2 Stability, Assurance, And the Concept of Legal Guidance

ADAM HILL

STABILITY, ASSURANCE, AND THE CONCEPT OF LEGALGUIDANCE

(Accepted 13 January 2014)

ABSTRACT. Legal theorists standardly hold that stability is one of eight necessaryconditions for legal guidance. We lack an adequate explanation, however, of why,exactly, stability is necessary in order that law possess the capacity to guidebehavior. Standard explanations, which rely on a claim about reasonable expec-tations, fail to connect the concepts of stability and legal guidance. In this paper,I argue that, according to the leading conception of legal guidance, stability is, infact, not necessary in order for law to possess the capacity to guide. In response,I provide an alternative conception of legal guidance, and argue that stability is anecessary attribute if we assume this conception of guidance. More generally, thepaper sets out an abstract framework in which claims that this or that property isnecessary for legal guidance can be assessed.

I. INTRODUCTION

Legal positivism is, in the first instance, a thesis about the conceptualfoundations of legal validity.1 It seeks to parse the social mechanismsthat underlie the legal decisions that, together, determine the sub-stance and govern the processes of a given legal system.2 Natural lawapproaches, by contrast, seek to parse not only the social but also the

2 See Gerald J. Postema, ‘Coordination and Convention at the Foundations of Law’, Journal of Legal Studies11 (1982): 165–203, at 167 (‘‘Thus positivists in the tradition stemming from Bentham locate the [criteria ofvalidity] in matters of social fact, thereby rejecting the view that the validity of a law is a function of its truth ormoral soundness.’’); see also Andrei Marmor, ‘The Rule of Law and its Limits’, Law and Philosophy, 23(1)(2004): 5.

1 See, e.g., Mark Greenberg, ‘How Facts Make Law’, Legal Theory 10(3) (2004): 157–198; JohnGardner, ‘Legal Positivism: 5 ½ Myths’, The American Journal of Jurisprudence 46(1) (2001): 199–227;Christopher Kutz, ‘The Judicial Community’, Philosophical Issues 11(1) (2001): 442–469; David Enoch,‘Reason-Giving and the Law’, in Leslie Green and Brian Leiter, eds., Oxford Studies in Philosophy of Law(Oxford: Oxford University Press, 2011), vol. 1: 1–15.

Law and Philosophy � Springer Science+Business Media Dordrecht 2014DOI 10.1007/s10982-014-9204-y

moral mechanisms that underlie legal determinations of validity.3 Forboth positivists and natural lawyers, proposition P accurately statesthe law governing some topic in a given jurisdiction, if and only if Pmeets the applicable criteria of legal validity, C, in that jurisdiction.4

The fundamental difference between the two approaches is thatpositivists maintain that C obtains in only virtue of social mecha-nisms, while natural lawyers remain open to the possibility that Cobtains in virtue of moral mechanisms.

For positivists, the social mechanisms at the foundations of thelaw are explicated through descriptive facts: claims about thebehaviors or mental states of judges, legislators, executives, or vot-ers.5 So, for instance, positivists may point to facts about thebehavior of courts to explain the content of C, but not to the meritsof C itself. Positivists, in other words, reject grounding their expla-nations of the criteria of legal validity in value facts: normative orevaluative claims.6 These are taken to bear only an arbitrary rela-tionship with legal norms.7 Natural lawyers, by contrast, might wellexplain the content of C through reference to value facts. Perhaps,for instance, C states the criteria of legal validity because it isdemanded by democratic values. Explanations of these sorts are notopen to positivists, for whom descriptive facts possess explanatoryprimacy. In giving descriptive facts explanatory primacy, positivistsare committed to the claim that there is no law, properly so called,absent that which is created through social mechanisms.

One of the most important of these social mechanisms – and thefocus of our attention in this paper – is the process by which legalofficials posit a law, and individuals respond. Say that a legislature

3 See Kutz, supra note 1, at 444–445. John Finnis, ‘On the Incoherence of Legal Positivism’, 75 NotreDame L. Rev. (2000): 1597–1598.

4 Kutz, supra note 1, at 443.5 Greenberg, supra note 1, at 157. For the classic statement, see H.L.A. Hart, Positivism and the

Separation of Law and Morals’, 71 Harvard Law Review 71(4) (1958): 593–629. As Ronald Dworkin put it,for positivists, legal validity is a matter ‘‘not [of the] content [of norms] but with their pedigree or themanner in which they were adopted or developed.’’ Ronald Dworkin, Taking Rights Seriously (Cam-bridge: Harvard University Press, 1978): 17.

6 Greenberg, supra note 1, at 157. There is, of course, the matter of hard and soft positivism. Softpositivists allow that value facts can possess derivative significance in determining the content of thelaw. Hard positivists argue that value facts can never do even that. The debate over whether value factscan affect the content of laws, not whether they can affect the criteria of legal validity.

7 Id. at 157–159. Positivism’s insistence on the primacy of descriptive facts gives rise to a thornyquestion: How can facts about what we happen to have been doing in the past give rise to genuinereasons about what we should be doing in the future? Or, put differently, how can it be that what wehappen to do around here can provide reasons to keep doing it? See, e.g., Kutz, supra note 1, at 446.

ADAM HILL

duly enacts proposition P. What follows? What is the characteristi-cally legal way in which individuals are influenced by P? The way inwhich one answers that question determines her conception of legalguidance.

A conception of legal guidance explains what attributes law mustpossess, such that it has the capacity to guide or direct individualconduct. Although positivism is, in the first instance, a thesis aboutlegal validity, it is not only that. It is also a descriptive theory thatseeks to explain how law, ‘as something that must be posited throughsome social act or activity, either by enactment, decision, or prac-tice’,8 possesses the capacity to guide or direct individual conduct.

This question is particularly pressing for legal positivists. If anatural lawyer claims that value facts possess explanatory primacy inan account of legal validity, such that, at least in some cases, when Prequires one to /, one thereby has a genuine or unqualified reason to/, then the task of explaining the connection between law andpractical reasoning is, arguably, less urgent. For natural lawyers, theconnection between the content of P and individual beliefs and ac-tion is more straightforward. In short, if P is underwritten by somemoral logic, it prima facie possesses the capacity to guide conduct.9

Positivists, however, have nothing of the sort upon which to rely.They must, instead, explain how it is that laws, which are not nec-essarily underwritten by any moral logic, possess the capacity toguide individual behavior.

The centrality of legal guidance has not eluded positivists.Scholars working in the field have developed a comprehensiveconception of legal guidance and set out the social and legal con-ditions under which it obtains. The conventional wisdom concep-tualizes legal guidance as that state of affairs in which an individual‘is able to learn of his obligations or rights from [the relevant legalnorm] without engaging in deliberation’.10 This conception of legalguidance treats guidance as a purely epistemic phenomenon. Lawguides when it possesses the capacity to convey information about the

8 Scott Shapiro, ‘Law, Morality, and the Guidance of Legal Conduct’, Legal Theory 6(2) (2000): 127(emphasis in original).

9 There are a number of ways in which we might draw a link between natural law and the capacityto guide behavior. For instance, one might maintain that the capacity to guide conduct is a necessaryfeature of moral norms. If all moral norms possess the capacity to guide, then any legal norm under-written by a moral norm will possess the capacity by extension.

