SSRN 2995488

COMMENTARY
INCOMPLETELY THEORIZED AGREEMENTS

Cass R Sunstein*
We think utility, or happiness, much too complex and indefinite an end

to be sought except through the medium of various secondary ends, con-
cerning which there may be, and often is, agreement among persons who
differ in their ultimate standard; and about which there does in fact pre-
vail a much greater unanimity among thinking persons, than might be
supposed from their diametrical divergence on the great questions of
moral metaphysics. As mankind are much more nearly of one nature,
than of one opinion about their own nature, they are more easily brought
to agree in their intermediate principles .. than in their first principles
JOHN STUART MTLL1
Why didn't the [Sentencing] Commission sit down and really go and ra-
tionalize this thing and not just take history? The short answer to that
is: we couldn't. We couldn't because there are such good arguments all
over the place pointing in opposite directions. . . . Try listing all the
crimes that there are in rank order of punishable merit.... Then collect
results from your friends and see if they all match. I will tell you they
won't.
2
JUSTICE STEPHEN BREYER
* Karl N. Llewellyn Distinguished Service Professor of Jurisprudence, Law School and De-
partment of Political Science, University of Chicago. This essay is an expansion of certain sec-
tions of the first of my 1994 Tanner Lectures on Human Values, delivered at Harvard University
in November 1994; the lectures themselves will appear under the title PoliticalConflict and Legal
Agreement, in 17 THE TANNER LECTURES ON HTMAN VALUES (Grethe B. Peterson ed., forth-
coming 1996). I am especially grateful to my audiences at Harvard for their extraordinary gra-
ciousness and for their probing comments and questions. Of the many people who offered help
on that occasion, I single out for special thanks my commentators Jean Hampton and Jeremy
Waldron, and also Joshua Cohen, Christine Korsgaard, Martha Minow, Martha Nussbaum, John
Rawls, Tim Scanlon, and Amartya Sen. For extremely helpful comments on the manuscript, I am
grateful to Bruce Ackerman, Ruth Chang, Joshua Cohen, Jon Eister, Charles Fried, Amy Gut-
mann, Don Herzog, Stephen Holmes, Elena Kagan, Dan Kahan, Larry Lessig, Saul Levmore,
William Meadow, Frank Michelman, Martha Minow, Martha Nussbaum, Susan Moller Okin,
Wiktor Osiatynski, Richard Posner, Joseph Raz, Frederick Schauer, Stephen Schulhofer, Anne-
Marie Slaughter, Mark Tushnet, Candace Vogler, and Lloyd Weinreb. I am also indebted to par-
ticipants in a work-in-progress lunch at the University of Chicago and to members of legal theory
workshops at Oxford University and the University of California, Berkeley. I am also grateful to
Sophie Clark for research assistance. Parts of this commentary will appear in a book, LEGAL
REASONING AND POUTICAL CONFLICT (forthcoming 1996).
1 John S. Mill, Bentham, in UTILITARLISM AND OTHER ESSAYS 132, 170 (Alan Ryan ed.,
1987).
2 Justice Breyer is quoted in Jeffrey Rosen, Breyer Restraint, NEw REPUBLIC,July 1I, 1994,
at 19, 25.
'733
Electronic
Electronic copy copy available
available at:at:https://ssrn.com/abstract=2995488
https://ssrn.com/abstract=2995488
1734 HARVARD LAW REVIEW [Vol. io8:x733
I prefer not to [talk] ...like that; again, grand principles have to be

applied in concrete cases. My job involves reasoning from the specific
case ....
3
JUSTICE RuTH BADER GINSBURG
INTRODUCTION
There is a familiar image of justice. She is a single figure. She is a

goddess, emphatically not a human being. She is blindfolded. And
she holds a scale.
In the real world, the law cannot be represented by a single figure.
Legal institutions are composed of many people. Our courts are run
by human beings, not by a god or goddess. Judges need not be blind-
folded; what they should be blind to is perhaps the key question for
law. And judges have no scale. Far from having a scale, they must
operate in the face of a particular kind of social heterogeneity: sharp
and often intractable disagreements on basic principle.
The problem of social pluralism pervades the legal system. Some
of the relevant disagreements are explicitly religious in character.
Others might be described as quasi-religious in the sense that they in-
volve people's deepest and most defining commitments. What is the
appropriate conception of liberty and equality? How should people
educate their children? Should government punish people on the basis
of deterrence only or should it consider retribution as well? Is the free
speech principle about democracy or instead individual autonomy?
There is much dispute about whether well-functioning democracies
should try to resolve such disagreements, and about how they should
do so if they do try.4 Perhaps government should seek an "overlap-
ping consensus", among reasonable people, thus allowing agreements
to be made among Kantians, utilitarians, Aristotelians, and others.
Perhaps participants in a liberal democracy can agree on the right
even if they disagree on the good. 6 Thus a sympathetic observer de-
scribes the liberal "hope that we can achieve social unity7 in a democ-
racy through shared commitment to abstract principles."
3 Nomination of Ruth Bader Ginsburg to Be Associate Justice of the Supreme Court of the
United States: Hearings Before the Senate Judiciary Comm., io3d Cong., ist Sess. 18o (x993)
(statement of then-Judge Ruth Bader Ginsburg).
4 Of course I am referring here to JOHN RAWLS, PoITICAL LIBERALISM (1993) and the sur-
rounding debate. For instructive discussion, see Joshua Cohen, Moral Pluralism and Political
Consensus, in THE IDEA OF DEMOCRACY 270 (David Copp, Jean Hampton & John Roemer eds.,
1993); Jean Hampton, The Moral Commitments of Liberalism, in THE IDEA OF DEMOCRACY,
cited above, at 292; and JOSEPH RAZ, Facing Diversity: The Case of Epistemic Abstinence, in
ETHICS IN THE PUBLIC DOMAIN 45 (1994).
5 RAWLS, supra note 4, at 133.
6 See id. at 133-72.
7 Joshua Cohen, A More Democratic Liberalism, 92 MICH. L. REV. 1503, 1546 (1994) (sum-
marizing RAvI.s, cited above in note 4).
Electronic
Electronic copy copy available
available at:at:https://ssrn.com/abstract=2995488
https://ssrn.com/abstract=2995488
1995] INCOMPLETELY THEORIZED AGREEMENTS 1735
This is a promising approach, and in some settings, it may work.

But an investigation of actual democracy, and of law in actual democ-
racies; raises questions about this view. Democracies - and law in
democracies - must deal with people who very much disagree on the
right as well as the good. Democracies - and law in democracies -
must deal with people who tend to distrust abstractions altogether.
Participants in law are no exception. Judges are certainly not ordinary
citizens. But neither are they philosophers. Indeed, participants in
law may be unwilling to commit themselves to large-scale theories of
any kind, and they will likely disagree with one another if they seek to
agree on such theories.
Judges also have to decide many cases, and they have to decide
them quickly. Decisions must be made rapidly in the face of appar-
ently intractable social disagreements on a wide range of first princi-
ples. These disagreements will be reflected within the judiciary and
other adjudicative institutions as well as within the citizenry at large.
At least this is so if adjudicative institutions include, as they should,
some of the range of views that are included in society generally.
In addition to facing the pressures of time, these diverse people
must find a way to continue to live with one another. They should
also show each other a high degree of mutual respect and reciprocity.
Mutual respect may well entail a reluctance to attack one another's
most basic or defining commitments, at least if it is not necessary to
do so in order to decide particular controversies. Participants in law,
even more than in democratic debate generally, do well to follow this
counsel.
My suggestion in this Commentary is that well-functioning legal
systems often tend to adopt a special strategy for producing agreement
amidst pluralism. Participantsin legal controversies try to produce in-
completely theorized agreements on particularoutcomes.8 They agree
S Consider the notion of overlapping consensus as set out in RAWLs, cited above in note 4, at
133-72. The idea of an incompletely theorized convergence on particulars is related. Both ideas
attempt to bring about stability and social agreement in the face of diverse "comprehensive
views." But the two ideas are far from the same. I am most interested in the problem of produc-
ing agreement on particular outcomes and low-level principles (defined below at p. 1740) to justify
them, with the thought that people who disagree on general principles can often agree on individ-
ual cases. Rawls is interested in a related but different possibility - that people who disagree on
comprehensive views can agree on certain political abstractions and use that agreement for polit-
ical purposes. See id. at 43-46. Of course this is also true.
What I am suggesting here shares with Rawls's project the goal of producing social stability
and a degree of mutuality among people who differ on fundamental matters. But a goal of the
incompletely theorized agreement is to obtain consensus on a judgment among people who do not
want to decide questions in political philosophy and who (for example) are uncertain as between
political liberalism and perfectionist liberalism, or as between liberalism and certain alternatives.
Such people may want to decide a case involving tort law, or free speech, or equality, by refer-
ence to principles that can be accepted by political and perfectionist liberals, and others as well (if
this is possible). In this way, we might conclude that judgments in law and politics sometimes
bear something like the same relation to political philosophy as do judgments in political philoso-
Electronic copy available at: https://ssrn.com/abstract=2995488

1736 HARVARD LAW REVIEW (Vol. 108:1733
on the result and on relatively narrow or low-level explanations for it.

They need not agree on fundamental principle. 9 They do not offer
larger or more abstract explanations than are necessary to decide the
case. When they disagree on an abstraction, they move to a level of
greater particularity. The distinctive feature of this account is that it
emphasizes agreement on (relative) particulars rather than on (relative)
abstractions. This is an important source of social stability and an
important way for diverse people to demonstrate mutual respect,10 in
law especially but also in liberal democracy as a whole.
Consider some examples. People may believe that it is important
to protect endangered species, while having quite diverse theories of
why this is so. Some may stress obligations to species or nature as
such; others may point to the role of endangered species in producing
ecological stability; still others may point to the possibility that obscure
species will provide medicines for human beings. When (and if) peo-
ple who agree on the same course of action are able to do so from
different foundations, they need not choose among foundations. So
too, people may favor a rule of strict liability for certain torts from
diverse starting points, with some people rooting their judgments in
economic efficiency, others in distributive goals, and still others in con-
ceptions of basic rights. Similarly, people may invoke many different
foundations for their belief that the law should protect labor unions
against certain kinds of employer coercion. Some may emphasize the
potentially democratic character of unions; others may think that un-
ions are necessary for industrial peace; others may believe that unions
protect basic rights.
Of course particular issues, in these as in other areas, are disputed
as well. Complete agreement is unlikely in matters of this sort. Dis-
phy to questions in metaphysics. (On Rawls' view, see RAwLS, cited above in note 4, at xix-xx.)
Just as the political philosopher may attempt not to take a stand on metaphysical questions, so
the lawyer, the judge, or the political participant may urge outcomes that make it unnecessary to
solve large questions in political philosophy. In a liberal society committed to allowing people of
different fundamental views to live together with mutual respect, the Rawisian strategy may
sometimes founder on confusion, limitations of time and capacity, and fears that political liber-
alism is itself too sectarian to serve as a defining political creed. I do not mean to suggest that
the Rawlsian project is unable to surmount these concerns. Certainly there must be constraints
on the content of incompletely theorized agreements, and those constraints take the form of ab-
stractions. But I do mean to suggest that participants in liberal political culture often seek agree-
ment on what to do rather than exactly how to think. When they reach these agreements from
diverse starting points, they can promote liberal goals in a way that has some distinctive advan-
tages. I do not attempt here to sort out all of the relations between the idea of an overlapping
consensus and the notions I have in mind; more detailed discussion will appear in SUNSTEIN,
LEGAL REASONING AND POLITICAL CONFLICT, cited above in note *.
9 I am thus emphasizing relative rather than absolute particularity. See infra pp. 1739-41.
It is also possible that people may agree on abstraction while disagreeing on particulars, a kind of
incompletely theorized agreement that also has legal uses. See infra p. 1739.
10 There is an exception, having to do with certain kinds of invidious or palpably confused
abstractions. See infra p. 1747.

agreement on foundations may produce disagreement on particulars.

What I am emphasizing is that, when closure cannot be based on rela-
tive abstractions, the legal system is often able to reach a degree of
closure by focusing on relative particulars. Examples of this kind are
exceptionally common. They are the day-to-day stuff of law.
When the convergence on particular outcomes is incompletely theo-
rized, it is because the relevant actors are clear on the result without
being clear, either in their own minds or on paper, on the most general
theory that accounts for it."I Some people may not have decided on
the best general theory; in their individual capacities, they may think
that it is unnecessary to choose among competing general theories, be-
cause whatever theory is best, the same result follows. It is common
for a judge to believe that a certain outcome is right because all possi-
ble theoretical positions lead to it. Alternatively, a particular judge
may believe that he knows which theory is best, but may not be able
to say so in an opinion, for fear of being outvoted, or because other
judges on his court are uncertain whether he is right and would prefer
to leave the more abstract issues for another day.
Often judges can agree on an opinion or rationale offering low-
level or mid-level principles and taking a relatively narrow line. They
may agree that a particular rule is binding and makes sense - a
sixty-five mile-per-hour speed limit, a requirement that people be pro-
vided a hearing before losing their homes - without agreeing on or
entirely understanding any set of purported foundations for their be-
lief. They may accept an outcome - reaffirming Roe v. Wade,' 2 pro-
tecting sexually explicit art - without knowing or converging on an
ultimate ground for that acceptance. Reasons are almost always of-
fered,' 3 and in this sense something in the way of abstraction accom-
panies the outcome; reasons are by definition more abstract than the
outcome for which they account. But the relevant actors seek to stay
at the lowest level of abstraction necessary for the decision of the case.
They hope that the reasons that have been offered are compatible with
an array of deeper possible reasons, and they refuse to make a choice
11 Interesting issues of collective choice lurk in the background here. Important problems of
cycling, strategic behavior, and path dependence may arise in multimember bodies containing
people with divergent rationales, each of whom wants to make his rationale part of law. See
generally KENNETH J. ARRow, SOCIAL CHOICE AND INDIVIDUAL VALUEs passim (2d ed. 1963)
(analyzing problems of public choice). There may also be complex bargaining issues as some
officials or judges seek to implement a broad theory as part of the outcome, while others seek a
narrow theory, and still others are undecided between the two. These important issues are be-
yond the scope of the present discussion, though it would be most illuminating to have a better
grasp, theoretically and empirically, of the sorts of bargaining games that occur when officials and
judges decide on the scope of the theory to accompany an outcome. Cf. DOUGLAS G. BAIRD,
ROBERT H. GERTNER & RANDAL C. PICKER, GAME THEORY AND THE LAW 6-49 (1994) (apply-
ing game theory models to analyze how players choose among different legal regimes).
12 410 U.S. 113 (1973).
13 For exceptions, se6 the discussion below at pp. 1754-60.

