I. Max Weber: Domination, Legitimacy and Authority

Max Weber famously stated that the “domination” is “the probability that a command with a given specific content will be obeyed by a given group of persons.”^{Footnote 1} Domination is not a mere matter of advice or simple request; it implies an expectation of conformity to the standard set. However, putting too much emphasis on the predictive aspect of Weber’s statement would amount to understanding domination from merely an external point of view. Weber clearly saw the importance of also describing domination from another perspective. He completed his explanation of the notion of domination by highlighting how the manifested will of the political superior is meant to influence the behaviour of the ruled:

While the ruled may obey the ruler out of sheer expediency or out of habit, Weber considered that such types of order were “much less stable than an order which enjoys the prestige of being considered binding, or, as it may be expressed, of “legitimacy”’.^{Footnote 3} Reflexive and willful obedience are understood as being the product of the belief in the legitimacy of the norm posed by the political superior. Domination is considered “legitimate” by the subjects when the political superior’s claims to rule are supported by justificatory principles accepted by the ruled. Legitimate domination is what Weber also calls “authority”. It is a distinct form of power, a distinct way to influence others. There may be many justificatory principles that could support the different claims of authority, but Weber focused on three ideal-types of such justificatory modes: traditional, rational-legal and charismatic. We will mostly concern ourselves here with the rational-legal justificatory mode.

Weber’s ideal-type of rational legal authority is often taken as primarily meaning that legitimacy of a power of command derives from “a system of consciously made rational rules (which may be either agreed upon or imposed from above)” according to which “every single bearer of powers of command is legitimated by that system of rational norms, and his power is legitimate insofar as it corresponds with the norm”.^{Footnote 4} This mode of justification would appear not to rely on the content of the rule, but rather on the fact that it was correctly enacted.^{Footnote 5} The legitimacy of the rule would thus be content-independent. To support that point of view, we may recall that Weber’s theory of value “asserts the positivity of all norms”.^{Footnote 6}

But this would be far from a complete picture. Among other things, Weber also specifically wrote that

Thus, if rational-legal legitimacy rests mainly on legal formalities, Weber nonetheless recognized the role of the alleged grounds for the rule in legitimising law. If Weber does not take moral correctness to be a criterion of legitimacy per se, he nonetheless recognizes that legal norms are established “on grounds of expediency or value-rationality”; those are the grounds upon which the authority will base itself in establishing this or that specific rule. If all that was needed to establish the legitimacy of a rule was to point to the fact that it had been enacted according to the proper formalities, it would not matter that the political superior grounded legal norms on expediency or value-rationality. Thus, while legal-rational legitimacy appeals to legal formalism to justify the exercise of power, legal authorities nonetheless need to claim that their rules are properly grounded to appear as legitimate guides.

Weber’s quote is interesting not only because it highlights the possible role of the grounds of directives for their legitimacy, it is also important because it recognises that “value-reasons” may offer such grounds. Let us recall that a “value-rational” social action is one that is “determined by a conscious belief in the value for its own sake of some ethical, aesthetic, religious, or other form of behaviour, independently of its prospect of success”.^{Footnote 8} According to Weber, the more one takes a value as an absolute guide to action, the less one will be influenced by consequential reasoning, and the less one’s actions will appear rational from an instrumental-rationality point of view.^{Footnote 9} A law that would only be perceived as imposing costs or benefits would not influence an agent entirely moved by value-rationality. In fact, from a purely instrumental-rational point of view, value-rationality is always irrational.^{Footnote 10} However, it remains that once a value has been accepted, there is a form of rationality involved in orienting one’s action towards the fulfilment of that value’s demands.

Finally, a third element must be highlighted in Weber’s citation about authority. While the grounds of the authority’s directives contribute to their legitimacy, it remains that rational-legal legitimacy relies mostly on the presence of social structures that allow individuals to be seen as members of the organization on behalf of which the authority is exercised. Thus, rational-legal legitimacy may be more about social roles than it is about acting for the right reasons.

To recap before moving on to our examination of the Razian theory of authority, Weber explained that domination is much more stable when the directives posited by the political superior are considered legitimate, that is, binding because they are supported by adequate justificatory reasons. Legitimate domination is what Weber called “authority”. Legal-rational legitimacy relies on the presence of a set of norms within which authorities are bound to act. The grounds of their directives nonetheless contribute to the legitimacy of the latter. Such grounds may include value-reasons that have nothing to do with instrumental rationality. The existence of a set of norms that may legitimize the exercise of power presupposes the existence of a social organization and it is primarily as members of such organization that individuals are subjected to legal-rational authorities. Let us keep all this in the back of our mind while we examine the Razian theory of authority as it will later be used to develop our critique and alternative theory.

II. Legitimate authority and the grounds of authoritative directives according to the Razian theory

What does it mean for a legitimate authority to be able to claim that its directives are properly grounded? Raz’s highly influential normative theory of what may be considered a legitimate authority^{Footnote 11} offers us great insights into our common expectations about the grounds of authorities’ directives. Raz offers us a theory about when authorities could be considered legitimate and so he tries to make moral sense of how authorities are conceived in our Western traditions. For example, while he may not say so explicitly, his theory is aimed at explaining how authority can make moral sense in a normative universe where individuals are taken to be roughly equal, where autonomy is valued and where there are pre-political reasons for action. Obviously, a theory of authority based on the assumption that some individuals are clearly worth more than others and that authority can only be legitimized by supernatural forces would take a very different shape than the one Raz proposes. But because Raz’s theory is a normative one in which he makes decisions as to what would be the better moral interpretation of the connection between this or that concept^{Footnote 12}, he ends up promoting one particular view as to how the concepts cohere to the detriment of a possible actual aporia in people’s practices. It is precisely there that we will depart from Raz’s analysis.

