Ted Honderich's book,
Punishment: The Supposed Justifications Revisited has as the 7th of 8 chapters the following look at certain theories of punishment by the state, theories now of the commonest kind. They are not traditional or refurbished retribution theories, to the effect that punishment is justified simply as retribution, desert, proportion or the like. They are not traditional prevention theories, utilitarian or the like, that punishment is justified just by the fact that it reduces offences. They are not reform or rehabilitation theories, that punishment is justified by morally improving offenders so that they do not offend again. Rather, mixed or compromise theories seek to escape objections to these theories by depending on several of them taken together. Three prominent mixed theories are considered here, those of H. L. A. Hart, Robert Nozick, and Antony Duff. The book's inquiry is completed in its final chapter, which is not more of the same. There is a note on this new book's relation to its predecessor, Punishment: The Supposed Justifications at the end of this excerpt.


     There was a lot wrong with traditional backward-looking theories of the justification of punishment with which we began this inquiry, mainly the retribution theories to the effect that we are obliged rather than permitted to punish offenders because they deserve it. One such theory confused what was needed, a moral reason for punishing a man or woman, with a legal reason. Several theories begged the question by in effect taking the argument that punishment is right because it is deserved to be the argument that punishment is right because it is right. Another theory, about the intrinsic good of the suffering of the guilty, was open to the reply among others that different and opposed goods, and bads, can be postulated as easily.

    No plain sense of the factual kind could be attached to the theory that a man's culpability in an offence has a commensurable penalty that it equals. There was no gain in talk of an offender's forfeited rights, or preference scales, and none in the elaboration of a system of offences and penalties without any explanation of the supposed justifying relation in it. For several reasons we did not get further forward with attempted justifications of punishment in terms of annulment of the offence and  an offender's right of consistency to his punishment. Nor was there much gain in backward-looking theories of a non-retributive kind, those in terms of a hypothetical rational contract, the better idea of consent to loss of immunity to punishment, and satisfactions in putting down burdens.

    To the various difficulties with these theories had to be added another objection, mentioned at several points. It comes to this, that surely if punishment is justified this must be because it does some good, indeed has some good effect or effects, where that is definitely not only the proposition that it is related in some way to a past offence. A goodly number of jurisprudents, lawyers, and indeed a philosopher or two, have not shown themselves fully aware of the logical or conceptual failings and indeed the disasters of the traditional theories just recalled. But very many of them have been quick to recognize and assert that not enough can be said for punishment if all that can be said for it is that it is deserved, where that means only that it stands in some seemingly factitious relation to a past event, the offender's offence. It was indeed a recommendation of the retribution theories that they prohibited the punishment of those who did not deserve it, or rather the victimization of them. But this fact of the theories being against victimization could not at all make good their weak or worse reason for punishment.

    We ourselves did find something justificatory to be said for punishment in retribution theories by abandoning a kind of respectability, indeed piety, not to mention tolerance for obscurity and near-nonsense, and instead understanding the theories in terms of grievance-satisfaction. The retribution theories were reduced to the proposition that punishment is justified because it can satisfy the grievance-desires brought into being by an offence, and do no more and no less than that.

    One thing to be said against this this uniquely clear reason for punishment, a reason resting on a real effect, an actual if all too human good, was that it could not be a sufficient reason for punishment. It could not be that the denial of great goods to offenders could be justified by the smaller thing of affording grievance-satisfactions to others. That this could not be justified was not then explained in terms of a moral principle or an articulated morality, which indeed was not needed. As might have been remarked, grievance-satisfactions do not persist in the way of the distress or worse of 20 years in prison.

    Something else to be said of the uniquely clear retributive reason for punishment is that it did not convince respectable and decorous philosophers and other theorists of punishment. They have not been willing to suppose the tradition of retribution can be a matter of no more than a lower part of human nature. They have been, to my mind, a little too impressed, unreflectively impressed, by both the pomp and real value of the law. Many have continued to be convinced both that at long last there is a better retribution theory, their own particular one, and that it has a certain role, of which more in a moment, in combination with something else.

    The utilitarian prevention theory, the original one among theories seeking to justify punishment by actual good effects or consequences, has an uncertain but perhaps sufficient basis in a certain proposition of fact. That is the proposition that it does somehow prevent sufficient offences to be justified. The theory had the recommendation, obviously, of providing a substantial reason for punishment. However, it too had its weakness. Despite attempts of one kind and another to sustain it, it failed for a particular reason. This was not that it justified victimizations that already occur in the law, and are taken as defensible, but that it justified morally intolerable victimizations. There was the same sort of difficulty with justifications of punishment in terms of reform and rehabilitation, whatever else could also be said against them. In short, the utilitarian and reform theories provided a reason for punishment but not a reason against victimization.

    In such a circumstance, inevitably, something suggested itself to philosophers and others. This was that the general strength of retribution theories, a prohibition against doing something, could be combined with the general strength of some preventive theory -- recommending punishment on the basis of some actual good to be realized. The combination, evidently, would not have the weaknesses when separate of the things conjoined. The good would not necessarily be the good of the utilitarian theory, the greatest possible total of satisfaction, which might not fit well with the prohibition on giving a man more or less than he deserved. The good might be the prevention of offences otherwise conceived. It might include the effect of the reformation of offenders. Thus were born the mixed or compromising theories of punishment, necessary or unnecessary, successful or not. What they come to in general, whatever is to be said of them, is that punishment is justified because it is or may be both deserved by the offender and also preventive of offences in the future.

    Did the mixed theories owe their emergence to another idea or feeling, by the way? That would be the fact that they are different from traditional retribution theories in not being so vulnerable to the truth of determinism. Certainly there has long been a kind of apprehension that it is true, pretty independent of philosophy. It has been on hand, if in the background rather than the foreground, to trouble the advocates of retribution. If the mixed theories cannot escape the worry, since they have to do partly with desert, their parts having to do with prevention are reassuring. If the worst comes to the worst, something will be left of them. But this is speculation, not very serious. There is another introductory matter seemingly of more importance.

    Mixed theories have commonly been introduced in a certain way, favoured by the distinguished jurisprudent H. L. A. Hart and others before him, including the philosopher John Rawls.1 It will be worth looking at this way briefly for several reasons. One is that it sometimes appears to be intended as an argument, or something like an argument, for the conclusion that punishment must be justified by several principles or reasons. Another is that it is at least regarded as a preferred or indeed essential mode of procedure. Our own procedure has been, and mostly will be, to consider the single and inclusive question, 'What, if anything, can morally justify the practice of punishment?' A number of philosophers and legal theorists still regard this as an error, one at least likely to have considerable consequences.

    It was Hart's view, in some accord with the linguistic or analytic philosophy of his day in Oxford, that in our inherited ways of talking about punishment we persistently over-simplify separate issues. What we must notice, if we are to avoid confusion, is that there is more than one question to be considered.2 There are separate questions. This is so because there is the institution or practice of punishment itself and then there are its several different features. Only certain persons are pun-ished, and only certain penalties are imposed. The institution therefore raises several quite separate questions of justification.

    There is the overall question, 'What justifies the general prac-tice of punishment?' or 'What is the good of maintaining this institution?' This can be named the question of the general justifying aim of the institution. Secondly, there is the question of liability. It is this: 'To whom may punishment be applied?' Finally, there is the question of amount. 'How severely may we punish?' We are invited to see that there are these three distinct questions -- and thus also to see that different answers are in place with respect to them.

    Surely, it is said, the general justifying aim of punishment must be the prevention of crime. Who can really disagree? As for the second question, the question of who may be punished, surely the answer is that only those who deserve it or are guilty may be. And, with respect to the third question, a partly retributivist answer is surely in place. A man should not be given a larger penalty than he deserves. In some such way as this we are led toward the view that punishment is justified only by several different considerations -- in short that it serves some preventive end with respect to crime but is limited in doing so by considerations of retribution. What justifies punishment in part is what was earlier called negative rather than positive retribution. To see that there are separate questions about punishment is to see that there are separate and diffrent answers?

