In this paper, I set out to argue that we should stop talking about ‘legal facts’ in general jurisprudence. This goes especially for legal positivists who seem to think that positivism is a theory about the way ‘legal facts’ depend on—are determined by, supervene on, are grounded in—social or moral facts. This is a very neat way to frame the issues at hand: It is all about facts of various sorts. It is facts all the way down. Neat as it is, I will argue that this framing should be resisted because as such it is a Trojan Horse. The point about the Horse was that it was meant to be confusing: The Trojans were legitimately puzzled about what the Horse was for (was it a gift? an offering? a supernatural apparition?) and, out of such confusion, they allowed it in.
I will argue that the same goes for ‘legal facts’ as they were introduced in jurisprudential debates by prominent non-positivists (especially Mark Greenberg). ‘Facts’ about law (or facts about the ‘content’ of law) are of very different kinds. It is reasonable to assume that some facts about law can only be ascertained via recourse to moral considerations. So, if positivism is about ‘legal facts’ and at least some ‘legal facts’ depend on ‘moral facts’, then positivism is indefensible. Moreover, ‘legal facts’ are often used as a Trojan horse for a holistic conception of legal content which is, in fact, incompatible with positivism’s core tenets: Positivists are therefore asked to provide a justification for their theses in a framing which is overtly hostile to them. Positivism can be defended as a jurisprudential tenet if and only if we show that it was never about ‘legal facts’ per se, but about a specific subset of ‘facts’ about law, and that allowing the issue to be framed in terms of ‘legal facts’, period, is just a self-inflicted blow.
1. Introduction: On Framing Debates in Legal Philosophy
When you are on the losing end of a debate, a fruitful strategy is to change the way the debate is framed. Your opponent is then faced with the following dilemma: Either argue against the framing itself—and then lose track of the original debate—or try to win the debate on the new terms you just imposed onto it. Whatever they choose, you are more likely than not to win. Contemporary analytical jurisprudence is a good case in point.
For most of the twentieth century, legal positivists of all stripes framed the debate with their opponents in terms of the law being a system of norms—especially normative rules—whose existence is explained in terms of their (systemic) validity.Footnote 1 Now, ‘validity’ is a very ambiguous word, and there has been a constant disagreement among positivists of various stripes; however, it could be argued that the smallest common denominator meaning of validity is that of ‘membership’.Footnote 2 PositivistsFootnote 3 share the core idea that whether a norm belongs to a legal system—and therefore is a legal norm—depends on social facts of a certain kind, and is not predicated on the value or merit (moral or otherwise) of said norm.Footnote 4 This notion is captured by the Austinian motto: “The existence of law is one thing; its merit or demerit is another.”Footnote 5 The Austinian maxim is still too vague: the “existence of law” can refer both to the existence of a legal system as well as the existence of particular laws (or legal norms). Many, if not all, positivists adopt an atomistic conception of law: The basic story that positivism embraces is that law (the legal system) is made up of a discrete set of norms—or normative rules—identified through specific criteria of validity.Footnote 6 The way this story was framed meant that their opponents had to show that moral (or otherwise evaluative) criteria must be used in order to ascertain the membership of legal norms—that is, their legal quality, or their character as laws.
Non-positivists were to some extent prisoners of such a way to frame the debate. Gustav Radbruch for instance, had to show that the injustice of a given law had, if extreme enough, an impact on its validity.Footnote 7 The same goes for Lon Fuller, who went out of his way to show that a legal system’s “internal morality” was a key component of the validity of its laws.Footnote 8 Robert Alexy, the leading non-positivist on the Continent, frames his typology of non-positivisms (exclusive, inclusive, super-inclusive) in terms of their views regarding the moral components of legal validity.Footnote 9
Some tried to escape this way of framing the debate. John Finnis accepted the basic tenet that immoral laws are still laws, while arguing that it was a trivial, and not quite interesting, fact about law: What was not trivial, in his view, was that law’s point (as a whole) was to perform a moral function, and that an immoral law, while still being technically legally valid, was otherwise defective in terms of its moral function.Footnote 10 As Hart argued, this was all very interesting, but beyond the point: As far as he was concerned, since Finnis granted the basic point about validity, his natural law theory was largely compatible with legal positivism.Footnote 11
Enter Ronald Dworkin. At first, Dworkin tried to attack positivism from within the positivistic picture, so to speak. In the “Model of Rules” I and II, he accepted the basic story put forward by Hart and others, that law was a system of norms; thus, his attack rested on the distinction between two kinds of norms, pedigreed ones (rules) and non-pedigreed ones (principles).Footnote 12 The two kinds of norms behave differently and only the former fits the positivistic picture. But ultimately, Dworkin realised he had to change course—perhaps because positivists can very well account for the rule/principle distinctionFootnote 13 —and he sought to reframe the debate altogether. In “No Right Answer?” and in Law’s Empire (where the rule/principle distinction does not appear once), Dworkin put forward a holistic, amorphous picture of what law is.Footnote 14 According to Dworkin, law, being an interpretive concept, is not made up of a discrete set of norms, whose criteria of validity—their ‘grounds’—would have to be explained in terms of either their social sources or their moral value. Law is nothing but an interpretation, or rather a chain of interpretations. What makes ‘legal propositions’ true is not relevant facts about the validity or the content of given norms; it is the best (in terms of both fit and moral value) interpretation of past materials identified as relevant at the pre-interpretive stage.
Now, Dworkin had a lot of followers; but he could not force positivists to accept his framing. Post-Hartian positivists addressed the challenge mounted by Dworkin by refining their key concepts (validity, rule of recognition, etc.); but they did not try to present a positivist version of Dworkinian interpretivism:Footnote 15 they just did not accept the framing itself.Footnote 16 To some extent, Dworkin was victorious insofar as he managed to push positivists into intestine and byzantine debates (such as the inclusive-versus-exclusive-positivism debate). But his victory was a Pyrrhic one: No positivist agreed interpretivism was a viable meta-philosophical option.
This excursus into the recent history of analytical jurisprudence shows that, until recently, no attack on legal positivism had succeeded in changing radically the terms of the debate. That is, until Mark Greenberg published “How Facts Make Law” (here-after HFML) I and II.Footnote 17 In these papers, as well as his more recent ones culminating in the articulation of his moral impact theory, Greenberg presents the issue in a distinctively new way.Footnote 18 In HFML I, he introduces the notion that “legal content” (on which more later) depends not only on “descriptive facts” or “social facts”—that is, facts about specific practices, which he calls “law practices,” involving certain social agents, such as legal authorities and officials—but also on “normative” facts or “value facts.”Footnote 19 Interestingly enough, the phrase “legal facts” appears only once in HFML I, and in a rather cursory way.Footnote 20 In HFML II, however, ‘legal facts’ are everywhere: They are first introduced as the same thing as legal content;Footnote 21 then as “true legal proposition[s],” and in the same breath as “fact[s] about the content of the law.”Footnote 22 I will come back to these different definitions later. For now, what is important is that, especially in HFML II, Greenberg sets out to use the legal fact/legal content formulation to rephrase central theses in contemporary legal positivism. Thus, the rule of recognition is “understood as constitutive of legal facts: … a model of how law practices contribute to the content of the law is legally correct in virtue of the existence of a rule of recognition specifying that model.”Footnote 23 Greenberg nevertheless concedes that “Hart would not put the point this way.”Footnote 24
Not only would Hart not put the point that way, but virtually no legal positivist ever put any of their (numerous, sometimes conflicting) points that way. As I will explain further in section 2, some positivists had already used vague and confusing formulations, such as ‘the existence and content of law depends on social sources’. But as I will show, even positivists using this kind of formulation did disambiguate it in their writings, usually endorsing the atomistic picture mentioned above. No positivist before HFML II had, to my knowledge, framed the main positivist tenets in terms of the relationship between ‘legal facts’ and moral or social facts.Footnote 25
This makes it even more puzzling that many self-avowed legal positivists—as well as many people writing about legal positivism—jumped on the ‘legal facts’ bandwagon. In his 2009 article on the rule of recognition, Scott Shapiro was still using the usual terminology (legal rules, criteria of validity, etc.); two years later, he explicitly endorsed the “legal fact[s]” phrase, expressly attributing its adoption to Greenberg’s influence.Footnote 26 He then went on to characterize the position of legal positivists as follows: “all legal facts are ultimately determined by social facts alone.”Footnote 27 After seeming initially reluctant, Kevin Toh now uses the ‘legal facts’-based characterization of legal positivism.Footnote 28 So do Jules Coleman,Footnote 29 Michael Green (reinterpreting Kelsen),Footnote 30 Barbara Baum Levenbook, Gideon Rosen, David Plunkett, Tomasz Gizbert-Studnicki, Samuele Chilovi, Bill Watson, Felipe Jiménez, Mitch Berman, and many others.Footnote 31 I think it is a serious mistake, and this paper aims to show why.
I have two main objections, which I am going to spell out in the three main sections of this paper: First, the category of ‘legal facts’ is overbroad—as it lumps together many types of ‘facts’ about law, some with a different status with regard to the justifiability of jurisprudential takes on law—and it surreptitiously invites positivists to adopt a holistic conception of law which is alien to core positivistic tenets. Second, the ‘legal facts’ picture draws not only positivists but all legal philosophers into a dispute which is ultimately metaphysical, inviting them to situate some queer entities (‘legal facts’) within the broader catalogue of the world. I will argue that such a move is, for positivists, self-defeating—and should be resisted by legal philosophers of all stripes.
