Legal Theory, 9 (2003), 201–220. Printed in the United States of America
Published by Cambridge University Press 0361-6843/03 $12.00 + 00
LEGITIMACY, DEMOCRACY, AND
RAZIAN AUTHORITY*
Scott Hershovitz
Balliol College, Oxford
Joseph Raz has developed a highly detailed analysis of the nature of authority
that accounts for its structure, the role it plays in practical reasoning, and
the conditions under which it is legitimate. Raz’s analysis, which has been
evolving over the last few decades, is impressive in both scope and insight.
With it, Raz dissolves many of the paradoxes and conundrums that have
traditionally puzzled students of authority.
This essay explores the relationship between authority and democracy. It
investigates whether Raz’s account of legitimacy, specifically his normal justification
thesis, captures the conditions under which a democratic authority
is legitimate. I shall argue that it does not. Moreover, I shall suggest that the
inadequacy of the normal justification thesis as a test for the legitimacy of
democracies teaches us an important lesson about the legitimacy of political
authorities.
Before we examine the normal justification thesis, we need to understand
the whole of Raz’s account of authority. That in turn requires a quick look at
his work on reasons. In Section I, I explain some of Raz’s work on reasons.
In Sections II and III, I introduce Raz’s account of authority and legitimacy.
Readers well versed in Raz’s analysis may wish to skim ahead. Beginning in
Section IV, I argue that the normal justification thesis is inadequate as a
test of legitimacy for political authorities. Finally, in Section VIII, I consider
an unsuccessful attempt to redeem the normal justification thesis from the
objections presented.
I. REASONS
We engage in practical reasoning when we decide what to do. Practical
authorities impact practical reasoning by guiding behavior.1
In order to
understand how practical authorities guide behavior, we need to learn a
*I would like to thank John Gardner and Dale Smith for helpful comments on earlier drafts.
I would also like to extend special thanks to Joseph Raz for his generous assistance and patient
tutelage.
1. Practical authorities can be contrasted with theoretical authorities, which are authorities
about what to believe. A meteorologist is a theoretical authority about what we should believe
the weather will be like. A police officer directing traffic is a practical authority about how
201
202 SCOTT HERSHOVITZ
little bit about reasons and how they work. Raz’s analysis of reasons and
practical reasoning is nearly as comprehensive as his analysis of authority,
but we shall consider even less of it here.
Raz distinguishes between first-order and second-order reasons. First-order
reasons are reasons for action or reasons for refraining from action. That it is
raining is a reason to carry an umbrella. That it is snowing is a reason not to
wear a bathing suit. These are examples of first-order reasons. Second-order
reasons are reasons to act for reasons or reasons to refrain from acting for
reasons. This is a confusing notion. In order to understand it, we first need
to understand what it means to act for a reason. We will say that a person acts
for a reason if the reason is part of her motivation for acting. If I carry an umbrella
because it is raining, then I act in part for the reason that it is raining.
Further, we will follow Raz and say a person conforms with a reason if she does
what the reason requires, whereas a person complies with a reason if she acts
for it.2
Last, a person refrains from acting for a reason if it is not part of her
motivation, even if she conforms with it; refraining from acting for a reason
does not necessarily require refraining from the action the reason requires.3
What does it mean to have a second-order reason, a reason to act for or
not act for another reason? An illustration will help. Suppose Aaron’s grandmother
is in the hospital and that this provides Aaron reason to visit her.
Suppose further that Aaron goes to the hospital and visits his grandmother,
but only because he was hoping to run into Michelle, whom he has a crush
on. In this case, Aaron conforms to his reason to go the hospital to visit his
grandmother but he does not comply with it. Does Aaron have reason to
comply with his reason to visit his grandmother rather than just conform
with it? Raz suggests he does, and I agree.4
Because Aaron went to the hospital
to see Michelle and not his grandmother, his actions do not embody
appropriate respect for his grandmother. Aaron had a reason to comply with
his reason to visit his grandmother, that is, he had a second-order reason
to act for a reason: Only through visiting his grandmother for the sake of
visiting her could he show her proper respect.
We will be primarily concerned with the other type of second-order reason:
reasons not to act for reasons. Raz calls these exclusionary reasons.5
To
illustrate, imagine that you are the commissioner of Major League Baseball.
That your favorite team is the Milwaukee Brewers is a reason for you cheer
for that team and to support the team in any number of ways. However, your
drivers should move their cars. Some authorities are both theoretical and practical authorities
simultaneously; the Pope is a theoretical authority regarding what Catholics should believe on
theological questions and a practical authority about how they should lead their lives. Raz’s
analysis applies to practical authorities. Hence “authority” as used in the argument here refers
to practical authority unless otherwise noted.
2. Joseph Raz, PRACTICAL REASONS AND NORMS 178 (2nd ed., 1999).
3. Id. at 185.
4. Raz offers a similar example. See id. at 178–179. Raz argues that cases like this are special
and that in general reasons are reasons for conformity and not compliance. Id. at 179–182.
5. For Raz’s response to objections to his account of exclusionary reasons, see id. at 178–199.
Legitimacy, Democracy, and Razian Authority 203
role as commissioner gives you an exclusionary reason not to act for reasons
of favoritism in discharging your official duties. It might be okay to offer
support to the Brewers as commissioner, but not because of the fact that it
is your favorite team. Occupying the position of commissioner gives you an
exclusionary reason not to act for reasons that would otherwise apply to you.
One potential source of confusion is to think that an exclusionary reason
is a reason to act for no reason at all. This is not the case, as exclusionary
reasons are reasons to refrain from acting for a specific reason or set
of reasons. Exclusionary reasons are not reasons to refrain from acting for
any reason whatsoever. Another potential source of confusion is to think
that an exclusionary reason is a reason not to think about other reasons—to
exclude them from consideration. This is mistaken. It is acceptable to think
about an excluded reason, but not to act for it. As long as one does not
allow it to affect what one does, there is nothing wrong with thinking about
it.6
Last, we should note, nothing is wrong with conforming to an excluded
reason as long as one does not act for it. Suppose a mother tells her child, “I
do not trust Ms. Smith; do not mind her instructions.” She is not telling the
child never to do what Ms. Smith instructs; rather, she is telling her child
not to do what Ms. Smith instructs simply because Ms. Smith instructed it.