    Is this so? Consider the second question, 'To whom may punishment be applied?' It is a long step in the direction of irrelevance to see this as peculiarly a question for a judge.3 If one answers as if one were a judge, of course, there is more than a temptation to give the answer that punishment may be applied only to those found to have broken the law. This may be to regard the question in a way which makes it irrelevant to our inquiry. A judge may indeed be said to be justified in singling out for punishment those found to have offended. This is to say, ordinarily, that his action is in accord with his powers and duties, as defined by law. But we are not inquiring into legality, into legal powers and obligations. We wish to know what can be said in moral justification of what is done by judges in accordance with their legal powers and obligations -- and maybe personal moral obligations.

    We need to understand the question 'To whom may punishment be applied?' in another way -- a way such that it might have the moral answer toward which we are being urged. It can indeed be taken as another question, a little more explicit. 'What, if anything, morally justifies our punishing only those found to have broken the law?' Does the fact that we can or should ask this separate question -- that fact by itself -- give us reason to answer it in terms of desert? This is the crux of the matter.

    The answer is plain enough. Certainly it is the part or side of punishment, spoken of as liability, to which our attention has been directed, that has led many people to retributivism. But that the separate question can be asked does not guarantee or even make probable such an answer. It remains perfectly possible, given only the fact of the separate question, to side with Bentham and declare that the answer to the question about who is to be punished is simply that those are to be punished whose punishment will prevent future offences.

    That is not all that comes to mind. We are contemplating the question 'What if anything morally justifies our punishing only those found to have broken the law?' Could the fact that it is only these individuals who may be said to deserve punishment provide a sufficient answer to the question? Obviously not, although exactly that is what is suggested, since desert is not a sufficient justification for punishing any indivi-dual. We know it. That desert is a sufficent condition is the plain retribution theory, which has been rejected. It is rejected, indeed, by the philosophers whose compromise views we are considering. This introduction to mixed theories is not encouraging.

    Similar remarks need making in connection with similar results, to the third question, 'How severely may we punish?' The question does not commit us to an answer, or even need to incline us towards one.

    As for the first question, about the general justifying aim of punishment, it is ambiguous in a way peculiar to itself. It may be taken to be about just the aim of punishment, which is certainly to be distinguished from its justification. It is sometimes true that to mention the aim of a particular line of action is to mention an accepted justification. On other occasions this is not so. Our aim in a war may be to occupy some land, but our justification may be some past fact or supposed past fact. It does seem true to say that one aim of punishment is prevention. A goal we want to achieve is a state of affairs where there are fewer offences committed. But it is clear that if the question about the practice or institution of punishment is to specify a part of our present inquiry, which is a moral one, it clearly cannot be taken to be about only the aim or goal of the practice.

    If we are asking what adequate moral justification can be given of the practice, it is perhaps natural to think first of the prevention of offences. But that we can ask the separate question goes no way towards establishing that answer. That fact, rightly, would not have delayed Kant for a moment in giving his retribution theory of punishment. The situation is the same as with the first two separate questions.

    In any case, it is on reflection quite mistaken to think that an answer can be given to the question of the moral justification of the practice in terms of, say, the utilitarian prevention of offences alone. This is exactly what has been established by the argument about victimization. Some such argument is accepted by exactly those who espouse compromise theories -- those now espousing them to us right now. How then can they suppose that the moral justification of the practice can be preven-tion alone when rightly or wrongly they also require another different reason, in terms of desert? That this second reason may be related to par-ticular features of the practice, the convicting and sentencing procedures, is im-material. That a reason is necessary for a part of something entails that it is necessary for that thing. Something the same may be said, not that it matters much, for the reason having to do with prevention. Prevention has to do with the particular fact or side of punishment that it is public. It certainly does not follow that it is not necessary to mention it in defending the institutition.

    It is not at all clear, then, why anyone who feels that a compro-mise or two-value justification is necessary should also think the practice as a whole justified by only one of them. There is some explanation, no doubt, in the confusion of aim with justification. This confusion may be made easier, incidentally, by an assump-tion similar to one noticed above. That is, it is sometimes supposed that the question about the justification of the practice is peculiarly one for legislators. Legislators are concerned with a society's larger intentions, including the prevention of offences, although not exclusively so concerned. Even if they were concerned solely with aims, nothing would follow about the answer to our question.

    Does it need to be added that we can of course express a moral question, in some sense about the institution or practice, whose answer will be in terms of prevention alone? It is: What is to be said for the practice of punishment other than that it is governed in certain ways by limitations on who is to be punished? That the answer to that question may be prevention alone of course does not give us the proposition that the practice is justified by prevention alone. There is the same uselessness in similar questions that can be formed about who is to be punished and how much.

    So, the mere fact that we can ask separate questions about punishment is consistent with all the answers being solely in terms of desert, or solely in terms of utilitarian prevention -- or, as is obvious, and as we shall be contemplating, solely in terms of a certain kind of consequences of punishment. There is no reason for thinking that some separate-ques-tions procedure inevitably or logically leads one to a compromise theory. The existence of separate questions does not constitute any argument for the conclusion that punishment must be justified by several principles.

    Is it nevertheless true that we are likely to fall into confusion if we do not ask these questions, or some other set of separate questions? Must this be the best procedure? This seems to be partly the claim that philosophers who have given single-reason answers about punishment, inade-quate ones, would have done otherwise if they had asked separate questions. This too is unconvincing. In fact we here have a further and yet simpler objection to what is urged upon us. utilitarians did not ignore the fact that there are prohibitions with respect to who can be punished and to what extent. In effect they isolated the questions about the basis of these prohibitions and gave answers to them. Hart discusses Bentham's.4 One can say of the traditional retributivists, similarly, that they were not unaware of the claim that the practice of punishment does have certain preventive effects. Following Kant, they commonly introduced this fact into discussion in order to deny its moral relevance.

    Are we likely to fall into confusion if we do not ask separate questions? We shall not fall into the confusion of thinking that a single-reason answer in terms of retribution or utilitarian prevention to the single and inclusive moral question about punishment will do. Why should we? Nor does it seem likely that we will risk other confusions. Certainly we must keep in mind the different features of the prac-tice of punishment -- and not only those having to do with certain prohibitions. We must indeed consider what contribution may be made to the justification of the practice by particular features. But this we can do without a mechanical separation of questions. One question will do.

    Our fundamental and final concern, obviously, must be with what justification can be given for the practice of punishment taken as a whole. This is a satisfactory formulation of the question which all theories of punishment try to answer. It is precisely the question, of course, which any theory must end by trying to answer. Needless to say, once you think of the matter, it is also precisely the question finally answered by all compromise theorists without exception, however mixed or compromising their answer. Hart, for example, was not giving a partial or incomplete answer to the question of what justifies punishment or answers to questions that do not come together into a summation, into what every relevant theorist of punishment provides, a summary verdict on that thing.


    There are a lot of mixed or compromising theories of punishment. There are too many for surveying in this book. Is there another reason why we need not exhaust ourselves making our way through them? Whether or not that is so, let us look at three of them. If they are indeed mixed theories, all have or claim to have a mainly or importantly retributive character, which fact is reflected in their names. Each of them, further, at least attempts to add to the 15 understandings of desert-claims that we have, in fact 15 retribution theories.

    The first, the work of Hart, follows on from the assumptions and arguments about separate questions that we have been considering. Those assumptions and arguments, however, are not actually premises for the theory. It does not depend on them. They do not determine whether it stands or falls. It may be the right answer. Certainly it has been an influential theory, not only among jurisprudents and lawyers, and it remains so. It serves as our example of the most expected and obvious compromise theoires, those that somehow justify and limit punishment to what is both preventive and deserved.

    The practice of punishment, we hear again, involves a prohibition on punishing most individuals. We hear in effect that in general or for the most part we punish only those individuals who have intentionally or negligently broken the law. Admittedly, as you will remember from what was said in connection with the utilitarian theory of punishment, there are already within the law many cases of what are generally regarded, or regarded by many, as being justified victimizations. There are convictions under statutes of strict and vicarious liability. Impositions of what were spoken of as exemplary penalties, which go beyond what may be said to be deserved, are also defended.