2. Varieties of ‘legal facts’
2.1. ‘Facts’ about law
Scott Shapiro has an extensive notion of the kinds of entities that qualify as legal facts: “A legal fact is a fact about either the existence or the content of a particular legal system. It is a legal fact, for example, that Bulgaria has a legal system. Similarly, it is a legal fact that the law in California prohibits driving in excess of 65 miles per hour.”Footnote 32 I will come back later to the ‘existence and content’ phrase. But suffice it to say that, according to Shapiro, legal facts are meant to cover both the fact that a given legal system—such as Bulgaria’s—and facts about given laws—such as the speed-limit rule in California—do exist. Greenberg on the other hand firmly equates legal facts with “the content of the law” or “fact[s] about the content of the law.” By that definition, facts about the existence of legal systems are excluded. Now the content of the law “consists at least of all the general legal obligations, rights, privileges, and powers that exist in the legal system (at the time).”Footnote 33 It is clear that Greenberg rejects norm-atomism (Shapiro’s case is less clear): In the Greenbergian picture, obligations, rights, etc., are not the content (or the result) of norms, but free-floating entities of the world.Footnote 34 The ‘legal fact’ category is therefore non-redundant with regard to other jurisprudential categories (such as norms).
Let me be clear from the outset that I do not intend to get into the debates on the metaphysical nature of ‘facts’.Footnote 35 I will not aim to ascertain whether facts are entities or relations between properties and entities; I will not venture into ascertaining whether facts are true propositions (truth-bearers)Footnote 36 or whether they are what make propositions true (truth-makers).Footnote 37 With some notable exceptions, people writing about legal facts—even about the metaphysics of such facts—do not spend too much time trying to adjudicate these debates.Footnote 38 Dworkin, for instance, arguably conceives “propositions of law” as truth-bearers, but their truth-makers (the so-called “grounds of law”) are not facts, but other (more fundamental) propositions.Footnote 39 In an analogous vein, Greenberg equates “facts” with “true propositions.”Footnote 40 On the other hand, Plunkett clearly seems to define facts as truth-makers.Footnote 41 For now, I will presuppose a very thin notion of facts as bare truth-makers: Facts are the stuff that makes propositions true. But I will not be enquiring into the exact nature of facts, or even if facts exist, because it would be beyond the point. Although some variants of fact-eliminativism have been emerging,Footnote 42 this paper is not going to argue that legal facts do not exist, but rather that ‘legal facts’ are too broad a category to usefully frame the positivist/anti-positivist debate. To be quite explicit: ‘Legal facts’ are used to allow for a non-atomistic conception of legal content to be used as a default assumption in debates between positivists and non-positivists—a strategy that positivists should resist.
Consider the following propositions about law:
(1) It is a fact that legal system S exists.
(2) It is a fact that legal norm N belongs to legal system S.
(3) It is a fact that legal norm-formulation NF belongs to system S.
(4) It is a fact that N provides that thieves ought to be hanged.Footnote 43
(5) It is a fact that NF reads as “thieves ought to be hanged.”
(6) It is a fact that (in S) legally, thieves ought to be hanged.
(7) It is a fact that Jones is a thief.
(8) It is a fact that Jones ought to be hanged.
It seems that even if some of these propositions can be prima facie thought of as extensionally equivalent, none of them are intensionally equivalent. They mean different things. Therefore, their truth-makers may be different ontological kinds.
As propositions (2)-(8) show, there are basically four ways of thinking about legal content: legal content as norms, legal content as norm-formulation, legal content as normative content, and legal content as normative states of affairs. I am going to suggest that contrary to what it may seem, these four kinds of facts are not equivalent.
2.1.1. (Facts about) norms
Propositions (2) and (4) are about norms . Since I intend to suggest that classical debates in legal philosophy rest on an atomistic conception of norms, it may not be ill-advised to say a few words about ‘norms’ qua specific entities. Now, I am well aware that the nature of norms has been debated for decades. For instance, deontic logicians and jurisprudes have debated what makes norms the way they are—and especially if what makes them ‘normative’ is their semantic features (especially the semantic features of deontic operators) or the pragmatic features of norm-formulations. According to an ‘expressive’ conception of norms, such as that of Alchourrón and Bulygin, norms are speech-acts (such as commands or imperatives) whose propositional content remains identical across pragmatic contexts (descriptive and normative): A norm is a sentence formulated in the imperative, which commands a certain proposition.Footnote 44 What makes the norm ‘normative’ is not its propositional content (which Alchourrón and Bulygin call “norm-content”), but the fact that it has been commanded.Footnote 45 Semantic conceptions of norms, on the other hand, treat normative operators as semantically sui generis: Norms are propositions, whose normativity is a feature of the very propositional content which makes up the meaning of the normative sentence.Footnote 46 According to the semantic conception, norms are distinct kinds of propositions, whose normativity cannot be reduced to the pragmatic factors (whether the sentence’s illocutionary force is, for example, imperative rather than merely providing a suggestion). Both conceptions usually agree that norms are typically not truth-apt. Footnote 47 Whenever a norm is issued (such as the norm which provides that you ought to close the door), the norm may be satisfied but cannot be verified. This is so because they have a world-to-word direction of fit; whether this is due to semantic or pragmatic features of normative formulations is not something I will adjudicate here.
Indeed, I do not intend to settle this debate, or any related debate about the nature of norms. I will assume throughout that norms are at least related to propositional entities—something both sides of the debate agree on, even if semantic theories treat norms as propositions, whereas expressive conceptions call norms the very speech-acts which ‘command’ such-and-such proposition (the proposition being itself normatively inert). Rather I intend to avoid this debate altogether by going oblique—that is, by focusing on propositions which aim at describing norms. Footnote 48 One must distinguish between norms—whatever they are—and normative propositions—that is, descriptive propositions about norms, such as (2) and (4).Footnote 49 Norms are the truth-makers of normative propositions: So (2) and (4) are true if N exists. As Raz puts it (re: rules, but the point can be generalized to norms in general): “[Rules] are things the content of which is described by some normative statements.”Footnote 50 In what follows, I will adopt such an oblique approach throughout: In order to illuminate the nature of norms and related entities, I will focus on propositions about norms (that is, normative propositions) and about related entities.
Whether normative propositions describe semantically sui generis propositions or they describe (semantically) ordinary propositions which happen to have been ‘commanded’ by a speech act uttered by a legal authority does not make much of a difference here. Let us take a normative proposition (that is, a descriptive proposition about a norm) such as: “It is obligatory that p.” According to the semantic conception, this proposition is true if and only if, for example, an authority has uttered a normative sentence whose meaning is the proposition Op. According to expressive conception, the normative proposition “it is obligatory that p” is true if and only if, for example, an authority has uttered “p!”—that is, a speech act commanding p. But in both cases, normative propositions describe some propositional content, be it complete normative propositions (semantically sui generis), or ‘norm-contents’—that is, ordinary propositions qua commanded via a specific speech act. So, when (especially in section 2.2.) I talk about normative propositions—descriptive propositions about the existence and content of norms—the reader will be free to read into ‘norms’ their preferred ontology of norms.
2.1.2. Other kinds of ‘facts about law’
Propositions (3) and (5) are facts about (legal) normative formulations. They stand to norms (or, if you are an expressivist, ‘norm-contents’) as sentences stand to propositions. It is therefore commonplace to assume that norms are the meaning, or the content, of normative formulations. It does not mean that norms are necessarily language-dependent, but they are language dependent insofar as propositions in general are language-dependent.Footnote 51 If you are a Fregean about propositions, then you will argue that norms are not language-dependent. However, it makes sense to make norms at least prima facie language-dependent, whether you believe norms are propositions or—a fortiori—whether you think they are speech acts. However, not all norm-formulations are linguistic sentences stricto sensu, such as verbal or written utterances (like texts). They encompass all plausible units of (normative) meaning, from gestures to complex practices such as customs. Nevertheless, the most standard way to formulate norms in municipal legal systems remains the issuance of texts (e.g., statutes).
Proposition (6) is about normative content. Propositions about normative content simpliciter are about general rights, obligations, etc. Normative contents connect a given action—or set of actions—with a deontic operator. Normative contents can be described holistically or atomistically. If one has an atomistic conception of normative contents, then (4) and (6) are extensionally equivalent: Normative contents are always, well, the content of norms—or rather, they are what norms require, even if such contents hold only pro tanto and may have to be combined with other normative contents (the contents of other norms) to yield a complete operative content, that is, a complete individuated norm.Footnote 52 However, if one endorses a non-atomistic, or holistic, notion of normative content, then (4) and (6) are not equivalent, because (4) is either meaningless or false (on an error-theoretic basis). Notably, Dworkin’s and Greenberg’s strategy involves talking about normative content (the stuff that makes legal propositions true) without individuating norms.Footnote 53 What makes propositions such as “thieves ought to be hanged” true is not the fact referred to in (4) but a holistic interpretation of many different kinds of facts (linguistic, social, and moral), according to criteria of fit and justification. So, it would make sense for a holist such as Dworkin to reduce legal facts to normative content thus understood—a kind of normativity without norms.
