Hart and natural law: the three concessions reviewed

The story so far. Herbert Hart was one of the founding fathers of contemporary legal positivism – the doctrine that laws are created through a specialised social practice of setting down (or positing) enforceable norms for behaviour, and are known and recognised as such by their grounding in that social practice. Sociologically speaking, this may sound uncontentious; the key contribution of legal positivism is the lack of any further stipulations as to what the law, inherently, is or what it is (necessarily) for. In any law-governed society (legal positivism argues) there is a system of laws which are recognised as such and an apparatus of roles and procedures for making and amending those laws; and (as Private Eye would put it), er, that’s it. To say that the legal system of Nazi Germany was a bad legal system – on the basis that it facilitated what are almost universally held to be evil and unjust outcomes – makes no more sense than to say that the language spoken by the Nazis was a bad form of German.

Advocates of ‘natural law’ – the theory that there are coherent and discoverable pre- or extra-legal principles, generally based on morality, by which positive law can be judged – have held that this picture needs to be qualified in different ways. In writings from 1958 and 1961, Hart singled out three ways of arguing the connection between natural and moral criteria, on one hand, and positive law on the other, and made concessions – ranging from major to very minor indeed – to each of them.

Firstly, the Substantive Natural Law position holds that any imaginable legal system – or, perhaps, any imaginable adequate legal system; any legal system functioning as a legal system – will have a certain minimum content: there are certain things that ‘the law’ has always forbidden and always will forbid, murder being the most obvious example. Hart gave this argument extended consideration and was willing to concede – if the minimum content was defined sparingly enough – that it might be correct. However, Hart was at pains to point out that the minimum content itself derived from the brute facts of human existence rather than from morality. Hart’s definition of the minimum content of law, in terms of obligatory forbearance from exploiting fundamental universal vulnerabilities, is compatible with Kelsen’s austere severance of legal from moral obligations: one could argue, following Kelsen, that the moral obligation not to kill and the legal prohibition of homicide are not only two separate norms, but elements of two distinct systems of norms, each of which is offered as a solution to the problems of coexistence in society. Natural law in this light is ‘natural’ in a Hobbesian sense – it is the law that elevates above the state of nature for the sake of survival (rather than, for instance, underpinning the Aristotelian conditions of human flourishing). Nor is there any inherent connection between minimal substantive natural law and justice. A universal and impartial breach of natural law would not be unjust; in the case of any more selective breach, the injustice would derive from the selectivity.

The Procedural Natural Justice position, secondly, holds that there is justice in the correct administration of the law, irrespective of the justice of the outcomes or of the law itself – or, conversely, that there is injustice in the maladministration of the law, irrespective (again) of our assessment of the law itself or of the outcomes reached. There is – by design – no connection between our assessment of natural justice in the administration of the law and the content of the law, natural or otherwise. In 1958 Hart referred to rule-based impartiality as ‘natural procedural justice’ and suggested that this was a moral virtue: “there is, in the very notion of law consisting of general rules, something which prevents us from treating it as if morally it is utterly neutral, without any necessary contact with moral principles”. In The Concept of Law he wrote, more guardedly, of “apply[ing] a law justly” and referred to procedures to ensure impartiality as “requirements of justice”. In other words, Hart’s 1961 formulation withdraws his earlier concession to the ‘procedural natural justice’ position, leaving open two alternative lines of argument: either justice in this sense is a term of art – a technical term for the appropriate administration of those things called laws – or it is a virtue in its own right. This may be a distinction that makes no difference, however: in the second case justice is not conceived as a moral virtue.

Lastly, the (Natural) Morality of Law position suggests that the law as a system – the properties of the laws themselves as well as the way in which they are administered – can be critiqued on moral grounds. The argument here is not merely that it’s a bad thing for laws to be confusingly worded or badly implemented, although it draws strength from that intuition. Rather, the argument is that subjecting society to governance by law is itself a morally good enterprise, and that the particular moral virtues which it embodies can be found – or found to be lacking – in particular legal systems, in individual laws and in acts of law-making. A key concept in this respect is followability: the law is taken to be a system of norms which addresses its subjects as free, rational and responsible for their own actions – and which is the less ‘law-like’ the less effectively it does so (through the use of laws which are unknowable, incomprehensible, impossible to comply with, capriciously applied, etc).

