollis 1982: 396).

Hollis here directly raises the question of the permissibility of deception, and I recognize that to argue for secrecy is not to argue for deception. Some, seeing an important distinction here, might countenance government secrecy for reasons just mentioned but insist that government deception is another matter entirely. But I believe that the grounds legitimating state secrecy also naturally extend to justify government deception and lying, and that this further strengthens the case for the entrusted authority model, which can accommodate such deception more easily than the Rousseauian account.

A first reason to think the permissibility of deception follows from the permissibility of secrecy is that there may be instances where deception is the only way to maintain secrecy. Imagine that a government plans a covert rescue mission and that agents of the government are directly and publicly asked about such plans. Imagine also that attempts to avoid answering will only fuel speculation that the government is planning the action in question. If (1) government secrecy about these plans is permissible, and (2) deception is necessary to maintain that secrecy, then it follows prima facie that (3) government deception about these plans is permissible. Some may reply that governments should instead minimize conditions where such deception is needed, but that would necessitate fewer opportunities where government officials comment publicly and so would come at the cost of government transparency.

This point connects to a second reason favoring government deception, rooted in a fascinating argument that Glen Newey has offered in defense of political lying. Newey locates the wrongness of lying in its violating another’s legitimate expectation to the truth. From this it follows that a lie is not even pro tanto wrong if the hearer has no right to the truth. (This neatly explains why, contra Kant, it’s permissible to lie to a murderer at your door.) Newey then considers a scenario where citizens have authorized their government to pursue an end whose successful pursuit requires lying to citizens. His point is that respecting citizens’ will may require that the government lie to its citizens. Because his argument does not hinge on the claim that governments know what is good for citizens better than citizens do, Newey can plausibly claim that this defense of government lying is not a case of paternalism. To the contrary, in cases where citizens “have autonomously authorized the lie,” lying is not just compatible with respecting their autonomy but actually required by it (Newey 1997: 105)! Who wills the end wills the means: if citizens collectively authorize such goals (or, as in my argument, if the goals can be uncontroversially attributed to them), then they cannot be understood to have authored constraints that would make it impossible to achieve those goals. Newey’s argument naturally extends beyond lying to countenance deception, lack of transparency, and a range of practices that seem to violate democratic process norms.¹⁹

One concern about Newey’s argument is that it risks being dangerously expansive; to say that states may take any steps necessary to achieve an end citizens have authorized seems too permissive a principle. This worry is to some degree addressed by Newey’s reference to the fact that democratic authorization is bounded by “constitutionally entrenched side-constraints (e.g. protecting minority rights)” (Newey 1997: 109). Though offered as a constraint on the ends citizens may authorize, and not on what states may do to achieve those aims, a natural reading is to see those side-constraints as also limiting what states may do to pursue such ends. On this reading, if some goal is achievable only by violating entrenched side-constraints, then citizens’ authorization of that goal does not legitimate violating such side-constraints. It is thus a central question, for both Newey’s account the entrusted authority model, how to identify those side constraints. Newey’s appeal to constitutional principles is a start, but given that constitutions can be seriously flawed (the U.S. constitution originally regarded slaves as three-fifths persons), that appeal alone cannot be decisive for determining what is permissible. In addition, because they operate at high levels of generality, constitutional principles may be insufficient to resolve questions about the permissibility of specific state actions. There may often be no alternative to casuistic moral reasoning about what sorts of deception are and are not consistent with the idea of a self-governing community.

More optimistically, some have suggested the tension between state deception and democratic norms of publicity can be resolved by citizens’ authorizing such deception ahead of time. In discussing the relationship between democracy and dirty hands, Stephen De Wizje invokes Sisela Bok’s idea that deceit and manipulation can be rendered compatible with the central idea of democracy so long as they have been “approved and authorized in advance through democratic procedures.”²⁰ But in order for them to be not just sanctioned by democratic will but effective as well, such approval and authorization would again have to be pitched at an extremely high level of generality, potentially weakening the force of those constraints in their practical application. To permit specific acts of deception, in other words, any such policy would have to leave state actors ample discretion to determine whether conditions are such that deception is warranted. But this is just the entrusted authority model.

When May Governments Deceive?

It should be clear that I am not defending as a general principle the permissibility of governments either lying to their citizens or acting in ways that bypass citizens’ will. It is undeniable that transparency, truth-telling, direct accountability, and the like are important pro tanto constraints within liberal democracies, overridable only when certain conditions are met. The question is, what are those conditions? I suggest that in order for government deception to be permissible, the following conditions must all be met. (In cowardly fashion, I do not claim the list is exhaustive.)

