DOWN THE METHODOLOGICAL RABBIT HOLE

This article surveys methodological matters that shape, drive, and plague analytic legal philosophy. Section 2 briefly explicates conceptual analysis, analytic definitions, and family resemblance concepts. It also argues that central cases are used in more than one way. Section 3 presents criticisms of those concepts and methods, and suggests that some of these difficulties are due to the lack of a shared paradigm regarding a counterexample’s impact. Section 4 explains “metatheoretical” desiderata. It contends that, to date, legal philosophical appeals to such norms have not been as helpful as some suggest. Section 5 returns to the issue of concept selection by addressing whether legal theorising is an invariably “normative” enterprise. It argues that certain “normativist” methodologies, such as Dworkin’s constructive interpretation and Finnis’ appeal to the central case of the internal point of view, are unnecessary.


. Introduction
In order to assess whether general jurisprudence is "interesting", at least in the sense David Enoch thinks important (Enoch forthcoming), one must first determine whether it is a credible enterprise. To that end, this article surveys certain methodological matters that shape, drive, and plague the discipline. It also makes some modest claims of its own along the way. While legal philosophers have been more attentive to such issues lately, the task of even elucidating key ones remains incomplete. 1 This article also hopes to move to the forefront certain (interrelated) meta-philosophical concerns that have gained far less coverage -especially their role within problems that have already been addressed by legal philosophers.
Section 2 explicates certain methods and views, such as conceptual analysis, analytic definitions, and family resemblance concepts. Those sufficiently familiar with these can skip to section 2.4, where, in addressing the central case method, I argue that central cases are used in more than one way. Section 3 then presents some (welland lesser-known) criticisms of the above-mentioned philosophical methods and the data on which they rely. The goal is to make more perspicuous the problems of relying on intuitions and examples in various contexts. I also suggest that some of these difficulties are due to the lack of a shared paradigm governing the impact of a counterexample.
Section 4 then briefly explains "meta-theoretical" desiderata: norms by which to help construct, evaluate, and compare concepts and theories. Analytic legal philosophers have invoked these a lot recently in regards to (i) concept and theory formation and (ii) grounds for comparing rival legal theories. Some legal positivists also appeal to these desiderata in order to rebut a certain methodological critique levied against them by their "normativist" rivals (e.g., natural lawyers and interpretivists). I nevertheless argue that, to date, legal philosophical appeals to such norms have not been as helpful as might appear.
Finally, section 5 returns to the matter of concept selection by addressing one of the dominant methodological issues in analytic legal philosophy: whether legal theorising is an invariably "normative" (in the sense of morally or politically evaluative) enterprise. I present some reasons for thinking that certain "normativist" methodologies, such as Dworkin's constructive interpretation and Finnis' appeal to the central case of the internal point of view, are unnecessary.

. 1 . Concepts, Conceptions, Words, and Things
To begin, it will prove helpful to outline popular philosophical distinctions amongst a thing, a concept, and a term, phrase, and word.
A thing (or, what some philosophers, for better or ill, treat as a synonym, a phenomenon) is a person, place, object, or state of affairs. The concept (of) A CONCEPT is contested. 2 A popular philosophical view construes it to be a mental representation (of a thing, say), which involves beliefs and propositional attitudes, and which has an internal structure containing more basic representational elements. 3 This view additionally holds that a concept can be a cluster of beliefs constituting something. For example, there are concepts of things that do not exist, such as unicorns and witches. 4 Words, terms, and phrases are symbols used in a language, which can be used to express a concept. 5 There can, however, be more than one word available by which to express a given concept. For example, "You can't do that, it's against the law" and "You can't do that, the legislature has prohibited it" appear to express or utilise the same concept (Leiter 2007, p. 123). 6 A few philosophers also carve a concept-conception distinction. A concept concerns a sort of category (or an entire field of enquiry), while a conception refers to particular viewpoints about that concept. 7 For example, there are various conceptions of the kind of a legal right that correlates with a duty. Hohfeld's "claim", for example, differs in certain respects from H.L.A. Hart's "right-correlative-toan-obligation" and Joseph Raz's account of a right as the grounds 2 The contemporary philosophical convention is to use all capitals to mark concepts. This is done to distinguish them from denotations of words or terms (for which scare quotes are used instead). 3 For some alternative views about the nature of a concept, see Laurence and Margolis 2011. 4 The view of concepts as beliefs generates an obvious regress problem, as the constituent "representations" are themselves in need of explanation (see Laurence and Margolis 2011). 5 Can we have concepts without having associated words? This issue matters if you believe that philosophical analyses can apply to cultures or peoples beyond your own. Just because I use the term "a right" and you use the term "subjektive recht" it might not follow that we are necessarily referring to the identical concept. Moreover, just because a community lacks a given word, it does not follow that they lack a particular concept. Take DIMENSIONALITY, for example. A given tribal community may lack the words for "height", "length", etc. However, since the community appreciates that its water source is a two-hour's walk away, and that it takes about ten bear-hugs to climb a tree to get to its fruit, do the community members not have the relevant concepts of dimensions? See, e.g., Dagger 1989, pp. 296-298. 6 Andrei Marmor (2013, pp. 210-211) appears to claim that all of a word's various meanings exhaust the relevant concept. This is false if different words, terms, or phrases refer to the same concept, e.g., if the words "legislature" and "law" do not have identical meanings, but can nevertheless refer to the same concept (LAW). 7 See, e.g., White 1975, pp. 113-114. of duties and other normative positions. 8 The concept-conception distinction's soundness is nevertheless disputed. Some philosophers deny that there are different conceptions of a concept; instead, people merely have different beliefs about a concept. Other philosophers believe the distinction is misleading because there need not even be "a" concept to which the various "conceptions" refer. 9

. . Conceptual Analysis and Analytic Definitions
Amongst other things, philosophy aims to help us better understand concepts and the things/phenomena our concepts are concepts of (if anything). However, there is no agreed-upon, established set of universally endorsed or preferred philosophical procedures. 10 One (set of) method(s), traceable back to at least the Platonic dialogues, 11 is called conceptual analysis. It is worthwhile to focus on this form, as it continues to be a dominant one in analytic legal philosophy. 12 On the assumption that concepts bear internal structures, a conceptual analysis explains a given concept in terms of its more basic components; you decompose it into its constituent parts until you are left with more (if not irreducibly) basic concepts. 13 In the words of its most famous contemporary apologist, "conceptual analysis is the very business of addressing when and whether a story told in one vocabulary is made true by one told in some allegedly more fundamental vocabulary" (Jackson 2000, p. 28).