Propositions (7) and (8) are about normative states of affairs. States of affairs (SOAs) are compositional entities (they are the connexion between a given entity and a property). Whereas facts either obtain or do not obtain, states of affairs can be non-obtaining.Footnote 54 For instance, let us suppose that Jones did not steal anything. Then it is not a fact that Jones stole something: A fact either obtains or is not a fact. Whereas states of affairs, being mere connections between entities and properties, can be non-obtaining. It is therefore possible to say that “Jones stealing a loaf of bread” is a non-obtaining state of affairs. Normative states of affairs are entailed by either norms or normative contents: That Jones ought to be punished is a state of affairs which may obtain or not. In Kelsenian terms, normative SOAs are SOAs which are gesollt.Footnote 55 If thieves ought to be punished, and if Jones is a thief, then it is entailed that “Jones ought to be punished.”Footnote 56 This is not a fact: After all, perhaps Jones is not a thief. In that respect, normative SOAs are logical consequences of norms/normative contents: they obtain only on the basis that the premises are true; otherwise, they remain non-obtaining. Both (7) and (8) denote facts about normative SOAs.Footnote 57
I have not discussed (1). Per Greenberg, (1) would not be a legal fact (pace, Shapiro). It still is a fact about law, but not a fact about the content of law. However, it bears noting that (1) is among the truth-makers of at least some propositions in (2)-(8). If the content of law is norms, one must have the concept of a legal system in order to understand how a given norm can be distinctively legal.Footnote 58 The typical positivist answer is that legal norms exist because they belong to a certain system of normative creation. The fact that my own ‘norm’, according to which “you ought to call me Your Royal and Imperial Highness,” is not a legal norm is due to the fact that it does not belong to any legal system.Footnote 59 The same goes for normative contents and SOAs insofar as (and only insofar as) they relate to norms—that is, insofar as normative contents are contents of norms atomistically individuated. A holistic conception of normative content does not need the notion of a legal system—if only metonymically—to denote the persons, times, and places to which the normative content applies. In a holistic conception of (6), one does not need the concept of a legal system (as in, “in S”), except as a shorthand, as “for persons P1…Pn, in places L1…Ln”, etc., which indicates the jurisdictional factors depending on which (6) is true or not.
2.2. Existence and content
As I have hinted, legal positivism is essentially about the truth-makers of propositions like (2). It is about the existence (or membership) of legal norms, and not about normative content per se. I will give my reasons for this claim in the next section. There remain, however, a few issues that need to be clarified in order for it to make sense to defend an atomistic metatheory of the membership of legal norms. In particular, if, as I am going to argue, legal positivism is a thesis—or set of theses—about membership rather than content, I need a theory of where the distinction lies. I will propose such a theory here, but you do not need to agree with it—other theories of the membership/content distinction are possible—in order to share the basic point that positivism is about membership and not about content.
Before ‘legal facts’ got centre-stage as the subject matter of legal philosophy, it was often claimed that positivism was the following thesis (or a variant thereof): “The existence and content of law is determined by social facts alone.”Footnote 60 The phrase “the existence and content of law” is ambiguous. As Jorge Luis Fabra-Zamora neatly puts it, one must distinguish between the issue of “the existence of a legal system” and that of “the existence of individual legal norms.”Footnote 61 As for ‘content’, the content of the legal system is precisely a set of legal norms, whereas the content of a law (a legal norm) is about what exactly that law requires. As Fabra-Zamora writes:
For example, the question “Was there a Roman legal system?” queries the system’s existence, while “Did Roman law have a rule compensating property owners for damage caused by third parties?” questions the system’s content. In turn, the question “Was Lex Aquilia a rule of Roman law?” is a question about a law’s existence, while “Did Lex Aquilia protect only the property of Roman Citizens?”, “Which kind of property did the law protect?”, and “Did the law provide compensation for fault only, or did it also allow for strict liability?” are questions about the content of that law.Footnote 62
I basically agree with that cartography of the various issues covered by the ambiguous expression “the existence and content of law.” However, it can be challenged, and I want to address such challenges.
First, propositions about the content of a legal system and statements about the existence of the system’s legal norms are extensionally equivalent. This is because, as I have argued elsewhere, membership is the specific mode of existence of legal norms (as well as all norms, but that is beyond the point).Footnote 63 A legal norm exists qua legal norm insofar as it belongs to a legal system. So it is strictly equivalent to claim that N ‘exists’ in legal system S and that it belongs to (the content of) S. The questions “Did Roman law have a rule compensating property owners for damage caused by third parties?” and “Was Lex Aquilia a rule of Roman law?” are extensionally (though not intensionally) equivalent.
However, the question remains whether propositions about the existence/membership of norms and propositions about their content can really be distinguished. Since I aim to claim that legal positivism is basically about existence (qua membership) rather than content, solving this problem is crucial.
The challenge basically runs the following way. Normative propositions, as we saw, aim to describe legal norms. Describing what norm N is, in short, is exactly the same as describing what N prescribes: Norms and normative contents are the same thing. So it seems that (6)Footnote 64 is but an expanded formulation of (2): If “N” is only a shorthand for a proposition and that proposition is “thieves ought to be hanged,” then it seems that N ≡ thieves ought to be hanged. Therefore, the propositions “there exists a norm N” and “thieves ought to be hanged” are strictly equivalent. As Bulygin puts it,
The sentences
[a] You must not park here
and
[b] There is a norm that prohibits parking here
have the same meaning, [b] being an expanded form of [a].Footnote 65
In that respect, (4) above is a synthesis of (2) and (6): The phrase “provides that” in (4) is the metaphorical equivalent of the equivalence operator (≡). On that picture, norms do not exist independently of what they prescribe or require. Let me call this the thesis of the unity of normative propositions (UNP). If (UNP) is true, then legal positivism is necessarily a theory of both the existence and content of laws, since asserting the existence of a law (a legal norm) is the same thing as asserting its content. It then makes sense to claim that positivism is about the determinants of legal content (atomistically understood as the content of legal norms), which entails that the reframing of the debate in terms of legal facts is prima facie sound.
I aim to suggest that (UNP) is false. One can assert that a norm exists (i.e., that it belongs to a legal system) without making a proposition about the correlated normative content. This is what Georg Henrik von Wright wanted to show when he distinguished between two kinds of normative propositions (that is, descriptive propositions about norms), to wit, “normative statements” (which describe normative contents) and “norm-propositions” (which bear on the existence of norms).Footnote 66 Thus, (6) above is an example of what von Wright calls ‘normative statements’, whereas (2) is an example of a ‘norm-proposition’.
According to von Wright (whom Bulygin criticizes in the above quote), norm-propositions (there exists a norm N which provides p) are the truth-makers of normative statements.
By the truth-ground of a given normative statement I understand a truthful answer to the question why the thing in question ought to or may or must not be done. Let the normative statement be, for example, that I may park my car in front of your house. Why may I do this? The answer could be that there is a regulation according to which I am permitted to do this. The existence of this regulation (norm, prescription, permission) is the truth-ground of the normative statement.… The proposition that such and such a norm exists, I shall call a norm-proposition.Footnote 67
In a nutshell, I think this distinction is sound, but that ‘norm-propositions’ (propositions about existence) are only part of the truth-makers of ‘normative statements’ (propositions about contents).
Here is why. (Descriptive) propositions about norms are about propositional contents—be it sui generis propositions, or ‘norm-contents’ resulting from speech acts, depending on your preferred ontology of norms—but one can enquire whether a given content exists without trying to figure out what the content exactly is (what the content of the content is, so to speak). This is so because the kinds of facts which are the truth-makers of propositions about the existence of a norm (von Wrightian ‘norm-propositions’) and about its content (von Wrightian ‘normative statements’) are different. In short, the facts that make the norm-proposition “N exists,” such as (2) above, are (per positivists) basically a specific set of social facts about norm-creation, such as enactment, promulgation etc., whereas the facts that make the proposition “it is obligatory to ϕ,” such as (6) above, are linguistic facts about the meaning of the normative formulation(s) which aims at expressing a norm making it obligatory to ϕ, together with, possibly, other facts (including, as I will suggest later on, possibly moral facts in some contexts). When we ask whether the norm which provides that thieves ought to be hanged exists (that is, belongs to the legal system), the (positivistic) answer will be that certain social facts obtain in virtue of which N exists:Footnote 68 For example, the fact that the legislature voted on a given statute (a set of norm-formulations), that the head of state enacted it, and that N can be traced back to such facts. Whereas when we ask how to define ‘thief’ and ‘hanging’ in the context of the law, we mean to ascertain the semantic content of linguistic utterances. These are two different operations. Of course, many statements about law will typically perform both operations at the same time (such as (4) above), but they can be broken down into two different kinds of propositions: propositions about content and propositions about membership.
Propositions about the content of norms such as (6)—i.e., von Wright’s normative statements—aim to provide a definite description of the propositional content which is being described. For instance, in (6), ‘thief’ and of ‘hanging’ can be interchanged salva veritate with descriptions of their intensional properties. On the other hand, propositions about norm-membership merely describe the fact that the norm being described belongs to a given legal system. Such propositions will typically use designators such as “N,” or “Lex Aquilia” or “article 443 of the Penal Code” or “the Donoghue v Stevenson rule.” The same goes for phrases such as “the norm that provides that thieves ought to be hanged” or the “thieves-ought-to-be-hanged norm.” Such normative designators are usually shorthand for normative contents; one does not need to have a complete description of these contents to discuss whether they belong to a given legal system. In a sense, these designators behave like proper names in sentences about the existence of the person they denote.Footnote 69
This is why two jurisprudes may disagree whether a norm belongs to a legal system without disagreeing on that norm’s content. Let us take as an example a truly evil norm (such as: “all redheads ought to be executed”). A Radbruchian non-positivist will claim that such a norm does not belong to the legal system, whereas a positivist will usually claim it does. Their disagreement is not about what counts as ‘redhead’ for the purposes of the law. A disagreement on what counts as ‘redhead’ would revolve on competing normative statements about the norm’s content, spelling out, for example, the intensional properties of ‘redhead’ (e.g., does ‘light auburn’ count as ‘redhead’?). Ex hypothesi, our two jurisprudes agree that such is the content of the norm. The disagreement is whether such a norm exists in the first place, i.e., whether it belongs to the legal system.
