Melbourne University Law Review
MIRKO BAGARIC[*] AND KUMAR AMARASEKARA[†]
[Retributivism has been the dominant theory of punishment in the Western world for the past few decades. This is despite the fact that there is no single widely accepted retributive theory or defining characteristic shared by retributive theories. This paper analyses the most influential retributive theories of punishment: intrinsic retributivism, the unfair advantage theory and the theories advanced by Antony Duff and Andrew von Hirsch, and concludes that none of the theories is able to provide a justification for punishment. The most pervasive flaw with retributive theories is that they are unable to substantiate the link between crime and punishment without relying on consequentialist considerations. The shortcomings of the retributive theories are so grave that we should abandon retributivism as a worthwhile legal and philosophical concept and look elsewhere for a justificatory theory of punishment. To this end, it is contended that the utilitarian theory of punishment is the soundest theory of punishment.]
Retributivism, under the banner of just deserts, has replaced utilitarianism as the prime philosophical underpinning of punishment in the Western world. It is also generally perceived that the philosophical leaning towards retributivism has permeated most sentencing systems, despite the gulf that normally exists between theories of punishment and sentencing practice and the tendency of the sentencing systems of most jurisdictions not to adopt a primary rationale for sentencing.
This paper examines the leading contemporary retributive theories and argues that despite the widespread appeal of retributivism, it is unable to justify the link between crime and punishment. This forms a central plank of the wider thesis that there should be a reversion to a utilitarian-based system of punishment and sentencing. As a matter of logic, this requires not only a rejection of retributivism, but also a positive case in favour of the utilitarian theory of punishment, which would need to address many of the damaging criticisms that have been levelled against it. Considerations of space prevent this latter task being undertaken extensively at this point. However, due to the persuasiveness of these attacks, we provide an overview of the manner in which they can be met. The utilitarian theory of punishment is further fortified by illustrating the manner in which it deals with many of the problems plaguing retributive theories.
Before turning to substantive issues, we shall first clarify some definitional matters and then outline the reasons for the popularity enjoyed by retributivism. In Part IV of this paper we consider how the utilitarian theory can respond to the most damaging criticisms that have been made against it. After briefly considering other non-retributive theories of punishment which have been proffered, the final four substantive parts of this paper discuss why we believe that retributivism does not justify the institution of punishment.
A vast array of theories of punishment have been advanced which are classified as retributive. As a result of the diversity of these theories, it has proven remarkably difficult to isolate a distinctive feature of theories which carry the tag. Nonetheless, all retributive theories assert that offenders deserve to suffer and that the institution of punishment should inflict the suffering they deserve; however, they provide vastly divergent accounts of why criminals deserve to suffer.
A cornerstone of many retributive theories is the proportionality thesis, which asserts that the punishment must be equivalent to the level of wrongdoing. This is a claim enthusiastically endorsed by Andrew von Hirsch, one of the main contemporary proponents of retributivism, who asserts that:
Sentences according to [the just deserts] theory are to be proportionate in their severity to the gravity of the criminal’s conduct. ... In such a system, imprisonment, because of its severity, is visited only upon those convicted of serious felonies. For non-serious crimes, penalties less severe than imprisonment are to be used.
He argues that the basis for proportionality is essentially that punishment is the vehicle for condemnation and, as a matter of fairness, punishment must be proportionate, since the severity of the sanction expresses the stringency of the blame. However, the view that punishment should be commensurate with the seriousness of the offence does not provide a justification for punishment, rather, it too simply acts as a restraint on it. Such a claim is also not distinctly retributivist. Utilitarians have also been known to invoke the principle of proportionality.
It has also been asserted that a distinctive feature of retributivism is that punishing criminals is itself just: it cannot be inflicted as a means of pursuing some other aim. However, even a cursory consideration of some of the leading contemporary retributive theories reveals that few do not ultimately advert to extraneous reasons to justify punishment. Only one retributive theory (intrinsic retributivism) claims that punishment is justified because it is intrinsically good to punish wrongdoers. All other retributive theories appear to rely, at least partially, on the instrumental effects of punishment to justify the practice. These include the capacity for punishment to convey blame or reprobation, to induce repentance, self-reform, reparation and reconciliation, or to restore the fair balance of benefits and burdens which is disturbed by crime.
Accordingly, it is difficult to identify a principle that represents a retributive pedigree. The true picture seems to be that there are many different theories of punishment which could be labelled as retributive. There is no clear positive and internal unifying principle running through all of them. Nevertheless, they do have at least one thing in common: they are not utilitarian. Thus, retributive justifications for punishment do not turn on the likely achievement of consequentialist goals: punishment is justified even when ‘we are practically certain that attempts [to attain consequentialist goals, such as deterrence and rehabilitation] will fail.’ It may well be that the negation of utilitarianism is the distinctive badge worn by retributive theories. This alludes to another characteristic feature of retributive theories: they are essentially backward-looking; punishment is an appropriate response to a past offence, irrespective of other incidental effects. This is in contrast to utilitarianism, which is concerned only with the likely future consequences of imposing punishment. The contrast with utilitarianism also explains why theories which rely on factors or virtues outside the parameters of the act of punishment itself are still regarded as retributive. The virtues invoked are not, at least expressly, consequentialist in nature, but instead are those commonly associated with a deontological account of morality.
It is unclear whether this captures the full contrast between retributivism and utilitarianism. However, for the purpose of this paper the precise definition of retributivism is not critical, for it is argued that all of the leading retributive theories are unsound.
The hallmarks of retributivism are, however, made clearer by further contrasting it with the utilitarian theory of punishment.
Utilitarianism is the theory that the morally right action is that which produces the greatest amount of utility. The utilitarian theory of punishment is merely an application of the general utilitarian theory of morality to the specific issue of punishment. Utility has been defined in numerous ways and although in this domain it does not matter significantly which version of utilitarianism is adopted, we shall adopt what we consider to be the most persuasive and coherent version of utilitarianism: hedonistic act utilitarianism. This provides that the utility that should be maximised is happiness or pleasure, which is the sole intrinsic good, and that pain is the sole inherent evil. Henceforth, it is referred to simply as utilitarianism.
The utilitarian starting point regarding punishment is to consider the most direct and immediate effect of punishment and, from this perspective, punishment is a bad thing because it causes unhappiness to the offender. It is only justified because of the wider contingent benefits it produces, which it is felt outweigh the bad consequences. The good consequences of punishment, which are thought to outbalance the suffering inflicted on the offender, include discouraging potential offenders from committing crimes in the first place and, once the offender is apprehended, discouraging the offender from re-offending, rehabilitating him or her and, where necessary, incapacitating the offender. If there are several forms of punishment which produce the same good consequences, we must choose the one which imposes the least unpleasantness upon the offender. Thus, unlike retributivism, the utilitarian theory of punishment is forward-looking: the commission of a criminal act does not justify punishment; rather, punishment is only warranted if some good can flow from it.
The dominance of retributivism stems, first, from a vacancy which was created in the field of a justificatory theory of punishment, after widespread criticisms were levelled at the utilitarian theory of punishment. Secondly, retributivism presented a neat and simple solution to many of the perceived problems which plagued utilitarianism. Utilitarianism lost many supporters as a result of two broad types of attack: one pragmatic and the other theoretical.
The first criticism was the perceived failure of penal practice and the treatment-based goals of sentencing to measure up to the prime utilitarian objectives of deterrence and rehabilitation. Research findings relating to rehabilitation, in particular, were at one point so depressing that a ‘nothing works’ attitude was pervasive. Given the apparent failure to achieve such lofty and ambitious sentencing goals, the natural inclination was to set the sights on aims which were far more achievable. Future-oriented goals of punishment, such as rehabilitation and deterrence, made way for backward-looking considerations, where the main goal was to ensure that criminals got what they deserved. Thus the aim of doing more good through the prison system was replaced by the goal of doing justice, where justice broadly equated to imposing punishment that was proportionate to the severity of the crime. On this rationale, so long as the punishment fitted the crime, or almost did so, the sentencing system was a ‘success’, irrespective of the indirect consequences stemming from it. Retributivism was the clear beneficiary of such an approach. A retributive sentencer can be sure that the sentence will achieve the purpose for which it is imposed — punishing the offender to the extent of his or her wrongdoing. The simplicity and appeal of retributivism is evident in Mabbott’s justification for punishing students breaking a rule compelling attendance at chapel:
Many of those who broke this rule broke it on principle. I punished them. I certainly did not want to reform them; I respected their character and their views. I certainly did not want to drive others into chapel through fear of penalties. ... My position was clear. They had broken a rule; they knew it and I knew it. Nothing more was necessary to make punishment proper.