10 See Shapiro, supra note 8, at 153.

STABILITY, ASSURANCE, AND THE CONCEPT OF LEGAL GUIDANCE

content of the law to individuals. In order to possess this capacity,positivists hold, ‘there are certain conditions that the law has tomeet’.11 A wide consensus holds that eight attributes ‘state necessaryconditions’12 for the existence of legal guidance and thus legality.These attributes are, by now, familiar. Laws must be general; theymust be promulgated; laws must not succumb to retroactivity norcontradiction; they must be clear, and possible to follow; there mustbe congruence between the rules as written and as applied; and,finally, laws must be stable. That legal norms must possess each ofthese attributes in order to guide – in the sense of conveying toindividuals their legal rights and obligations – is conventional wis-dom among legal positivists.13

This paper challenges the conventional wisdom about epistemicguidance and the general picture painted by positivists explaininghow individuals are informed about and motivated by legal norms. Itdoes so through the lens of an alleged attribute that has provendifficult to explain14 yet nonetheless is held essential for epistemicguidance: the stability requirement, which holds that laws shouldpossess a certain ‘constancy through time’ and ‘not be changed toofrequently’.15 This paper develops a means for testing whether anygiven attribute of guidance is, in fact, necessary for law to possess thecapacity to guide, and then applies that test to the case of stability.The paper argues that, under the standard conception of epistemicguidance, stability is not, in fact, required to guide conduct.16 Yet,legal theorists have long linked stability with guidance. In an effort torationalize this long tradition of linking stability with guidance, this

11 Marmor, supra note 2, at 5.12 Scott Shapiro, Legality (Cambridge: Harvard University Press): 395. Earlier, Shapiro observes that

‘‘regimes that flout these principles are simply not engaged in the basic activity of law.’’ Id. at 394.13 See generally H.L.A. Hart, The Concept of Law (2) (Oxford: Oxford University Press, 1997); Joseph

Raz, ‘The Rule of Law and Its Virtue’, in The Authority of Law (1979); Jeremy Waldron, ‘The Conceptand the Rule of Law’, Harvard Journal of Law & Public Policy 30 (2006): 15–30; John Gardner, ‘Hart onLegality, Justice, and Morality’, Jurisprudence 1(2) (2010): 253–265.

14 See, e.g., Marmor, supra note 2.15 Lon Fuller, The Morality of Law: Revised Edition (New Haven: Yale University Press, 1969): 79.16 Thus, while the paper is avowedly abstract in its primary concerns, it possesses implications for

public policy. The conventional wisdom argues that legal norms must meet a certain set of parametersif they are to possess the capacity to guide conduct. This is a straightforwardly descriptive claim. Theseparameters place constraints on the form legal norms can take. If they are true, they thus limit thesubstantive policies the state can pursue, assuming the state wants to use law to guide behavior. If wethink of law as a technology or tool for governance, the eight attributes of guidance essentially state thelimitations of the tool. Because the criteria of legality place substantive constraints on policy, theypotentially exclude democratically selected policies.

ADAM HILL

paper presents an alternative account of legal guidance. Rather thanconceptualize guidance as purely epistemic, the paper presents amotivational account of guidance. Whereas an account of epistemicguidance explains merely how law conveys its requirements, anaccount of motivational guidance explains how law could be thereason why individuals comply. Such an account, the paper argues, isvaluable not only because it best explains legal theorists’ long-standing tendency to link stability with guidance, but also because itoffers an account of the way in which the state and its citizenscooperate to overcome the harm to reliance interests possiblybrought about by unstable law. The account of motivational guid-ance presented here models the state and its citizens as playing anassurance game, and argues that legal change, even change that isresponsive to widely-held problems, will tend to only occur whenboth sets of parties are sufficiently assured that the other willcooperate. We begin our argument, however, with a much morestandard idea: the way in which law, as a descriptive matter, is ableto shape individual deliberation.

II. EPISTEMIC LEGAL GUIDANCE: MEDIATING BETWEEN LAW ANDDELIBERATION

When the U.S. federal government enacts new criminal laws, orseeks to change health care policy, or changes the rules governingsecurities laws, it seeks to regulate the behavior of hundreds ofmillions of individuals.17 Regulation at this scale poses a problem:how is the state to inform this number of individuals of their newduties, rights, or obligations? Even if each individual governed by anew regulation could be located, it is clearly beyond the capacity ofthe state to send officials door-to-door providing updates on newregulations.18 As a substitute, the state crafts legal norms that pro-vide citizens with epistemic guidance. Laws provide epistemic guid-ance when a citizen ‘is able to learn of his obligations or rights from

17 See Shapiro, supra note 12, at 72 (describing ‘‘the regulation of mass populations’’). See also JeremyWaldron, ‘The Rule of Law in Contemporary Liberal Theory’, 2(1) Ratio Juris (1989): 79–96. The basicdemographic contours of the U.S. population are available at United States Census Bureau, www.census.gov. It should be noted that U.S. law, in addition to governing those within the territorial UnitedStates, also applies to certain individuals outside its borders.

18 Although, on certain occasions, the state does organize events, like community workshops, tohelp inform individuals of their new duties, rights, and obligations.


[the relevant legal norm] without engaging in deliberation’.19 Morecolloquially, laws provide epistemic guidance to individuals when anindividual is able to read the law and thereby learn how to act incompliance with the law. Such impersonal guidance is a substitute forthe personal guidance rendered infeasible by the size of the modernstate and its large-scale regulatory schemes.20

Epistemic guidance offers a solution to the problem of regulatinglarge-scale populations. But not all laws provide epistemic guidance.A wide consensus holds among legal positivists that epistemicguidance is possible only if the legal norms that a state promulgatesmeet certain parameters. In order to guide epistemically, positivistshold, ‘there are certain conditions that the law has to meet’.21 Let uscall these conditions – which positivists argue ‘state necessary con-ditions’22 for the existence of epistemic guidance – the attributes ofguidance.

Conventional wisdom among positivist legal philosophers holdsthat, in order to guide individual conduct, ‘there are certain condi-tions that the law has to meet’.23 A wide consensus holds that thereare eight attributes that ‘state necessary conditions for the existenceof [law]’.24 We need only briefly review the substance of theseconditions. Laws must be (1) general, setting forth rules of conduct,applicable to some defined segment of the population, prohibiting orfacilitating certain modes of behavior. These rules must be pub-lished, or (2) promulgated, so that they are available to those whoseconduct they govern on a (3) prospective, rather retrospective, basis.Laws must avoid obscurity or unintelligibility; they must be (4) clear.A legal code must be (5) non-contradictory, in that its provisions donot conflict. Closely related is the demand that law be (6) possible tofollow. Official enforcement of these laws should be (7) congruentwith the laws as written. And, finally, laws should be (8) stable, which

19 See Shapiro, supra note 8, at 153.20 On occasion, law is used, of course, to regulate single individuals or small groups. Legislatures

pass bills aimed at single individuals. See, e.g., Jeffrey S. Hill and Kenneth C. Williams, ‘The Decline ofPrivate Bills: Resource Allocation, Credit Claiming, and the Decision to Delegate’, 37(4) American Journalof Political Science (1993): 1008–1031. Judicial decisions affect only the parties to the case, in the firstinstance. And administrative agencies adjudicate claims of single persons.

21 Marmor, supra note 2, at 5.22 Scott Shapiro, supra note 12, at 395. Earlier, Shapiro observes that ‘‘regimes that flout these

principles are simply not engaged in the basic activity of law.’’ Id. at 394.23 Marmor, supra note 2, at 7.24 Shapiro, supra note 12, at 395.

ADAM HILL

requires that norms not change too frequently. As a knife must besharp if it is to cut well, so too must laws meet these criteria if theyare to provide epistemic guidance.25

We, of course, are concerned here with the attribute of stability inparticular. Ordinarily one might pause here and explicate the idea ofstability in some detail. The difficulty is that positivists haveremained rather coy about just what stability amounts to. AndreiMarmor captures contemporary sentiment about stability: ‘Thisrequirement of the rule of law is basically a rough standard’, arguesMarmor, because ‘it would be absurd to assume that we can have aprecise notion of the ideal pace of change’.26 Our inability to for-malize the stability constraint seems to stem from the fact that,unlike other attributes of guidance, it is difficult, if not outrightimpossible, to define stability in the abstract. To characterize a law asstable seems, in some fundamental way, to involve an assessment ofthe conditions in a society, and the expectations of the governed, in away that the other items on the list do not. In the second half of thispaper I will offer a conception of motivational guidance that cashesout stability in just these terms – but prior to reaching that point, wewill need to investigate the conventional approach to stability.

The conventional wisdom holds that, if a legal norm meets thecriteria stated by the attributes of guidance, then it provides episte-mic guidance and, thus, is capable of making a difference to individ-uals’ beliefs and conduct – ‘a difference, that is, in the structure orcontent of deliberation and action’.27 Epistemic legal guidance, then,and the attributes of guidance that make it possible, provide a crucialexplanatory link for positivists. It explains how laws, which begin asstrings of characters posited by officials, govern society.

