Dworkin V Philo

College of William & Mary Law School
William & Mary Law School Scholarship Repository

Faculty Publications Faculty and Deans
2007
Dworkin v. e Philosophers: A Review Essay on

Justice in Robes
Michael S. Green
William & Mary Law School, [email protected]
Repository Citation
Green, Michael S., "Dworkin v. 1e Philosophers: A Review Essay on Justice in Robes" (2007). Faculty Publications. 22.
h2ps://scholarship.law.wm.edu/facpubs/22
Copyright c 2007 by the authors. 1is article is brought to you by the William & Mary Law School Scholarship Repository.
h2ps://scholarship.law.wm.edu/facpubs
DWORKIN V. THE PHILOSOPHERS: A
REVIEW ESSAY ON JUSTICE IN
ROBES
Michael Steven Green*
In this review essay, Professor Michael Steven Green argues that

Dworkin's reputation among his fellow philosophers has needlessly
suffered because of his refusal to back down from his "semantic
sting" argument against H. L. A. Hart. Philosophers of law have uni-
formly rejected the semantic sting argument as a fallacy. Nevertheless
Dworkin reaffirms the argument in Justice in Robes, his most recent
collection of essays, and devotes much of the book to stubbornly, and
unsuccessfully, defending it. This is a pity, because the failure of the
semantic sting argument in no way undermines Dworkin's other ar-
guments against Hart.
Justice in Robes, by Ronald Dworkin. Belknap Press of Harvard Univer-

sity Press, 2006.
Ronald Dworkin's theory of law has not been adequately appreci-

ated by other philosophers of law. At first glance, this statement might
seem absurd. After all, Dworkin is, by a wide margin, the most famous
living philosopher of law. 1 But his influence is primarily upon non-
philosophers. The philosophers of law themselves would choose Joseph
Raz as their most influential member. 2 While countless philosophers of
* Cabell Research Professor, William & Mary School of Law. Ph.D. (Philosophy), Yale Uni-
versity, 1990; J.D., Yale Law School, 1996. Thanks to Matthew Adler, Jules Coleman, and Ken
Himma for helpful comments on an earlier draft.
1. In a list Brian Leiter compiled of the most cited law professors by specialty for the academic
year 2002-2003, the number of citations to Dworkin (4750) was larger than the number of citations to
the next ten law professors writing in the philosophy of law combined. Brian R. Leiter, Top 10 Most
Cited Faculty by Areas, 2002-03, http://www.leiterrankings.com/faculty/2002faculty_impact_areas.
shtml (last visited Mar. 13, 2007).
2. See Leslie Green, Three Themes from Raz, 25 OXFORD J. LEGAL STUD. 503, 503 (2005) ("To
find another legal theorist [besides Raz] who has not only produced an indispensable body of work,
but who taught and encouraged so many jurisprudents of the next generation, one has to go back to
his forebear, H.L.A. Hart."); Brian Leiter, The Law School Observer, 5 GREEN BAG 101, 103 (2001)
("Among philosophers, [Dworkin) has long been overshadowed by Raz, who is generally thought by
specialists in the field to be the most important living legal philosopher.").
1477
1478 UNIVERSITY OF ILLINOIS LAW REVIEW [Vol. 2007
law have declared their debt to Raz, the number of self-described

Dworkinians is smalP
Why has Dworkin's fame not translated into comparable influence
in the philosophical community? It is not enough to point to Dworkin's
view that moral considerations must be part of the criteria for determin-
ing valid law4 (a position that we can call, perhaps misleadingly, natural
law theory). 5 It is true that most philosophers of law reject natural law
theory. 6 But John Finnis/ Stephen Perry,8 and Michael Moore 9 have
each embraced natural law positions, and their writings have not met
with the same level of resistance as Dworkin's. 10
The problem, I believe, is that Dworkin has made it far too easy for
other philosophers to dismiss him, because he stubbornly refuses to back
down from some bad arguments. The best example is Dworkin's "se-
mantic sting" argument against H. L. A. Hart, first presented in Law's
EmpireY As we shall see, philosophers of law have consistently rejected
the semantic sting argument as a fallacy. 12 There is, literally, no dis-
agreement in the profession on the matter. And yet Dworkin continues
to present the argument, even though he would have plenty of good rea-
sons for criticizing Hart without it.
3. See Brian Leiter, The End of the Empire: Dworkin and Jurisprudence in the 21st Century, 36
RUTGERS L.J. 165, 166 (2004) ("The only good news in the story about Dworkin's impact on law and
philosophy is that most of the field declined to follow the Dworkinian path-something, interestingly,
that those not working in legal philosophy generally do not know."); Thorn Brooks, Book Review, 69
MOD. L. REV. 140, 140 (2006) ("[Dworkin's) wide readership has not translated into more than a small
number of disciples. It is quite rare to find anyone in the field identifying herself as a 'Dworkinian."').
4. RONALD DWORKIN, JUSTICE IN ROBES 35 (2006).
5. See Ronald A. Dworkin, "Natural" Law Revisited, 34 U. FLA. L. REV. 165, 165 (1982) (noting
that critics have characterized his theory as a natural law theory).
6. Most philosophers of law are positivists. Inclusive legal positivists claim that the law may
contain moral criteria of validity, but need not do so. Whether the law includes morality is ultimately
answered by social facts-namely the conventions that form the foundation of a legal system. See, e.g.,
JULES L. COLEMAN, THE PRACTICE OF PRINCIPLE 67-148 (2001) [hereinafter COLEMAN, PRACTICE
OF PRINCIPLE); H. L. A. HART, THE CONCEPT OF LAW 250-54 (2d ed. 1994); W. J. WALUCHOW,
INCLUSIVE LEGAL POSITIVISM (1994); Jules L. Coleman, Negative and Positive Positivism, 11 J. LEGAL
STUD. 139 (1982). Exclusive legal positivists agree that the criteria of legal validity are ultimately de-
termined by social facts, but argue that morality cannot be incorporated into the law at all. See Joseph
Raz, Authority, Law, and Morality, in ETHICS IN THE PUBLIC DOMAIN: ESSAYS IN ni:E MORALITY OF
LAW AND POLITICS 195 (rev. ed. 1994); Joseph Raz, Legal Positivism and the Sources of Law, in THE
AUTHORITY OF LAW: ESSAYS ON LAW AND MORALITY 37 (1979); Scott J . Shapiro, On Hart's Way
Out, in HART'S POSTSCRIPT: ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF LAW 149 (Jules Cole-
maned., 2001).
7. See, e.g., JOHNFINNIS,NATURALLAW AND NATURAL RIGHTS (1980).
8. See, e.g., Stephen R. Perry, Interpretation and Methodology in Legal Theory , in LAW AND
INTERPRETATION (Andrei Marmor ed., 1995).
9. See Michael Moore, Law as a Functional Kind, in NATURAL LAW THEORY:
CONTEMPORARY ESSAYS 188 (Robert P. George ed., 1992).
10. Only Dworkin's arguments have been described as "riddled with philosophical confusions,"
COLEMAN, PRACTICE OF PRINCIPLE, supra note 6, at 155, and "largely without philosophical merit,"
Brian Leiter, supra note 3, at 166.
11. See RONALD DWORKIN, LAW'S EMPIRE 45-46 (1986).
12. Indeed, according to Brian Leiter, it "has by now been subjected to so many withering criti-
cisms, that . .. if any argument is no longer worth discussing, it is this one." Brian Leiter, Beyond the
Hart!Dworkin Debate, 48 AM. J. JURIS. 17,31 n.49 (2003).
No.5] DWORKIN V. THE PHILOSOPHERS 1479
Dworkin's stubbornness is sadly in evidence in Justice in Robes,13 his

most recent collection of essays. As with all of Dworkin's writings, there
is much in the book to admire. But I doubt that its virtues will be noticed
by philosophers of law. What they are likely to remember is the futile
defense of the semantic sting argument to which a large part of the book
is devoted. 14
I. THE SEMANTIC STING ARGUMENT
In his semantic sting argument, Dworkin claims that H. L. A. Hart's

theory of law is a consequence of Hart's semantic views about how words
(in particular the word "law") get their meaning. According to Dworkin,
Hart's semantic theory was conventionalist-the meaning of the word
"law" is determined by agreement. As a result, Dworkin argues, Hart
was committed to a conventionalist theory of law, in which the law of a
jurisdiction is determined by agreement. Dworkin takes the semantic
foundations of Hart's theory of law as a reason for rejecting it. Because
semantic conventionalism is inadequate, so is Hart's theory of law.
Although Dworkin's semantic sting argument is unquestionably a
fallacy, why it is a fallacy takes some explaining. In particular, care is
needed in describing the conventionalist theories of law and convention-
alist semantics that Dworkin wrongly sees as related.
A. Conventionalist and Nonconventionalist Theories of Law

A theory of law is a general account-that is, an account that applies
to any jurisdiction (the United States, Uzbekistan, the Roman Empire)-
of the conditions that must be satisfied for something to be the valid law
of the jurisdiction. 15 Because they are general, theories of law have a
hurdle to overcome: the conditions for valid law change as one moves
from jurisdiction to jurisdiction. To use a very simple example, in 1788
the valid laws of Great Britain were arguably what the King-in-
Parliament commanded. At the same time, the valid laws of France were
arguably what the King of France commanded. "What the King-in-
Parliament commands is valid law" cannot be a general theory of law, for
it does not work for France in 1788, just as "what the King of France
commands is law" would not work for Britain at that time. For this rea-
son, it appears that a general theory of law cannot directly identify the
laws of a jurisdiction. Instead, it must offer a more general criterion that
allows us to identify the criteria, specific to each jurisdiction, that in turn
identify the laws of that jurisdiction.
13.DWORKIN, supra note 4.