In examining what it could mean within our culture for directives to be sufficiently grounded, it will not be necessary to engage with all of Raz’s specific views, critiques and replies. Our purpose is not to offer a critique of Raz’s theory per se, nor to present a better normative theory of the concept of authority, but it is rather to flesh out the architectonic and background assumptions of our practice of having binding authorities. This means that Raz’s views will be treated more like a foil to examine our own views than as a specific theory to be analysed and debated. Similarly to Hart, who was more interested in his famous Concept of Law in a broadly Austinian theory of law as an intellectual steppingstone than in Austin’s precise views on the concept of law, we will be more concerned here with discussing a Razian theory of authority than Raz’s views on authority. That being said, for the sake of expository simplicity, we may often refer to “Raz’s views”, but the reader will have to keep in mind that the real target here is a cluster of views associated with his approach, rather than Raz’s particular views. All that being said, let’s move on to what the Razian theory has to tell us about what an authority is and how the grounds that it may invoke contribute to its perceived legitimacy.

Raz is not mainly concerned with cases where a subject does not know what to do and decide to defer to the directives of an authority, but cares much more about the conditions under which subjects may consider it legitimate to obey an authority despite the fact that their own personal appraisal of the balance of reasons on the merits of a particular action goes against what is required.^{Footnote 13} This is the real challenge for the legitimacy of an authority claimant. According to the Razian theory, a legitimate authority would not require a complete surrender of one’s judgment—and thus would be respectful of autonomy—in the sense that it could only be legitimate if there were sufficient reasons to follow its directives, regardless of the balance of reasons applicable to a specific case.^{Footnote 14} What kind of reasons could be sufficient for a subject to accept the directives of an authority as legitimate? Joseph Raz’s theory of legitimate authority is built around three main theses: (1) the normal justification thesis; (2) the dependence thesis; and (3) the pre-emptive thesis.

The normal justification thesis claims that

The Razian theory therefore suggests that authorities are conceived as offering a service to help subjects and, on the whole, enhance their chances of fulfilling their responsibilities. A particular rule may not be right in the circumstances, but an authority would nonetheless be legitimate if the results, over time, were superior than if the subject had acted on the basis of his own judgments. Thus, the Razian theory is based on the idea that authorities are legitimate if they offer the right type of cumulative benefits (i.e., increasing subjects chances of fulfilling their duties).

How can a legitimate authority maintain that it can increase the subject’s chances of acting correctly? Raz proposes that this claim can be explained through a combination of the dependence thesis and the pre-emptive thesis. The dependence thesis claims that the authority’s directives are meant to be mainly based on the balance of underlying reasons that are applicable to the subject.^{Footnote 16} Rules may ultimately not be based on such a balance, but they are meant to be. The combination of the normal justification thesis and the dependence thesis is what Raz calls his “Service Conception” of authority. We may say that the Razian theory conceives authorities as useful heuristic devices.

The pre-emptive thesis provides that once the authority’s directive has been enacted, it is meant to displace the underlying balance of reasons and to exclude some or all other reasons for action.^{Footnote 17} The rule adopted by the authority must displace the reasons that the authority has balanced for the subject because otherwise, those reasons would be counted twice. By pre-empting considerations that apply to the subject, legitimate authorities incidentally offer an additional service to their subjects. According to Raz, not only do they increase the subjects’ chances of acting for the right reasons, they may also save them time and energy in helping them avoid the full deliberation process that lead the authorities to their conclusions. Indeed, whether or not authorities may require subjects not to deliberate about the right course of action by themselves, authorities nonetheless offer their subjects this incidental service that we will call the “resource-saving” feature of the Service Conception of authority.

The authority adopts rules that function both as first-order reasons for action and as second-order reasons for excluding some or all other reasons. A reason that fulfils both functions (orient action and exclude other reasons) is called a “protected reason”. The fact that a rule contains a protected reason helps to explain the difference mentioned earlier between, on the one hand, an advice or a request, and on the other, a rule given by an authority: the protected reason contained in the rule is meant to exclude some or all other reasons for action and to substitute them with the reason given by the authority. For example, protected reasons may exclude all reasons for action but for the one contained in the rule adopted by the authority, or, to take a different example, it may simply exclude a subset of reasons for action like religious or economic reasons. The latter case could be illustrated by a Legislature delegating discretionary powers to a civil servant to grant liquor licenses; that delegation may very well allow the civil servant to examine on his own a certain set of reasons for action related to the commerce of alcohol, but it may also exclude from consideration reasons that may only have to do with his own religious beliefs.^{Footnote 18} The authority may even exclude all reasons for action but for a subset to which the subject will be allowed to have recourse to deliberate about the right course of action. The set of rules adopted by the authority may also include rules that are meant to create exceptions to the exclusionary features of certain rules. For example, the rule against killing is qualified by other rules so as to allow the subject to act on reasons related to his own survival. The rule providing for the possibility of self-defense is an exception to the protected reason against killing. But the authority normally claims to be the only one with the capacity of adopting such rules having defeasibility power over other rules adopted by that same authority—unless that authority recognizes its subordinated status to a higher authority. When the authority does not accept a subordinated status to a higher authority, it will thus claim that it exclusively has the power to determine when and under what conditions its protected reasons may be defeasible.

In short, the authority claims that its set of rules is conclusive because it is made of protected reasons that only it can render defeasible on the conditions it sets. Authority, thus, claims not to be brute force. It also claims to respect autonomy to the extent that it does not necessarily require a person to abandon one’s judgment to it. Rather, it claims to be able to modify the set of reasons upon which one may act.

Raz qualifies his theses by first stating that when an authority claims to help a subject acting for the reasons applicable to him, it does not mean that a legitimate authority necessarily has to help promote the subject’s interests. For example, Raz writes that “[a] military commander (…) should put the defense of his country above the interests of his soldiers. But then soldiers are supposed to put their country above their personal interests, and only because of this should they obey their commander”^{Footnote 19}. However, Raz does not tell us why should a soldier put one’s country above his or her personal interests. But one thing is for sure, this cannot be explained by putting one’s own individual life as the ultimate value. We will come back to that point later.