25 Raz helpfully compares law to a knife, in the sense that both can possess certain properties thatrender them, as tools deployed by individuals, more or less useful. Our question can usefully be framed:If sharp knives cut well, what sort of laws govern conduct well? See Raz, supra note 13, at 225.

26 Marmor, supra note 2, at 34. It is instructive to observe that Marmor suggests that, in discussingstability, we are searching for an ‘‘ideal’’ pace of change. This interpretation of the attributes ofguidance seems overly moralized. The items on the list explain how laws should be crafted if they are toguide individuals; they do not instruct officials how to make ideal policy.

27 Jules Coleman, ‘Incorporationism, Conventionality, and the Practical Difference Thesis’, 4(4) LegalTheory (1998): 383. We can test whether or not an individual was guided by a rule by asking about thecounterfactual case: A legal rule, P, guides a person to do some act, A, only if that person ‘‘might nothave done’’ A ‘‘had he not appealed to’’ P qua legal rule. See Shapiro, supra note 8, at 132.


Crucial to this explanatory chain is the idea that individuals, oncethey learn of their rights and obligations from law, take the contentof law as reasons for action.28 Law transmits its directives by pro-viding individuals with things called reasons.29 Individuals, in turn,are equipped to process and make sense of these reasons. Althoughphilosophers dispute the exact set of processes that individuals use toprocess reasons – is reasoning machine-like or driven in part byemotions? – it is generally agreed that the processing of reasonsinvolves some, most likely imperfect, process of weighing. Discreteconsiderations can be measured against each other, and added andsubtracted, in order to help individuals arrive at this or that belief.Hence we often speak of a balance of reasons. Legal norms provideepistemic guidance when they possess the capacity to add a reason toone side or the other of the balance.

At this point one might well wonder, when we speak of ‘legalnorms’, if we are referring to all legal norms, or only a subclass ofthem. One could, following Joseph Raz, argue that only the legalnorms governing the processes that produce legal rules must be sta-ble, not the legal rules themselves.30 Guidance, Raz argues, requiresmerely that ‘the making of particular laws should be guided by openand relatively stable general rules’.31 What distinguishes a generalrule? Raz specifies that ‘two kinds of general rules create theframework for the enactment of particular laws’, namely, ‘thosewhich confer the necessary powers for making valid orders’ and‘those which impose duties instructing the power-holders how toexercise their powers’.32 Under Raz’s framework, general rulesgovern the creation of the primary rules that regulate individualconduct. And it is general rules that must be stable in order that lawguide conduct.

Notice, however, that it is also the case that any given processrule can be re-described as a discrete rule that applies to some legalofficial.33 All process rules are discrete rules, in this sense. Thus, any

28 Enoch, supra note 1; Gerald J. Postema, ‘Implicit law’, Law and Philosophy, 13(3) (1994): 361–387;Stephen Perry, ‘Hart’s Methodological Positivism’, in Hart’s Postscript: Essays on the Postscript to theConcept of Law (Oxford: Oxford University Press, 2001).

29 Pamela Hieronymi, ‘Reasons for Action’, Proceedings of the Aristotelian Society 111 (2011): 407–427.30 Raz, supra note 13, at 214–216.31 Id. at 215.32 Id.33 Ultimately, some official enforces the rules that govern process.

ADAM HILL

argument we can run to show that stability is not a necessary con-dition for epistemic guidance works for both discrete norms andprocess rules. Insofar as stability is concerned, Raz’s distinctiondoesn’t make a difference.34

Moreover, instability of this sort – instability so pervasive that itaffects the conventions sustaining the creation of laws – is the type ofinstability that undermines an entire legal system. Legal systems tendnot to survive such dramatic shake-ups. And legal systems, as such,are standardly defined as stable entities.35 If Raz’s distinction doesmake a difference, it proves too much. Thus, absent a compellingreason to distinguish types of legal norms, we shall use the term torefer indiscriminately to both general and primary rules.

It is important, however, to distinguish between legal and non-legal norms. Unsurprisingly, legal norms do not monopolize indi-viduals’ deliberations. Non-legal reasons compete with legal reasonsfor individuals’ allegiance. In contemporary states, individualsreceive guidance not only from legal norms, but also from cus-tomary norms, ethical norms, religious norms, and social norms.Each of these types of norms might also purport to regulate indi-vidual conduct.

The state assists individuals in locating the legal norm by adorningit with ‘an official marking that designates certain standards as thoseto which one must conform’.36 Common marks include ‘inscriptionin some authoritative text’ or ‘declaration by some official’.37 Officialmarks allow citizens to identify that a norm is, in fact, a legal normwithout having to engage in deep deliberation and speculation as tothe pedigree of the norm. For positivists, a legal norm is valid just invirtue of certain social facts, usually certain political procedures, suchas passage of bills, and the like. Adorning legal norms with officialmarks is designed to distinguish legal norms from other sorts of

34 Raz’s reason for drawing this distinction is well taken. Raz is concerned that if we do not draw thedistinction, most of administrative law will fail to be counted as law, since it is notoriously volatile. Razsingles out administrative regulations as the site at which stability of general legal processes are ofparticular import for the stability requirement. We will show later in the argument that one can sustainadministrative law as law without resorting to the discrete rules versus process rules distinction.

35 Shapiro, supra note 12, at 65.36 Jules L. Coleman, The Practice of Principle: In Defense of a Pragmatist Approach to Legal Theory

(Oxford: Oxford University Press, 2003): 137.37 Scott Shapiro, ‘What is the Rule of Recognition (and Does it Exist?)’, Yale Law School Public Law

and Legal Theory Research Paper Series (unpublished manuscript): 2, available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1304645.


norms, and to ensure that individuals need not engage in the deephistorical and conceptual tasks that would be involved in an inves-tigation of whether a given norm is, in fact, a legal norm.38 Officialmarks attempt to solve the informational problem that arises whencompeting norms purporting to regulate proliferate. An official legalmark conveys to the individual the idea, as far as the law is con-cerned, this norm is dispositive.39

Epistemic guidance, therefore, involves presenting individuals withthe legal norm alongside whatever non-legal norms to which theyadhere. In other words, epistemic guidance models a thin integrationbetween the reasons created by legal norms and an individual’sdeliberative processes. Individuals pursue their antecedently chosenends, and, in the pursuit of those ends, encounter legal norms.Background beliefs and desires determine whether, and to whatextent, individuals comply with legal norms. Legal norms providesomething of a map of possible routes for individuals, assuming thatthe individual knows her destination. Which route any given agenttakes is a function of antecedently given desires. And the individualmay choose to simply not consult the legal map at all, depending onthese antecedently given desires. Epistemic guidance does notrequire that individuals choose their ends because of legal norms –only that individuals can make themselves aware of what the lawrequires.40 This would seem to be the case even on a Razian accountpremised on peremptory reasons.41 For even if it is the case that law

38 See Hart, supra note 13, at 95.39 Nonetheless, an individual might wonder why comply with the legal norm rather than some

other norm. Official markers do not provide an all-things-considered reason to comply with the legalstandards, but, in designating certain standards as those with which law requires conformity, officialmarkers convey that legal reasons do not take themselves as subject to the balance of total reasons.Official markers demonstrate that legal reasons, simply in virtue of their status, not only are not subjectto the balance of all reasons, but take themselves to have changed the balance of reasons in a Razianmanner or do not take themselves to have changed the balance of reasons but do simply disregard theother reasons.

40 Here is another way of putting the point. Assume an individual with a set of beliefs, B, about whatthe law requires. Artificially separate the individual’s desires into some set of desires toward compliancewith the law, D, and all of her other desires, O. Individuals’ deliberations concerning the pursuit of Oare constrained by the interaction of B and D. So, e.g., if the individual wants to get rich, but also wantsto comply with the law, then things like robbing a bank will be ruled out. Now, in the ordinary courseof things robbing a bank will also be ruled out by moral or social desires. In such a case, that theindividual doesn’t rob a bank is overdetermined. But let’s assume that, for this individual, moral orsocial pressure doesn’t rule out robbing a bank; the desire to comply with law is the only constraint. SayB changes, to B-prime. Now, O is constrained by B-prime. O does not change; only the content of theconstraint is changed.

