The spark for this series of posts was a brief comment at the end of the Introduction to Hart’s Essays in Jurisprudence and Philosophy:

I hope that in what is a second exchange of friendly polemics between myself and Fuller … I have not been unfair in my criticisms of his conception of an inner morality of law; but I see now largely as the result of Professor Lyon’s [sic] essay on Formal Justice that an argument similar to mine against Fuller might be used to show that my claim made in ["Positivism and the separation of law and morals"] and repeated in my Concept of Law that a minimal form of justice is inherent in the very notion of a general legal rule applied according to its tenor to all its instances is similarly mistaken. I am not sure that it is so, but I am clear that my claim requires considerable modification.

As well as wanting to investigate the possible implications of this suggestion myself, I was intrigued to see what writers on Hart had made of it. A quick literature review – carried out by the unscientific method of searching for the name Hart and the phrase “requires considerable modification” on Google Scholar – brought back three papers, by Matthew Kramer, Leslie Green and John Gardner. (This is not a discussion of Hart’s approach to procedural justice more generally, which would have to range much more widely. To name only the most obvious omission, Gardner’s paper is a response to Nigel Simmonds’s book Law as a moral idea - which I read some time ago and, I’m afraid, have not re-read for this blog post.)

Here, then, are our three authors on Hart on Lyons, and on Hart’s later inclination to row back from the idea of a form of justice “inherent in the very notion of a general legal rule applied according to its tenor to all its instances”. I’ve resisted the temptation to write a conclusion to this post; see the previous post in the series for my considered thoughts (which, I suppose I should say, haven’t altered as a result of reading these papers!). Consider this post an addendum.

Kramer, M. (1997), “Justice as constancy”, Law and Philosophy 16:561-80.

The focus of Kramer’s paper is on Lyons’s argument against classing formal regularity as procedural justice, and on Lyons’s challenge to Hart. I have argued that there is some equivocation in The Concept of Law as to whether procedural justice is merely a technical merit of a well-functioning system of law – just as killing without detection is a technical merit of a well-functioning poison – or a quality which deserves the name of justice, albeit without any necessary moral weight. Lyons argued for the first of these positions; Kramer argues persuasively for the second. Procedural regularity, Kramer argues, “ensures that official conduct in the administration of laws is no worse (and no better) than what is required by the substantive standards of fairness in the laws themselves”; “procedural justice ensures a minimum of substantive justice – a minimum that is also a maximum, to be sure”.

The key to Kramer’s argument is in the second part of these two formulations, which might at first sight look like rhetorical flourishes: procedural justice, although it does indeed produce justice in the application of a rule to different individuals, sets a minimum that is also a maximum. As such it may be profoundly immoral, to the extent of enforcing the violation of moral obligations:

Although an official of course has a moral duty not to enforce a wicked mandate in situations where it is inapplicable, he does not have a moral duty – even a prima-facie moral duty – to enforce the mandate in situations where it indeed is applicable. He has a moral obligation not to go below the minimum of decency secured by procedural justice, but he also has a moral obligation not to treat that minimum as a maximum.

Where procedural justice sets a minimum (preventing unrestrained official harshness), it is moral; where it sets a maximum (preventing unrestrained official benevolence) it is immoral. Since it always does both of these things, it must be considered as, in itself, neutral – neither moral nor immoral.

Lyons argued that Hart’s briefly sketched examples of procedural justice – treating like cases alike, following a rule, impartiality – were empty in the absence of substantive criteria of justice; he went on to argue, contra Hart, that the justice of rule-following was no such thing and that rule-infractions were no more ‘unjust’ than infractions of the rules of grammar. Kramer’s paper arrives at similar conclusions to my posts, by a slightly different route. Where I had argued that Hart’s idea of procedural justice is best understood as a combination of rule-following and impartiality (in cases where discretion is required), Kramer focuses on rule-following and dismisses impartiality; however, he dismisses it as a stand-alone virtue and does not discuss the relationship between impartiality and discretion. Similarly, with regard to the neutrality of laws and the comparison between criminal and grammatical offences, I wrote:

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

(“Allocative justice” refers here to the assignment of legal statuses (e.g. ‘guilty’) and of criminal penalties, which are justly allocated when allocated to those who deserve them. The same argument holds if we think in terms of “corrective justice” instead.) Kramer wrote (in 1997):

a charge of procedural injustice can be perfectly apposite in connection with an infringement or a putative infringement of a linguistic norm – provided that the charge pertains to the groundlessness of someone’s condemnation (or exoneration) of the person who has allegedly infringed (or actually infringed) the norm. When we determine whether a procedural injustice has occurred in relation to such a norm, what matters is not the norm’s morally neutral status but the misdirectedness of someone’s judgment about someone else’s compliance or noncompliance with the norm’s requirements.

Green, L. (2010), “The germ of justice”, Oxford Legal Research Paper 60/2010.

Green takes his title from one of Hart’s remarks about procedural justice: “though the most odious laws may be justly applied, we have, in the bare notion of applying a general rule of law, the germ at least of justice”. This ‘germ of justice’ argument stands alongside Hart’s denial of any necessary moral content to the law – “it is in no sense a necessary truth that laws reproduce or satisfy certain demands of morality, though in fact they have often done so” – and may seem to contradict it, particularly if Hart is held to his earlier formulation: “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”.