    Still, the practice of punishment does involve a wide prohibition against penalizing most individuals. In general, we in effect hear first from Hart, we punish only those who could have kept to the law when in fact they broke it. Also, he remarks, this belief that men could have done otherwise than they did is no less than the keystone of retributive theories of punishment.5 Subsequently we hear that for the most part, the far greater part, our punishments are of persons who are 'morally guilty'. And further, they are 'those who have broken the law -- and voluntarily broken it.'6

    There is more to be said, however, than that punishment in general is imposed on those who could have done otherwise, are morally guilty, and have voluntarily broken the law. This has to do with the side of punishment that earlier was called liability. There is also the matter of how much they are to be punished, the matter of amount. Here there are rules of justice of various kinds.

    (1) Penalties must not cause more distress than would occur if they were not imposed. (2) A penalty must not be imposed if a man's action falls under a special rule of justification, such as the rule having to do with injuring in self-defence. (3) A man is to be excused a penalty if his act, although of a kind the law seeks to prevent, was unintentional, involuntary, or something of the sort. (4) A man is to receive a mitigated penalty if he faced a special difficulty in keeping the law, perhaps because of provocation. (5) Finally, a man is to be treated as others in his situation are treated. Similar offenders are to be treated similarly.

    As you will be aware, some of these various considerations give us some ways in which punishment can be taken to be justified, in part, by its having a preventive effect. Exactly what effect Hart has in mind is not clear. He speaks of Bentham and utilitarianism, indeed at length, but we need not conclude that what he contemplates as the effect of punishment is exactly the the greatest total of satisfaction when compared with alternatives to punishment. Let us leave this unsettled.

    More particularly, to return to the wide or general limitation of punishment to those who have intentionally or negligently broken the law, and so on, we are to understand that a part of the justification is that this serves prevention. It is those who have broken the law in the given way who are more likely to do it again. The same sort of principle plays a part with respect to (1) the rule against penalties that cause more distress than would occur if they were not imposed. So with (2) the rule about justification in offending, such as the one about self-defence. It is remarked that one of the aims of punishment is the protection of individuals from attack. It would be inconsistent, then, to punish a man for defending himself from attack, whatever else is to be said about it.

    But of course this is not the end of the story of justification. What is called retribution is also part of the story. It is part of what justifies the wide or general fact of punishment that it is for the most part done to offenders. Retribution is also part of the justification of (3) excusing a man if his action was unintentional or involuntary or something of the sort. So too is retribution a justifying consideration with (4) a mitigated penalty because of, say, provocation. The same can be said of (5) similar offenders being treated similarly -- you will remember that such a consideration entered into our own final analysis of traditional retribution theories.

    We now come to the crux. What actually is the reason of retribution for the parts or features of punishment just noticed? The answer, if it may be surprising, is clear. In general punishment is to be imposed only on those who could have kept the law when in fact they broke it for the reason that they could have kept the law when in fact  they broke it. That fact, you will remember, is given as the keystone of retribution theories and retribution. The same sort of reason is given when what is in question is why those who are morally guilty or have voluntarily broken the law are punished. The reasons, more particularly, are that they are morally guilty or they have voluntarily broken the law. Is 'voluntarily' here used in the special way in which we have been using it?

    To come on to the rules, and in particular the rule (3) that a man is to be excused a penalty if his action was unintentional, involuntary or something of the sort, the justification of this feature of punishment is that his action was unintentional, involuntary or something of the sort. As for (4) the reason for a mitigated penalty on account of provocation, it is that the offender was provoked -- that he acted in something less than perfect freedom. And as for (5), I take it that there is to be equal treatment because it is equal treatment.

    What all this comes to is an answer, good, bad or indifferent, to what is meant when a punishment is justified as a retribution. What it comes to, presumably, is a further answer to the question -- we have had a number -- of what it is for a punishment to be deserved. To revert to our previous style, the claim that a penalty is deserved for an offence, is understood roughly as follows.

                    (16) The offence was freely and responsibly committed.

    To look back over our inquiry, we have at various points taken a proposition about a punishment's being deserved for an offence to depend on or presuppose, roughly, that the offence was freely and responsibly committed. That indeed was a main point of our inquiry into determinism. At the end of that inquiry, further, it was concluded that retribution theories have rested on an assumption of that particular kind of freedom that is free will or origination as against what was called voluntariness.

    What we have now, in the theory of punishment we are considering, is that the retribution reason for punishment does not assume or presuppose a proposition about freedom and responsibility, but is that very proposition. If the theory is not explicit about the kind of freedom and moral responsibility it propounds or asserts, is there an indication that it is indeed free will? Certainly it is origination, not voluntariness, that goes most naturally with talk of moral guilt. It has to be said, however, that another paper than the one we have been considering gives reason to think that the freedom and responsibility taken to be the essential content of the talk of retribution may be no more than voluntariness. This is surely atypical of thinkers about retribution, and is such as a raise a question about the stability or durability of the theory.

    So much for a quick exposition of a supposed justification, or rather part-justification, of punishment. We come to this question: Why is it the case that man's having broken the law when he could have kept it is itself a reason for punishing him? Why is it that his having somehow broken the law is itself a reason for punishing him, and for punishing him to a certain extent? And so on. All of this needs to be kept separate from something else.

    It needs to be kept in mind that in the theory we are considering, to look at the other side of it, full attention is given to the voluntariness of actions or the lack of it because voluntariness is necessary with respect to the other justification of punishment, that it is preventive. The central proposition here is that it is those who freely and responsibly offend who are most likely to commit more offences unless deterred or the like. More is said by Hart along the same line.

    There is the recommendation of our requiring a free and responsible action that members of society are thus able to determine by their choices, at least to a considerable extent, what their futures will be. They are able, that is, to opt for law-abidance and its consequences or disobedi-ence to law and its possible consequences. Thus there are consider-able satisfactions to be gained -- individuals can enjoy a particular security, one that would be denied them under some system which did not make their own choices the determinants of their futures. The satisfactions of security are, essentially, satisfactions which accrue from confidence in prediction. I can rule out the possibility of at least certain unpleasant surprises if I feel capable of keeping the law. There are also gains of another kind. That I choose what will happen to me, to some considerable extent, is itself a source of satisfaction. In order to see this clearly, one need simply compare the experience of choosing with the experience of, say, being coerced.

    This sort of thing, the worth of a free and responsible action in arguing in a preventive way for punishment, is pretty plain. But, to revert to our questions, it is far from plain why the fact of voluntariness in an offence by itself is a justification of part-justification of punishment. As the advocates of forward-looking theories say, what is the worth of this dead fact of the past? To say better than that, all of us value such great goods as not being in pain and being free. Is this not our shared nature? These great goods do indeed give us reasons and justifications for things. If the elusive importance of a past voluntary act considered independently of anything else exists at all, must it not by comparison be near to insigificant?

    Perhaps in anticipation of just such a response, Hart writes elsewhere of what he speak of an important general principle.

'Human society is a society of persons; and persons do not view them-selves or each other merely as so many bodies moving in ways which are sometimes harmful and have to be prevented or altered. Instead persons interpret each other's movements as manifestations of intentions and choices, and these subjective factors are often more important to their social relations than the movements by which they are manifested or their effects. If one person hits another, the person struck does not think of the other as just a cause of pain to him; for it is of crucial importance to him whether the blow was deliberate or involuntary.... If you strike me, the judgement that the blow was deliberate will elicit fear, indig-nation, anger, resentment: these are not voluntary responses; but the same judgement will enter into deliberations about my future voluntary conduct towards you and will colour all my social relations with you. Shall I be your friend or enemy? Offer soothing words? Or return the blow? All this will be different if the blow is not voluntary. This is how human nature in human society actually is and as yet we have no power to alter it. The bearing of this fundamental fact on the law is this. If as our legal moralists maintain it is important for the law to reflect com-mon judgement of morality, it is surely even more important that it should in general reflect in its judgements on human conduct distinc-tions which not only underlie morality, but pervade the whole of our social life. This it would fail to do if it treated men merely as alterable, predictable, curable, or manipulable things.'7

    We certainly do distinguish between those who injure us deliber-ately, those who injure us involuntarily, and, one may add, those who do not injure us at all. We act differently in response. In so far as we are justified in so doing, however, this may have to be explained by something other than the mere fact by itself that some injuries are the consequences of voluntary actions. We are certainly not justified in responding in the same way to all those who injure us voluntarily, including policemen and our victims who are acting in self-defence. What can be said for acts of blaming may be what can be said for certain punishments. The fact taken by itself that an act was voluntary may not in itself be morally important in either case.