41 See Joseph Raz, ‘Authority, Law and Morality’, 68(3) The Monist (1985): 295–324. I am very gratefulto an anonymous reviewer for raising this objection.

ADAM HILL

presents itself as a practical authority, and is ‘meant to replace thereasons on which it depends’, it still may be the case that the citizendetermines that the reputed practical authority is no authority atall.42 By the rules of the legal game, legal norms purport to be practicalauthorities; but epistemic guidance does not require that citizensthemselves conclude, as an all-in judgment, that legal rules arealways authoritative.43 If it did, it would imply that epistemic guid-ance is only satisfied when all citizens comply all the time. All thatepistemic guidance requires is that legal norms be written in such away as to convey their normative demands to citizens.

The conventional wisdom holds that epistemic guidance is nec-essary in order that individuals be made aware about what the lawrequires of them. In a world in which personal guidance is infeasible,individuals learn of their rights and obligations impersonally,through epistemic guidance. Because individuals’ beliefs and actionsare determined by weighing reasons, legal norms affect beliefs andaction by creating new reasons for individuals. Thus, on thisdescriptive model, legal norms operate via ‘self-directed action’.44 Inother words, individuals must apply general legal directives to theirown particular situations, thus overcoming the problem of massregulation. Law thus guides by seeking to affect individuals withreasons.

III. DOES EPISTEMIC GUIDANCE REQUIRE THAT LAW BE STABLE?

In order to guide by reason – in order to provide the epistemicguidance crucial for governing mass populations – must legal norms,in fact, meet the criteria set forth by the attributes of guidance?Given this picture of epistemic guidance, which mediates betweenlaws on the books and individuals’ deliberations and actions, is it, infact, true that the attributes of guidance state necessary conditionsfor epistemic guidance? More formally, we may ask of any of theeight attributes:

42 Id. at 297.43 As Raz writes, ‘‘No blind obedience to authority is here implied. Acceptance of authority has to be

justified …’’ Id. at 299. Epistemic guidance, for Raz, requires that law ‘‘must be capable of guiding thebehavior of its subjects. It must be such that they can find out what it is and act on it.’’ Raz, supra note13, at 214.

44 Postema, supra note 28, at 369.


What state of the world does epistemic guidance require such that some attribute,X, is necessary, assuming the operation of some other set of attributes, Y, where Xj2 Y?

We answer this question by posing a counterfactual. An attribute isnecessary for a conception of legal guidance when the absence ofthat attribute renders law incapable in principle of guiding individ-uals. The idea here is that there are certain states of the world thatmust obtain in order that legal guidance be possible, and the itemsconstituting the attributes of guidance either uniquely cause thosestates or are necessary for those states to exist. Posing a counter-factual in which legal norms do not possess the attribute in questionallows us to test whether guidance is possible in its absence.Applying this framework to the case of stability, we can generatetwo cases: one in which a legal norm is stable and a counterfactualcase in which a legal norm is unstable. We then compare how thesetwo cases affect epistemic guidance.

First, imagine a successful case of epistemic guidance, in whichthe applicable law, P, is stable. Say that Alice wants to invest somemoney in a stock that will grow in value over time. Alice is indif-ferent between stocks, so long as the one she picks will grow invalue. One investment option is a company’s stock that Alice readabout in the newspaper. Another option is a company’s stock thather friend, who possesses insider information, told her would growin value because it is going to be acquired by another companyshortly. Prior to consulting the applicable law, Alice is indifferentbetween the two options, since all she wants to do is invest in amoney-making stock. However, because Alice is motivated tocomply with law, she seeks epistemic guidance. Alice first seeksindirect guidance – from her associates, secondary sources, and soforth – and receives a uniform set of reports as to the content of thelaw. They tell her that trading stocks on insider information is illegal,and could land her in jail. Just to be sure, Alice reads the applicablelaw and applicable judicial opinions, and finds that her indirectguidance is supported by the direct guidance. One alternative,investing in the stock she read about in the newspaper, is within thebounds of compliance, and the other alternative, trading on insiderinformation, is prohibited. Alice chooses the legal alternative, andavoids jail.

ADAM HILL

Our successful case of epistemic guidance provides a baselineagainst which we can pose a counterfactual. Can we alter the casesuch that instability interferes with the transmission of legal content,thereby impairing epistemic guidance? It is difficult to imagine such acase. Even if the contents of P changed at random intervals, Alicecould, in principle, locate the content of the law and thus receiveepistemic guidance. This conclusion follows from the definition ofepistemic guidance. No matter how volatile a discrete legal rule, if itis duly enacted, then it bears law’s official mark. And the fact that anorm bears law’s mark implies that the informational problem aboutwhich norm is authoritative has been solved, and epistemic guidanceis, in principle, possible. Duly-enacted laws are published, of course,thanks to the promulgation requirement. Laws altered in any ran-dom interval are still subject to promulgation, and the fact of pro-mulgation ensures that individuals can in principle locate the contentof the law, which is all that epistemic guidance requires. Epistemicguidance simply does not require any state of the world for whichstability is required. Contra the conventional wisdom, then, it seemsthat no level of instability interferes with the capacity of legal normsto provide epistemic guidance.

If our account thus far is correct, the conventional wisdom, whichholds that stability is a necessary attribute of epistemic guidance, isincorrect. Where did those accounts go wrong? We need an errortheory. Here, we will explore why it is commonly believed thatepistemic guidance requires stability. The culprit is that instabilityraises the costs of epistemic guidance. Instability makes it moredifficult to locate the law, thus making it more difficult to complywith the law. On this line of thought, instability sets off a chainreaction that ends with decreased compliance.

This line of thought dovetails with the most common way tojustify stability as an attribute of guidance. As earlier noted, the mostcommon way to do so is on grounds that, if law is volatile, indi-viduals will not rely on it. This claim, or some version of it, mightwell be true. The problem with this argument is that epistemicguidance does not require increasing individuals’ reliance on law orotherwise providing motivational guidance. It does not require legalofficials to cultivate reliance or that individuals be disposed to rely. Ifthe attributes of guidance are conceptualized as attributes that tend


to increase guidance or compliance, then it would make sense tojustify stability on reliance grounds. The difficulty here is that if weconceptualize the attributes of guidance in this way, then our list ofeight attributes is, and always was, woefully incomplete. If thestandard is attributes that tend to enhance compliance, then surelyPareto optimality, or something along these lines, is missing fromthe list.

Nonetheless, one might still be worried about the possibility thathigh volatility makes epistemic guidance more difficult, thus ren-dering compliance less likely. This worry is not altogether mis-guided. Even if stability is not required for epistemic guidance, itmight possess a contingent, yet predictable, relationship to compli-ance and thus law’s efficacy.45 Sketching out such a contingentrelationship does help to explain the error, and explain why wemight care about stability even if it is not a property of legality. Thekey to sketching out this contingent relationship between instabilityand reduced compliance is to treat legal norms as costly information.

IV. LEGAL NORMS AS COSTLY INFORMATION

Ideally, locating legal norms would be frictionless. The environmentin which legal norms operate, however, ensures that locating law’scontent is rather complex. In contemporary states, individualsreceive guidance not only from legal norms, but also from cus-tomary norms, ethical norms, religious norms, and social norms.Each of these types of norms purports to regulate individual con-duct. Laws can only claim primacy over other types of norms if theymake themselves known. Earlier, we observed that official marksattempt to solve the informational problem that arises when com-peting norms purporting to regulate proliferate. Here, we show thatofficial marks are signals from the state to citizens. As signals, theinformation they provide is not always perfect. It is sometimes noisy,especially when laws are unstable. This situation raises the costs of

45 Ultimately, the level of compliance sought by legal officials is a function of the regulatory regime.Thus, higher or lower search costs might be tolerated. We cannot answer that in the abstract, so wecannot specify how much non-guidance is compatible with a regulatory program. We can only observethat there will be, in the normal case, some desired level of guidance and compliance.

ADAM HILL

obtaining legal information, and thus can lead to reduced compli-ance.