1738 HARVARD LAW REVIEW Vol. 108:3733
among those deeper reasons if it is not necessary to do so. This phe-

nomenon raises questions about many prominent accounts of judging,
particularly that of Ronald Dworkin, which I discuss below. 14
By emphasizing incompletely theorized agreements, I intend partly
to describe existing practice. These agreements are a pervasive phe-
nomenon in Anglo-American law and, in particular, in judge-made
law. Indeed, they play an important function in any well-functioning
democracy consisting of a heterogeneous population. The persistence
of such agreements offers a challenge to people who think that areas
of law actually reflect some general theory, involving (for example)
is
Kantian or utilitarian understandings.
I want, however, to make some normative claims as well. These
claims are connected with the distinctive morality of judging and per-
haps the distinctive morality of politics in a pluralistic society. It is
customary to lament an outcome that has not been completely theo-
rized, on the ground that any such outcome has been inadequately jus-
tiffied; but there are special advantages to incompletely theorized
agreements in law (and elsewhere). Such agreements are especially
well suited to the institutional limits of the judiciary, which is com-
posed in significant part of multimember bodies, consisting in turn of
highly diverse people who must render many decisions, live together,
avoid error to the extent possible, and show each other mutual respect.
The virtues of incompletely theorized agreements extend as well to so-
cial life, to workplace and familial life, and even to democratic poli-
tics. I will note these possibilities without discussing them in detail
here.
Of course I do not suggest that there is no place in society or law
for ambitious theories. Such theories are an important part of the aca-
demic study of law, which can influence legal and political institutions.
Certainly such theories are legitimately a part of democratic delibera-
tion. To reveal practices as confused or unjust, or to decide whether
practices are confused or unjust, it may be necessary to think at high
levels of abstraction. Indeed, judges might well raise the level of theo-
retical ambition if this is necessary to decide a case, or if a high-level
theory can be shown to be a good one and they can be persuaded to
agree to it. But participants in adjudicative institutions are - and
should be - very cautious before taking this step.
14 See infra pp. 1757-6o.

Is See, e.g., RONALD DwORaNq, LAW'S EmPIRE 24o-50 (1986) (arguing that tort law requires
inquiry into abstract "justice"); RICHARD A. POSNER, ECONOMIC ANALYsIs OF LAW 23 (4th ed.
1992) (arguing that the common law actually promotes economic efficiency). Of course, advocates
of a deep theory might urge that incompletely theorized agreements, taken as a class, tend to
reflect a certain theory even if the theory has not been adopted self-consciously, and that those
decisions that are inconsistent with the theory should be abandoned.

,9951 INCOMPLETELY THEORIZED AGREEMENTS 1739
I. AGREEMENTS WITHOUT THEORY

A. In General
Incompletely theorized agreements play a pervasive role in law and
society. It is rare for a person, and especially for a group, to theorize
any subject completely - that is, to accept both a highly abstract the-
ory and a series of steps that relate the theory to a concrete conclu-
sion. In fact, people often reach incompletely theorized agreements on
a general principle. Such agreements are incompletely theorized in the
sense that people who accept the principle need not agree on what it
entails in particular cases. People know that murder is wrong, but
they disagree about abortion. They favor racial equality, but they are
divided on affirmative action. Hence there is a familiar phenomenon
of a comfortable and even emphatic agreement on a general principle,
accompanied by sharp disagreement about particular cases.16
This sort of agreement is incompletely theorized in the sense that it
is incompletely specified - a familiar phenomenon with constitutional
provisions 17 and regulatory standards in administrative law. Incom-
pletely specified agreements have distinctive social uses. They may
permit acceptance of a general aspiration when people are unclear
about what the aspiration means, and in this sense, they can maintain
a measure of both stability and flexibility over time. At the same
time, they can conceal the fact of large-scale social disagreement about
particular cases.
There is a second and quite different kind of incompletely theo-
rized agreement. People may agree on a mid-level principle but dis-
agree both about the more general theory that accounts for it and
about outcomes in particular cases. They may believe that govern-
ment cannot discriminate on the basis of race, without settling on a
large-scale theory of equality, and without agreeing whether govern-
ment may enact affirmative action programs or segregate prisons when
racial tensions are severe. The connections are left unclear, either in
people's minds or in authoritative public documents, between the mid-
level principle and general theory; the connection is equally unclear
between the mid-level principle and concrete cases. So too, people
may think that government may not regulate speech unless it can
show a clear and present danger, but fail to settle whether this princi-
ple is founded in utilitarian or Kantian considerations, and disagree
about whether the principle allows government to regulate a particular
speech by members of the Ku Klux Klan.
16 For a valuable discussion, see Henry S. Richardson, Specifying Norms as a Way to Resolve
Concrete Ethical Problems, x9 PHIL. & PUB. AFF. 279, 306 (iggo).
17 Thus, such terms as "equality" and "freedom" are "essentially contested" in the sense made
famous by W.B. Gallie, Essentially Contested Concepts, 56 PRoc. ARISTOTELIAN Soc'Y 167
(ig9-56).

My special interest here is in a third kind of phenomenon - in-

completely theorized agreements on particular outcomes, accompanied
by agreements on the low-level principles that account for them.
These terms contain some ambiguities. There is no algorithm by
which to distinguish between a high-level theory and one that operates
at an intermediate or low level. We might consider Kantianism and
utilitarianism as conspicuous examples of high-level theories and see
legal illustrations in the many (academic) efforts to understand such
areas as tort law, contract law, free speech, and the law of equality to
be undergirded by highly abstract theories of the right or the good.1 8
By contrast, we might think of low-level principles as including most
of the ordinary material of legal doctrine - the general class of princi-
ples and justifications that are not said to derive from any particular
large theories of the right or the good, that have ambiguous relations
to large theories, and that are compatible with more than one such
theory. 19
By the term "particular outcome," I mean the judgment about who
wins and who loses a case. By the term "low-level principles," I refer
to something relative, not absolute; a principle is low-level only when
compared to more abstract alternatives. I also mean the terms "theo-
ries" and "abstractions" (which I use interchangeably) in a relative
sense: the notions "low-level," "high-level," and "abstract" are best un-
derstood as comparative, like the terms "big" and "unusual." The
"clear and present danger" test is abstract when compared with the
judgment that members of the Nazi Party may march in Skokie, flli-
nois, 20 but the test is relatively particular when compared with the
constitutional abstraction "freedom of speech." The idea of "freedom
of speech" is relatively abstract when measured against the notion that
campaign finance laws are acceptable, but the same idea is less ab-
stract than the grounds that justify free speech, such as the principle
of autonomy. 2 1 The notion of a completely theorized judgment can be
understood in the abstract: it refers to a judgment on a particular set
of facts combined with all relevant vertical and horizontal judgments,
as in a judgment in a free speech case accompanied by an understand-
ing of all other free speech cases, or perhaps of all cases, and of all
principles, at all levels of generality, that explain or justify that
judgment.
What I am emphasizing here is that when people diverge on some
(relatively) high-level proposition, they might be able to agree if they
18 See, e.g., Thomas Scanlon, A Theory of Freedom of Expression, x PHIL. & PuB. AFF. 204
(1972); DWORicN, supra note iS, at 407-10 (1986); CHARLES FRIED, CONTRACT As PROMISE 7-8
(198I).
19 See JOSEPH RAZ, On the Autonomy of Legal Reasoning, in ETHICS IN THE PUBLIC Do-
MAIN, supra note 4, at 310, 314-24 (1994) (discussing doctrine and the autonomy of law).
20 See Collin v. Smith, 578 F.2d 1197, 1202-03 (7th Cir. I978).
21 See Scanlon, supra note 18, at 215-20.

1995] INCOMPLETELY THEORIZED AGREEMENTS 1741r
lower the level of abstraction. People are sometimes able to converge

on a point of less generality than the point at which agreement is diffi-
cult or impossible. In law, the point of agreement is often highly par-
ticularized - absolutely as well as relatively - in the sense that it
involves a specific outcome and a set of reasons that typically do not
venture far from the case at hand.
As I have said, reasons are, by their very nature, more abstract
than the outcome for which they account. In analogical thinking, a
judge cannot go from one particular to another; it is necessary to iden-
tify a principle or a reason to unify or separate the particulars. To the
extent that law prizes reason-giving, it contains an impulse toward ab-
straction. Full particularity is rare.2 2 What I am emphasizing is the
lawyer's impulse to offer reasons on which people can unite from
widely diverse foundations.
Perhaps the participants in law endorse no high-level theory, or
perhaps they believe that no such theory is yet available. Perhaps they
find theoretical disputes irrelevant, confusing, or annoying. Perhaps
they think that it is unnecessary to select a theory, because a certain
outcome follows whatever theory they choose. Perhaps they disagree
with one another as they enter into high-level debates. What is critical
is that they agree on how a case must come out and on a low-level
justification.
The argument emphatically applies to legal rules, whether set
down by judges, legislators, or administrators. Legal rules are typi-
cally incompletely theorized in the sense that they can be accepted by
people who disagree on many general issues. People may agree that a
sixty-five mile-per-hour speed limit makes sense, and that it applies to
defendant Jones, without having much of a theory about criminal pun-
ishment, and without making judgments about the domain of utilitari-
anism and the scope of paternalism. A key social function of rules is
to allow people to agree on the meaning, authority, and even the
soundness of a governing legal provision in the face of disagreements
about much else. 23 Much the same can be said about other devices
found in the legal culture, including standards, factors, and emphati-
cally analogical reasoning. 24 Indeed, all of the lawyer's conventional
22 See infra section ILC.

23 See JOSEPH RAz, THE MORALITY OF FREEDOM 58 (x986) ("[T]he practice [of proceeding
through the mediation of rules] allows the creation of a pluralistic culture. For it enables people
to unite in support of some 'low or medium level' generalizations despite profound disagreements
concerning their ultimate foundations, which some seek in religion, others in Marxism or in Liber-
alism, etc.").
24 1 discuss the latter in Cass R. Sunstein, On Analogical Reasoning, xo6 HARv. L. REV. 741
passim (r993).

1742 HARVARD LAW REVIEW [Vol. 108:1733
tools can allow the achievement of incompletely theorized agreements

on particular outcomes, though in interestingly different ways. 25
It should now be clear that I am especially concerned with the use
of ambitious thinking to produce "depth" - full accounts of the foun-
dations of a decision, in the form of attempts to find ever deeper rea-
sons behind the outcome. Incompletely theorized agreements do not
offer such accounts. But such agreements also fail along the dimen-
sion of "width" - that is, they do not try to rationalize the law by
showing how an outcome in one case fits coherently with particular
outcomes in the full range of other cases. Judges, of course, attempt to
produce local coherence, especially through reasoning by analogy. But
they do not try for global coherence.
B. How People Converge
It seems clear that people may converge on a correct outcome even
though they do not have a full account for their judgments. Jones
may know that dropped objects fall, that bee stings hurt, and that
snow melts, without knowing exactly why these facts are true. Much
the same is true for law and morality. Johnson may know that slavery
is wrong, that government may not stop political protests, and that
every person should have just one vote, without knowing exactly why
these things are so. Judge Wilson may know that under the Constitu-
tion, discrimination against the handicapped is generally permitted,
and that discrimination against women is generally banned, without
having an account of why the Constitution is so understood. We may
thus offer an epistemological point: people can know that X is true
without entirely knowing why X is true. This is very often so for
particular conclusions about law.
There is a political point as well. People can agree on individual
judgments even if they disagree on general theory. Diverse judges
may agree that Roe v. Wade26 should not be overruled, though the
reasons that lead each of them to that conclusion sharply diverge.
Perhaps the judges have different large-scale theories and can agree
only on a low-level principle. Perhaps some of the judges have not
developed ambitious accounts of the relevant area of the law at all.
Thus some people emphasize that the Court should respect its own
precedents; others think that Roe was rightly decided as a way of pro-
tecting women's equality; others think that the case was rightly de-
cided as a way of protecting privacy; others think that the case has
everything to do with state neutrality toward religion; others think
that restrictions on abortion are unlikely to protect fetuses, and so the
25 I emphasize "can" rather than "will." Analogies are frequently disputed, and people often
disagree on the meaning of a standard ("unreasonable risk," "discrimination," and so forth).
26 410 U.S. 113 (x973). On the refusal to overrule Roe, see Planned Parenthood v. Casey, 112
S. Ct. 2791, 2804 (1992).