14. /d.at30-33, 165-66,225-26.
15. See MichaelS. Moore, Interpreting Interpretation, in LAW AND INTERPRETATION: ESSAYS IN
LEGAL PHILOSOPHY 1, 26--27 (Andrei Marmor ed., 1995).
In Hart's theory of law, the general criterion is acceptance by offi-

cials in the jurisdiction: something is the law of a jurisdiction if it satisfies
the criteria that the jurisdiction's officials Qudges, legislators, sheriffs,
and the like) have accepted for enforcing norms. 16 For example, the Se-
curities Exchange Act is valid law in the United States because it satisfies
the criteria, such as promulgation in accordance with the U .S. Constitu-
tion, that American officials have accepted for norms that may be backed
up by governmental power. 17 Hart calls this official practice of enforcing
norms on the basis of these accepted criteria a rule of recognition. 18
An alternative, but not incompatible, description of a theory of law
(one favored by Dworkin) is to speak of it as an account of the truth
conditions for propositions of law. 19 Consider the proposition "The Se-
curities Exchange Act is valid American law." Hart's theory of law can
be understood as describing the conditions under which that proposition
is true-namely if the Securities Exchange Act satisfies the enforcement
criteria accepted by American officials.
It follows from Hart's theory that when there is disagreement
among judges about how to resolve a hard case, and the disagreement
stems from differing views about enforcement criteria, there is no law an-
swering the case, for the rule of recognition is silent on the matter. 20 This
is not to say that the court must dismiss the action for failure to state a
claim, for the court's resolution of the case may be a legally permissible
act of law making (as opposed to law application). It would be an act of
law making if so identified by the rule of recognition, that is, if there is
agreement among officials that judicial resolutions of legally undeter-
mined cases are enforceable (something true of the American rule of
recognition, for example).
16. HART, supra note 6, at 94. The existence of a legal system also requires that the primary
rules that are valid according to the criteria are generally-although not necessarily a/ways-obeyed
by the population. /d. at 116-17.
17. On the complexity of these criteria in the American legal system, see Michael Steven Green,
Legal Revolutions: Six Mistakes About Discontinuity in the Legal Order, 83 N.C. L. REV. 331, 343-51,
358--60, 374-82 (2005) , and Kent Greenawalt, The Rule of Recognition and the Constitution , 85 MICH.
L. REV. 621, 630--60 (1987).
18. In fact, sometimes Hart speaks of the rule of recognition not as a social practice but as a
proposition specifying the criteria of legal validity within the legal system. See Benjamin C. Zipursky,
The Model of Social Facts, in HART'S POSTSCRIPT: ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF
LAW, supra note 6, at 219, 227-28. But he also uses the term to refer to the social fact that a certain
rule of recognition (in the propositional sense) is practiced by officials. The proposition is practiced in
the sense that officials agree to enforce only that which satisfies the criteria in the proposition. On the
distinction between a rule of recognition in the propositional sense and the practice of that rule, see
COLEMAN, PRACTICE OF PRINCIPLE, supra note 6, at 77-78.
19. DwORKIN, supra note 4, at 9.
20. Judges might disagree about how a case should tum out even though they agreed on en-
forcement criteria, for they might disagree about whether those criteria are in fact satisfied. They
might agree, for example, that t.he case could be resolved only by a statute validly enacted by the legis-
lature, but disagree about whether the legislative quorum required for valid enactment was satisfied
because the presence of a legislator at the time of voting was in dispute.
Let us call a theory of law conventionalist if it looks to the currently

accepted views of a group in a jurisdiction to find the criteria identifying
the laws of that jurisdiction. In contrast, under a nonconventionalist the-
ory of law these criteria can diverge from the currently accepted views of
any group in the relevant jurisdiction. Hart's theory of law is conven-
tionalist because the criteria identifying the laws of a jurisdiction must be
accepted by officials in the jurisdiction. 21
An extreme example of a nonconventionalist theory of law would
simply use the same set of criteria for identifying the laws of all jurisdic-
tions. Such a theory might say, for example, that God's word, as ex-
pressed in the Bible, is the law of every jurisdiction in the world. But a
theory of law can be nonconventionalist even if it is sensitive to the cur-
rent views of people in a jurisdiction, provided that it does not simply
equate the criteria identifying the laws of a jurisdiction with these current
views the way a conventionalist theory of law does.
Dworkin's theory of law is an example of such a nonconventionalist
theory of law. In Justice in Robes, Dworkin asks us to consider an imagi-
nary case in which the plaintiff, Mrs. Sorenson, has suffered harm from a
generic drug. 22 She cannot prove which company manufactured the pills
that caused her harm. Does the law of her jurisdiction entitle her to
damages from each of the manufacturers on the basis of its market share,
given that in past cases "judges stated that no one is liable for injuries he
did not cause" and there are "no past decisions in which the judge
awarded anyone damages based on market share rather than direct cau-
sation "?23
Dworkin offers the following account of how the matter would be
decided according to his theory of law and according to Hart's:
In my view, legal argument is characteristically and pervasively
moral argument. Lawyers must decide which of competing sets of
principles provide the best-morally most compelling-justification
of legal practice as a whole. According to Hart's [theory], on the
other hand, substantive legal argument is normative only when so-
cial sources make moral standards part of the law. No legislature or
past judicial decision has made morality pertinent in Mrs.
Sorenson's case so, on Hart's view, no moral judgment or delibera-
21. In speaking of Hart's theory as conventionalist, I mean only and vaguely that the rule of rec-
ognition rests upon agreement. I do not take a stand on whether it should be understood as a conven-
tion in the sense articulated by David Lewis, that is, as a solution to a coordination problem. DAVID
LEWIS, CONVENTION: A PHILOSOPHICAL STUDY (1969). For such an account of the rule of recogni·
tion, see Jules Coleman, lncorporationism, Conventionality, and the Practical Difference Thesis, in
HART'S POSTSCRIPT: ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF LAW, supra note 6, at 99, 114-
22; Gerald J. Postema, Coordination and Convention at the Foundations of Law, 11 J. LEGAL STUD.
165 (1982). Nor do I take a stand on whether it should be understood as a shared cooperative activity
in Michael Bratman's sense. Michael E. Bratman, Shared Cooperative Activity: Three Features, 101
PHIL. REV. 327 (1992). For such an account of the rule of recognition, see COLEMAN, PRACTICE OF
PRINCIPLE, supra note 6, at 96-99.
22. DWORKIN, supra note 4, at 7.
23. !d. at 8.
tion enters into the question whether she is legally entitled to what
she asked. So far as the law is concerned, he would have said, she
must lose. 24
The truth is, for Hart's theory to say that Sorenson loses, Dworkin must
also assume that the statutes and past decisions did not leave her case le-
gally indeterminate. After all, it is common for the rule of recognition of
a jurisdiction to allow courts to create new law, on the basis of moral
considerations, when the law is silent on a matter.
Nevertheless, the fact remains that Sorenson's case is treated differ-
ently under Hart's theory compared to Dworkin's. According to Hart's
theory, morality is relevant to Sorenson's case only if officials agree that
it is. For Dworkin, in contrast, the law concerning Sorenson's case is al-
ways answered "by asking whether the best justification of negligence
law as a whole contains a moral principle that would require that result
in her circumstances." 25 Moral considerations are always relevant, even
if officials in the jurisdiction disagree on this matter, and so it always is
possible that the law is on Sorenson's side.
Dworkin's theory of law is nonconventionalist because the criteria
identifying the law of a jurisdiction are not necessarily those currently
accepted by people in the jurisdiction. Currently, no one in the jurisdic-
tion agrees about how Sorenson's case should be answered or they agree
that she should lose. But there nevertheless can be law on the matter on
the basis of which she should win- namely if such law follows from the
best moral justification of negligence law. But Dworkin's theory of law is
not completely insensitive to the attitudes of people in the jurisdiction,
for the materials upon which the moral interpretation works are identi-
fied by these attitudes. The question is the best moral justification of the
negligence law of Sorenson's jurisdiction, not whatever is morally best
simpliciter. For this reason, the criteria identifying the valid law of a ju-
risdiction, although always taking morality into account, will do so in dif-
ferent ways as one moves from one jurisdiction to the next.
B. Philosophy as Conceptual Analysis

Hart titles his book The Concept of Law because he takes his con-
ventionalist theory of law to be the articulation of the content of that
concept.26 In so doing, Hart assumes a popular theory of philosophical
activity, according to which philosophers analyze the contents of prob-
lematic and significant concepts, such as freedom, knowledge, and law, in
order to arrive at peculiarly philosophical knowledge concerning the es-
24. !d. at 144.