If there is independent ground to do what Ms. Smith instructs, then the
child may do so; her mother has given her a reason not to comply with Ms.
Smith’s orders, but conformity may still be warranted.
One last type of reason will be important for understanding Raz’s analysis
of authority: content-independent reasons.7
Content-independent reasons are
facts that achieve their status as reasons in virtue of features other than their
content. As any parent who has ever uttered the words, “because I said so,”
will attest, parental edicts are reasons for their children to act independent
of what the parent requires. This general truth is not without limits. If a
parent instructs her child to do something that will cause her great harm,
the instruction may be no reason for the child to do so. Nevertheless, within
a great range of activities, the fact that a parent has ordered an act is a reason
to perform it regardless of what the act is (perhaps not a conclusive reason—
there may be weightier countervailing reasons—but a reason nonetheless).
This is exhibited clearly by cases in which a parent can order diametrically
opposed actions (“go with your brother to the store” or “do not go with your
brother to the store”) and provide a reason for the child to act through both.
II. RAZIAN AUTHORITY
Let us put aside our analysis of reasons for a moment and return to authority.
Many describe authority as the ability to change what other people do. On
this view, if I can make you eat bean sprouts, I have the authority to make you
6. Id. at 184.
7. J. Raz, THE MORALITY OF FREEDOM 35 (1988).
204 SCOTT HERSHOVITZ
eat bean sprouts. This conception of authority is too broad. Perhaps I get
you to eat bean sprouts by blackmailing you, or bribing you, or just by telling
you about the positive effects of eating them. It would hardly seem correct
to say that I exercised authority over you. This view of authority confuses it
with power and influence.8
Nevertheless, authority is closely related to power and influence. Some
suggest that authority is the power to change what people ought to do or,
more accurately, to change the reasons that apply to them—to change their
normative situations. This suggestion comes closer to capturing our concept
of authority. When a mother orders her daughter to take out the trash, the
daughter has a reason to take out the trash that she did not have beforehand;
her normative situation has changed. This analysis of authority is still too
broad, however. It encompasses a variety of activities that we do not consider
instances of authority. For example, you can often change what someone
ought to do by requesting something of them. If I request that you help
me carry my groceries, I have given you a reason for action which you did
not have before: my request. The ability to issue a request, however, does
not indicate that one has authority over the addressee. Requests are often
issued precisely because one lacks authority to order the requested action.
Authority is something narrower still than the power to change people’s
normative situations.
Raz highlights the difference between ordering and requesting to draw
out the structure of authority. It will help to summarize his remarks briefly.9
When one makes a request, one gives the addressee a reason for action
that she did not have before. However, the individual making the request
recognizes that it is not an exclusionary reason to refrain from acting for
other reasons. Countervailing reasons may outweigh the request. If I request
that you help me carry my groceries, I expect you will consider my request
along with all the other reasons you have for action. I expect you to act upon
my request only if it tips the balance of reason in favor of doing so.10
Ordering is different. An order is intended to exclude other reasons as
a basis for action.11
Suppose Mary tells her son Patrick to wash the dishes.
Patrick, prior to the order, may have had reasons to wash the dishes and he
may have had reason to do many other things as well, perhaps mow the lawn
or play tennis with a friend. Does Mary intend Patrick to consider her order
8. See id. at 23–28 for criticism of such views.
9. For further discussion of the difference, see J. Raz, THE AUTHORITY OF LAW: ESSAYS ON LAW
AND MORALITY 15–16, 23 (1997).
10. Someone who requests may believe that their request is conclusive reason to perform the
requested action; they may think the request outweighs all other considerations. Nevertheless,
one who requests does not intend to exclude other reasons from consideration.
11. Some contend that the difference between ordering and requesting is not one of kind,
as Raz asserts, but rather is a difference of weight. Orders are simply weightier reasons than
requests, on this view, and though they may be intended to outweigh other considerations,
they are not intended to exclude them. Raz points out that valid requests do not necessarily
carry less weight than valid orders. There are weighty requests and orders that exclude little.
Raz, supra note 9, at 23.
Legitimacy, Democracy, and Razian Authority 205
as a reason alongside all the other reasons he has for acting and comply only
if it tips the balance as to what he should do? If she did, she would not have
issued an order. Mary’s order to Patrick is intended to exclude and replace
the reasons he had previously.
Mary’s order to Patrick is a first-order reason to wash the dishes and an
exclusionary reason not to act for some reasons he had previously. Raz calls
reasons that are both first-order reasons for action and exclusionary reasons
protected reasons. Authority over others, according to Raz, is the power
to change their protected reasons.12
To avoid confusion, we need to revise
this slightly. Legitimate authorities change protected reasons; merely de facto
authorities do not (though they purport to change or are taken to change
protected reasons). Legitimate orders are protected reasons to act as the
order requires and not to act for (some) other reasons for or against the required
action.13
Requests are not exercises of authority because they are firstorder
reasons only and not protected reasons.
Raz captures the exclusionary aspect of an order in his preemption thesis:
The fact that an authority requires performance of an action is a reason for
its performance which is not to be added to all other relevant reasons when
assessing what to do, but should exclude and take the place of some of them.14
The preemption thesis is, of course, true only of legitimate authorities,
because the directives of authorities lacking legitimacy may fail to create
reasons at all.15
Raz notes that this view of the nature of authority is captured
in the legal doctrine of res judicata.16
Once a binding judgment has been
issued by a court, the original cause of action cannot be relied upon for any
12. Id. at 19. We can also change protected reasons for ourselves through promising.
13. Raz, supra note 2, at 191. Raz suggests that orders exclude all reasons within the jurisdiction
of the authority issuing the order. Id. at 192. That is to say, reasons that the authority was
able to base its decision upon are excluded, whether they are reasons for or against the action
the order requires. Orders that are clearly mistaken, however, may fail to exclude reasons at
all. Raz, supra note 7, at 62.
Authorities do other things besides issue orders. They grant permissions and confer powers.
Both actions can be helpfully elucidated in terms of the power to change protected reasons.
When an authority grants permission for an action, it either cancels protected reasons against
performing the action or excludes some reasons against doing the action from a subject’s deliberation.