One way in which individuals might determine if a norm is, infact, a legal norm is to engage in a historical inquiry into the pedigreeof the norm. This historical inquiry would involve questions like, DidCongress pass this norm?, Does the Constitution forbid norms of this sort?,Has any court modified the norm?, and so forth. Such investigations,however, are likely to result in confusion. Instead, we need a proxyor a signal to stand in for all of the historical investigation that goesinto determining if a norm is, in fact, a legal norm. The best proxy orsignal of legality is that the norm ‘bear the mark’ of legal authority.When a legal norm bears the mark of legal authority, it announces toindividuals that it possesses the proper pedigree.46 In theory, officialmarks guide individuals by signaling those norms, and only thosenorms, that are sanctioned by the state. As a signal, however, it isliable to convey incomplete or erroneous information.47 In suchcases, epistemic guidance suffers.

Signals can be erroneous because inscriptions in authoritativetexts do not stop signaling if a law is superseded. A superseded statutedoes not report that it has been superseded. A superseded statutedoes not stop bearing the mark of law once it has been superseded.Ideally, we would devise a technique to signal the content of the lawthat could update itself. Official marks as they currently exist,however, are not self-updating.48

One might be tempted to argue that a superseded statute bears aninvalid mark. But marks are not valid or invalid. Rather, they areaccurate or inaccurate, but there is no way of discerning whetherthey are accurate or not from the mark itself. Thus, every change inthe content of a law results in the creation of some erroneous

46 One might be tempted to avoid taking the term ‘‘mark’’ literally, since, at base, to be lawfulsimply means to have the correct pedigree, to be a duly enacted law. But the official mark is notequivalent to (or even a logical extension of) the fact that a norm is duly enacted. The mark is a proxydesigned to solve informational problems and decrease deliberation. Only duly enacted norms receiveofficial marks, but we can imagine a case in which a duly enacted law does not, for whatever logisticalshortcoming, receive a mark but is nonetheless law. A law that is not published would not receive amark, for instance. If the official mark is to solve an informational problem about which normsindividuals are supposed to comply, then we must take the idea of a mark literally.

47 The legal philosophy literature has not addressed these shortcomings with official marks. It hastreated the idea of official marks quite generously.

48 It is worth observing that the technology is readily available to make official marks self-updating,at least with regard to those official marks published electronically. A simple piece of code embedded inthe metadata of the standing law might well be able to provide self-updating signals.


signals, unless inaccurate signals are immediately destroyed. Fre-quently changing legal norms results in a proliferation of inaccuratesignals. If, for instance, an official report of the law is printed onceevery year, and a given legal norm is changed every year for10 years, then there will exist nine superseded reports, unless theyare destroyed. These reports are inaccurate signals of the content ofthe law.49 Instability thus recreates the informational problem thatofficial marks were intended to solve. Superseded marks are, in es-sence, competitor norms. They state norms that conflict with thestate’s position on what counts as compliance with the law.

In addition to the possibility that an official mark provides erro-neous information, official marks might also, or alternatively, pro-vide incomplete information. Although a legal norm contains much ofthe information that is necessary to comply with law, it does notalways contain all of the information required for compliance. Thatinformation is located in interpretations of the norm, in judicialopinions, and the like. In other words, an official mark is partialwhen it points to a string of text, and refers to that text as law, when,in fact, the text in question only conveys part of the information anindividual needs to know in order to comply with the law. In theUnited States, many official marks are partial, because some legalcontent is contained in judicial opinions. For instance, much of thecontent of the Eighth Amendment’s prohibition on ‘cruel and unu-sual punishment’, is contained not in the text of the law but, rather,in the text of judicial opinions.

In theory, official marks, or signals, point to a string of text, andtell the citizen: This string of text contains all of the information youneed in order to comply with the law. In practice, however, theseofficial marks are partial. They convey only some of the informationneeded to comply with the law.50

If instability increases the erroneousness or incompleteness ofofficial signals, it increases what we can describe, following JulesColeman, as search costs.51 If these search costs are high enough, theefficacy of a particular legal norm is reduced. Efficacy is a necessarycondition of legal systems, although not itself an attribute of legalnorms. This line of thought does not show that stability is an

49 Inaccurate, that is, assuming that the content of the change is not a reversion to some past norm.50 Cf. Appalachian Power Co. v. E.P.A., 208 F.3d 1015, 1020 (D.C. Cir. 2000).51 Jules Coleman, Risks and Wrongs (Oxford: Oxford University Press, 2002): 125–131.

ADAM HILL

attribute of legality, but it may help us to identify the point at whichthe costs of finding the law become such that compliance is com-promised to the point that legal norms are rendered inefficacious.

Volatility can increase the costs of locating the law, thus rampingup the costs of compliance. Volatility can artificially reduce com-pliance by making it costlier than standing law would imply that it is.Assume some ‘natural’ rate of compliance with a given law. The rateof compliance is natural in the sense that it is a function of indi-viduals’ background beliefs and desires. Volatility of the sort we havedescribed here reduces compliance below this natural rate, eventhough the content of the standing norm is the same.

It’s possible that instability creates costs sufficiently high that theydeter individuals from locating valid law. Locating valid law in theface of instability can take time and effort, and can even result inindividuals mistakenly acting on what they believe to be legal rea-sons. An example will illustrate how these costs could mount undernon-ideal conditions.

Imagine a variation on Alice’s successful case of epistemic guid-ance. Assume that Ben, acting in good faith, seeks epistemic guid-ance. The content of the applicable legal norm is highly volatile, butBen is unaware of this fact. Ben seeks indirect guidance by consultingsecondary sources, which provide mixed reports. Some treatisesreport that the law permits only option P. Reports from those in arelated industry indicate that P and Q are permitted. Sensing thatsomething is amiss (Ben is not a legal official, but is aware that thelaw is supposed to be clear and predictable), Ben seeks direct guid-ance from what he believes to be an authoritative legal text. Butbecause the law is volatile, Ben encounters a superseded statute,which, of course, does not report that it has been superseded. Thestatute bears the mark of law, and indicates that both P and Q arepermissible. Because Q is the preferable option all things equal, Benadds a consideration to his stack of reasons and acts accordingly. Infact, Ben erroneously relied on a superseded statute. The validstatute prohibits Q. So Ben fails to comply.

The costs individuals are willing to incur in their search for epi-stemic guidance presumably have limits. To be sure, the absence ofepistemic guidance does not imply a lack of compliance in everycase. It might well be the case that most individuals can comply


without guidance. But over the long haul compliance suffers whenguidance decreases. And when volatility increases, compliance de-creases because individuals prefer to save the costs to time and effort.If, at a certain point, the costs become high enough and the contentof the legal norm is non-intuitive enough that individuals cannotfigure out the compliant behavior absent the norm, compliancecould decrease to such a degree that the system can no longer bedescribed as efficacious.

A second way in which search costs can impair epistemic guid-ance is when individuals erroneously rely on a superseded officialmark. Here, individuals desire to comply, and believe that they arecomplying, but in fact fail to comply because instability decreases thestrength of the signal sent by officials to individuals. Unlike caseswhere individuals decide to stop the search due to the costs ofsearching and bite the bullet on compliance, here individuals believethat they have been guided and are complying.

In our hypothetical, for instance, Ben took himself to be relying ona reason created by law. In fact, he erroneously relied on a super-seded official mark. In such cases, compliance suffers not because ofa decision that is too costly, but, rather, because volatility leads to asituation in which individuals’ operative reasons – the reasons onwhich they, in fact, act – are different from the reasons that do, infact, justify the action from the perspective of law. The mistakennon-compliance induced by volatility raises the question of whetherBen was epistemically guided. We can see that there was no reasonfor Ben to act in the way that he did, yet it seems as though Benacted for a reason.52

V. THREE TYPES OF REASONS

Was Ben epistemically guided when the reason on which he actedwas, in fact, not a reason? By way of analogy, consider a game ofchess between Beginner and Expert.53 If, say, Beginner is consideringwhether to move a bishop vertically, he might consult the rules ofchess and find that bishops are only allowed to move diagonally. Therules of chess tell Beginner that moving the bishop diagonally (or at

52 See, e.g., Jonathan Dancy, Practical Reality (Oxford: Oxford University Press, 2002).53 Here the account follows Pamela Hieronymi, supra note 29.

ADAM HILL

least omitting to move it non-diagonally) is an appropriate action.The rules justify the action.