Green asks: if the law (as a system of general rules) offers ‘the germ of justice’ and as such has some necessary contact with moral principles (the ‘as such’ here is the connection broken by Kramer’s argument), how can this be reconciled with Hart’s “separation of law and morals” – now commonly referred to as the ‘no necessary connection’ thesis? To be more precise, Green asks two questions: whether the ‘germ of justice’ thesis can be held consistently with legal positivism, and how (if so held) it could be true. An answer to the first question is to point out that the formulation ‘no necessary connection’ may be altogether too broad: to say that law does not necessarily “reproduce or satisfy certain demands of morality” is not to deny a number of other possible claims about law and morality, which a positivist might feel able to concede (e.g. “law necessarily deals with moral matters; law necessarily makes moral claims on its subjects, law is necessarily apt for appraisal by moral standards”). It may then be that a purely procedural morality of law – summed up in a statement such as “a functioning legal system is morally superior to a poorly-functioning system” – can similarly be entertained alongside a commitment to legal positivism (although Kramer, for one, would not accept this). Another answer is to stress that the ‘germ of justice’ is consistent with key principles of Hart’s legal positivism – specifically, what Green terms the ‘social thesis’ and the ‘fallibility thesis’. In other words, we can hold that the application of the law is always (in some, unspecified and perhaps very minimal, sense) a moral endeavour while also holding that the law is a social construction on which morality has no prior claims, and that actually existing legal systems may be highly fallible in terms of the justice they deliver. The ‘germ of justice’ in this latter formulation is reminiscent of Hart’s half-concession towards Fuller (“if this is what the necessary connection of law and morality means, we may accept it. It is unfortunately compatible with very great iniquity.”); it is a germ that makes no justice.

Green’s answer to his second question covers similar ground to Kramer’s response to Lyons, as well as the conclusion of my post on the purpose of law. Green notes the strong association between norms of allocation and the law, on one hand, and justice on the other: “perhaps we can say that every legal system contains allocation norms, that they are among the most important norms in the system, and that it is among these norms we find norms of justice”. Pursuing this line of thinking, Green notes that “Law is not just any old bunch of social rules”: “if there were general virtues of rule-application, and even if these had something to do with justice, then they would turn up promiscuously whenever and wherever rules are being applied. … If it were simply the ruly character of law that produced the germ of justice, then justice would have no more intimate association with law than it has with grammar or football.” Rather, the function of the legal system is to equip society with “institutions that can identify and authenticate the rules of the system, and that can render binding determinations in any disputes about them”. Hence the connection to allocation norms: “[a]uthoritative adjudication is in its essence an allocative enterprise, for its core function is to settle who is to get what, and on what grounds.” Perhaps, then, the establishment of a legal system (founded on the basic principles of procedural justice) represents the ‘germ of justice’ because it is the fundamental precondition of subjecting society to the rule of law. This remains compatible with the ‘social’ and ‘fallibility’ theses, and with a narrow reading of the ‘separability’ thesis:

To say that a legal system must of its nature settle disputes about certain kinds of concerns, and that these are moral concerns, is not to say that it must do so well or even tolerably. And as Hart insisted time and again, to say that every legal system must deliver the goods to some people does not show that it must deliver the goods to everyone, and that one of the characteristic ways that law can fail, consistently with satisfying the minimum content, is by not delivering them justly. But if we have law-producing and law-applying organs, then we do at least have social machinery by which justice could be made effective

This is an interesting and cogent argument, whose strengths (at least from my point of view) include the fact that I was thinking along similar lines before I read it; great minds, etc. Its key weakness is that – as Green concedes – the route it takes from premise to conclusion does not track Hart’s own thinking; indeed, the rule of law was a topic which Hart seems to have largely avoided, possibly mistrusting its potential for adverse judgments of ‘unlawlike’ legal systems.

Gardner, J. (2010), “Hart on legality, justice, and morality”, Oxford Legal Research Paper 44/2010.

After some fairly knockabout remarks aimed at Nigel Simmonds, Gardner focuses on Hart’s brief and often equivocal remarks about procedural justice (and about Fuller’s ‘inner morality of law’). He concedes that Hart can be interpreted as expressing outright scepticism about the concepts in question (the alternative being acceptance with limited enthusiasm). If a sceptical reading is preferred, Gardner argues, the question is where – in his formulation – the chain is to be broken: is Hart supposed to have believed that law had no necessary connection with the ideal of legality; that the ideal of legality had no necessary connection with justice; or that justice had no necessary connection with morality? In an illuminating argument, Gardner argues that Hart’s conception of law as a system of rules – “general standards of conduct communicated to classes of persons, who are then expected to understand and conform to the rules” – brings with it an assumption that laws are intended to be followed, and not simply cited by officials as a justification for coercive behaviour modification. It follows logically – whether or not Hart himself would have wished to draw this inference – that, to the extent that more laws within a system are being obeyed intelligently and willingly, that system is more lawlike. In Gardner’s words,

the ideal of legality or the rule of law is an ideal for law because there is a conceptually necessary feature of a legal system, namely that it is a system of rules, which entails that it has a proper way of functioning as a legal system, namely by guiding or (as we also put it) by ruling those who are subject to it.

As for the relationship between legality and justice, Gardner argues that the principles of legality are principles of justice: they guide free and rational individuals in such a way as to make it possible to reach a satisfactory adjustment of claims between multiple individuals. At this point Gardner endorses Lyons’s critique of formal justice (more or less in passing), dismissing any connection between justice and the law in and of itself. As with Lyons himself, the argument here seems to rest on a broader and more expansive definition of ‘justice’ than is usually implied in the phrase ‘procedural justice’. The question of whether, if an unjust society began to follow its rules more uniformly, there would be any gain in justice is not directly addressed.

Gardner sews up the relationship between legality and justice fairly tightly, but in an argument which I found difficult to paraphrase and not much easier to get to grips with. On the relationship between justice and morality he is more inconclusive and to my mind more interesting. He canvasses three readings of Hart on morality and the law, and the ‘inner morality of law’ in particular. One is that Kramer’s reading is correct – Hart believed in justice as a (more than technical) virtue of functioning legal systems, but not as a moral virtue. This reading is well supported by Hart’s own words; Gardner seems to reject it because he finds it too outlandish. A second is that Hart’s lukewarm endorsement of Fuller’s ‘inner morality of law’ should be taken at face value: Hart believed that Fuller had indeed identified a morality of law, but not a complete set of moral criteria for judging the law (which would require attention to the substantive aims of the law). This is weakly supported by Hart’s writing and – perhaps more importantly – would have represented a serious misreading of Fuller, whose ambitions for his ‘inner morality’ were less ambitious than this. The third reading, to which Gardner inclines, is itself inconclusive; perhaps, Gardner argues, the key is that ‘morality’ was a difficult concept for Hart, in whose presence his thinking became more than usually tentative and disconnected.