The decline of utilitarian punishment and sentencing was also greatly accelerated by the fact that, at the theoretical level, there was a move towards rights-based moral theories and widespread support for arguments that utilitarianism commits us to abhorrent practices, such as punishing the innocent. The main argument in support of rights-based moral theories is aptly stated by John Rawls, who claims that only rights-based theories take seriously the differences between human beings and, in addition, protect certain rights and interests which are so paramount that they are beyond the demands of net happiness.
Charges of this nature have been extremely influential. In the last half of this century, following the Second World War, there has been an immense increase in ‘rights talk’, both in sheer volume and in the number of supposed rights. Assertion of rights has become the customary means of expressing our moral sentiments: ‘there is virtually no area of public controversy in which rights are not to be found on at least one side of the question — and generally on both’. The domination of ‘rights talk’ is such that it is accurate to state that ‘[t]he doctrine of human rights has at least temporarily replaced the doctrine of maximising utilitarianism [sic] as the prime philosophical inspiration of political and social reform’.
The narrower theoretical objection to utilitarian punishment — that it permits punishment of the innocent — has been so persuasive that it alone has led many to reject utilitarianism as a general theory of morality. The real force of this objection is found in the more general criticism that utilitarianism fails to protect basic individual rights and interests, and since it does not prohibit anything per se, may lead to horrendous outcomes. This contrasts with the apparent ease with which retributivism deals with this dilemma: by confining punishment only to wrongdoing, it ensures that no person is sacrificed for the good of others.
It is also often claimed that, under a retributive regime, offenders are not treated as more (or less) blameworthy than is warranted by the seriousness of the offence and, hence, the principle of proportionality is thought to sit most comfortably in a retributive system of punishment. The appeal of retributivism is also said to lie in the simple fact that it accords with our common sense notions of equity.
It has been argued at length elsewhere that the criticisms which have been levelled against utilitarianism are unpersuasive. However, given the influential nature of these attacks, an outline of what they are and how they can be countered is provided.
The supposed good consequences that utilitarians normally invoke to justify punishment are incapacitation, rehabilitation and deterrence. The empirical evidence relating to the first two objectives is still not promising.
Incapacitation is not an objective of punishment and sentencing, but rather a means of protecting the community. Its efficacy cannot be judged by the height of the prison wall. In order for incapacitative sentences actually to protect the community, it must be demonstrated that the offenders who are subject to such sanctions would have offended if they had not been restrained.
To this end, the evidence suggests that we cannot distinguish with any meaningful degree of confidence those offenders who will re-offend from those who will not. Studies have shown that in predicting dangerousness, psychiatrists are wrong nearly 70 percent of the time. Despite some initial optimism, predictive techniques, which draw on more concrete supposed risk factors, such as employment history and the age at which a person first started offending, have also had a low success rate.
The fact that a person has previously committed a serious offence is a particularly poor guide to identifying future serious offenders. A recent study which tracked the offending behaviour of 613 offenders released from prison in New Zealand for a two and a half year period revealed that those who would be classified as serious offenders were no more likely to receive a further conviction within two and a half years after release than ordinary offenders and were, in fact, less likely to be imprisoned within that time. It was also found that, of all serious offences committed by the entire sample group, the vast majority was committed by offenders who were imprisoned for non-serious offences. In total, only 30 of the sample of 613 offenders committed a serious offence within the follow-up period. It was noted that there is very little hope of achieving crime control through altering the definition of a serious offence.
The most recent extensive review of incapacitation research notes that current predictive techniques (including those which go beyond simply looking at prior criminality to such things as employment history and so on) ‘tend to invite overestimation of the amount of incapacitation to be expected from marginal increments of imprisonment’. Our ability to predict which offenders are likely to re-offend is so poor that it has been estimated that the increase in crime rate if prison use was reduced or abolished could be around five percent.
The evidence concerning rehabilitation is more promising than was the case a quarter of a century ago. In a recent wide-ranging review of the published studies in rehabilitation (which compared the recidivism rate of offenders who were subject to rehabilitative treatment with those who were not), Howells and Day conclude that there has been a significant degree of success with cognitive-behavioural programs. These programs target factors that are (presumably) changeable and are directed at the ‘criminogenic needs’ of offenders, that is, factors which are directly related to the offending, such as anti-social attitudes and a lack of problem-solving skills and self-control. Promising programs have been developed in the areas of anger management, sexual offending and drug and alcohol use. These appear to be more successful than programs based on, for example, confrontation or direct deterrence, physical challenge or vocational training. Three judges in Missouri have even imposed transcendental meditation programs as part of the probation conditions of minor offenders, apparently with great success.
Despite such developments, the most that can be said confidently at this point regarding the capacity of criminal punishment to reform is that there is some evidence that it will work for a small portion of offenders and that there is no firm evidence which shows that it cannot work for the majority of offenders. However,
treatments do not ... exist ... that can be relied upon to decide sentences routinely — that can inform the judge, when confronted with the run-of-the-mill robbery, burglary, or drug offense, what the appropriate sanction should be, and provide even a modicum of assurance that the sanction will contribute to the offender’s desistence from crime.
A more fundamental problem with invoking rehabilitation as an objective of punishment is that rehabilitation (at least of the type which appears to be having some success) and punishment may be inconsistent. Punishment by its very nature must hurt. There seems to be an inherent contradiction between deliberately subjecting someone to pain and at the same time trying to get him or her to see things your way. The more tolerant, understanding and educative we are in trying to facilitate attitudinal change in others, the closer we come to providing them with a social service. For example, cognitive-behavioural programs focus on the needs of offenders. They attempt to meet these needs by providing education and counselling which is aimed at reshaping the beliefs, attitudes and values of offenders and improving their problem-solving capacity, in order that they no longer engage in criminal behaviour. Such programs seem to work better in community settings than when delivered in institutions. There is very little difference between such programs and educational courses within the community (which are enthusiastically undertaken by many law-abiding members of the community). This is all the more so, given that it is a feature of many rehabilitative ‘sanctions’ that they cannot be ‘imposed’ unless the offender consents to them. By making the interests of the offender paramount, modern rehabilitative programs are more akin to welfare services than punitive sanctions. In order for the goal of rehabilitation to justify punishment, it must be shown at minimum that reform is attainable in a setting that is directed primarily at imposing unpleasantness on the offender. There is no evidence to support this. One suspects that this tension will prove irreconcilable.
There are two aspects to deterrence. Specific deterrence aims to discourage crime by punishing offenders for their transgressions and thereby convincing them that crime does not pay. General deterrence seeks to dissuade potential offenders from engaging in unlawful conduct by using the threat of anticipated punishment and by illustrating the unsavoury consequences of offending. Deterrence is clearly a forward-looking sentencing objective, focusing solely on preventing harm, by either punishing the offender or dissuading others who come to know of the punishment. For Jeremy Bentham, general deterrence was the primary good consequence of punishment:
General prevention ought to be the chief end of punishment, as it is its real justification. ... [W]hen we consider that an unpunished crime leaves the path of crime open, not only to the same delinquent, but also to all those who may have the same motives and opportunities for entering upon it, we perceive that the punishment inflicted on the individual becomes a source of security to all. That punishment which, considered in itself, appeared base and repugnant to all generous sentiments, is elevated to the first rank of benefits, when it is regarded not as an act of wrath or of vengeance against a guilty or unfortunate individual who has given way to mischievous inclinations, but as an indispensable sacrifice to the common safety.
It is inordinately difficult to obtain information about the effectiveness of sanctions in deterring offenders from committing offences at the expiry of a sanction. Offenders may not re-offend for numerous reasons, apart from the fear of being subject to more punishment. The offending may have been a one-off in any event, a suitable opportunity to offend may not again present itself, or rehabilitation may have occurred. The likelihood of recidivism is also reduced if the offender gets a job or simply ‘grows up’. Empirical evidence strongly supports the view that criminal behaviour is a young man’s endeavour. Studies in the United States have shown that the peak age for violent offences is 18, the peak age for property offences is 16 and the rate of offending typically declines at the age of 30.