Conceptual analyses are said to admit of immodest and modest versions. "Immodest" ones purport to be able to tell us something about the very things our concepts are concepts of. By contrast, a "modest" conceptual analysis merely aims to elucidate a concept qua 8 Hohfeld 1913;Hart 1982;Raz 1986;Raz 1994. 9 E.g., Halpin 1997, p. 11. 10 E.g., Huemer 2015, p. 52. 11 "Preface", in DePaul and Ramsey 1998 One alternative is (soft) methodological naturalism, which calls for philosophers to employ or emulate the methods of the hard and social sciences. E.g., Leiter 2017. For decades, Brian Leiter has called upon legal philosophers to see the naturalist light and conceptual analysis' problems. Even so, he makes two concessions. First, modest conceptual analysis, to be explained below, suffices for certain "hermeneutic" concepts: those "whose reference is fixed by the role it plays in how people make sense of themselves and their social world" (Leiter 2011, p. 512). Second, some modicum of analysis appears to be necessary to get a naturalized jurisprudential inquiry off the ground. E.g., Leiter 2007, p. 65. 13 E.g., King 2016 concept; it does not pretend to be able to describe "the world" itself (Jackson 2000, p. 42). 14 Such analyses are also said to offer "conceptual clarifications". As words and/or concepts admit of ambiguity, vagueness, etc., many philosophers believe they can tidy up this messiness by either (A) rendering a concept's features more perspicuous (so that the concept might be better understood), 15 or (B) by replacing the concept with a more precise one. 16 Others can then employ the refurbished concept in their philosophical, scientific, or social scientific work. 17 Conceptual analysis often proceeds by trying to distinguish what is necessary to a concept from that which is merely contingently associated with it. Being "necessary" means that a given feature (a) is constitutive of the concept and (b) is exemplified in every token of the type. 18 In other words, a necessary feature's failure to obtain in a given instance of the thing the concept represents is deemed to be inconceivable. Two caveats are warranted here, though. First, this is not to say that conceptual analysts assume that a target concept, 14 Once again, though, see Leiter 2011, p. 512: conceptual analyses of "hermeneutic" concepts can yield knowledge about the world, while analyses of "natural kind" concepts only explain ordinary folks' "talk" about them, as opposed to the best understanding of these concepts, which are instead the "deliverance[s]" of the sciences. 15 For example, P.M.S. Hacker suggests that we clarify a concept, expressed by means of a word (the usage of which we as language users have relatively mastered), by examining its uses to make explicit its combinatorial possibilities, its entailments and incompatibilities, the presuppositions of its use, its purposes, etc. Interview with Peter Hacker, Emeritus Professor of Philosophy, University of Oxford at his home (Oxford, March 7, 2016). (The problem of relying on only one word in order to understand a concept, however, has already been suggested above. See supra note 6.) 16 E.g., Jackson 2000, pp. 38, 46-47 (certain "intuitions" might have to be "massaged" in order to make "sensible adjustments" to a "folk" concept). See also Stoljar 2013, p. 237, citing Carnap and Quine on "explicative definitions", which "make more exact a vague or not quite exact concept used in everyday life [ . . . ] or rather of replacing it by a newly constructed, more exact concept" (Carnap 1956, pp. 7-8), or "improve upon the definiendum by refining or supplementing its meaning" (Quine 1951, p. 25). 17 E.g., Leiter 2007, p. 134. But see infra note 60 and the accompanying text on scholars (such as Raz and Dickson) who think that legal philosophising should (at least in part) focus on the way people use concepts to understand themselves. This sort of focus might lead one to oppose the "replacement" option. 18 The terms "essential" and "necessary" will be used interchangeably in this article. However, some scholars treat essential properties as a subset of necessary ones: those that are specific or unique to a particular concept. E.g., Patterson 2012, p. 52, n. 8. Historically, essential or accidental properties were predicated of things, not concepts. Bix 2003, p. 541, n. 17. in contrast to one or more of its features, is itself indispensable. 19 Second, it is not true that all philosophical analysis aims to focus on, let alone capture, a concept's necessary features. Sometimes, and particularly in legal philosophy, there is great value in examining the important yet conceptually contingent. 20 Whether for the sake of rendering a concept more perspicuous, or as part of an effort to advocate for its replacement with a more precise one, conceptual analyses often levy definitions. Not only are there different kinds of definition, but philosophers also claim to be able to define words, concepts, and things (Gupta 2015). 21 A philosophical or "analytic" definition aims to provide a set of necessary and sufficient conditions for a concept. These specify the concept's (as opposed to a word's) extension. 22 This approach actually makes several assumptions: (I) that we can "capture precise, informative, tractably specifiable, noncircular necessary and sufficient conditions for the application of [concepts]"; (II) that "statements expressing these analyses are analytic truths"; and (III) that these analyses can be produced via a priori reflection. 23 Do analytic definitions suffice for identificatory purposes, i.e., for locating instances of things/phenomena that the concept represents? Some philosophers seem to suggest as much.
Although sufficient conditions, necessary conditions, etc., are sometimes called "criteria" [ . . . ] the sense of "criterion" in which an analytic definition provides a criterion for something's being the sort of thing to which a term applies is a very strong one: (a) the "criteria" I am speaking of are necessary and sufficient conditions of something's 19 For some legal philosophers' defences of the notion of contingent-conceptsbearing-necessary-features see, e.g., Raz 2009a;Giudice 2015;Bix 2003. 20 E.g., Giudice 2011 (on the theoretical worth of conceptually contingent aspects of law); Schauer 2012 (on the worth of "generics" and the statistically significant, as not all philosophising, or even most general jurisprudential work, concerns the search for a concept's necessary features). 21 Sometimes stipulative definitions are employed. These are rules for the usage of a word, along lines a philosopher designs. Stipulations ostensibly aim to help elucidate (by suggesting particular usages), not to police word usage per se. Being concerned with word usage, moreover, definitions of this sort need not (aim to) include all of a concept's relevant features. See, e.g., Raz 2009b, p. 41. Space does not allow further elucidation of the matter here, but some philosophers also claim to present real definitions: definitions of things (as opposed to words or concepts). E.g., Rosen 2015.
22 See Laurence and Margolis 2011. For an excellent summary of the history of this kind of definition in philosophy, see Baker and Hacker 2009, pp. 201-208. 23 Graham and Hogan 1998, pp. 271-272. being an A; and (b) by means of them people can and do determine that something is an A. (Putnam 1979, p. 67) Philosophers test an analytic definition by subjecting it to actual or hypothetical cases to see if the definition can cover them. 24 If it fails to do so, then any such case (qua counterexample) is treated as being sufficient grounds for revising or falsifying the definition. (The mark of a successful analytic definition, then, is that is impervious to counterexamples). However, some believe that not every counterexample, or set thereof, suffice for this task. Instead, one can propose an error theory to explain why the intuition is itself mistaken. 25 Two or three kinds of cases are employed to do the falsifying. First, imagine that a candidate analytic definition of (the concept of) X construes feature φ to be necessary to X. To test this, just present or imagine a plausible case of X that lacks φ. The counterexample's existence is taken to show that, pace the definition, feature φ is not necessary to X after all. Alternatively, present a case that is not thought to be an example of X, but which nevertheless satisfies the analytic definition's conditions. This kind of counterexample is taken to show that the definition's features are insufficient for X.