25. !d. at 14.
26. See H. L. A. Hart, Jhering's Heaven of Concepts and Modern Analytical Jurisprudence, in
ESSAYS IN JURISPRUDENCE AND PHILOSOPHY 265, 274 (1983); Nicos Stavropoulos, Hart's Semantics.
in HART'S POSTSCRIPT: ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF LAW, supra note 6. at 59, 63-
88.
sential existence conditions for these things-that is, knowledge of what

must be the case for something to be freedom, knowledge, or law. 27
An allied position, also endorsed by Hart,28 is that philosophers ar-
rive at existence conditions by analyzing the meanings of words. But this
view suffers from the awkward problem that linguistic meanings rarely
correspond to the contents of concepts with which philosophers are con-
cerned. It is implausible with respect to the philosophy of law, for exam-
ple, because the word "law" can be used to refer not merely to statutes,
judicial decisions, and the like, but also to scientific or mathematical
laws. Philosophers of law are generally interested in the word "law" only
as applied to items of the former type. 29 For this reason, it is best to as-
sume that Hart is concerned, not with the meaning of the word "law,"
but with the content of one of the concepts associated with the word
"law."
Much of Justice in Robes is devoted to criticizing aspects of Hart's
philosophical method, as Dworkin understands it. For example,
Dworkin argues that philosophers of law should be understood as inves-
tigating the doctrinal concept of law, that is, the concept that one em-
ploys when one talks about what is valid or invalid law in a certain juris-
diction.30 He claims that some confusion in Hart's philosophy of law has
resulted from the failure to distinguish the doctrinal from other concepts
of law. 31 Dworkin also rejects Hart's "archimedianism" -that is, Hart's
belief that the analysis of the concept of law is an activity divorced from
27. E.g., Stavropoulos, supra note 26, at 59, 64.

28. HART, supra note 6, at vi.
29. See, e.g., Jules L. Coleman & Ori Simchen, "Law," 9 LEGAL THEORY 1,1 n.1 (2003); Joseph
Raz, Two Views of the Nature of the Theory of Law: A Partial Comparison, in HART'S POSTSCRIPT:
ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF LAW, supra note 6, at 1, 7-8.
30. See DWORKIN, supra note 4, at 2.
31. E.g., id. at 2-5. The sociological concept of law is used to identify " a particular type of insti-
tutional social structure." /d. at 3. It is employed, for example, when we ask whether a primitive tribal
society has law. Dworkin believes, perhaps rightly, that this concept of law is insufficiently determi-
nate for such questions to have an interesting answer.
Dworkin also identifies the taxonomic concept of law as that used to identify which standards "are
legal standards as opposed to moral or customary or some other kind of standards." /d. at 4. For ex-
ample, even when a judge must use arithmetic to decide the appropriate damages that must be paid by
a defendant, it does not mean that arithmetic is part of the law in the taxonomic sense. /d. at 4-5.
Likewise, a Polish court might use Greek law to decide a case, but that does not necessarily mean that,
taxonomically, Greek law has become part of Polish law. /d. at 235. Once again, Dworkin does not
think that the question of what is law in the taxonomic sense is an interesting one.
One can understand Hart's theory as concerning the concept of law in both the sociological and
taxonomic sense. Hart's idea of a rule of recognition was intended to identify those societies in which
law, rather than a more informal system of norms, exists. Furthermore, Hart's theory might be under-
stood as concerning the taxonomic concept, in the sense that only those norms identified by the rule of
recognition should be considered laws. But Dworkin accepts that Hart's conception of a rule of rec-
ognition provides an account of the doctrinal concept of law and so is a theory of law in the proper
sense. /d. at 26. Hart's theory provides truth conditions for propositions of law. It simply gives the
wrong truth conditions.
the more concrete efforts of lawyers and judges to determine what is le-
gal or illegal.32
But even though Dworkin sees the investigation of the doctrinal
concept of law as situated within concrete judgments of legality and ille-
gality, he nevertheless agrees with Hart that a general theory of law can
be arrived at by revealing the content of this concept of law. 33 Dworkin
thinks that his theory of law is the proper account of the content of this
concept. 34
C. Conventionalist (Meta)semantics
Both Hart and Dworkin understand their theories of law as reveal-
ing the content of the concept of law. For this reason, both theories of
law can be understood as semantic. To make a semantic claim is to iden-
tify the meaning of a word-or the content of a concept associated with
the word. For example, to say that the content of the concept of bache-
lor is unmarried male (or that "bachelor" means unmarried male) is to
make a semantic claim. Both Hart's and Dworkin's theories of law can
be understood as semantic, for both are accounts of the content of the
concept of law.
But there is a different sense of the word "semantic," which refers
not to claims about what the content of a concept (or the meaning of a
word) is, but rather to claims about how this content or meaning is gen-
erated. Sometimes the term "metasemantic" is used to identify semantic
claims in this second sense. 35 The view that "bachelor" means unmarried
male says nothing about why "bachelor" has this meaning and so is not a
metasemantic claim. Likewise, Hart's and Dworkin's theories of law are
not metasemantic, for in saying what the content of the concept of law is,
they do not say anything about how it came to have this content.
Dworkin primarily uses the term "semantic" in this second, meta-
semantic, sense. For example, in Justice in Robes, he describes as "se-
32. /d. at 140--86. As we shall see later, it might be the case that Hart, like Dworkin, did not in-
sist upon a neat separation between the analysis of the concept of law and concrete applications of the
concept.
33. For example, in Justice in Robes, Dworkin notes that " [i]n one respect ... [Hart and I] are in
the same boat. We both believe that we will understand legal practice and phenomena better if we
undertake to study, not law in some particular manifestation, like the law of product liability in Scot-
land, but the very concept of law." Id. at 145; see also id. at 8--9 (stating that an account of the doc-
trinal concept of law is the development of. a " general theory of law").
34. He does not join Quine, who questions the very existence of philosophical truths that follow
from the content of a concept. He is not a skeptic about conceptual analysis. W. V. Quine, Two Dog-
mas of Empiricism, 60 PHIL. REV. 20, 20-34 (1951). For a Quinean position in the philosophy of law,
see Brian Leiter, Legal Realism, Hard Positivism, and the Limits of Conceptual Analysis, in HART'S
POSTSCRIPT: ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF LAW, supra note 6, at 355, 357.
35. David Kaplan, Afterthoughts, in THEMES FROM KAPLAN 565, 573-76 (Joseph Almog et al.
eds., 1989); see also Coleman & Simchen, supra note 29, at 12, 18; Michael Steven Green, Dworkin's
Fallacy, Or What the Philosophy of Language Can't Teach Us About the Law, 89 VA. L. REV. 1897,
1905 n.21 (2003).
mantic" the view that a concept is "criterial." To say that a concept is

criteria! is to say how the concept gets its content. If a concept is crite-
ria!, its content is fixed by agreement "on a definition-rough or pre-
cise-that sets out the criteria for the correct application of the associ-
ated term or phrase."36 The concept of a bachelor is an example of a
criteria! concept: "People share the concept of bachelorhood only when
they know that a bachelor is an unmarried male." 37
To repeat, to say that the concept of a bachelor is criteria! is to
make a metasemantic claim about the concept. It tells us how the con-
cept gets its content, namely from the commonly accepted criteria for
employing the concept. To say the concept of a bachelor is criteria! is not
a semantic claim (in the first sense of the word) because it does not as-
sign a content to the concept the way "'bachelor' means unmarried
male" does. However, armed with the metasemantic knowledge that the
concept is criteria!, we can figure out its content by examining the criteria
that people agree upon for using it.
Notice that by virtue of accepting a semantic account of a concept,
one is not yet committed to any metasemantic theory. Someone who be-
lieves that "bachelor" means unmarried male might have any number of
theories of why the word has this meaning. She might believe, for exam-
ple, that God determines the meanings of our words. According to this
(implausible) metasemantic theory, "bachelor" would still mean unmar-
ried male even if we all used the word to refer to married females.
The metasemantic view that concepts are criteria! can be under-
stood as conventionalist because concepts are shared only to the extent
that there is current agreement concerning the criteria for the concepts'
use. Other people share my concept of bachelor only if they also accept
the criteria unmarried and male. 38 If they do not, their concept has a dif-
ferent content. When they think about bachelors, they are thinking
about something different from what I am.
Dworkin's fundamental criticism of conventionalist metasemantics
is that it makes meaningful disagreement about a concept's content im-
possible, for such disagreement would simply mean that different con-
cepts were being used and people were talking past one another. 39 Of
course, sometimes this is just what we want to say. Someone who insists
that married women can be bachelors must be working with a different
concept of bachelorhood and so cannot really be disagreeing with me at
36. DWORKIN, supra note 4, at 9. In Law's Empire, Dworkin also describes this metasemantic
position as the view that "[w]e follow shared rules ... in using any word: these rules set out criteria
that supply the word's meaning." DWORKIN, supra note 11, at 31.
38. For a fuller description of the metasemantic conventionalist approach, see Green, supra note
35. at 1899-1903.
39. DWORKIN, supra note 4, at 9-12.
all. Dworkin argues, however, that conventionalist metasemantics is im-