Of the three papers, I find Kramer’s the most convincing and most challenging – the idea of defining any form of justice in morality-free terms is counter-intuitive, to say the least, but the argument is well made. Green fills out the legal positivist background very usefully, as well as alighting on the potential connection between a procedural morality of law and the minimum content of natural law (a connection I identified myself). Gardner’s paper does an excellent job of presenting Hart as a liberal idealist, albeit one with a neuralgic reaction to the word ‘morality'; the argument is very much in line with my own thinking about the law, but as an interpretation of Hart I found it less persuasive.

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. Can we then 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?

In this & the following posts I’m going to look (sometimes quite obliquely) at Hart’s position on natural law – on the ways in which, and the extent to which, law can and should be taken to rest directly on morality, rather than constituting its own free-standing structure of posited norms.

Hart is strongly associated with legal positivism, and with what Jules Coleman has called the ‘no necessary connection’ argument in particular. Legal positivism tells us that the law is constituted, and can be identified, through rule-based social practices which confer meaning and significance on particular statements, acts and roles. The ‘no necessary connection’ argument tells us that the rule of recognition – the set of rules, assumptions and practices which qualify law as law in a given society – either may or may not include reference to moral standards. Contra Kelsen on one hand and natural law theorists on the other, Hart argued that morality and the law might (in some systems) be connected – there might be a settled and officially-recognised practice of deciding certain points by reference to moral arguments – but that the two had no inherent, general or necessary connection.

In this post I’m going to introduce three apparent concessions by Hart to natural law arguments, then introduce a late comment in which he suggests that one of these might be in need of revision. The next two posts will look at two papers by the philosopher David Lyons, one of which prompted Hart’s remark qualifying his position, and ask what implications Lyons’s arguments had for Hart. After summing up how I think Hart might have made use of Lyons’s arguments, I shall review some later responses to Hart’s qualification, from Matthew Kramer, Leslie Green and John Gardner.

The Giant Land Crab Postulate

Hart makes three concessions to natural law arguments, covering the content of law, the justice with which it is administered and the followability of law. We can refer to these as substantive natural law,  procedural natural law and the morality of law, respectively. In his 1958 paper “Positivism and the separation of law and morals”, Hart introduces the concept of a necessary minimum level of law – and hence a minimal framework of substantive natural law – by reference to the fact that we are not giant land crabs.

suppose that men were to become invulnerable to attack by each other, were clad perhaps like giant land crabs with an impenetrable carapace, and could extract the food they needed from the air by some internal chemical process. In such circumstances (the details of which can be left to science fiction) rules forbidding the free use of violence and rules constituting the minimum form of property – with its rights and duties sufficient to enable food to grow and be retained until eaten – would not have the necessary non-arbitrary status which they have for us, constituted as we are in a world like ours. At present, and until such radical changes supervene, such rules are so fundamental that if a legal system did not have them there would be no point in having any other rules at all. Such rules overlap with basic moral principles vetoing murder, violence, and theft; and so we can add to the factual statement that all legal systems in fact coincide with morality at such vital points, the statement that this is, in this sense, necessarily so.

The argument is developed further in The Concept of Law, in which Hart identifies three fundamental facts of social life from which we can derive a minimal system of mutual forbearances, and hence the “minimum content of natural law”:

1. human vulnerability
2. approximate equality
3. limited resources
4. limited altruism
5. limited understanding and strength of will

It has to be said that Hart’s second and fifth qualities don’t entirely seem to belong. The last is introduced in the context of the need for sanctions to back up voluntary compliance with a code of forbearances, rather than the need for the code itself. The second seems more to be a background condition of any human society. Hart refers to the impossibility of imposing a generally-accepted framework of rules on individuals of vastly differing capacities; in the absence of general rough equality, in other words, the 20-foot noblemen (or the mutant superheroes) could lord it over all of us and no law of ours could bind them. This seems a counterfactual too far, and Hart seems to have brought it in mainly in reference to international law; he argues that the absence of this condition is precisely the key problem in that setting.

If we remove these, we are left with a core list of three facts of life:

1. human vulnerability
2. limited resources
3. limited altruism

These three can all be seen as vulnerabilities – vulnerability to direct physical harm; to material deprivation (ultimately, to hunger and cold); and to social abandonment and neglect. The three vulnerabilities have the interesting quality that the removal of any one of them would address the other two. (Try it: for the third one, “imagine all the people sharing all the world”. Also, “imagine there’s no scarcity” and “imagine we’re all solar-powered land crabs”. Some would say that John Lennon missed a trick there.)

There are certain adverse outcomes to which we are all vulnerable, in any imaginable human society, and which – crucially – we can all bring about in others: anyone can kill or be killed, steal or be stolen from, abandon or be abandoned. Hence a certain minimum, presumptively universal, content to the law, which can without too many problems be called natural. (It might seem that deprivation of human kindness – abandonment by one person of another – is considerably less serious than robbery or violence. But consider that, in most cases where one adult can be said to abandon another, it will be unclear who has deprived whom of kindness. Ideas of abandonment come into play – and into the realm of the law – where one party is need of care and/or the other has a duty of care.)

Substantive natural law, then, gives a minimum content to positive law. It has a considerable degree of overlap with the precepts of morality, but it derives – as does morality – from the facts of human existence. Substantive natural law is the minimum framework of mutual forbearances required to regulate society in response to the irreducible fact of human vulnerability (physical, material and social).