Moreover, it makes sense to ask whether the thieves-ought-to-be-hanged norm belongs to S without having a complete idea of the normative content it designates. Let us suppose that in S the legal definition of thief (such as it is formulated in another part of the criminal code) is “someone who steals royal property” (stealing private property is either permitted, or covered by another statute), and that you ignore the existence of this definition. You can still make the true proposition that in S the thieves-ought-to-be-hanged norm exists—i.e., that it belongs to S—while being wildly off the mark when it comes to the normative content itself. You are still referring to the same norm—because it is traceable to the same set of facts—even if the proposition you would have made about the content of the norm would have been partially false. Now suppose that, among the scholars of S, there is widespread disagreement about the content of N—some, for example, claiming that N only applies to someone who steals royal property, whereas others claim that N also applies to someone who steals sacred relics. What we have here is not a debate on the membership of either one of two different norms, since such norms can be traced to the same social facts; the disagreement is about the precise content of the same norm.
To be sure, we usually are able to differentiate—and individuate—various existing norms based on a (more or less limited) grasp of their content, because among the social facts that make a norm belong to a legal system there is usually the fact that a text—a set of normative formulations such as a statutory provision or a judicial ruling—has been adopted, that is, has been agreed upon as expressing a legal norm. So, unless one thinks of legal communication as being radically opaque, one must accept that normative formulations usually give a rough idea of the kind of behaviour prescribed by the norm: When we ask whether S contains a norm providing that thieves ought to be hanged, we will not be satisfied by an answer pertaining to the way the no-vehicles-in-the-park norm was enacted. This rough grasp of normative content allows us to use the right designators in propositions about the existence/membership of norms; the same way that we need to have an idea of who Pierre is when asserting propositions about Pierre, even if such propositions do not aim to give a description of what (or who) Pierre is.
Now, suppose someone asks the following question Q: “whether in S thieves ought to be hanged.” There are basically two ways to answer in the negative: (2’) ‘no, this norm does not exist/belong to S’ (i.e., because such a norm was never enacted in S); (4’) ‘yes, there is such a norm in S, but it only applies to people who steal royal property’. (2’) and (4’) cannot be true at the same time. Given what we know about S, (4’) is a true description of N’s content. This entails that propositions about the existence of norms are independent of propositions about normative contents.
One will object that it all depends on the way one individuates the relevant norm.Footnote 70 Since ‘thief’ means “anyone who steals royal property,” then the norm is not really about thieves, but about thieves of royal property. So, in the end, there is only one way to answer Q: “No, there is no norm in S which provides that thieves as such ought to be hanged.” Or so the objection goes. But this does not seem remotely as informative as (2’) and (4’) above: Indeed, (6’) ‘In S, thieves ought not to be hanged’ seems patently false, precisely because information about existence and information about content are two different kinds of information: (6’) is false because we know that the “thieves-ought-to-be-hanged norm” belongs to S, even though such information does not tell us much about the exact content of this norm. This is why propositions about membership are only part of the truth-makers of propositions about normative content.
To be clear: I do not claim that all positivists (if any) endorse this way of distinguishing between existence and content. Some do not endorse the distinction, and others frame it quite differently (see section 3.1, below). My only point is that positivism is best understood as being about membership and not content, which presupposes that the two can be presumptively distinguished. What I have been expounding in this subsection is, however, my own way of delineating the distinction.
3. Legal positivism is only about the membership/existence of norms
Legal positivism is about what makes propositions like (2) above true. It is about the criteria of membership of norms within particular legal systems. Hence Gardner’s famous definition of legal positivism:
(LP*) In any legal system, whether a given norm is legally valid, and hence whether it forms part of the law of that system, depends on its sources, not its merits (where its merits, in the relevant sense, include the merits of its sources).Footnote 71
To put it another way, legal positivists claim that whether a norm belongs to the legal system depends on social facts (or rather a specific set of social facts relating to legislation, custom, precedent, etc.), and not moral considerations.Footnote 72 A norm may belong to the legal system even if it is morally unmeritorious, and it may be the case that a morally meritorious norm does not belong to the legal system. This is not to say that this is the only claim that positivists make. Some positivists will typically endorse thicker definitions, but I take (LP*) as the smallest common denominator definition of what legal positivism is about.Footnote 73
The departure from Gardner’s framing by the new ‘legal facts’ crew did not go unnoticed. As David Plunkett noted, Shapiro’s embrace of the ‘legal facts’ framing allowed him to find a common ground for debate with non-atomistic theories of law such as Dworkin’s or Greenberg’s:
[S]ome accounts such as Dworkin’s and Greenberg’s own that understand legal content holistically rather than atomistically are difficult to make good sense of using the idea of a single norm either having the status of legally valid or not.… Shapiro’s account of what is at stake in the debate over legal positivism gives him a broader definition than Gardner’s definition, allowing Shapiro to more smoothly make sense of a range of possible positions in the field.Footnote 74
‘Smooth’ is an interesting word in this context. As in Michael Jackson’s song, legal positivism has “been hit by a [very] smooth criminal.”Footnote 75 At no point does Plunkett enquire whether a non-atomistic conception of ‘legal content’ is indeed compatible with legal positivism. For the reasons I will explore in this section, it is, in fact, not compatible. The reason why Gardner frames positivism in an atomistic way is that atomism is an essential or crucial property of positivism. One cannot find common ground in a debate by reframing the terms in such a way that one abdicates central parts of one’s own position.Footnote 76
In this section, I will defend Gardner’s framing: Of all the ‘facts’ denoted by (2)-(8) above, legal positivism bears only on (2). As I will suggest, positivism (even exclusive positivism) may allow for ‘moral facts’ to be among the truth-makers of at least some of propositions (3)-(8). Therefore, framing legal positivism as the thesis that “legal facts are determined by social facts only” is misleading, because every positivist in their right mind ought to acknowledge that this thesis is false.
3.1. Membership of norms or normative formulations?
Let me address here a common challenge to the idea that positivism is about the membership of norms. Such a challenge does not deny that positivism is about membership; however, it contends that what belongs to legal systems are normative formulations, not norms. This point has been put forward by various iterations of legal realism on the Continent.Footnote 77 Its most recent and elegant exposition is by Giovanni Battista Ratti and Luka Burazin.Footnote 78 Let us suppose that the meaning of a given normative formulation NF can be either one of two norms, N1 or N2. Depending on the way one interprets NF, the system will either comprise N1 or N2. Both N1 and N2 cannot belong to the system at the same time. Whether it is N1 or N2 that belongs to the system cannot be settled without interpreting NF.
Since norms are meanings or contents of normative formulations, they must be developed via interpretation, usually by an authorized interpreter, such as a court. So, unless the membership/content distinction collapses and the rule of recognitionFootnote 79 incorporates canons of interpretation (which Ratti and Burazin are adamant is not the case, and I agree), then what is identified as belonging to a legal system is only a set of normative formulations (which they call “sources”).Footnote 80 Norms are the result of the application of various canons of interpretation: They do not pre-exist the authoritative interpretation of normative formulations by various officials.
This theory goes even further in the direction at which I have pointed—that is, in distinguishing issues about membership and content: Legal positivism is about the membership of normative formulations, and it remains agnostic on the various ways of deriving normative content. It seems to be a close ally of the views I defend in this paper; I nevertheless disagree. Explaining in detail why I disagree would take us beyond the scope of the paper. In a nutshell, legal texts are not the stuff that belong to legal systems, but part of the stuff in virtue of which norms belong to legal systems. Basically, on a positivistic theory of norm-membership, norms belong to legal system if and only if they can be traced back to a given set of social facts (usually called ‘sources of law’): the fact that a text exists, that Parliament voted on it, that the Head of State promulgated it, etc.Footnote 81 The rule of recognition picks out these facts and makes them criteria of membership for legal norms. Sometimes, the existence of a text will not be part of the relevant criteria of validity: Customary norms are not typically created in virtue of the fact that a text has been promulgated.