For the Razian theory, the pre-emptive aspect of a directive does not entail that an authoritative directive cannot change the normative universe. For example, a rule adopted for the purpose of solving a particular collective action problem (i.e., increase the saliency of a conventional solution to a coordination problem or solve a prisoner’s dilemma type of situation^{Footnote 20}) would modify the normative landscape by the very fact of giving directives. It is possible, in such cases, that the precise content of such directives may not be determined at all by a balance of dependent reasons. In other words, prior to the enactment of the rule, agents would not have a reason to act one way or another, but once the rule is established, they have a reason to follow it because the very existence of the rule would be justified by underlying reasons. Indeed, the very existence of such a rule that aims at settling a collective action problem is justified by the reasons why we ought to solve such problems. Such a rule would be justified in the sense that it would help subjects fulfil their responsibilities by enhancing their capacities. Here, justification of the rule would not depend on its specific content, but rather on its capacity to settle the collective action problem. We can call these directives “content-independent” in the sense that their content is arbitrary from the point of view of the ends sought to be achieved. This is well-illustrated by the fact that the rule that requires us to drive on the right or the left side of the road, in itself, is not based on dependent reasons; it is the need to identify a common side for driving that is based on the dependent reasons associated with the need to optimize transport speed. Unless we can identify reasons why driving on the right or left side of the road is, in itself, better, what matters here is that one side is chosen and accepted by all agents.^{Footnote 21} The service offered here is not primarily one of an epistemic nature (i.e., informing subjects how the proper balance of reasons requires them to act). Instead, the service offered by the authority here amounts to a modification of the circumstances of deliberation and action to facilitate the achievement of certain goals by the subjects. An authority, according to Raz, may thus alter the normative universe by providing content-independent reasons for actions (e.g., a rule providing for which side of the road one must drive) that are, in turn, supported by dependent reasons (e.g., optimization of transport resources).

If authoritative directives may legitimately change the normative universe, irrespective of the content of the directive issued in certain circumstances, it is because such directives are considered useful heuristics. Authorities, according to the Razian views, appear mainly as mere stand-ins, substitutes for something else. According to this logic, the authority would only be legitimate because, on the whole, its directives are. Legitimacy, in the end, appears to be attached directly to the directives and only indirectly to the bearer of authority itself.

III. Difficulties with the inner workings of the Razian Service Conception of authority

There are difficulties with the inner workings of the Razian theory. First, the normal justification thesis may be in tension with the second-order exclusionary rules that preclude acting for any other reasons than the ones indicated in the first-order reasons. The fact that one obeys a specific authority because it enhances his chances of acting correctly on the whole (let’s call this a “third-order reason”) may put one in violation of the second-order rules that were meant to enunciate exclusive reasons for action. In other words, accepting a particular authority—whatever its directives—may be conditioned by reasons that are of a higher order than the reasons for action or deliberation given by that authority. And the very act of choosing to obey an authority on the basis of some higher reasons (e.g., effectiveness or any other reason) may violate the requirements of the exclusionary reasons adopted by the authority (when those reasons were meant to be peremptory).^{Footnote 22}

Moreover, to actually determine whether or not an authority is worth accepting, according to the Service Conception, the subject will ultimately have to engage in the very deliberations that the “resource-saving” feature of the Service Conception was meant to help the subject avoid.^{Footnote 23} But if the subject is convinced, after examining all the dependent reasons upon which the supposed authority constructs its directives, that it offers directives that are well-founded, then, the subject no longer needs the authority’s directives as one has already done all the work it was supposed to provide. The subject would then act for the reasons that one has oneself examined and not for reasons given by the entity that no longer serves the authority function. Thus, to verify whether or not an entity is a legitimate authority for oneself, one has to destroy the authority function of that entity. While the Service Conception of authority tries to justify authority on the basis of instrumental reasons (the service it offers), acceptance that an entity has a legitimate authority over us would be ultimately a matter of faith.

Another problem flows from Raz’s recognition that one may have reasons against accepting the claims of authority of a person or institution if another presents a better claim to be recognized as an authority.^{Footnote 24} We may indeed imagine that more than one authority claimant could be potentially legitimate in the Razian sense of being capable of providing the service of enhancing the subject’s capacity of acting for the right reasons. It is true that the issue may not arise as long as the competing authorities do not require inconsistent actions, but when they do, the subject may face a truly tragic situation if there are no sufficient reasons to decide which institution has a better claim to be the relevant authority.

The situation may be truly tragic because a relation of authority is not merely one of “choice” as the language of “acceptance” implies. Take, for example, the case of someone who is a citizen of two countries at war and who is required by each country to act in a way that the other country would consider an act of treason punishable by death.^{Footnote 25} Authority relations are not always symmetrical; the subject may not have the power to exit that relation at will. The fact that one may be said to be an authority—and not merely to have authority—underlies the potentially temporally extended character of the authority status. Therefore, individuals may find themselves in situations where they cannot obey all the legitimate authorities at once and that two or more legitimate authorities may impose consequences that will make it impossible to act for the right reasons in the future—this would be the case when disobedience (i.e. “treason”) may be sanctioned by death. Thus, while each authority may be legitimate when taken separately, taken together, they mutually destroy their legitimacy, as understood by the Service Conception. This would lead to the bizarre conclusion that the fact that someone is subject to two legitimate authorities may, for that very reason, find oneself under no legitimate authority at all!