    Let us grant that in ordinary life we do react differently, and quite spontaneously, simply and solely because an act was volun-tary. That is, let us admit that we have distinctive emotional responses. Is it suggested, on the basis of these facts, that the bare consideration that a man has freely offended gives us a moral reason for punishment? Such an argument, in my view, would certainly be mistaken. We may be able to give a partial explanation of a feature of our practice of punishment by reference to these facts. Nothing is more likely than that there will be a connection between that practice and our ordinary attitudes and behaviour. To concede that is not to concede the point at issue. It is not to concede that the fact, taken by itself, that a man has freely and responsibly acted in a certain way is itself a moral reason for our further action. As many more say, if our action will do no good, if no good will come of it, what is the point of it -- indeed what is the good of it?

    The theory of punishment we have been considering has the recommendation of being well-informed with respect to the working institution of punishment. It has the recommendation too that it does not indulge in speculative thinking about retribution and desert. It is a long way from earlier pieces of retributive thinking we have considered. It has in it no flights of fancy.

    As it seems to me, however, the theory very clearly pays the price of its restraint. It produces no substantial or significant reason for punishment in terms of retribution. It invites replacement by the argument, such as it is, based on the satisfaction of grievance-desires. Let us leave the result of such a replacement unconsidered for a time. The theory does of course have the recommendation of seeing that more than a reason of punishment in terms of desert is needed. It does not have another recommendation that we have not so far considered, or not much considered. Except for suggestions of utilitarianism, it does not contemplate what it is that a justified theory of punishment is to prevent -- what its consequences are to be. We shall be giving that question more attention.


    The conservative and libertarian philosopher Robert Nozick's explanation of the justification of punishment is such that beginning with a summary of it is a good idea.

    The justification begins with a familiar idea of a deserved punishment, technically expressed, and contemplates in passing that such a punishment may be self-justifying -- justified because deserved in this sense. However, Nozick then seeks a separate rationale for such a punishment, first by distinguishing it from revenge. He goes on to find that a deserved punishment is an act of communication involving several intentions and the like on the part of the communicator. This act, perhaps imprisoning a man for 20 years or killing him, tells or shows the man how wrong his offence was. It reconnects him with the correct values from which he disconnected himself by his offence, correct values he flouted -- and still flouted if he was subject to determinism. By his imprisonment, these values have a significant effect on his life. Nothing but the punishment could bring about the significant effect of reconnection with them. It may happen that such a punishment also does something different. It may morally improve the offender. But this if it happens will be a bonus. It is enough for a justified punishment that an offender is reconnected with correct values.

    To these principal propositions are added various others. An offender is not only to be reconnected to correct values by his deserved punishment. We are to try to force him to compensate victims of his offence. Further, in addition to the deserved punishment, we may consider what is called matching punishment or the original lex talionis, which is acting against the wrongdoer in the very way he acted against his victim, so far as this is feasible. In punishing retributively, we not only have the reason of connecting the offender with correct values, but also others. One is that we connect ourselves with value. And, with such great evil-doers as Hitler, we may also have to respond to wrong as wrong, by killing him even if he could be reformed. Finally, to come to an objection contemplated by Nozick, there is the idea that we might be obliged by this thinking to punish people who are disconnected from correct values but have not committed any offence. The reply given is that this would be a violation of their rights.

    All of which is not to say that any current punishment is well designed and functions well in terms of retributive punishment as now explained. That choice of words indicates, I take it, that current punishment as it is can none the less have a significant justification in terms of desert, communication, connection with correct values, the possible but unnecessary bonus of moral improvement of those punished, and so on.

    So much for the summary. Let us now make our way through some of these propositions and think a little bit about them, the first having to do with a familiar idea of a deserved punishment, technically expressed.

'The punishment deserved depends on the magnitude H of the wrongness of the act, and the person's degree of responsibility r for the act, and is equal in magnitude to their product, r x H. The degree of responsibility r varies between one (full responsibility) and zero (no responsibility) and make take intermediate numerical values corresponding to partial responsibility.'8

    The idea expressed in this manner is best and most kindly taken as the familiar one that a deserved punishment is one that is equivalent to the culpability of the offence -- where that is ordinary talk and not the idea of a hopeless calculator who supposes that there do exist measurements that do not exist. That is, we need to remember, of course, that Nozick's sentences expressing the idea are not literally true. There are no units for the counting or measurement of wrongness, there are no units for the counting of responsibility, there is no multiplication of harm times responsibility, there are no units for the counting of punishment, and there is no equality or inequality in magnitude between r x H and a punishment. Taking the words literally, none of those things exist.

    That is not to say, of course, that we cannot put numbers on the things in question and then operate with those numbers. We can put numbers on them in the way that we put numbers -- examination marks -- on particular answers to examination questions. Or the way that we put numbers -- grades -- on apples or other fruit that we assign to different categories or qualities. Or the way we put numbers -- points -- on an ice-dancer's performance. Or numbers -- demerits -- on a pupil's contribution to the school community.

    Is the point that we cannot count or measure wrongness, responsibility and so on a trivial or small one since it is to be granted that we can put numbers -- in terms of dollars or years, say -- on wrongnesses of acts and punishments? I doubt it. To talk in terms of magnitudes, products, numerical values and so on is to invite someone to think that what is in question is a single and clear argument for a punishment. That is less likely to happen when we speak more accurately, say in terms of putting numbers on things. Also, the untrue way of speaking leads us away from our role in responding to offences and as punishers in these so-called magnitudes, etc.

    That is not what is most important, of course. It is not the main objection to the assertion of an equivalence between the offence and the punishment. What is most important can be put in terms of things that can literally be measured and are literally equal. Take a morning's activity that produces a table four feet long and a stool two feet wide. Take an afternoon's activity that produces a bench six feet long. Is the morning's activity an argument for the afternoon's activity in virtue of the equality by itself? Obviously not. There may some other reason why the equality in question is a good thing, or a reason why it is a bad thing, or a reason why the equality has no value at all, positive or negative.

    So with an equality between or rather an equality put on r x H and a punishment. It is only the dulling force of convention, and conventional thinking in whatever form, including a form that trivially mimics mathematics or physics, that may conceal the absurdity from us. To add an example, a war does not become right for the sole and only reason that it will equal another war in deaths and destruction. Add your own further examples.

    As against this, Nozick does contemplate in passing that an assigned equality or equivalence between r x H and a punishment may indeed be self-justifying and need no further explanation -- that is, the equality may in itself be a justification for imposing the punishment.

'Is it necessary, though, to offer an explanation at all of retributive punishment? Perhaps its appropriateness is just a fundamental fact, with nothing further underlying it: people who commit wrongs simply deserve to be punished.'9

    At best, this is or amounts to a particular theory noticed in our earlier inquiry into traditional retribution theories, earlier, intrinsic retributivism, where the suffering of the guilty is taken as an intrinsic good. What we have is the proposition that there is intrinsic good produced by adding the punishment to r x H. The fatuity of this argument is made evident by a possible response of the same force in argument -- or lack of force. It is that there is intrinsic goodness in forgiveness or whatever, and hence that there is intrinsic good in not adding the punishment to r x H.

    Very rightly, despite his moment of contemplation, Nozick does suppose that a better answer, an actual answer, is needed to the question of why it is right to impose a punishment that can be talked of as equal to the r x H of the offence. This better answer is what he calls a rationale of retributive punishment, indeed the rationale. It is a long way from the clear sense that we make punishment in some sense or way equal to r x H in order to prevent offences and do do so at no greater cost of suffering than is needed.

    The way to Nozick's chosen rationale of punishment is cleared by getting rid of the obstacle of suspicion that that retributive punishment is revenge, that making a punishment equal an offence is taking the correct revenge. The various points made about punishment and revenge, some of interest and many contentious, reflect the general fact accepted by almost all of us that punishment however justified is the work of an authority, most notably a national state, according to law. It is not the work of an individual who has himself been harmed by his victim or has another personal link to him.