case rightly reflects the fact that any regulation of abortion would be
ineffective in promoting its own purposes. We can find incompletely
theorized political agreements on particular outcomes in many areas of
law and politics - on both sides of the affirmative action controversy,
both sides of the dispute over the death penalty, and in all facets of
the debate over health care reform.
C. Rules and Analogies
There are two especially important methods by which law might
resolve disputes without obtaining agreement on first principles: rules
and analogies. Both of these methods attempt to promote a major
goal of a heterogeneous society: to make it possible to obtain agreement
where agreement is necessary, and to make it unnecessary to obtain
agreement where agreement is impossible.
For purposes of law, reliance on rules might be incompletely theo-
rized in three different ways. People might agree that rules are bind-
ing without having a full or agreed-upon account of why this is so.
They can often agree on what rules mean even when they agree on
very little else.2 7 They can even agree that certain rules are good
without agreeing on exactly why they are good. And in the face of
persistent disagreement or uncertainty about what morality generally
requires, people can sometimes reason about particular cases by refer-
ence to analogies. They point to cases in which their judgments are
firm, and proceed from those firm judgments to the more difficult
ones. From different foundations, they may be able to agree on the
plausibility of an analogical claim because they share a judgment
about a governing low-level principle in the face of disagreement
about the abstractions underlying that principle.
We might consider, in this regard, Justice Stephen Breyer's discus-
sion of one of the key compromises reached by the seven members of
the United States Sentencing Commission.2 8 As Justice Breyer de-
scribes it, a central issue was how to proceed in the face of disparate
theoretical commitments. Some people asked the Commission to fol-
low an approach to punishment based on "just deserts" - an ap-
proach that would rank criminal conduct in terms of severity. But
different commissioners had different views about how different crimes
should be ranked, and a rational system was unlikely to follow from a
collaborative ranking effort. Other people urged the Commission to
use a model of deterrence. There was, however, no empirical evidence
to link detailed variations in punishment to prevention of crime.
27 Of course, substantive disagreements may break out during interpretation. See FREDERICK
SCHAUER, PLAYING BY THE RULES 207-28 (i99i); Cass R. Sunstein, Problems with Rules, 83
CAL. L. REv. (forthcoming July 1995).
28 See Stephen G. Breyer, The FederalSentencing Guidelines and the Key Compromises upon
Which They Rest, 17 HOFSTRA L. REv. 1, 15-i8 (1988).

1744 HARVARD LAW REVIEW [Vol. 1083733
Though Justice Breyer does not stress the point, it seems clear that the
seven members of the Commission were unlikely to agree that deter- 29
rence provides a full account of the aims of criminal sentencing.
In these circumstances, what route did the Commission follow? In
fact, the Commission adopted no general view about the appropriate
aims of criminal sentencing. Instead, the Commission adopted a rule
- one founded on precedent: "It decided to base the Guidelines pri-
marily upon typical, or average, actual past practice."30 The Commis-
sion reached an incompletely theorized agreement on the value of
starting (and usually ending) with past averages. Hence unusual judi-
cial sentences would be filtered out through adoption of typical or av-
erage practices. Consciously articulated explanations, involving low-
level reasons, were used to support particular departures from the
past. Justice Breyer saw this effort as a necessary means of obtaining
agreement and rationality within a multimember body 31
charged with
avoiding unjustifiably wide variations in sentencing.
The example suggests a quite general point. Through both analo-
gies and rules, it is often possible to achieve convergence on particular
disputes without resolving large-scale issues of the right or the good.
For judges and officials at least, this is an important virtue.
The fact that we can obtain an agreement of this sort - about the
usefulness and meaning of a rule or the existence of a sound analogy
- is no guarantee of a good outcome, whatever may be our criteria
for deciding whether an outcome is good. The fact that there is agree-
ment about a rule does not mean that the rule is desirable. Perhaps
the rule is bad, or perhaps the judgments that go into its interpreta-
tion are bad. The resolution of the Sentencing Commission deserves
approval only if average practices were not pervasively unjust. Per-
haps the Sentencing Commission incorporated judgments that were
based on ignorance, confusion, or prejudice.3 2 Perhaps a more33
deeply
theorized approach would have produced better guidelines.
Some of the same things may be said about analogies. People in
positions of authority may agree that a ban on same-sex marriages is
analogous to a ban on marriages between uncles and nieces. But the
analogy may be misconceived because there are relevant differences,
because the similarities are far from decisive, or because the principle
that accounts for the judgment of similarity cannot be sustained. The
fact that people agree that case A is analogous to case B does not
29 See id. at 17.
30 Id.
31 See id. at 18; supra p. 1733.
32 For arguments to this effect, see Albert IV. Alschuler, The Failure of Sentencing Guidelines:
A Plea for Less Aggregation, 58 U. Cm. L. REv. 9o, 905-07 ('99'); Daniel J. Freed, Federal
Sentencing in the Wake of Guidelines: Unacceptable Limits on the Discretion of Sentencers, ios
YALE L.J. 1683, 1703-05 (z992).
33 I do not discuss the choice between rules and rulelessness here.

mean that case A or case B is rightly decided. Problems with analo-
gies and low-level principles might lead us to be more ambitious. Par-
ticipants in law may well be pushed in the direction of general theory
- and toward broader and more ambitious claims - precisely be-
cause low-level reasoning offers an inadequate and incompletely theo-
rized account of relevant similarities or relevant differences.
All this should be sufficient to show that the virtues of incom-
pletely theorized outcomes - and the virtues of decisions by rule and
by analogy - are partial. Those virtues should not be exaggerated,
and sometimes participants will have to raise the theoretical stakes.
But no system of law is likely to be either just or efficient if it dis-
penses with incompletely theorized agreements. In fact, it is not likely
even to be feasible.
II. JUSTIFICATIONS AND INSTITUTIONS
What might be said on behalf of incompletely theorized agree-

ments, or incompletely theorized judgments, about particular cases?
As I have indicated, incompletely theorized agreements may be unjust
or otherwise wrong. Indeed, we are accustomed to thinking of incom-
plete theorization as reflective of some important problem or defect.
When people theorize by raising the level of abstraction, they do so to
reveal bias, or confusion, or inconsistency. Surely participants in a
legal system should not abandon this effort.
There is a good deal of truth in these usual thoughts, but they are
not the whole story. On the contrary, incompletely theorized judg-
ments are an important and valuable part of both private and public
life. 34 My principal concern is how judges on a multimember body
should justify their opinions in public; the argument therefore has a
great deal to do with the problem of collective choice. But some of
the relevant points bear on other issues as well. They have implica-
tions for the question of how an individual judge not faced with the
problem of producing a majority opinion - a judge on a district
court, for example - might write; they bear also on the question of
how a single judge, whether or not a member of a collective body,
might think in private as well as write for the public; and they relate
to appropriate methods of both thought and justification wholly
outside of the adjudicative setting. I will say more about these issues
below.
34 I have discussed some of these points in the specific context of analogical thinking. See
Sunstein, supra note 24. I try here to broaden and generalize the discussion and also to correct
some ambiguities and errors in that discussion.

1746 HARVARD LAW REVIEW [VOL. 108:1733
A. The Case for Incomplete Theorization

x. Multimember Institutions. - I begin with the special problem
of public justification faced by a multimember body. The first and
most obvious point is that incompletely theorized agreements are well
suited to a world - especially a legal world - containing social disa-
greement on large-scale issues. By definition, such agreements have
the large advantage of allowing a convergence on particular outcomes
by people unable to reach anything like an accord on general princi-
ples. This advantage is associated not only with the simple need to
decide cases but also with social stability, which could not exist if fun-
damental disagreements broke out in every case of public or private
dispute.3 5
Second, incompletely theorized agreements can promote two goals
of a liberal democracy and a liberal legal system: to enable people to
live together, 36 and to permit them to show each other a measure of
reciprocity and mutual respect.37 The use of rules or low-level princi-
ples allows judges on multimember bodies and hence citizens generally
to find commonality - and a common way of life - without produc-
ing unnecessary antagonism. Perhaps more important, incompletely
theorized agreements allow people to show each other a high degree of
mutual respect, civility, or reciprocity. Ordinary people frequently dis-
agree in some deep way on an issue - the Middle East, pornography,
gay marriages - and sometimes they agree not to discuss that issue
much, as a way of deferring to each other's strong convictions (even if
they do not at all respect the particular conviction that is at stake).38
If reciprocity and mutual respect are desirable, it follows that judges,
perhaps even more than ordinary people, should not challenge a liti-
gant's or one another's deepest and most defining commitments, at
3s The institution of the concurring opinion shows that judges sometimes reject a colleague's
low-level justification, perhaps because they accept the result but would prefer a different low-
level justification, perhaps because they would prefer a lower-level or higher-level justification.
As concurring opinions proliferate, we may begin to reach full particularity in the sense of an
agreement on an outcome without agreement on a low-level rationale. See, e.g., Furman v. Geor-
gia, 4o8 U.S. 238, 240 (Douglas, J., concurring); id. at 257 (Brennan, J., concurring); id. at 3o6
(Stewart, J., concurring); id. at 3o (White, J., concurring); id. at 314 (Marshall, J. concurring);
infra pp. 1754-57. This phenomenon is revealing insofar as it demonstrates the possibility of
agreement on an outcome unaccompanied by agreement on any set of reasons. But judges strug-
gle very hard to avoid full particularity, and they seek incompletely theorized agreements as a
way of providing an explanation and a degree of guidance despite their differences.
36 This aspect of liberalism is emphasized in CHARLES E. LARMORE, PATTERNS OF MORAL
COMPLEXITY 42-47 (1987).
37 See RAWLs, supra note 4, at 6-17 ("[T]he idea of reciprocity lies between the idea of
impartiality, which is altruistic[,]... and the idea of mutual advantage understood as everyone's
being advantaged with respect to each person's present or expected future situation as things
are.").
38 Cf Stephen Holmes, Gag Rules or the Politics of Omission, in CONSTITIONAUSM AND
DEMOCRACY ig passim (Jon Elster & Rune Slagstad eds., 1988) (discussing rules designed to pre-
vent discussion of issues that paralyze political participants).

I995] INCOMPLETELY THEORIZED AGREEMENTS T747
least if those commitments are reasonable and if there is no need for
them to do so. Thus, it would be better if judges intending to reaffirm
Roe v. Wade could do so without challenging the belief that the fetus
is a human being,3 9 or if judges seeking to invalidate the death pen-
alty could do so without saying that the punishment of death is inva-
lid because of its sheer brutality. 40
Institutional arguments in law, especially those involving judicial
restraint, are typically designed to bracket fundamental questions and
remove them from the realm of the judiciary. 4 1 The allocation of cer-
tain roles has the important function of allowing outcomes to be
reached without forcing courts to make decisions on fundamental is-
sues. 42 Those issues are resolved by reference to institutional compe-
tence, not on their merits.43
To be sure, some fundamental commitments might appropriately be
challenged in the legal system or within other multimember bodies.
Some such commitments are ruled off-limits by the authoritative legal
materials. Many provisions involving basic rights have this function.44
Of course, it is not always disrespectful to disagree with someone in a
fundamental way; on the contrary, such disagreements may reflect
profound respect. For example, objections to judgments based on
prejudice may reflect respect for the person at the same time that they
attempt to overcome the judgments at issue. When defining commit-
ments are based on demonstrable errors of fact or logic, it is appropri-
ate to contest them. The same is true when those commitments are
rooted in a rejection of the basic dignity of all human beings, or when
it is necessary to undertake the contest to resolve a genuine problem. 45
But many cases can be resolved in an incompletely theorized way, and
that is all I am suggesting here.
2. Multimember Institutions and Individual Judges. - I turn now
to reasons that call for incompletely theorized agreements whether or
39 This is the goal of the equal protection argument. See CASS R. SUNSTEIN, THE PARTIAL
CONSTITUTION 259-61, 272-85 (I993).
40 This is the goal of the argument that the death penalty is invalid because it is arbitrary in
its application. See Callins v. Collins, 114 S. Ct. 1127, 1129-3 o ('994) (Blackmun, J., dissenting
from denial of certiorari); Godfrey v. Georgia, 446 U.S. 420, 437-40 (198o) (Marshall, J., concur-
ring in the judgment).
41 See, e.g., Harris v. McRae, 448 U.S. 297, 326 (i98o) ("It is not the mission of this court or
any other to decide whether the balance of competing interests reflected in the Hyde Amendment
is wise social policy.").
42 See DON HERZOG, HAPPY SLAVES I2 (1989) ("[J]udges must simply ignore a host of politi-
cally charged facts.").
43 This is part of the legal process tradition. See William N. Eskridge, Jr., & Philip P.
Frickey, The Making of The Legal Process, 107 HARV. L. REV. 2031, 2037-38 (1994). The merits
are thus excluded as a reason for action. For a general discussion of exclusionary reasons, see
JOSEPH RAz, PRAcTiCAL REASON AND NORMS 73-76 (2d ed. 199o).
44 See, e.g., U.S. CONST. amend. XIII.
4S Cf. RAwLs, supra note 4, at 4 (discussing reasonable pluralism); see also Hampton, supra
note 4, at 292-312 (responding to Rawls).