Hart did not address this argument in his 1958 paper. In The Concept of Law he considered it as a technical argument, on the basis that any system of ‘[social] control by rule’ needs to have certain characteristics – ‘[the rules] must be intelligible and within the capacity of most to obey… [so that], for the most part, those who are eventually punished for breach of the rules will have had the ability and opportunity to obey’. Might these requirements be taken as rising to the level of morality? Hart’s response is brief and devastating: “if this is what the necessary connection of law and morality means, we may accept it. It is unfortunately compatible with very great iniquity.” In other words, the fact (as Hart saw it) that unjust and immoral outcomes are entirely compatible with a pursuit of a ‘followable’ system of laws makes the idea of ‘followability’ as a virtue irrelevant or frivolous.

To sum up: Hart concedes the argument for Substantive Natural Law, but only after narrowing its scope to the point of guaranteeing mere survival rather than promoting human flourishing. In 1958 he concedes the argument for Procedural Natural Justice; in 1961, however, he situates procedural justice either as technical excellence in the administration of justice or as a (non-morality-based) virtue in its own right. His position on the Morality of Law is more complex: he argues that the merits described thereby are purely the technical merits of a functioning system of rules, but leaves open the possibility that they can be called a morality – subject to the withering proviso that such a morality, and such a necessary connection between law and morality, appears to have no actual effect on the law. In other words, Hart does not challenge the logic of Fuller’s argument but questions whether it is necessary to analyse the law in the real world, with the strong presumption that it is not.

As we have seen, David Lyons’s papers address the second and third of these concessions, from a position which can perhaps best be described as a radically sceptical formalism. Lyons’s key move – made in both papers – is the technical argument, which he raises to a higher level of abstraction than had Hart: he grants that certain features might represent a defect in the law (or its effective administration), but denies that this necessarily tells us anything about justice.

On Procedural Natural Justice, Lyons’s bracingly sceptical denial that there was any moral virtue in proceeding by rule or in treating like cases alike would certainly have appealed to Hart; a post-Lyons revision of the argument in The Concept of Law might have stressed the idea of procedural justice as a technical merit of a well-administered legal system, to the exclusion of granting it even nominal consideration as a form of justice. However, I’m not convinced that this move is one Hart would have wanted to make. Lyons’s central argument can be summed up in a question posed rhetorically in his 1973 paper: “Why should we suppose that the pattern of treatment prescribed by the law is the same as (or even compatible with) that prescribed by any principle of justice?” As I noted above, considerations of justice come into play in some fields but not others. If we, as a kind of thought-experiment, think of law as a field in which justice may not be involved, there is then no particular reason to think of the administration of law in terms of justice of outcome, or by extension to think of the process of the administration of law in terms of procedural justice. But I think, for Hart, that would have been a formalist move too far. I do not think Hart would have found it necessary or useful to drive a wedge, as Lyons does, between the concepts of ‘law’ and ‘justice’, both of which Hart saw as human, social, culture-bound concepts. While Hart promoted the classification of merits in the administration of the law in neutral, technical terms, I don’t think anything turned for him on not calling these merits principles of justice. Ironically, I think he would have sympathised with Lyons’s devalorised version of rule-following as a way of describing the administration of the law, but rejected as idealism Lyons’s unqualified references to justice as a virtue.

Lyons shares Hart’s lack of enthusiasm for the Morality of Law but criticises it from a different angle. While an inadequately followable law can reasonably be seen as defective, Lyons argues, additional information would be required in order to call this defect an injustice. Hart’s position, by comparison, is both weaker and stronger: weaker, in that he takes no definite position on whether Fuller’s ‘morality of law’ is a moral framework or merely a set of technical considerations; stronger, in that he effectively dismisses the debate as not worth having.