    One contentious point is that revenge unlike retribution is done for a harm and not necessarily a wrong. This at least overlooks the fact that both are done for supposed wrongs, however personally judged. I have to persuade myself of some defect in the neighbour whose windows I break or even the child I shoot as a soldier in the course of my state-terrorism. Another contentious point, in this case one that likens revenge to retributive punishment as it will be conceived, is that revenge is always done with the intention that the victim know why it is occurring and know that he was intended to know. That seems untrue. On the contrary, I may take revenge in so careful and self-preserving a way that the finger does not point at me at all.

    None of this is of importance, first of all because it seems very hard or impossible to suppose that that punishment, certainly punishment by the state, is revenge. No clearing of an obstacle of suspicion is needed. And, most relevantly, that some of us suppose it is like revenge is perfectly consistent with the differences adduced between punishment and revenge, including the contentious ones. Further, as Nozick admits, it does not follow, from the fact that retributive punishment is not revenge, that the former is justified. As he might also have thought, the upright ways of doing philosophy do not include clearing non-obstacles in order to clear the way to something else.10

    Near the centre of the view we are considering are intriguing thoughts of the philosopher of language Paul Grice. They have to do with what it is to mean something by some sounds or marks. The thoughts, arrived at by means of contemplating examples and counter-examples, include such intricacies as that my communicating with you in an ordinary way includes not only my intending that you come by way of my sounds or marks to have a belief or whatever, but also that you know I intended my sounds or marks to have the effect.11 There is no great difficulty in arriving by way of this account of the conditions of communication at the proposition that that punishing someone, and in particular punishing someone so that the penalty can be said to equal the responsibility r X the wrong H, is a kind of act of communication. If we may be helped to forget that punishing is also other things, this proposition is made clear to us.

'Retributive punishment is an act of communicative behaviour. ... The (Gricean) message is: This is how wrong what you did was. ... We might see punishment as an attempt to demonstrate to the wrongdoer that his act was wrong, not only to mean the act is wrong but to show him its wrongness.'12

    There is no great problem about taking punishment to be communication. As Nozick might have remarked, there are innumerable examples elsewhere of such non-linguistic communication. Holding out water to a thirsty man is, among other things, a kind of communication, something like the speech-act of saying 'Here is water for you'. So is pointing a gun at someone's head, or marrying them, or leaving them, or going to war. Punishing someone, among other things, is indeed a kind of act of communication.13 But suppress any question you may have with respect to that, or, more likely, the use to be made of the fact, and consider instead what this act of communication is also said to be or achieve.

'The wrongdoer has become disconnected from correct values, and the purpose of punishment is to (re)connect him. It is not that this connection is a desired further effect of punishment: the act of retributive punishment itself effects this connection..14

Or, as can be added,

'Correct values are themselves without causal power, and the wrongdoer chooses not to give them effect in his life, in a secondary way. When he undergoes punishment these correct values are not totally without effect in his life (even though he does not follow them) because we hit him over the head with them.'15

    Let us not linger over these thoughts either, but press on to something else that throws a strange light on them. The reconnection about which we are hearing is not something else. Reconnecting an offender with correct values is not getting him to realize that his offences was wrong.16 That is not what happens as a result of the penalty that is said to equal his past wrong and responsibility. Reconnecting, in a word, is not reforming the offender, not the moral improvement of the offender.

    Reformation is something that may happen, but is not necessary for the justification of punishment being claimed.17 It is a bonus, not something to be counted on or needed. Nozick must take this view, as in effect he admits, because it requires at least optimism to suppose that retributive punishment, indeed any punishment, however communicative, does in fact make men conventionally moral.18

    A question now arises. Indeed, it cannot be avoided. If we allow that imprisoning a man for 20 years in a sense means something, in a sense communicates something to him, what is the further fact that it reconnects him with correct values -- if that does not mean he is made to realize that what he did was wrong, that he is somehow made morally better?

    Reader, I am getting tired of this stuff, and will not persist much longer with it. Speculate, if you want, about what reconnection with correct values comes to. Follow Nozick's habit elsewhere and bring bits of philosophical imagination to bear on the question, maybe a joke. My own conclusion at this point can be brief. This obscurity, which is indubitable, and thus this weakness or absence of argument, this mere performance, suggests very strongly that retributivism, and the doctrine we have been contemplating, has in it or underneath it something it does not express.

     Another question may have arisen in your mind before now, one raised by any theory of punishment as communication. If your aim is to tell somebody something, is it necessary to do so by putting him in jail for 20 years? If you do not even have the aim of changing his mind about correct values, why is it necessary to put him in jail for 20 years? Nozick's answer, it seems, is in the following lines.

'Correct values are themselves without causal power, and the wrongdoer chooses not to give them effect in his life. So others must give them some effect in his life, in a secondary way. ...we hit him over the head with them. Through punishment, we give the correct values, qua correct values, some significant effect in his life.... The complicated (Gricean) intentions enable us to act as a vehicle.... Only those complicated conditions enable the correct values to act upon him.... There is now no puzzle about why we do not simply speak or telegram the (Gricean) message, without adding a punishment.'19

    What that says, perhaps, is that values are such things as to need a realization in conditions of communication in order to have a certain effect on someone. The conditions are achieved in punishment. Therefore there is no puzzle as to why we punish. That appears to be a non sequitur, one that forgets about the conditions of communication realized in language and a good deal else, a non sequitur that also hits you over the head and which I leave for your consideration.

    Whatever point is intended about the necessity of punishment to the wholly obscure effect, the point is itself sufficiently obscure as to be unlikely to be forceful. To which can be added the thought that it is dishonourable to present the alternative method of communication as 'seeing films, reading novels, and hearing explanations of the causes of his behaviour and the tales of his victims'.20 What a serious communication theorist opposed to punishment could perfectly reasonably propose is an institution of something like the size of punishment, no doubt a little less expensive. What we can instance, in objection to Nozick's proposition of necessity, is such an institution.

    What you have heard so far, properly reduced, seems to me as follows. A deserved punishment, as nothing else can, communicates to the offender that an offence was wrong to some particular extent. Since this is unlikely to reform him, a proposition is needed in explanation of why we go in for this communication. It is that the punishment reconnects him with correct values.

    This piece of mystery-mongering raises a further question so far unasked. What are these values? If everything else were as clear as glass, this question would still need answering. Presumably there are values of a good society or indeed the perfectly just society. These can be the values, that is, of the society that is morally ideal. Nozick wrote another book specifying this society.21 The morally ideal society is the one where all the goods, say food, are where they are, in the ownership of particular people or entities, as a result of a certain history. They are where they are because the owning people mixed their labour with them or with something related to them, and so came to own them, or they are where they are as a result of sales and gifts, maybe a long sequence of these tracing back to the labour-saving. Or, thirdly, the goods are where they are as a result of acts of rectification that put right departures from a history of sales and gifts, say robbery, maybe by governments in the form of taxes.

    There are brief and unquestionable summaries of this society of unquestionably correct values. It is the one where no person and no group or entity has any outstanding moral obligation and no person has a moral right to anything they do not already own. The only moral rights people have are to the things to which they have legal property rights. This is a society in which a family starving to death may have no moral right to food. It is a society in which no person and no entity may have any moral obligation to save the family, maybe with a father who is psychotic, paranoid, phobic or dim, from starving to death.22

    What I mean to do by this reminder of our instructor's correct values is mainly one thing. It is to note again that if we are told that a system of punishment has to do with values in whatever way, we will need to know something of what those values are. They may be such as to raise a question, to say no more than that, with respect to the moral decency of the system of punishment.

    So much for Nozick's principal propositions in justification of punishment. I leave his other propositions to your own consideration, to your reading of his book. You may also wish to consider the question of the consequencs of determinism for his propositions.


    There is another communication theory, said also to be a kind of retribution theory but not a pure retribution theory. If a metaphor helps, it takes punishment, anyway in part, to be the voice of the law and what is hoped to be the voice of the community. It is the work of Antony Duff, a philosopher who adds to his logic the politics of what is called communitarianism as well as the more traditional politics of liberalism. The theory, in its first full statement in the book Punishment, Communication and Community23, is that punishment

'should communicate to offenders the censure they deserve for their crimes and should aim through that communicative process to persuade them to repent those crimes, to try to reform themselves, and thus to reconcile themselves with those whom they wronged.'24

When this is true of punishment, it is justified. It is justified when it is this 'species of secular penance'.25 This is not at all or not mainly penance in a usual sense -- a voluntary self-punishment inflicted as an expression of repentance for having done something wrong. Rather, what we have is the kind of penance that is imposed by somebody on somebody else with a certain aim.