not we are dealing with a multimember body. The first consideration

here is that any simple general theory of a large area of the law -
free speech, contracts, property - is likely to be too crude to fit with
the best understandings of the multiple values that are at stake in that
area. Monistic theories of free speech or property rights, for example,
fail to accommodate the range of values that speech and property im-
plicate. Human goods are plural and diverse, and they cannot be
ranked along any unitary scale without doing violence to our under-
standing of the qualitative differences among those very goods.4 6 Peo-
ple value things not just in terms of weight but also in qualitatively
different ways. In the area of free speech, a simple top-down theory
- stressing, for example, autonomy or democracy - is likely to run
afoul of powerful judgments about particular cases. For this reason,
47
monistic theories are usually inadequate.
Of course, a "top-down" approach might reject monism and point
to plural values. 48 But any such approach is likely to owe its genesis
and its proof - its point or points - to a range of particular cases,
and considered judgments about particular cases, on which it can
build. Top-down approaches based on plural values may seem con-
vincing, but incompletely theorized judgments are well suited to a
moral universe that is diverse and pluralistic, not only in the sense
that people disagree, but also in the sense that each of us is attuned
49
to
pluralism when we are thinking well about any area of law.
Second, incompletely theorized agreements serve the crucial func-
tion of reducing the political cost of enduring disagreements. If judges
disavow large-scale theories, then losers in particular cases lose much
less. They lose a decision but not the world. They may win on an-
other occasion. Their own theory has not been rejected or ruled inad-
missible. When the authoritative rationale for the result is
disconnected from abstract theories of the good or the right, the losers
can submit to legal obligations, even if reluctantly, without being
forced to renounce their deepest ideals. To be sure, some theories
should be rejected or ruled inadmissible; this is sometimes the point of
authoritative legal materials. But it is an advantage, from the stand-
46 See EUZABETH ANDERSON, VALUE IN ETHICS AND ECONOMICS 55-59 (1993); Amartya
Sen, Plural Utility, 81 PRoc. ARISTOTELIAN SOC'Y 193, 193-210 (1981); Cass . Sunstein, Incom-
mensurability and Valuation in Law, 92 ICH. L. REV. 779, 795-812 (1994).
47 See T.M. Scanlon, Jr., Freedom of Expression and Categories of Expression, 40 U. PITT. L.
REV. 519, 519-20 (1979) (rejecting autonomy theory as a full account). Analogical reasoning is
especially desirable here. This way of thinking allows judges and lawyers to build doctrine with
close reference to particular cases and thus with close attention to the plurality of values that may
well arise. This plurality will confound "top-down" theories that attempt, for example, to under-
stand speech only in terms of democracy, or property only in terms of economic efficiency.
48 See, e.g., AMARTYA SEN, COMMODITES AND CAPABILITIES 33-37 (1985); Sen, supra note
46, at 199--204.
49 See JOSEPH RAz, The Relevance of Coherence, in ETHICS IN THE PUBLIC DOMAIN, sUpra
note 4, at 298-303 [hereinafter RAz, Relevance of Coherence] (discussing pluralism and coherence).

point of freedom and stability, for a legal system to be able to tell
most losers - many of whom are operating from foundations that
have something to offer, or that cannot be ruled out a priori - that
their own deepest convictions may play a role elsewhere in the law.
The third point is that incompletely theorized agreements may be
valuable when what is sought is moral evolution over time. Consider
the area of constitutional equality, where considerable change has oc-
curred in the past and is likely to occur in the future. A completely
theorized judgment - at least if it takes rule-like form - would be
unable to accommodate changes in facts or values.50 Incompletely the-
orized agreements are a key to debates over constitutional equality,
which raise issues about whether gender, sexual orientation, age, disa-
bility, and other characteristics are analogous to race. Incompletely
theorized agreements have the important advantage of allowing a large
degree of openness to new facts and perspectives. At one point, we
might think that homosexual relations are akin to incest; at another
point, we might find the analogy bizarre. A completely theorized
judgment would, of course, have many virtues if it were correct. But
at any particular moment in time, this is an unlikely prospect for
human beings, including judges, in constitutional law or elsewhere.5 1
Fourth, incompletely theorized agreements may be the best ap-
proach available for people with limited time and capacities. The
search for full theorization may be simply too difficult for participants
in law to complete, and so too for others attempting to reason through
difficult problems. A single judge faces this problem as much as a
member of a multimember panel. In this respect, the principle of stare
decisis is crucial: attention to precedent is liberating, not merely con-
fining, since it is far easier for judges to decide cases if they can take
much law as settled. The rule of precedent thus assists in the process
of obtaining agreements among people who disagree on first principles.
Indeed, precedents can lower the level of theorization by making more
foundational views irrelevant or even inappropriate5 2 and by binding
judges to outcomes that they would like to reject.5 3 In any event, in-
completely theorized agreements have the advantage, for ordinary law-
yers and judges, of humility and modesty: they allow past judgments
to be treated as given and make it unnecessary to create the law anew
in each case.
50 I am assuming here that complete theorization results in fixed rules rather than principles
whose application can change over time. An agreement on an abstraction would not create fixity
if the abstraction were indeterminate or capable of new meanings in new contexts.
S1 Rawls says that, in all cases, the "struggle for reflective equilibrium continues indefinitely."
RAWLS, supra note 4, at 97-
52 See Planned Parenthood v. Casey, 112 S. Ct. 2791, 28o6-ii (1992) (using stare decisis to
avoid new judgment on merits of abortion right).
S3 See Antonin Scalia, The Rule of Law as a Law of Rules, 56 U. Cm. L. REV. 1175, 1177-80
(1989) (discussing virtues of rules).

1750 HARVARD LAW REVIEW [Vol 108:1733
Fifth, and finally, incompletely theorized agreements are well

adapted to a system that should or must take precedents as fixed
points. This is a large advantage over more ambitious methods, since
ambitious thinkers, in order to reach horizontal and vertical coherence,
will probably be forced to disregard many decided cases. In light of
the sheer number of adjudicative officials, law cannot speak with one
voice; full coherence in principle is unlikely in the extreme.5 4 The area
of contract law is unlikely to cohere with the field of tort law, or prop-
erty law; contract law is itself likely to contain multiple and sometimes
inconsistent strands. Multiple and sometimes inconsistent strands are
a natural outgrowth of incompletely theorized agreements, which are
themselves a way of minimizing the extent and depth of conflict. To
be sure, the existence of such agreements may increase conflict by vir-
tue of inconsistency.
It is important, then, that analogical thinkers and rule-followers
usually take precedents as given even if they disagree with those
precedents as a matter of first principle. The fact that precedents are
fixed points helps to bring about incompletely theorized agreements as
well, by constraining the areas of reasonable disagreement. The result
is a degree of stability and predictability that are important virtues for
law.
We can find many examples in ordinary life. Parents' practices
with their children may not fully cohere. Precedents with respect to
bedtime, eating, homework, and much else are unlikely to be suscepti-
ble to systematization under a single principle. Of course parents do
not seek to be inconsistent - of course a child may justly feel ag-
grieved if his sibling is allowed to watch more hours of television for
no good reason - but full coherence would be a lot to ask. The prob-
lem of reaching full consistency is much more severe in law, where so
many people have decided so many things, and where disagreements
on large principles lurk in the background. We need not say this with
regret.
None of these points suggests that incompletely theorized agree-
ments always deserve celebration. The virtues of such agreements are
partial. If an agreement is more fully theorized, it may provide greater
notice to affected parties, at least if the fuller theorization yields rule-
like judgments about a wide range of cases.-5 Moreover, fuller theori-
zation - in the form of wider and deeper inquiry into the grounds for
legal judgment - may be valuable or even necessary to prevent in-
consistency, bias, or self-interest. We should be wary of judicial out-
S4 The point counts against Dworkin's Hercules metaphor, see DWORKIN, supra note z5, at
239-4o, and also against efforts to find an analogy between legal reasoning and the search for
reflective equilibrium.
SS See Scalia, supra note 53, at 1177-8o; cf. Sunstein, supra note 27 (criticizing extravagant
enthusiasm for rules).

comes based on grounds that have not been stated publicly. If judges
on a panel have actually agreed on a general theory, and if they are
truly committed to it, they should say so (even if, for reasons I have
suggested, they should usually be reluctant to commit). Judges and the
general community will learn much more if they are able to discuss
the true motivating grounds for outcomes. All these are valid consid-
erations, and nothing I am saying here denies their importance.
B. Judges, Theory, and the Rule of Law
There is an association between the effort to attain incompletely
theorized agreements and the rule of law ideal. Insofar as our system
prizes rule by law rather than rule by individual human beings, it tries
to constrain judgments before particular decisions are made. Indeed, a
prime purpose of the rule of law is to rule off limits certain deep ideas
of the right or the good, at least in the sense that those ideas ought not
to be invoked by officials occupying particular social roles, 6 Among
the forbidden or presumptively forbidden ideas are, often, high-level
views that are taken as too hubristic or sectarian for judges precisely
because they are so high-level. The presumption against high-level
theories is an aspect of the ideal of the rule of law to the extent that it
is an effort to limit the exercise of judicial discretion at the point of
application.
In this way, we might begin to make distinctions between the role
of high theory in the courtroom and the role of high theory in the
political branches; these distinctions are central to the claims I am
making here. In democratic arenas, there is no taboo, presumptive or
otherwise, against invoking high-level theories of the good or the
right.5 7 Such theories have played a role in many social movements
with defining effects on American constitutionalism, including the
Civil War, the New Deal, the women's movement, the civil rights
movement, and the environmental movement.5 8 Many of the most ab-
56 See HERZOG, supra note 42, at no0-47 (connecting the law-politics distinction with the
need to distinguish between the sphere of adjudication and the sphere of legislation).
57 I am putting to one side the questions raised by "comprehensive views." See RAWLS, supra
note 4, at 13-14, 175.
S8 See BRUCE ACKERMAN, WE THE PEOPLE 44-80 (1991). It might be responded that there
are many cases in which high principle was invoked judicially - consider, for example, the fairly
dramatic and rapid shift in the treatment of sex equality. See GEOFFREY R. STONE, Louis M.
SEIDMAN, CASS R. SUNSTEiN & MARK V. TUSHNET, CONSTITUTIONAL LAW 676-7i8 (2d ed.
iggi). In cases of this sort, the Court does use principles of a relatively high-level, even if it
thinks analogically. High-level arguments can certainly be found in some important constitutional
cases, including, for example, DeShaney v. Winnebago County Dep't of Social Servs., 489 U.S.
x89 (x989), in which the Court accepted a large-scale view of the state as "passive" rather than
"affirmative." See id. at r95-97. Note, however, that the Court, in invoking theories, is usually
likely to be following democratic trends rather than initiating new ones, and indeed, in the area of
sex discrimination, the Court was building self-consciously on democratic developments. See, e.g.,
Frontiero v. Richardson, 411 U.S. 677, 687-88 (r973). It is notable, too, that the Court was also

1752 HARVARD LAW REVIEW [Vol. 1oS83733
stract arguments of high principle have come from participants in de-

liberative democracy - James Madison, Thomas Jefferson, Abraham
Lincoln, Franklin Delano Roosevelt, Martin Luther King, Jr., and
Rachel Carson. 9 To be sure, incompletely theorized agreements play a
key role in democratic arenas too; consider laws that are supportable
by reference to diverse foundations, such as those protecting endan-
gered species or granting unions a right to organize. But high-level
theories are an indispensable part of democratic politics.
By contrast, use of large-scale theories by courts is usually prob-
lematic and understood as such, within the judiciary (as exemplified
by judicial practice) if not within law schools. The skepticism is partly
a result of the simple fact that judges who invoke large-scale theories
may have to require large-scale social reforms, and courts are usually
ineffective at implementing such reforms on their own.60 An impor-
tant part of the problem lies in the fact that courts, in common law or
constitutional cases, must decide on the legitimacy of rules that are
aspects of complex systems. In invalidating or changing a single rule,
courts may produce unfortunate systemic effects that are not visible to
them at the time of decision and that may be impossible for them to
correct thereafter. 61 Legislatures are in a much better position on this
score.
For those who believe that the Warren Court provides an enduring
model for American constitutionalism, this cautionary note may be in-
adequate; perhaps courts can take important steps in producing justice
where there is now injustice, and perhaps a degree of theoretical ab-
straction is necessary for them to perform this role. Even if this is so,
constitutional law is only a small part of adjudication, not the whole
picture, and in any case, a degree of social reform in the name of the
Constitution might be produced through judgments that are not com-
pletely theorized. 62 Moreover, constitutional courts are unlikely to
want to produce large-scale social change; the mode of selecting
reluctant to develop a theory to account for its new approach to sex equality, but relied instead on
a set of relevant factors. See id. at 682-87. (1 am grateful to Susan Moller Okin for raising this
issue.)
s9 Carson is probably the most contentious example on this list, but her book was in many
ways the foundation of the environmental movement. See RACHEL CARSON, SILENT SPRING
277--97 (1962); Averting a Death Foretold, NEWSWEEK, Nov. 28, 1994, at 72, 72-73.
60 See GERALD N. ROSENBERG, THE HOLLOW HOPE: CAN COURTS BRING ABOUT SOCIAL
CHANGE?, 5-7, 336-38, 342-43 (iggi) (discussing problems in judge-led reform).
61 Examples are offered in R. SHEP MELNICK, REGULATION AND THE COURTS: THE CASE OF
THE CLEAN AIR ACT 110-12, 176-78, 19O-92 (983), and DONALD L. HOROWITZ, THE COURTS
AND SOCIAL POLICY 260-73 (1977). The point is described from the theoretical point of view in
Lon L. Fuller, The Forms and Limits of Adjudication, 92 HARv. L. REV. 353, 393-405 (1978), and
JOSEPH RAz, The Inner Logic of the Law, in ETHICS IN THE PUBLIC DoMAIN, cited above in note
4, at 222, 222-25.
62 See Sunstein, supra note 24, at 767-69, 78o-82 (claiming that analogy and low-level princi-
ples need not be conservative).