It’s the Law

Procedural natural law is introduced by Hart in his 1958 paper as follows:

If we attach to a legal system the minimum meaning that it must consist of general rules … this meaning connotes the principle of treating like cases alike, though the criteria of when cases are alike will be, so far, only the general elements specified in the rules. It is, however, true that one essential element of the concept of justice is the principle of treating like cases alike. This is justice in the administration of the law, not justice of the law. So 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. Natural procedural justice consists therefore of those principles of objectivity and impartiality in the administration of the law which implement just this aspect of law and which are designed to ensure that rules are applied only to what are genuinely cases of the rule or at least to minimize the risks of inequalities in this sense.

In The Concept of Law, Hart develops this argument; he also includes a second argument in response to Fuller. (Fuller’s Anatomy of Law had not yet appeared at this stage, but Fuller had replied to Hart’s 1958 paper in the same year, arguing for an ‘inner morality of law’.)

We may say that [the idea of justice] consists of two parts: a uniform or constant feature, summarised in the precept “Treat like cases alike” and a shifting or varying criterion used in determining when, for any given purpose, cases are alike or different. … In certain cases, indeed, the resemblances and differences between human beings which are relevant for the criticism of legal arrangements as just or unjust are quite obvious. This is pre-eminently the case when we are concerned not with the justice or injustice of the law but of its application in particular cases. For here the relevant resemblances and differences between individuals, to which the person who administers the law must attend, are determined by the law itself. To say that the law against murder is justly applied is to say that it is impartially applied to all those and only those who are alike in having done what the law forbids; no prejudice or interest has deflected the administrator from treating them equally. Consistently with this, the procedural standards such as ‘audi alteram partem‘ ‘let no one be the judge in his own case’ are thought of as requirements of justice, and in England and America are often referred to as principles of Natural Justice. This is so because they are guarantees of impartiality or objectivity, designed to secure that the law is applied to all those and only to those who are alike in the relevant respect marked out by the law itself. The connection between this aspect of justice and the very notion of proceeding by rule is obviously very close. Indeed, it might be said that to apply a law justly to different cases is simply to take seriously the assertion that what is to be applied in different cases is the same general rule, without prejudice, interest or caprice. (pp. 160-1)

Further aspects of this minimum form of justice which might well be called ‘natural’ emerge if we study what is in fact involved in any method of social control … which consists primarily of general standards of conduct communicated to classes of persons, who are then expected to understand and conform to the rules without further official direction. If social control of this sort is to function, the rules must satisfy certain conditions: they must be intelligible and within the capacity of most to obey, and in general they must not be retrospective, though exceptionally they may be. This means that, for the most part, those who are eventually punished for breach of the rules will have had the ability and opportunity to obey. Plainly these features of control by rule are closely related to the requirements of justice which lawyers term principles of legality. Indeed one critic of positivism has seen in these aspects of control by rules, something amounting to a necessary connection between law and morality, and suggested that they be called ‘the inner morality of law’. Again, if this is what the necessary connection of law and morality means, we may accept it. It is unfortunately compatible with very great iniquity. (pp. 206-7)

For Hart in 1958, “natural procedural justice” consists in applying general rules objectively and impartially so as to treat like cases alike; this is “justice in the administration of the law”, which can be distinguished from “justice of the law”. For Hart in 1961, “to apply a law justly” is to apply the same rule to different cases without prejudice; however, he does not refer to this (in his own voice) as a principle of natural justice. He is also reluctant to extend the label of ‘natural’, or the term ‘morality’, to the ‘requirements of justice’ identified by Fuller as an ‘inner morality of law'; Hart prefers to characterise these as ‘features of control by rule’, requirements which are (‘in fact’) characteristic of any form of rule-based social control. However – in a seemingly nugatory concession which has the effect of making Hart’s position much stronger – Hart grants that Fuller’s ‘inner morality of law’ may be accepted as such, with the dismissive proviso that it appears to have no effect (moral or otherwise) on the outcomes produced by a legal system.

Friendly Fire

By 1961, then, Hart’s attachment to any idea of procedural natural justice was already qualified and reluctant. What, then, to make of the closing paragraph of the “Introduction” to 1983’s Essays in Jurisprudence and Philosophy?

I hope that in what is a second exchange of friendly polemics between myself and Fuller … I have not been unfair in my criticisms of his conception of an inner morality of law; but I see now largely as the result of Professor Lyon’s [sic] essay on Formal Justice that an argument similar to mine against Fuller might be used to show that my claim made in ["Positivism and the separation of law and morals"] and repeated in my Concept of Law that a minimal form of justice is inherent in the very notion of a general legal rule applied according to its tenor to all its instances is similarly mistaken. I am not sure that it is so, but I am clear that my claim requires considerable modification.

What indeed? We’ll find out next time, by way of a reading of David Lyons’s 1973 paper “On formal justice”.

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.

First, some links to the individual posts in this series.

1. Liberty or liberties?
2. Restricting liberty for liberty’s sake
3. Restricting liberty for harm’s sake
4. Choosing (more) liberty
5. The priority of liberty (same post as previous)

These five points can be boiled down to three key questions: the quantification of liberty; the joint possibility of individual liberties, posing the need for protective as well as permissive rights; and the human preference for liberty over (for example) material wealth or peace and quiet. On the first of these I’d say that Rawls carries the day, at least in the sense that Hart’s challenge prompted some valuable elaboration and clarification of his model. On the other two – which are inter-related, at least in Rawls’s presentation of them – I’m less sure. Rawls’s argument is airtight – and he appears to escape the charge of designing a model world for model citizens – but I’m not entirely convinced; I think Hart’s scepticism may be the X that marks the spot of quite a deep equivocation.