Lyons’s argument, like his position on formal justice, is grounded in his scepticism about the relationship between law and justice. This (as I have argued in both instances) over-reads the moral neutrality of well-formed rules and rule-following. Certainly rules may be contradictory or incomprehensible without any injustice being done, just as rules can be followed or broken without moral implications. However, in social practices where distributive or allocative justice is at stake – where a single standard is being applied to multiple individuals, with potential consequences affecting them – procedural justice is necessarily engaged, and rule-writing and rule-following become questions of justice. This is the case with regard both to practices which may affect the outcome and (perhaps less obviously) practices which cannot – if only because there is no way to be sure, either during or after the process, that any discriminatory practice can be ruled out as not affecting the outcome. To be singled out for special treatment (good or bad) in the course of a court case – even if one has a well-founded confidence in one’s innocence and the fairness of the court – is necessarily, and of itself, to be a party to (procedural) injustice. Similarly, if the state asks citizens to obey a law and obedience is impossible – or only intermittently possible, or unascertainable – each citizen is placed in an anomalous position, and one of unjust disadvantage relative to the large number of citizens who obey the law (or who may be obeying the law, or who may be treated as if they are).

However, I don’t think Hart’s position on the morality of law would have been affected either by Lyons’s sceptical argument or by my response. His position on the virtues of a good legal system, as identified by Fuller, is that these are indeed virtues of a good legal system, in the same sense that sharpness is a virtue of a good knife and undetectability is a virtue of a good poison. He might even have granted the point that Lyons denies – that procedural virtues in the process of making law and administering justice can be considered, by extension, as forms of procedural justice. He might have been resistant to this line of thinking (his references to it in The Concept of Law are guarded at best), but I think fundamentally he would have thought it to be beside the point – neither true nor importantly false, merely irrelevant.

Lyons draws a line – I think incorrectly – between ‘law’ in the world and ‘justice’ in the heaven of ideas. The line Hart draws encompasses law and justice, both located in the world, both variable from one system to another, both amenable to more or less technical assessment. Arguably even morality is inside the line – it too exists in the world and can be invoked (although it need not be) in the workings of the law, as a source of principles and as a means of assessing outcomes. Outside the line are the virtue of procedural justice and the morality of well-written law – and they’re outside not because Hart situated them in a Platonic realm of concepts but because he was indifferent to whether they existed or not; his model had no need for them. The debate between Hart and Fuller was not between two rival framings of the law, but between an advocate of one framing and someone who believed that it was irrelevant – that nothing turned on whether it was used or not. This may account for the occasional asperity of the debate between Hart’s and Fuller’s partisans. Hart’s famous formulation that the morality of law was “compatible with very great iniquity” is, logically, a challenge to the position that it is impossible for an iniquitous regime to have a followable system of laws – this being the only position which would have made it necessary to incorporate the inner morality of law into his thinking. However, the position advanced by Fuller and his successors is merely that an entirely followable system of laws (whatever the laws’ content in all other respects) is more just than the same system would be if it were less followable. This implies that an iniquitous regime, with no respect for its subjects, will have no reason to avoid introducing unfollowable laws; this in turn suggests that it is unlikely that an established iniquitous regime will not have taken the opportunity to introduce unfollowable laws. However, this is not to say that a followable system of laws is not “compatible with very great iniquity”; at its strongest, Fuller’s argument leaves Hart’s position untouched. Hart’s challenge is unanswerable; if Fuller’s model is to be used, other justifications are needed for doing so.