As I sought to explain in section 2.2, propositions about the existence of norms are not the same thing as propositions about normative contents. So, one can say that a norm belongs to a legal system without having a complete definite description of its content, even if the normative formulation usually gives a rough (though possibly incomplete, and as we shall see, indeterminate) notion of the content. This is why I remain unconvinced by the picture put forward by Guastini, Burazin, Ratti, and others, according to which the meaning of norm-formulations is made up of a disjunctive set of discrete norms—that is, of discrete normative contents. Except in cases of pure lexical ambiguity (e.g., “bats are forbidden here”, on which see infra), a normative formulation NF does not typically yield a disjunctive set of discrete norms N1 or N2 or N3. First, it is not conceptually impossible that NF expresses a single norm N, but whenever it does not, what NF expresses typically is a cluster of overlapping normative contents and not a disjunctive set of discrete norms. These contents cannot usually be neatly individuated one from another (as Guastini and others argue they can), because they are made up of sets of overlapping properties, rather than of a disjunctive set of fully individuated norms. This explains why we can make membership-statements about norms (understood as such clusters) prior to any authoritative interpretation while there still remains possible uncertainty about the norm’s content (since it is not fully stabilized prior to interpretation, although there will be a ‘core’ of determinate properties). So a norm-formulation such as “violations of human dignity are prohibited” does not yield distinct norms according to the interpreters’ diverging concepts of ‘dignity’ (or of ‘violation’), but a single cluster of overlapping normative contents, where the decisive set of properties—the complete description of “violation of dignity”—is left unspecified. As I will intimate in the next section, law rests on a dynamic system of implementation and concretization, performed by various layers of law-applying institutions. Their job is precisely to fix the content of norms they apply, by selecting (among a given cluster) the set of properties which makes up this content.
Therefore, two questions ought to be distinguished: a) does norm N belong to S? and b) is it possible for me to act on N (does N gives me a reason to act)? Regarding the question a), “N”, as we saw, is used as a shorthand designator. We can say that the norm prohibiting gavagai belongs to S without having any idea of what gavagai is. It does not mean that what belongs to S is the formulation “gavagai are prohibited”: What belongs to S is the norm that gavagai are prohibited (or the gavagai-prohibiting-norm) in virtue of the fact that there is a norm-formulation reading as “gavagai are prohibited” which was enacted in such-and-such way (and possibly additional social facts).
Now regarding the question b): The fact that a norm exists does not necessarily provide its addressees with a reason for action. In the case of the gavagai-prohibiting-norm, the addressee will typically not know how to act. In order to do so, they will have to ask what the normative content is: What is gavagai? What are the intensional properties of gavagai? In other words, they will have to interpret the normative formulation in order to grasp what the norm—which exists in virtue of the existence of this formulation plus some additional facts—requires. In a nutshell, what interpretation does is not ‘posit’ or ‘create’ the norm—the norm existed prior to interpretation, since its existence/membership can be described independently of a complete description of its content. What interpretation does (or at least aims to do) is give a (hopefully complete) definite description of the normative content—that is, of what the norm requires. The fact that interpretation (by officials as well as by private citizens) is required in order to act on the norm does not mean that interpretation is required in order to establish that the norm itself is part of the legal system.
This is why, although the ‘continental realist challenge’ does distinguish—as I do—between membership and content, it nevertheless ought to be resisted. Besides, if legal positivism is understood as merely claiming that legal texts exist in virtue of certain social facts, then it is trivially true—and not even remotely informative. Nobody would deny that whether a given text has been written, submitted for approval to Parliament, and then enacted by the head of state is a matter of social fact. What makes positivism controversial is the claim that all legal norms exist if and only if they can be traced back to such social facts.
3.2. Legal positivism is not about normative content (nor about normative SOAs)
Granted, some important legal positivists have claimed positivism was a theory about both the existence and content of law/laws. Raz famously claimed about the sources thesis, which he claimed to be central to positivism: “A law is source-based if its existence and content can be identified by reference to social facts alone.”Footnote 82
However, I will argue that there is no direct route from existence (that is, legal membership) to legal content. Establishing the existence of a norm is a matter of social fact—or so classical positivists will argue. The proposition “N belongs to S” can be negated by showing that the social facts picked out by the rule of recognition (RoR) do not obtain. When we ask whether the norm prohibiting infringement on human dignity belongs to the German legal system, we only ask whether social facts obtain such that the ‘dignity-protecting-norm’ belongs to the legal system without asking what, in fact, are the intensional properties of human dignity. Likewise, even if my own norm-formulation “Swifties ought to be punished” does aim to ‘posit’ a norm, the social facts do not obtain such that the “Swifties-ought-to-be-punished norm” exists in any legal system.Footnote 83 Membership demarcates between the norms which are part of the legal system and those which are not—that is, between legal norms and non-legal norms. In this respect, not all social facts are relevant: Only the social facts (usually called ‘sources’) picked out by the rule of recognition are.
What about normative content? It could be objected that if, as I claim, normative content can be severed from norm membership, then one could conceive of the law as being a catalogue of opaque normative designators (for instance, “N1”, “N2”, “N3”, etc.) referring to entities the content of which would be left altogether unspecified. Norms would be individuated atomistically, but normative content could be discovered holistically (or in any possible way, including cartomancy and divination). Of course, I do not suggest such a theory makes sense; it patently does not. Normative content is at least partially determined by the content of normative formulations (but ‘partially’ is key).Footnote 84 When people try to ascertain what the law requires of them, they establish the existence of a given norm and then try to discover its content by grasping the meaning of the normative formulation which aims at positing that norm.
Many positivists have insisted, however, that mere semantic and pragmatic facts about normative formulations are often not enough. We can put aside cases in which ambiguity makes it altogether impossible to assign any normative content to a normative formulation—such as cases of homophony: “bats are forbidden in the building” (the normative formulation is such that, if the context of utterance is opaque, there actually are two possible norms). Such cases are rare, and even cases of lexical or syntactic ambiguity rarely make it impossible to establish that a norm exists, albeit with a somewhat indefinite content. In many cases, however, normative formulations will not be enough to allow for a complete definite description of the normative content. Hence Kelsen talked about norms as frames, and Hart wrote at length about the open texture of rules.Footnote 85 The idea of ‘clusters of normative properties’ mentioned earlier points to the same direction.
Of course, additional distinctions would have to be made. As Guastini rightly insisted, normative interpretation (the interpretation of normative texts/formulations in order to identify normative contents) and normative subsumption (identifying which cases fall under the scope of the norm at the application stage) are two different operations.Footnote 86 In our own parlance, identifying normative content is different from deriving propositions about normative states of affairs. This distinction is important, but it can be occasionally blurred. For instance, (soritical) vagueness is a straightforward extensional problem (a subsumption problem), but Hartian open texture, for instance, is a more complex issue: Although it is mainly an extensional problem (i.e., is token x an instance of concept C?), it nevertheless stems, as I have argued elsewhere, from a specific kind of intensional indeterminacy or meta-ambiguity, the intension of the open-textured concept being made up of various partially overlapping sets of properties.Footnote 87 In cases of open texture—and a fortiori in cases of ambiguity—the problem is not only one of subsumption, but also one of interpretation—that is, of fixing the intensional features (the semantic properties) of the normative content.
Now it is undisputed that, in most legal systems, many normative formulations use moral terms. Take article 1.1. of the German Grundgesetz (“Human dignity shall be inviolable. To respect and protect it shall be the duty of all state authority”)Footnote 88 or the former (now repealed) article 1133 of the French Civil code (“A cause [of contract] is unlawful when … it is contrary to good morals or to public order”).Footnote 89 Giving a complete definite description of the content of such normative formulations does entail some kind of moral reasoning on the relevant properties of ‘human dignity’ and ‘good morals’. Whereas the membership of the norm(s) ‘posited’ by art. 1.1 of the German Basic Law is a matter of social fact (there is absolutely no doubt that the ‘dignity-protecting-norm’ belongs to the German legal system), the exact normative content (what the norm actually requires) may be partially determined by moral facts.
Some positivists seem to have been bothered by this. Such is the case of Joseph Raz who, in The Authority of Law, reaches the (somewhat ‘surprising’, as Timothy Endicott puts it) conclusion that whenever a norm-formulation uses a moral concept, such as in the case of a norm forbidding immoral contracts or protecting dignity, it creates a gap in the law—the judge being therefore required to exercise discretion.Footnote 90 Whenever a normative formulation uses a moral concept, then the law runs out, and judges are entrusted with a “directed power” to develop the law.Footnote 91 This may be the case: But even if it were not—even if moral reasoning was required to identify a normative content that was ‘already there’—it would not be much of a problem for legal positivists.
We can agree that in hard cases, the law runs out: When the norm formulation uses moral concepts, and judges are faced with a borderline case of the application of such concepts, then judges are directed to exercise moral deliberation in order to settle the matter before them.Footnote 92 But then the normative content is fixed by the judicial decision, which changes the law by issuing a new norm (applicable to the case at hand and possibly all similar cases).
But what about easy cases? When normative formulations use moral concepts, they can still yield easy cases—they do not always entail discretion on the part of law-applying institutions. In easy cases, judges do seem to discover pre-existing law, the law ‘already there’, so to speak. This normative content seems, then, to be determined by moral facts. Is that a problem for positivists? No.
First, such moral (and, more generally, evaluative) considerations often are somehow embedded in the (social) linguistic rules governing the use of moral concepts: Deferring to a shared evaluation embedded in the conventional rules of language is a success condition of the application of all evaluative concepts, and perhaps of certain non-evaluative concepts as well. Let me take an example: In the United Kingdom, the Criminal Law Amendment Act of 1885 made it prohibited to engage in acts of “gross indecency.”Footnote 93 In the middle of the twentieth century, a consensual homosexual relationship would be a clear case of ‘gross indecency’. Let us imagine that such a case is brought before a judge who happens to be a liberal devotee of John Stuart Mill. In their own assessment of the case, this judge would never think of the two men’s behaviour as grossly indecent; however, unless they are to break the law, they will declare them guilty. In applying the concept of gross indecency, they have not relied on their own straightforward moral evaluation of the case but have only made a proper use of the rules of language (such as existed at the time of ruling), and this use certainly involved a minimal form of moral evaluation. In many cases, even though the law has not run out—it is clear that the norm requires punishing the gay couple—the kind of moral evaluation displayed in identifying the normative content (or as is the case in the indecency offence, in qualifying states of affairs normatively) is itself supervenient on social facts (that is, linguistic rules governing the use of evaluative concepts such as ‘gross indecency’). In any case, these social facts have no bearing on the membership of the norm at hand: Legal positivism is not about the social nature of normative content, but about a specific set of social facts (‘sources of law’, such as legislation, precedent, custom, etc.) picked out as criteria of membership by the system’s rule of recognition.