    Certainly an idea of desert is fundamental or at least integral to this theory. We have it, further, that desert is in a necessary rather than a contingent connection with a forward-looking idea of reform or whatever, and so the theory is not what we have been calling a mixed theory. Punishment is justified as

'a communicative enterprise focussed on the past crime, as that for which the censure that punishment communicates is deserved; but also looking to a future aim to which it is related, not merely contingently as an instrumental technique, but internally as an intrinsically appropriate means. ... Punishment will now look both back (as retributivists insist it must) to a past crime as that which merits this response, and forward (as consequentialists insist it must) to some future good that it aims to achieve.'26

    Whatever is to be said about the supposed mistake of taking this as a mixed theory, of which a word more will be explained, you may, reader, have fallen into another mistake. It has to do with the necessity or otherwise of reform in the theory. You may be confirmed in the mistake by understanding that the theory is not the commoner idea just that punishment expresses to the offender a deserved condemnation.27  It is not just that punishment is the voice of the law and maybe the community, but rather that the voice is really heard by the offender. Expression of something as we are reminded, requires only somebody to express something. On the other hand, 'communication involves, as expression need not, a reciprocal and rational engagement'.28 In communciation, at any rate, we try to engage with someone else, get a response mediated by the other's grasp of what we say.

    Despite this fact about communication, it does seem to be a mistake to suppose that on this theory punishments are justified only when they are likely to result in the repentance, reform and reconciliation of an offender with respect to his wrong-doing.

'...the law should aim to bring him to recognize and repent that wrongdoing: not just because that is a method of persuading him not to repeat it, but because that is owed to him and to his victim. To take wrongs seriously as wrongs involves responding to them with criticism and censure; and the aim internal to censure is that of persuading the wrongdoer to recognize and repent his wrongdoing. This is not to say that we should censure a wrongdoer only when we believe that there is some chance of thus persuading him. We may think that we owe it to his victim, to the values he has flouted, and even to him, to censure his wrongdoing even if we are sure that he will be unmoved and unpersuaded by the censure. But our censure still takes the form of an attempt (albeit what we believe is a futile attempt) to persuade him.'29

    Whether or not in perfect consistency with what is said elsewhere,30 that is pretty definite. A punishment on this view does not have to have even a chance of reforming in order to be justified. It is not that we are to go ahead with it because we cannot be sure it won't work. What the view comes to, we can take it, is that punishment is right when it tells an offender something, and in some way he understand it, even if it and his understanding have no reformative effect on him. The thought may incline us to suppose this is more of a retribution theory than so far supposed, purer retribution than supposed.

    The quoted passage, incidentally, also indicates the sense in which the theory is not a mixed one. What this comes to, anyway in part, is that taking something as having been wrong involves or even is responding to it with censure. You logically cannot do the first thing without doing something like the second. So to take a penalty as deserved, anyway in a certain sense, or to make a judgement involved in taking a penalty as deserved, is to do something that carries censure with it, rather than something that may have censure added to it. The point does not preclude us from thinking, however rightly, that in the theory punishment is being justified by more than one thing.

    This is a good moment to ask what general understanding of the relationship of desert informs the view, however and wherever it informs it. How are we to understand the repeated statement that a penalty may be deserved for an offence, owed for an offence, however and wherever such a proposition functions in the theory? The question actually comes up in a somewhat puzzling way. It is not presented, so to speak, as a fundamental or first-order question to which the theory needs an answer in order to exist, an answer that is indeed fundamental to the theory. Rather, the question comes up in a kind of addendum to the theory or the main theory, an addendum on what is called its sentencing policy -- the offences-and-penalties system that goes with the recommended justification of punishment.31 Still, it seems clear to our instructor, more enlightened than many predecessors, that his many previous references to desert have to have an understanding supplied for them, that such a thing is needed for them, indeed that it is intrinsic to any version of retribution.32

    The general answer to the question of what the talk of desert comes to is given in terms of proportionality, a penalty's being proportionate to an offence, and proportionality is explained in terms of the offences-and-penalty system.33 In short,  we have something that was considered in our look at traditional retribution theories, an attempt to explain propositions of the form 'Penalty P is deserved for offence O' as the penalty's being according to a system of penalties and offences. In the course of our guide's exposition of the penalty system, the ordinary one of which we all have an idea, we learn interesting things about kinds of proportionality, relative and absolute and positive or negative. We are reminded that we have do indeed have what are called intuitions or anchoring points -- such as that five years in prison would be grossly disproportionate for a parking offence, and a fine of $10 a grossly disproportionate sentence for a brutal murder. But it is admitted that this leaves us with a lot of questions. We cannot make determinate judgements. Perhaps it is is admitted that we cannot make an initial and general judgement that will guide us in thinking about or defending or amending or reconstructing the system.

'We must recognize, I think, that any attempt to to work out a penalty scale (either its anchoring points or its content) from scratch is doomed to failure. There is no archimedean point, independent of all existing penal practice, from which we could embark on such an enterprise. We can only begin from where we are now, with a penalty scale whose content and upper and lower limits have been determined by a host of historical contingencies.'34

    That recognition is indeed not very clear. Is it the recognition that there is no general principle or proposition of proportionality or desert that informs the making or altering of a penalty system? Certainly none is supplied. It is not said, for example, that we are guided by actual facts of equivalence between culpability and distress, something like facts of of actual weight or length. Does the retributive communication theory of punishment then lack content of a fundamental and essential kind? Before we answer that question and others that arise from what we have before us, there is another matter that needs attention. The communication theory has a recommendation so far unnoticed, however well it fits together with what have on the table already. It is a recommendation lacked by all of the traditional retribution theories of punishment considered earlier.

    The central theme of retribution, as we know, is that a punishment is deserved for an offence. Whatever else that comes to, it includes a reference not merely to what is against the law, but what is wrong. As remarked before now, no reflective retributivist maintains that an offender deserves punishment no matter the moral worth of the law in question, no matter the moral standing of the offence or crime. No serious retributivist would suppose, for example, that someone deserved punishment in his sense for the crime of not going to the authorities about the whereabouts of a person who would end up in a gas chamber if located.

    So it is very clear indeed, and should long have been clear despite many oversights, that any retribution theory of punishment depends on a conception of what things a society ought to make into crimes. Whether or not the answer to the question is partly obvious -- rape and torture presumably cannot be left out -- the question must have a full answer. Seeing and saying so is a recommendation of Duff's thinking.

'A normative theory of punishment must include a conception of crime as that which is to be punished. Such a conception of crime presupposes a conception of the criminal law -- of its proper aims and content, of its claims on the citizen. Such a conception of the criminal law presupposes a conception of the state -- of its proper role and functions, of its relation to its citizens. Such a conception of the state must also include a conception of society and of the relation between state and society.'35

    In short, a justification of punishment rests on a justification of a society. It is of considerable interest that the problem of the justification of punishment was so often for so long considered entirely independently of the question of what actions are wrong, what actions are to be punished, indeed what a decent society is. Justifications of punishment in terms of retribution were advanced as good as always without specifying the category of things for which punishment was said to be deserved. It was very unlike the utilitarian prevention theory, which explicitly rested on the idea that the laws of a society should serve the end of the greatest total of satisfaction or well-being.