plausible concerning some concepts.40
For example, according to the criteria accepted by most white
Southerners in 1850, slavery should be judged just. If conventionalist
metasemantics applies to the concept of justice, then slavery was .indeed
just according to their concept of justice, even though it is not according
to ours. But this, Dworkin argues, misdescribes our disagreement. We
think that they were wrong about justice in a shared sense, not right
about justice in some different sense. 41 The inability of conventionalist
metasemantics to explain disagreement about the content of a concept is
the "semantic sting."42 As we shall later see, Dworkin believes that be-
cause there is significant conflict concerning the content of the concept of
law, it too cannot be understood as criteria!.
What are the nonconventionalist metasemantic alternatives? Ac-
cording to one approach, commonly applied to natural-kind concepts,
meaningful disagreement about the content of a concept is possible be-
cause the content is determined, not by common criteria for use, but by
the natural structure of paradigm samples falling under the concept. 43
For example, we can say that we share the same concept of water with
people in the sixteenth century, even though their criteria for using the
concept (odorless, colorless, potable liquid) are different from ours (hav-
ing the molecular structure H 20). The concept had the same content for
them as it does for us (namely having the molecular structure H 2 0) be-
cause paradigm samples of water have the molecular structure H 20. It is
true that people in the sixteenth century might have occasionally de-
scribed as "water" some odorless, colorless, potable liquid that was not
H 2 0. But they misapplied our common concept of water. They did not
correctly apply a different concept.
Although natural-kind metasemantics might appear to allow for
meaningful disagreement concerning the contents of controversial con-
cepts-like equality, liberty, or law-Dworkin rejects such an approach
in Justice in Robes:
Do these concepts describe, if not natural kinds, at least political
kinds, which, like natural kinds, can be thought to have a basic in-
grained physical structure or essence? Or, at least, some structure
that is open to discovery by some wholly scientific, descriptive, non-
normative process? Can philosophers hope to discover what equal-
ity or legality really is by something like a DNA or chemical analy-
sis? No. That is nonsense. 44
40. /d.
42. DWORKIN, supra note 4, at 223-26; DWORKIN, supra note 11, at 43-46.
44. /d. at 152.
What these concepts are about has no natural structure that could be sci-
entifically investigated: "(P)hilosophical analysis of political concepts
cannot be shown to be descriptive on the model of scientific investigation
into natural kinds. Liberty has no DNA." 45 And neither does the law.46
Another. reason not to apply natural-kind metasemantics to the
concept of law is that this metasemantics is still conventionalist in the
sense that content depends upon agreement concerning paradigm sam-
ples. And certain concepts can lack even that level of agreement: "Peo-
ple can share ... a concept even when they disagree dramatically about
its instances."47 A metasemantic theory about such a concept "cannot
simply ... excavate the deep structure of what people mainly agree are
instances."48
Dworkin's metasemantic theory is more radically nonconventional-
ist than the natural-kind approach. The content of a concept can outstrip
the current beliefs and attitudes of those using the concept even concern-
ing paradigm samples. The fact that there is disagreement, even perva-
sive disagreement, about what falls under a concept does not mean that
the concept is not shared or that there is no fact of the matter about what
falls under the concept.49
D. Dworkin on the Relationship Between (Meta)semantic

Conventionalism and Conventionalist Theories of Law
But Dworkin's semantic sting argument is more than a criticism of
conventionalist metasemantics. In Law's Empire he argues that accep-
tance of conventionalist metasemantics leads to conventionalist theories
of law. 50 Dworkin says the same thing in Justice in Robes,51 except he in-
cludes natural-kind approaches among the metasemantic culprits:
(In Law's Empire) I hypothesized that (legal positivists) assume that
all concepts, including the doctrinal concept of law, are criteria!
concepts and that proper analysis of the doctrinal concept must
therefore consist in elucidating the tests that lawyers share, except
in borderline cases, for judging whether propositions of law are
45. !d. at 153.

46. !d. at 166.
47. !d. at 11-12.
48. !d. at 12.
49. I shall not say anything more here about the specifics of Dworkin's nonconventionalist meta-
semantics. I explore that issue in Green, supra note 35, at 1908-29. In particular, I argue that, much as
Dworkin confuses metasemantic conventionalism with a conventionalist theory of law, he confuses
metasemantic nonconventionalism with a nonconventionalist theory of law. Dworkin has criticized
my interpretation in Justice in Robes, Dworkin, supra note 4, at 226-27, 289 n.5, and I have responded
in Michael Steven Green, Does Dworkin Commit Dworkin 's Fallacy? , 28 OXFORD J. LEGAL STUD.
(forthcoming 2008). I will not discuss this disagreement between the two of us here.
50. See DWORKIN, supra note 11, at 45-46.
51. DWORKIN, supra note 4, at 31 ("Hart assumed, in effect, that the doctrinal concept of law is a
criteria! concept and that analyzing that concept means bringing to the surface the criteria that lawyers
actually use, even if unselfconsciously, in applying it.").
true. I called that assumption- that all concepts are criteria!-the

"semantic sting." I then offered an analysis of law as an interpre-
tive rather than a criteria! concept to show how lawyers' disagree-
ments can be genuine even if they do not agree on criteria for iden-
tifying true propositions of law.
I tailored that account of the semantic sting, in 1986, to fit the
arguments for legal positivism then current. But a fresh generation
of legal philosophers has become more sophisticated in the phi-
losophy of language, which has itself moved on since I wrote, and
my description has proved too narrow. I should now characterize
the sting more broadly: it lies in the assumption that all concepts
depend on a convergent linguistic practice... : a practice that marks
out the concept's extension either through shared criteria of appli-
cation or by attaching the concept to a distinct natural kind. The in-
fection of the semantic sting, I shall now say, is the assumption that
all concepts of law, including the doctrinal, depend on a convergent
practice in one of those two ways. The pathology of the semantic
sting remains the same. Lawyers who are stung will suppose that an
analysis of the concept of law must fit-and only fit-what lawyers
mainly agree is law.52
In particular Dworkin argues that Hart's conventionalist theory of
law was the result of his conventionalist metasemantic views53 : "Hart as-
sumed, in effect, that the doctrinal concept of law is a criteria! concept
and that analyzing the concept means bringing to the surface criteria that
lawyers actually use, even if unselfconsciously, in applying it."54 Al-
though Dworkin recognizes that his interpretation of Hart has been criti-
cized since the publication of Law's Empire, he insists that his is "the
best available" 55 and that his "original diagnosis was correct."56 Hart's
conventionalist theory of law was a consequence of his conventionalist
metasemantics.
II. DWORKIN'S ARGUMENT FAILS

But Dworkin's argument is a fallacy, for, as many philosophers of
law have argued, conventionalist theories of law do not follow from con-
ventionalist metasemantic views.
52. /d. at 225-26.

53. In Law's Empire, Dworkin describes theories of law like Hart's as "semantic theories of law"
because they allegedly follow from conventionalist (meta)semantic views. DWORKIN, supra note 11, at
31-35.
55. /d. at 166.
56. /d. at 31.
A. Conventionalist Metasemantics Does Not Generate Conventionalist

Theories of Law
Keep in mind that metasemantic conventionalism is a view about
how concepts-including the concept of law-get the contents they have.
According to this view, the content of the concept of law is determined
by the criteria for using the concept agreed upon by those employing it.
In saying this, one has not yet said what the content of the concept is. In
contrast, a conventionalist theory of law says what the content of the con-
cept of law is, without yet saying how the concept gets that content.
If this distinction is kept in mind, it should be clear that convention-
alist metasemantics does not entail conventionalist theories of law. After
all, the metasemantic fact that the content of the concept of law consists
of the criteria agreed upon by those using the concept has yet to tell us
what these criteria are and therefore cannot yield any particular theory
of law. The criteria people have agreed upon for using the concept of
law might be that the law of every jurisdiction in the world is God's
word, as expressed in the Bible. One can accept conventionalist meta-
semantics and a nonconventionalist theory of law.
Indeed, people could agree that the criteria for using the concept of
law are those specified in a nonconventionalist theory of law like
Dworkin's. Dworkin's theory of law would be the content of a criteria!
concept. 57
57. Indeed, it is arguable that there is nothing that conventionalist metasemantics excludes as the
content of a criteria! concept. Assume that it is claimed to exclude a candidate content. To be a plau-
sible candidate, what the content is must be described by describing what it is about. (For example,
the content bachelor can be described by saying that it is about unmarried males.) One has not ade-
quately described a candidate content simply by saying that it is about an indescribable something, for
the metasemantic conventionalist could simply deny that thinking about this indescribable something
is thinking about anything at all. But once a description of the content is offered, the description itself
could be the criteria agreed upon for using a concept.
The fact that any description can be made the criteria of a concept has been taken advantage of by
defenders of conventionalist metasemantics. Consider the metasemantic theory for natural-kind
terms, discussed earlier, according to which the meaning and reference of these terms is fixed by the
underlying structure of a paradigm sample. See supra text accompanying notes 43-49. Although such
accounts appear incompatible with conventionalist metasemantics-since the underlying structures of
these paradigm samples can be unknown to those using the term-metasemantic conventionalists can
reply that the description by means of which this alternative metasemantic theory is articulated is itself
part of the criteria for using the term. The word "gold," for example, means " whatever has the same
structure as the stuff picked out by the appropriate causal-historical relationship with the first uses of
the term 'gold."' See, e.g., Green, supra note 35, at 1950-51; Frederick W. Kroon , Causal Descriptiv-
ism, 65 AUSTRALASIAN J. PHIL. 1 (1987); David Lewis, Putnam's Paradox, 62 AUSTRALASIAN J. PHIL.
221 (1984).
Indeed, if Dworkin argued that his theory of law could not be the content of a criteria! concept, he
would be saying, in effect, that conventionalist metasemantics is unable to explain how one could think
his theory of law, even to reject it. I doubt-and Dworkin never suggests-that he thinks metasemantic
conventionalism is that bad of a theory.
To say that conventionalist metasemantics renders no content unthinkable is not to say that a con-
tent cannot be unthinkable in particular circumstances. What conventionalist metasemantics makes
impossible is thinking a content that outstrips currently accepted criteria for the concept's use. Any
content can be thought if it is formed in the right way, and any content can be unthinkable if it is
formed in the wrong way.
The error in Dworkin's argument is his assumption that because the