Raz calls the latter an exclusionary permission. Additionally, sometimes permissions
serve as an indication that no order against an action will be issued. Conferring power does
not change protected reasons but it allows the empowered to change protected reasons. See
J. Raz, Permissions and Supererogation, 12 AM. PHIL. Q. 161 (1975); Raz, supra note 9, at 18.
14. Raz, supra note 7, at 46.
15. See id. at 46. Raz presents the preemption thesis as a consequence of the normal justification
thesis and the dependence thesis, which will be discussed below. Raz argues that if the
dependence thesis were true of legitimate authority and the preemption thesis false, certain
reasons would be double counted. See also J. Raz, ETHICS IN THE PUBLIC DOMAIN: ESSAYS IN THE
MORALITY OF LAW AND POLITICS 215 (1995). I have changed the order of presentation for ease
of exposition. I do not think that it affects the substance of Raz’s views to do so. Raz’s analysis
is an interlocking whole whose parts mutually reinforce each other. See Raz, supra note 7, at 55.
16. Raz, supra note 7, at 42.
206 SCOTT HERSHOVITZ
purpose. The judicial decision is a new reason for action for the participants
that excludes and replaces the reasons they had previously.
III. RAZIAN LEGITIMACY
Now that we understand the structure of authority on Raz’s analysis, we can
ask what makes an authority legitimate. We can ask what circumstances justify
accepting someone else’s order as a protected reason for action. Before we
start, I should clarify a potential ambiguity. Raz understands legitimacy as
entailing an obligation to obey. Some theorists use “legitimate” to denote
authorities that are in some sense justified even though subjects are not
obligated to obey their directives.17
I shall follow Raz’s usage.
According to Raz, an authority may be legitimated or justified in many
ways;18
however, one criteria captures the normal and most important way
of justifying an authority. Raz calls it the normal justification thesis :
The normal way to establish that a person has authority over another person
involves showing that the alleged subject is likely better to comply with the
reasons which apply to him (other than the alleged authoritative directive)
if he accepts the directives of the alleged authority as authoritatively binding
and tries to follow them, rather than by trying to follow the reasons which
apply to him directly.19
That is a mouthful. The idea is simply that if we will do better by following
an authority than by working out what to do on our own, the authority is
legitimate for us.
Illustrations will help, but before we start, we should make a minor revision
in Raz’s canonical expression of the normal justification thesis. The normal
justification thesis, as Raz presents it, is fulfilled whenever a subject would
do better complying with the reasons she has by following the order than
she would otherwise. But if one follows an order, the most one can do is
conform with the reasons one has apart from the order. This is because one
is acting on the order and not on the other reasons one has. Remember,
in Raz’s framework, complying with a reason requires acting for it. This
problem is easily remedied: We should replace “comply” with “conform”
in the formulation of the normal justification thesis. With this potential
17. Raz sees the question of the legitimacy of political authority as part of a “wider issue
of the justification of government.” Raz, supra note 15, at 356. Justifying a government is the
wider issue, in Raz’s view, because governments do things that are not properly construed as
exercises of authority; for example, governments own property, participate in markets, and
employ people. Raz believes governmental action stands in need of justification regardless
of whether it involves an exercise of authority. The legitimacy of authority presents special
problems, however, because it involves an obligation to obey. Id.
18. See Raz, supra note 7, at 53–56, 75, for discussions of other ways an authority might be
justified.
19. Id. at 53.
Legitimacy, Democracy, and Razian Authority 207
source of confusion cleared, we can examine the operation of the normal
justification thesis.
Suppose Andy is ill. Andy has reason to take the medication that will best
aid his recovery, if he has reason to take medication at all. Since Andy knows
nothing about medicine, he will be more likely to take the best medication
if he consults Eboo, a doctor. Eboo is a legitimate authority for Andy regarding
what medicine he should take, because Andy will be more likely
to conform with the reasons that antecedently apply to him (reasons about
what medication to take) by following Eboo’s orders than he would were he
to choose which medicine to take on his own.
Contrast this with the case of Dave. Dave is also ill, but he is a doctor.
Dave’s mother has no medical training but likes to dispense folk remedies.
Dave has reason to take the medicine that will best aid his recovery, but
given his medical knowledge, it is likely that Dave will do better deciding
which medicine to take on his own than he would following his mother’s
orders. Dave’s mother is not a legitimate authority for Dave regarding what
medicine he should take.
Suppose, however, that Eboo is telling Dave what medicine to take, instead
of his mother. Remember, Eboo is a doctor too. Whether Eboo is a legitimate
authority for Dave depends in part on who has greater expertise regarding
the illness in question. But it does not depend solely on that. Perhaps Dave
knows slightly more than Eboo, but Eboo will be better able to make a
decision free from bias due to Dave’s emotional involvement in his own
illness. If this is so, Eboo will be a legitimate authority for Dave regarding
what medicine he should take, despite Dave’s superior expertise.
Two things are apparent. First, the justification of authority is personal, in
the sense that an alleged authority may be justified for one person and not
another. Eboo is an authority for Andy but perhaps not an authority for Dave.
Second, a common way for an authority to fulfill the normal justification
thesis is through superior expertise to its subjects, though, as we have seen,
this is not the only way the normal justification thesis can be fulfilled.
Another way of fulfilling the normal justification thesis which has attracted
a lot of attention from legal theorists is for an authority to be in a position
to provide and enforce a solution to a coordination problem. Following
Raz, we can define a coordination problem as “a situation in which the vast
majority have sufficient reason to prefer to take that action which is (likely
to be) taken by the vast majority.”20
A classic example of a coordination
problem is which side of the road people drive on. Considerations of safety
and efficiency demand that people drive on the same side of the road, but
it does not matter much if it is the left or the right.21
An entity that has the
ability to enforce a solution may be a legitimate authority regarding which
20. Id. at 49.
21. I have been told but have not confirmed that it matters a little. Supposedly studies have
shown that driving on one side is better suited to our reactions and reflexes, but which side the
studies have shown seems to correlate with the nationality of the person reporting the result.
208 SCOTT HERSHOVITZ
side of the road people drive on, despite its lack of expertise or any claim
to better judgment. It is legitimate because following its directives allows
people to conform with reasons they have (driving on the same side of the
road as everybody else) better than they could on their own.