Quantifying liberty (posts 1 and 2)
The quantification of liberty appears to pose problems for Rawls in two respects: allocation and comparison. Suppose that the basic liberties can each be considered as contributing a quantum of fungible ‘Liberty-stuff’ to an overall total. In that case, there is no reason to take the actual list of basic liberties as definite, and it may be that a model which maximised the amount of Liberty-stuff allocated to all citizens would take us in directions that Rawls wouldn’t want to go (e.g. away from the institution of private property). If, on the other hand, the basic liberties are seen as individually and discretely valuable – because what they provide is not fungible Liberty-stuff but different and distinguishable types of freedom – then there seems to be no basis on which to strike a balance between them. This second possibility becomes more pressing if we consider Rawls’s dictum that a liberty should only be restricted for the sake of another liberty: if liberties are incommensurable, how can this be achieved except by random selection?

Rawls addressed both these points in the 1982 lectures by introducing the notion of ‘significance’, and in the process making it clearer that liberties function in his model as a means to an end. He wrote: “a liberty is more or less significant depending on whether it is more or less essentially involved in, or is a more or less necessary institutional means to protect, the full and informed and effective exercise of the moral powers” – these in turn being the powers to co-operate reasonably and to seek the good rationally. The basic liberties are liberties which are in fact conducive to reasonable co-operation and rational deliberation; the issue of comparison is resolved by considering the conditions created by the exercise of a particular liberty. The question of allocation (and fungibility) is not banished as easily; however, what Rawls can offer is a strong presumption that the basic liberties he lists do have at least some ‘significance’ in his sense, and the criteria with which an alternative list would need to be justified.

Compossibility (posts 3 and 4)
The two questions of whether restrictions on liberty could be justified in order to prevent harm, and of whether the choice of greater liberty for all would be rational, both turn on the Kantian question of the conditions for the joint possibility (or compossibility) of individual wills freely exercised. We know from experience that societies can function on the basis of a shifting balance of freedom and coercion; the question is whether there is a coherent and non-arbitrary solution to the problem, a framework of basic rights or liberties which will tend to produce social harmony out of the free independent actions of individuals. Hart’s comments suggest that, in his view, Rawls has assumed that his model has this virtue but has not demonstrated it.

Rawls’s answer to this point is, once again, to invoke the ‘significance’ of the basic liberties and their exercise. The point, in other words, is not to maximise liberty or liberties, but to create those conditions which are best created by the exercise of the basic liberties. It follows that a balance between the basic liberties, and hence the limitation of particular (less significant) liberties, is an integral part of the model. The basic liberties are both self-limiting and mutually limiting: to the extent that a liberty furthers “the full and informed and effective exercise of the moral powers”, to that extent it will tend to be protected over and against liberties whose exercise is less significant.

I find this argument convincing but unpersuasive; to put it another way, it seems to answer the question within its own framework but at the cost of making that framework less attractive – and, perhaps, distancing it from the world in which the question was asked. It may be significant that the question of harm is one on which (at least according to Hart’s account) Rawls is all over the place: starting from the presumption that a liberty should only be curtailed for the sake of another liberty, we can cover the idea of infringing liberty to protect from harm by invoking the liberty-reducing effects of harm, an association between citizenship and the exercise of liberties, and ideas of a duty of care to animals and the natural world, but it ends up looking like a bit of a hack. (The hackwork is mostly mine, but the gap it fills seems to have been left by Rawls.)

The preference for liberty (post 4)
Lastly, Hart poses two questions which can both be taken as calling into question Rawls’s assumption of a preference for liberty (as distinct from, say, material wealth or a quiet life). Hart suggests that Rawls has tacitly built his model society on a liberal model of active civic virtue, thereby resolving all such questions in favour of the – undeniably virtuous but potentially strenuous and unrewarding – pursuit of reasonable co-operation and rational deliberation.

The charge is made lightly but it is potentially devastating, reducing Rawls’s model to a utopian vision of how people would live if only they were good. Rawls rejects it, quietly but firmly; while he concedes that his presuppositions are liberal, they find expression not in the characters of the subjects populating his model but in their nature. Specifically, it is in their nature to work together with other individuals (and hence to value reciprocity and fairness) and to value some states of affairs more highly than others (and hence to value morality and logic). The only qualities Rawls reads into the subjects in his model are the capacities to co-operate reasonably, where ‘reasonableness’ includes a sense of justice, and to deliberate rationally, where ‘rationality’ includes an idea of the good. Everything else in the model follows – which is to say, everything in the model follows from some facts about people as they are.

Again, I find myself convinced but unpersuaded. One reason why I’ve harped on Rawls’s particular definitions of the ‘reasonable’ and ‘rational’ is that they’re easily overlooked, but make Rawls’s model much easier to understand if they’re taken into account – cf. Hart’s puzzlement over how the model would work on assumptions of ‘self-interested rationality’. But another reason is more critical. Certainly there are such things as co-operative reasonableness and rationality oriented towards an idea of the good, and practices informed by them; an account of society based exclusively on individual self-interest would be not so much impoverished as false. But bracketing out means/end rationality and reasonableness in the pursuit of one’s own desires seems like an equal and opposite distortion. The question is not whether Rawls’s moral powers are a human reality but whether they could ever do the work he wants them to – and saying that they could do if their exercise were given priority over less moral pursuits is begging the question.

I sense that Hart saw a deep equivocation here, between a model which could exist (in the sense that it rests on valid assumptions about human nature) and one which could exist (in the sense that the model itself represents an imaginable society). It may be that Rawls only saw himself as developing the first of these; however, to the extent that such an abstract standard can be a driver for reforms to the society we have, it must surely be possible to envisage reforms which would represent steps towards it, even if they were fated never to reach it. And, if Rawls’s model is supposed to represent something approachable (even if not attainable), we’re back to the original question: why are his subjects so nice? The answer seems obvious – it’s because they’ve chosen to prioritise conditions favouring the exercise of the two moral powers – but this only defers the question: in the light of most of human history to date, why have they chosen to do that? If we’re going to have a society founded on a complex balance of basic liberties – and it sounds like a good idea – where are we going to get the people who want one? And, if we haven’t got those people, maybe we should be working on something for the people we have got – a theory of justice as between flawed, lazy, selfish and intermittently deceitful people, for example.