However, the evidence that is available supports the view that severe punishment (namely imprisonment) does not deter offenders: the recidivism rate of offenders does not vary significantly, regardless of the form of punishment or treatment to which they are subjected.
In terms of analysing the evidence concerning the efficacy of general deterrence, there are two broadly different levels of inquiry. Marginal deterrence is directed at determining whether a direct correlation exists between the severity of the sanction and the prevalence of an offence. Absolute deterrence relates to the threshold question of whether there is any connection between criminal sanctions and criminal conduct.
There have been numerous attempts to ascertain if there is a connection between penalty levels and crime rate. There is no firm evidence that increasing penalty levels results in a reduction in crime. Following a comprehensive review of the available evidence, the report of a panel of the National Research Council in the US concluded that the research evidence that does exist regarding marginal deterrence is, overall, inconclusive:
We cannot yet assert that the evidence warrants an affirmative conclusion regarding deterrence. We believe scientific caution must be exercised in interpreting the limited validity of the available evidence and a number of competing explanations for the results. Our reluctance to draw stronger conclusions does not imply support for the position that deterrence does not exist, since the evidence certainly favours a proposition supporting deterrence more than it favours one asserting that deterrence is absent.
In a similar vein, Franklin Zimring and Gordon Hawkins stated that:
Studies of different areas with different penalties, and studies focusing on the same jurisdiction before and after a change in punishment levels takes place, show rather clearly that the level of punishment is not the major reason why crime rates vary. In regard to particular penalties, such as capital punishment as a marginal deterrent to homicide, the studies go further and suggest no discernible relationship between the presence of the death penalty and homicide rates.
In a more recent study it was postulated that an escalation in penalty level would result in an increase in offending behaviour. The well-publicised Crime (Serious and Repeat Offenders) Sentencing Act 1992 (WA) was targeted at reducing the number of high-speed pursuits involving stolen vehicles by significantly increasing the penalties for such offences. Following its introduction there was a substantial increase in the rate of offences (such as motor vehicle thefts) which were a trigger for high-speed pursuits.
The news on absolute deterrence, however, is much more positive. There have been several natural social experiments where there has been a drastic reduction in the likelihood (perceived or real) that people would be punished for criminal behaviour. The key thing about these events is that the change happened abruptly and the decreased likelihood of the imposition of criminal sanctions was apparently the only changed social condition.
Perhaps the clearest instance of this is the police strike in Melbourne in 1923, which led to over one third of the entire Victorian police force being sacked. Once news of the strike spread, mobs of thousands of people poured into the city centre and caused widespread property damage, looted shops and engaged in other acts of civil disobedience, including assaulting government officials and torching a tram. The civil disobedience lasted for two days and was only quelled when the government enlisted thousands of citizens, including many ex-servicemen, to act as ‘special’ law enforcement officers. The events of those two days were in complete contrast to the normally law-abiding conduct of the citizens of Melbourne. Other similar examples include the police strike in Liverpool in 1919 and the internment of the Danish police force in 1944.
The results of these social observations also counter the argument that deterrence does not work because it only addresses those who do not need it: that is, law-abiding citizens. It would seem that there are many citizens who would flaunt the law readily if they thought they could do so with impunity.
The view that there is a general link between punishment and the crime rate is also in accordance with econometric research which shows that there is an inverse relationship between the incidence of violent offences and the use of imprisonment, but that the length of imprisonment imposed is irrelevant.
Thus, although deterrence does work, the evidence suggests that there is only a limited direct connection between crime and some penalty, and that there is no connection between crime and penalty level.
The critics are only partially right when they claim that there is little evidence to support the efficacy of punishment to attain the utilitarian objectives of sentencing. On the basis of the current empirical evidence, the objectives of incapacitation, specific deterrence and rehabilitation cannot be invoked by the utilitarian to justify punishment. Incapacitation is flawed since we are very poor at predicting which offenders are likely to commit serious offences in the future. There is nothing to suggest that offenders who have previously been punished are less likely to re-offend; thus, there is no basis for pursuing the goal of specific deterrence. Also, there are no far-reaching rehabilitative techniques which have proven to be successful. Even more telling is the fact that the goals of punishment and rehabilitation may be internally inconsistent.
However, experience shows that absent the threat of punishment for criminal conduct, the fabric of society would readily dissipate. Crime would escalate and overwhelmingly frustrate the capacity of people to lead happy and fulfilled lives. Thus, while there is only one objective of punishment which the utilitarian can invoke, this is more than sufficient to justify the practice of state-imposed unpleasantness on those who violate the criminal law.
Despite this, there is insufficient evidence to support a direct correlation between higher penalties and a reduction in the crime rate. This means that while deterrence justifies punishing offenders, it is of little relevance in fixing the amount of punishment. This must be done by reference to other utilitarian ideals, and to this end the principle of proportionality is likely to be the guiding determinant.
The most telling theoretical objection to utilitarianism is that it permits punishment of the innocent. A similar, if insensitive, illustration of this is H J McCloskey’s small town sheriff example:
Suppose a sheriff were faced with the choice of either framing a Negro for a rape which had aroused white hostility to Negroes (this Negro believed to be guilty) and thus preventing serious anti-Negro riots which would probably lead to loss of life, or of allowing the riots to occur. If he were ... [a] utilitarian he would be committed to framing the Negro.
A common utilitarian response to this dilemma is that such examples are impossible in the real world and, hence, need not be addressed. Punishing the innocent may at times provide short-term benefits, such as securing social stability, but these are always more than offset by the likelihood of greater long-term harm as a result of the loss of confidence in the legal system and the associated loss of security to all members of the community who will fear that they may be the next person framed. However, with only a little imagination McCloskey’s example can be tightened, for instance, by introducing considerations that significantly reduce or totally obviate the possibility of disclosure, so that the only logical utilitarian conclusion is to punish the innocent. Even if the process of modifying the examples appears to far remove them from the real world, it is still a situation which the utilitarian must address. As C L Ten notes, ‘fantastic examples’, as he labels them, which raise for consideration fundamental issues, such as whether it is proper to punish the innocent, play an important role in the evaluation of moral theories, since they sharpen the contrasts between them and illuminate the logical conclusions of the respective theories. In this way they test the true strength of our commitment to the theories. Thus, fantastic examples cannot be dismissed summarily on the basis that they are ‘simply’ hypothetical.
The more promising utilitarian response is not to attempt to deflect or avoid the conclusion that there may be some extreme situations where utilitarianism commits us to punishing the innocent. Rather, it is to accept this outcome and to contend that, as horrible as this may seem on a pre-reflective level, it is not a matter which, on closer consideration, really insurmountably troubles our sensibilities to the extent that it would necessarily render flawed any theory which approves of such an outcome. By drawing comparisons with other situations in which we take the utilitarian option, it is contended that punishing the innocent is not a practice which is necessarily unacceptable.
The view that punishing the innocent is the morally correct action in some circumstances is consistent with and accords with the decisions we as individuals and societies as a whole readily have made and continue to make when faced with extreme and desperate circumstances. We come to grips with the fact that our decisions in extreme situations will be compartmentalised to desperate predicaments and will not have a snowball effect, and serve henceforth to diminish the high regard we normally have for important individual concerns and interests. When we are between a rock and a hard place we do and should — though perhaps somewhat begrudgingly — take the utilitarian option. In the face of extreme situations we are quite ready to accept that one should, or even must, sacrifice oneself or others for the good of the whole.
For example, in times of war we not only request our strongest and healthiest citizens to fight to the death for the good of the community, but we often demand that they do so under threat of imprisonment or even death. Quite often they must battle against hopeless odds, in circumstances where we are aware that in all probability they will not return. We compel them to give their lives, not because they want to, not because they are bad, but merely because we believe it would be good for the rest of us — classic utilitarian reasoning. Faced with the reality of the decisions we do make in such horrible situations, the examples proffered against utilitarianism and the terrible things it entails, such as punishing the innocent, lose their bite. Horrible situations make for appalling decisions whichever way we turn, but when pressed we do make the utilitarian choice because of our lack of true commitment to any higher moral virtue. By opting for the utilitarian line we are soothed by the one saving grace: at least the level of harm has been minimised. When the good of many or the whole is under significant threat, we have no difficulty selecting certain classes of innocent individuals, whose only ‘flaw’ is their sex, state of health and date of birth, to sacrifice themselves for the rest of us. Their protests that they should not be compelled to go because it impinges on their civil, legal or human rights, such as the rights to life and liberty, or their desperate appeals to other virtues such as justice or integrity, fall on deaf ears. For this is serious stuff now — our lives (or other important interests) are at stake. Such appeals should be saved for rosier times.