One last thesis is necessary to fill out Raz’s account of authority: the dependence
thesis. Raz writes:
All authoritative directives should be based on reasons which already independently
apply to the subjects of the directives and are relevant to their action
in the circumstances covered by the directive.22
Raz calls reasons that already apply to subjects dependent reasons. The name
draws an appropriate contrast with content-independent reasons, because
dependent reasons are often, though not exclusively, reasons for a person
in virtue of their content as opposed to their pedigree.23
We do not require a detailed examination of the dependence thesis here.
It is easy to see how it is intertwined with the normal justification thesis. An
authority’s role, on Raz’s analysis, is to assist a subject in better conforming
with the reasons that apply to her. This necessitates basing decisions on
dependent reasons.24
The normal justification thesis, the preemption thesis, and the dependence
thesis together form what Raz terms the service conception of authority.
On the service conception, legitimate authorities serve their subjects by mediating
between them and the reasons that apply to them. In the remainder
of this essay, I aim to show that the service conception is inadequate for
understanding the legitimacy of political authorities.
IV. RETHINKING LEGITIMACY
Remember that Raz holds that the normal justification thesis is only one
way in which an authority can be justified, though he thinks it the normal
and most important way. In the next few sections, I argue that Raz’s service
conception of authority is inadequate for understanding important features
of law in democratic societies, because the normal justification thesis does
not represent either the normal or most important way of justifying political
authorities in such communities.
Both Roberto Unger and Jeremy Waldron have suggested that the “dirty
little secret” of modern jurisprudence is that it is uncomfortable with
22. Raz, supra note 7, at 47.
23. Not exclusively, because the directives of a different authority may be dependent reasons
that an authority can base its decisions upon.
24. The dependence thesis is often taken to lead to a corollary, the no-difference thesis, according
to which “[t]he exercise of authority should make no difference to what its subjects ought
to do.” Raz, supra note 7, at 48. The inference from the dependence thesis to the no-difference
thesis is invalid. For discussion, see id. at 30–31, 48–49.
Legitimacy, Democracy, and Razian Authority 209
democracy.25
I am not sure that “uncomfortable” is the proper adjective,
but less attention has been paid to what I will argue is the special character
of law in democracies than is warranted. Indeed, contemporary legal theorists
devote little attention to the legislative process; they focus mainly on
the activities of judges. Waldron’s Law and Disagreement and its companion
volume The Dignity of Legislation26
are, in part, conceived as an attempt to
redress this defect. The lack of attention to legislation and to the distinctive
character of law in democracies is partially explained by the desire many
theorists have to engage in general jurisprudence, to say what is true of law
wherever and whenever it is found.27
The project of general jurisprudence
is certainly worthwhile. However, in searching for commonalities among all
legal systems, we may overlook features of law in certain political communities
that contribute more to the shape of what law is for those communities
than the features they share in common with law across the board. I believe
along with Waldron that just this has happened with democracies.
An important part of what we have neglected about law in democracies is
that it is the mechanism people use to make collective decisions about their
governance and that these decisions are (ideally) the result of participatory
procedures. Below I shall discuss the implications this has for our understanding
of law and its legitimacy, but I want to begin by placing the issues
in context.
As near as I can tell, there are a few reasons the special character of law
in democracies is often overlooked (aside from the desire to do general jurisprudence).
Modern jurisprudence developed prior to the rise of statutes
as the most significant source of law, and it is the modern statute produced
in an elected legislature (along with law enacted through referendum) that,
among the various types of laws, lays the greatest claim to participatory development.
Further, much of modern jurisprudence has been a reaction
against John Austin’s command theory of law. In one important respect,
it seems that the discipline is still held in its sway. On Austin’s view, laws
are commands backed up by threats. Of course, jurisprudence has moved
far beyond Austin.28
Raz’s theory of law’s authority is light-years ahead of
Austin’s in sophistication. Nevertheless, Raz’s analysis, our most advanced,
still conceives of law as telling us what we may and may not do. This affects
25. Roberto Mangabeira Unger, WHAT SHOULD LEGAL ANALYSIS BECOME? 72–73, 115 (1996);
Jeremy Waldron, LAW AND DISAGREEMENT 8–9 (2001).
26. J. Waldron, THE DIGNITY OF LEGISLATION (1999).
27. As Waldron notes, Raz has suggested that a legal system need not have a legislature.
Waldron, supra note 25, at 34; Raz, supra note 2, at 139–141. Waldron discusses general jurisprudence
and what he considers its misguided aim for institutional neutrality at Waldron,
supra note 25, at 45–48.
28. Hart assailed Austin’s command theory because it neglects the normativity of law. He
also showed that Austin’s theory lacks the nuance necessary to explain law. Law does more than
issue commands; it also grants permission, creates powers, and regulates its own development
and application. H.L.A. Hart, THE CONCEPT OF LAW (2d ed. 1994).
210 SCOTT HERSHOVITZ
how Raz understands legitimacy.29
We can see this clearly if we look at the
questions Raz asks at the outset of his inquiry into legitimate authority: “What
can justify holding some people duty-bound to obey others? Under what conditions
can some have a right to rule others?”30
Let me be clear: These are
important questions that any account of legitimate authority must answer.
Yet in a way they seem vaguely unsatisfying as questions about the legitimacy
of democratic authorities because they presuppose a division between
rulers and subjects—a division between the duty-bound and the binders—
that many deny can be drawn in a suitable democracy.
An ideal (perhaps not the only one) of democracy views government
action as collective action, as the action of the community taken through
various procedures and decision functions that operate on the preferences
or views of its members. The idea behind this view of democracy is that we
are binding ourselves through acts of legislation. Contrast this with the view
expressed by Raz: “Governments decide what is best for their subjects and
present them with the results as binding conclusions that they are bound to
follow.”31
This is no doubt true generally, and yet an adherent of the view
I am expounding will deny that it is true of well-constructed democracies.
Rather, in democracies, people decide together what they shall be bound
to do. Law in a democracy does not merely tell us what we may and may not
do—though it does that; law is how we decide what we may and may not do.