One final point, before I conclude this series of posts by looking at other writers’ comments on these issues. A striking virtue of Fuller’s argument is that it considers legal systems as a whole, arguing that they may exhibit the same merits and defects in many different ways. An individual law may be unfollowable for reasons of content, structure, administration or enforcement: because it clearly requires the impossible, or because it is drafted so badly as to be incomprehensible, or because it is liable to be changed without warning, or because it is only capriciously enforced. With this in mind, it is worth recalling the first aspect of the minimum content of natural law – the substantive element – and asking whether it may have any bearing on the other two, wholly or partly procedural, elements. If laws – some laws – are required in any conceivable human society, for the sake of bare collective survival, does this tell us something about the nature of law? Might it be appropriate – natural, indeed – to take as a starting point the assumptions that (contra Lyons) law does in fact embody the value of justice in society, and that (contra Hart) this value is of supreme moral importance?

Two counter-arguments can be envisaged, one minimal and one maximal. On one hand, it could be argued that justice is comparative and distributive: when an outcome is applied evenhandedly to the whole of a society, justice has nothing to say about what that outcome is. It is unjust for one person to be killed; it cannot be unjust for an entire society to perish. Consequently, although the minimal content of substantive natural law embodies the network of collective forbearances required to keep society in existence, this is not a goal of justice. An answer to this is that what the minimum framework of forbearances wards off is not instantaneous extinction but a lawless chaos, in which many injustices would in fact be done: the survival of society is not a goal of justice but a condition of justice, or of the prevention of injustice. On the other hand, it could be argued that justice is – as Lyons suggests – something we know rather little about: from a Marxist point of view, for example, a genuinely just settlement might involve the withering away of the state and the communalisation of private property. The answer to this is that justice is scalar rather than binary: an innocent prisoner detained without trial is a victim of injustice, but she may still become a victim of further injustice (by being singled out for mistreatment, isolation etc). Conversely, a structurally unjust society (e.g. one dominated by a single low-wage employer) may become more just without changing that fundamental condition (e.g. by the introduction of more effective factory regulation).

Let us say, then, that the survival of a functioning society is a precondition of justice, and that the law – as well as being a means to that survival – embodies, however imperfectly, the pre-legal standard and value of justice. If these points are granted, can we still ground Lyons’s scepticism about the relationship between law and justice – or Hart’s bracketing of the inherent virtues of law, so as to discuss laws alongside the rules of games? Does Hart’s “minimum content of natural law” in fact lead us back to natural law theory?

Mutterings in favour of Kelsen

Hart’s Essays in Jurisprudence and Philosophy includes two pieces engaging, in tones of patient and courteous bafflement, with the ‘pure theory of law’ propounded by Hans Kelsen. Hart focuses on four main assertions:

1. Law is a system… All valid laws, qua valid laws, form a single system. Kelsen was willing to extend this argument to cover international law, with the implication that there is only one system of law in the world.
2. …which is logically coherent, Every legal system must be logically self-consistent throughout, such that no two valid laws can contradict each other.
3. rests on a basic norm… While a legal system must be observed (treated as binding on conduct) in order to have any validity, every legal system is also founded on a presupposed ‘basic norm’ on which all legal powers granted within that system, and all laws laid down by those powers, ultimately rest.
4. and has no moral content. Legal obligation and moral obligation are distinct and separate concepts, deriving from separate logical systems: if moral obligations were allowed to impinge on the law, it would be possible for obligations to conflict, which would destroy the logical coherence of the law as a system. The statement that a valid law is immoral is thus of no more legal significance than the observation that it was enacted on a Tuesday.

For completeness, I should add that Kelsen subsequently revised the second and third points here, allowing the possibility of valid contradictory laws and redefining the ‘basic norm’, not as a norm which was presupposed rather than having been enacted, but as a fictitious norm enacted by a fictitious lawmaker. The second of these is more promising than it sounds; I haven’t been able to find any commentary on the first.