There is a lot to unpack here. For one thing, how do we know that the first case is actually a genuine instance of X lacking feature φ, or that the second case is not really an instance of X despite meeting the analytic definition's conditions? Philosophers commonly claim to rely on their intuitions in order to make these categorisation judgements. 26 However, they do not agree about what they mean by "intuition", 27 while the various philosophical accounts of INTUITION "range from little more than hunches to at most expert opinions on disputed issues" (Gutting 1998, pp. 6-7). 28 Additionally, quite a few philosophers treat "ordinary" or "folk" linguistic usages, patterns, or understandings as the data to be used 24 E.g., Daly 2010, pp. 48-49;Jackson 2000, pp. 28-29, 35-36. 25 E.g., Booth 2014, p. 1. 26 E.g., Ramsey 1998he process of appraising definitions requires comparing and contrasting the definitional set of properties with intuitively judged instances and noninstances of the target concept"). Cf. Leiter 2007, p. 184;Weinberg 2016, pp. 290-291. 27 Carrie Jenkins provides a catalogue and explanation of the different conceptions and candidate roles of "intuition" in philosophy. Jenkins 2014, pp. 92-97. Cf. Weinberg 2016 Rethinking Intuition has become a sort of locus classicus for its debates on the merits of utilising intuitions (as data, as evidence, as reasons, etc.), conceptual analysis, and the method of reflective equilibrium. in such analyses. 29 Sometimes, though, the data set is restricted to technical discourses, e.g., lawyers' talk for the purposes of a legal theory. 30 Philosophers then "test" their intuitions about cases using this linguistic data. Indeed, it is sometimes said that they judge what would be appropriate or valid to say vis-à-vis their "linguistic intuitions".
Here is a legal philosophical example of the method of hypothetical cases. Arguing against a long-standing belief amongst legal philosophers and others that law is necessarily coercive, Joseph Raz provides a purported counterexample: the law of a community of angels. Angels, he suggests, would need laws to coordinate and structure their social order. Being highly virtuous, though, their legal system would not require (the utilisation of) coercive measures in order to establish conformity with given laws or for the system as a whole to be effective (Raz 1999, pp. 156-161). This seems to be a case of law without coercion. Coercion, Raz concludes, is therefore not a necessary feature of "our" concept of LAW; conceptually, it is only contingently related to law, albeit present in most or all human legal systems.

. 3 . Family Resemblance Concepts
Analyses do not always necessitate the levying of "analytic" definitions, which try to capture a concept's essence in terms of necessary and sufficient conditions. In fact, quite a few legal philosophers follow Wittgenstein in rejecting the idea that all concepts must possess necessary conditions. That is to say, they reject the notion that the various things falling under a concept must exhibit some common feature, or group of features, which invariably obtain in every token of the type.
Instead, the various things falling under a given concept relate to each other in diverse, overlapping ways. A concept presents "a complicated network of similarities overlapping and criss-crossing" (Wittgenstein 2009, § 66), where there is "no one thing in common in virtue of which we use the same word for all -but there are many different kinds of affinity [ . . . ]" ( § 65). This is called a family resemblance concept. For just as members of a biological family can 29 Jackson 2000. See, for example, Hart 1997, pp. 3, 240 (about what any educated man would be able to identify vis-à-vis the salient features of legal systems). Others reject this approach for certain sorts of concepts, e.g., for "natural kind" concepts. See Stoljar 2013, pp. 235-236. 30 E.g., Hohfeld 1913. exhibit shared physical characteristics (e.g., certain members having the same nose shape, some the same shaped eyebrows) without there being one that is common to the whole family (e.g., everyone having the same eye colour), so too the things falling under a given concept may exhibit overlapping features without there being features that are common to all of them ( § 67). 31 Wittgenstein's famous example of a family resemblance concept is that of GAME (and thus the multifarious things we call games) (2009, § § 69-71), but here is a legal philosophical example. Wesley Hohfeld (1913) claimed to have identified four conceptually basic types of rights: claims, privileges, powers, and immunities. 32 Many have since asked, what makes these all types or examples of "the" concept, A RIGHT? 33 Is it that they are all entitlements? That they are all normative advantages? Do they all protect the right-holder's will or interests? If A RIGHT is a family resemblance concept, then there is no common necessary feature that all rights exhibit. Instead, they are just related to one another in distinct, overlapping ways.

. 4 . Central Cases
Analytic definition (bearing necessary and sufficient conditions) and family resemblance are not the only ways by which to understand or explain a concept. An alternative, or supplementary, basis is the central case method (aka the paradigm case or focal case method). 34 The method is of ancient provenance. 35 As Alex Langlinais and Brian Leiter explain: "A central case analysis of some phenomenon identifies some subset of possible or actual instances of that phenomenon as explanatorily privileged. The members of this subset are the paradigm or central cases of the phenomenon [ . . . ]" (2016, p. 682).
Like a family resemblance concept, a central case is not, or at least it need not be, "essentialist", i.e., it need not be said to be 31 Wittgenstein also analogises the extension of a concept to the spinning of a thread by twisting different fibres. "And the strength of the thread resides not in the fact that some one fibre runs through its whole length, but in the overlapping of many fibres" ( § 67). Space does not allow for a discussion for ascertaining when, if ever, a concept is of the family resemblance or some other variety. For a discussion of this and related issues, see Baker and Hacker 2009, pp. 219-222. 32 Cf. Pound 1959; Frydrych 2017. 33 E.g., Hart 1983, p. 35, n. 15. 34 Cf. Langlinais and Leiter 2016, pp. 672-673. 35 John Finnis traces the idea of a "focal" meaning (central case) back to Aristotle. Finnis 2011, pp. 9-11. composed of immutable, necessary properties. 36 A central case of X contains important features, but not every such property need obtain in every instance of X. 37 For example, Tony Honoré presents "the" central case of ownership in a modern "liberal" legal system. The case includes eleven standard instances or features, e.g., the right to possess, the right to sue, the right to manage, etc. Still, he does not insist that every feature must obtain in every case within a liberal legal order (Honoré 1961).
While philosophers may be becoming more aware of the extent to which the selection and employment of such cases is reliant upon "intuition pumping", what nevertheless seems to have escaped their notice is whether such cases are always employed for the same purposes. I wish to suggest that there are actually two different philosophical uses of "central" cases. Usage type A is well known.
[Central cases] are privileged in two respects. First, the central cases are privileged insofar as a theory of the phenomenon is primarily concerned with explaining the important features of these cases. Second, the central cases are explanatorily prior to those instances of the phenomenon that are not members of the set of central cases. (Langlinais and Leiter 2016, p. 682) This requires supplementation. For one thing, central cases are also used to help explain "peripheral" cases. A philosopher adjudges φ to be a central case of X. He or she then compares φ with ψ, a contested case of X. Using φ's features as a benchmark, the philosopher can look at the features ψ does or does not possess in order to note the overlap, similarities, or differences in the two cases' features. 38 Second, philosophers additionally use φ here as a basis for identifying or disqualifying other candidate cases from counting as X, i.e., to help determine whether ψ really is a token of the type X.
For example, much of H.L.A. Hart's exposition of LAW in The Concept of Law concerns the "modern municipal" legal system. He 36 E.g., Stitch 1998, p. 104;Ramsey 1998, pp. 168-172. P.M.S. Hacker appears to distinguish Wittgenstein's family resemblance concept from Aristotle's focal meaning on the grounds that, while the former includes "paradigms" by which we can compare to peripheral cases, it has "no single centre of variation" as a focal case does (Baker and Hacker 2009, p. 215). 37 Socio-legal pluralists reject the central case method, claiming that it is nevertheless essentialist. E.g., Tamanaha 2001, pp. 149-151. 38 See Langlinais and Leiter 2016, p. 682. For some concerns about how to (I) determine a central case's contents and (II) compare it with other cases see Ramsey 1998, pp. 168-169. then uses the municipal system qua central case to compare with -what he deemed to be-peripheral cases, such as international law. 39 (Hart also claimed to present some necessary conditions of LAW or a legal system. 40 So, whether or not one believes that the central case method must stand in contrast to analytic definitions or family resemblance approaches to concepts, it should be noted that legal philosophers sometimes take a syncretist, or methodologically pluralist approach.) Philosophers, I wish to suggest, also use central cases in yet another way. Usage type B: some (supposedly) uncontroversial cases, understood "pre-theoretically", are used as benchmarks for evaluating a philosophical account. Such cases are treated as part of an adequacy condition: a philosophical theory's inability to account for them is treated as a reason for adjudging it to be a failure. Just like those who levy a counterexample to an analytic definition, philosophers who present such cases would not accept that they, unlike perhaps other counterexamples, merely count as exceptions to the theory; for the presenters rather believe that these sorts of cases suffice to falsify the theory.