More profoundly, the Razian Service Conception encounters a series of difficulties associated with what appears to be a limited conception of the reasons that individuals may find applicable to them. Indeed, the tragic situations that competing authorities may impose on their shared subjects may not simply be caused by the fact that such subjects may not know, in fact, which of the competing authorities, has a better claim to enhance the subjects chances of complying with reasons which apply to them.^{Footnote 26} The ultimate source of the tragedy may indeed lie in the fact that there may not be an answer to this enquiry. The capacity to determine which authority claimant has a better claim to enhance the subjects’ abilities to comply with reasons applicable to them presupposes a common matrix of reasons. But there may not be a unified set of reasons for action. Different authorities may rely on distinct sets of reasons that have their own logic. Indeed, the subject may see herself, for example, as having to choose between being guided by what she perceives as the competing demands of esthetics and ethics. If philosophical monism is not true, that is, if beauty, goodness and truth do not all point in the same direction, the determination of the proper authority to follow may be indeed daunting! Or, to take another classic example, the demands of one’s secular state and the demands of one’s religious obligations may lead to opposite actions. And there may not be a common yardstick to determine whether the secular or the religious authority has a better claim to lead to the proper action. The Service Conception assumes that the choice of the proper set of reasons has already been made or that there is a single set of such reasons.

An additional problem with the Service Conception is that it implicitly seems to assume that it is meant to help subjects deal with reasons that are otherwise applicable to them. However, authorities do not simply deal with reasons that pre-exist any form of sociability and authority relations. While it may be conceived that if one were alone in the universe, one would nonetheless have reasons for acting one way or another. At least, if not also value-reasons, one may have instrumental reasons to act in certain ways. For example, one may have reasons to walk in one direction rather than another if one is thirsty and needs to drink from a river. Now, the very fact of sociality modifies the matrix of possible actions. This, in turn, has an impact on the set of instrumental reasons applicable to the subject. Thus, sociality modifies the normative landscape by introducing new (social) facts. Indeed, those new facts force us to consider certain reasons for action that would not exist but for those facts.

This is where the Weberian insights introduce themselves in the conversation with the Razians. Human interactions are not merely material facts; they exist within particular imaginaries. Ideas about particular memberships, roles and relations are all parts of specific repertories of actions, beliefs and implicit assumptions.

Sociality thus also breeds value-reasons. Such reasons arise from the meaning given to the different forms of social interactions. We imagine ourselves having certain reasons for actions based on the particular ways we stand with others; they are thought of as reasons for actions that only exist because of the nature of the relations one entertains with others. In other words, there are reasons for actions that are dependent on the relation between the subject and another entity. They are what we would call “relational reasons”. For example, a woman may have reasons to sacrifice herself to save her own child while she may not have the same reasons to act the same way to save someone else. Indeed, there might be common reasons to sacrifice oneself to save one’s own child and a stranger, but we imagine also having reasons to save one’s child that are not applicable to the stranger. In other words, it is not simply that we have more intense feelings towards our own children, it is that we also imagine our duties to be different towards them because of the nature of the relation that brings us together. Obviously, there are many ways of imagining how one may be related to another and all those diverse forms of sociality may breed their own expectations.

Why is that a problem for the Service Conception of authority? After all, can a legitimate authority not also help subjects act for those relational reasons applicable to them? The problem is caused by the fact that authorities do not appear as mere factual objects that subjects take into consideration in their instrumental reasoning. They are not simply seen as detached parts of the furniture of the universe. Subjects are subjects precisely because there is a form of relation that binds them to the authority. In other words, an entity cannot be an authority without being involved in a relation to others. And authority relations bring with them a wide range of reasons for action that would not exist but for the existence of that relation. For example, an authority relation may create a particular reason for a subject to sacrifice oneself to protect that authority from destruction. Such a reason would not exist but for the existence of the authority relation. Or, to take a different example, it is only because an authority has decided that certain violations of its orders would result in incarceration in a prison that the issue of the proper conduct of a prisoner may be raised. In the absence of any authority structure, the right reasons for action of a prisoner, qua prisoner, would be unintelligible. This means that the very presence of an authority changes the possible dependent reasons upon which rules may be made. But the Service Conception can hardly explain how an authority can be justified by the help it gives in answering the very problem it has created. The Service Conception assumes that the authority takes the world as it finds it and issue directives without taking seriously the fact that the very existence of the authority (and the relations it entails) modify not only the factual landscape, but the normative landscape in which subjects find themselves.

Moreover, sociality does not only breed relational reasons for action based on roles attributed to individuals as discrete units of existence. Social facts also constitute collective entities that have relative autonomy from the individuals that compose them. Such collective entities also produce new reasons for action. To take an example that we have discussed earlier, imagine a military officer ordering part of his troops to attack a particular enemy position to create a diversion. Prior to the giving of the order, we could say that no individual soldier had reasons to attack that particular position and expose herself to the attendant risks. But once the order is given, each soldier to whom the order applies now has a reason to attack the said position. This example does not only show that authorities may solve collective action problems, but that they may also legitimately provide reasons for action that not only differ from, but may actually directly contradict the subject’s individual reasons for (in)action. Thus it would be misleading to suggest that the order reproduces a reason for action that the subject already had and, likewise, it would be incorrect to state that the order is illegitimate because it fails to reproduce dependent reasons.^{Footnote 27} Not all protected reasons offered by the legitimate authority are pre-emptive of the reasons that existed for the individual, as a mere individual, prior to the order. Not only that, the individual may have no reason, tactical or otherwise, to attack a specific position all by herself, and more generally, she may have no reason of her own to fight total strangers without seeing herself as a soldier directed to do so by the relevant authority. Then, one may ask “what sum of right reasons would require someone to sacrifice her life?” How can any legitimate authority based on the Service Conception ever require the sacrifice of its subjects and claim that this demand is based on pre-emptive reasons? If there are reasons that may justify the absolute sacrifice of the individual, those reasons are either part of what Weber called value-reasons—ungrounded posited values—or a form of instrumental reasons that do not take the individual’s survival as its ultimate goal. In other words, it makes no sense of individuals qua individuals to fight strangers in a war, but it may make sense as a member of a unit of existence that transcends one’s own individuality.