It turns out that although moving the bishop diagonally was avalid move, it was not a smart one, and Expert has placed Beginnerin a very precarious situation. Beginner sees that his one way out ofExpert’s trap is to undertake what he believes to be a ‘castle’.Beginner attempts to think through the rules of castling, wonderingif he may do so despite the fact that he had already moved his king.Beginner mistakenly concludes that the rules of chess do, in fact,allow him to castle despite already having moved his king. Here,Beginner acts for what he took to be a reason, but, which, under therules of chess, is not actually a reason at all. We can call such reasonsoperative reasons.

Beginner erroneously believed he could castle after moving theking, due to his lack of familiarity with the rules of chess and ageneral sense of anxiety. These reasons explain Beginner’s choice. Byintroducing a distinction between considerations grounded in anormative system (justificatory reasons) and considerations that anagent took to be grounded in a normative system (operative rea-sons), we can see why Ben was not epistemically guided by thesuperseded statute.

Volatility may induce individuals to mistakenly believe that theyare being epistemically guided by the legal system when in fact theyare not. Ben, like Beginner, sought to comply with the rules. Thenon-compliance in both cases resulted from the complexity of therules relative to the individuals’ capacities to comply. And while wecan now see how to avoid the awkward implication that Ben was infact guided by an invalid legal rule, the practical worry that volatilitydecreases guidance remains.

We have our error theory, then. We have identified why onemight be worried by unstable legal norms, even if stability is not anattribute of legality. The claim that stability is necessary appearsreasonable if one improperly assumes that stability is necessary inorder to maintain the efficacy of the legal system, rather than tofacilitate legal guidance. The core concern is that volatility reducesthe strength of the signal created by law’s official mark. Without astrong signal, ordinary folk will have difficulty locating the content ofthe law, which leads to a decline in epistemic guidance, which


reduces the efficacy of the system. This is a serious concern, yet itdoes not explain why stability is necessary for law to possess thecapacity to guide conduct. We are thus left with a dilemma.

This dilemma is this: We must reject either the claim that stabilityis an attribute of guidance or the claim that epistemic guidance is anaccurate account of legal guidance. Stability is widely thought to bean attribute of guidance, and epistemic guidance is the most prom-inent conception of legal guidance. Perhaps the way out of thisdilemma is to modify our conception of legal guidance. Thus, wewill consider an alternative conception of legal guidance. On thisconception, stability is, in fact, a necessary attribute of guidance.

VI. INTEGRATED GUIDANCE

Integrated Guidance (‘IG’) reconceptualizes law’s guidance func-tion.54 IG treats guidance as possessing both epistemic and motiva-tional dimensions. In addition to providing epistemic guidance, lawmotivates individuals by disposing them to comply with it. IGmodels individuals as planning agents with the capacity to incorpo-rate legally sanctioned means in their plans. Law guides when itsnorms, owing to their status as dispositive settlers of normativecontroversies, exert pressure on individuals to use the norms’ con-tent as the bases of plans. If the goal of individual deliberation is toefficiently coordinate a large number of beliefs and desires, legalnorms should outperform any other type of norm, since no othertype of norm promises to be dispositive of controversy. Broadlyspeaking, this process can occur via either of two routes: filtering andfacilitating. Legal norms that prohibit certain behavior filter outunsanctioned means from individuals’ plans. Legal norms thatfacilitate conduct offer individuals opportunities to employ sanc-tioned means to achieve their ends. The content of legal normsfilters out some means, and creates others. The result is a legallysanctioned set of means from which individuals can choose. Ifindividuals create higher order intentions structured around thislegally sanctioned set of means, then they will be more disposed tocomply with the law. Because plans require a certain degree ofstability, the means by which individuals accomplish those plans

54 We find strong hints of IG-like positions in the literature, most notably in the work of later ScottShapiro, but also in the work of Postema and Raz.

ADAM HILL

must also possess a degree of stability. But we cannot simply leavethings there, since individuals will have different preferences overreconsideration. We will attempt to sketch how much specificity isrequired, given a population with certain preferences over recon-sideration.

IG is most naturally modeled around Michael Bratman’s planningconception of individual agency.55 Very roughly, the idea is thatindividuals possess certain desires, or outcomes that they hope toachieve. Most of the time, these desires are not immediately satisfied;certain intermediate steps need to be taken to render their realizationmore likely. The planning conception of agency explains how westructure means and ends together to help us to achieve our desires.Building on the planning conception of agency, IG explains how,given the way in which individuals ordinarily structure their meansand ends, legal norms guide behavior. Individual action is primarilygoverned by a set of hierarchical plans, to which one is non-triviallycommitted, and by which one is guided in the making of morespecific decisions. The existence of such plans is essential, because weall face severe resource limitations: limitations of time, energy,cognitive capacity, and so forth. And, yet, despite these limitationswe nonetheless must coordinate a large set of diverse activities. Plansare special sorts of individual intentions that help to overcome theselimitations.56 Individuals weave plans of various levels of generalityand complexity together in order to create an agenda for futureaction. Rather than lurch from decision to decision, unguided by anybackground agenda, individual action is designed to bring about thegoals contained in the plans, and do so efficiently. Inefficient means,or impossible means, are ‘filtered out’.57 Individuals are left withplans that exert strong, although not absolute, control over anygiven decision that an individual makes. Plans can be reconsidered,as we shall shortly see, but IG treats reconsideration as the exception,not the rule. In case an individual does reconsider a plan, she mightreach for a new plan from her ‘plan library’, or a set of beliefs about

55 Michael E. Bratman, Intentions, Plans, and Practical Reason (Cambridge: Harvard University Press,1987); Shapiro, supra note 12.

56 Michael E. Bratman, ‘Planning and the Stability of Intention’, Minds and Machines 2 (1992): 1–16.57 Michael E. Bratman, David J. Israel, and Martha E. Pollack, ‘Plans and Resource-bounded Practical

Reasoning’, Computer Intelligence 4 (1988): 350.


which actions bring about which effects under a given set of cir-cumstances.58

Because plans by their very nature strive to reduce the costs ofaccomplishing one’s ends, impossible, or inefficient, means are fil-tered out. This is the feature of individual agency that allows law todispose individuals toward compliance. Those subject to law tend tobelieve that (a) purposive agents, namely, legal officials, (b) createlegal norms addressed to them (c) in order to settle questions of howto act.59 If an individual has these three beliefs, legal norms willoccupy a privileged place in an individual’s deliberations.

Because such agents accept that a hierarchical relationship existsbetween themselves and the state, such agents take the content ofthe law to provide plans for their own actions and seek to ensurethat other plans are consistent with legal norms.60 The content oflegal norms thus has the capacity to induce rational agents to inte-grate their conduct within the parameters of the law because suchagents take the law to be an authority within its domain. Law guideswhen its norms, owing to their status as dispositive settlers of nor-mative controversies, exert pressure on individuals to use theircontent as the bases of plans. Law is designed to settle normativedebates. If law does what it claims to do, it will provide a stablefoundation for derivative plans. If the goals of a deliberative systemare to efficiently coordinate a number of desires, legal norms shouldoutperform any other type of norm, since no other type of normguides with the backing of the state. Individuals thus desire toincorporate its content into their plans. This process of incorporationcan occur via either of two routes: filtering and facilitating.

VII. FILTERING AND FACILITATING

Legal norms that prohibit certain behavior filter out possible meansfor the accomplishment of desires. For instance, if an individual has a

58 Id. For instance, if Alice possesses a plan to travel to a party at 9 p.m., then she can filter out alloptions that are incompatible with the possibility of attending the party, such as planning a dinner for9 p.m. Faced with a smaller set of alternatives, Alice can more easily focus on the relevant questions:how to get to the party, and so on. And when 9 p.m. arrives, Alice will, in fact, travel to the party,absent special circumstances.

59 Shapiro, supra note 12, at 200.60 Id., at 141.

ADAM HILL

desire to ‘get rich quick’, criminal prohibitions on insider tradingmight filter out the means of, inter alia, buying or selling securitieson material, nonpublic information in breach of a fiduciary duty.61

By filtering out certain possibilities, non-filtered possibilities facefewer competitor means. Legally sanctioned means are, in a sense,elevated as non-legal means are winnowed away.