The decisions we make in a real life crisis are the best evidence of the way we actually prioritise important competing principles and interests. Matters such as rights and justice are important, but in the end are subservient to, and make way for, the ultimate matter of significance: general happiness. Bad as it seems, ‘framing the Negro’ and imprisoning the innocent are certainly no more horrendous than the decisions history has shown we have made in circumstances of monumental crisis.
A pointed example is the decision by the British Prime Minister of the day, Winston Churchill, to sacrifice the lives of the residents of Coventry in order not to alert the Germans that the English had deciphered German radio messages. On 14 November 1940 the English decoded plans that the Germans were about to air bomb Coventry. If Coventry were evacuated or its inhabitants advised to take special precautions against the raid, the Germans would have known that their code had been cracked and the English would have been unable to obtain future information about the intentions of its enemy. Churchill elected not to warn the citizens of Coventry, and many hundreds were killed in the raid which followed. Their lives were sacrificed in order not to reveal a secret that would hopefully save many more lives in the future. Significantly, such decisions subsequently have been immune from widespread or persuasive criticism. This shows not only that when pressed we do take the utilitarian option, but also that it is felt that this is the option we should take.
What we actually do does not justify what ought to be done. Morality is normative, not descriptive in nature: an ‘ought’ cannot be derived from an ‘is’. Nevertheless, the above account is telling because of the forcefulness of the ‘punishing the innocent’ objection. Punishing the innocent, according to this objection, supposedly so troubles our moral consciousness that utilitarianism can therefore be dismissed. The outcome, it is argued, is so horrible that we are forced to say ‘there must be a mistake somewhere’. However, the objection loses its force when it is shown that punishing the innocent is in fact no worse than other activities we condone.
It has been pointed out that it is not only a utilitarian system of punishment that may permit punishment of the innocent:
Retributivists who advocate punishment are relevantly like utilitarians who sacrifice the welfare of innocents for the greater good, since retributivists are willing to trade the welfare of the innocent who are punished by mistake for the greater good of the punishment of the guilty. While never intending to punish the innocent, they nevertheless do not choose to withdraw their support for arrangements that have this result.
There are several possible retributivist responses to this problem, although, in our view, none is convincing.
It is inevitable, given the fallibility of any institution, that any criminal justice system will at times inflict punishment on the innocent. This difficulty could largely be circumvented by increasing the number and level of safeguards in the criminal justice process. For example, the standard of proof could be raised from beyond reasonable doubt to, say, beyond any possible doubt; admissible evidence could be limited to direct observations of the relevant act; and a confession could be made a mandatory precondition to a finding of guilt.
However, such a response is only open to retributivists if ensuring that the innocent are not punished does not hinder the main retributivist aim. Radically increasing the number of legal safeguards in criminal cases could result in fewer guilty people being punished, and consequently an increase in the amount of crime.
A common retributive response to the problem of punishing the innocent is that offered by Antony Duff, who denies that punishing the innocent is a concern for the retributivist, since, unlike the utilitarian situation, punishment of the innocent is not intended and occurs despite the aims of a retributive system of punishment. The credibility of this response turns on the persuasiveness of the distinction between consequences which are intended and those which are merely foreseen.
Underpinning Duff’s argument is the doctrine of double effect, which provides that it is morally permissible to perform an act which has two effects, one good and one evil, where the following elements are satisfied: the good consequence is intended and the bad consequence is merely foreseen, the bad consequences are insignificant when compared with the good consequences, and those consequences occur almost simultaneously.
The doctrine has a rich history and is frequently appealed to as a purported justification for acts or practices which produce foreseen undesirable consequences. For example, this is the reason why it is permissible, supposedly, to bomb an enemy’s ammunition factory in wartime, even though it will result in the certain death of civilians, and why it is justifiable to abort a foetus where this is necessary to save the mother and why self-defence is legitimate. In the case of euthanasia, the doctrine is employed as a justification for alleviating pain by increasing doses of pain-killers, even when it is known that this will result in death, because the intention is to reduce pain, not to kill.
The legal status of the doctrine is unclear. In R v Nedrick the House of Lords held that foresight, even of near certainty, was not the same as intention, whereas in Hyam v The Queen Lord Hailsham was of the view that one who blows up an aircraft in order to obtain money intends to kill. However, the courts have endorsed the doctrine in relation to euthanasia. In Airedale National Health Service Trust v Bland it was held that it is permissible to relieve suffering even if the measure incidentally shortens life. This view has been endorsed in a number of cases.
The moral significance of the doctrine of double effect is much in dispute. Jonathon Glover gives the example of a terrorist who, for the purpose of making a (legitimate) political protest, throws a bomb into a crowd and kills several people. He correctly points to the difficulty in ascertaining whether the deaths are intentional or merely foreseen. The above examples illustrate that the inevitability of the deaths cannot be used to impute intention, for the doctrine provides that foreseen consequences which are certain need not be counted as intentional. Thus, the fact that the terrorist is possibly more certain to kill innocent people than the institution of punishment is to punish innocent people is irrelevant.
It is also beside the point that the institution of punishment does not aim to punish specific innocent individuals. For not only would the terrorist be pleased if no person were killed, but as far as he or she is concerned the crowd consists of random unidentified targets rather than particular, predetermined victims. Thus, there appears to be no principled reason to maintain that the terrorist intends to kill, whereas the institution of punishment does not intend to punish the innocent: in both instances if the respective objectives could be achieved without the harmful by-products the agents would be pleased. This alludes to the central flaw in the doctrine of double effect: it is not possible to provide a general account of the distinction between what is intended and what is merely foreseen which applies in all circumstances. It is illusory to claim that intentions are divisible along the lines of good and bad consequences of an act.
The preferable view is that there is no inherent distinction between consequences which are intended and those which are foreseen. We are responsible for all the consequences which we foresee but nevertheless elect to bring about. Whether or not we also ‘intend’ them is irrelevant. Underlying the doctrine of double effect, and the only coherent basis for distinction adverted to by the doctrine, is nothing more than the consequentialist view that it is permissible to do that which is ‘merely foreseen’ if the adverse consequences of the act are outweighed by the ‘intended’ good consequences. Utilitarianism deals with the difficulties that are sought to be overcome by the doctrine in a far more comprehensible and straightforward manner. The reason that neither the doctor who administers a lethal dose of pain-killers nor the legal system which punishes the innocent (believed guilty) is blameworthy has nothing to do with the difficulty in determining what is intended rather than foreseen. It simply follows because in all the circumstances the good consequences outweigh the bad. Further, from the perspective of the innocent person who is punished, it certainly does not matter whether his or her punishment was intentional or merely foreseen: it hurts just the same. Notwithstanding this, an institution which causes such hurt is still morally justifiable, because it leads to a situation which is happier overall than that which results from the abolition of punishment.
In the end the motivation for the doctrine of double effect seems to be to provide a means for deontological theories which employ notions of absolute (or near absolute) rights to deal with the difficult, but inevitable, situations where there are conflicts between different rules or rights, or even different applications of the same rule or right. The doctrine maintains absolutism by utilising the fiction of merely foreseen consequences and absolving liability for them.
There may yet be another way in which the retributivist may attempt to defend a system of punishment which, unfortunately but invariably, will result in the punishment of some innocent people. This adverts less crudely to the distinction discussed earlier regarding the identity of victims who are incidentally harmed as a by-product of what is thought to be a generally desirable act, and invokes the Kantian concept of means and ends.