Second, even if it were the case that more robust moral facts would determine the content of norms (that is, normative contents), it would still not be a threat to positivism, and it would not justify ‘softening’ positivism. Indeed, some ‘inclusive positivists’ have claimed that the use of moral concepts in the law (such as article 1.1 of the Basic Law) buttresses inclusive positivist claims.Footnote 94 I do not think this is the case. The debate between inclusive and exclusive legal positivists is best understood as a debate about criteria of membership. The mere fact that moral considerations are required (at least in some easy cases) to ascertain the content of moral concepts used in legal normative formulations does not mean that a (moral) norm protecting dignity is part of the German legal system because of its own merits rather than its traceability to social facts.Footnote 95 If the dignity-protecting norm belongs to the German legal system, it is squarely in virtue of social facts (the fact of the existence of the constitutional text, plus facts about enactment and promulgation). In some cases, it will be impossible to fix normative contents without engaging in moral deliberation: This is the case, for instance, when normative formulations use ‘thin’ moral concepts (e.g., “one ought to do what is good”; “consideration of the best interest of the child ought to be paramount”). Determining what is ‘good’ or what is a child’s ‘best interest’ cannot be settled by the use of linguistic conventions. However, this has no bearing on the reasons why the “do-what-is-good” norm or the “best-interest-of-the-child” norm belong (or do not belong) to a given legal system.
So positivism is not about legal content; positivists should have no qualms accepting that “people involved in existing legal practice such as judges and lawyers actually reason and argue when they try to discover which legal propositions are true, they seem almost inevitably to rely partly on … normative and evaluative thought.”Footnote 96 However, positivists insist that in many cases (but not necessarily all cases), they do not try to “discover which legal propositions are true,” because the law is (relatively) indeterminate, and sometimes the job of judges and other law-applying officials is to make new law. ‘Legal content’ is not something that is always ‘already there’, waiting to be ‘discovered’. There is, in other words, no direct route from norm membership to normative content: Methods of interpretation, that is, methods for ascertaining normative content, are not part of the criteria of membership picked out by the rule of recognition.Footnote 97
3.3. Atomistic membership
The upshot is: Legal positivism is about what makes a given norm (however indeterminate its content) belong to a legal system, not about normative content. What determines normative contents (that is, what the norm actually requires according to a complete definite description of its intensional properties) is—at least partially—a set of social and linguistic facts about the relevant normative formulation plus, possibly, some additional facts, including moral considerations. This is no threat to positivism. Granted, not all positivists would see things that way; I am sure that many would find my distinction between membership of content too coarse.Footnote 98 But I think that, as a historical matter, legal positivism (especially in its polemic against natural law) bears mainly on what makes legal norms exist as such—rather than being a specific theory of interpretation or legal reasoning. All that precedes is but an elaboration on this basic idea; as I said earlier, one need not agree with the elaboration to share the basic idea.
Now, as we saw, the notion of norm-membership in a legal system presupposes thinking about norms atomistically, even if it does not mean that legal content is itself always wholly determined atomistically—that is, by sole reference to the meaning of normative formulations. This is why what Greenberg calls the ‘Explanatory Directness Thesis’ (EDT)—a key part of the ‘standard picture’ he wants to refute—is not, in fact, a necessary component of legal positivism.Footnote 99 The EDT is defined as follows:
In the complete constitutive account of the obtaining of a given legal norm: 1) the authoritativeness of the pronouncement is prior in the order of explanation to the obtaining of the legal norm; 2) the authoritativeness of the pronouncement is independent of the pronouncement’s (specific) content and consequences; 3) there are no explanatory intermediaries between the authoritative pronouncement’s being made and the norm’s obtaining.Footnote 100
Insofar as by “the obtaining of a legal norm,” Greenberg means the existence of the norm—its membership within a legal system—he is right that the EDT is an important part of what legal positivism is about. Now, if by ‘legal norm’ Greenberg refers to what determines normative content, it is not true that positivists are necessarily committed to endorse the EDT. Some have endorsed it. But, again, legal positivism is not a theory of interpretation. It is not inconsistent with legal positivism to accept that there are, indeed, explanatory intermediaries between ‘the authoritative pronouncement’ (the social fact ‘positing’ the norm) and the resulting normative content. Greenberg is right that atomism—which he defines as the idea “that individual legal norms [by which he means individuated legal norms] are explanatorily prior to the content of the law as a whole”—is a “natural consequence” of the EDT: Membership is prior to content.Footnote 101 But it is not true that content is solely ‘explained’ by membership. Content must be in part explained atomistically: Legal content is the meaning of legal normative formulations aiming at ‘positing’ legal norms.Footnote 102 But it does not follow that only social facts (such as social or linguistic facts about normative formulations—or ‘authoritative pronouncements’) determine normative content; moral facts, determined holistically, may come into account.
Holistic theories such as Greenberg’s or Dworkin’s, on the other hand, do not distinguish between membership and content: They take a given normative SOA, such as “may Elmer Palmer inherit?”, and try to establish how the ‘content of the law as a whole’ provides the right answer.Footnote 103 The ‘content of the law as a whole’ is a set of legal facts. Normative formulations do not posit any norms; they are just part of the pre-interpretive materials whose integrity-based interpretation is required to ascertain this ‘content as a whole’ (in Dworkin’s view at least), or they are part of the ‘relevant actions of legal institutions’ which change our moral obligations, the ‘content of the law’ being nothing more than the set of moral obligations ‘impacted’ by pronouncements of legal authorities (in Greenberg’s view). Framing the issue in terms of legal facts is just a way to make the holistic view the default framing. By accepting it, not only do positivists renounce atomism (which is key to their theories of norm membership), but they are invited to provide a theory of legal content—which was always, at best, a secondary endeavour.Footnote 104 Positivists are now required to defend the idea that legal content is always predicated on social facts—although they were never interested in social facts per se, but only in the small subset of social facts (i.e., facts about legislation, precedent, custom, etc.) picked out as criteria of membership by the rule of recognition. In doing so, positivists lose the point that positivism always was about ‘membership facts’ and never about ‘content facts’: There is no specifically positivist theory of interpretation, and nothing anti-positivist to a theory of interpretation which accounts for the role moral considerations play in the determination of legal content, if only when judges use their directed powers.
4. Resisting the metaphysical turn in jurisprudence
Let us take stock. If I am right, then the legal positivism versus non- (or anti-) positivism debate cannot be framed in terms of ‘legal facts’ simpliciter (or ‘social facts’ simpliciter, for that matter), because legal positivism is only about a very limited set of such facts: facts about the existence/membership of norms. This is why the category of ‘legal facts’ will not do: It is overbroad.
One can make the following objection: “Maybe so. But it does not mean that the category of ‘legal facts’ ought to be abandoned. Rather, it means that we should get rid of the positivism/non-positivism debate altogether.” This is a powerful objection. The fact that existing debates are disfigured by a new framing may show that these debates must be abandoned, and that new debates about the new framing should be part of the research agenda of jurisprudence. In particular, we should accept that the new way to theorize about law is resolutely metaphysical—whereas the ‘old way’ is conceptual—and if it does not neatly map onto the parochial and old-fashioned debate between positivism and non-positivism, then so much for that debate. Who cares about ‘norm membership’ anyway? ‘Legal facts’ are here to stay; we should talk about ‘the content of law as a whole’ as if such a thing existed without there being any atomistically individuated norms in the first place. Classical debates in jurisprudence are, after all, uninteresting;Footnote 105 metaphysics, on the other hand, is interesting. The metaphysical approach allows jurisprudence to stop being so parochial—and it makes it part of mainstream philosophy—by asking how legal facts fit in our catalogue of the world, in the reality at large. The metaphysical approach makes jurisprudence much more philosophically interesting than dusty old debates about positivism and natural law, which have no philosophical significance whatsoever. Or so the objection goes.