Does the idea that we are binding ourselves make sense or is it just happy
rhetoric obscuring the brute fact that in all political systems one group of
people rules over another? We might think so, because we believe all actual
democracies fall short of the ideal. I want to set this aside, however,
and focus on the ideal itself, asking whether there are theoretical reasons
to believe that a group cannot bind its members through collective action.
Taken out of the political context, it is clear that a group can collectively
make decisions which bind its members. A football team can decide which
tournament it wants to enter, a husband and wife can decide together where
they shall settle, a corporation can decide to introduce a new product, and
so on. All of these collective decisions, in the right circumstances and subject
to limitations, can bind the individual members that compose the group.
Though there are interesting philosophical issues regarding the relationship
between collective action and individual action and between collective
decision-making and individual obligation, no one doubts that we can make
decisions in concert with others which bind us. Why should a political community
be any different?
I raise all this only to suggest that Raz’s questions are misleading when
applied to democracies. Democratic governments, as I am conceiving them,
may have different criteria of legitimacy from nondemocratic governments
29. Cf. Waldron, supra note 25, at 100.
30. Raz, supra note 15, at 356.
31. Id. at 359.
Legitimacy, Democracy, and Razian Authority 211
precisely because of the fact that one group of people is not ruling over
another, because of the fact that people are binding themselves. Clearly, I
can bind myself in ways others cannot bind me, through promising and such.
Though I would not suggest that lawmaking involves any sort of promising
on the part of an individual (the analogy may not be so bad when it comes
to the polity as a whole), a law that I enact in concert with others may have
greater claim on me, or a different sort of claim, than one promulgated by
a government I play no role in.
I do not want to minimize the difficulty in specifying the conditions of
legitimacy for democracies. Here are some of the important questions we
must confront.
1. What types of decisions are properly made collectively and what types of decisions
are properly reserved to individuals?
2. What counts as a collective decision?
3. What procedures should be used to reach a collective decision?
4. How does a person become a part of, or cease to be a part of, the community
which is binding itself ?
5. How far does the obligation of an individual extend if she disagrees with the
collective decision reached?
Answers to all of these questions, and to many more, would be necessary
parts of a theory of the legitimacy of democratic authorities. The last is particularly
important; we might call it the problem of losers, and it exists because
consensus is not a realistic method of making decisions in modern political
communities. Under any decision procedure that is serviceable, some
will lose out. Their favored outcome will not be selected. There are serious
questions as to (1) whether the losers can be said to have bound themselves;
and (2) how far their obligation extends. The answers would, of course,
depend on the nature of the group and its decision-making process, and
even a general attempt to tackle the questions is too large a project to take
on here. For now, I shall simply note that the aspiration of democracy on
the ideal I am presenting is for people to make decisions together through
fair procedures, and we need to consider the normative implications of this
aspiration for law’s authority. Moreover, how close actual democracies on
their grand modern scale do come or can come to achieving this aspiration
is a further question, one that will hover in the background of the considerations
here but will certainly not get the treatment it deserves.32
My aim in
32. An important difference between democratic communities of differing size is the character
of participation opportunities. Les Green somewhat skeptically describes democratic
citizens as “law-takers just as individual consumers in a competitive market are price-takers.”
Les Green, Law, Legitimacy and Consent, 62 S. CAL. L. REV. 795, 802 (1989). I have misgivings
about this description, but as the size of a community grows, it probably becomes more apt.
However, even in gigantic democracies, more robust (and effective) forms of participation
than law-taking at the ballot box are frequently available (e.g., protest, petition, and lobbying).
The legitimacy of a democratic political authority will depend in part on the character of the
participation opportunities it affords.
212 SCOTT HERSHOVITZ
the following sections is simply to raise doubts that the normal justification
thesis is an adequate test of legitimacy for democratic communities.
V. VARIETIES OF LEGITIMACY
The normal justification thesis is a substantive theory of legitimacy. A theory
of legitimacy is a substantive theory if it confers legitimacy on the basis of
the directives an authority issues. The normal justification thesis assesses
whether an individual better conforms with the reasons she antecedently
has by following the directives the authority issues than she would by following
her own reasoning on the matter. No assessment is made of how the
directives are reached; the thesis judges only the content of the directives
themselves.33
We might also have procedural theories of legitimacy. A procedural
theory of legitimacy assesses how the directives an authority issues are
arrived at. Such a theory might judge, for example, whether the process was
fair, deliberative, or public.34
Last, a theory of legitimacy could be a hybrid
theory, partially judging the substance of a directive and partially judging
the procedure that produced it.
Procedural theories of legitimacy play an important role in our thinking
about the legitimacy of democratic governments. We critique governments
based on the degree to which people have access to the political process and
equal efficacy within it. We often think, for example, that decision-making
behind closed doors is inappropriate even if the decisions reached are good
ones. Notice that we do not oppose decisions made behind closed doors
only because we think that open decision-making processes produce better
decisions; on some occasions we recognize that a better decision might be
reached without the process proceeding publicly. We support open government
partially because we think it disrespectful of the dignity of citizens
as agents in their own self-government for the government to make important
decisions affecting their lives without allowing them to witness and
hence contribute to the debate shaping the outcome, even if only indirectly
through public comment and protest.35
We do not always have reason to care how an authority reaches its decisions.
It makes no difference to me whether my doctor reads all the relevant
literature herself or instead consults an expert, as long as she is making the
33. Of course, the method of reaching the directives may be relevant in helping us to determine
whether the directive fulfills the normal justification thesis. If we believe that the
authority’s method of deciding a question helps it to reduce bias that would afflict our own
reasoning, we may believe that its directive will help us to better conform with the reasons we
have for action.
34. I use “procedural” broadly. Amy Gutmann and Dennis Thompson distinguish between
procedural, constitutional, and deliberative theories of democracy. Amy Gutmann and Dennis
Thompson, DEMOCRACY AND DISAGREEMENT 27–49 (1996). As I use the term, whether a political
process is deliberative in Gutmann and Thompson’s sense is a question about procedure.
35. For an illuminating discussion of the value of open government and countervailing
concerns, see id. at 95–127.