• (1) The actions supported by deception substantially advance important goals.

• (2) Deception is vital to the success of those actions.

• (3) Citizens have not demonstrated their clear opposition to their state’s taking such actions.

• (4) The actions are not in themselves impermissible. ²¹

I shall consider these individually in a moment. Before doing that, observe that people may reasonably disagree over how far any of these conditions are satisfied. For example, we can imagine strong debate over whether various actions (such as eavesdropping, sabotage, or targeted killing) will substantially enhance citizens’ security, how far their success hinges on their being kept covert, and so on. Disagreement on such questions of fact may lead to differing judgments about the permissibility of deception in any case without thereby indicating any disagreement over either the necessary conditions for the permissibility of deception or the account below.

• (1) The actions in question substantially advance important goals.

Given the substantial pro tanto weight of truth-telling, transparency, and the like, this first condition seems uncontroversial. Still, it’s worth noting that the distinction drawn earlier—between goals that reflect discretionary interests and those that reflect interests we can confidently attribute to states anywhere—may generate at least three potential areas of dispute over the application of condition (1). First, regarding discretionary goals, there may be disagreement about whether it is correct to attribute any specific goal to a community or about the relative importance of different discretionary goals within any community. Second, there may be disagreement over which goals can be universally attributed to all communities. Though above I listed concern for an extensive health care system as a discretionary goal to which some communities may give special weight, others may argue that this is a goal attributable to all political communities. Last, there might be questions about relative priority within particular communities between discretionary goals and universal goals. To the degree that all of these disputes can be resolved through public discussion marked by transparency, that is surely ideal.²² But even if we grant the complexity of these disputes, it’s hard to imagine that maintaining national security is not a goal that citizens care deeply about and one that advances their interests everywhere. If deception can be justified in any context, it is surely this one.

• (2) Deception is vital to the success of those actions.

At first glance this condition seems obvious: if an action’s success doesn’t require deception, there seems nothing to override the pro tanto commitment to transparency in democratic states. Some may then wonder how the interest in deception could arise at all if not connected to some action whose success relies on it. There are, however, a variety of contexts in which state agents might be motivated to deceive citizens even if the deception is not aimed at enabling a particular state action, one of which in particular raises deep questions about the account I am offering.

One altogether pedestrian case, all too familiar, is that of political actors engaging in deception entirely for reasons of self-interest and personal gain; this is philosophically uninteresting. More interesting are cases where agents of the state seek to hide actions because they deeply violate a nation’s self-image and/or professed norms; the US government’s campaign of torture in the 2000’s under the George W. Bush administration is perhaps a good example. Some theorists have argued that maintaining national myths is a salutary and in some cases appropriate goal in political action, so there may be more to be said for such deception than first appears.²³ But there is, I think, a conclusive reason against it. Given normal processes of moral education and enculturation, it is overwhelmingly likely that citizens of a political community will see their governing norms as broadly correct. And as noted earlier, we should assume that what is permissible for political agents depends closely on what citizens believe it is permissible to do.²⁴ In the context of entrusted authority that means acting in ways that are consistent with those norms. Independent of the objective correctness of those norms (which, I argue below, can entail their own prohibitions), political actors entrusted with representing their citizens’ will have powerful reasons not to violate those communal norms.

Far more interesting questions arise in cases where political actors consider deception because they fear citizens might express strong opposition to what is being done, making it harder to satisfy condition (3). (This fear obviously connects with the worry about violating public norms, but it’s a different concern that raises a quite different challenge.) If we accept that state agents should not act in ways that contravene the will of their citizens, we may think it blindingly obvious that deception in this last kind of case is impermissible. After all, what could legitimize state deception undertaken not to advance its citizens’ will but to thwart their ability to express their will on some issue? The issue, though, is more complicated.

To see why, consider John Mearsheimer’s discussion of fearmongering in his bracing account of the lies governments tell: “Leaders engage in fearmongering when they think they recognize a serious threat to national security that the public does not see, and that the public cannot be made to appreciate with straightforward and honest discourse. They reason that the only way to mobilize their citizens to do the right thing is to deceive them for their own good” (Mearsheimer 2011: 54). Now in the model of entrusted authority, citizens are understood to have authorized political agents’ discretion to act in ways that advance national security. Following Newey, I have argued that this authorization can legitimize deception. But now a serious problem emerges: if deception is undertaken to ensure that citizens will either tolerate or support steps that genuinely advance their security, which they otherwise would not, then why not see such fearmongering as a form of permissible deception à la Newey’s account? After all, it too employs deception to advance a goal that citizens have authorized their agents to pursue. If it seems outrageous to so defend fearmongering, how can we stop the model of entrusted authority from sliding towards such a defense?