The difference between the two types of central case employment can be summed up thus. Usage type A treats case φ as a tool for: explaining both other cases and certain things/phenomena of which they are cases; for determining whether those other "peripheral" cases really are genuine tokens of type X; and for delimiting the focus of a theory about X. Usage type B determines (or assumes) that case φ is not only an incontestably genuine token of type X, but one of such importance that -perhaps unlike other cases that might serve as counterexamples-its inclusion is deemed to be indispensible to a successful account of X while its exclusion marks the account's failure.
Here, to my mind, are some examples of usage type B. In response to Raz's thesis that law claims supremacy over all other normative systems governing or affected by those subject to it, Brian Tamanaha presents the medieval European legal situation as a (set of) counterexample(s): Raz's strictures would force the conclusion that there were no legal systems throughout much of Europe during the medieval period. Dur- 39 Hart 1997, pp. 3, 17, 81, 279-280, andchap. 10. Cf. Finnis 2011, pp. 10-11;Bayles 1992, p. 13. See Langlinais and Leiter 2016, pp. 672-674, on whether Hart's method was in fact essentialist. 40 E.g., Hart 1997, p. 116. ing this period several recognized bodies of law coexisted -including ecclesiastical, feudal, merchant, manorial, royal, and municipal-which did not typically claim to regulate all types of behaviour, and did not claim supremacy over all other normative systems. (Tamanaha 2001, p. 139) 41 Regardless of whether Tamanaha's claims are correct, his cases are not presented as mere exceptions to Raz's thesis; they purport to falsify it. In other words, Tamahana would not accept, as a potential reply, the claim that these cases are either mere aberrations, or just insignificant outliers that Raz's account can comfortably ignore. As a second example, pace the will theory of rights, Neil Mac-Cormick raises the cases of children's and mentally incompetent persons' legal rights (MacCormick 1977(MacCormick , 1982. The will theory is not merely deemed to be under-inclusive for (supposedly) excluding such cases, but to be false for that very reason. MacCormick's argument is not simply a morally or politically evaluative judgement of the consequences of the will theory's excluding certain sorts of agents from rights-bearing status. He is also claiming that it is an analytic or explanatory failing to be unable to account for such cases. (Whether or not the will theory can actually account for them is irrelevant for our purposes. Again, what matters are rather how and to what ends MacCormick presents these cases.) One might of course challenge the idea that this second type of case usage is really a matter of the central case method. What makes it so, rather than, say, a category fitting in between central cases and other sorts of cases that (ought to) count as exceptions to an account but which do not suffice to falsify them? Why, moreover do these instances not simply represent the method(s) of actual and hypothetical cases discussed in section 2.2?
One reason for deeming these to be instances of central cases is their being presented as being indispensable to an account -regardless of whether there is (e.g., scholarly) concurrence about their status as such. A philosopher could, after all, treat one case as being indispensable to a theory and a second as being genuine-but-nonethelessignorable for theoretical purposes. 42 Regardless of what you think 41 For Raz's thesis see Raz 2009b, pp. 117-121. 42 Of course, not all (purported) counterexamples are presented as though they were central cases. For example, arguing against Raz's thesis that law necessarily claims moral legitimacy, Matthew Kramer presents an analysis of a hypothetical wicked legal system. He holds that the particularly heinous regime need not even feign belief in the morality of its laws, or of a given law. Kramer deems this case the most apt label is for Usage type B, it is important to note the very existence and practice of using cases for such purposes.

. 1 . Do Legal Concepts Bear Necessary Features?
The issues concerning central cases only scratch the surface, as a great many methodological problem points in analytic (legal)  to be sufficient evidence to falsify Raz's "necessity" claim, but he might very well deny that his imaginary regime represents a central case of law or of a legal system (Kramer 1999;cf. Kramer 2007, p. 186). But see Shapiro 2011, pp. 391-392, who argues that there are no reasons why unjust regimes cannot count as "central" cases of law. 43 E.g., Tamanaha 2015. This is not the place to prove whether concepts bear necessary features. However, it may be worthwhile addressing one strand of antiessentialist criticism of conceptual analysis that relies on W.V.O. Quine's attack on analyticity. Quine's argument is sometimes taken to entail that concepts do not bear essential features and cannot be known a priori. For one thing, though, there are reasons for thinking that Quine's argument about analyticity either applies to a narrow category of concerns, or fails altogether to undermine the concept. E.g., Grice and Strawson 1956;Hacker 2013. For another, Michael Giudice provides reasons for doubting the supposed entailments from Quine's argument. Giudice argues that conceptual analysis looks for necessary features, not analytic ones (necessity and analyticity being different). Conceptual analysis tests intuitions by the use of (hypothetical and/or real-world) cases to look for the necessary propositions of a concept. But these propositions do not necessarily bear meanings that would be inconceivable otherwise, i.e., are not necessarily true, analytically. For example, when certain legal philosophers hold that law is not necessarily coercive, they are not at the same time claiming that it is inconceivable that law is or could be inherently coercive (Giudice 2015, pp. 9-10, 90-109;cf. Quine 1951). Giudice's particular response to Quine's challenge here nevertheless seems to fail to answer whether legal concepts: bear necessary features; are instead family resemblance concepts; or (if construed as a genuine "middle ground" between the two previous options) admit of central cases. Now, some philosophers nevertheless provide the beginnings of a plausible argument that, even if concepts are culturally, linguistically, or temporally (e.g., epochally) delimited, it does not follow that they lack essential properties. Quite a few legal philosophers, for example, hold that the concepts in their area of inquiry (LAW, A COURT, A RIGHT, etc.) are historically or culturally contingent and can evolve over time. Hence, a certain feature can be deemed to be "necessary" to a concept, as opposed to being merely contingently related to it, even though the concept itself may be inessential. 44 This has a prima facie appeal. That is, until these philosophers go on to claim to have presented analyses of "our" concept of X, the "Western" concept of X, the concept of X of the early modern period, etc., as if those delimiting terms necessarily reflect singular notions. But who counts as the "our"? Is the concept jurisdictionally delimited, regionally so, civilizational, or just individualistic? How can it be established that it is either a concept with a given linguistic-temporal boundary, or instead one that has obtained across different eras (centuries? millennia?), peoples, and/or cultures, even if modified slightly? 45 In other words, even if one buys the idea of contingent-concepts-bearing-necessary-features, it remains unclear how to delimit their "spatial" or temporal extensions. 46 Without this information, it seems impossible to identify what a conceptual clarification or analysis actually applies to.