    The communication theory we are considering does not leave out a kind of account of the society for which it offers a justification of the punishment in it. It is a society such that both of two political philosophies are realized in it, or on the way to being realized in it, these being liberalism and the more recent and conceivably transient doctrine of communitarianism. In this way the theory takes forward earlier work in the defence of punishment by way of the several ideologies, notably the work of Nicola Lacey.36

    The political philosophy of liberalism, in fact a general prescription, is often summed up as being to the effect that a society to some considerable extent should not impose a general theory of how life ought to be on its members. Rather, the society should be pluralistic or tolerant -- within certain limits it should leave its members to pursue their own ideas of what is called the good. It should leave individuals to themselves except when they cause harm somehow conceived to others. What is fundamental to this enterprise, as is often said, is individuals and their rights -- they are to have freedom or autonomy, privacy, and the protection of a neutral state more engaged in protecting their contracts than in any larger endeavour.37

    The second political philosophy, communitarianism, is said to be even harder to define than liberalism.38 However, we hear that a state and society have an obligation to forward some theory of the good and perhaps that they cannot help but do this. It is said to be essential to this view that it gives some greater value and role to a community, say a country, than does liberalism. This prescription goes with a certain conception of persons as somehow less individual, as in fact somehow owing their natures to their communities. It is not easy to be confident that the basic ideas of communitarianism, on balance, locate it to the right or the left of liberalism in the political spectrum.39

    That is not the end of the matter of punishment and society. The society that enters into the justification of punishment provided by the communication theory is in fact to be taken as one that is true to both liberalism and communitarianism. The society we are to think about has the name of being a liberal political community.

'A liberal political community will recognize individual freedom and autonomy as crucial values: as human goods to be fostered and encouraged and as rights that must be respected by other citizens and by the state. ... Although it will insist upon autonomy, freedom and privacy as central goods, it will not seek to enforce any simple all-embracing or comprehensive conception of human good.'40

However, for a start, it is also true that the members of a liberal political community will

'constitute a community insofar as they aspire, and know that they aspire, to share the community-defining values of autonomy, freedom and privacy...and insofar as they aspire, and know that they aspire, to an appropriate mutual concern for one another in the light of those values. That mutual concern will involve a readiness to assist one another in pursuing and preserving the community's distinctive goods -- though such assistance will often be organized and directly provided by the state -- and, more crucially for present purposes, a respect for one another as fellow members of the community that precludes simply exploiting others for one's own own ends or treating them in ways that are inconsistent with the community's defining values.'41

    So much for an exposition of a ramified, qualified and distinctive theory of the justification of punishment. We have accumulated elements of it that give rise to more questions than have been partly answered. Let us proceed through them, following something like the order of the quoted passages.

    As already granted in connection with a different kettle of fish, there does seem to be sense in saying that punishment is among other things a communication, censure or condemnation. It can indeed be said to tell offenders something, and to get their understanding of it. That thing is that a society or rather a state takes their offences to have been wrong. Of several responses we might make to this fact, the main one is surely that this is not enough to be said to punishment. If this is all that is to be said for it, not enough can be said. The addition of this good to some fact of punishment's being deserved is not enough to justify it.

    That punishment is the voice of the law is not enough to make right imprisoning a man for 20 years, quite likely a man who has never quite known what he was doing in the world, or killing him by putting him in an electric chair or injecting him with something loathesome.. It is essential to keep in mind here, of course, that the communication in or by punishment is distinct from and is not at all recommended as having any effect in reforming offenders, any effect in preventing offences. On the theory we have, as you have heard, we are to punish when we believe there is no chance at all of the punishment's having a reformative effect. Despite the quoted passage there may be some uncertainty about the point. But I take it that punishment would have the given value of being a communication if it never had a reformative effect.

    It must be, if so, that the justification of punishment on offer depends on some other thing than communication and penance. Indeed it does. What goes with the insufficient fact of communication is that somehow punishment is deserved. It is curious that that additional thing, despite familiar lines in the book to the effect that retribution theories fail in not finding some other real good in punishment, is not a good consequence or effect of punishment. What is added to the seemingly insufficient fact of communication is that somehow punishment is deserved. A first question here has to do with the first passage quoted above and what might be called the role or part played by some or other relation of desert in the theory. Punishment when justified, you heard, will 'communicate to offenders the censure they deserve for their crimes'. What does that means? Certainly it is not clear and unambiguous. Let us disambiguate.

    Remember, first, that the communication or censure is the punishment. The communication is not additional from or different from the 20 years in prison, maybe the judge's words on the offender's character and like things, or the accusing attitudes of the guards. If we concentrate on this proposition, surely indubitable, what we have in references to a deserved communication, censure or penance is at bottom a traditional retribution theory, most likely about culpability and distress in some way, onto which some thoughts on communication are imposed. What goes together with the proposition that punishment is communication is just that punishment is deserved. This understanding of the theory goes together nicely with what is regularly mentioned in passing but not officially declared as a main proposition, that punishments are 'owed' to the offender or his victim.

    We can remember or suppose something else, however. It can be supposed that the theory is not the punishment as an imposition of distress is deserved, but only the punishment as a communication is so. The punishing of the man has several sides or properties, and it is its being a communication, which we have supposed it can be, that is deserved. But what does it mean to say that a communication is deserved? What is a deserved communication? It is not easy to say. If we keep firmly in mind that it is not the distress of the punishment that we are contemplating, but exactly and only the communication or censure, is its being deserved perhaps its being true, merited, or well-based? Is it that the message fits the facts?

    If that new idea is not the right one, here is another. What is deserved is exactly the distress of being censured or condemned for wrong-doing, as against the more ordinary distress of punishment, say being kept in prison for 20 years. It is to be noted that both of these new ideas can be supposed to make for what we have heard about, an intrinsic connection between the past wrongful offence and what is now deserved. If what is deserved is a true condemnation, that is one with or is logically entailed by taking the offence to have been wrong.

    These three answers issue in judgements that can be expected. The first of them makes the communication theory into a traditional retribution theory with the addition of a not very substantial proposition about communication. This is surely not enough to justify the 20 years in prison. The second one, about a true communication, might be thought to go even less far towards a justification. The third, whether or not separated from the first, calls for the same response. That is not all.

    These answers issue in more of what you have noticed often enough before now, a conviction that what is said of retribution and desert by justifiers of punishment is obscure. It does indeed invite some sort of reductive analysis, if not diagnosis. The second and third answers also issue in something else. It is a question with which our theorist of punishment seeks to deal. If our aim is only to engage in a true or well-founded condemnation, or to cause to someone only the hurt or the like of just that condemnation, why are we putting him in jail for 20 years? But leave that.

    Another question about the theory's reliance on desert or retribution can be dealt with more quickly. Put aside the first new idea -- a deserved condemnation is merely a true or well-based one. The other two ideas raise the question asked so often in the course of our inquiry. What is the relation of desert between an offence and a penalty, the relation that makes the penalty right and justified? Well, what is said of that was reported in the exposition of the communication theory. But no account of the relation of desert can possibly be given by taking up the mere synonym that the deserved penalty is the proportionate one, of course. Nor is a general account given by way of citing a penalty system. You cannot convey what is supposed to be the justifying principle of a penalty-system by pointing at it. The communication theory therefore does lack content of a fundamental and essential kind. A preventive theory of punishment points at such a system, very similar indeed, and actually gives its rationale in terms of rational prevention.

    The obscurity about the relation of desert, and about what it is suppose to relate, certainly bears on another matter, a practical one. The obscurity is not just a failing in a theory but may also at least contribute to a practical problem. Penalty systems are not settled once and for all, but come up for revision. As new proposals come up, how are we to decide on them if we have no general guide? It is no good saying you look at the rest of the system and do something new that fits in with the rest of it, is it? We have to have a general understanding of the rest of it that guides us in the change, indicates why one thing would be inconsistent with another.

    What of communitarianism and liberalism? You have heard something of them, in the words of their expositor, and of their synthesis in the doctrine of the liberal political community. You have had the words delivered to you in order to as to make less unfair a question or two that might be better answered after a lot more reading. What does communitarianism come to? What does liberalism come to? What does the liberal political community come to?

    We need answers to the questions because we need to know what the wrongful acts are that this retributive communication theory will take to be rightly punished. Let me say plainly that in my opinion the given exposition of communitarianism and liberalism shares the character of all other expositions of which I know. It is at least elusive. This is partly a matter of indeterminate ideas -- say freedom and indeed community -- and the piling on of qualifications. I hope to speak for you, reader, in saying that we do not have much useful distinction between communitarianism, liberalism and the doctrine of the liberal political community. We do not have plainness in the account of any of them. What we have, it might be said, is a report of good intentions carried not carried forward into determinateness.42

    Do you say in reply that we do know what is in question, or know well enough? Do you say, despite the fact of political conservatism in our societies, that what is in question with liberalism etc. is in fact societies that are more or less or like ours -- societies with more of this or that kind of provision or institution? Well, that does not accord with Antony Duff's reservations.43 But suppose we can get an idea of the subject-matter of the communication theory, in so far as it concerns a society, by thinking of the things that are crimes in Britain and America at the start of the 21st Century.