Legitimacy, Democracy, and Razian Authority 213
best decisions possible regarding my care. Likewise, I do not care whether
my guitar instructor has studied methods of teaching music extensively and
suggests exercises in accord with an academic theory or whether she has
developed her expertise in teaching through years of experience, as long
as her instructions help me to progress as efficiently as possible toward my
goals. Judging the directives of a doctor or a music instructor solely on the
basis of their substance makes a lot of sense. This is because we regard the
doctor and the music instructor as experts; how they became experts is of
no consequence as long as they truly are experts.
Political authority is different. We do care how governments reach the
decisions they make. A full accounting of why we care is far too large a task to
tackle here, though I shall make some exploratory remarks about the reasons
for being concerned with procedure, specifically reasons for preferring the
procedures of government to be democratic ones. I shall not focus on any
specific conception of democracy, much less any specific democratic procedures.
Controversy of course surrounds most questions about the nature of
democracy and what it demands. Here I conceive of democracy simply as
a class of political systems that are participatory, where each citizen has the
ability to participate (preferably, at some foundational stage, equally) in
the creation of government and policy. All democracies embody in some
sense the ideal expressed by the word’s Greek ancestor, demokratia, meaning
“rule by the people.” Democracies stand in opposition to political systems
that display rule by one, a few, and so on. We probably ought to think of political
systems as being democratic in degrees, that is, more or less democratic,
but even this modest point extends deeper into a discussion of democracy
than appropriate for our task here. I want merely to call attention to
reasons for preferring democratic decision-procedures, whatever form we
ultimately decide best, whatever other values might mitigate against them.
VI. THE VALUES OF DEMOCRACY
We might believe that democratic procedures will produce better decisions
than other forms of political decision-making. On this view, the fact that
a decision is reached through democratic procedures marks it as having a
claim to expertise. An element of this line of thought partially underlies
some theorists’ attraction to democracy; they believe that if we throw our
ideas into the public arena, vigorous debate will allow us to reach better
decisions than we would otherwise. The value of democratic procedures, on
this reasoning, is instrumental; democracy serves the goal of improving the
substance of our decisions.
There are further reasons to be concerned with procedure, reasons that
do not view procedure merely as a proxy for judging substance. One of
these is the value we place on autonomy. People are better off, we believe, if
they are able to organize their own affairs and develop and pursue their own
214 SCOTT HERSHOVITZ
conceptions of the good. Since governments play a pervasive role in organizing
our lives, denying people the opportunity to participate in the process
of government closes off opportunities to organize one’s life autonomously.
Democratic procedures afford people opportunities to participate in lifedefining
decisions.36
We also value democratic procedures because we believe they show proper
respect for the dignity of people as rational agents. It is harder to pick out
experts in moral and political matters than it is in medical or musical matters.
Indeed, some argue that there are no moral experts. Since we believe that
none of us are (generally) better situated than others to discover moral
truth, when there are moral decisions to be made that affect us all, nobody
(generally) has a claim to be better able to make those decisions than anyone
else. We are equals, and respecting this requires that we participate as equals
in the decision-making process that orders much of our moral universe.
Not only does democracy allows us to participate as equals, but it reinforces
that status. As Amy Gutmann and Dennis Thompson put it: “When citizens
deliberate in democratic politics, they express and respect their status as
political equals.”37
Justice may also demand democratic procedures. Robert Dahl suggests
that democratic procedures instantiate a form of procedural justice.38
Moreover,
he suggests that distributive justice requires a fair distribution of crucial
resources, including not only wealth, income, and education, but power as
well.39
Power, Dahl argues, is determined in part by the structure of authority
in a community. Hence distributive justice, in his view, demands that the
exercise of authority be controlled through procedures that allocate power
fairly.
In addition to the intrinsic values that democracy instantiates, there are
further consequentialist reasons to support democratic decision-making
procedures aside from any positive impact they may have on the substance
of the decisions reached. People who can access their political system are
empowered to take control of their lives. Those who feel that they have no
voice in their governance, no efficacy in the political process, may experience
alienation and resignation.40
Often it is little comfort to be assured
36. I do not intend to invoke what Raz calls the “least persuasive argument for democracy,”
which is, according to him, “one which regards it as the nearest we can get to selfdetermination.”
J. Raz, Disagreement in Politics, 43 AM. J. JURISPRUDENCE 25, 45 (1998). The
value of people participating in decisions that have an important impact on their lives can be
grounded without any suggestion that self-determination, whatever that might be, is required
or desirable.
37. Gutmann and Thompson, supra note 34, at 18.
38. Robert Dahl, DEMOCRACY AND ITS CRITICS 164 (1989).
39. Id.
40. I have in mind here phenomena like adaptive preference formation, in which people
define their goals by what they believe is reasonably available to them, as well as much of the
research on the importance of what sociologists and historians call “social capital.” On adaptive
preference formation, see Jon Elster, SOUR GRAPES: STUDIES IN THE SUBVERSION OF RATIONALITY
(1983).
Legitimacy, Democracy, and Razian Authority 215
that important decisions that will affect your life but are being made without
your input will be good ones. For related reasons, John Rawls finds additional
value in democracy. He believes that democracy “enhances the selfesteem
and the sense of political competence of the average citizen.”41
Rawls
speculates that being expected to vote and therefore to have political opinions
encourages “the development of . . . intellectual and moral faculties.”42
The point of all of this is to highlight the fact that political procedures
can shape the lives of citizens just as substantive policy does. Democracy’s
positive contributions to character and psyche are further sources of its
value.
Another view about the value of democracy deserves discussion, but it is
one that I shall disclaim. A long tradition of thought holds that political
legitimacy rests on the consent of the governed. There is no important conceptual
link between consent and democracy; we can consent to monarchy
just as easily. However, as Raz notes: “In common political discourse, government
by consent often means no more than democratic government.”43
I imagine that democracy and consent are linked in people’s minds because
the participatory nature of democracy makes it appear as though people
consent to the political process. In fact, few of us act in a way that can reasonably
be interpreted as consent to the near-limitless authority over our
lives that government purports to exercise.44
Even if we did, it is doubtful
that consent contributes significantly to political legitimacy. We need not
be detained by the details of the argument, but the idea is simple. Consent
to be governed by a wicked authority is akin to promising to do something
morally atrocious; neither generates the reasons it might under more felicitous
conditions. Indeed, Raz argues persuasively that consent to a political
authority is effective only if the authority satisfies an independent test of
legitimacy.45
Raz thinks the independent test is specified (for the most part)
by the normal justification thesis. I think the normal justification thesis is
inadequate as a test of political legitimacy for the reasons we are in the midst
of exploring. However, I agree with Raz that consent is generally insufficient
to render an illegitimate political authority legitimate. Hence I do not believe
that the value of democratic procedures derives from any connection
between consent and legitimacy.