content of a concept is determined by convention, it is about convention.
That this is a mistake is easiest to see in connection with concepts whose
content, being obviously unrelated to human practices, could not possi-
bly be about conventions. No one is tempted to commit Dworkin's fal-
lacy in connection with the concept of a bachelor, for example. Even if
the content of this concept is determined by convention, it is, quite obvi-
ously, about unmarried males, not conventions.
But even if a concept is about human practices, the metasemantic
fact that its content is determined by convention does not mean it is
about the conventional aspects of those human practices. Consider, for
example, the concept of the nonconventional, that is, a concept that re-
fers to those characteristics of practices that do not depend upon agree-
ment among the practices' participants. The nonconventional would in-
clude, for example, the injustice of slavery in the South in 1850. The
metasemantic conventionalist would say that the concept of the noncon-
ventional gets its content from language-users' conventions. But it would
be a mistake to conclude from this that the concept's content is about the
conventional aspects of human practices. It is, after all, the concept of
the nonconventional. To think that a conventionalist theory of law fol-
lows from conventionalist metasemantics is a similar mistake, for conven-
tionalist metasemantics on its own gives one no reason _to think that the
concept of law is about conventional rather than nonconventional as-
pects of legal practices.
By the same token, the fact that the content of a concept is deter-
mined nonconventionally does not mean that it is about the nonconven-
tional. Indeed, the content of the concept of a convention might be de-
termined nonconventionally, that is, by something other than the criteria
agreed upon for the concept's use. 58 This would allow for meaningful
disagreements about the content of the concept. The fact that we have
different theories of conventions would not mean that we are simply
talking past one another. But it would be absurd to conclude from this
that the concept of a convention is about the nonconventional.
Just as nonconventionalist metasemantics is compatible .with our
concept of the conventional, it is compatible with conventionalist theo-
ries of law. If a philosopher· of law accepts nonconventionalist metase-
For this reason, conventionalist metasemantics can make conventionalist theories of law unthink-
able. Consider Dworkin, who currently accepts a nonconventionalist theory of law. According to me-
tasemantic conventionalism, when Dworkin is thinking about the law, he cannot be thinking about the
law as identified by a conventionalist theory like Hart's, because that attributes a content to Dworkin's
concept of law that outstrips the criteria for using the concept that Dworkin currently accepts. For the
metasemantic conventionalist, Dworkin thinks about Hart's theory of law (in order to reject it) only
when he generates a concept whose criteria are, by stipulation, those spelled out in Hart's theory. In
contrast, nonconventionalist metasemantics would open up the possibility that Dworkin is actually
thinking about the law as identified in Hart's theory when he tries to think about the law as identified
by his own theory.
58. Green, supra note 35, at 1917-18.
mantics, he thinks that people can share a concept of law even though
they do not currently agree about the criteria for its use. This would al-
low for meaningful debate about the proper theory of law. But all this is
compatible with insisting that, according to the nonconventional criteria
that actually make up the content of the concept of law, something is the
law of a jurisdiction only if it satisfies the criteria for enforcement agreed
upon by people within that jurisdiction.
Indeed, not only is such a person possible, H. L.A. Hart was (or at
least claimed to be) just such a person. True, Hart thought that an ex-
amination of the content of the concept of law showed that the proper
theory of law was conventionalist. But he did not think that the reason
why the concept of law had this content was because language users
agreed that it did. As he put it, "the criteria of the application of a con-
cept with a constant meaning may both vary and be controversial."59 He
was no less aware than Dworkin that there was fundamental disagree-
ment between theorists of law. But he did not think that they were sim-
ply talking past one another.
If Hart was indeed a metasemantic nonconventionalist, then
Dworkin is wrong to accuse him of archimedianism, that is, of a com-
mitment to philosophy as occurring on a "second-order platform of
'meta' discourse, in which first-order concepts are defined and ex-
plored."60 Philosophy of law would be archimedian if it simply reported
the criteria people accept for using the concept of law, in a way that did
not depend upon concrete employment of the concept. Hart, however,
repudiated such a method.
Nevertheless, even if Dworkin is right that Hart was a metasemantic
conventionalist-and there are some reasons to believe that Hart was-
Dworkin's argument remains a fallacy. Dworkin is still wrong that Hart
arrived at his conventionalist theory of law because he was a metaseman-
tic conventionalist, for metasemantic conventionalism is compatible with
nonconventionalist theories of law like Dworkin's.
B. Why Does Dworkin Commit His Fallacy?

Given that Dworkin's argument is a fallacy, why does he repeatedly
succumb to it? One reason is that he confuses two practices-the linguis-
tic practice of applying the concept of law and the legal practice of en-
forcing norms in a jurisdiction (which Hart calls a rule of recognition). 61
Although they may appear similar, the two practices are in fact very
different. First, the linguistic practice of talking about the law is a prac-
tice in which any language user can participate, including those in differ-
ent jurisdictions or indeed those in conditions of anarchy. In contrast,
59. HART, supra note 6, at 246.

61. Green, supra note 35, at 1919-29.
the legal practice is one in which only officials in a particular jurisdiction

can participate. Second, one engages in the linguistic practice just as
much when one says law does not exist as when one says it does. In con-
trast, the legal practice consists not in opining that certain laws are valid
or not valid, but in enforcing, through coercive sanctions, norms that sat-
isfy certain conditions. Third, the criteria at issue in the linguistic prac-
tice are general (they allow one to identify, for each jurisdiction, what the
laws of that jurisdiction are). In contrast, the criteria in the legal practice
identify only the conditions for enforcing norms in the jurisdiction. They
are not the criteria for identifying law in some general sense at all.
For clear examples of Dworkin's conflation of the two practices,
consider the following descriptions, in Justice in Robes, of the rule of rec-
ognition:
Hart thought that in every community in which claims of law are
made the great bulk of officials of the community all accept, as a
kind of convention, some master rule of recognition that identifies
which historical or other facts or events make claims of law true. 62
But this is not a description of the rule of recognition. A community in
which claims of law are made is a community engaged in the linguistic
practice of using the concept of law. Or again: "Hart argued that every
legal system necessarily depends upon a master rule, or 'rule of recogni-
tion,' for identifying any and all valid propositions of law."63 Once again,
this is not a description of the rule of recognition. Criteria for identifying
valid propositions of law are criteria for employing the concept of law.
Dworkin has misdescribed the rule of recognition as the linguistic prac-
tice of employing that concept.
If the rule of recognition is treated as a linguistic practice, it does
indeed look like there is a connection between conventionalist metase-
mantics and conventionalist theories of law. The metasemantic demand
for agreement in the practice of employing the concept of law (on pain of
talking past one another) starts to look equivalent to Hart's demand that
there be agreement among officials concerning criteria for enforcement
(on pain of there being no law to enforce in that jurisdiction).
But a rule of recognition is not something that exists "in every
community in which claims of laws are made,'' nor is it a rule "for identi-
fying any and all valid propositions of law."64 Consider a community liv-
ing in anarchy in Somalia. In saying "We have no valid law here, but the
Securities Exchange Act is valid law in the United States," the members
of this community would be making claims of law or stating valid propo-
sitions of law. For this reason, metasemantic conventionalism would

63. /d. at 190; see also id. at 32, 165--66, 214. He makes the same claim in Law's Empire: "Se-
mantic theories [of law] suppose that lawyers and judges use mainly the same criteria (though they are
hidden and unrecognized) in deciding when propositions of law are true or false .... " DWORKIN, su-
pra note 11, at 33.
64. See supra text accompanying notes 62--63.
have a good deal to say about the rules that they use. Unless they shared
the same criteria for employing the concept of law, they would be talking
past each other. This metasemantic requirement would be satisfied, for
example, if Hart's The Concept of Law was accepted by people in Soma-
lia as a proper theory of law.
But their agreement concerning the criteria for employing the con-
cept of law would do nothing to create a rule of recognition. They have
no rule of recognition in Somalia, because officials are not actually en-
forcing norms on the basis of accepted conditions. Indeed, precisely be-
cause they all agree concerning the criteria for using the concept of law,
they would all conclude that there is no valid law in their jurisdiction at
all.6s
Because a rule of recognition is so different from the linguistic prac-
tice of using the concept of law, Dworkin finds it difficult to describe just
who the participants in this muddled linguistic/legal practice are. He of-
ten settles on lawyers, perhaps because they stand half-way between the
officials in a jurisdiction who are relevant for the rule of recognition and
the general language users who are relevant for conventionalist metase-
mantics. For example, in Justice in Robes, Dworkin claims that positiv-
ists like Hart assume that "analyzing (the concept of law) means bringing
to the surface the criteria that lawyers actually use, even if unselfcon-
sciously, in applying it."66
But this misdescribes both Hart's conventionalist theory of law and
conventionalist metasemantics. It distorts Hart's theory of law because
Hart did not think that lawyers were participants in a rule of recognition.
Assume that American officials (judges, legislators, sheriffs, regulators
and the like) took the Securities Exchange Act to be identified by the
American rule of recognition and enforced it accordingly. Hart would
say that the Securities Exchange Act was law and that lawyers who dis-
agreed were simply wrong-and were going to lose a lot of casesY
Metasemantic conventionalism is also distorted because the content
of a common concept of law is determined by the criteria agreed upon by
65. This is true even if one expands the agreement required in a conventionalist theory of law
beyond officials to include the general population in the jurisdiction. (On the question of whose prac-
tices are constitutive of a legal system, see Matthew Adler, Popular Constitutionalism and the Rule of
Recognition: Whose Practices Ground U.S. Law?, 100 Nw. U. L. REV. 719 (2006).) The point remains
that a population can agree concerning the concept of law and nevertheless not have the type of
agreement that is necessary for law according to a conventionalist theory of law. That has to be possi-
ble, for they must be able to agree that they have no law.
66. DWORKIN, supra note 4, at 31; see also id. at 225 (noting that positivist theorists of law as-
sume that the analysis of the concept of law must "consist in elucidating the tests that lawyers share,
except in borderline cases, for judging whether propositions of law are true"); DWORKIN, supra note
11, at 43 (noting that the project of semantic theories of law like Hart's is "digging out shared rules
from a careful study of what lawyers say and do").
67. Of course, because the primary rules that are valid according to the rule of recognition must
be generally-although not necessarily always-obeyed by the population, HART, supra note 6, at
116-17, it would probably be the case that most lawyers must obey the Act, however grudgingly. It
would not, however, be necessary that they think it is actually enforceable.
language users generally-not merely lawyers, much less lawyers within a