In this section I will briefly address this objection by arguing that we should resist the metaphysical turn in jurisprudence. Although I believe that we should—to some extent—go beyond the positivism/antipositivism debate, I do not think metaphysical debates about legal facts are the right way to go. I fully realize I am going against the tide. Grounding is now all the rage; before that, it was supervenience.Footnote 106 Greenberg himself attempted to introduce a new metaphysical approach—“rational determination.”Footnote 107 But what are the alleged benefits? According to Patterson and Tripkovic (who nevertheless ultimately share my skepticism), the main alleged benefit of grounding-based theories of law is “clarifying and recasting some old debates in a common vocabulary”—old debates such as positivism and anti-positivism, for instance.Footnote 108 Chilovi claims that:
Since LP [legal positivism] and AP [anti-positivism] are widely taken to make claims on the determinants of legal facts, since the main candidates for playing the law-determination role other than grounding—causation, supervenience, and necessitation—are ill-suited to the task, and since law-determination essentially does as grounding does, it is very plausible to think of LP and AP as views about grounding.Footnote 109
However, I have been trying to show that metaphysical theories—be they about grounding or any other metaphysical relation—do not help to efficiently clarify or reframe old debates in more precise terms. Since ‘grounding’ (as well as other metaphysical concepts before it) is mainly about how various facts relate to each other by belonging to more or less fundamental strata of reality, it must operate at a level of generality (‘legal facts’). This works well for anti-positivist holistic theories—where there are free-floating legal contents waiting to be grounded in something more fundamental—but it does not work at all for positivists, precisely because positivists never intended to develop a theory of ‘legal facts’ in the first place. I am not saying that grounding-based reformulations of positivism as a theory of the grounds of legal facts—reformulations such as Plunkett’s or Chilovi’s—are not interesting.Footnote 110 But they have nothing to do with what the great philosophers in the positivist canon (from Kelsen to Ross to Hart to Bobbio to Raz to Bulygin to Gardner) understood themselves to be doing.
Now, let us put the positivism/antipositivism debate behind us—at least for now. What are the downsides of metaphysical approaches in legal philosophy at large?
4.1. The invisibilisation of jurisprudence
First, there is the invisibilisation of jurisprudence as a specific set of questions and issues (not all related to the positivism/antipositivism debate—far from it). I fully understand the urge to make jurisprudence more philosophically robust—with the help of technical concepts and apparatuses borrowed from other, more sophisticated, parts of philosophy. Jurisprudence should sometimes be less parochial and more integrated within other philosophical subfields, insofar as they can illuminate specific issues about the ‘nature’ of law—or, if one dislikes essentialist overtones, ‘the way law operates’. But hardcore metaphysics are not the right way to proceed. Grounding, for example, presupposes various levels of reality (some more, some less fundamental), with a special relation (grounding) allowing the more fundamental levels to determine the less fundamental. It is very easy to see that ‘legal facts’ will never be part of the more fundamental levels of reality: Law is an infinitely small, insignificant part of our catalogue of the world. Metaphysicizing about the law is like casting pearls before swine.Footnote 111 The grounding of ‘legal facts’ is no more interesting than the grounding of the fact that my dishwasher is currently broken.
A good illustration of this issue is Gideon Rosen’s idea that ‘laws’ (in general) can figure in grounding explanations—that is, in explaining why certain more fundamental facts ground less fundamental facts.Footnote
112
By ‘laws’, Rosen means logical laws (such as
$$\left[ {\square \;\forall \varphi \forall \psi \;\left( {\varphi \supset \left( {\varphi \vee \psi } \right)} \right)} \right]$$
), as well as cases of ‘formal’ or ‘soft’ normativity (such as law), and ethical cases (implying a more robust kind of normativity). It seems to me that these are very different kinds of laws. Rosen may be right that laws proper (i.e., legal norms) are required to explain how certain pre-legal facts (such as “A killed B”) ground legal facts (“A ought to be convicted of murder”). However, since laws are themselves grounded in something more fundamental (for instance the actions of officials, or what Greenberg calls “law practices”),Footnote
113
they are to some extent dispensable—whereas, according to Rosen, ethical laws are required to ground ethical facts. But it remains that Rosen’s main point is not about law, properly understood (i.e., the legal system), but about all “nomic explanation[s]” of grounding.Footnote
114
This is understandable: Why should metaphysicians specifically focus on the law?
Another important feature is that metaphysical explanations such as grounding are inherently controversial among metaphysicians themselves. Jurisprudes who dabble in legal metaphysics must first adjudicate among various understandings of the right metaphysical explanation and then try to provide a metaphysical picture of the law. They will ask whether grounding entails (or not) supervenience, whether facts are truth-makers, whether rational determination entails grounding (or supervenience), whether and how grounds, anchors, and enablers differ from one another—very interesting debates to be sure, and yet jurisprudential non-starters.Footnote 115
4.2. The obfuscation of important philosophical issues about the law
Contrary to morality, law is a dynamic phenomenon, which involves various institutional apparatuses of implementation and concretization. For any class of cases, whether the relevant legal normative content is p or q is often not settled before the implementation stage. This is due to many factors—for instance, the generality of law and its relative indeterminacy. Whether p or q is normatively required is not always (if ever) a matter of fact. This dynamic feature of law raises a vast number of philosophical issues about the law that are impossible to grasp with such a coarse category as ‘legal facts’. I am going to take only two examples, which, I think, cut across the positivism/anti-positivism debate: The issue of discretion on the one hand and normative conflicts on the other.
4.2.1. Discretion
Let us examine discretion (or judicial law-making). As I suggested in the previous section, positivists usually put forward a picture of the interpretation and application processes that gives an important role to discretion—whereas people like Dworkin usually deny that ‘strong’ discretion is a typical feature of law and adjudication.Footnote 116 To my mind, the debate over discretion is nevertheless orthogonal to the positivism/anti-positivism debates: Some positivists disagree on the exact extent of the indeterminacy of the law—and absolute formalism can be compatible with positivism even if no positivist has ever endorsed it;Footnote 117 conversely, not all anti-positivists endorse the one-right-answer thesis or need to downplay the role of discretion.Footnote 118
Granted, the ‘legal facts’ picture does not necessarily entail the absolute determinacy of the law. It is to some extent compatible with the acknowledgment that there is epistemic uncertainty about the obtaining of a given legal fact, or whether the more fundamental facts (social or moral) ground this or that legal fact.Footnote 119 It is even compatible with the more robust metaphysical (and not epistemological) claim that sometimes grounding fails because reality is indeterminate (metaphysical indeterminacy).Footnote 120 I am not going to argue that metaphysical explanations of ‘legal facts’ cannot account for the existence of discretion. My point will be that such explanations obfuscate what is at stake in jurisprudential debates about discretion.
Roughly speaking, the issue about judicial discretion is whether in hard cases, judges make ‘new law’ or whether they apply ‘already existing law’. Take the question: “May Elmer Palmer inherit?”. If one frames the question in terms of legal facts, then the answer to that question will be grounded in more fundamental facts, as if there was a fact to be grounded in the first place. But it can be argued that the proposition “Elmer Palmer may inherit” had no truth-makers at all (however epistemically inaccessible or metaphysically indeterminate) before the New York court ruled on the Riggs v. Palmer case.Footnote 121 As I suggested, framing this issue in terms of legal facts does not make it possible to have such an argument in the first place, because it is instrumental in providing a picture of legal content as being already there, as it were. A proponent of the legal facts picture can account for the existence of discretion if—and only if—it appears that legal content is underdetermined by its metaphysical determinants (whatever the metaphysical explanation for such underdetermination is). Indeterminate facts remain facts: There must be a fact about Elmer Palmer’s right to inherit even though the fact is either epistemically inaccessible (which, I guess, would be Dworkin’s position, if he was a metaphysician) or underdetermined by its metaphysical determinants. This begs the question whether such a fact obtains in the first place.
As Patterson and Tripkovic (who have a much more sympathetic view than mine) put it:
The talk of facts can elide the distinction between these two explanations. A grounding analysis may lead to the view that all legal facts that are established in judgments are existing legal facts, while they may become so only ex post facto. Law is an inherently dynamic social phenomenon and—at least from the standpoint of positivism—many “facts” acquire their status by retrospective recognition.Footnote 122
I think this is right. Let us assume that judicial discretion is possible. When judges exercise discretion, they will usually have recourse to moral arguments—even if such arguments are usually cloaked under the guise of legal principles (“[t]he judge will wrap his social work in the black robe of principle and precedent”).Footnote 123 It is usually easy to see through the judge’s manoeuvres. If one then has a grounding explanation of the resulting ‘legal fact’, the default position is to make moral facts its main (if not its only) determinant. Although—as I have insisted—it is not a threat per se to positivism, this default framing is nevertheless misleading. If a moral argument is used in finding against Elmer Palmer, or if the judge uses a Dworkinian integrity-oriented interpretation—which is moral through and through—then it seems that the legal fact that Elmer cannot inherit is grounded in the moral fact reflected in the judge’s moral deliberation. But lawmakers typically have recourse to moral discourse. It is part of law-making; it does not follow that the fact that the resulting law has such-and-such content is determined by moral facts. The same goes for judicial law-making. In some cases (typically the cases in which moral evaluations are embedded in the social linguistic meaning of the normative formulation, as we saw above), moral facts may determine legal content, and judges will, indeed, apply pre-existing law; but in many others, moral deliberation on the part of the judge is at the background of the issuance of a new norm—which is a thoroughly social fact. By failing to distinguish between the two scenarios, talk of ‘legal facts’—insofar as it bears on normative content—blurs what is at stake in debates about discretion. The debate is not about whether either epistemic limitations or the metaphysical indeterminacy of the grounding facts give the judge leeway in fixing legal content—that is, in ascertaining that a legal fact, such as the fact that Elmer is entitled to inherit, obtains or not. The debate is whether judges are bound by the norms applicable to the case or whether judges sometimes create new norms. On the pro-discretion view, there is simply no legal fact regarding Elmer’s right to inherit prior to the judge’s decision. The question the judge must answer is not whether it is a fact or not that Elmer has a right to inherit; the question is whether the judge ought (or ought not) to grant Elmer such a right, to make it a fact that Elmer has a right to inherit. The issue is about the extent of the normative powers of the judge—a question that it does not make much sense to ask if all that is at stake is the epistemic limitations of judges or the metaphysical indeterminacy of pre-existing facts.