The foregoing paragraphs encompass an embarrassingly short account of
the value of democracy. Articulating a full defense of democracy would be
41. John Rawls, A THEORY OF JUSTICE 234 (1971).
42. Id.
43. Raz, supra note 15, at 366.
44. See Raz, supra note 9, at 239; Green, supra note 32, at 808.
45. Raz, supra note 15, at 355–369. Raz does allow that consent can have some impact on
legitimacy. He believes that it can strengthen obligations to obey and that it can express a
citizen’s trust for his government. Id. Raz also thinks that there are limited cases in which
consent can establish authority—cases where it is an “optional good” for people to decide
the matters in question themselves. The authority of the modern state does not fall into this
category. See J. Raz, Facing Up: A Reply, 62 S. CAL. L. REV. 1153, 1183 (1989).
216 SCOTT HERSHOVITZ
a significant undertaking, but fortunately, the world does not lack for good
ones.46
Though short, the discussion has reminded us, I hope, of some of
the important reasons we have for being concerned with the procedures that
political authorities use to make decisions as well as with the substance of
those decisions. We can now turn more explicitly to questions of legitimacy.
VII. LEGITIMACY AND DEMOCRACY
Our discourse about the legitimacy of governments indicates that we believe
a government can fail to be legitimate on procedural grounds. If a
government’s electoral system favors some interests in society, or appears
corruptly financed, or causes portions of the population to be marginalized
and voiceless, we are quick to judge it illegitimate, or at least less legitimate
than it might be otherwise. Where these deficiencies are present, it counts
for little that a government may produce substantively good decisions, decisions
that the normal justification thesis would hold authoritative. Nondemocratic
governments that rule well are tolerated only as second-best
options when we see no opportunity for democratic reform. This shows us
one way in which the normal justification thesis is incomplete as a theory
of legitimacy for political authorities: Governments that fulfill it may fail to
be legitimate on procedural grounds. They fail to be legitimate because,
even though they produce good law, they do it without showing proper respect
toward their citizens, without allowing them appropriate opportunity
to participate in the structuring of their own lives, without regard for a just
allocation of political power, and perhaps in ways that breed resentment and
alienation.
We ought to ask whether the normal justification thesis might be incomplete
in the other direction as well. Can governments that fail its test still be
legitimate on procedural grounds? In Law and Disagreement, Jeremy Waldron
argues that democratically enacted laws deserve respect even if they fail the
test of the normal justification thesis, and it seems that his argument accords
legitimacy (of sorts—it is unclear how the demand for respect differs from
an obligation to obey) solely on procedural grounds.47
Waldron, like many
before him, bases his argument in part on the normative importance of the
fact that people disagree about the principles that should be used to order
society.48
46. See, e.g., Dahl, supra note 38; Gutmann and Thompson, supra note 34.
47. Waldron, supra note 25, at 85–118.
48. Raz has responded in a number of forums to Waldron and other theorists (including,
most prominently, John Rawls) who accord normative importance to disagreement. He argues
that disagreement can ground neither the neutrality of the sort that Rawls seeks nor an obligation
to submit to fair decision procedures Waldron might be taken to argue for. In part, Raz
objects because, he claims, arguments from disagreement undermine themselves. People will
inevitably disagree over the implications of disagreement, and hence attempts to ground political
theories in the fact of disagreement crumble in an endless regress. The debate over what
Legitimacy, Democracy, and Razian Authority 217
Waldron suggests that the circumstances of politics are marked by:
a felt need among the members of a certain group for a common framework or
decision or course of action on some matter, even in the face of disagreement
about what that framework, decision or action should be.49
Waldron believes that democratically enacted laws achieved in the circumstances
of politics demand a certain sort of respect. He writes:
When something is enacted as law or as a source of law, I believe it makes on
us a demand not to immediately disparage it, or think of ways of nullifying it
or getting around it, or mobilizing the immune system of the corpus juris so as
to resist its incorporation. . . . It is a demand for a certain sort of recognition
and, as I said, respect—that this, for the time being, is what the community has
come up with and that it should not be ignored or disparaged because some
of us propose, when we can, to repeal it.50
This respect, Waldron argues, “is in part an acknowledgment of the need
for a common solution and respect for the conditions of fairness in which a
common solution was arrived at among those who disagreed about what it
should be.”51
Importantly, it is the pedigree of the law that demands respect
of us, and not its substantive merit. Waldron claims: “Once voted on in
the legislature, [a law] is entitled to whatever respect this communitarian
status confers on it, without regard to—indeed bracketing away from—the
substantive merits of its content.”52
I agree with Waldron. Democratically enacted law demands our respect
because it “stands for the time being in the name of the whole community”53
and represents how we have bound ourselves through fair procedures in the
face of disagreement (of course, this is true only where the procedures are
fair). Yet the fact that law is achieved in a democratic community in the
circumstances of politics is not the only claim it has on our respect or obedience.
The values canvassed in the last section—the value of autonomy, the
importance of respect for persons as rational agents, and the consequentialist
value realized through democratic procedures—place demands on
us as well. To my thinking, these values are so important that they demand
disagreement implies for democracy is one of the most important and challenging in political
theory. It is also one that, I believe, can be avoided entirely for our purposes here. I should
point out, however, that if one could successfully show that disagreement over the principles
that should be used to order society grounds an obligation to participate in fair democratic
procedures, one would also have shown that political authority can be legitimate at least in part
on procedural grounds. See J. Raz, Facing Diversity: The Case of Epistemic Abstinence, in ETHICS
IN THE PUBLIC DOMAIN, supra note 15, at 60–96; J. Raz, Liberalism, Scepticism and Democracy, in
ETHICS IN THE PUBLIC DOMAIN, supra note 15, 97–124; Raz, supra note 36.