All this calls for a re-evaluation of the descriptive aspect of the Razian theory of authority. While it seems to capture certain important aspects of the way authorities are imagined in our world (legitimate authorities help us to do the right thing), it appears to hide other important aspects about the way we imagine authorities being able to bind us. This is even more obvious when we consider authorities within a collective entity like a “family”, a “church”, a “corporation”, etc. Individuals as members of a collective self appear to have primary reasons for action that would not exist but for the creation of the collective self. That collective self could not exist without the individuals who constitute it, and in turn, that collective self contributes to the constitution of the individual. This creates a set of reasons for actions related to the very existence of such collective selves and it may give us a hint as to why the soldier mentioned earlier may have a reason to put her country ahead of her own individual interests. This brings us back to Weber’s insight that legal-rational legitimacy depends primarily on membership of the subject to the group governed by the relevant authority.

IV. Challenges to the centrality of the Service Conception of Authority

Raz states that, on the main, a legitimate authority’s directives aim at reflecting the right balance of reasons applicable to the subjects. The problem with the descriptive aspect of Raz’s theory is that it takes the Service Conception as the only normal way to explain the legitimacy of authority. In the end, this theory claims that the normal way to justify authority relies on the quality of the service it offers. But Raz also recognizes that an authority may legitimately offer directives that are not dependent on reasons directly applicable to the subjects. He thus recognizes that authorities may adopt directives for their own reasons. Raz tries to reconcile this possibility with the dependence thesis by claiming that “[a]ll it requires is that it shall act primarily for dependent reasons.”^{Footnote 28} He suggests, for example, that considerations arising out of the functioning of a bureaucracy may not arise for the individual acting on his own.

This, in our view, shows an important limitation of the Razian theory of authority as a theory: a wide range of situations that Raz—and most other people sharing in our modern imaginary—consider to be cases of “legitimate authority” cannot be explained by the dependent thesis that grounds the Service Conception of authority. The fact that the authority “primarily” acts for dependent reasons may legitimise the directives it issues in such circumstances, but this cannot explain why the other directives may be justified. Indeed, an authority may “primarily” act for dependent reasons, but incidentally issue a set of rules that are not based on dependent reasons and that may have disastrous consequences for the subjects—the battlefield case discussed earlier may be a case in point. If such authoritative directives are to be legitimate, it seems that something else has to be said in their favour over and above the mere fact that they were adopted by an authority that primarily acts on dependent reasons. In other words, there is a (wide) range of authoritative directives that are considered legitimate even though they do not conform to the dependence thesis and are thus in need of a different justification.

One may argue that some authoritative directives that do not appear to conform to the dependency requirement may nonetheless be legitimate as they could be, in one way or another, necessarily connected to the primary directives given by a legitimate authority. As such, the primary directives could not be issued without the necessary incidental directives that may come with them. This may be the case, for example, when the incidental directive may not be logically detachable from the main directive. An order to alleviate the pain of a patient in palliative care with the help of a sedative may incidentally amount to an order to shorten the patient’s life. While the legitimate authority may primarily aim at issuing a directive to alleviate the pain (on the basis of dependent reasons), its order may incidentally—and predictably—amount to ordering the killing of the patient (an order that, arguably, would not be based on dependent reasons). This would be an application to authoritative directives of the “principle of double effect” famously introduced by Aquinas^{Footnote 29}. However, in all such cases where the two directives are necessarily connected in one way or another, the incidental directive’s justification could be traced back to the primary directive’s justification. In other words, the incidental directive could indirectly be relying on the dependence thesis and thus would not fit the hypothesis explored here, namely, the justification of directives that do not rely on reasons directly applicable to the subject alone. Again, such cases call for a different justificatory discourse than the main one proposed by Raz.

Because of his normative project of proposing a unified and consistent theory of our practices, Raz is forced to recognize that while other justifications for authorities are also frequently invoked, there are to be seen as “deviant” or “secondary”. What he calls the deviant forms of justifications would be the ones that are foreign to the functions of authorities; they are reasons that may completely replace the normal reasons for justifying authorities even if they are inconsistent with the proper functions of authorities. Obviously, this is a normative claim since the ascription of the proper functions of authorities depends on a conception of what it is good for, what is its raison d’être.

Raz draws a parallel here with someone seeking a piece of advice. He says that the normal reason for accepting an advice or for giving it is that it is likely to be sound advice. To compare obeying an authority to seeking an advice is to load the dice in favour of the Service Conception. That is why he can say that

This example highlights the fact that the Service Conception of authority has a hard time seeing love, loyalty, patriotism—three value-reasons—as proper justifications for the acceptance of an authority. While love, loyalty and patriotism may not appear at first to be directly related to the rational-legal character of an authority, nor to the grounds of a directive, they nonetheless seem to be able to play a central role in legitimizing political authorities. Love, loyalty and patriotism are commitments that provide reasons for obeying an authority; they legitimise the relation between the subject and the authority. Patriotism, for example, is not necessarily limited to a mere set of sentiments; it also provides reasons to follow the directives of one’s political authority rather than those of a different political authority, and this, irrespective of whether one’s authority is generally better at offering appropriate guidance than the political authority that governs a neighbouring community. Thus, the register of values, beliefs, practices, assumptions, etc., associated with patriotism may offer Americans living in the United States non-instrumental reasons to accept the authority of the Congress rather than the authority of the Canadian Parliament. The reason why the Service Conception seems to be incapable of accepting love, loyalty, patriotism, etc., as primary justifications for authorities is probably because it is too committed to the view that authorities may only be legitimated through instrumental-rationality: authorities must help the individual to achieve something in order to be legitimate.^{Footnote 31} The Service Conception, relying so heavily as it does on the dependence thesis misses the point of patriotism. Patriotic persons may be indifferent to the content of the norm, what matters to them is that respecting the directive given by the proper authority corresponds to who one is. It is an affirmation of one’s sense of belonging that is constitutive of that person’s self. Thus, respect for the appropriate authority is not seen merely as a means, it is understood as an end. From this perspective, what constitute a legitimate authority would not be explained by the quality of its directives, but rather by the nature of the relationships that binds it to its subjects.