4.2.2. Normative conflicts
Take another issue which, though related, is a bit more technical: the issue of normative conflicts . It is close to impossible to frame this issue in terms of legal facts. Normative conflicts occur when contrary, contradictory, or incompatible norms bear on the same case.Footnote 124 In order to conceptualize normative conflicts (at least genuine normative conflicts), one must have the concept of a norm requiring specific actions or omissions (either from officials or ordinary citizens).Footnote 125 It must be the case that it is—at the same time—obligatory to ϕ and not to ϕ (contraries), or at least obligatory to ϕ and not obligatory to ϕ (contradictories). Metarules (Lex posterior, Lex specialis, Lex superior) can help solve the conflict, but in order for them to intervene, so to speak, there must be a conflict in the first place. In a previous work, I have argued that metarules are (Bulyginian) norms of applicability: They direct judges to apply a particular norm in case of a conflict.Footnote 126 They have no bearing on the membership of either norm. For instance, I have argued that Lex posterior was not to be interpreted in terms of implicit repeal: when a citizen is faced with two norms that make it both obligatory to ϕ and not to ϕ, for instance; or when a judge is called upon to adjudicate the matter, saying that one of the two norms does not exist is misleading at best. Besides, metarules may themselves conflict with one another.
Take two norms as an example: the norms N1, “One must not cover one’s face” (with no exception whatsoever); and N2, “facemasks are compulsory”. It is impossible to reduce both norms to one single normative content. But to frame things in terms of ‘legal facts’, one must presuppose the conflict is solved at inception. It must be a legal fact that either Jones ought to wear a facemask or that he ought not to. There is no room for contradictory or contrary facts obtaining at the same time. In that sense, talk of ‘legal facts’ is inherently irenicist: It must presuppose that the normative conflict is solved beforehand, in order to establish that a given fact obtains. But metarules cannot serve to ground the truth of any of the two conflicting propositions. For example, it is not just the case that because N2 is posterior, then Lex posterior makes it the case that N1 does not belong to the legal system (if only because the conflict is what Alf Ross calls a “total-partial contradiction”),Footnote 127 and that other face coverings than facemasks are ex hypothesi clearly forbidden precisely because N1 exists.Footnote 128 So, until the judge has adjudicated the matter, there is simply no legal fact which obtains, but merely potentially non-obtaining states of affairs.
Making ‘legal facts’ the object of philosophical analysis about law makes it therefore impossible to philosophically understand situations of normative conflicts. Once again, legal contents are not always ‘already there’, waiting to be discovered. Unless one renounces the principle of non-contradiction, the fact that facemasks are prohibited and the fact that they are compulsory cannot obtain at the same time. Greenberg is quite clear—although he acknowledges that his framing of the issue is tentative—that the relevant moral obligations which are ‘impacted’ by the actions of lawmakers are all-things-considered, and not pro tanto, obligations.Footnote 129 Legal facts are therefore rationally determined by all-things-considered moral obligations such that they are affected by the actions of legal institutions. Although there may be conflicts of pro tanto moral obligations, such conflicts are resolved by morality itself. “[T]he contribution of a statute to the content of the law will depend on the on-balance best resolution of conflicts between moral considerations. Morality provides answers to questions of how conflicts between competing considerations are to be resolved”—thus yielding all-things-considered obligations.Footnote 130 Therefore, at a metaphysical level, genuine conflicts of legal norms—or of legal contents—are impossible, since legal facts are rationally determined by all-things-considered moral facts.
Of course, it is perfectly possible to dispute that genuine normative conflicts are possible, i.e., that two conflicting norms can belong to a legal system at the same time (for instance, on logical grounds). But the dispute does not even arise in these terms if they are framed in terms of legal facts, because the metaphysical apparatus makes it impossible to yield conflicting all-things-considered legal contents. My points about discretion and normative conflicts are therefore purely meta-theoretical: Although my views on the first-order issue are, I think, clear, my point was merely that adopting a metaphysical stance on ‘legal facts’ frames these debates in such a way that it tips the balance in favour of one specific side of the debate by obscuring what is really at stake.
Therefore, it makes much more sense to have a norm-first picture than a fact-first picture of the law (as well as of all normative fields). First there are norms, which may clash one with another, whose content may be more or less determinate, and which often are not dispositive of any given particular case. These norms belong to the legal system in virtue of social facts. Whether Elmer may inherit cannot be decided before various social processes (interpretive, adjudicative, enforcing-oriented) come into play and fix the norm content, fill potential gaps, solve potential normative conflicts, and exercise discretion. At the end of the process, a decision is taken, and then, Elmer may or may not have a right to inherit. Sometimes—indeed, quite often—the process is quite straightforward. It is possible from the outset to say, “yes (or no), Elmer may (or may not) inherit”. At other times, the process is complicated, and various factors—including normative conflicts—make the final decision somewhat indeterminate. To put it as plainly as possible, whether or not Elmer may inherit is not a matter of fact.
So, we should drop ‘legal facts’ as the main metaphysical components of law. Law is not made up of facts. It is made up of norms as well as various institutional and decisional apparatuses aimed at fixing normative content and reaching a decision on given states of affairs falling under the content’s scope. As such it is a wholly social phenomenon, a phenomenon which is not in need of any distinctive metaphysical explanation. Of course, law can be used as an example in metaphysical explanations of the social reality—but although this will be of great interest for metaphysicians, it will not necessarily be the same for jurisprudes.
5. Conclusion: Gardner was right
The bulk of the argument I have tried to make in this paper is in sections 2 and 3. The reframing of the main issues in legal philosophy (especially the positivism/anti-positivism debate) in terms of ‘legal facts’ ought to be resisted, especially by positivists, for two main reasons. First, legal positivism was never a theory of legal facts simpliciter, but only a theory about a very specific and distinctive set of facts about law: the fact that certain norms belong to legal systems, and others do not. What makes a given norm exist—or not—qua legal norm is (according to positivists) a matter of social fact. This is the core thesis of legal positivism, as John Gardner rightly insisted. ‘Legal facts’ are usually defined as facts about the ‘content of the law’, which is very ambiguous and covers issues completely orthogonal to positivism (such as issues about normative content and interpretation). Since the determination of normative content may, in certain circumstances, rest on moral considerations—without there being any threat to the core claim of legal positivism—it is simply not true that legal positivism is the thesis that legal facts simpliciter are always determined by social facts and never by moral facts.
Secondly—and more insidiously—the imprecision of the ‘legal facts’ framing is a Trojan Horse for holistic theories of normative content which flout the membership/content distinction and do not rest on the existence of norms atomistically individuated. Since positivism is mainly a theory of the existence of legal norms, talk of ‘legal facts’ distorts the debate with anti-positivists (such as Dworkin and Greenberg) by making it the default picture that the ‘content of the law’ is not a set of legal norms but a set of ‘facts’ (such as “the fact that Jones legally ought to pay $35 to Smith”), which can then be grounded holistically in moral facts. This in turn forces positivists to make extreme contortions in order to salvage the thesis that “legal facts are grounded in social facts alone,” thus abandoning the atomistic picture of law which is part and parcel of the core thesis of legal positivism. There can be no theory of norm-membership without there being atomistic norms to begin with. This confirms Gardner’s idea that the ‘problems’ that positivists are often called upon to address by modifying their framework “are not problems of legal positivism’s own making. They are problems of systematic misrepresentation by others.”Footnote 131
The fourth section of the paper is, in some sense, subsidiary. Talk of ‘legal facts’ has been embraced by many positivists (despite the weakening of their theory that it entailed) out of metaphysical appetite. Facts are great because, contrary to other entities—for example, norms, whose ontological status has always been disputed—they are perfect for metaphysical treatment. If grounding is about the way some facts determine others, then legal facts are obvious candidates for grounding-based explanations of the law. I have suggested that jurisprudes—of all stripes—should resist this new metaphysical turn in jurisprudence, which impoverishes our common frameworks of conceptual analysis and obfuscates many important philosophical questions about law.
A final word on positivism. Positivism has a limited domain. It is mainly a thesis about legal membership—what Gardner calls legal validity. It is not a theory of interpretation or legal reasoning. It is not a theory of what makes states of affairs such as Jones owing Smith 500 euros obtain (or not). It is only a small part of the many interesting philosophical questions that law, as a distinctive phenomenon, raises. “Legal positivism is not a whole theory of law’s nature, after all. It is a thesis about legal validity, which is compatible with any number of further theses about law’s nature.”Footnote 132 In this respect, as in many others, Gardner was right.
Acknowledgments
I presented this paper at the Lisbon Legal Theory general workshop, and I received tremendous feedback there. I am most grateful to David Duarte, Pedro Moniz Lopes, Pedro Caballeros, Jorge Silva Sampaio, and Sara Azevedo for engaging the paper with such depth and rigor. I would also like to thank Brian Bix, Mitch Berman, Dennis Patterson, and Paolo Sandro for their very useful comments and their encouragements. I am also grateful to an anonymous reviewer who identified several key weaknesses in the paper, which I hope to have at least partially addressed in this final version. My biggest debt of gratitude goes to Felipe Jiménez. Recounting in detail the ways in which he contributed to the improvement of this paper would take pages. I have learned more from our disagreements than from agreeing with anyone else.
Open access