judges," and American history itself, suggest that courts will rarely at-
tempt to replicate the Warren Court's role. And judges who use such
theories may well blunder. Judges are not trained as political philoso-
phers, and in many cases, use of abstractions not developed by close
63
reference to particular problems has led to major mistakes.
Even more fundamentally, judges lack a democratic pedigree, and
it is in the absence of such a pedigree that the system of precedent,
analogy, and incompletely theorized agreement has such an important
place. The right to a democratic system is one of the rights to which
people are entitled, and in such a system, judicial invocation of large
theories to support large decisions against democratic processes should
be a rare event.6 4 To be sure, judges have a duty to interpret the
Constitution, and that duty authorizes them to invoke relatively large-
scale principles, seen as part and parcel of the Constitution as demo-
cratically ratified. Many people think that judicial activity is best
characterized by reference to use of such principles, 65 and it would be
wrong to deny that there are occasions on which this practice occurs
and is legitimate.
To identify those occasions, it would be necessary to develop a full
theory of legal interpretation. For present purposes, I urge something
more modest. Most judicial activity does not involve constitutional in-
terpretation, and the ordinary work of common law decision and stat-
utory interpretation calls for low-level principles. Indeed,
constitutional argument is itself based largely on low-level principles,
not on high theory, except on those rare occasions when more ambi-
tious thinking becomes necessary to resolve a case, or when the case
for the ambitious theory is so insistent that a range of judges do and
should converge on it.66
At this point, it should be remarked that incompletely theorized
agreements are not necessarily conservative, and it would therefore be
63 Consider the large-scale reference to the passive state in DeShaney v. Winnebago County
Dep't of Social Servs., 489 U.S. i89, i95-97 (1989), a view that cannot withstand analysis in light
of the role of the state in protecting property rights, contract rights, and rights of plaintiffs in
certain settings. See David P. Currie, Positive and Negative ConstitutionalRights, 53 U. CHI. L.
REv. 864, 872-80, 886-87 (1986); David A. Strauss, Due Process, Government Inaction, and Pri-
vate Wrongs, 1989 Sup. CT. REV. 53, 72-76.
64 Of course the point must be qualified by reference to the democratic failures of existing
majoritarian arrangements. For variations on this theme, see JOHN H. ELY, DEMOCRACY AND
DsmTRUST I05-24 (2980), and CASS R. SUNSTEIN, THE PARTIAL CONSTITUTION 133-53 (1993).
6S This is the vision of judicial review in ACKERMAN, cited above in note 58, at 4, 38-39,
131-32, 284. Note that it differs dramatically from the understanding in DwOmKIN, cited above in
note IS, at 355-99, in the sense that Ackerman insists that the large-scale principles have sources
in actual judgments of "we the people." There is, however, a commonality between Ackerman
and Dworkin in the sense that both see the use of such principles as a large part of the Court's
work. It is along that dimension that I am doubting both of their accounts.
66 See infra p. 1766 (discussing the phenomenon of contested views becoming uncontested and
vice versa).

wrong to identify enthusiasm for such agreements with a belief in

Burkeanism or traditionalism for law. 67 Many incompletely theorized
agreements are highly critical of existing social practices; consider the
judgment that sex discrimination is unacceptable, a judgment that
may rest on diverse theoretical foundations but at the same time lead
to sharp challenges to many current social and legal practices.
Whether incompletely theorized agreements produce approval or dis-
approval of existing practice depends on their content, not on the fact
of their incomplete theorization. 68 Of course, high-level theories need
not be critical of current practices; many such theories are designed to
69
defend them.
C. Full Particularity?
Reason-giving is usually prized in law, as of course it should be.
Without reasons, there is no assurance that decisions are not arbitrary
or irrational, and people will be less able to plan their affairs.
In a few areas of law, however, institutions are permitted to oper-
ate with full particularity. People converge on the result, but they
need offer no reasons for their decision. 70 With full particularity, the
judgment is not merely incompletely theorized; it is not theorized at
all. In denying certiorari, for example, the Supreme Court is silent; in
issuing verdicts, juries do not give reasons; college admissions offices
produce results but rarely justifications. Each decision applies to the
case at hand and to that case alone. 71 Participants may actually have
reasons but refuse to give them; or they may lack reasons in the sense
that they are unable to articulate what accounts for their decision; or
each of them, on a multimember institution, may have reasons but be
unable to agree with one another about them, and hence they leave an
outcome officially unexplained. In any of these cases, this approach
67 A form of Burkeanism is defended in ANTHONY T. KRONMAN, THE LOST LAWYER 19-52

('993). Much of what Kronman says about practical reason is convincing, see id. at 53-1o8, but
one may accept what he says, or much of it, without accepting his form of traditionalism.
68 See Sunstein, supra note 24, at 767-69, 780-82 (discussing conservatism and analogy). Jo-
seph Raz defends analogy as responsive to the possible problems of "partial reform" and hence as
conservative in JOSEPH RAz, Law and Value in Adjudication, in THE AUTHORITY OF LAW 180,
200-06 ('979), but if the judgments of principle that underlie claims of analogy are critical of
existing practice, there may be nothing conservative about analogy. See, e.g., Andrew Koppel-
man, Note, The Miscegenation Analogy: Sodomy Law as Sex Discrimination, 98 YALE L.J. x45,
147 (1988) (arguing that sodomy laws are invalid by analogy to the invalidation of miscegenation
statutes in Loving v. Virginia, 388 U.S. i (1967)).
69 Henry Sidgwick so understood utilitarianism, broadly speaking. See HENRY SIDGWICK,
THE METHODS OF ETHICS 475-77 (Hackett Publishing ig8x) (7th ed. 1907). Richard Posner uses
economics to defend much of the common law. See POSNER, supra note xs, at 23-25, 31-32,
251-64.
70 I draw here on the illuminating discussion of full particularity in Frederick Schauer, Giving
Reasons, 47 STAN. L. Rxv. (forthcoming x995).
71 See id.

offers full particularity because, by their very nature, reasons are more
abstract than the outcomes that they justify. Once offered publicly,
reasons may therefore apply to cases that the court, in justifying a
particular decision, does not have before it.72 This is why reason-giv-
ing promotes planning. Perhaps the most famous (or infamous) illus-
tration of full particularity comes from Justice Stewart:
I shall not today attempt further to define the kinds of material I under-
stand to be embraced within that shorthand description [of hard-core
pornography]; and perhaps I could never succeed in intelligibly doing so.
But I know73it when I see it, and the motion picture involved in this case
is not that.
All well-functioning legal systems value the enterprise of reason-
giving. But there is a good reason to be wary of reason-giving: rea-
sons may be both over- and under-inclusive. In this way, reasons are
like rules, which are also over- and under-inclusive if measured by
reference to their justifications. 7 4 Whenever a court offers reasons,
there is a risk of future regret - not simply because the court may be
confined in a subsequent case and thus have to avoid inconsistency,
but because the reasons offered in case A may turn out, on reflection,
to generate a standard, a principle, or. a rule that collides with the
court's considered judgment about case B. The constraint produced
by the reason may limit discretion and promote predictability, but it
may also produce a bad result.
The distinction between holding and dictum helps reduce this
problem. Indeed, the distinction squarely addresses the problem of ex-
cessively theorized judgments, and it helps to ensure that legal deci-
sions are incompletely theorized. If we understand the holding to be
the narrowest possible basis for the decision, a subsequent court is
able to offer sufficiently narrow reasons for the outcome in the previ-
ous case - that is, reasons that ensure that the outcome in the previ-
ous case does not apply to a case that is genuinely different. In
initially giving a reason, court one may be unaware of possible appli-
cations that will falsify that reason because of its imprecision and ex-
cessive generality. But court two, able to offer some narrower and
better-fitting explanation for the outcome, can eliminate the difficulty.
It can label the excessive generality "dicta."
There is another difficulty with reason-giving. One might some-
times know something without knowing the reasons for it. For exam-
72 Consider John Dewey's rendition of "the old story of the layman who was appointed to a
position in India where he would have to pass in his official capacity on various matters in
controversies between natives. Upon consulting a legal friend, he was told to use his common-
sense and announce his decisions firmly; in the majority of cases his natural decision as to what
was fair and reasonable would suffice. But, his friend added: 'Never try to give reasons, for they
will usually be wrong.'" John Dewey, Logical Method and Law, io CORNELL L.Q. 17, 17 (1924).
73 Jacobellis v. Ohio, 378 U.S. 184, r97 (1964) (Stewart, J., concurring).
74 See SCHAUER, supra note 27, at 31-34.

pie, one may know that this is Martin's face, and no other face,
without knowing, exactly, why one knows that fact. Or one might
know that a certain act would be wrong, without knowing, exactly,
why it would be wrong. It is certainly possible to know that some-
thing is true without knowing why it is true.
Is there an analogy in law? No simple answer would make sense.
As I have suggested, a special quality of most legal systems is a pre-
sumptive requirement of reasons for legal outcomes. This requirement
makes it hard to prize a capacity to know what the law is without
knowing why it is as it is, or how a case should come out without
knowing why it should come out that way. On the other hand, some
people think that many of our judgments about similarity and dissimi-
larity - a key to legal reasoning - do not rest on reasons but instead
simply constitute our descriptive and normative worlds. 75 Conceiva-
bly, there is an aspect of socialization into law that enables people to
see that case A is "like" case B, and not at all "like" case C, without
always having much of an account of why this is so.
More concretely, it is possible that experienced judges, like exper-
ienced lawyers, develop a faculty - best described as wisdom, percep-
tion, or judgment - that allows them to reach decisions very well and
very quickly. This is a distinctive faculty. It seems to be associated
with the ready and sympathetic apprehension of a wide range of di-
verse particulars and with an appreciation of the appropriate weight
to be given to each.7 6 Certainly, we can imagine a class of people who
have a wonderful capacity to tell whether one case is relevantly like
another, or to decide who should win cases, but who lack much of a
capacity to explain what underlies their ultimate judgments, or their
convictions about relevant similarity and difference. They are not the-
orists at all. But they have a "good ear," unlike some others who have
a "tin ear" for law.
Perhaps it would ultimately be possible for such people (or at least
outside observers) to explain what underlies these good apprehensions,
but this may not be so. It is important not to mystify these issues.
What is asserted to be a capacity for perception may in fact be a
product of bias or confusion, and reason-giving helps diminish this
risk. The faculty of wisdom, perception, or judgment probably
75 See Mark Johnson, Some Constraintson Embodied Analogical Understanding,in ANALOGI.

CAL REASONING 25, 39 (David H. Helman ed., 1988) (C[A]nalogies cannot be understood as pro-
positional or conceptual mechanisms for reflecting on already-determinate experiences; rather, we
can actually speak of them as constitutive of our experience, because they are partially constitu-
tive of our understanding, our mode of experiencing our world. Analogy is a basic means by
which form, pattern, and connection emerge in our understanding and are then articulated in our
reflective cognition and in our language.").
76 See KRONMAN, supra note 67, at 53-62 (describing this skill as "statesmanship"); see also
JON ELSTER, SOUR GRAPES: STUDIES IN THE SuBvERsION OF RATIONALITY 16-17 (I983) (discuss-
ing the existence and importance of this skill in professions other than law).

amounts to a capacity to think very quickly of a resolution that takes
account of everything that matters, including a wide array of compet-
ing considerations, and that coheres well with the rest of our particu-
lar and general judgments. Compare this striking description of
President Franklin Delano Roosevelt:
Frances Perkins later described the President's idea . . .as a "flash of
almost clairvoyant knowledge and understanding." He would have one
of these flashes every now and then, she observed, much like those that
musicians get when "they see or hear the structure of an entire symphony
or opera." He couldn't always hold on to it or verbalize it, but when it
came, he suddenly understood how all kinds of disparate things fit to-
gether.... Roosevelt made up for the defects of an undisciplined mind
with a profound ability to integrate a vast multitude of details into7 a
larger pattern that gave shape and direction to the stream of events.
When someone is thought to be wise about particulars in law, it is
because she is able to see how to resolve a case without doing violence
to other judgments at the horizontal and vertical levels. She is thus
able to decide cases, and to decide who wins and who loses, while
minimally endangering other valued goods and goals. She may not
be
able to explain what underlies her decisions (though we should hope
that she can). Something of this sort underlies the conviction that a
person is "judicious," and it helps explain why full-scale theorization is
so rare in law. Of course, someone generally described as judicious
might be wrong much of the time, and more ambitious theory might
be introduced to show where the errors lie.
I. HERCULES AND THEORY

A. An Ambitious Alternative
Enthusiasm for incompletely theorized agreements meets with
many adversaries. An especially distinguished example is Ronald
Dworkin, who urges, at least as an ideal, a high degree of theoretical
self-consciousness in adjudication. 78 Dworkin argues that when law-
yers disagree about what the law is, they are disagreeing about "the
best constructive interpretation of the community's legal practice."7 9
Dworkin claims that interpretation in law consists of different efforts
to make a governing text "the best it can be."80 This is Dworkin's
conception of law as integrity.
Dworkin's account appears to require judges to develop high-level
theories, and it does not (to say the least) favor theoretical modesty.
77 Doms K. GOODwiN, No ORDINARY TIME 193 (i994).