Therefore, although respecting an authority’s directives for instrumental reasons—i.e., increasing one’s chance of acting correctly—is one possible explanation for why individuals consider authorities to be legitimate within our cultural tradition, it does not seem to amount to the (only) central case of legitimate respect for an authority. Indeed, it seems that taking an authority’s directives as the maxims of one’s action for reasons other than the very fact that the authority required the subject to do so may actually be considered a distortion of the authority relation. Maybe a legitimate authority will be right more often than the subject on what the subject should do, but the so-called authority may no longer find itself in the typical position of an authority if the subject accepts its powers for that reason. In other words, from the internal point of view of the subject, what distinguishes an authority imposing rules to any service-provider may be that the authoritative relationship is not conceived as one of exchange between relatively equal partners.

The more central case of a relation of authority seems to entail the presence of a dominant agent and a dominated subject—even if the individuals and entities only occupy such positions for the limited purposes for which the authority detains power over another. A practical authority is not an advisor precisely because its directives are not mere advices; the directives are meant to change the reasons applicable to the subjects. A practical authority can issue directives that render certain subjects’ actions no longer optional.

The Service Conception may have been elaborated to explain how this power imbalance between the rulers and the ruled could make sense in a society made of relatively equal individuals. But this may be where the theory goes wrong. Authorities may not only be a matter of relations between individuals qua individuals. Individuals are engaged in all sorts of relations from which their identities are constituted through the diverse roles they occupy and the self-perception they gain from such engagements. The identities of individuals consist in their own intrinsic particularities as well as the multiple social roles in which they are cast (mother, teacher, neighbour, Italian, etc.). There is nothing new in suggesting that the same individual possesses multiple identities. Some identities may be thought to be discrete and singular, while others may be only thought as flowing from a shared experience. Some identities presuppose that an individual is a whole in oneself, other presupposes that the individual is only a part, a member, of a collective self.

The existence of each form of self may be valued, despite the fact that the “will” of each self may sometimes conflict with another. For example, imagine a minister in a government where all major decisions are taken, after deliberation, either by consensus or by a majority vote of all the members of the Cabinet.^{Footnote 32} The members of that government see themselves as forming an identity distinct from its individual members^{Footnote 33}, a whole where all members—not mere individuals—are expected to show solidarity for the decisions taken together. This is what we call “Cabinet solidarity”. Presume that after the Cabinet deliberation, a minister is convinced that the Cabinet has clearly^{Footnote 34} made the wrong decision and has adopted the wrong rule as a consequence. As an individual, that person may or may not treat the Cabinet as an authority according to the normal justification thesis—she may not consider that following its directive will increase her chances, on the whole, of acting for the right reasons. However, as a member of the Cabinet, this way of thinking is antithetical to her role. Indeed, her role as a member of the Cabinet is to implement the collective decision whether or not she agrees with it. Refusing to do so may not only amount to a breach of her obligations, but also to a negation of her identity as a member of the group. From the member’s perspective, qua member, the question would not be whether or not implementing the collective decision is the right action to take, nor whether or not, on the whole, she will increase her chances of acting in the right way if she obeys the group; from the perspective of a member the question would rather be: am I truly a member or not? Should I remain a member of the Cabinet or not? We are not suggesting here that the minister in question ultimately should follow the rule adopted by the Cabinet. We are not making a normative claim. We are simply trying to highlight how the relevant reasons for action may depend on the perceived identity of the agent and the nature of the relations that determine such identity.^{Footnote 35} All this becomes probably even clearer in cases requiring a great sacrifice.

The exercise of power by an authority within a perceived collective self may then be conceived as not being wholly heteronomous, but rather as flowing from our own selves. The “I” is then replaced by the “We”, and as the saying goes: there is no “I” in “We”. That is why, as Weber noted in the quote cited above, the authority is understood as extending its demands of obedience when it requires those who are not members of the group in question, those who do not belong to the “We”, to submit to its authority.^{Footnote 36} Maybe authority must be justified to non-members through something like the so-called “normal justification thesis”. We are not going to a take position on this issue. But for members of the collective entity, acceptance of the legitimacy of domination is not merely a matter of instrumental rationality and it is not merely based on the idea that following the authority’s orders will enhance one’s chance of doing the right thing. Rather, accepting a political authority as legitimate may involve recognizing an existential connection to that authority. In other words, individuals who are members of collective entities may see themselves as having what we called “relational reasons” for action.

Members of a collective entity may privilege the “We” identity over the “I” depending on the circumstances. According to this conception, and whatever formulation is used, it is the fact that the authority’s directives are flowing from a version of the self that makes them justified. Theorists who have tried to ground the legitimacy of authority on consent, assent, will and so on, have systematically struggled to try to account for the fact that authorities issue directives that may go directly against the individuals’ consent, assent, will and so on. In a very crude way, we can say that this form of theorizing pits the legitimacy of one form of self (a collective self) against another (the individual self), and it tries to explain which one ought to have precedence over the other and under what conditions. As we are not engaged in such a normative enterprise, it will suffice here to highlight the facts that conflicting authorities may also be matched to conflicting selves and that authorities are not always imagined as being heterogeneous to the subjects.

To summarize, from the perspective of the members of the group subjected to the legal-rational form of legitimate domination, legitimacy flows mainly from the fact that those who exercise power within one’s group, do so according to legal rules purposefully enacted to achieve expedient or some internally accepted value-rational results. The exercise of authoritative power through legal means may be taken as a minimal condition of legitimacy. But it appears not to be a sufficient consideration. Two other considerations may compete for a conclusive justification of that authority: (1) the subject’s appraisal of the political superior’s (general or particular) judgement on the proper instrumental-rational or value-rationality course of action; and (2) the subject’s self-identification as a member of the group which recognizes the political superior as a legitimate authority for the group.