More on Hart and his 1973 paper on John Rawls’s A Theory of Justice. Hart put forward five queries, of which I’ve now covered the first two:

1. Liberty or liberties?
2. Restricting liberty for liberty’s sake In the second post in this series, we saw that the idea of restricting one liberty for the sake of another raises commensurability issues. In his reply to Hart, Rawls addressed these – partially addressing Hart’s first criticism in the process – by introducing the metric of ‘significance’.
3. Restricting liberty for harm’s sake Rawls appears not to grant that liberties should sometimes be limited for harm-related as well as liberty-related reasons. Is this sustainable?
4. Choosing (more) liberty Rawls argues that subjects in the original position would, in their own interest, tend to choose more rather than less extensive liberties. Given the potential adverse effects of liberties extended to the whole of society, is this valid?
5. The priority of liberty Following on from the previous point: Rawls appears to believe that, all else being equal, subjects in the original position would choose a quantum of liberty over a quantum of material benefit. Is this an unstated presupposition on Rawls’ part? How is our perception of his model affected if this is granted?

Now for #3.

Restricting liberty for harm’s sake
Although (as we have seen) Rawls later modified this position, in the first edition of A Theory of Justice he argued that a basic liberty could only legitimately be curtailed for the sake of another basic liberty – or rather, for the sake of bringing about a more extensive and/or more equal distribution of basic liberties. As Hart pointed out, this implies that it is possible to weigh up basic liberties against one another, a problematic position which I covered in the previous post. Hart also criticises this position on the grounds of what it does not include: specifically, it makes no provision for restrictions to basic liberties for the sake of preventing harm, or (more broadly) in order to restrict or regulate behaviour which causes loss, pain or suffering to others.

This is a fairly big deal. The possibility of living together in freedom and under law has been a focus of political philosophy since Kant. If we say, broadly following Kant, that the law should protect each individual’s right to pursue his own interests without prejudice to anyone else’s right to pursue theirs, we run into the problem that interests conflict. I stress ‘conflict’ as distinct from ‘compete': competing interests, as between two companies in the same line of business – where the parties have a rivalrous interest in the same resource – pose no problem in theory. Conflicting interests arise when the parties define the situation differently, as between a business (with a legitimate interest in maximising profit) and a trade union (with a legitimate interest in protecting its members). The problems this situation poses are, fairly clearly, political problems, with no easy answer on the philosophical plane. We could lower our sights somewhat and go with Mill, arguing for complete freedom of action up to the point where another person is harmed – a position which has entered the language in the formulation “The right for me to swing my fists ends where your face begins“. But, despite its surface plausibility, this gets us no further forward, as it depends entirely on interpretation of the word ‘harm’. Define ‘harm’ as actual physical harm and many undesirable activities would be permitted, from harassment to the sale of contaminated food. Define ‘harm’ as the setting back of interests and we’re back with Kant. Should trade unions be banned, as their tendency is to raise members’ wages, lower the amount available to pay dividends and hence harm (set back the interests of) shareholders? Should shareholders be expropriated, on the grounds that their extraction of value from businesses tends to lower the amount available to pay wages and hence harm (set back the interests of) workers? I’m not familiar enough with Mill’s work to say how he would have answered these questions, but I’m fairly confident that it would have been a political answer, rather than one dictated by the terms of the question.

There is no theoretical or practical difficulty encroaching on liberties so as to prevent harm; societies do it all the time. However, justifying those restrictions in a coherent and generalisable way has proved to be a serious challenge for political philosophy. Rawls, oddly, doesn’t seem to say much about it, other than to rule it out on principle – because a liberty should only be curtailed for the sake of a liberty (of greater significance). Can this be accepted, and if so how? Hart canvasses two – unsatisfactory – solutions, to which I’ll add another couple.

1. They’re all liberties! Perhaps, when we curtail a liberty so as to prevent harm, we are actually doing so for the sake of another liberty. Steal my car and my effective liberty to exercise property rights is curtailed; hit me hard enough and my effective liberty to participate in society becomes moot, at least temporarily. Hit me when doing something you want to deter (voting, say) and the fear of future harm may have even more liberty-impairing effects. In other words, might harm be derivative of impairment of liberty? The argument is superficially attractive but ultimately implausible; as Hart comments, the actual valuation of injury and harm is clearly independent of any consequential liberty-impairing effects they may have. I would add that the argument is also covertly circular, inasmuch as liberties would be meaningless if we were not vulnerable to harm from one another. If property were inalienable (or all property were held in common), there would be no need for a right to property; if incitement could never result in injury, there would be no need to restrict speech and hence no need for a right to free speech.
2. …unless they’re duties. In addition to the basic liberties, Rawls suggests that subjects behind the veil of ignorance would take on certain natural or moral duties, judging that the benefit to all from doing so outweighed the costs; examples include the duty to aid those in need and the duty not to cause unnecessary suffering. Clearly, the existence of duties would have some liberty-curtailing effects. Arguably this is unsatisfactory on grounds of theoretical parsimony; it certainly suggests that the ‘only curtail liberties for the sake of liberties’ formulation might need to be amended. In any case, the range of natural duties seems far too narrow to cover all those cases in which unrestricted liberties could foreseeably cause avoidable harm, from invasion of privacy to exhibitionism.
3. We’re all citizens here. Perhaps Rawls overlooked the prevention of harm because, within his scheme, harm wasn’t relevant. He was designing a model to be inhabited by full citizens in effective possession of their liberties; a citizen convalescing in hospital, or immobilised by car theft, is not effectively a full citizen, and as such is of no relevance to the model. All we can do is hold their place in the model open for when they’re ready to occupy it again. The point could even be generalised to cover harms which do not directly attack liberties: a citizen recovering from a head wound may still be able to take part in political deliberations, but her mind won’t really be on the task in hand. This is probably the weakest solution of all: the point is, of course, to prevent one’s citizens from suffering these interpersonal harms in the first place, and one could well argue that a scheme of liberties which doesn’t do this job isn’t worth the candle – particularly if one were sitting in a planning meeting with a bandaged head at the time.
4. …unless we’re minors. In the context of the second point here, Hart notes that Rawls acknowledges the existence of duties owed to the non-human world, “which are outside the scope of a theory of what is owed to a rational individual”. Building on this suggestion, we can imagine a modified combination of the first and third solutions, which would tie liberties to citizenship. In this model, criminal harms would be conceived as attacks on liberty-enjoying citizens, thus meeting Hart’s objection to the first solution – that our valuation of harms does not depend on their consequential liberty-impairing effects. Conversely, those who are routinely deprived of liberties – minors, prisoners, hospital patients etc – would be seen as deprived of citizenship because unable (for the moment) to exercise it effectively. Restrictions on liberties implemented in order to protect free citizens from harm, or to restrain those not able to exercise the liberties of citizenship effectively, could then be defined as restrictions on liberties for the sake of liberties, as Rawls would (presumably) wish. This is ingenious (if I say so myself) but unsatisfactory. Although it accounts for the assimilation of harms to attacks on liberties, it does so at too high a cost: the implication that prisoners, hospital patients et al are non-citizens is troubling, and the further implication that harms to those people are of less account is unacceptable. This model also fails to account for (among others) students, employees and the patients of psychoanalysts, all of whom are free citizens who are routinely subjected to un-lawlike regulation of their movements and activities.