Hart’s arguments against Kelsen are both meticulous and dense, but they take two main forms: demonstrations that one of Kelsen’s assertions cannot be logically sustained, or has unsatisfactory implications if assumed to be true; and demonstrations that, even if true, the assertion would not do the work Kelsen claims that it does. I’m certainly not in a position to say anything authoritative about Kelsen, let alone rebut any of Hart’s criticisms. In this post I want to take on an easier target: Hart’s bafflement. Repeatedly in Hart’s two papers we seem to hear him muttering Why would you think that? Or perhaps, Yes, that’s very neat, but why would you think it’s true? (Something similar can be heard, rather less sotto voce, in Neil Duxbury’s 2007 paper on the later ‘fictitious’ basic norm.) In Hart’s hands, Kelsen’s pure theory comes across as the proverbial beautiful hypothesis brought down by an ugly fact – or rather, a dense and elaborate hypothetical architecture undermined by a closely-marshalled assault by ugly facts. Kelsen’s model seems to do some of the work done by Hart’s own theory of law, but only some of it and not as adequately.

For me, this prompts the question: why would you think that? If we took Kelsen’s pure theory as a starting point, what would it give us – would it take us anywhere that Hart’s legal positivism doesn’t? Here are some thoughts about three of the four counter-intuitive propositions above (I’m not touching the second, on grounds of not being familiar with Kelsen’s later revisions to it).

Law as a single system: well, plainly, all valid laws don’t compose a single system. Kelsen (on Hart’s account) advances very few arguments in support of this proposition, and Hart finds it easy to dismantle those which are offered. But the complex of ideas which Kelsen reduces to this assertion – the mutual recognition (or constitution) of international and domestic law; the tendency (and on occasion the need) for one jurisdiction’s laws to be honoured by another – may be worth more attention than Hart gives them. We could argue, for example, that law tends to universality, and that this tendency (as well as practical necessity) underlies the tendency for discrete legal systems to find points of contact and forms of understanding; international law, in this framework, would be a separate enterprise undertaken to provide a single universal solution to this problem, like a connectivity standard – SOAP to municipal law’s XML. (And there goes my very last reader!) We could then go further, treat the tendency to universality as a norm (part of a ‘morality of aspiration’ in Fuller’s terms) and suggest that, to the extent that a legal system refuses arrangements of mutual recognition with other systems, to that extent there’s something unlawlike about that system. This certainly isn’t a move Hart would make – for Hart there wasn’t much more to say in formal terms about the legal system in Nazi Germany than that it was one – but that doesn’t mean it’s necessarily a bad idea. In other words, perhaps what Kelsen had in his sights here is, precisely, an aspect of the morality of law.

The basic norm: digressing slightly, I found Duxbury’s paper both enlightening and frustrating. It’s frustrating because it appears to solve its central mystery halfway through and then carry on as if it was still unsolved. (Unless I’m missing something obvious – also a frustrating thought!) To fill in the background, Kelsen saw the law as composed of ‘norms’ – conceived mainly as statements that behaviour X would attract sanction Y – which had been enacted by acts of will; the acts in question had been validated by prior norms, setting down how and by whom law could be made. However, these norms themselves had (by definition) been enacted by acts of will, creating an infinite regress (or founding the law on the brute facts of history, although this doesn’t seem to be an alternative Kelsen entertained). Hence a basic norm, never actually enacted, had to be presupposed as the foundation for the entire system. Late in life Kelsen changed tack: rather than presupposing a basic norm which (although not enacted) was both real and valid, he proposed that we treat the basic norm as a fictitious norm, enacted like all other norms, but enacted by a fictitious act of will.

Duxbury’s paper looks for support for the proposition that this isn’t as daft as it sounds, and largely fails – or rather (in my view) succeeds halfway through, almost without noticing. The key is in Kelsen’s definition of ‘fictitious’: he follows a philosopher called Vaihinger (not otherwise familiar to me) who distinguished between the partially fictitious (things which could exist but just happened not to) and the wholly fictitious (things which couldn’t exist). The basic norm, in Kelsen’s late formulation, is wholly fictitious: in other words, it’s a contradiction in terms, the paradox of a valid norm resting on an act of will which itself has no validation. Viewed in this light, the late formulation is, perhaps, a more satisfactory restatement of the earlier one. By putting our trust in the basic norm, we are not arbitrarily presupposing a norm which happens to have arisen without a prior valid act of will – which, in Kelsen’s terms, is a bit like arbitrarily presupposing a triangle which happens to have four sides. Rather, we are deliberately relying on a norm which we have defined as paradoxical and impossible: our triangle has four sides and we know it.