The nature of the distinction between the victims who are harmed intentionally and those who are harmed unintentionally is illustrated by the following example. It is necessary to build a bridge between two suburbs. Two different types of bridges are possible. If proposed bridge A is built, actuarial studies show that it is certain that two people will die during the construction. If bridge B is constructed, it is known in advance that a particular worker will die. It is contended that the utilitarian on this information alone would elect route B. On the other hand, a powerful deontological argument can be made in favour of bridge A, because unlike in case B, no individual is being used simply as a means for a particular end. This follows from the fact that each person who is involved in the project or is in some way affected by it may too ultimately benefit from the project. For example, he or she may use the bridge or be paid a salary for working on its construction. It may be argued that the terrorist example is analogous to situation B and the retributive system of punishment to situation A. Although the terrorist kills victims who are unknown to him or her, they are nevertheless specific people whose identity is ascertainable at the time of the act and they have no prospect of benefiting from the legitimate protest. This is not so in the case of a retributive system of punishment which unintentionally punishes an innocent person. The identity of the (innocent) ‘offender’ is not known at the time of conviction and sentence, and may never be known. Even though the innocent who are punished ultimately do suffer, they are part of a general practice through which they too may have prospered. In this sense, so the argument runs, they are not sacrificed for the good of the whole.
However, even putting to one side the difficulties associated with the means and ends distinction, this retributive approach to the dilemma is also unsatisfactory. At the time an innocent person is punished there is always at least one person who is aware of the injustice: the true offender. It is not the point that the system is oblivious to the innocence of the ‘offender’ at the time of conviction and sentence. If the system was really concerned with the unfairness it would have taken measures to avoid the predicament by implementing safeguards, of the type mentioned earlier, to prevent wrongful conviction. By persisting with such a defence of their theory, retributivists are expressing either feigned concern or blissful ignorance. Even more generally, it is immaterial that the offender could have potentially benefited from the institution of punishment. If the offender did not, he or she is a victim of the institution, and it is unrealistic to expect meaningful solace to be attained through such unrealised potential: in any meaningful sense of the word, the innocent person is being sacrificed for the good of the whole. Thus, the retributivist is less equipped to deal with the problem of punishing the innocent than the utilitarian.
The other broader theoretical objection to utilitarianism, mentioned above, is that it is inconsistent with the concept of rights. This is a criticism of utilitarian theory in general, rather than being peculiarly applicable to the practice of punishment. Accordingly, this criticism is not challenged in detail at this point, save to mention that it has been argued that deontological rights-based theories are intellectually vacuous. Despite their dazzling veneer, they are unable to provide convincing answers to central issues such as: what is the justification for rights? How can we distinguish real from fanciful rights? Which right takes priority in the event of conflicting rights? It has been further argued that rights have a place in a utilitarian ethic and, what is more, it is only against this background that rights can be explained and their source justified.
Accordingly, there are solid grounds to fend off the attacks that have historically proven so damaging to the utilitarian theory of punishment. This opens the door for utilitarianism to present itself again as the dominant justificatory theory of punishment. To capitalise on this opening, it is necessary to topple the prevailing theory. Against this background, it is argued that despite the perceived advantages of retributivism, all retributive theories have severe shortcomings and, ultimately, are unable to provide a coherent account of why criminals should be punished.
Utilitarianism and retributivism do not exhaust the range of justificatory theories of punishment. Hence, before tackling retributivism, some of these other theories will be discussed briefly.
Apart from the retributive and utilitarian theories, probably the most influential theory of punishment advanced this century is H L A Hart’s theory of punishment. Hart believes that it is untenable to proffer a theory of punishment which is either absolutely utilitarian or retributive; rather, a compromise is necessary. For Hart, the utilitarian aim of crime reduction provides the general justification of punishment (he terms this the ‘general justifying aim of punishment’), however, this is mitigated by the notion of justice (called ‘retribution in distribution’) which determines two aspects of punishment: who should be punished and how much they should be punished.
This is ultimately the theory endorsed by the New South Wales Law Reform Commission, which felt that it is
impossible to identify among the varying philosophical approaches to punishment a dominant rationale which should or could rationally guide the reform of sentencing law ... The court must impose a sentence which emerges as a compromise between the competing factors, regardless of which punishment theory is currently in vogue.
Hart’s prime reason for invoking the principles of justice to act as side constraints on the ‘general justifying aim’ is that he believes that the principles of justice, such as the proscriptions against sacrificing the innocent and punishing family members of offenders vicariously, cannot be derived from utilitarianism (or for that matter from retributivism). However, as is discussed above, ostensibly abhorrent practices such as punishing the innocent are not so shocking at the pre-philosophical inquiry that we ought to permit the intuitive unease stemming from them to trump principle. The utilitarian theory, in the end, soundly deals with such dilemmas. This being so, there is no reason (let alone justification) for expediently invoking external principles in order to appease our intuition. To do so would not only result in a loss of doctrinal coherency, but as with all compromise theories, leave one with little guidance on critical matters. For example, despite the emphasis of Hart’s theory on the virtue of justice, he accepts that in ‘extreme cases’ it may be right to violate the principles of justice; however, no meaningful indication is given when such a point is reached. Ultimately, all theories must invoke a principle as being cardinal. In light of this, the search for a single coherent theory of punishment should not be abandoned readily.
Perhaps the most influential non-retributive contemporary theory of punishment is the ‘dominion’ theory advanced by John Braithwaite and Philip Pettit. Their theory is consequentialist, but the utility which they believe ought to be maximised is not happiness, but what they term ‘dominion’. This is a republican concept of liberty where people are free from non-interference by others by virtue of the protection of the law and related social institutions, and where people enjoy equality before the law. Braithwaite and Pettit believe that the sole objective of the criminal justice system is to maximise dominion. The only acts which should be subject to criminal punishment are those which violate the dominion of others. Punishment should only be imposed in order to promote liberty. This can be done via several different means, such as incapacitation and deterrence. However, in their view it is best done through shaming or censuring criminals. In contrast to von Hirsch’s theory, reprobation is not an integral aspect of punishment, but has an instrumental role to the extent that it is useful in promoting observance of the law.
The motivation for the switch from happiness to dominion appears to be essentially twofold. They feel that hedonistic utilitarianism accords insufficient weight to individual interests (thus, a measure is chosen which gives greater emphasis to the individual’s capacity for personal choice) and they believe that their theory circumvents the problem of punishing the innocent.
As we have already seen, hedonistic utilitarianism is able to deal with the latter dilemma. Further, it has been argued elsewhere that hedonistic utilitarianism can answer the criticism that it cannot accommodate the separateness of persons. Moreover, dominion is a poor substitute for happiness as the ultimate virtue which ought to be maximised. It is implausible to suggest that all important human interests are reducible to, or derivative from, dominion.
Braithwaite and Pettit argue that interference with the dominion of an offender is only justified when he or she violates the dominion of another. Infringement of other interests (such as physical integrity) does not justify punishment unless dominion is also diminished. However, as has been pointed out by Ten, good things do not always go together. Rape and murder certainly violate dominion, but expressing the wrongness of such crimes in terms of the violation of one’s dominion fails to capture adequately the force of the objections to such conduct. It is even more telling that Braithwaite and Pettit are unable to explain intelligibly why it is wrong to inflict pain and suffering on sentient beings that have no concept of dominion, such as animals. This is in sharp contrast to the hedonistic utilitarian: ‘the question is not, Can they reason? nor, Can they talk? but, Can they suffer?’
These matters aside, the general thrust of the theory advanced by Braithwaite and Pettit is persuasive. The justification for punishment relates to forward-thinking considerations which are aggregative in nature. However, we believe that the utility which should be aggregated is the broader notion of happiness, as opposed to dominion.
We now turn to the main focus of this paper — critically evaluating retributive theories of punishment.
Most of the significant groundwork in the development of retributive theories has been undertaken in the philosophical context. While retributivism is an established legal concept, it has been subject only to superficial analysis in the legal arena. Although it is not uncommon for judges and legal commentators to advert to the concept, there is normally a pronounced lack of convergence regarding which version of retributivism is being invoked and an aversion to detailed discussion regarding the intricacies underlying the concept. In this domain the concept is normally tossed around with enormous generality and in the rare cases that a specific form of retributivism is considered, it often seems to be in ignorance of other competing versions. Thus, judges and legal commentators have contributed little of substance to the debate on retributivism. This is not, however, meant as a criticism of the judicial approach to sentencing.