When legal formalities are respected and both competing considerations point in the same direction, we face an easy case of legitimacy. But these conditions obviously do not always obtain. How these elements combine in particular situations lead to an important variety of rapports between authority claimants and subjects.

V. Attitudes towards authorities’ conflicting justifications

The types of conflicting rapports that are relevant here are those between authority claimants and subjects that could be classified as either (1) cases where a subject’s conclusion is that the authority claimant makes wrong judgment calls on the directives it gives to the subjects, but nonetheless the authority claimant is seen as having endogamous powers over the group to which the subject is a member, and (2) cases associated with a negation of the endogenous character of the authority claimant.

The first type of situations is associated with the subject’s belief that the political superior clearly made the wrong decision based on the appropriate reasons for action. While the subject may recognize the legitimacy of the authority’s power to enact rules for one’s social group, one may also recognize that the rule in question was adopted and implemented according to the proper legal requirements, but one may nonetheless disobey the rule. In such cases, the subject recognizes the rule as a conclusive—yet, obviously defeasible—reason for action. In consequence, rules would not be considered determinative of the subject’s actions. The subject could then be said to recognize the legitimacy of the authority to make rules, but not the legitimacy of the particular rule in question. In such circumstances, the political superior would be seen as enjoying general legitimacy, but it would not be taken as an effective authority in those precise cases.

However, the subject may also decide, for value-reasons, to respect the directive that one sees as clearly establishing the wrong rule. There are a variety of such value-reasons. Thus, respect for the authority may be justified by fair-play: one should share in the burden if one shares in the benefits of collective actions^{Footnote 37}. To justify democratic authority, the argument goes:

This, however, seems often insufficient to justify the acceptance of certain negative consequences of following an authority’s orders.^{Footnote 39} For example, when the burden is much more important than the benefits received, fair-play may no longer require its acceptance.^{Footnote 40} This may be the case when one is asked to give one’s life. Also, when the supposed “benefits” bestowed upon individuals are considered more of the nature of a loss by them, they may not think that fair-play could adequately justify incurring of that additional loss. For example, fair-play would probably not be able to justify forcing one to pay for a public fanfare that plays on the street facing his bedroom window every Saturday and Sunday at 7 in the morning.

There are other value-reasons that may move subjects to respect the directives of an authority that they believe mistaken. Loyalty and the belief in the importance of being supportive of one’s political community’s decisions may be one of them. A good example of this may be Socrates’ acceptance to drink the hemlock. We could hardly argue that he accepted the death sentence as his fair contribution for all the benefits he received from the polis, but Socrates nonetheless believed that it was justified on other grounds.

One may then go one step further and accept the decision of the political authority as an expression of his solidarity; one may recognize that his individuality cannot be wholly detached from, because fused to, the political community. Acceptance of the mistaken directive would be justified as one’s participation in the collective act of self-assertion.^{Footnote 41}

In all such cases, despite being perceived as being wrong on the determination of proper conduct, the political superior would be taken as being both a legitimate and effective authority by the subject.

Also, when the subject disagrees with the political authority as to what the proper conduct ought to be, she may nonetheless conform to the requirements of the rule established by the political superior for instrumental-rational reasons aimed at survival, self-protection, etc. The decisions of the political superior would not be perceived as legitimate by the subject, but they would nonetheless be effective.

The second type of authoritative rapports concerns cases where the subject believes that a political superior indeed strikes the right balance of reasons, but does not recognize the political superior as having a legitimate power over the social group to which one belongs. In those cases, the subject may act consistently with the requirements of the rule but could not be said to be respecting the rule; one is not treating the rule in itself as the maxim of action. It just happens that the requirements of the rule coincide with the practical judgment of the subject. The political superior then is neither an effective nor a legitimate authority from the point of view of the subject. The authority makes no practical difference to the situation of the subject except, maybe, by offering a second opinion as to how all the rights reasons for action ought to be balanced.

A crucial variation on this type of rapport is concerned with cases where subjects do not care whether the directives given by the authority are substantially right; the subjects simply refuse to act according to the directives out of self-assertive reasons. It is entirely intelligible that someone may even believe that an authority claimant may have in fact correctly considered the right reasons for action, but that the very heteronomic character of the authority’s request changes its normative value. Indeed, a person may have reasons, in certain cases, to ensure that the imperative of self-assertion trumps other reasons for action. This is typically invoked to justify anti-imperialistic, anti-paternalistic actions. This may lead subjects to reason along those lines: “this political decision may not be the right one, but at least, it’s our decision.” The claim of autonomy is obviously reflected in the rejection of the heterogeneous authority: “this may well be the right decision to take, but we do not want your decisions to be imposed on us.” Acting out of self-assertive reasons is not understood as an application of instrumental-rational considerations but of value-rational impetuses: “We want to be masters of our own destiny, even if it means facing a bleaker future”.

These may all be political clichés, but for this very reason they reflect real and meaningful attitudes towards authority claimants, towards institutions. All this support the hypothesis that the way an authority claimant will be treated may have little to do with its intrinsic features, but much to do with the way it is imagined. In fact, the formal rules setting up the existence of an institution meant to exercise authority may not ultimately determine conclusively the way that the created authority will be imagined and treated by its subjects. It thus matters greatly whether an institution is to be understood as being justified because it serves certain functional purposes—and would thus be evaluated according to the “normal justification thesis” of the Service Conception—or because it embodies a collective existential commitment. Two institutions may share identical constitutive rules but yet could be seen as embodying different forms of authority depending on the meaning given to each by their subjects.