particular jurisdiction. Nothing about lawyers' views is essential to de-
termining the content of this concept. Although lawyers might know a
lot about the rule of recognition in their jurisdiction, they are not in a po-
sition superior to other language users in determining the content of the
concept of law-determining, for example, whether a rule of recognition
is at all relevant to what should be called "law."68 If metasemantic con-
ventionalism is true, lawyers' knowledge of the rule of recognition is
relevant to law only because those using the concept of law, which in-
cludes all language users, say it is.
Dworkin's confusion is evident in the centerpiece of his semantic
sting argument against Hart. Dworkin asks us to consider two judges
who come to different conclusions in a case because they disagree about
the criteria in the rule of recognition of their jurisdiction. Because he
confuses legal and linguistic practices, Dworkin assumes that these
judges must disagree concerning the criteria for using the concept of law
and so must be treated by the metasemantic conventionalist as having
different concepts of law. The metasemantic conventionalist must treat
the judges' dispute as "pointless in the· most trivial and irritating way, like
an argument about banks when one person has in mind savings banks
and the other riverbanks. " 69 This, Dworkin argues, would be a nihilistic
view about legal practice, since judges in hard cases do not think they are
talking past one another. 70
But this is a fallacy. Assume our judges all accept Dworkin's theory
of law. They will show that they share the same criteria for applying the
concept of law-and so are not talking past each other according to con-
ventionalist metasemantics-precisely by concluding that there is an an-
swer to the hard case despite their disagreement.
· By the same token, disagreement concerning the concept of law
does not have to lead to disagreement in a rule of recognition. Consider
two judges: Judge Lockean believes that valid laws must be promulgated
in accordance with a constitution to which the population has consented.
Judge Hartian holds Hart's theory of law. Judge Lockean considers
promulgation in accordance with the u:s. Constitution to be an appro-
priate criterion for American law because the Constitution received the
requisite consent when it was ratified in accordance with Article VII.
Judge Hartian also considers promulgation in accordance with the U.S.
Constitution to be an appropriate criterion for American law because it
is accepted by American officials. There is no reason not to conclude
that Lockean and Hartian participate in the same rule of recognition, as
68. Of course, it is possible that language-users' criteria for the concept of law specified that they
should defer to experts concerning the content and scope of the concept. Cf Coleman & Simchen,
supra note 29, at 10-11. But, once again, that would be true not because the experts said so, but be-
cause language users did.
70. /d.
they are both enforcing norms if those norms are promulgated in accor-
dance with the Constitution-even though, according to the metaseman-
tic conventionalist, they are talking past one another when they call the
norms they are enforcing "law.'m
Indeed, applying the concept of law is such a different activity from
participating in a rule of recognition that officials can have a rule of rec-
ognition without a concept of law at all. It is perfectly possible for an of-
ficial to enforce norms according to accepted criteria in her jurisdiction
without being at all aware of the general criteria that would allow her to
identify what should be called "the law" of various jurisdictions. She
might never have thought about-and be completely disinclined to think
about-how her jurisdiction and other jurisdictions have "laws" in some
common sense. This is no more impossible than someone who is able to
play the game of chess but has never come up with the concept of a
game. 72
Furthermore, even if an official has the concept of law, there is no
reason to think that she regulates her enforcement of norms on the basis
of whether they fall under the concept. She might care passionately
about abiding by the criteria in her rule of recognition-about enforcing
only those norms promulgated in accordance with the U.S. Constitution,
for example-but not care a whit about making sure that what she is en-
forcing is or is not law in a general sense. That is no more odd than the
idea of someone caring passionately about playing chess properly with-
out caring about whether what he is doing is a game.
C. Was Hart a Metasemantic Conventionalist After All?

As we have seen, Dworkin's argument that Hart's theory of law was
the consequence of his conventionalist metasemantics remains a fallacy
even if Dworkin was right that Hart was indeed a metasemantic conven-
tionalist. It is important to keep this point in mind because the question
of Hart's metasemantic views is a matter of some debate.
For example, in The Concept of Law Hart argues that legal inde-
terminacy results from the "irreducibly open-textured" nature of the lan-
guage in which laws are formulated. A law prohibiting "vehicles" in the
park, for example, leaves it open whether bicycles or roller skates are in-
cluded under the concept of a vehicle.73 The concept is indeterminate on
this matter, Hart appears to suggest, because the criteria we currently ac-
cept for using the concept do not answer the question. 74
71. For a similar argument, see COLEMAN, PRACfiCE OF PRINCIPLE, supra note 6, at 181.
72. For a similar argument, see Kenneth Einar Himma, Ambiguously Stung: Dworkin 's Semantic
Sting Reconfigured, 8 LEGAL THEORY 145, (2002).
74. Paradigm samples (such as motor cars) clearly fall under the term because we agree that they
do. /d. at 129.
1496 UNIVERSITY OF ILLINOIS LAW REVIEW (Vol. 2007
Nicos Stavropolous has argued that Hart's metasemantics was con-

ventionalist on these grounds. 75 He claims that Hart's metasemantic
conventionalism can be found not merely in his approach to concepts-
like the concept of a vehicle-employed in legal rules, but also in his ap-
proach to the concept of law itself.16 In arriving at his theory of law, Hart
sought to capture the criteria that people using the concept currently ac-
cept. Joseph Raz has come to the same conclusion about Hart,77 al-
though several other philosophers disagree. 78
Dworkin cannot, therefore, be accused of refusing to acknowledge a
clear error in insisting that Hart was a metasemantic conventionalist.
But the fact remains that Dworkin is clearly in error in assuming that
Hart's conventionalist theory of law was a consequence of his metase-
mantic conventionalism.
D. The Influence of Metasemantics on the Scope of the Law

One more qualification is necessary in identifying Dworkin's mis-
taken argument. Even though conventionalist theories of law do not fol-
low from conventionalist metasemantics, there is reason to believe that
conventionalist metasemantics makes a difference to one's understand-
ing of the scope of the law within a conventionalist theory of law. As-
sume one has already adopted Hart's theory of law. According to this
theory, the law of a jurisdiction is picked out by the criteria in the rule of
recognition. But whether an item satisfies these criteria depends upon
the contents of the concepts employed in the criteria. Since metaseman-
tics makes an important difference to our understanding of the contents
of these concepts, it should make a difference to what counts as law.
Consider a crude rule of recognition, under which every norm on a
particular tablet, and nothing else, is enforceable. If conventionalist me-
tasemantics applied to the concept on the tablet, the scope of that concept
would be exhausted by language-users' shared criteria. For this reason, a
norm could not be on the tablet if there was fundamental disagreement
among language users on this matter. On the other hand, if nonconven-
tionalist metasemantics were correct, there could still be an answer to the
question of whether a norm was on the tablet (and so was law) despite
fundamental disagreement on the matter.
Assume further that the command "Do not steal" appears on the
tablet. If conventionalist metasemantics applies to the concept of steal-
ing, then, once again, something cannot be stealing if language users dis-
75. Stavropoulos, supra 26, at 59.