What does this get us? I think it gets at an aspect of the moral significance of law – another theme on which Hart wasn’t particularly keen. In terms of actually, practically grounding the validity of law, Hart cut Kelsen’s knot and warded off the infinite regress by proposing that every law-governed society has its own ‘rule of recognition’ – a rule, or practice, or assumption, or set of rules and practices and assumptions, which determines how law is made and who by. The reference is sideways rather than upwards, in other words (“Who can revise law A?” “Refer to rulebook X.” “Who can revise rulebook X?” “Refer to rulebook X.”). But, whatever the rule of recognition might in practice be, I think the idea of the basic norm – and in particular the fictitious basic norm – captures something important about why the law matters, or is believed to matter. Ultimately, perhaps, the question the basic norm answers isn’t “where did this law come from?” but “why should I obey this law?”. And here there still is a reference upwards, unless it’s cut off by a learned shrug (“why should you obey this law? because you’re a citizen of a law-governed country, this law is a validly enacted law according to that country’s rule of recognition, and as such obeying validly enacted laws is what you should do”). “Why should I obey this law?” “Because you should obey the Law[= those laws currently in force in your society].” “Why should I obey the Law?” “Because it’s right to obey The Law[=the laws laid down in pursuance of the project of subjecting social behaviour to just, consistent and uniform regulation]”. “Why is it right to obey The Law?” “Because the basic norm demonstrates how important the enterprise of The Law is – it’s important enough to be founded on a paradox; important enough for its foundations to be treated as real even though we know they’re not.”

No moral content: here Kelsen might seem to be cutting with the grain of Hart’s positivism, as well as against the grain of any kind of natural law theory. Things aren’t that simple, though; although Hart certainly maintained that the rule of recognition (and hence the law) could have no moral content, he was at pains to deny that it must have none. The rule of recognition itself could encompass moral precepts, in other words. Although we can see why Hart might have made this move – presumptively de-moralising the rule of recognition would have created hostages to fortune in his disputes with natural law theorists – I tend to think it creates unnecessary confusion. Although lawyers and legislators can and do invoke moral assumptions in their decisions (this being the avenue Hart would have wanted to close off) they do not do so in any predictable or systematic way, which is surely what would be required for moral arguments to form part of a rule of recognition. I would argue that the ‘rule of recognition’ model sits more comfortably with Kelsen’s austere division between moral and legal domains: whatever morality might dictate, the law is what is made law through the practices of legislators, lawyers and juries, and nothing else.

Despite appearances, treating law entirely as positive law is not the end of the line for the moral critique of law, or even for natural law. If law is autonomous of morality, this does not make the project of subjecting law to morality any less valid or important – if anything, the reverse is true. Moreover, if there is a ‘natural’ core to positive law, as Hart conceded; or if the conditions of possibility for a community living under the law can be considered as a ‘morality of law’, as Fuller argued; or if (as Jules Coleman has argued) morality itself requires that the law be applied consistently and interpreted without reference to moral norms; then the dichotomy between morality and positive law cannot be treated as absolute. And – paradoxically perhaps – if we want to think about how the dichotomy should be modified and reinterpreted, we’re better off starting with a theory which emphasised it (Kelsen’s) than with one which blurred and downplayed it (Hart’s).

In short, the reason why Kelsen’s theory looks logically airtight, unsustainable in the real world and ultimately slightly crazy may actually be that it is all of those things – but it’s still useful to think with, and in some ways more useful than Hart’s all-purpose scepticism.

Next up: Hart on Fuller. That should be quick and easy…

Anyone still here? Do feel free to comment, link, retweet etc.