    That does indeed give some content to the communication theory. We know more of what we are talking about. But there remains a large problem. We now do know in a sense what the basic aim of the communication theory is, but this is not to know very much, not enough. The aim is to defend our societies as they are, or to defend societies to which we can extrapolate from our societies. But we do not know, in any enlightening way, what is supposed to make these societies worth defending. That leaves the communication theory incomplete in the most important way. If it has the distinction over almost all of its predecessors in seeing a need, it does not satisfy that need.

    Return for a moment to the utilitarian prevention theory of punishment. As already remarked, we know exactly what it is that recommends a certain system of law, criminalizing of behaviour, prevention of offences. It is that the system is the one that is taken to maximize satisfaction or well-being. What is the comparably enlightening generalization with respect to the liberal political community? It has certainly not been produced.

    We shall be coming back to this crux of theories of punishment. But, to end, here, what is to be said of the communication theory is that it shares the failure of traditional retribution theories of punishment by having a hole at its centre, a hole where there ought to be clarity about desert. It adds that hole to what needs a great deal more to recommend it, which is punishment as communication. The theory invites being rewritten into something else.

    Would further study make it clearer to me? Would it help with other problems you may have, say about communicating with the many offenders who are mentally ill? Well, further study might be worth it, and I commend some study to you.


     Punishment: The Supposed Justifications was first published in 1969. It would be alarming for an author if a book written 35 years ago and revised 20 years ago seemed just as right now as it did then. It would not be reassuring to a reader either. Fortunately for its peace, and as you will have gathered, my mind has changed about things. If this additionally revised and enlarged edition (Pluto Press, 2005/6) is no recantation, but remains on the side of its predecessors and does not go over to any other side, it does take up a different position in the one it stays on.
     You could say there is more edge to its moral conviction. Is this partly because the world has changed? That, as is plain,
punishment has become more punitive and there is more of it? Or because the world has not changed? Because, that is, the political moralities of conservatism and liberalism do not change their spots, but keep their societies serving their interests, and because this has sunk in on the author?
    As well as in moral conviction, the book is also very different in other contents and in organization. Certainly there was reason to change the title. Two chapters of the last edition, the first and last, about backward-looking theories of punishment, have become two different ones, the first and second. The sixth chapter, on determinism, freedoms, and ways of holding people morally responsible, entirely replaces the old one. It reflects my own progress, if that is what it has been, in thinking about these subjects on their own, the separate problem that is nearly as much a problem in the philosophy of mind as in moral philosophy. The seventh chapter, on compromise theories, and the eighth, giving conclusions, are entirely new replacements of their predecessors. Throughout the book some mistakes have been put right. Some propositions are worked out more fully.

     Another way of saying much of this is that the book remains true to a realism and plain sense, as it does indeed seem to me, about almost all talk of retribution, desert, proportionality and the like in punishment. That is a realism that has never endeared the book to a better class of readers, including some too self-respecting lawyers.
    But, to go back to differences, the book is now truer to the proposition that the subject of punishment is not to be treated without an actual consideration of the subject of crime, which is to say what a society or part of it deals with by means of punishment. That is to also to say that the subject of punishment is not considered without actually considering and judging, if too briefly, what a society or some part of it takes to be rightly defended -- indeed takes to be a decent society.
    The book is also different in being aware of some late 20th and early 21st Century thinking and also of some empirical facts and other things that have become plainer, notably facts having to do with rehabilitative and deterrent hopes for punishment and also the subject of determinism and an interpretation of Quantum Theory. No really new doctrine of the justification of punishment has come into being since the book was first published. But certainly there has been enlargement, modification and some admission of error with respect to the old doctrines, and also that wonderful intellectual progress that is terminological.

1 H. L. A. Hart, 'A Prolegomenon to the Principles of Punishment', in his Punishment and Responsibility: Essays in the Philosophy of Law (Oxford University Press, 1963); John Rawls, 'Two Concepts of Rules', The Philosophical Review, 1955.

2 'A Prolegomenon to the Principles of Punishment', pp. 1-4.

3 'Two Concepts of Rules'.

4 'A Prolegomenon to the Principles of Punishment', pp. 17-21.

5 'A Prolegomenon to the Principles of Punishment', p. 1.

6 'A Prolegomenon to the Principles of Punishment', p. 9.

7 Hart, 'Punishment and the Elimination of Responsibility', in Punishment and Responsibility, pp. 182-3.

8 Robert Nozick, Philosophical Explanations (Oxford University Press, 1981), p. 363.

9 Philosophical Explanations, p. 366.

10. As remarked earlier, however, Leo Zaibert maintains that no effective distinction can be made between punishment in general, as distinct from punishment by the state, and revenge. See his Punishment and Retribution (Ashgate, 2005).

11 H. P. Grice, Studies in the Ways of Words (Harvard University Press, 1989).

12 Philosophical Explanations, pp. 370-1.

13 There is much earlier thought on punishment as communication, some in the work of the sociologist Emile Durkheim. On Durkheim, see David Garland, Punishment and Modern Society (Oxford University Press, 1990), Ch. 2, especially p. 45. See also Jean Hampton, 'The Moral Education Theory of Punishment', Philosophy and Public Affairs, 1984, Anthony Skillen, 'How To Say Things With Walls', Philosophy, 1980 and Igor Primoratz, 'Punishment as Language', Philosophy, 1989.

14 Philosophical Explanations, pp. 374.

15 Philosophical Explanations, p. 375.

16 Philosophical Explanations, p. 372, p. 373.

17 Philosophical Explanations, p. 373, cf. pp. 378-9.

18 Nozick's setting aside of actual moral reformation by punishment sets his view aside from A. C. Ewing's reform theory, noted earlier, despite very notable similarities about punishment as communication. See above p. 000.

19 Philosophical Explanations, pp. 375-6.

20 Philosophical Explanations, p. 373.

21 Anarchy, State and Utopia (Blackwell, 1974)

22 On Political Means and Social Ends (Edinburgh University Press, 2003), pp. 31-3, 164.

23 Oxford University Press, 2001.

24 Punishment, Communication, and Community, p. xvii.

25 Punishment, Communication, and Community, p. xix.

26 Punishment, Communication, and Community, p. 88, cf. p. 129.

27 J. Feinberg, 'The Expressive Function of  Punishment', in his Doing and Deserving (Princeton University Press, 1970); I. Primoratz, 'Punishment as Language', Philosophy, 1989, and also Justifying Legal Punishment (Humanities Press, 1989)

28 Punishment, Communication, and Community, p. 79.

29 Punishment, Communication, and Community, pp. 81-2, cf. p. 123.

30 Punishment, Communication, and Community, pp. 121-5.

31 Punishment, Communication, and Community, p. 131.

32 Punishment, Communication, and Community, p. 137.

33 Punishment, Communication, and Community, pp. 11-12, 131-4.

34 Punishment, Communication, and Community, p. 134.

35 Punishment, Communication, and Community, p. 35.

36 Nicola Lacey, State Punishment: Political Principles and Community Values (Routledge, 1988).

37. See, for a start, 'Liberalism' by Will Kymlicka in Honderich, ed., The Oxford Companion to Philosophy (Oxford University Press, 2005). For something on a liberal theory of punishment, see Hugo Adam Bedau, 'Punishment', The Stanford Encyclopedia of Philosophy (Summer 2003 Edition), Edward N. Zalta (ed.),  URL =  <http://plato.stanford.edu/archives/sum2003/entries/punishment/>

38 Punishment, Communication, and Community, p. 42.

39 Elizabeth Fraser, 'Communitarianism', in The Oxford Companion to Philosophy.

40 Punishment, Communication, and Community, p. 47.

41 Punishment, Communication, and Community, pp. 47-8.

42 On Political Means and Social Ends, mainly the papers on Mill and Rawls.

43 Punishment, Communication, and Community, Ch. 5.

HOME to Det & Free website front page
HOME to T.H. website front page