Choosing a particular theory of punishment represents an important policy choice and is accordingly a matter for legislatures. Given the reluctance of most legislatures to take a pointed stance in this regard, it is not surprising that the courts have at least tinkered with the issue of why it is that we do and should punish wrongdoers. Despite the relatively cursory analysis of retributivism by the courts, the judicial approach to the issue forms a useful backdrop to the foregoing discussion, if mainly to underline the nebulous and evolving nature of the concept. An overview of the courts’ approach to retributivism also shows that many of the major themes running through retributivism have, at some point and to varying extents, been recognised.
The crudest form of retributivism is intrinsic retributivism, or retributivism as vengeance. This has been acknowledged by the courts, though its reception has been mixed. It has been asserted that offenders deserve harm and that there is an element of revenge in sentencing. In R v Gordon it was said that:
Retribution, or the taking of vengeance for the injury which was done by the offender, is ... an important aspect of sentencing ... Not only must the community be satisfied that the offender is given his just deserts, it is important as well that the victim, or those who are left behind, also feel that justice has been done.
A contrary view was taken in R v Wheatley: ‘[t]he relatives of the victims ought to approach this type of case not in the spirit of Old Testament morality, seeking vengeance, but rather in light of ordinary straightforward retribution’. However, in R v Collins Tadgell J seemed to assume that revenge and retribution were equivalent:
What are we about when we say our task is to uphold the criminal law? We mean that we must implement it; but in doing so we must be sure that we are not concerned with the exaction of vengeance. The punishments that the courts impose are not primarily to be seen as retributive.
In R v Sargeant it was suggested that the lex talionis (the concept of an eye for an eye and a tooth for a tooth), which is often used interchangeably with intrinsic retributivism, is misguided:
The Old Testament concept of an eye for an eye and a tooth for a tooth no longer plays any part in our criminal law. There is, however, another aspect of retribution which is frequently overlooked: it is that society, through the courts, must show its abhorrence of particular types of crime, and the only way in which the courts can show this is by the sentences they pass. The courts do not have to reflect public opinion. On the other hand courts must not disregard it.
In a similar vein, in R v Roberts it was stated that:
At one time punishment was regarded in the light of vengeance or retribution against the wrongdoer and offenders were sentenced to be hanged for comparatively minor offences. This was an outgrowth of the old Biblical concept expressed in the words ‘eye for eye and tooth for tooth’. Retributive justice has faded into comparative insignificance in the present-day administration of criminal justice.
The satisfaction theory of retributivism provides that punishing wrongdoers satisfies ‘the feeling of hatred — call it revenge, resentment, or what you will — which the contemplation of such conduct excites in healthy constituted minds’, and thereby diminishes the prospect of harmful vendettas by victims and their associates, who may be tempted to exact their own revenge. It too has occasionally been endorsed by the courts:
One of the objects of punishment, and by no means the least important object of punishment, is to prevent, so far as possible, the victims of crime from taking matters into their own hands. It is no great step from private vengeance to vendetta, and there is no knowing where vendetta will stop.
The version of retribution which has received possibly the greatest judicial support is denunciation. Denunciation is a mark of public disapproval of criminal conduct. It is a means by which the courts can maintain public confidence in the administration of justice and reflect the moral sense of the community in the sentence. In R v Nichols Lee AJ stated that in the case of serious crime, the court must show its denunciation of the crime committed: ‘[t]he moral outrage of the community must be taken into account’. English courts have expressed similar sentiments, holding that one of the purposes in sentencing is ‘to mark the disapproval of society’ or ‘to mark the abhorrence which society feels for this type of attack’.
In assessing the degree of community outrage the courts look not to actual community feeling, but rather to informed public opinion: ‘it would be discreditable for the law to seek to give effect to extremist views of those palpably disabled from a fair judgment by reason of prejudice, ignorance or other such factors’. The public are assumed to have knowledge of the circumstances of the case and an appreciation of the range of penalties imposed in previous similar cases.
The irrelevance of actual community sentiment to denunciation is highlighted by the fact that evidence of community feeling is not even admissible:
The evaluation of the criminality of the offence and whether imprisonment is called for is for the judge to determine upon the relevant evidence in relation to the crime. It is not a matter to be determined by reference to the views of others given directly in evidence or as hearsay, which views in any event may be based on wrong facts or facts not in evidence. ... [T]he extent of community abhorrence of a crime or type of crime is not a matter of evidence.
Support for a denunciatory theory of punishment has not always coincided with an endorsement of retributivism. Denunciation has often been backed because it is felt that it will promote more fundamental objectives of sentencing, many of which sit more comfortably in a utilitarian ethic. For example, it has been contended that denunciation prevents crime by publicly declaring that criminal activity will not be tolerated. In R v McKenna it was stated that ‘a non-custodial sentence ... lacks the element of denunciation of the crime which is of vital importance in the case of laws designed to protect young persons and thus necessary if deterrence is to be achieved’.
Denunciation has also been regarded as a means of satisfying victims and maintaining public confidence in the criminal justice system. It is also said to have a strong educative role. In R v Collins Tadgell J, quoting the words of Sir John Barry, stated that punishment ‘serves by its solemn procedures as a teacher of minimal standards of morality and behaviour; as an agency for the expression of public indignation and condemnation; and as a force operating to produce cohesion within society.’
Reports by sentencing committees and the like have on the whole been far less supportive of the retributive cause than the judiciary. This may be largely due to the sweeping manner in which retributive theories have been considered by such bodies. The Australian Law Reform Commission provided that:
Retribution is a justification which provides that a person is to be punished for his or her wrongful acts simply because he or she deserves it. It is based on the ancient principle of an eye for an eye and a tooth for a tooth. Retribution in its pure form has very little application today.
More recently, the New South Wales Law Reform Commission also failed to distinguish between different retributive theories and simply provided that ‘retribution is the notion that the guilty ought to suffer the punishment which they deserve’. It stated that ‘just deserts [of the type adopted in some Australian jurisdictions] is merely a reflection of the common law principle of proportionality which places limits, in terms of the gravity of the offence in issue, on the severity of the punishment’. The Canadian Sentencing Commission was quick to conclude that retributivism must borrow utilitarian arguments to address the issue of the justification of punishment convincingly.
Such criticism lacks persuasive force as a result of its all-embracing treatment of retributivism. A convincing rejection of any theory requires, at minimum, an exploration of its distinctive features and supposed advantages and, where relevant, important modifications and subtleties.
It is to this task that we now turn. This requires a detailed consideration of the philosophical debate in this area, where dozens of different retributive theories have been advanced. It is not possible within the scope of this article to consider each of them meaningfully. With one exception, our focus is on those retributive theories which have proven to be the most influential. Thus we consider von Hirsch’s theory, which is widely acknowledged as being responsible for the revival of retributivism. This is followed by a consideration of Duff’s theory, which picks up many of the themes discussed by von Hirsch. Finally, we discuss the unfair advantage theory of punishment which George Sher has recently attempted to revive.
Our starting point is the crudest and most straightforward version of retributivism: intrinsic retributivism. This has few contemporary adherents: retributive theories are now more ‘sophisticated’. Although few would consider intrinsic retributivism a tenable theory of punishment, in our view it merits detailed consideration because with a little adjustment it is ultimately the most persuasive retributive theory.
The paradigmatic and most basic retributive theory has been dubbed intrinsic retributivism. This is the simple claim that punishment is justified because there is intrinsic good in making wrongdoers suffer: ‘the principle that wrongdoers deserve to suffer seems to accord with our deepest intuitions concerning justice.’
As individuals we have a wholly proper desire to seek revenge when wrongs are inflicted on us: as a society we demand that constituted authority punish those who unjustifiably inflict injury on others or otherwise act in ways we think are wrong. If other benefits are incidentally derived from making the wicked suffer, well and good; but those benefits must not be sought for their own sake.
Intrinsic retributivism is commonly associated with the lex talionis: the concept of an eye for an eye, a tooth for a tooth, and so on. However, this is not necessarily the case, since intrinsic retributivism says nothing about the amount of punishment, which is a central feature of the lex talionis. Because the lex talionis is often expressed without further justification, it lends itself easily to the claim that wrongdoers deserve to suffer is a self-evident truth.