76. /d. at 67-69.
77. Raz, supra note 29, at 1, 1-27.
78. See, e.g. , Timothy A.O. Endicott, Herbert Hart and the Semantic Sting, in HART'S
POSTSCRIPT: ESSAYS ON THE POSTSCRIPT TO THE CONCEPT OF LAW, supra note 6, at 39: Veronica
Rodriguez-Blanco, A Defence of Hart's Semamics as Nonambitious Conceptual Analysis, 9 LEGAL
THEORY 99 (2003).
agree about whether it is stealing. On the other hand, if nonconvention-

alist metasemantics applies to the concept, it remains possible that the
act is stealing despite disagreement on the matter. 79
Stavropoulus has offered an interpretation of Dworkin's metase-
mantic critique of Hart that emphasizes these points: According to
Dworkin, Hart's metasemantic conventionalism led him to conclude that
there was no answer to hard cases that resulted from disagreement about
the proper criteria for using a concept. For example, Hart was forced to
conclude that it is legally indeterminate whether the rule prohibiting
"vehicles" in the park applied to bicycles because language users dis-
agree about whether bicycles are vehicles. 8° Furthermore, nonconven-
tionalist metasemantics of the sort embraced by Dworkin does a better
job accounting for judges' attitudes in such hard cases, because judges
engaged in disagreement about whether a concept applies feel that there
is an answer despite their disagreement. 81
In fact, one can be skeptical about whether metasemantics will have
such an effect on the resolution of these hard cases. It is true that non-
conventionalist metasemantics makes it possible for an act to fall under
the concept of stealing even if there is disagreement on the matter. But it
does not mandate that the concept refer in this way. It is always possible
for us to generate a criteria! concept of stealing if we want to. A meta-
semantic theory cannot prohibit me from thinking only of what people
agree is stealing. So, the possibility will always remain that a concept
employed in the law is criterial.
Second, one might argue that the scope of concepts in laws is itself a
question of law. Even though the tablet's drafters meant stealing in a
metasemantically nonconventionalist sense, it may be the law that any-
one interpreting the command should look only to what is currently
agreed to be stealing (or what the drafters agreed to be stealing). One
reason for this rule may be that such a limitation on interpretation brings
with it predictability. Because the appropriate interpretation of laws is
itself a legal question, it is difficult to see how metasemantics can have an
influence upon it at all. 82
The fact remains, however, that even if metasemantics does have
this influence on the scope of the law within Hart's theory of law,
Dworkin's argument that Hart's theory of law itself followed from con-
ventionalist metasemantics remains a fallacy. The metasemantic conven-
tionalist is free to insist that the best account of the criteria we associate
with the concept of law is a nonconventionalist theory of law like
79. See David 0. Brink, Legal Theory, Legal Interpretation, and Judicial Review, 17 PHIL. & PUB.
AFF. 105 (1988).
81. Stavropoulos, supra note 26, at 59, 61.
82. See Brian Bix, Can Theories of Meaning and Reference Solve the Problem of Legal Determi·
nacy?, 16 RATIO JURIS 281 (2003); Green, supra note 35, at 1946-48.
Dworkin's, according to which the criteria for the valid laws in a jurisdic-
tion are not limited to what officials in the jurisdiction consider to be
law. 83
By the same token, Dworkin's claim that metasemantic convention-
alism forces us to conclude that judges who disagree about the criteria of
legality in their jurisdiction must be talking past one another remains
false. It is true that the metasemantic conventionalist might have to treat
judges disagreeing about whether an item falls under a certain concept
(such as "stealing" or "vehicle") as talking past one another. These are
the examples of hard cases that Stavropoulos emphasizes. But Dworkin
emphasizes hard cases that result from judges disagreeing about the cri-
teria in the rule of recognition. And, as we have seen, there is no reason
to think that this type of disagreement means that judges have different
concepts of law and so are talking past one another. For judges might
share a concept of law-like Dworkin's-that embraces such disagree-
ment.
E. Dworkin Is All Alone

Dworkin's argument that Hart's conventionalist theory of law was
the result of conventionalist metasemantics has been widely and consis-
tently rejected by philosophers of law.84 Furthermore, these philosophers
have all agreed that it is a fallacy for the reasons I have outlined above.
Dworkin's mistake is assuming that a legal practice like the rule of rec-
ognition is equivalent to the linguistic practice of using the concept of
law. As Timothy Endicott has put this point: "Although [Hart] did claim
that legal systems are based on rules of recognition, he did not say that
those rules are linguistic rules .... " 85 Endicott correctly describes this as
a "simple and compelling" objection to Dworkin's argument.86
Because these linguistic and legal practices are different, a metase-
mantic demand for agreement in the linguistic practice (in order for
there to be a shared concept of law) has nothing to do with the sort of
agreement in the legal practice demanded by conventionalist theories of
83. Indeed, at one point in Justice in Robes, Dworkin suggests he is such a person. He argues
that interpretive concepts, like the concept of law as he understands it, "require that people share a
practice: they must converge in actually treating the concept as interpretive." DWORKIN, supra note 4,
at 11. Dworkin appears to argue that the concept of law has the content to which he assigns it because
people agree that it does.
The same idea pops up in Law's Empire. He argues that the concept of law is interpretive because
"(j]udges normally recognize a duty to continue rather than discard the practice they have joined. So
they develop, in response to their own convictions and instincts, working theories about the best inter-
pretation of their responsibilities under that practice." DWORKIN, supra note 11, at 87. Once again,
Dworkin suggests that language-users' current attitudes are what determine the contents of their con-
cepts.
84. I rejected Dworkin's argument in Green, supra note 35, at 1927-29.
85. Timothy Endicott, Law and Language, in THE STANFORD ENCYCLOPEDIA OF PHILOSOPHY
(Edward N. Zalta ed., Winter 2007 ed.), http://plato.stanford.edu/entries/law-language/.
86. /d.
No.5) DWORKIN V. THE PHILOSOPHERS 1499
law. As Ken Himma has put this point: "Dworkin conflates the claim
that the concept of law can be explained in terms of shared criteria with
the claim that the grounds of law are exhausted by shared factual criteria
of legal validity."87 Or as Dennis Patterson has put it: "Dworkin confuses
the view that the content of the rule of recognition is determined by
shared criteria, which is a positivist view, with the view that the meaning
of the word 'law' is determined by shared criteria. The latter is no part of
the theory of positivism."88 Not only can officials share the same criteria
in a rule of recognition while having different concepts of law, they can
share a rule of recognition without having a concept of law at all. As Jo-
seph Raz has put it: "The point is that Uudges') duty (under the system in
whose courts they sit) is to judge in accordance with the rules of that sys-
tem, and it matters not at all whether these rules are legal ones ....
[T]here could be legal systems in cultures that do not have the concept of
law."89
Conversely, the fact that there is disagreement in a rule of recogni-
tion does not mean that officials are employing different concepts of law
and so talking past one another. As Hart himself put it, Dworkin was
mistaken to think that "if the criteria for the identification of the grounds
of law were not uncontroversially fixed, 'law' would mean different
things to different people."90
Is it really true that Dworkin is all alone against these philosophers?
Does he have no allies? I do not know whether Dworkin has found some
private defenders. But I have found no one defending the argument in
print. Furthermore, the two allies that Dworkin points to in Justice in
Robes, abandon him here.
The first is Nicos Stavropoulos. The essay by Stavropoulos that
Dworkin cites as defending his interpretation of Hart has been discussed
above. 91 In it, Stavropoulos accuses Hart of metasemantic conventional-
ism and argues that because of this metasemantics, Hart cannot account
for why judges disagreeing about the applicability of a concept (such as
the concept of a vehicle) in a hard case are not talking past one another. 92
But Stavropoulos never suggests that Hart held a conventionalist theory
of law because he had a conventionalist metasemantics, nor does he ar-
gue that disagreement over the criteria in a rule of recognition means that
those disagreeing hold different concepts of law.
87. Himma, supra note 72, at 165.