. 2 . Intuitions and their Clashes
Philosophers claim to use their intuitions about cases to determine a concept's extension, to sieve the conceptually necessary from the contingent. This sometimes involves using actual and hypothetical cases, the latter exemplified by Raz's angels case above. Even assuming arguendo that some concepts bear necessary features, however, both the technique and tools used to sieve them from the conceptu-44 E.g., Bix 2003, p. 549;Coleman 1998, p. 393, n. 24;Raz 2009a, pp. 27-46, 70-71. Raz also thinks that we can use our concepts to understand other cultures' or language users' concepts via a comparison of similarities and differences (Raz 2009a). 45 E.g., Barber 2015, pp. 812-816, on inconclusive "common" understandings. 46 A related problem is that if you only look to ordinary or technical word usage in order to run a conceptual analysis, all you might produce is an "ethnographic lexicography" -as opposed to an explanation of a generally held concept. E.g., Finnis 2011, p. 426;Leiter 2007, pp. 196-197. But see Farrell 2006Farrell , pp. 999-1003 argues persuasively that modest conceptual analysis is not mere lexicography and that it has elucidatory power over (at least) "hermeneutic" concepts. ally contingent are suspect. If philosophers really must employ their intuitions, then their analyses may be completely flawed for utilising unreliable sources.
We can, to be sure, test our intuitions about possible cases to fix the concept of "space" or the concept of "representational content," but since such intuitions are hostage to parochial bias, lack of empirical knowledge, and all variety of selection effects, there is no reason to think such intuitions and their deliverances deserve epistemic weight. (Leiter 2007, p. 184; internal citations omitted.) 47 There also seems to be a significant difference between what conceptual analysts purport to do and what they actually do. Particularly, determining that a candidate case of X fails to actually constitute a genuine instance of X for lacking feature ψ seems to be a matter of judgement -but not a deductive one. There being no "pretheoretical" agreement about what counts as a genuine token of a type, how philosophers go about their analyses -in part by employing tokens as tools-is the opposite of what they claim to be doing.
On the classical theory of concepts, we decide whether to apply a concept to an object by comparing the object's properties with the properties listed in the concept's definition. But philosophical practice suggests the reverse: we intuitively judge whether a concept applies to an object, independent of any definition, and we evaluate a definition by how well it fits with the correct usage of the concept. (Huemer 2015, p. 53) 48 For example, Raz's argument that the society of angels has "law" reflects his judgement that the example really constitutes a hypothetical case of law, and not of something else. Others disagree, and therefore deny that the Angels case constitutes a bona fide counterexample to 47 Cf. Harmon 1999, p. 140. 48 Cf. Ramsey 1998, p. 164); Baz 2016, pp. 115-116: When asked, different people can and do disagree about whether candidate X, an actual or hypothetical case, counts as a genuine token of the type Y. See also Priel 2007, p. 187: Conceptual analysts claim to start with "pre-theoretical" samples of law, and then develop a theory that tries to show the necessary and important features these samples (and all other instances) have. However, conceptual analysis lacks the means by which to resolve disputes about what counts as law in the first place, at a pre-theoretical level. For a response to Priel, see Giudice 2015, pp. 51-55. the view the law is necessarily coercive. 49 Conceptual analysis alone probably cannot resolve such disputes (aka "clashes of intuitions").
It might be suggested that the method of reflective equilibrium can alleviate philosophical worries about intuitions. 50 Here is how the "narrow" version of reflective equilibrium is supposed to work. 51 First, identify a set of intuitions/beliefs about a given matter, e.g., the concept of LAW. 52 Second, try to generate a set of principles and/or theories that systematize and account for those intuitions/beliefs. This process will likely expose conflicts: (I) amongst one's initial intuitions; (II) amongst one's initial set of theories/principles; and (III) between those initial intuitions and theories/principles. Therefore, reflect upon those intuitions and principles in order to eliminate, add, or revise some of them until you are left with a coherent set (i.e., consistent based on their contents). On the "wide" version of the method, by contrast, one seeks to place into equilibrium one's initial beliefs about a matter, theories or principles about that matter, and some third consideration: whether that be intuitions about some other matter, theories of some other sort, or something else. 53 "Reflective equilibrium" is actually the end-point of this deliberative process (Daniels 2016). While that might turn out to be an unachievable ideal state (Cath 2016, p. 215), proponents nevertheless deem the method to be valuable, if not indispensible.
There are reasons to doubt reflective equilibrium's merits. The critiques are well known, but here are some important ones. For one thing, on a coherentist understanding, the method can incorporate beliefs that are known to be dubious or false (Stitch 1998, p. 100). 49 E.g., Himma 2001, p. 308. While I only wish to flag the difficulties in establishing something as an instance of a case and the lack of shared criteria for doing so, there are nevertheless other grounds for thinking that Raz's Angels argument is not a genuine counterexample. For example, in the Christian narrative God punished Satan by expelling him from Heaven. . . (coercive liability for rule violation). 50 My thanks to an anonymous reviewer for raising this point. 51 This account largely follows Yuri Cath's "initial sketch" of the process (Cath 2016, pp. 214-217). 52 The method's proponents dispute: (I) whether the items in question are intuitions, judgements, or beliefs (where the latter is deemed to be different in kind from the former); (II) whether any and all intuitions should be plugged into the reflective process, or if only some subset of "considered" instances warrant inclusion; and (III) whether one tries to account for the intuitions/beliefs with theories or principles. 53 "The important point is that the philosopher seeks to construct an ever more comprehensive system of beliefs and to bring these beliefs into equilibrium via a process of mutual adjustment [of intuitive judgements and theories]" (DePaul 1998, p. 296).
For another, it is questionable whether you can compel yourself to believe all sorts of things, such as whether an intuition is inaccurate (Cummins 1998, p. 125). Third, privileging one's own, or one's culture's intuitions over others for the sake of reflection has been accused of evincing a form of bias, be it ego-or ethnocentricity, or some other form (Cath 2016, p. 222). 54 Fourth, it is wholly unclear how one could establish that one's beliefs are actually in a state of wide reflective equilibrium (Stitch 1998, p. 101).
There are, I think, further worries of tasking reflective equilibrium to resolve issues surrounding the reliability of intuitions as data. First, since a given counterexample may be insufficient to falsify a theory or principle under the method (Cath 2016, p. 228), on what grounds are we to determine which intuitions and which theories/principles to retain, discard, or amend? Second, if intuitions are unreliable qua data (let alone as evidence), why not deem them to be defective tools for the purpose of trying to achieve a reflective equilibrium? Why think that a given intuition can be remedied by employing other instances (i.e., intuitions about cases, the weight of a case, the explanatory power of a theory, one theory's simplicity relative to another) in order to decide upon what to keep, discard, or amend? If these charges are sound, and since the method calls for mutual adjustment of intuitions and theories/principles, it would seem that a coherentist account of reflective equilibrium for its own sake is no real virtue, and that undertaking the method on a truthoriented approach would be a fool's errand.

. . What Does a Counterexample Do?
As already seen above in the contexts of assessing analytic definitions and using central cases, what does a counterexample do and entail? Does it show that an account or definition of X: is wholly mistaken; is under-inclusive but salvageable; or that it faces a prima facie challenge but is otherwise unaffected by the counterexample's existence? In relation to rules, propositions, and theories, when does a counterexample: (A) constitute an exception to them; (B) delimit them; (C) falsify them; or (D) have no impact upon them? Further, if more than one counterexample is required for falsification, how many are needed? What is the threshold? What are the criteria?