There is no question that the sentiment that the guilty deserve to suffer is widespread and extends beyond the parameters of legal punishment. If a wrong-doer suffers some accidental harm, for example, if a burglar breaks his or her leg breaking into a house, this is often described as ‘poetic justice’. We naturally feel that the good should prosper and, perhaps even more strongly, that the bad ought to suffer. If a wicked person wins the jackpot, for example, we tend to think of this as being undeserved or unfair. Given that there is widespread support ‘among the people whose moral intuitions constitute the main data we have for settling questions of value’ for the assertion that wrongdoers deserve to suffer, it is very likely that that assertion is correct. Thus, retributivists assert that the strong innate chord struck by the view that wrongdoers deserve to suffer is so strong that support from some more fundamental principle or instrumental goal is not necessary to substantiate it: the justification is self-evident. There are, however, several problems with this theory.
Ten suggests that the intrinsic retributivist faces the difficulty of explaining whether deserved suffering is influenced by past undeserved suffering. A ‘whole life view of suffering’ approach requires us to determine desert, not by merely focusing on the wrongdoer’s level of culpability for the particular offence, but rather by weighing all of his or her wrongs against the suffering experienced throughout his or her lifetime, which incorporates such things as undeserved suffering from social causes. On this basis it may be that the wrongdoer, as a consequence of social deprivation, may have amassed credits for suffering and accordingly ought not be punished for a particular crime. However, this objection is not decisive. It looks to the quantum of punishment rather than the need for it and in this regard it is not an objection peculiar to intrinsic retributivism. Rather, it is a matter that must be addressed by all theories of punishment. Intrinsic retributivism makes one simple claim — wrongdoers deserve to be punished. It does not go on to provide how much they should be punished. Nor does it directly need to do so, as this can be dealt with by invoking other principles.
The fact that other principles may be needed to supplement the primary claim of a theory is not necessarily a shortcoming. No general theory of punishment provides clear cut answers to every aspect of punishment: this can be left to other subordinate principles. It is only where the subordinate principles that lead to the ‘appropriate result’ (in this case, that offenders from deprived backgrounds should be punished less severely) are not compatible with the primary claim of the theory that one gets into strife. However, the proposition that previous hardships sustained by an offender are relevant to the amount of punishment leaves intrinsic retributivism unscathed. First, it can simply be asserted that the whole life view of suffering is untenable. Moreover, the intrinsic retributivist could accept the whole life view of desert by extrapolating his or her primary claim to the effect that ‘punishment’ includes the pain naturally endured throughout life; thus, where an offender has suffered at least as much incidental hardship as is deserved by an offence, then no additional punishment is called for. There are, however, several other more telling objections to intrinsic retributivism.
The first objection relates to the reliance on self-evident or intuitive ‘truths’. Reliance on such truths can only occur in the most limited of circumstances, since their persuasiveness is roughly commensurate with the incongruity of an assertion to the contrary. Given that there is no absurdity in the claim that wrongdoers do not deserve to suffer, intrinsic retributivism is not self-verifying. While our feelings often incline us to respond favourably towards those who treat us well and wish harm on wrongdoers, such sentiments are not universal. That a person has broken the law or for that matter has behaved immorally does not compel any particular moral conclusions. We do not necessarily feel a pang of resentment towards those who drive too quickly, steal from shops or lie to others. Although we disapprove of speeding, shoplifting and lying, we recognise that it is a fact of life that such things occur and do not expect or demand that such conduct should always result in unpleasantness to the agent. Thus we are often indifferent when law-breakers go unpunished.
Even if the desire to punish wrongdoers is pervasive, this does not justify punishment, since truth does not necessarily follow consensus. This is the case with empirical and supposed normative beliefs. The widely held view that the world was flat did not make it so and widespread acceptance of slavery similarly did not provide a justification for it. Similarly, even if we are so built that wrongdoing necessarily prompts a punitive impulse, it does not follow that intrinsic retributivism justifies punishment, any more than an innate sense of jealousy justifies locking up our partners. What is necessary from a normative theory is a justification of the relevant practice, not an explanation of it.
We frequently must set aside our natural human responses (for example, jealousy, lust, rage and anger) and adopt more considered and reflective dispositions, because of the harm which they cause. So too could it be argued with intrinsic retributivism. Of course, it could be countered that as a psychological matter, our make-up is such that it is not possible to react in a manner other than to punish wrongdoers: inferring, therefore, that it is wasteful even to consider (other) justifications for punishment, since there is no point in arguing against that which cannot be curtailed. This, however, is repudiated by the sheer number of counter-examples where people bear no animosity towards those who have violated their important interests. Indeed, there have been loud calls for the abolition of punishment.
Even if it can be shown that punishing the guilty is an intrinsic good, it does not necessarily follow that punishment is justified. For as David Dolinko points out, an intrinsic good need not be a particularly important good or objective:
I have an itch; I scratch myself; the itching ceases. The cessation of the itching sensation, I believe, is an intrinsic good. Yet it is surely a quite unimportant good, and if for some reason I could not scratch myself without creating a high risk that innocent people would die, it would be unconscionable for me to scratch anyway on the ground that doing so would bring about ‘an intrinsically good state of affairs.’
This emphasises the point that in evaluating the morality of any practice, consequences cannot be totally ignored. Followed to its logical conclusion, intrinsic retributivism entails that offenders should be punished even when no good comes from this. If it were positively established that, overall, punishment had bad social consequences (for example, because it increased recidivism and the overall crime rate), utilitarians would abandon punishment in preference for a more effective manner of dealing with offenders. In contrast, the commitment to punishment by intrinsic retributivists would remain unchanged. It seems wrong to impose a harsher punishment if an offender could be reformed by a lesser sanction: ‘[r]etributive justice may be a very good thing, but the saving of souls is a much better thing’. This criticism can be levelled at most retributive theories, since they do not rely on the effects of punishment for justification. The criticism is particularly damaging with respect to intrinsic retributivism, where the only justification for punishment — even where the punishment would lead to disastrous consequences — is the mere impulse to punish.
Retributivists do not have to deny that consequences are totally irrelevant. Thus, they could attempt to invoke consequentialist considerations to veto punishment, say, where no good at all would come from it. However, as with most compromises, this risks fatally destabilising the theory. It would be unclear at what point consequences become decisive. The retributivist also would be required to give a coherent account of why consequences matter only some of the time, otherwise, he or she would appear to be advancing not a moral justification of punishment, but rather an expedient and disparate set of retorts in order to justify an existing intuition.
The source of the judgment that the guilty deserve to suffer is unclear. However the most likely origin is the desire for revenge. It follows that, stripped bare, intrinsic retributivism amounts to the claim that offenders deserve to be punished because a wrong action naturally calls for revenge or a hostile response. This was a point acknowledged by Mackie, who claimed that retributive sentiments are an ingrained part of our moral thinking. He labels as ‘the paradox of retributivism’ the proposition that ‘on the one hand, a retributive principle of punishment cannot be explained or developed within a reasonable system of moral thought, while, on the other hand, such a principle cannot be eliminated from our moral thinking.’
In order to resolve this dilemma, he relies on the distinction between beliefs and desires or feelings, and claims that retributive sentiments are desires or feelings, not beliefs, and that such feelings arise from the advantages to be gained through retaliatory behaviour. It is natural to desire to punish wrongdoers, but our value system must look beyond such sentiments.
It can be argued that this analysis of intrinsic retributivism shows it to be the mere expression of the primitive desire to exact revenge, which emphasises reprisal above all else and that the main impact of its adoption as a rationale for punishment would be to desensitise and brutalise society. Viewed in this light, most have been prepared to dismiss intrinsic retributivism as an untenable theory of punishment.
At this point it may seem that the above account of intrinsic retributivism is too expansive and accords it more respect than is commensurate with its standing among contemporary theories of punishment. However, with some deft fine-tuning, it may be possible to revive the theory. In order for this to occur, the intrinsic retributivist must embrace the concept of revenge rather than seek to mask it, and also affirm that — despite the harsh veneer of the link with vengeance — revenge is in fact not only a natural but also an appropriate response to wrongdoing.
Indeed, over a century ago, Eugen Dühring claimed that the origin of the concept of justice lies in the notion of revenge, that the desire to retaliate against those who have done wrong occurs naturally, and that, ultimately, criminal justice is simply the public organisation of revenge. Dühring also contended that the lex talionis should not be dismissed because it is too barbaric, but rather because it does not go far enough: more harm than that corresponding to the crime is necessary to restore equality, since the natural desire for revenge ‘does not limit itself to the magnitude of the offence: it normally goes further, and rightly so’.