Here one might simply concede that there’s no way of stopping that slide and that this shows how far the model of entrusted authority departs from self-rule. Strong advocates of self-rule might then see this conclusion as a reason to reject entrusted authority. Weaker advocates may instead see it as indicating the need to abandon self-rule in matters of national security. But might there be a way to stop that slide, to account for the permissibility of deception in a way less vulnerable to such expansion? Any such argument will, I think, need to do two things. It will have to clarify and shore up our common-sense distinction between means and ends, and it will have to distinguish objectionable cases of paternalism from the kinds of cases that reasonably fall under entrusted authority and which seem to share some of paternalism’s features. Here is the beginning of such an argument.

In the normal course of events we unavoidably, and quite usefully, distinguish between means and ends. In the political context, this is the distinction between goals and policies. Often strong agreement exists on goals (we want peace with the Soviets, economic growth for the poor, greater racial equality, etc.), and debate concerns the best ways to reach those goals. My argument thus far has condoned deception chiefly as a valuable means in the service of an end citizens obviously endorse. Fearmongering, however, is a categorically different enterprise, because it employs deception to convince citizens to adopt an end they would otherwise not have.²⁵ A fully adequate account of the means-ends distinction would likely require a notion of something like proximity or contiguity to distinguish between cases where an end political agents seek to bring about through deception is directly nested within a broader goal endorsed by citizens, and where on the other hand an aim is sufficiently distant from citizens’ ends as to raise questions about whether its pursuit is genuinely an extension of citizens’ will. A full response to the fearmongering case thus requires an account that can distinguish between (i) appropriate instrumentality in the service of an end endogenous to citizens, and (ii) the substitution by political agents of their own end in the place of citizens’.²⁶

We can get further purchase on this distinction, as well as defuse some of the worries of paternalism, by considering a stylized example from medicine. Doctors routinely prescribe treatment for reasons patients do not and cannot understand. Patients’ trust their doctors to do precisely this, and few of us find this a case of objectionable paternalism.²⁷ Imagine now that there were some highly effective treatment whose success depended on patients’ misunderstanding why it was beneficial; for the treatment to work, doctors would need to mislead their patients. Is this objectionable paternalism? Our intuitions are torn here, in part because the doctor is so clearly working to advance a goal endogenous to the patient. Observe now that some of the structural features of this case are echoed in the context of national security, on account of the existence of bad actors whose threat level would be increased were government measures to thwart them more publicized. In both cases there are good reasons for deception grounded not in the aim to deceive but in the simple fact that such deception is necessary to advance goals endogenous to the parties in question.

This line of argument suggests that we might further block the expansion to fearmongering by appeal to something like the doctrine of double effect. In the standard case of state deception that I have imagined, the government aim of protecting citizens from persons who threaten their security requires depriving those persons of information that would make them more dangerous. It can do this, however, only by keeping such information from its citizens as well. So even though deception in both the standard case and fearmongering case has the effect of creating a false belief, the nature of the acts is different: the first is an act of deceiving enemies (which has its consequence deceiving citizens), while fearmongering is an act of getting citizens to develop false beliefs. All of this, I recognize, requires far more elaboration than I have provided, but I hope I have said enough to suggest strategies to prevent the argument I have offered from expanding to sanction measures that seem insupportable from the standpoint of democratic ideals.

• (3) Citizens have neither expressed strong concern over their government’s actions nor demanded greater transparency from their government regarding them.