Yet, as Hart famously observed, legal norms do more than pro-hibit conduct: they also create and facilitate new forms of conduct.By making available new forms of action, legal norms create newmeans by which individuals can accomplish their ends. Although it isodd to speak of ‘compliance’ with legal norms that facilitate conduct,it is not odd to imagine that legal officials in some sense favor the useof such ends over the use of non-legal ends. It makes sense, then, toconceptualize guidance as disposing individuals to employ legally-created forms of action, even if the language of compliance isinapposite.

The content of legal norms filters out some means, and createsothers. The result is a set of legally sanctioned means. If individualscreate higher order intentions structured around this legally sanc-tioned set of means, individuals will be more disposed to complywith the law. Law influences dispositions by influencing the struc-tural framework of – the higher level plans – individual deliberations.This dynamic has the effect of crowding out alternative sources ofplanning norms. Recall that legal norms compete against othernorms to guide individual conduct. One of the primary competitiveadvantages of law is that it usually has the state’s force behind it.Nonetheless, as the fact of non-compliance shows, legal norms donot always claim victory. IG shows how legal norms stay competi-tive by reducing the foothold that alternatively-sourced norms mighthave in individuals’ planning mechanisms.

Raz hints at such an idea in his discussion of stability. ‘[O]nly ifthe law is stable are people guided by their knowledge of the content ofthe law’, Raz argues.62 Implicit in Raz’s argument is the idea that, inthe absence of legal guidance, individuals will plan according toalternatively-sourced normative standards. If law is unstable, indi-

61 Such prohibitions may also filter out certain ends. One might be enticed by the thought of a life ofcrime, for instance. Our focus on means here should not be read as denying that prohibitions also filterends.

62 Raz, supra note 13, at 229 (emphasis in original).


viduals are guided by their knowledge of some other norm. Legalnorms, though, can crowd out competing norms, and thus makethem less probable to influence individual behavior. By creating a setof legally sanctioned means, law disposes individuals toward creatinghigher order intentions that are compliant with law. The moreindividuals craft their intentions in a legally compliant manner, themore dependent they become upon legally compliant behaviors. Aswith many systems engineered to produce certain outputs, the moreengaged one is with a system, the more dependent one becomes onthat system, the harder it becomes to exit that system. And so it iswith law.

Plans allow individuals to coordinate vast swaths of desires in anefficient manner. They are efficient solutions in part because theyallow individuals to incur the costs of weighing options, deciding onmeans, and coordinating those choices with other of their plans andother individuals. It thus seems as though plans only efficientlystructure decision-making if they are relatively stable. After all, ‘if wewere constantly to be reconsidering the merits of our prior plansthey would be of little use in coordination and in helping us copewith our resource limitations’.63 ‘Nonreconsideration’, in otherwords, ‘will typically be the default’, even in cases where, were theindividual to reconsider the decision, she would plan differently.64

Under IG, legal norms must be capable of forming the basis ofplans through legally-sanctioned means. And plans, we just saw, arein the normal course of things stable. Legal norms, then, too, itwould seem, must be stable. Predicating plans on unstable meansseems like a recipe for incurring the costs of reconsideration. Therole of stability under IG is, at bottom, about the way in whichreconsideration affects individuals’ ability to construct plans aroundlegally-sanctioned means. Plans are premised on the existence ofcertain conditions. When the conditions upon which a plan is pre-mised change, an agent may reconsider her previously formed plans,but she need not. Non-reconsideration is the default position, but wecan sketch with slightly more specificity the conditions under whichindividuals override this default.

63 Bratman, supra note 56, at 3.64 Id.

ADAM HILL

Changes in the world exact one of two types of costs in relation toindividual plans. On the one hand, changes can exact costs related tothe agent’s deliberative structure. Reconsideration of plans premisedon changes takes time, energy, and inhibits coordination with others.If an agent prefers to avoid those costs, he might incur a differenttype of cost: the costs of attempting to bend a changed world back toa state in which it is compatible with his original set of plans. For allindividuals interested to keep their plans and the state of the world insome sense connected, change exacts one or the other of these typesof costs. Individuals choose different strategies to manage thesecosts.65 At one end of the spectrum, individuals’ plans are only asstrong as the set of beliefs and desires that formed the original basisfor the plan. If the agent’s plans are non-robust in the face change,then the agent will frequently be attempting to bring his plans in linewith the changed state of the world. The costs typically borne ofsuch an agent include lost efficiency gains, since he expends re-sources recalculating means and ends, and lost coordination gains,since others that rely on his previous plans will no longer reliably doso.66 Such plans will struggle to serve the usual purposes of a plan,since they filter few actions out and ask the agent to continuallydeliberate over new options. Nonetheless, some agents may holdsuch a view, at least about some matters.

At the other end of the spectrum, plans possess an ‘intrinsic sta-bility’, which commits the agent to attempt to bring about theirsatisfaction conditions, come what may.67 If an agent never recon-siders his plans, then volatility will impact the agent in a character-istic way; namely, the agent will be forced to attempt to bring theworld in line with her pre-established plans. The costs typicallyborne of this type of agent include those involved with the open-ended challenges of attempting to make whatever aspect of theworld is no longer compatible with his original set of plans onceagain compatible. Outside of these two poles, individuals toleratevarying levels of reconsideration. This variance need not be con-sidered the product of irrationality. Some individuals may be quite

65 Bratman, supra note 54.66 For instance, consider the deliberative difficulties facing an agent with a higher order plan

grounded in act consequentialism, or some other set of values that requires that an agent constantlycalculate the value of her action.

67 Id. at 11.


adept at shouldering the costs of reconsideration, while others maybe skilled at bending the world to their wills.

VIII. DELIBERATIVE TYPES

We see, then, that individuals prefer to allocate change costs indifferent ways. This allows us to imagine a spectrum of approachesto the allocation of change costs.68 We will identify two intermediatetypes, producing four types in total.69 We can label these types asfollows:

1. Non-Robust: Prior plans are always subject to reconsideration, given arelevant change.

2. Weakly Robust: Prior plans are subject to reconsideration, given arelevant change, only if the anticipated costs of reconsideration are lessthan the anticipated benefits of reconsideration, with regard to a spe-cific action.

3. Strongly Robust: Prior plans are subject to reconsideration, given arelevant change, only if the anticipated costs of reconsideration are lessthan the anticipated benefits of reconsideration, with regard to anagent’s overall dispositions about reconsideration.

4. Intrinsically Stable: Prior plans are never subject to reconsideration, nomatter the change.

The existence of deliberative types complicates IG’s account ofthe value of planning. Recall that future-direct intentions, or plans,allow us to deliberate more efficiently and coordinate intra- andinter-personally.70 It would seem to follow that an agent will notadopt or will tend not to adopt a plan unless the plan facilitatesefficient deliberation and intra- and inter-personal coordination. Anyplan, the content of which is likely to induce inefficient deliberationor inhibit coordination is likely to remain in an agent’s ‘plan library’of possible, but unadopted, plans.71 Given an agent’s knowledge of

68 This section draws heavily on Bratman’s discussion of stability of intentions. See Bratman, supranote 56. Any given individual might be a certain type with regard to one issue, but a different type withregard to another issue, since the costs of being one type or the other will vary based on the issue andindividual capacities.

69 We will assume that individuals know their deliberative types, and find it rational to be that type.If an individual knows his deliberative type, and the costs associated therewith, he will tend to structurehis plans in such a way as to bring about the satisfaction of rational desires.

70 Bratman, supra note 55.71 Id.

ADAM HILL

her type, she will rule out some plans in the plan library on groundsthat they are subject to too much volatility. Given a potential planfrom the plan library, there is some threshold of probability at whichthe agent, given her tolerance for reconsideration, will reject theplan. Given the choice of plan A, there is some probability, Pr(A),that, out of the set of possible future events, at least one event thatforces the agent to reconsider will materialize. When that Pr(A)reaches the threshold that the agent sets based on her deliberativetype, that potential plan is rejected.