78 1 discuss some of these points in Sunstein, cited above in note 24, at 784-87, and I draw on
that discussion here. My current goal is to generalize some of the points made in a very narrow
context.
79 DwoR!iN, supra note 15, at 225.
80 Id. at 229.

1758 HARVARD LAW REVIEW [Vol. io8:i733
There is no presumption against abstraction. In his hands, the rele-

vant theories are large and abstract; they sound like political philoso-
phy or moral theory.81 These theories are derived from or brought to
bear on particular problems. But this is not how real lawyers and real
judges proceed. If they can, they avoid broad and abstract questions.
Let us distinguish here among four situations: the individual judge
thinking through a case on his own and sitting on a single-judge dis-
trict court; the individual judge writing for a single-judge district
court; the individual judge thinking through a case privately but also
sitting on a multimember court; and the individual judge writing or
participating in an opinion to be joined by members of a multimember
institution. The constraints in the four situations are different. In all
of them, the judge would probably prefer to decide a case knowing
that the outcome and supporting rationale can be justified by reference
to a range of diverse and more general theories. If so, it is unneces-
sary for the judge to choose among them; if so, the judge may put
aside the more theoretical debates, knowing, for example, that whether
he is a Kantian or a Benthamite - whether he cares about fairness or
deterrence - someone who has assaulted another must pay compensa-
tion for the wrong done.
Now it may be that some individual judges would like to think
things through very deeply in order to be sure that they are right.
There is certainly nothing wrong with an aspiration of this sort,
though limits of time and capacity may make the aspiration unrealis-
tic, and the system of precedent will complicate matters. But even the
individual judge sitting on a single-judge court may be reluctant to set
forth his preferred abstractions in public - on the theory that the
judicial role calls for the avoidance of dicta, understood here as state-
ments about the law that are not necessary to the decision. Perhaps
the outcome can be justified on narrower grounds that involve more
modest and more reliable principles and that are less likely to create
problems for unforeseeable future cases.
The point is even clearer for the participant on a multimember
panel. To some extent, the avoidance of more abstract or theoretical
claims stems from purely strategic or pragmatic considerations, as
81 See, e.g., id. at 403-04 (discussing tort law and equal protection). Dworkin writes about
constitutional interpretation:
The Constitution is foundational of other law, so Hercules' interpretation of the document
as a whole, and of its abstract clauses, must be foundational as well .... [lit must be a
justification drawn from the most philosophical reaches ot political theory .... In constitu-
tional theory philosophy is closer to the surface of the argument and, if the theory is good,
explicit in it.
Id. at 380. These are points that I am denying here. In fact, Dworkin's account of the judicial
role says very little about the various fact-finding and theory-building weaknesses of judges, who
are after all unelected, and about the judges' own knowledge of their weaknesses, even though
much of judge-made law is based on that knowledge. This seems to me a major gap in Dwor-
kin's view.

when individual judges on a multimember body have a highly abstract

theory but are unable to persuade their colleagues to accept it.82 But
often, the avoidance is not strategic in this way but stems instead from
the fact that highly abstract questions can be too hard, large, and
open-ended for legal actors to handle. Members of a multimember
panel may especially fear that any responses will go wrong when
tested against concrete cases not before the court, will be too sectarian
and hubristic, and will take too long to offer, because they are too
deeply theorized. On multimember bodies, efforts to answer such ab-
stract questions can prevent people who disagree on large principles
from reaching consensus on particular outcomes when, in fact, those
outcomes, and low-level explanations, could command agreement from
a variety of perspectives.
Because of his enthusiasm for abstractions, Hercules - Dworkin's
patient and resourceful judge - could not really participate in ordi-
nary judicial deliberations. He would probably be seen as a usurper,
even an oddball. On a single judge court, he would suffer from the
vice of hubris. On a multimember panel, he would lack some of the
crucial virtues of a participant in legal deliberation. These virtues in-
clude collegiality and civility, which incline judges
3
toward the lowest
level of abstraction necessary to decide a case.
Dworkin anticipates an objection of this kind. He notes that it
might be paralyzing for judges to seek a general theory for each area
of law, and he acknowledges that Hercules is more methodical than
any real-world judge can be.84 But Hercules, in Dworkin's view,
"shows us the hidden structure of [ordinary] judgments and so lays
these open to study and criticism." 5 Of course, Hercules aims at a
"comprehensive theory" of each area of law, while ordinary judges, un-
able to consider all lines of inquiry, must seek a theory that is "par-
82 If the judge has thought the relevant issues through in an ambitious way and made the
moral judgments that are raised by the issue (to the extent that they are legally relevant), we may
have more assurance that the process of adjudication is reliable, in the sense that it is likely to
lead to results that are just. What I am emphasizing is that the judge might sign an opinion that
contains a more modest account.
83 I should say that I do not know to what extent this counts as a criticism of Dworkin, since
Dworkin presents Hercules as a thought experiment, and it is unclear to what extent Hercules'
approach is supposed to be an ideal for real-world judges. To my knowledge, Dworkin has not
discussed the question of appropriate levels of abstraction in legal justification. But see supra
note 81 (noting Dworkin's suggestion that, in constitutional law, the relevant justification is
"drawn from the most philosophical reaches of political theory").
84 See DwomaN, supra note i5, at 264-65; see also id. at 380-81 (referring to the need "to
gain the votes of other justices and to make their joint decision sufficiently acceptable to the
community so that it can continue to act in the spirit of a community of principle at the constitu-
tional level").
85 Id. at 265.

176o HARVARD LAW REVIEW [Vol. 108:1733
tial." But Hercules' "judgments of fit and political morality are made
on the same material and have the same character as theirs."8 6
It is these points that I am denying here. The decisions of ordinary
judges are based on different material and have a different character.
They are less deeply theorized not only because of limits of time and
capacity, but also because of the distinctive morality of judging in a
pluralistic society. I will qualify this claim below.8 7 But for the mo-
ment, I suggest that the ordinary judge is no Hercules with less time
on his hands, but a different sort of figure altogether.
B. Conceptual Ascent?
Borrowing from Henry Sidgwick's suggestions about method in
ethics, 8 8 advocates of ambitious thinking might respond in the follow-
ing way. There is often good reason for judges to raise the level of
abstraction and ultimately to resort to large-scale theory. As a practi-
cal matter, discrete judgments about particular cases will often prove
inadequate.8 9 Sometimes, people do not have clear intuitions about
how cases should come out. Sometimes, seemingly similar cases pro-
voke different reactions, and it is necessary to raise the level of theo-
retical ambition to explain whether those different reactions are
justified or to show that the seemingly similar cases are different after
all. Sometimes, different people simply disagree. By looking at
broader principles, we may be able to mediate the disagreement. In
any case, there is a problem of explaining our considered judgments
about particular cases - to make sure that they are not just an acci-
dent9 O - and at some point, the law may well want to offer that
explanation.
Ambitious thinkers might therefore urge that low-level principles
may conflict with one another or be demonstrably wrong. In these
circumstances, judges might well resort to higher theory. When our
modest judge joins an opinion that is incompletely theorized, he must
rely on a reason or a principle that justifies one outcome rather than
another. Perhaps the principle is wrong because it fails to fit with
other cases, or because it is not defensible as a matter of political mo-
rality.9 1 A distinguished judge will seek to add a good deal in the way
of both width and depth by exploring other cases and by deepening
the theoretical ambition of his analysis. He will therefore experience a
kind of conceptual ascent in which the more or less isolated and small
86 Id.
87 See infra pp. 1765-66.
88 See SIDGWICK, supra note 69, at 96-zo4.
89 See id. at oo.
90 "[T]he resulting code seems an accidental aggregate of precepts, which stands in need of
some rational synthesis." Id. at 102.
91 See DwoRKIN, supra note is, at 251-54 (discussing local priority).

low-level principle is finally made part of a more general theory. Per-

haps this would be a paralyzing task, and perhaps our judge need not
often attempt it. But it is an appropriate model for understanding law
and an appropriate aspiration for evaluating legal practice.
The conceptual ascent is especially desirable in light of the likeli-
hood that incompletely theorized agreements will contain large pockets
of injustice and inconsistency. Unambitious principles and high levels
of particularity may ensure that the similarly situated are not treated
similarly. Or perhaps some areas of the law will make internal sense,
but because the categories with sense are small, they may run into
each other if they are compared. We may have a coherent category of
law involving sex equality and a coherent category involving racial
equality, but these categories may have a strange relation to each
other or to the categories involving sexual orientation and the handi-
capped. Various subcategories of tort law may make sense, but they
may not fit together, or they may not cohere with the law of contract.
We might conclude that judges should think of incompletely theorized
agreements as an early step toward something both wider and deeper.
Many academic understandings of law undertake the task of develop-
92
ing that wider and deeper conception.
There is some truth in this response. Very probably, moral reason-
ers should try to achieve vertical and horizontal consistency, not just
the local pockets of coherence offered by incompletely theorized agree-
ments. In democratic processes, it is appropriate and sometimes indis-
pensable to challenge existing practice in abstract terms. But the
response ignores some of the distinctive characteristics of the arena in
which real-world judges must do their work. Some of these limits in-
volve bounded rationality and thus what should happen in a world in
which people face various constraints; others involve appropriate judi-
cial (and in some ways political) morality and mutual interaction in a
world in which judges are mere actors in a complex system and in
which people legitimately disagree on first principles. In light of these
limits, incompletely theorized agreements have the many virtues de-
scribed above, including the facilitation of convergence, the reduction
of costs of disagreement,9 3 and the demonstration of humility and mu-
tual respect. As I have noted, incompletely theorized agreements are
especially well adapted to a system that must take precedents as fixed
points; lawyers could not try to reach full integrity without severely
compromising the system of precedent. Usually, local coherence is the
most to which lawyers may aspire.9 4 Just as legislation cannot be un-
92 See, e.g., DWORXIN, supra note 15, at 276-312 (discussing the law of tort); POSNER, supra
note i, passim.
93 See AMY GuTmANN & DENNIS THOMPSON, DEMOCRACY AND DISAGREEMENT (forthcom-
ing 1995).
94 See RAZ, Relevance of Coherence, supra note 49, at 298-303.

derstood as if it came from a single mind, so precedents, compiled by

many people responding to different problems in many different peri-
ods, will not reflect a single authorial voice.
There are many lurking questions here. What are the criteria for
evaluating moral or political judgments? What is the relation between
provisional or considered judgments about particulars and correspond-
ing judgments about abstractions?95 Some people write as if abstract
theoretical judgments, or abstract theories, have a kind of reality and
hardness that particular judgments lack, or as if abstract theories pro-
vide the answers to examination questions that particular judgments,
tested against the theories, may pass or fail. On this view, theories
may be treated as searchlights that illuminate particular judgments
and show them for what they really are.9 6 But we might think instead
that there is no special magic in theories or abstractions, and that the-
ories are simply the (humanly constructed) means by which people
make sense of their ethical and political worlds. The abstract deserves
no priority over the particular: neither is foundational; neither is
harder or more real. A (poor or crude) abstract theory may simply be
a confused way of trying to make sense of our considered judgments
about particular cases, which may be much better than the theory. Of
course, particular judgments may be confused too. 97
C. Legitimacy
Dworkin's conception of law as integrity offers a theory of what it
means for law to be legitimate. Hercules can produce vertical and
horizontal consistency among judgments of principle in law. Do in-
completely theorized agreements comport with any plausible concep-
tion of legitimacy? We have seen that a legal system pervaded by
such agreements need not yield anything like full coherence. Perhaps
this is a decisive defect.
A complete response would require a detailed discussion of one of
the largest issues of political morality; for the moment, a few brief
remarks must suffice. In fact, the idea of integrity, insofar as it is a
judicial product, is unlikely to provide a convincing theory of legiti-
macy. Indeed, it seems plausible to say that integrity, as Dworkin de-
95 In Rawls' understanding of the search for reflective equilibrium, we consult "our considered
convictions at all levels of generality; no one level, say that of abstract principle or that of partic-
ular judgments in particular cases, is viewed as foundational. They all may have an initial credi-
bility." RAwLS, supra note 4, at 8 n.8.
96 See, e.g., BRUCE A. ACKERMAN, PRIVATE PROPERTY AND THE CONSTITUTION 15-20 (1977)
(comparing "Scientific Policymakers" with "Ordinary Observers"); POSNER, supra note x, at
23-24 (arguing that economic theory often underlies opinions that appear to be based on other
grounds).
97 For illuminating remarks on abstractions, see RAWLS, cited above in note 4, at 44-46.
Rawls thinks that we turn to abstractions "when we are torn within ourselves." Id. at 44. This is
undoubtedly true, but sometimes we turn instead to particulars when we are so torn.