This condition tries to reconcile the idea that governments should not act against citizens’ will with the central impulse behind the entrusted authority model. For that reason it is less stringent than a requirement that citizens must have demonstrated clear support for the actions in question. I have opted to capture the idea of consistency with citizen will in a less demanding way because the argument for entrusted authority begins in the recognition that states sometimes need to act in ways that are not immediately and directly accountable, which means that citizens will not be in a position to know what is happening and so will be unable to demonstrate their support. A stronger condition would therefore rule out precisely the kinds of action that motivate the model of entrusted authority.²⁸

Even this comparatively weaker requirement, though, may seem to rest on a logical howler: since citizens can express opposition only to actions of which they are aware, how can this condition have any force if we are imagining actions that are not made public? The response is that when state actors substantially depart from authentic citizen will and act in ways that may engender strong opposition if known, the actions of various agents (government agents talking off the record, lower-level personnel leaking information, the hard work of investigative journalists, political representatives with privileged information, and so on), along with reports of the effects of such actions on those affected, can work together to create enough awareness so that concerned citizens can press their government for greater accountability. Admittedly, this places some important responsibility on a group of custodians watching over the res publica. But it is in the nature of any relationship of trusteeship to rely, to some extent, both on the virtuous behavior of those who directly bear such trusts and on those outside the relationship who can exercise additional oversight. The alternative, i.e. demanding full transparency in all state action, forgoes important advantages covert action might offer. This is precisely the logic behind entrusted authority: to forgo direct accountability so as to retain those advantages.

When sufficient numbers of citizens clamor for greater details and information about such state actions, what then should happen? The entrusted authority model would cede too much if any expression of objection were deemed sufficient to render deception impermissible. One option, mentioned earlier, is for governments to meet those objections by making the case for secrecy with as much publicity as the matter allows. Dennis Thompson suggests a two-level solution along these lines: permit secrecy within democracies (at the 1st level) so long as the decisions for such steps have been made publicly (at the 2nd level) (Thompson 1999). But Thompson notes that this approach often will not resolve all competing values, in part because public discussion of the grounds for secrecy will sometimes compromise the secrecy sought. Thompson thus concludes that the need for secrecy may inevitably carry some cost to democracy, insofar as striking the balance between the competing goals involves relying on state agents who would sometimes assess the proposal for secrecy out of the public eye.²⁹ Such steps constitute a substantial retreat from the Rousseauian model, and to endorse them is already to move some way towards the entrusted authority model.

Any move in that direction and towards any such two-tier model of deliberation might be challenged, however, by appeal to an ideal of public reason that many see as central to liberal democracies. On this issue Simone Chambers has offered an especially nuanced discussion (Chambers 2004). Chambers’ concern is that governmental secrecy may allow sectarian considerations to influence state policy in a manner inappropriate for a democratic community governed by a shared conception of public reason. She identifies two main ways this can happen: secret deliberations might be informed either by private-interest bargaining, or by considerations reflecting parochial values or controversial truth claims at odds with other citizens’ reasonable views (Chambers 2004: 396). In either case, private reasons not shared by all reasonable citizens improperly replace public reason. But it’s important to see, as Chamber stresses, that public and private reason are defined neither by the settings in which they occur nor by the number of people involved, but by the kinds of reasons they are (392). It is a mistake, then, to conclude that closed deliberations must be marked by private reasons in an improper way, and it is perfectly conceivable that entrusted authority under conditions of secrecy will be discharged in a way that satisfies the constraints of public reason that Chambers recommends. What the worry over public reason suggests, then, is the need for attending to various mechanisms that will make it harder for private reason to find purchase in those discussions (through systems of oversight, rules about the composition of such deliberators, and so on). On the positive side of secrecy, Chambers notes that carrying out deliberations away from the public eye can bring a range of important epistemic benefits, by virtue of diminishing the effects of electioneering, manipulating, pandering, and the like. In the context of national security, where political grandstanding is an ever-present temptation, this last consideration only gains in strength.³⁰ So while the worry over public reason cannot be ignored, there is no reason to think that the requirements of secrecy recommended by the entrusted authority must violate public reason, and there are good reasons for believing that such secrecy may sometimes improve the quality of deliberation.

• (4) The actions are not in themselves impermissible.

Under this condition we should distinguish two kinds of cases. The first is when citizens overwhelmingly see as immoral actions which a political actor believes to be permissible. The resolution of this sort of case is, I think, independent of the correctness of citizens’ judgments. Where the citizens are correct, deception is impermissible because the action which allegedly justifies the deception is itself impermissible, and the pro tanto wrongness of state deception cannot be overcome by virtue of its importance to another wrong act. But for reasons already given, the same conclusion holds for cases where citizens are wrong in judging the action impermissible. Since political agents’ authority to act derives from their role as representatives of citizens’ wills, deception undertaken to enable the action in question is also ruled out, so long as we assume that citizens would not authorize action they take to be impermissible and that an agent’s authority to act on behalf of some principal extends only to what the latter believes she can permissibly authorize: if a principal does not see x-ing as permissible, she cannot authorize someone else permissibly to x or permissibly to deceive her in the service of x.³¹