Strongly Robust and Intrinsically Stable types may be wary ofincorporating plans that they anticipate will be subject to volatilitythat destabilizes those plans. This wariness may result from the factthat they tend not to reconsider plans. Bending the world to one’splans can be difficult, even where one prefers that course of action toreconsidering her plans. Of course, some changes in the world mayinduce such high costs that reconsideration is likely even for StronglyRobust and Intrinsically Stable types. As a percentage of all changes,however, these cases are likely to be low, given these types’ policiesabout change.

Strongly Robust and Intrinsically Stable types, precisely becausethey make it a policy not to reconsider, may plan for contingencies.If this is so, then these types may, in fact, not be considerably morewary of potentially unstable plans. Yet, contingency planning iscostly, just as buying insurance is costly. Thus, it will be sought to beavoided where possible. In general, then, we can hypothesize thatStrongly Robust and Intrinsically Stable types will possess a com-paratively low volatility threshold. In other words, for these delib-erative types, a comparatively low level of volatility will cause themto leave plans in their planning libraries, rather than adopt them. Forinstance, say an Intrinsically Stable type is considering whether tobuy a piece of land on which sits a large lake, for the purposes offishing. An assessment of the regulatory climate, however, suggeststhat fishing from lakes such as this one might soon be prohibited.The question is, given our individual’s policy of not reconsideringplans, our individual decides not to buy the property. It seem asthough, because of his policy to not reconsider plans, our individualwill decide not to buy the property. But, as an Intrinsically Stabletype, he might be comfortable incurring the risk and attempting to


bend the regulatory process to his preferences. For symmetric rea-sons, Non-Robust and Weakly Robust types will possess a compar-atively high volatility threshold.

Law’s capacity to guide is typically thought of as a capacity thatextends uniformly over all individuals. IG requires a differentiatedapproach to guidance, since it is premised on the idea that individ-uals possess different tolerances for the risks associated with vola-tility. By introducing Deliberative Types, we have introduceddifferences in the population. We must find some way of discussingguidance as a capacity possessed by law while also taking account ofthese individual differences. The solution requires analysis of thedynamic and cooperative character of legal guidance.

The state faces a limit on how often it can change the content ofthese legal norms, if it desires individuals to use them as part of theirplans. Given a potential plan from the plan library, there is somethreshold of probability at which the agent, given her tolerance forreconsideration, will reject the plan. This probability threshold,however, is not permanently fixed or wholly independent of indi-viduals’ wills, as we have thus far imagined.72 Even if we hold one’sdeliberative type constant – even if one maintains that an individual’sdeliberative type is beyond that individual’s capacity to change – theindividual can, nonetheless, structure his affairs such that the possi-bility of legal change is less likely to reach the threshold beyondwhich the legal norm is perceived to be too volatile to form the basisof a plan. In other words, individuals can plan for legal change itself.But why do so? Purchasing insurance, after all, is costly. StronglyRobust and Intrinsically Stable types, precisely because they make ita policy not to reconsider previously established plans, may chooseto plan for certain sorts of contingencies. If this is so, then these typesmay not be considerably more wary of potentially unstable plans.However, all things equal, such planning is costly. Why incur thesecosts?

72 Whether an individual is of a certain deliberative type is, in some sense, a function of thatindividual’s desires. If an agent really is committed to act consequentialism, then it makes sense to treatprior plans as always or almost always subject to reconsideration. But whether an agent is in factcommitted to act consequentialism is itself subject to reconsideration. Likewise, individuals can seek toreject fewer legal norms on grounds that they are insufficiently stable. An individual can lower theprobability threshold at which he rejects possible legal norms due to its likely volatility if he movescloser to the Non-Robust type.

ADAM HILL

IX. LEGAL GUIDANCE AS ASSURANCE GAME

We can explain why individuals incur these costs by observing thatthe state’s provision of legal norms takes the form of an assurancegame. Both the state and individuals benefit when the state respondsto widely held problems and individuals are disposed to adopt theregulations. But both of these endeavors are costly, if unrecipro-cated. One might hypothesize, then, that legal change will tend tooccur when both sets of parties are sufficiently assured that the otherwill cooperate.

Let us assume that the state faces a choice between enacting alegal norm that is responsive to some widespread problem. The statecan either change the content of the law or not change it. Citizens,meanwhile, can choose to plan for volatility or not plan for it. Ofcourse, some individuals, namely, those Non-Robust types, need notplan for volatility in order to cooperate. They are cooperative bytheir very type. Outside of the special case of Non-Robust types,however, the actions of the state and individuals are interdependent.The state is better off changing legal norms in response to wide-spread problem if and only if individuals tend to plan for volatility.Likewise, individuals are better off incurring the costs of planning forvolatility if and only if the state enacts responsive legal regulations.Otherwise, they have spent resources – deliberative and perhapsmaterial – preparing for legal guidance that never arrives.

We can represent legal guidance as an assurance game:

In order that the state be willing to change a legal norm, it likelyanticipates that individuals will be guided by it. As we observedearlier, there is some volatility threshold above which individuals of

Individuals

Change Not-Change

State Change

Not-Change

5, 5 -5, 0

0, -5 0, 0


a certain deliberative type will not incorporate a legal norm intotheir planning apparatus. Individuals encourage the state to enactnew regulations when they raise that threshold.

An unreciprocated attempt at cooperation is worse than main-taining the status quo. This is true of both the state and its subjects. Ifindividuals believe that the state cannot or will not enact responsiveregulation, it is preferable not to incur the costs of planning, all otherthings equal. Likewise, the state will seek to avoid attempting toprovide legal guidance if it is likely to fail.

The dynamic conception of stability incorporates the possibilitythat individuals will raise their volatility thresholds. Although, as weobserved, raising one’s volatility threshold is costly, we have nowseen why individuals may be inclined to do so. Individuals cooperatewith the state by planning for beneficial legal change. Where this istrue, volatility thresholds decrease. If so, then we have a solution toour problem of differentiated guidance: volatility thresholds are nolonger so differentiated. If individuals plan for legal change, theirvolatility thresholds clump together, thus providing officials with atarget range of stability that must be met. And, thus, we have de-rived an answer to the question of why stability is a necessarycondition in order that law guide individual behavior. The degree ofstability necessary in order that law guide individuals is just thatdegree required to meet the volatility threshold of some significantportion of the population.

X. CONCLUSION

The conventional wisdom among positivists holds that instabilitygenerates uncertainty and this uncertainty itself impairs guidance. Inthis paper, we have rejected that claim, and suggested, instead, thatindividuals’ expectations about the volatility of legal norms is far moreimportant to legal guidance. If individuals anticipate volatility, orhave non-legal reasons to prefer volatility, then unstable legal normswill pose few problems.

Raz argues that stability requires ‘that the making of particularlaws should be guided by open and relatively stable general rules’.73

The concern is that instability produces uncertainty, which results in

73 Raz, supra note 13, at 213.

ADAM HILL

individuals failing to guide their conduct by law. When this is not thecase, Raz argues, it becomes ‘difficult for people to plan ahead on thebasis of their knowledge of the law’.74 Our analysis suggests a morecomplicated picture. Whether individuals are able to plan ahead onthe basis of probabilities about what the law will be depends on theirgeneral deliberative type and the level of resources that they havedevoted to the issue. If the issue matters to individuals, they willdevote resources to understanding the probabilities that the law willchange; they will take out insurance to compensate against change;and they will seek to push the law in one direction or the other.What’s more, as we have shown, individuals can recognize that thestate’s responsiveness to issues can be a public good, and respond tothis realization by planning for legal change.

As one of the primary tools of democracy, law is deployed torespond to problems. The sorts of problems that contemporarydemocratic regimes solve do not arrive in neat intervals; they areoften significant and difficult, and the solutions partial and tempo-rary. Instability is high because long-term solutions are technically orpolitically impossible. To demand stability in the name of legalguidance is to effectively curtail the process of problem solving.Because it is usually preferable that the state be flexible in its ap-proach to problem solving, it is important to know just how muchinstability is compatible with legal guidance. We have tried to movepast the simple refrain that law must be stable, so common in thelegal philosophy literature, and explain just how much stability is, infact, compatible with legal guidance.

UC-Berkeley, 2240 Piedmont Ave, Berkeley, CA, 94704, USAE-mail: [email protected]

74 Id. at 216.


2 Stability, Assurance, And the Concept of Legal Guidance

Documents