Like everyone else, philosophers seem to lack a shared paradigm governing a counterexample's impact. Some believe that counterexamples can be used to examine a concept's extension and can serve as (at least prima facie) reasons for disbelieving claims of certain features being "necessary" to a concept. They take either a single counterexample, or some unspecified number of them, as sufficient grounds to falsify a theory, an analytic definition, an account of a concept (e.g., by presenting a purported central case as a counterexample), or a proposition. Other philosophers do not, proposing error theories instead. Still others think counterexamples merely circumscribe or delimit a theory, or mark exceptions. Others still are not bothered by counterexamples at all. 55 Determining how to account for the seeming hierarchy (or hierarchies) of cases, and the differently tiered cases' employment and respective effects appears to be under-discussed in legal philosophy, and in meta-philosophy more generally. Philosophers seem to intuit their way to a hierarchy of cases based on how important the instances are to a concept, theory, practice, etc., and how weighty a role they can serve qua counterexamples, i.e., as exception-carvers, circumscribers or delimiters, or falsifiers. The use of cases also gives rise to another methodological puzzle: why think that a given counterexample evidences an account's bearing exceptions, or being false, rather than the counterexample itself being a "degenerate", "imperfect" case? Both legal positivists and normativists (i.e., interpretivists, natural lawyers, etc.) employ the central case method(s). With the notable exception of John Finnis (i.e., his account of the central case of the internal point of view), however, many philosophers do so without fully explaining or justifying their selection, i.e., what makes theirs a bona fide "central" case, let alone try to establish the technique's validity. 56 55 Brian Weatherson makes a narrower point. He thinks the effects of a given counterexample are treated differently in the various branches of philosophy (epistemology, moral philosophy, metaphysics, etc.), and that various kinds of counterexamples have distinguishable weights. Weatherson nevertheless seems to believe that there is a correct answer to this methodological matter -at least in epistemology, such that the famous Gettier cases should not be treated as falsifying the theory that knowledge is justified true belief (Weatherson 2003, pp. 1-2, 8

. 4 . Why Re-Craft Concepts?
As mentioned above, some philosophers believe they are adept at re-crafting concepts. Is it worth undertaking such projects, though -especially if the reformed product does not track the usual ways people actually employ the (unreconstructed) concept? "Conceptual clarification", in this sense, may deleteriously affect what researchers get out of their queries (Leiter 2011, p. 516). For example, some Hohfeldians favour restricting the concept of A RIGHT to a Hohfeldian claim. 57 As they admit that Hohfeld's conception is itself a "clarification" or "correction" of ordinary and lawyerly discourse, however, why believe we can get to the truth of the matter about rights by restricting our focus to it?
Bald faith in empirical work is no escape from the difficulties regarding how best to determine what a concept really is, let alone how it should be clarified. Additionally, if (at least some) counterexamples do not suffice to distinguish what is conceptually necessary from what is conceptually contingent, yet legal philosophers continue to present concepts/conceptions, of LAW, say, containing only those purportedly necessary features -ascertained via conceptual analyses-might these not also be distorted pictures? Might this not be, for example, what Raz's critics could say about his philosophical conception of LAW, i.e., it being in part the product of the Angels case upon his viewpoint?

. Meta-Theoretical Desiderata
Meta-theoretical desiderata are norms that can be used to help construct, evaluate, and compare rival "conceptions" and theories. Though usually found in the philosophy of science, analytic legal philosophers have invoked such desiderata a lot recently. Some legal positivists particularly appeal to them in order to rebut a certain methodological critique levied against them by their "normativist" rivals (e.g., natural lawyers and interpretivists), 58 which will be addressed in the next section. Examples of meta-theoretical desiderata for theories include: Simplicity: We prefer simpler explanations to more complex ones, all else being equal (i.e., without cost to other theoretical desiderata). 57 E.g., Kramer and Steiner 2007, pp. 295-299. 58 See, e.g., Leiter 2009;Dickson 2001, p. 32. Consilience: We prefer more comprehensive explanations -explanations that make sense of more different kinds of thingsto explanations that seem too narrowly tailored to one kind of datum.
Conservatism: We prefer explanations that leave more of our other well confirmed beliefs and theories intact to those that do not, all else being equal (i.e., without cost to other theoretical desiderata). (Leiter 2009(Leiter , p. 1239 Wayne Sumner provides the following norms for assessing philosophically constructed conceptions: Extensional adequacy: a conception of a concept is extensionally adequate when it includes every item which seems preanalytically to be an instance of the concept and excludes every item which does not. Theoretical adequacy: comparing the merits of two conceptions [ . . . ]. If one of these maps (of the theoretical terrain) identifies more significant theoretical boundaries than the other, and if it seems advisable to use the concept [ . . . ] to mark these boundaries, then we will have good reason for preferring the conception which yields that map. (Sumner 1987, pp. 49-50, 96-97) As mentioned in Section 2, many legal philosophers take the ordinary understandings of concepts, or of linguistic practices, as an indispensable starting point for theorisation. However, they also sometimes employ them as a metric by which to evaluate an account's soundness, e.g., asking how intelligible the theory is to those who use such concepts, and how widely it diverges from their beliefs (Barber 2015, p. 806). Raz and Dickson additionally believe that people use the concept LAW to help them understand themselves. Hence, for them, a criterion of explanatory adequacy for legal theories is whether they pick out a concept's important features, which are required for capturing (and advancing) the way in which people understand themselves. 60 59 Dickson's list of such desiderata includes: simplicity, clarity, elegance, comprehensiveness, and coherence (Dickson 2001, pp. 32-33). 60 See Raz 1994, p. 237;Dickson 2001, pp. 48, 59. Sean Coyle posits that a philosophical account must demonstrate conformity with "received" understandings of our concepts, but nevertheless claims that: We are, of course, permitted to abandon that received understanding in favour of a modification of our concepts, but any such modification would have a profound impact upon our ordinary discourse about [e.g.,] rights. Whether such a departure is justified therefore depends upon two things: (a) whether such talk is coherent; and (b) whether the departure significantly enriches our talk of [e.g.,] rights, or enhances our existing understandings [ . . . ]. (Coyle 2002, p. 33;cf. pp. 21, 27)

. 1 . Difficulties in Application
These all seem like helpful and sound guidelines or metrics. Unfortunately, determining what counts as the successful application or utilisation of any such desideratum, or combination thereof, is unclear. How a legal theory can meet any of them is left unexplained. It is uncertain: how to apply the desiderata; how to quantify them; how to show when a theory meets them; how to prove that theory A better meets them than theory B; or what weight to affix to them in order to show which matters are of greater philosophical concern when comparing two or more theories. This is not simply a matter of being unable to show whether theory A or B better meets desideratum φ. Even if it could be shown that theory A better meets Dφ than does theory B, if B better meets Dψ than A does, which of the two desiderata is weightier for the overall comparative evaluation, Dφ or Dφ? This is not to claim the impossibility of the tasks, i.e., hold that shared metrics and paradigms for establishing how to convincingly meet these candidate criteria and desiderata could never, even in principle, obtain. It is merely to note that they have not yet even been attempted. Just like the issues surrounding counterexamples and central cases, the reasonable employment of such desiderata seems -inescapably-to be a matter of having to make (scholarly) judgements (or "intuition pumping"). Indeed, the whole process of theorisation involves a great many more judgement calls than could easily be enumerated here. As Andrew Halpin puts it: [I]n too many ways the judgement of the theorist rather than the imperative of methodology will be a determining factor, in shaping what feature of the subject matter is regarded as worthy of theoretical inquiry, or in shaping the theoretical construct that is regarded as offering greatest illumination on the subject matter as the theorist perceives it. Even at the low level methodology of metatheoretical precepts there remains room for the theorist's judgement to influence the impact those precepts will have upon the construction of theory. And in recognizing technical semantic or philosophically sophisticated analytical approaches, the pervasive influence of the theorist's judgement is still to be found: in selecting a particular type of semantics; or in discerning an essential property and elaborating its quality in the tension between its recognition and the basis for its selection. Even where the apparent strictures of methodology are the strongest, in directing the theorist to one side or another of the normative/descriptive divide, we have seen that the particular position adopted here is influenced by the choice of the theorist over how to focus on the subject matter of the theory. (Halpin 2008, p. 617)

. Concept Formation and Selection: A Wholly Normative Affair?