It’s all rather unsatisfactory, and I’m afraid that’s how I’m going to have to leave it – Rawls’s 1982 lectures are as far as I’m going for further reading at the moment, and he makes no reference in them either to original-position duties or to the prevention of harm. Fortunately he’s a bit more forthcoming on Hart’s next question.

In his paper on A Theory of Justice, Hart put forward five queries, all of which Rawls (some years later) responded to directly; I’ll incorporate some comment on Rawls’s responses as we go along.

Hart’s questions can be summarised under the following headings.

1. Liberty or liberties? Is Rawls talking about a single quality of liberty which takes multiple forms, or about multiple discrete liberties? If it’s the latter, what are they and where do they come from – and what implications does Rawls’s selection of specific liberties have for his model?
2. Restricting liberty for liberty’s sake Rawls argues that the only justification for limiting a liberty is an overall extension of liberties. What issues does this raise in terms of resolving potential conflicts between liberties?
3. Restricting liberty for harm’s sake Rawls appears not to grant that liberties should sometimes be limited for harm-related as well as liberty-related reasons. Is this sustainable?
4. Choosing (more) liberty Rawls argues that subjects in the original position would, in their own interest, tend to choose more rather than less extensive liberties. Given the potential adverse effects of liberties extended to the whole of society, is this valid?
5. The priority of liberty Following on from the previous point: Rawls appears to believe that, all else being equal, subjects in the original position would choose a quantum of liberty over a quantum of material benefit. Is this an unstated presupposition on Rawls’ part? How is our perception of his model affected if this is granted?

For reasons of space and time, in this post I’ll only address the first of these. (Whatever else you can say about Hart, his writing is extraordinarily good to think with – particularly when he’s got something as substantial as A Theory of Justice to chew on.)

Liberty or liberties?
This may be little more than a textual quibble. Rawls’s response (in “The Basic Liberties and Their Priority”) can certainly be read in this way:

the equal basic liberties in the first principle of justice are specified by a list … No priority is assigned to liberty as such, as if the exercise of something called “liberty” has a pre-eminent value and is the main if not the sole end of political and social justice. … Hart noted, however, that in A Theory of Justice I sometimes used arguments and phrases which suggest that the priority of liberty as such is meant; although, as he saw, this is not the correct interpretation

Rawls goes on to characterise the ‘basic liberties’ as (in crude terms) means to an end rather than ends in themselves – the point is not to achieve an equal distribution of liberties for its own sake but to establish the conditions of possibility for the collective construction of a just society and for collective deliberation on ideas of the good. Rawls proposes an equal distribution of basic liberties not as an end state, but because it is among those conditions of possibility – or, perhaps, a necessary condition for the eventual development of those conditions of possibility. We are a long way from a utilitarian framework of the maximisation of capital-L Liberty, in other words.

I’m not sure that this disposes of Hart’s comments, though. The question concerns the fungibility of a generic liberty (I’ll refer to this as Liberty) – or, to put it another way, the commensurability of distinct liberties. Declining to specify an optimum scheme of basic liberties, Rawls argues that multiple alternative configurations or schemes of liberties may be equally valid. However, this implies that a restriction of one liberty may be compensated by an expansion of another, which as a minimum implies that different combinations of liberties may have the same beneficial effect. The model does not require simple commensurability between liberties (a bit more freedom from arbitrary arrest is worth a bit less freedom of expression). However, it does seem to imply that Liberty is somewhere in the background, being enhanced by the expansion of one liberty and reduced by the restriction of another. At least, it suggests an underlying resource of interchangeable Liberty-stuff, such that – given satisfactory levels of two different liberties – the overall settlement could only be improved by increasing one without lessening the other. If Liberty is seen as a single state or goal, approached in different ways through the instantiation of different liberties (each producing its own contribution of Liberty-stuff), it follows that some individual liberties may be much less effective in its realisation than others; some may even be expendable. This is the case even if we know what we mean by Liberty.

A realised state of Liberty is (more or less by definition) unknown to us, however. It’s safe to assume that liberties we have not considered may deserve a much more prominent place in our thought and action. In this context, Hart has an interesting passage on Rawls’s comments on lifestyle-related freedoms – in sexual conduct, drug and alcohol use, and so on – which he does not elevate to the level of a basic liberty. This may be explained by referring back to the civic focus of the basic liberties – seen as preconditions for the development of a just community rather than as individual liberties tout court – although the idea that sexual identity is irrelevant to public deliberation was not universal even in 1973.