The second kind of case arises when citizens disagree over whether the act supported by the deception is permissible. Consider again the policy of targeted killing, or the kind of detention without trial the U.S. carries out at Guantanamo Bay. While some citizens see these acts as clearly wrongful violations of basic human rights, others disagree, perhaps seeing the rights at issue either as possessed only by certain persons (e.g. U.S. citizens) or as properly overridable in existing conditions of conflict. One might opt here for a conservative approach and say that so long as the impermissibility view is plausible, that’s reason enough not to engage in the action, because non-action does not raise the possibility of engaging in a morally wrong act. The problem with this approach, however, is that if the action in question really does advance an important goal connected to national security (which it must do to be a candidate for permissibility), then failing to act may also be wrong. There may, in other words, be no option without moral risk.

Even if such risk is unavoidable, however, there may be better and worse ways of managing it. Here the concern over public reason that Chambers chronicled reappears with special urgency. I have granted that for entrusted authority to avoid the charge of paternalism it must be possible to sketch, in some plausible and non-question-begging way, a distinction between ends citizens have (which government agents are bound to pursue) and different policies for pursuing those ends (with respect to which such agents may exercise discretion and even secrecy when required). But as we have just seen, such a distinction may leave in place the problem of citizens who disagree about the permissibility of various means of pursuing some common end. For example, even if we attribute to all reasonable citizens the end of combating terrorism, they may disagree over whether measures like targeted killing or extended imprisonment without trial are consistent with their community’s basic norms of political morality. The greater the degree of secrecy attaching to deliberations over such matters, the less grounds for confidence that the decisions arrived at reflect the collective will of the citizenry. In terms of our earlier discussion, the worry here is not so much that norms reflecting an appropriate conception of public reason will not get a hearing, but that the fact of secret deliberations may impede the process of collective will-formation conducive to the identification of such norms in the first place.

It may be that here the idea of something like publicly recognized constitutional side-constraints, intended categorically to rule out certain actions once and for all, can play an especially important role. Such constraints can also serve as a bulwark against the moral slippage that is a fact of human nature, i.e. the fact that our judgments about the intrinsic wrongness of certain acts, which can seem quite confident when such acts offer no clear benefits to us, often wobble when we find ourselves in positions where we can advance important goals by engaging in them. (To skeptics of ticking time-bomb scenarios, this capacity for slippage is precisely what such scenarios rely on in seeming to dislodge our opposition to torture.) There may, then, be a good case for constitutionally enshrining firm and clear side-constraints as a form of collective pre-commitment by which citizens, in the clear light of dispassionate moral reasoning, remove options that might later be tempting to political actors. The exercise of entrusted authority, including permissible deception on behalf of state action, would then be bound by such constraints.³²

There is, however, a darker possibility that must be acknowledged. I have been proceeding on the idea that the case for entrusted authority connects to the values of expertise, secrecy, quick responsiveness, and the like. But that model may also offer citizens a far more problematic benefit. I refer to the degree to which entrusted authority shifts responsibility for state actions away from citizens and onto political agents more directly involved. This shift creates conditions where citizens can enjoy the benefits of actions done on their behalf without seeing themselves as directly responsible for them. In thus separating moral responsibility for action from the goods that action makes possible, entrusted authority creates a risk of moral hazard.

Comparing the two models in terms of the value to citizens of shaping state action (as I have done here) can obscure this risk, because that approach encourages us to see the choice between the two models in terms of one good (direct political input) which citizens might partially sacrifice to achieve some other (security). This makes it easier to see the issue as chiefly involving choices citizens are free to make for themselves. But along with thinking about how citizens might wish to balance these goods, we must also consider their responsibility to ensure that their state acts in morally defensible ways consistent with norms of justice. The moral hazard in entrusted authority is that citizens can eat their cake and have it too: enjoy the benefits of morally questionable acts their government carries out without directly bearing the responsibility that should go with such acts. For their part, political agents enjoy the prestige of power and its various perquisites. How far this is an accurate description of current political arrangements in the world I cannot judge, but we would be foolish to ignore the risks of this moral division of labor. The central idea of the entrusted authority model, i.e. that it is consistent with self-rule for elected officials to make decisions in the absence of direct citizen input, is predicated on the idea that citizens may freely designate others to make certain decisions on their behalf. If the model diminishes citizens’ incentive to ensure the decency of their state’s actions, that would constitute a serious drawback. That question awaits further discussion.