88. Dennis M. Patterson, Dworkin on the Semantics of Legal and Political Concepts, 26 OXFORD
J. LEGAL STUD. 545, 546 n.7 (2006).
89. Raz, supra note 29, at 1, 35.
91. See Stavropoulos, supra note 26, at 59. Dworkin mentions this essay at DWORKIN, supra
note 4, at 288 n.39.
92. Stavropoulos, supra note 26, at 59.
The second ally is Stephen Perry. 93 But, once again, in the essay
Dworkin cites, Perry does not defend Dworkin's argument that Hart's
conventionalist theory of law follows from conventionalist metaseman-
tics. Indeed, the topic does not come up in the essay at all. 94
F. Dworkin's Response
So how does Dworkin, besieged and friendless, respond to these
criticisms? His most pointed response is in chapter seven of Justice in
Robes, where he reviews Jules Coleman's book The Practice of Principle.
In this book, Coleman criticized Dworkin's "semantic sting" argument
for the same reason the rest of the philosophers have, because Dworkin
confuses agreement in a rule of recognition with agreement in the
linguistic practice of using the concept of law. As Coleman puts it,
people "can use different factual criteria for determining whether
something is legally binding without disagreeing about the meaning or
concept of law. " 95
93. See Stephen R. Perry, Hart"s Methodological Positivism, in HART'S POSTSCRIPT: ESSAYS ON
THE POSTSCRIPT TO THE CONCEPT OF LAW, supra note 6, at 311. Dworkin mentions this essay at
DWORKIN, supra note 4, at 31 & 265 n.23.
94. Indeed Perry suggests, contrary to Dworkin, that Hart was not a metasemantic conventional-
ist. Rather, he engaged in a form of conceptual analysis very similar to Dworkin's own approach.
Perry, supra note 93, at 312-13.
It is worth noting that in a review of Coleman's book, Perry acknowledges that Coleman's critique
of Dworkin's semantic sting argument is correct:
Coleman asserts it is possible to disagree about the content of the concept of law while agreeing
about the criteria of legality in a particular legal system. He also asserts, more importantly, that it
is possible to agree about the content of the concept while disagreeing about the criteria in a par-
ticular system. He is clearly right on both counts. He is also right that in Law's Empire Dworkin
did not distinguish as clearly as he should have done between the criteria for applying the term
"law" and the criteria of legality in particular legal systems.
Stephen R. Perry, Method and Principle in Legal Theory, 111 YALE L.J. 1757, 1800-01 (2002). He
agrees with Coleman that two Dworkinians can be understood as agreeing concerning the content of
the concept of law even though they disagree about the criteria of validity within their legal system.
Perry does offer a different interpretation of Dworkin's argument, however, according to which
"this carelessness on Dworkin's part may well be nothing more than harmless error." /d. at 1801. As
Perry notes, it is clearly possible for there to be theoretical disagreement about the criteria of validity
in a legal system, that is, disagreement that is motivated by different theories of law. Perry offers as an
example the disagreement between inclusive and exclusive legal positivists:
It follows that we cannot determine which theory of law we should adopt simply by looking for
agreement about the content of the concept of law, because there is no such agreement. In this
sense, we can agree with Dworkin's rejection of "semantic" theories (without, it should be noted,
having to accept his claim that substantive positivist theories have heretofore all been semantic in
character).
/d. at 1802. Perry's point is that the disagreement between those offering different theories of law sug-
gests that conventionalist metasemantics cannot be true. For these people do not feel as if they are
talking past one another. But this is not all that there is to Dworkin's argument. Dworkin does not
merely suggest that conventionalist metasemantics cannot explain theoretical disagreement concerning
the law. On that matter Coleman agrees. Dworkin also argues that conventionalist theories of law
like Hart's follow from conventionalist metasemantics. Perry refuses to sign on to this part of
Dworkin's argument. He refuses to accept "[Dworkin's] claim that substantive positivist th.::ories have
heretofore been [meta]semantic in character." /d.
95. COLEMAN, PRACTICE AND PRINCIPLE, supra note 6, at 181.
To show why this is the case, Coleman asks us to consider two peo-
ple who disagree concerning the criteria for enforcement in their jurisdic-
tion and who nevertheless insist that there is law answering a hard case.
It does not follow that they disagree concerning the content of the con-
cept of law. For, as we have seen, they may agree on a nonconventional-
ist theory of law, like Dworkin's, that allows for right answers even in the
face of their disagreement. As Coleman puts it:
Suppose, for example, that we share the view that law is a contest-
able concept in the sense that wherever there is law, what the law is
is always a matter of potential dispute, and requires an interpretive
practice. Indeed our disagreement about what the criteria of legal-
ity in our community are makes perfectly good sense to us in part
because such disagreement is part of what we take law to be-part
of our shared understanding of the kind of thing it is. Thus, not
only is disagreement about the criteria of legality in our community
compatible with our sharing the same criteria for applying the con-
cept of law, in this case our disagreement about the criteria of legal-
ity in our community is intelligible to us just because we share the
same criteria for applying the concept.96
In short, a Dworkinian theory of law can be the content of a criteria!
concept.
Dworkin begins his reply by noting that both he and Coleman reject
conventionalist (or criteria!) metasemantics:
Lawyers share the concept of law as what I call an interpretive (or
essentially contested) concept. They do not agree on criteria for
applying the claim "it is the law that," but rather offer rival
interpretations of paradigm propositions of that form that they both
accept, and then extract from these paradigm propositions different
criteria for applying the judgment "it is the law that" to fresh cases.
about which they disagree. Coleman ... also endorses this view
about how lawyers employ the concept of law. 97
Dworkin is correct that Coleman rejects conventionalist metasemantics.
Coleman agrees that the true content of the concept of law cannot be
simply read off the accepted criteria for using the concept. People can
share a concept even though they disagree on criteria for its use.
But Dworkin takes his metasemantic agreement with Coleman to
mean that the people in Coleman's example disagree about the criteria
for using the concept of law. This is false. It is true that Coleman
believes that agreement on the criteria for using the concept of law is not
necessary for sharing the same concept of law. But it hardly follows from
this that people can never agree on these criteria. And Coleman is
offering an example of two people who in fact do agree on these criteria.
96. /d. at 182.

The point of his example is to show that it is possible for these people to
disagree in hard cases despite this other form of agreement.
Having understood Coleman's example as concerning people who
disagree concerning the criteria for using the concept of law, Dworkin
finds it easy to dismiss:
So what is my "fundamental" -or, as he also says, "deeply embed-
ded" -confusion? His explanation comes to this: if I am right that
lawyers share the concept of law as an interpretive concept, then,
contrary to what I say, they actually do agree about criteria for ap-
plying the concept. They agree that it should be applied in the in-
terpretive manner that I described. But this explanation misunder-
stands what I meant when I said that lawyers do not share criteria; I
clearly meant that they do not agree on a single set of tests for de-
ciding which propositions of law are true. Since Coleman agrees, it
remains mysterious why he supposes that I am confused.98
But Coleman does not accept that the people in his example "do not
agree on a single set of tests for deciding which propositions of law are
true." As he made clear, they "shar[e] the same criteria for applying the
concept of law" and have a "shared understanding of the kind of thing
[the law] is." 99 True, Coleman himself thinks that agreement on criteria
is not necessary for sharing a concept. But the point of his example is
solely to show that disagreement in a hard case does not entail disagree-
ment about criteria for using the concept of law.
Of course, by saying that the two people "do not agree on a single
set of tests for deciding which propositions of law are true," Dworkin
might mean that they do not agree on the enforcement criteria in their
jurisdiction. This is indeed something that Coleman accepts about his ex-
ample. But the fact remains that despite- indeed because of- this dis-
agreement concerning the rule of recognition, the people in his example
nevertheless agree concerning the criteria for applying the concept of law
because they share a nonconventionalist theory of law, like Dworkin's, in
which the law is whatever follows from the best moral interpretation of
standing lawY)() The fact that they disagree in a hard case would not
mean that they assign different contents to the concept of law. This
shows that Dworkin's semantic sting argument is wrong. And Dworkin
has managed to say nothing in response.
CONCLUSION
Despite his fame, Ronald Dworkin is something of a tragic figure in

the history of the philosophy of law. The intransigence of his defense of
the semantic sting argument falsely suggests that the argument is some-
98. /d. at 221-22.

99. COLEMAN, PRACTICE AND PRINCIPLE, supra note 6, at 182.
100. ld. at 181.
how crucial to his rejection of Hart. In fact, Dworkin has offered other
powerful arguments against Hart's theory of law, arguments that appeal
to Hart's inability to arrive at a coherent account of legal obligation. I
have indicated my own sympathies to this part of Dworkin's critique of
Hart elsewhere. 101 But I'm afraid that until Dworkin abandons his se-
mantic sting argument, his critique of Hart-and his philosophy of law as
a whole-are not going to get the hearing they deserve.
101. See Michael Steven Green, Legal Realism as Theory of Law, 46 WM. & MARY L. REV. 1915,
1939-56 (2005).

Dworkin V Philo

Uploaded by

Copyright:

Available Formats

Dworkin V Philo

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

Dworkin V Philo

Uploaded by

Copyright:

Available Formats

College of William & Mary Law School

William & Mary Law School Scholarship Repository

Dworkin v. e Philosophers: A Review Essay on

In this review essay, Professor Michael Steven Green argues that

Justice in Robes, by Ronald Dworkin. Belknap Press of Harvard Univer-

Ronald Dworkin's theory of law has not been adequately appreci-

law have declared their debt to Raz, the number of self-described

Dworkin's stubbornness is sadly in evidence in Justice in Robes,13 his

I. THE SEMANTIC STING ARGUMENT

In his semantic sting argument, Dworkin claims that H. L. A. Hart's

A. Conventionalist and Nonconventionalist Theories of Law

13.DWORKIN, supra note 4.

In Hart's theory of law, the general criterion is acceptance by offi-

Let us call a theory of law conventionalist if it looks to the currently

B. Philosophy as Conceptual Analysis

24. !d. at 144.

sential existence conditions for these things-that is, knowledge of what

27. E.g., Stavropoulos, supra note 26, at 59, 64.

mantic" the view that a concept is "criterial." To say that a concept is

all. Dworkin argues, however, that conventionalist metasemantics is im-

D. Dworkin on the Relationship Between (Meta)semantic

45. !d. at 153.

true. I called that assumption- that all concepts are criteria!-the

II. DWORKIN'S ARGUMENT FAILS

52. /d. at 225-26.

A. Conventionalist Metasemantics Does Not Generate Conventionalist

The error in Dworkin's argument is his assumption that because the

B. Why Does Dworkin Commit His Fallacy?

59. HART, supra note 6, at 246.

the legal practice is one in which only officials in a particular jurisdiction

62. DWORKIN, supra note 4, at 163.

language users generally-not merely lawyers, much less lawyers within a

C. Was Hart a Metasemantic Conventionalist After All?

Nicos Stavropolous has argued that Hart's metasemantics was con-

D. The Influence of Metasemantics on the Scope of the Law

75. Stavropoulos, supra 26, at 59.

agree about whether it is stealing. On the other hand, if nonconvention-

E. Dworkin Is All Alone

87. Himma, supra note 72, at 165.

96. /d. at 182.

Despite his fame, Ronald Dworkin is something of a tragic figure in

98. /d. at 221-22.

You might also like

Dworkin v. e Philosophers: A Review Essay on