49. Waldron, supra note 25, at 102.
50. Id. at 100.
51. Id. at 85.
52. Id. at 101.
53. Id.
218 SCOTT HERSHOVITZ
democracy and can overwhelm concerns about substantive legitimacy. Simply
put, making decisions together can be more important than getting
them right.
If the values of democracy support making decisions together even at
the expense of getting them right, then the most important tests for the
legitimacy of political authorities will be procedural or, at the very least,
hybrids of substance and procedure. Of course, the argument so far has
been too cursory to serve as a demonstration that the values of democracy
demand that we make political decisions together, that we submit to fair
democratic procedures and abide by their outcome. A fuller demonstration
is a task left for others, or perhaps for another day. I hope simply that I
have made plausible the claim that political legitimacy has a procedural
dimension.54
If this is right, the normal justification thesis is inadequate as
a test of political legitimacy because it is insensitive to procedure.
I hasten to add the caveat that the obligation to obey the results of fair
democratic procedures can go only so far. We can tolerate just so much deviance
from good decision-making. A fair procedure that consistently produced
disastrous results might undermine the very value of the autonomy
and respect that led us to seek fair procedures in the first place. Democracy
may in some instances be a luxury too expensive to afford. We should also
not underestimate the degree to which we want decisions made by people
with expertise. Indeed, part of the rationale for representative government
over referendum is that we delegate authority to make decisions to a group
of people we expect to engage in the extensive study and consideration
we do not have time to engage in ourselves. We cannot (and need not)
reach any final conclusions here about the proper balancing of substantive
and procedural legitimacy. It is important to recognize, however, that the
normal justification thesis leaves out important tests we ought to apply in
determining whether our political authorities are legitimate.
I want to be careful not to overstate the impact of this objection on Raz’s
analysis. Raz never claims that the normal justification thesis is the sole test
for legitimacy, nor does he claim that it is not a defeasible criterion. Indeed,
Raz indicates some sympathy with the view I have expressed here. He writes:
First, because participation is valuable in itself, providing opportunities for it
through democratic constitutional arrangements is a value which may justify
putting up with some shortfall in other dimensions of performance.55
I suspect that I am willing to put up with more shortfall than Raz. He follows
this passage with the suggestion that “[democracy] relies, for its legitimacy,
54. As Gutmann and Thompson put it: “The moral authority of collective judgments about
policy depends in part on the moral quality of the process by which citizens collectively reach
those judgments.” Gutmann and Thompson, supra note 34, at 4.
55. Raz, supra note 15, at 117. Raz also says that “how we decide what to do is, sometimes, of
independent value.” Id. at 366 n.7.
Legitimacy, Democracy, and Razian Authority 219
on its ability to deliver sound decisions.”56
Democracy, in my view, draws
some of its legitimacy from its ability to deliver sound decisions, but the
main source of its legitimacy lies elsewhere.
The normal justification thesis is neither the normal nor most important
test of political legitimacy. The importance of the normal justification thesis
for understanding authority should not be underestimated—some types of
authority can be legitimate simply because they meet its requirement. However,
we have seen that we have reason to be concerned with the procedural
legitimacy of political authorities. Indeed, political authorities that fall short
of fulfilling the normal justification thesis may nevertheless be legitimate on
procedural grounds. Conversely, political authorities that meet the requirement
of the normal justification thesis may be illegitimate on procedural
grounds.
VIII. AN OBJECTION REJECTED
One might object to this conclusion by claiming that the normal justification
thesis is flexible enough to accommodate the procedural concerns I have
been addressing. The test it offers for determining the legitimacy of an
authority is whether by complying with its directives I will better conform
with the reasons that apply to me than I would otherwise. The objector might
say that where I have reason to comply with the results of a particular decision
procedure, doing so will help me to conform better than I might otherwise to
the reasons I have. We should reject this strategy because it makes the normal
justification thesis empty. So conceived, the normal justification thesis would
accommodate all theories of legitimacy that turned out to be true; hence it
ceases to be a competitor with other candidate theories of legitimacy.
Raz does not appear to understand the normal justification thesis in this
way. He suggests that the need for certain life-defining choices to be made
autonomously by individuals (like choosing one’s career or one’s friends) is
an exception to the normal justification thesis.57
Raz recognizes that some
people might conform better to the reasons that apply to them if the government
chose their career; nevertheless, he suggests, it would be illegitimate
for a government to do so. If Raz was open to the strategy we are exploring
here, he might have argued that there is no way an individual could ever
do better by complying with a government’s directive because of the importance
of making the decision autonomously. He did not argue this for
good reason; it would make the normal justification thesis empty. In order
for the normal justification thesis to be a competitor with other theories of
legitimacy, it cannot be the case that it subsumes or gives way to any theory
that on its face conflicts with it but turns out to be true.
56. Id. at 117.
57. Raz, supra note 7, at 57.
220 SCOTT HERSHOVITZ
This objection misconstrues Raz’s analysis in another way as well. According
to the normal justification thesis, we have reason to treat a directive as
an exclusionary reason when following it helps us better to conform with
the reasons we have for acting on the matter the directive regards. Hence, if a
directive concerns the use of prescription drugs, the test for its legitimacy is
whether it helps us to make better decisions about prescription drugs than
we would on our own. The test is not whether by following the directive
we make our decision the way we have reason to decide such questions, as
it would have to be in order to sustain the objection under consideration.
If something like that were part of the test of the normal justification thesis,
then we would need on all occasions a further (and different) criterion
of legitimacy to assess whether the normal justification thesis was fulfilled,
whether following the directive amounted to making the decision the way
we have reason to make it. All this is just another way of saying that the
normal justification thesis is offered as a candidate theory of legitimacy that
competes with others. It is a substantive test that cannot be stretched to
account for procedural concerns.
IX. CONCLUSION
The normal justification thesis is inadequate as a test for the legitimacy of
political authority. The legitimacy of some authorities, perhaps doctors or
music teachers, may not turn on the procedures they use to reach their
decisions. But political authorities are different. We care about how they
make decisions, and we are right to care. The normal justification thesis
fails as a test for the legitimacy of political authorities because it fails to
account for the procedural dimension of legitimacy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.