1995] INCOMPLETELY THEORIZED AGREEMENTS . 1763
scribes it, is neither necessary nor sufficient for legitimacy, judicial or

otherwise. Legitimacy stems not simply from principled consistency on
the part of adjudicators (or someone else) but from a justifiable exer-
cise of public force. That theory should be founded in a theory of
authority9" and hence (if we are democrats) in suitably constrained
democratic considerations. In this light, Dworkin's conception of in-
tegrity offers an excessively court-centered conception of legitimacy. It
sees legitimacy not as an exercise of legitimate authority, or as an out-
come of well-functioning democratic procedures, but instead as a pro-
cess of distinction-making undertaken by judges. 9 9 Even if done
exceptionally well, distinction-making by principled judges is an inade-
quate source of legitimacy.
Of course, principled consistency should not be disparaged, and a
regime of principle, as Dworkin understands it, has many advantages
over imaginable alternatives. Problems of legitimacy may arise pre-
cisely because of the absence of such consistency. But a theory of ad-
judicative legitimacy should be part of a theory of just institutions,
and an approach focused solely or principally on adjudication will not
provide any such theory. By contrast, those who stress incompletely
theorized agreements insist that adjudication is part of a complex set
of authoritative institutional arrangements, most prominently including
democratic arenas. They attempt to design their theory of judicial
conduct as an aspect of a far broader set of understandings about ap-
propriate institutional arrangements and about institutions in which
the (suitably constrained) public can deliberate about its judgments.
Above all, they see the American system as one of deliberative de-
mocracy in which judges play a partial role. 100 For reasons of both
policy and principle, the development of large-scale theories of the
right and the good is most fundamentally a democratic task, not a
judicial one. 101 These are brief and inadequate remarks, but they tend
to suggest the ingredients of an account of legitimacy of which incom-
pletely theorized agreements would be a part.
Full integrity, moreover, consists of much more than a legal system
of numerous, hierarchically arranged courts can be expected to offer.
Judges generally should and will reason from cases with which they
98 See RAZ, supra note 23, at 21-69 (offering a theory of authority).

99 There is some ambiguity in Dworkin's treatment, since integrity is a political virtue, not
simply a legal one. I am not sure of the intended relation between political integrity and integrity
in law. See DwoIuaN, supra note 15, at 186-219. Hence I am not sure of Dworkin's views on
the general question of political legitimacy.
100 See, e.g., JOSEPH M. BEssETTE, THE MILD VOICE OF REASON: DELIBERATIVE DEMoc-
RACY AND AMERICAN NATIONAL GOVERNMENT 6-17 (1994) (discussing political deliberation and
the separation of powers).
101 ACKERMAN, cited above in note 58, at 86-99, takes the same basic position but sees the
judicial role as the "synthesis" of constitutional moments. I am rejecting the latter claim and
saying that, with some exceptions, courts do and should avoid large-scale synthesis.

1764 * HARVARD LAW REVIEW [Vol. io8:1733
disagree - because of the need for predictability and stability in law,

because of the need to avoid judicial hubris, and because similarly sit-
uated people should be treated similarly. 0 2 But across the broad ex-
panse of the law, the resulting judgments are unlikely fully to cohere.
IV. INCOMPLETELY THEORIZED AGREEMENTS OVER TIME

Incompletely theorized agreements have virtues, but as I have
noted, their virtues are partial. Stability, for example, is brought
about by such agreements, and stability is usually desirable, but a sys-
tem that is stable and unjust should probably be made less stable.
Agreement is important, but disagreement is important too, for it can
be a creative force in revealing error and injustice. In this final sec-
tion, I offer some qualifications to what has been said thus far. In
brief: some cases cannot be decided at all without introducing a fair
amount in the way of theoretical ambition. Moreover, some cases can-
not be decided well without increasing the level of theoretical ambi-
tion. If a good abstraction is available, and if diverse judges can be
persuaded that the abstraction is good, there should be no taboo
against its judicial acceptance. The claims on behalf of incompletely
theorized agreements are presumptive rather than conclusive.
A. Change
Thus far, the discussion has offered a static description of the legal
process - a description in which judges are deciding what to do at a
certain time. Of course, low-level principles are developed over long
periods, and a dynamic picture shows something different and more
complex. First, the understanding of these low-level principles may
shift and perhaps deepen. At one point, libel of a public figure might
seem analogous to libel of a private figure and therefore be unpro-
tected; at another point, it might seem more analogous to political
speech and therefore be protected. In this process, courts may move to
higher levels of abstraction.
102 A challenge to this view can be found in Joseph Raz's illuminating discussion of coherence,
but in my view, Raz concludes too quickly that "speaking with one voice" is not an independent
ideal, and I think that he rejects too quickly the view that judges should reason from cases with
which they disagree as a matter of morality. See RAZ, Relevance of Coherence, supra note 49, at
297-98. Raz emphasizes the problems created when judges reason from cases that embody bad
moral judgments. The answer to Raz's claim is institutional. Judges should reason from previ-
ously decided cases not because it is good to sacrifice good moral values, but because the judge's
conception of good moral values is not always reliable, and because keeping (local) faith with
precedent is a way of disciplining the judges, promoting predictability, and ensuring a form of
equality. Of course, the extent to which a particular judge should keep faith with a past decision
(which she believes wrong) cannot be decided in the abstract What I am suggesting is that
decisions that the judge believes wrong may not in fact be wrong, and that the obligation to
reason from past cases makes best sense if it is seen as questioning the (ordinary) judge's ability
to make this judgment well.

Second, a characteristic role of observers of the legal process is to
try to systematize cases in order to see how to make best sense of
them, to show that there is bias and inconsistency, or to show that no
sense can be made of them at all. In any process of systematization,
more abstraction will be introduced. Observers may try to invoke
some higher-level idea of the good or the right in order to show the
deep structure of the case law, or to move it in particular directions,
or to reveal important, even fatal inconsistencies. A demonstration
that the law makes deep sense might be a source of comfort and occa-
sional reform. A demonstration that the law makes no sense, or re-
flects an ad hoc compromise among competing principles, might
produce discomfort and large-scale change.
Third, the law sometimes reflects more ambitious thinking on the
part of judges or reacts to these more ambitious efforts by outsiders.
The law of antitrust, for example, is now based in large part on a
principle of economic efficiency. 10 3 This development occurred
through the gradual incorporation of economic thinking into the cases,
beginning with the judicial suggestion, following academic observa-
tion, that some important cases "actually" or "implicitly" were founded
on economics, and continuing until economics appeared to offer a large
ordering role.0 4 Some especially ambitious or creative judges will in-
voke theories too. Some of our great judges were principally nonthe-
oretical thinkers - Harlan, Friendly, Cardozo - but some of them -
Holmes, Brandeis, Marshall - had at least ingredients of a large-scale
vision of the legal order. They used analogies, to be sure, but often
with reference to a relatively high-level theory about some area of law.
Many areas of law now show the influence of Holmes, Brandeis, and
Marshall, in part because courts, whether or not they deploy low-level
principles, have adopted aspects of the relevant theory. 05
It is very rare for any area of law to be highly theorized. Most of
the time, people starting from divergent points can accept (or reject)
relevant outcomes. But I do not mean to say that it would always be
bad for theories to be introduced, nor do I deny that small-scale, low-
level principles can become part of something more ambitious. A de-
scriptive point first: it is important to note that, after a period of time,
the use of low-level principles may well result in a more completely
theorized system of law. The process of low-level judging can yield
greater abstraction, or a highly refined and coherent set of principles.
103See, e.g., Continental T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36, 56-59 (1977).
104 See, e.g., James C. Miller III & Paul Pauter, Predation:The Changing View in Ecofiomics
and the Law, 28 J. LAW & ECON. 495, 495-502 (1985) (diicussing the prominent role that eco-
nomic analysis and scholarly research have played in shaping antitrust predation law).
105 The best example may be the law of free speech. See Whitney v. California, 274 U.S. 357,
372, 375-78 (1927) (Brandeis, J., concurring); Abrams v. United States, 250 U.S. 616, 624, 630-31
(1919) (Holmes, J., dissenting). Note, however, that even here, no unitary theory of free speech is
offered, and nothing very much like a philosophical account appears in the relevant opinions.

1766 HARVARD LAW REVIEW [Vol. 108:X733
In the areas of free speech and discrimination, some such process has
occurred, resulting in occasionally ambitious claims, even if it would
be far too much to say that full theorization or coherence can be
10 6
found.
An especially interesting phenomenon occurs when a once-contesta-
ble analogy becomes part of the uncontested background for ordinary
legal work - or when the uncontested background is drawn into
sharp question, sometimes via analogies. Thus, for example, the view
that bans on racial intermarriage are "like" segregation laws is now
taken largely for granted; it is part of the way that lawyers order their
conceptual world. So too, perhaps, with the view that sex discrimina-
tion is "like" race discrimination - a view that would have been un-
thinkable in Supreme Court opinions as late as, say, i965. Ordinarily,
the slippage from the uncontested to the contested occurs in law
through encounters with particular cases that reveal gaps or problems
with the conventional view.1 0 7 In American law, views that once were
taken as natural - or not even as views at all - sometimes become
dislodged in this way. The original attack on the monarchical legacy
took this form;108 so too did the attack on racial hierarchy; so too did
the New Deal, which depended on an insistence that common law cat-
egories were far from natural and prepolitical but instead were a con-
scious social choice.' 0 9 Eventually, the contested can become
uncontroversial as new categories emerge and harden through repeated
encounters with particular cases.
Now turn to the question of what judges should do. I have urged
that there are reasons for judges to offer the least ambitious argument
necessary to resolve cases, in the hope that different people from their
diverse standpoints can converge on that argument, and with the be-
lief that abstractions may prove troublesome for later cases. But if
judges can agree on an abstraction, and if the abstraction can be
shown to be a good one, judicial acceptance of that abstraction may
hardly be troubling but, on the contrary, an occasion for celebra-
106 See, e.g., City of Cleburne v. Cleburne Living Ctr., Inc., 473 U.S. 432, 439-47 (xg85) (fail-
ing to offer a full theory of equality); New York Times Co. v. Sullivan, 376 U.S. 254, 27o-83
(1964) (failing to offer a full theory of free speech).
107 See, e.g., Reed v. Reed, 404 U.S. 71, 76-77 (1971) (challenging preexisting acceptance of sex
discrimination); New York imes, 376 U.S. at 268-69 (finding libel law subject to the First
Amendment); Brown v. Board of Educ., 347 U.S. 483, 495 (1954) (overruling Plessy v. Ferguson,
163 U.S. 537 (I896)).
10 Gordon Wood's description of this development traces the shift from the earlier view: "So
distinctive and so separated was the aristocracy from ordinary folk that many still thought the
two groups represented two orders of being. . . . Ordinary people were thought to be different
physically, and because of varying diets and living conditions, no doubt in many cases they were
different. People often assumed that a handsome child, though apparently a commoner, had to be
some gentleman's bastard offspring." GORDON S. 'WOOD, THE RADICALISm[ OF THE AMERICAN
REVOLUTION 27 (1992).
109 See SUNSTEIN, sup'a note 64, at 4o-67.

tion." 0 Who could object to judicial adoption of what is by hypothe-

sis a good theory? If agreement is possible on a good abstraction, a
legal system will have done its job especially well; consider, as pos-
sibilities, antitrust or equal protection law. But any abstractions will
likely have been developed through the generalization and clarification
of incompletely theorized outcomes, accomplished by constant refer-
ence to concrete cases against which the theory is measured. At least
part of the test of the theory - if it is a theory of law meant for
judicial adoption - is how well it accounts for the cases and for con-
sidered judgments about the cases, though of course judicial mistakes
are possible, and these may be corrected by the theory, subject to the
constraints of stare decisis.
I am thus declining to endorse what might be called a strong ver-
sion of the argument offered here: a claim that incompletely theorized
agreements are always the appropriate approach to law and that more
ambitious theory is always illegitimate in law. What I am urging is a
more modest point, keyed to the institutional characteristics of judges
in any legal system we are likely to have. Judges should adopt a pre-
sumption rather than a taboo against high-level theorization. Usually,
they should invoke the lowest levels of theoretical ambition necessary
to decide the case. In many contexts, they will not be able to think of
a good theory. In many cases, they will not be able to- agree on any
theory. The effort to reach agreement on relative abstractions may
make it hard for judges or other people to live and work together, and
unnecessary contests over theory can suggest an absence of respect for
the deepest and most defining commitments of other people."'
It may even happen that an area of judge-made law will become
less theorized over time. A once-acceptable general theory may come
to seem inadequate, and confrontations with particular cases may
show its inadequacy and make it unravel. The Supreme Court's deci-
sions in the Lochner era were relatively well-theorized, in the sense
that they were founded on a recognizable general theory of the permis-
sible role of the state." 2 The general theory was not replaced with a
new one all of a sudden, or even at all; instead, it came apart through
particular cases that attacked the periphery and then the core." 3 This
is a familiar phenomenon, as the process of case-by-case decision tests
110 Consider, for example, Alexander Meiklejohn's suggestion that New York 7imes Co. v. Sul-
livan was "an occasion for dancing in the streets," as reported in Harry Kalven, Jr., The New
York Times Case: A Note on "The Central Meaning of the First Amendment," 1964 SuP. CT.REV.
191, 221 n.125.
IIIThere are intrapersonal parallels. In our ethical lives as individuals, each of us may avoid
choice among theories if we do not need to choose in order to decide what to do in particular
cases. But the interpersonal case is perhaps more vivid.
112 See STONE, SEIDMAN, SUNSTEIN & TUsHNET, supra note 58, at 787-807.
113 See West Coast Hotel Co. v. Parrish, 300 U.S. 379, 391-93 (1937); Nebbia v. New York,
291 U.S. 502, 535-39 (1934).