Let us return to the concept CONCEPT, the features concepts bear, and the means of their discernment. A staple methodological dispute amongst legal philosophers is whether they must make normative -"normative" in the sense of being morally or politically evaluative-judgements when (re-)forming or selecting concepts. Is it possible to describe or interpret the concept X without engaging in normative judgements about (i) what X is, or (ii) its value, worth, or goodness?
No legal philosopher thinks data can just be identified and described "as is". Philosophers must judge what count as such and how to interpret them. Those who come closest to the "as is" view, legal positivists -or, at least contemporary ones-claim only to defend the following proposition: the law 61 is determinable simply by looking at its social sources, without needing to assess its merits (Raz 2009b, pp. 47-48;Gardner 2012). This is false to the extent that positivists are also (generally? necessarily?) committed to the idea that one can come to know, analyse, and present an account of X without adjudging X to be morally good or bad, valuable or valueless. They additionally seem to assume that the "folk" concept or "common understanding" of X (e.g., H.L.A. Hart's explication of the educated man's understanding of what LAW is) is discernible and analysable without (first) morally evaluating it. 62 Pace positivists, "normativist" legal philosophers (i.e., interpretivists and natural lawyers) think it is impossible to know what X is without making "normative" -again, in the sense of being morally or politically evaluative-judgements about what it is. These include assessments of X's point or purpose. To their minds, there are no "anormative" concepts in legal philosophy: every concept that is germane to legal theory is inseparably entwined with, and shaped by, evaluative features and judgements about its nature. Hence, the process of determining a concept's content and contours is an inherently moral/political enterprise. 63 Normativists nevertheless disagree amongst themselves about why this must be so. Ronald Dworkin offers a normativist method called constructive interpretation: provide an account of X that fits most of the facts of a practice and that shows it in its best light vis-à-vis its point or purpose. 64 There are three stages to constructive interpretation. The first involves information collection, which Dworkin calls the "pre-interpretive" data. However, even (some of) the preinterpretive data must be interpreted, as it were. 65 Rather than 62 Legal positivists also used to hold that one must first know what X is before one can evaluate it ("for how can you know whether it is good or bad if you do not even know what it is?"). However, that may no longer be the case. Instead, some contemporary positivists seem to believe that coming to know what X is and evaluating it are two distinct, but potentially concurrent, inquiries. E.g., Dickson 2013, pp. 364-366). For the idea that positivists are offering interpretations of law, not unqualified descriptions (i.e., accounts constructed without having made an evaluation of any variety, let alone just moral and political ones) of it, see Raz 2009a, p. 60;Gardner 2012, p. 28 and 28, n. 16;Dickson 2001, chap. 2. 63 They also think it is merely trivially true that laws are posited. E.g., Finnis 2003, pp. 128-129. 64 Dworkin 1986. While Dworkin speaks in terms of "practices" and "rules and standards" of a normative domain or legal system, or tokens of a type (e.g., the text of Moby Dick being identified and distinguished from the text of other novels), his main example is really a concept (COURTESY) (Dworkin 1986, pp. 65-68). 65 "First, there must be a "preinterpretive" stage in which the rules and standards taken to provide the tentative content of the practice are identified" (Dworkin 1986, pp. 65-66). "[The interpreter] needs assumptions or convictions about what counts as part of the practice in order to define the raw data of his interpretation at the preinterpretive stage; the interpretive attitude cannot survive unless members of the same interpretive community share at least roughly the same assumptions about this" (p. 67). This is followed by the interpretive and post-interpretive stages. The interpretive stage involves settling on a general justification for the practice's main elements merely presenting a mere propaedeutic to undertake research in this fashion, Dworkin seems to suggest that constructive interpretation is what all social (including legal) philosophers are really doing, regardless of how they otherwise conceive of, or characterise, their work.
John Finnis presents an alternative normativist methodology, one grounded in the natural law tradition. Unlike Dworkin, he does not think legal theorists must present their subject matter in its best light. However, only the right sort of person, one with the appropriate sort of mindset, can properly undertake social theorising. Finnis endorses -to some extent-(legal positivist) H.L.A. Hart's claim that a legal theorist must utilise a given community's or system members' "internal point of view" in order to better understand its norms. 66 Without it, a theorist cannot understand or explain either how the people use rules to praise and blame each other, or critical features of how such norms shape such people's lives. (Without such information, in other words, a theory would be underinclusive and explanatorily inadequate). Finnis goes farther, though. He believes there is an indispensable central case of the internal point of view: that of the practically wise person, the spoudaios (Finnis 2011, pp. 14-19). Only such a character can (A) see law's point (which is to solve coordination problems and create institutional structures that allow for certain basic goods to come about that otherwise would not) (2011, pp. 85-90, 245-250, 351-352); and (B) only he or she can see how all other, less practically-oriented agents internalise (or fail to, or reject) the relevant norms and utilise them. (By contrast, other kinds of agents cannot fully grasp the practically wise person's viewpoint (2011, p. 15, n. 37)). What is often missed in discussions of Finnis' work is that, not only does the spoudaios select data and shape a theory based on that central case point of view, but he or she also shapes the very concepts employed in the account based on that privileged point of view. 67 (those identified in the first stage). This consists in "an argument why a practice of that general shape is worth pursuing, if it is" (Dworkin 1986, p. 66). This justification "must fit the standing features of the practice to count as an interpretation of it rather than the invention of something new" (p. 67). In the post-interpretive stage the agent then adjusts his or her "sense of what the practice 'really' requires so as better to serve the justification he accepts at the interpretive stage" (p. 66). He or she "will need more substantive convictions about which kinds of justification really would show the practice in its best light [ . . . ]" (p. 67).
66 See Hart 1997, pp. 56-57;Finnis 2011, pp. 12-13. 67 Chapter 1, section I of Natural Law and Natural Rights is titled "The Formation of Concepts for Descriptive Social Science". In it Finnis says: The challenge normativists raise to legal positivists is whether (legal philosophical) concepts/conceptions can either be formed or selected without making morally or politically charged evaluations. Their debate with positivists about concept selection, and about whether the law is determinable without making those sorts of value judgements -be it in terms of a constructively interpreted account, or via the purported central case of the internal point of view-has dominated analytic legal philosophy for the last forty years. Again, equally important, yet seemingly neglected by scholars, is the issue of what role such morally or politically evaluative judgements play in the very formation of theorists' (re-)conceptions (that is, if and when philosophers do offer replacement concepts/ions).