VI. “Existential Communities” and “Functional Regimes”

As it should be clear by now, authorities whose legitimacy is explained through the Service Conception are not subject to the same expectations as those that are understood as embodying existential commitments. The criteria used to evaluate one or the other will not be the same. This leads us to introduce two notions that may help understand how people’s competing expectations regarding authorities and their legitimacy are played out in our social lives. We propose to distinguish between “functional regimes” and “existential communities”.^{Footnote 42} Human groups are usually portrayed as being situated somewhere on a spectrum between those two pure types.

At one end of the spectrum, “functional regimes” are institutional mechanisms that are meant to solve functional problems like the need to coordinate behaviour or solve epistemic deficiencies. Those collective institutions are means to achieve certain ends. The internal logic of functional regimes is based on instrumental rationality and the quest for efficiency orients actions and beliefs. Those institutions fit very well with the Razian Service Conception of authority. A paradigmatic case here would be the International Organization for Standardization (ISO). If any sense of identity results in the participation in a functional regime, this is simply an incidental effect (except if the regime has been put in place specifically for the purpose of inducing a sense of identity to its participants).

At the other end of the spectrum, we find the “existential communities”, a community through which individual selfhood is constituted by a deep sense of “love”,^{Footnote 43} of loyalty to the other members of the group, of identity between the self and the group. In other words, an existential community is what makes it possible for the self to transcend the individual. Those communities, from the internal point of view of committed members, are ends and not means. Their internal logic is of an existential kind with a repertoire of actions and beliefs oriented towards what the agent considers as self-assertion. A paradigmatic case here would be the Catholic view of the Church as the “corpus mysticum” of Christ.^{Footnote 44}

These notions are similar to Ferdinand Tönnies’ distinction between Gesellschaft (society) and Gemeinschaft (community)^{Footnote 45}, but they differ in part because our notions are not taken to be objective descriptions of the form of “being-together”, but rather as the subjective interpretation given to such groups by their participants and outsiders.

Because the proper categorization of a group, or the nature of one of its institutions, is a matter of subjective interpretation, it may also be subject to controversies among participants. This may complicate public deliberation as the controversy about the proper nature of the group or institution in question may cause participants to talk past each other.

Indeed, with each qualification of the nature of a group or institution comes a very different logic, a separate repertory of actions, beliefs and evaluative criteria. As the imaginary of love is different from that of formal legality^{Footnote 46} or utility-maximisation, the internal logic of existential communities is radically different from the one of functional regimes. Saying that you love someone because your relationship results in a positive balance of costs and benefits devalues that relationship. It is not that instrumental rationality is an evil that corrupts love and friendship when it is the driving force behind a relationship (although it does exactly that to love and friendship). It is rather that instrumental rationality does not belong to the same imaginary realm; it is to be replaced by value-rationality. Instrumental rationality finds its place in other kinds of relationships where value-rationality appears at lost. For example, expecting your bank to increase your credit line simply because you are friendly would be to seriously misunderstand the type of relation sought by your financial institution.

Often, the type of relation involved is precisely the subject of debates or misunderstanding. Just think of how many works of literature are built around exactly those conflicting expectations between two protagonists—one who sees their relationship through the lenses of love, while the other sees it through an efficiency matrix. If such conflicts are possible, it is in large part because the material conditions that give rise to such relations under-determine the type of relation involved; the meaning of the relation and the imaginary associated to it are ascribed by the different agents. Oftentimes, there may be a certain consensus among members of a group about the nature of different actions, relations or institutions, but such consensus is always dependent on those subjective beliefs.

For example, the decision to speak and to promote the survival and flourishing of a minority language is not habitually based on the logic of utility-maximisation for the native speaker—after all, this is a rather “expensive” choice when considering the opportunity costs it entails. Rather, the promotion of one’s own minority language is usually based on existential considerations. However, majority language speakers do not always understand why linguistic minorities want to take special measures to protect their language as opposed to simply letting the “linguistic market” do its job. Language for many members of the linguistic minority is not seen as a mere instrument of communication; it is constitutive of the transcendent self. Thus, the issue, for the member of the linguistic minority, is often not one of costs/benefits of teaching children the language of the majority or that of the minority. It is not a question about which language is more worth possessing. When language is thought as the basis of an existential community, it no longer stands in a relation of “possession” to the individual—one does not have a language—, it rather stands in a relation of identity—one is a Francophone, a Catalan, a Fleming, etc. Thus, ditching a minority language is not an issue of replacing some possession by something more efficient, more practical; it means denying the sense of selfhood of those who identify with that language. While, from one perspective, it may look like a simple issue of efficiency, from another, it looks like a negation of one’s existence. Not being able to identify correctly the register used by our interlocutors is therefore quite understandably a frequent cause of political dispute.

Now, since institutions may be thought of as being situated along a spectrum between functional regimes and existential communities, this obviously means that our discourse may be peppered with incompatible assumptions flowing from the two distinct logics. Indeed, the recognition of the primary character of an institution will invite the use of an imaginary that may enter into conflict with the imaginary associated to the secondary character of that institution. For example, the rise of the democratic ideal came with expectations that the government will incarnate the will of the people; it will not only be for the people, but also of the people and by the people. Thus, the government in our democratic societies is primarily associated to an existential imaginary. It is imagined as the official voice and arms of the collective entity’s will. But this primary ascription does not occupy the entire horizon of expectations. Governments are also seen as tools, as instruments to achieve set goals. As a consequence, democratic governments are asked to be both the will of the People and to be efficient, even if such objectives may sometimes point toward contradictory results.^{Footnote 47} At any rate, understanding the relative proportions of expectations of such a mixture is key to make sense of the type of authorities that may be accepted in our democratic cultures. And this is because authorities are not merely justified by the services they offer, but often more so by the relation of identity that has been established between the authority claimant and the subject who sees oneself as belonging to same existential community.