1768 HARVARD LAW REVIEW [Vol. 1o083733
any general theory and exposes its limits. Over time, an area of law
may become more theorized or less so; over long periods of time, it
may go from one to the other, and back again.
If all this is right, we ought not to think of incompletely theorized
agreements on particulars as a kind of unfortunate second-best,
adopted for a world in which people disagree, are confused or biased,
and have limited time. The alleged first-best - Hercules, or the (ex-
hausted?) judge who has reached reflective equilibrium - calls for an
extra-human conception of law. It is extra-human because it is so ob-
viously unsuited to the real world. To say the least, it is hard to know
whether a top-down or highly theorized approach is appropriate for
morality. But often, at least, it is easy to know that such an approach
is inappropriate for law. The institutional features of the legal system
- a human institution with distinctive constraints - require an ac-
count of law that is highly sensitive to the characteristics of the system
in which it is situated. Among those characteristics are confusion or
uncertainty about general theory, deep disputes about the right and
the good, and a pressing need to make a wide range of particular
decisions.
B. Disagreement
What of disagreement? The discussion thus far has focused mostly
on the need to obtain agreement or convergence. This is only part of
the picture. In law, as in politics, disagreement can be a productive
and creative force - revealing error, showing gaps, moving discussion
in appropriate directions. The American political order has placed a
high premium on "government by discussion,"" 4 and when the process
is working well, this is true for the judiciary as well as for other insti-
tutions. Progress in politics and even law is often fueled by failures of
convergence and by sharp disagreement on both the particular and the
general.
It should be unnecessary to emphasize that legal disagreements
may have many legitimate sources. Two of these sources are especially
important. First, people may share general commitments but disagree
on particular outcomes. This is no less pervasive a social phenomenon
than its converse, which I have stressed here. People may think, for
example, that it is wrong to take innocent life, but disagree about
whether the Constitution protects the right to have an abortion. They
may think that everyone has a right to be free from invidious discrimi-
nation, but disagree about the legitimacy of affirmative action. It is
common to find agreement on the general alongside disagreement on
the particular.
114 SAMUEL H. BEER, To MAKE A NATION: THE REDISCOVERY OF AMERICAN FEDERALISM 74

(1993).

Second, people's disagreements on general principles may produce

disagreements over low-level propositions as well. People who think
that an autonomy principle accounts for freedom of speech may also
think that the government cannot regulate truthful, nondeceptive com-
mercial advertising - whereas people who think that freedom of
speech is a democratic idea may have no interest in commercial adver-
tising at all. People who think that tort law is founded on Kantian
understandings may think that strict liability is required, whereas utili-
tarians may disagree. Academic theorizing about law can have a salu-
tary function in part because it tests low-level principles by reference
to more ambitious claims. Disagreements can be productive by virtue
of this process of testing.
It is important, then, to emphasize that disagreements are some-
times desirable and that incompletely theorized agreements may be
nothing to celebrate. If everyone having a reasonable general view
converges on a particular (by hypothesis reasonable) judgment, nothing
is amiss. This is the basic phenomenon to which I am calling atten-
tion here. But if an agreement is incompletely theorized, there is a
risk that everyone who participates in the agreement is mistaken, and
hence that the outcome is mistaken too. There is also a risk that
someone who is reasonable has not participated and that if that person
were included, the agreement would break down. Incompletely theo-
rized agreements should therefore be subject to scrutiny over time.
Nor is social consensus, or something approaching consensus, a consid-
eration that outweighs everything else. Usually, it would be much bet-
ter to have a just outcome, rejected by many people, than an unjust
outcome with which most or even almost all agree. Consensus or
agreement is important largely because of its connection with stability,
itself a valuable but far from overriding social goal.
Of course, it would be foolish to say that no general theory can
produce agreement, even more foolish to deny that some general theo-
ries deserve general support, and most foolish of all to say that incom-
pletely theorized agreements warrant respect whatever their content.
What seems plausible is something more modest: except in unusual
situations, and for multiple reasons, general theories are an unlikely
foundation for judge-made law, and caution and humility about gen-
eral theory are appropriate for judges, at least when multiple theories
can lead in the same direction. This more modest set of claims helps
us to see that incompletely theorized agreements are important phe-
nomena with their own special virtues. Such agreements are the law-
yer's distinctive solution to the problem of social pluralism.
C. Principle, Politics, Law
Henry Hart and Albert Sacks of course defended the view, influen-
tial for the last generation, that courts engage in principled reasoning

and elaboration of basic social commitments, while the political pro-

cess involves an ad hoc set of judgments often producing unprincipled
compromises.' 1 5 This view finds its most extreme statement in Dwor-
kin's Hercules metaphor, which sees constitutional courts as "the fo-
rum of principle." 1 16 But as description, this view is historically
myopic. Courts do offer reasons and principles, but those principles
tend to be modest and low-level. By contrast, high principle has had
its most important and most defining moments inside the political
branches of government, not within courtrooms. The founding, the
Civil War, the New Deal, and other periods of social reform" 17 - pro-
gressivism, the civil rights movement, the women's movement - sug-
gest that the key forum of principle in American government has been
democratic rather than adjudicative.
The question might therefore be asked whether it would be possi-
ble to reverse the Hart and Sacks formulation, and to propose instead
that high principle plays an appropriately large role in the democratic
arena while low-level principles are the more appropriate and usual
stuff of adjudication. Any such view would be too simple, but there is
a good deal of truth to it. Of course, high principle does not charac-
terize ordinary politics, but high principle has had an important defin-
ing role in American political life, and in any case, the principles that
mark American constitutionalism owe their origins to political rather
than legal developments. There are important exceptions, when courts
make or refer to arguments of high principle,"" but ordinarily those
exceptions consist of vindications of (certain readings of) constitutional
judgments made by previous generations. As I have suggested, the ar-
gument on behalf of incompletely theorized agreements is therefore
part of a theory of authority and of just institutions, rooted in a claim
that fundamental principles are best developed politically rather than
judicially.
I's See HENRY M. HART, JR. & ALBERT M. SACKS, THE LEGAL PROCESS: BASIC PROBLEMS
INTHE MAKING AND APPLICATION OF LAW 643-46 (William N. Eskridge, Jr. & Philip P. Frickey
eds., 1994).
116 RONALD DwoRKiN, The Forum of Principle, in A MATTER OF PRINCIPLE 33, 69-71 (1985);
see also supra note 81. Note in this regard that Dworkin's discussion of "passivism" in constitu-
tional law, see DwORKIN, supra note IS, at 369-79, does not deal with the need for devices to
limit the power of judicial review in light of the judges' likely biases and institutional weaknesses,
and it fails to see the possibility that, because of those weaknesses, there is a gap between the
actual meaning of the Constitution and judicial pronouncements about that meaning. See id. at
370; cf Lawrence G. Sager, Fair Measure: The Legal Status of Underenforced Constitutional
Norms, 91 HARv. L. REv. 1212, 1213-28 (1978) (discussing that gap). The presumption in favor
of theoretical modesty is one way of disciplining judicial power in light of judicial weaknesses.
117 See ACKERMAN, supra note 58, at 40-56.
11aBrown v. Board of Education, 347 U.S. 483, 495 (i954), is, of course, the most important
example.

CONCLUSION: THE LIMITS OF THEORY

A key task for a legal system is to enable people who disagree on
first principles to converge on outcomes in particular cases. Incom-
pletely theorized agreements help to produce judgments on relative
particulars amidst conflict on relative abstractions.
Notwithstanding these points, the impulse to theorize should not be
disparaged. In some areas of law, we cannot think very well without
embarking on a kind of conceptual ascent from the incompletely theo-
rized outcome, and the conceptual ascent will require us to say
broader and deeper things. It is sometimes desirable to think about an
issue very deeply; this is certainly true for academic observers, and
also for people in politics, in which abstractions and aspirations play a
crucial role. But judges must decide many cases quickly; they have
limited time and capacities. They must also work with each other,
and contests over the right and the good can make it hard for them to
do so. Like ordinary people, judges should obey a norm of mutual
respect, or of reciprocity, and this norm can incline judges to avoid
large-scale contests, at least when they involve people's deepest or
most defining commitments.
These considerations might be made part of a role-specific account
of public reason - an account of public reason designed specifically
for participants in adjudication. But it would not be surprising if at
least some of these points turned out to connect with some of the
broader aspirations of a well-functioning deliberative democracy. Like
judges, ordinary people have limited time and capacities; they too
must live with one another. At least as much as judges, they should
discuss their disputes in ways that entail mutual respect, and thus
avoid unnecessary controversies over fundamental commitments.
I have emphasized that general theory - involving broad claims
about the right and the good - plays an appropriate role in the polit-
ical sphere. In law, I have argued that there is, and should be, a pre-
sumptive taboo against making broad and general claims; there is, and
should be, no such taboo in the world of public reason. But citizens
and representatives, like judges, must make decisions about concrete
controversies in the face of sharp or even intractable disagreements on
first principles. Entirely outside of law, and in the world of ordinary
democratic discussion, incompletely theorized agreements play an im-
portant role.
There is, however, an important exception to the general claim that
I have made throughout this essay. In order for participants in law (or
democracy) to accept that general claim, they must accept at least one
general theory: the theory that I have attempted to defend. This is the
theory that tells them to favor incompletely theorized agreements.
That theory probably cannot itself be accepted without reference to
general theoretical considerations, and its acceptance or rejection

1772 HARVARD LAW REVIEW [Vol. 108:X733
should not be incompletely theorized. Many people claim that law

should reflect a high-level theory of the right or the good, and they
will not be satisfied with incompletely theorized agreements. 119 The
choice between the two approaches will turn on issues that are both
high-level and quite controversial.
This is an important matter. But it is notable that the belief in
incompletely theorized agreements, while tacit, is quite widespread.
The best evidence is the legal culture itself. Such agreements are the
usual stuff of law, and participants in the legal culture are suspicious
of much in the way of theoretical ambition. There are reasons for
their suspicion. What I have tried to do here is to spell out those
reasons and to connect them to some of the most notable characteris-
tics of thinking in law.
119 Recall, however, that the grounds for such agreements are diverse, and that it is possible to
accept some while rejecting others. See supra pp. 1745-5I. In this sense, an incompletely theo-
rized judgment in support of incompletely theorized agreements is certainly imaginable.

SSRN 2995488

Uploaded by

Copyright:

Available Formats

SSRN 2995488

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SSRN 2995488

Uploaded by

Copyright:

Available Formats

COMMENTARY

INCOMPLETELY THEORIZED AGREEMENTS

We think utility, or happiness, much too complex and indefinite an end

JOHN STUART MTLL1

I prefer not to [talk] ...like that; again, grand principles have to be

There is a familiar image of justice. She is a single figure. She is a

This is a promising approach, and in some settings, it may work.

Electronic copy available at: https://ssrn.com/abstract=2995488

on the result and on relatively narrow or low-level explanations for it.

Electronic copy available at: https://ssrn.com/abstract=2995488

agreement on foundations may produce disagreement on particulars.

Electronic copy available at: https://ssrn.com/abstract=2995488

among those deeper reasons if it is not necessary to do so. This phe-

14 See infra pp. 1757-6o.

Electronic copy available at: https://ssrn.com/abstract=2995488

I. AGREEMENTS WITHOUT THEORY

Electronic copy available at: https://ssrn.com/abstract=2995488

My special interest here is in a third kind of phenomenon - in-

Electronic copy available at: https://ssrn.com/abstract=2995488

lower the level of abstraction. People are sometimes able to converge

22 See infra section ILC.

Electronic copy available at: https://ssrn.com/abstract=2995488

tools can allow the achievement of incompletely theorized agreements

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

II. JUSTIFICATIONS AND INSTITUTIONS

What might be said on behalf of incompletely theorized agree-

Electronic copy available at: https://ssrn.com/abstract=2995488

A. The Case for Incomplete Theorization

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

not we are dealing with a multimember body. The first consideration

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

Fifth, and finally, incompletely theorized agreements are well

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

stract arguments of high principle have come from participants in de-

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

wrong to identify enthusiasm for such agreements with a belief in

67 A form of Burkeanism is defended in ANTHONY T. KRONMAN, THE LOST LAWYER 19-52

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

75 See Mark Johnson, Some Constraintson Embodied Analogical Understanding,in ANALOGI.

Electronic copy available at: https://ssrn.com/abstract=2995488

I. HERCULES AND THEORY

77 Doms K. GOODwiN, No ORDINARY TIME 193 (i994).

Electronic copy available at: https://ssrn.com/abstract=2995488

There is no presumption against abstraction. In his hands, the rele-

Electronic copy available at: https://ssrn.com/abstract=2995488

when individual judges on a multimember body have a highly abstract

Electronic copy available at: https://ssrn.com/abstract=2995488

Electronic copy available at: https://ssrn.com/abstract=2995488

low-level principle is finally made part of a more general theory. Per-

Electronic copy available at: https://ssrn.com/abstract=2995488

derstood as if it came from a single mind, so precedents, compiled by

Electronic copy available at: https://ssrn.com/abstract=2995488