This article is not the place to offer a full-blown rebuttal to normativists about the concept (re-)formation or selection processes, or about which viewpoint to adopt qua social theorist. Still, there are a few points worth mentioning. First, while normativists go too far, both Dworkin and Finnis offer convincing reasons why some sort of interpretation is involved in both the selection and very formation of concepts for philosophical accounts. 68 Their critiques of legal positivism's problems establishing the moral-political neutrality of concept formation, selection, and understanding have, in the least, not been completely rebutted.
So when we say that descriptive theorists (whose purposes are not practical) must proceed, in their indispensable selection and formation of concepts, by adopting a practical point of view, we mean that they must assess importance or significance in similarities and differences within their subject-matter by asking what would be considered important or significant in that field by those whose concerns, decisions, and activities create or constitute the subjectmatter [ . . . ].
[T]he evaluations of the theorist himself are an indispensable and decisive component in the selection or formation of any concepts for use in description of such aspects of human affairs as law or legal order [ . . . ]. But when all due emphasis has been given to the differences of objective and method between practical philosophy and descriptive social science, the methodological problems of concept-formation as we have traced it in this chapter compel us to recognize that the point of reflective equilibrium in descriptive social science is attainable only by one in whom wide knowledge of the data, and penetrating understanding of other persons' practical viewpoints and concerns, are allied to a sound judgment about all aspects of genuine human flourishing and authentic practical reasonableness. (Finnis 2011, pp. 12, 16, 17-18;emphasis added.) However, Finnis does not appear to employ such methodology when later endorsing Hohfeld's concepts/conceptions of rights and other normative positions wholesale (2011, pp. 198-205). At least, no such justification is offered. 68 Contemporary positivists agree. See, supra note 62.
Legal positivists offer the following rejoinder. It is true that some modicum of evaluation is needed to determine what count as the data, to determine how concepts are to be clarified, etc. Still, this can be done without assessing the moral or political worth or goodness of those features. Philosophers' evaluations here are "normative" -just not in the sense of being morally or politically evaluative: their evaluations simply concern "meta-theoretical" desiderata. Thus, it seems possible to be able to interpret a concept without making moral judgements about its worth, goodness, or point. 69 As the argument in Section 5 suggests, however, baldly appealing to meta-theoretical desiderata as a counter to normativists about the kind of normativity that must play a role in concept formation and selection for the purposes of legal theory is inadequate. Positivists must do more work to make the case that judgements concerning meta-theoretical desiderata alone suffice to undertake legal theorising.
Although underdeveloped, there nevertheless is merit in the positivists' rebuttal. They must (and, I think, can) show that the selection and employment of such desiderata are neither the function of morally or politically evaluative judgments, nor predicated upon them. For even if philosophers are biased about how they shape or select their conceptions or definitions, there is no good reason to think that such biases must either be formed by, or reflect, moral or political convictions. For example, defending the idea that a duty is a weighty reason rather than an exclusionary one need not be for moral or political reasons. In doing so, you may be correct or mistaken, but you might have decided thusly simply because you believe (rightly or wrongly) that there is no such thing as exclusionary reasons.
Furthermore, there is no need to employ constructive interpretation or a central case of the internal point of view. Pace Dworkin, legal philosophers need not look at anything in its best light, let alone present it as such -especially if they deny it has one. 70 How, moreover, could that approach even work with regard to the selection and interpretation of "pre-theoretical" data unless one had already undertaken the other interpretive stages? Dworkin himself notes that the data of the pre-interpretive stage itself requires interpretation (and therefore does not really view his three stages as actually being "stages" that follow a fixed sequence). Even so, his method for doing 69 E.g., Dickson 2001, chap. 3. 70 Jules Coleman (2002, p. 316) shows the fallaciousness of Dworkin's argument that because criterial semantics is a non-option constructive interpretation is the only alternative (for there are other alternatives). so seems to preclude constructing that data in its best light if the work product's concepts are to be recognisable and salient to a given community.
I enclose "preinterpretive" in quotes because some kind of interpretation is necessary even at this stage. Social rules do not carry identifying labels. But a very great degree of consensus is needed -perhaps an interpretive community is usefully defined as requiring consensus at this stage-if the interpretive attitude is to be fruitful, and we may therefore abstract from this stage in our analysis by presupposing that the classifications it yields are treated as given in day-to-day reflection and argument. (Dworkin 1986, p. 66) The method is also unnecessary. For example, fully understanding WEIGHTINESS as a feature of A DUTY does not require that either be presented in their best lights.
Pace Finnis, there is no need to employ the (purported) central case of the internal point of view. One reason is because law may not be the exclusive, let alone the best, means for providing the relevant coordination schemes for generating the basic goods. An anarchocapitalist would deny this, at any rate. To hold otherwise (i.e., that law in its best cases does so) is therefore to beg the question, methodologically. In response, Finnis might say that even if there are alternative (and sounder) means for generating such schemes or procuring such goods, if we want a philosophical account of law, then this is how we must approach the subject. This is false because it presumes a core teleological function that law may not have. For example, understanding whether A DUTY includes the feature of weightiness requires neither that it be seen from the perspective of, nor delineated by, the practically wise person (spoudaios). Do we really need the spoudaois' perspective of why a given duty, or set of duties, may (or may not) be weighty to understand WEIGHTINESS? (Note the difference between that question and asking whether we need the central case of A DUTY to understand WEIGHTINESS).
Furthermore, in attempting to defend Finnis' view, George Duke has instead undermined it somewhat (Duke 2003, p. 189). Duke notes that it is the sophos who reasons from first principles, not the practically minded spoudaios (or phronemos). If this is correct, then how could the spoudaios be the appropriate person to do the work of concept (re-)formation and selection, rather than the sophos?

. Conclusion
This article addressed some basic methodological problems and questions affecting analytic legal philosophy. Justifying the practice and moving the discipline forward requires addressing them head-on. First, what effects do or should counterexamples have upon a rule, proposition (e.g., about what is "necessary" or merely contingently related to a concept), or theory? When do counterexamples carve exceptions, when do they delimit, and when (if ever) do they falsify a theory, proposition, etc.? Is the central case method (or methods: I identify two types of usages), which also relies on the use of examples, of any merit? If so, what are the grounds for identifying a genuine instance? Does "the" method also rely upon the use of counterexamples?
The article then shows why that set of questions applies equally to matters surrounding the (i) selection and (ii) (re-)formation ("clarification") of concepts for philosophical accounts. Are philosophers really doing what they claim to be doing when undertaking conceptual analyses? (I nevertheless argue that "normative" -in the sense of morally or politically evaluative-modes of concept formation, selection, and theory construction are unnecessary). Additionally, how can one establish: that a legal philosophical account satisfies a particular meta-theoretical desideratum, that it does so better than a rival account, or how various desiderata are to be weighed relative to each other?
While the article's main aim concerns the mere elucidation of these problems, and does not pretend to have any grand solutions, it nevertheless suggests that all of these issues -of cases and counterexamples, their employment in analysis and clarification, and the employment and weighing meta-theoretical desiderata, all rely on making scholarly judgements. The pervasive nature of these judgements warrants far greater attention than has heretofore been afforded. 71