Conversely, some of what we now consider to be liberties may well be irrelevant to the achievement of Liberty, or even counter-productive. Citing Rawls’s reference to a ‘principle of greatest equal liberty’, Hart refers to a Kantian model of universal freedom under a common law advanced by Herbert Spencer. Spencer’s critics, including Henry Sidgwick, pointed out that the institution of private property represents a stumbling-block for any pure theory of freedom: in the simplest form it represents a hard limit to individual freedom, expanding the freedom of the owner and reducing that of everyone else. (In less simple forms, private property brings with it inherited wealth, the accumulation of capital and wage labour, all of which bring their own forms of unfreedom.) Spencer’s response to this critique was to admit its force and revise his model, arguing that the greatest equal liberty would only be attainable when all property was held in common.

The value of the right to property in Rawls’s model is circumscribed, inasmuch as there is no basic right to individual ownership of the means of production: a ‘Rawlsian’ society may have an entirely state-owned economy. However, Rawls resists Spencer’s move, including the right to own personal property in his brief list of basic liberties. This is certainly justified on pragmatic (and on Pragmatic) grounds, making his model seem less utopian – and hence more useful to think with – than Spencer’s. Whether the liberty to own property (or other discrete liberties) can be justified on other grounds is less clear. As I have argued, although Rawls treats his basic liberties as discrete and distinct, to the extent that they can be balanced against one another there must be a Liberty behind the curtain which they jointly make it possible to approach – or at least a Liberty-stuff which they each in their different ways produce.

If this is the case, the basic liberties are not fundamental, but different aspects or facets of the production of fungible Liberty-stuff, or of the approach to an ineffable Liberty. And if that’s the case, clearly Rawls’s list can’t be taken as definitive; the possibility that it might need to be lengthened, and – more disruptively – the possibility that it might be appropriate to trade down one or more of our current list altogether, can’t be avoided. Might an equal distribution of basic liberties be achievable on the basis of a Fourierian phalanstery, designed to guarantee freedom from want, idleness and anomie, but not too great for political and personal liberties? And if not, why not? To be more precise: given that we’re objecting on philosophical rather than political grounds – political disagreements start some way down the road and should be expressible within the framework we’re developing at this stage – the contention must be that the map of small-l liberties not only should not but cannot be redrawn to that extent. But how can we justify that belief without recourse to political arguments?

One superficially attractive argument justifying a Rawlsian list of basic liberties – and countering ‘fungible Liberty-stuff’ and ‘ineffable Liberty’ arguments – runs like this. Let’s suppose that, somewhere out there in concept-space, there is such a thing as Liberty, although we don’t know what it looks like. However, we do have an idea what freedom of the press, freedom of assembly, freedom of conscience and the rest of them look like. Moreover, we know they’re all valuable: we may be open to the striking of different balances between different liberties, even (perhaps) to the point of suspending one so as to preserve another, but we know that any solution which dispenses with a particular liberty altogether is a bad solution. (At least, it’s a solution with at least one bad feature – that one – and we want to avoid bad features in our solutions.) The difference between ‘balancing discrete liberties’ and ‘fungible Liberty-stuff’ models, in other words, is that the former provides a backstop, or several backstops: these liberties are things we know to be valuable, so we know that no solution in which any one of them is traded off entirely can be satisfactory.

The trouble with this argument is that it’s circular, and rather a tight circle at that. (Why shouldn’t we think in terms of fungible Liberty-stuff? Bad outcomes would result. Why would the outcomes be bad? Because there is no fungible Liberty-stuff.) However, it points the way to an argument or group of arguments which, while they lack the logical closure of this one, may be more persuasive. In its weakest form, the argument draws attention to the fact that Rawls uses a list of basic liberties at a fairly early stage in the development of the argument of A Theory of Justice - not to mention an early stage in the development of the ideal framework of social relations proposed in the book. Perhaps the choice of a list of basic liberties, and scepticism about that choice, are not positions which can usefully be counterposed in argument but simply alternative starting-points, to be judged primarily by their fruits. In other words, to the extent that Rawls makes it work, it works, and challenges at this foundational level are little more than doodling in the margin – or signs that the person putting the challenge may be reading the wrong book (“Aeschylus is so sexist!” – English student, Cambridge (overheard)).

A stronger version of the same argument would return to the first part of the circular argument set out above and plant its flag there. We do know what’s meant by each of Rawls’s basic liberties and we do know that each of them is valuable and therefore worth preserving, all else being equal. By contrast, nobody has ever seen Liberty, and we’ve got no compelling reason to believe in the existence of Liberty-stuff. Direct experience tells us about basic liberties; only human ingenuity tells us about Liberty, and we know how unreliable human ingenuity can be. The alternative to a conservative reliance on the liberties we know, on this argument, is not philosophically-grounded radicalism but reckless concept-mongering. To put it another way, trading a known basic liberty for a speculative increase in Liberty-stuff would be an irresponsible gamble. (Another criticism of Rawls, which I’ll come to in the next post, is that he’s not conservative enough in this respect.)

At its strongest, this argument would take direct aim at the shrouded numinosity of Liberty, asking whether we really have any reason to think there’s anything there. Perhaps the basic liberties have no more in common than a sandwich without mustard and a dog without a lead. Or rather (cutting off a potentially interesting but irrelevant line of questioning), perhaps we have no need to think of the basic liberties as having any properties in common in order to group them together. Perhaps, rather than working on the assumption that there is something positive and theoretically realisable called Liberty, we should think of the basic liberties as a group of absences which we hold to be significant and valuable – just as nitrogen and carbon dioxide share the common property of not being oxygen, rather than that of being saturated with phlogiston. This would, of course, leave open the question of whether Rawls’s list is the right list of basic liberties, but it would make the task of amending the list much more challenging.

Hart’s second criticism… but enough! Not sure how long this series is going to be; all I can say is that the next post will definitely address at least one of the remaining four arguments.