As to the nature of revenge, numerous explanations have been offered. Apart from Dühring’s description, revenge has also been explained as an attempt to regain one’s honour or social prestige and an instinct which assists in the struggle for survival, since vengeful groups are supposedly less likely to be harmed by others.
The claim that intrinsic retributivism stems from the desire for revenge can form the basis of an extremely coherent normative argument when coupled with the further claim that morality consists precisely of such retorts: desires rather than beliefs. This, in our view, puts intrinsic retributivism in its best possible light. This interpretation of morality has a significant impact on the debate about intrinsic retributivism. If morality is ‘simply’ an expression of our feelings or desires, as many contemporary moral philosophers believe is the case, then the fact that intrinsic retributivism stems from a desire (for revenge) as opposed to a (rational) belief is an advantage, not a drawback.
Even if we accept that moral judgments stem from desires, it is not the case that any desire will do. Certain desires are so repugnant that they are disqualified from forming the basis of normative prescriptions. The desire to kill or torture the innocent, no matter how pervasive, could not tenably be proffered as a normative justification for any action. To show that the desire for revenge is a candidate for moral recognition, it is first necessary to devise a test for distinguishing between desires that are morally relevant and those that are not, then it must be shown why the desire for revenge belongs to the former group. Non-cognitive theories of morality, such as emotivism and projectivism, have fared poorly in attempting to distinguish morally relevant desires from those which are irrelevant. However, it is certain that one consideration relevant to determining the moral relevance of a desire must be the effect of implementing that desire.
In addition to this, as a general rule, our moral standpoint looks favourably upon expressions of feelings and desires that are likely to be welcomed by the recipients of these sentiments, such as feelings of love, compassion and respect. On the other hand, desires which prompt behaviour which is generally unwelcome are morally condemned. Given the measures most people take to avoid vendettas, it is obvious that the desire for revenge belongs to the latter group.
It has also been claimed that revenge is always misguided because no action can reconstruct the past; no deed can be annihilated. Harming others does not eliminate one’s own suffering and revenge may even be self-defeating, since if it is satisfied it merely ‘increases the amount of suffering in the world’. This view can be challenged on the basis that it wrongly assumes that revenge has a particular rationale which can be measured by the extent to which it promotes some more fundamental virtue, such as happiness, rather than being merely a fundamental desire, the aim of which is no higher than its own satisfaction. This response, however, runs head-on into the objection, outlined earlier, that a consequence of this theory is that punishment is justifiable even if it results in overall bad consequences.
The intrinsic retributivist could try to soften his or her position by asserting that what is being appealed to is ‘retribution’, not revenge. It is not exactly clear in what sense retribution is being used here; however, Robert Nozick advances five differences, and Ten another one, between revenge and retribution. Broadly, it is claimed that retribution is a manifestation of a reflective view that the guilty should be punished, as opposed to the expression of our primitive desire to punish wrongdoers. More specifically, it is claimed that:
However, in relation to the first four factors it is not clear that they do represent genuine differences and, even if they do, they are irrelevant for the purpose of this discussion. The effect of the last two factors is merely to weaken the case for intrinsic retributivism.
With respect to point (a) it is unclear why retaliation for something which is not a wrong (such as sneezing in public) is properly defined as revenge any more than it is retribution. More properly, it would seem to simply be a case of misplaced anger. Likewise regarding point (d), which highlights the fact that retribution is supposedly more targeted than revenge: once again it is not clear why harm directed at a person who is not a wrongdoer is not more appropriately defined as misplaced anger, rather than revenge. In any event, the disadvantages of revenge as a foundation for intrinsic retributivism do not concern the possibility of punitive measures against people other than wrongdoers.
Consideration (b) focuses on the purported distinction between retribution and proportionality, but this link is more hypothetical than real, since retribution is primarily concerned with why we should punish, not how much. Point (c) is also not significant. No sentencing system has yet been able to provide consistency in sentencing; not even those loosely based on retributive ideals. Also, consistency is not the primary objective of sentencing. A consistent sentencing system which pursues flawed objectives simply perpetuates pre-existing injustice.
Points (e) and (f) are more significant, but only because they underline why retribution forms an even less tenable foundation for intrinsic retributivism than revenge. Point (e) charges that revenge is normally meted out by the aggrieved party, unlike retribution which is administered by another party. In this way retribution is dispensed in a more controlled and measured fashion which eliminates some of the more unpleasant features of private revenge. Nevertheless, the problem here is that however one elects to prop up intrinsic retributivism, the claim that punishment should be administered by the state does not follow from intrinsic retributivism: even if it shows that the guilty deserve to suffer, it cannot support the claim that the suffering should be deliberately inflicted on wrongdoers by the state.
Furthermore, once the desire for revenge is renounced as the basis for intrinsic retributivism, its most alluring aspect (the apparent pervasiveness of the desire at its core) is lost, since it cannot be asserted as convincingly that there is a universal desire for retribution. This is the point made in (d) and is supported by the apparent disinterest of many victims of crime in the outcome of the prosecution of the offender. The statement that wrongdoers deserve retribution lacks the emotive appeal of the assertion that they deserve revenge.
The upshot of the above discussion is that retributivists are put to the task of justifying the link between punishment and wrongdoing, beyond merely asserting the appropriateness of the connection. There have been numerous attempts at establishing such a link; we shall consider three of them.
According to von Hirsch, the principal justification of punishment is censure: that is, to convey blame or reprobation to those who have committed a wrongful act. Von Hirsch thinks that censuring holds offenders responsible and accountable for their actions and in this way, because offenders are given an opportunity to respond to their misdeeds by acknowledging their wrongdoing in some form, it recognises their moral agency.
For von Hirsch, there is a second justification for punishment, namely, to prevent crime. He believes that human nature is such that the normative reason for compliance must be complemented with a prudential one, otherwise ‘victimising conduct would become so prevalent as to make life nasty and brutish’. He states that ‘it is the threatened penal deprivation that expresses the censure as well as serving the prudential disincentive’. Although he believes that deterrence is not a sufficient reason for punishment, he claims it is a necessary one: ‘if punishment has no usefulness in preventing crime, there should ... not be a criminal sanction’. Instead, other means should be adopted to express censure.
As such, von Hirsch is clearly a consequentialist in part since his theory of punishment is contingent upon punishment having a deterrent effect. However, he attempts to water down the consequentialist tag by asserting that although general deterrence is a necessary precondition for punishment, it is not sufficient and indeed only serves as a secondary justification.
Before considering von Hirsch’s theory in detail, it is worth noting that there are good consequentialist reasons which support his thesis that censure or condemnation of the criminal is a proper response to crime. Condemning criminal activity stigmatises such conduct and serves to discourage future criminal acts. Von Hirsch’s challenge is to provide reasons for censure which do not rely on the good consequences which follow from blaming criminals.
The most general point made by von Hirsch in defence of his theory is that responding to wrongdoing by reprobation ‘is simply part of a morality that holds people accountable for their conduct ... It is addressed to the actor because he or she is the person responsible’. The assertion that blaming people may hold them accountable for their misdeeds does not provide a normative justification for punishment, but rather only identifies the function of blame. The link between blame and accountability provides no better answer than intrinsic retributivism to the question of why we should blame wrongdoers. To address this von Hirsch identifies the ‘positive moral functions of blaming’.
Von Hirsch asserts that the benefits flowing from censure are threefold. First, censure addresses the victim: ‘[c]ensure, by directing disapprobation at the person responsible, acknowledges that the victim’s hurt occurred through another’s fault’. This recognition, however, need not come in the form of blame directed at the offender; a mere declaration to the victim that his or her suffering was caused by the wrongful actions of another would presumably suffice. The fact that von Hirsch requires the further step (that the offender is told of his or her wrongdoing) indicates that there are also other goals at work here.
To this end, the satisfaction theory of punishment provides two reasons why blaming and punishing offenders is desirable. Recognising the victim’s unfair fate pleases the victim and an institutional system of punishment also serves to quell the desire for socially harmful vendettas which may otherwise be unrestrained.
It is essential to note that both these considerations are consequential in nature. Of course, there is normally nothing wrong with this unless, like von Hirsch, one happens to be, at least primarily, a non-consequentialist. Another narrower problem with closely aligning the justification of punishment with the interests of the victims is the large array of victimless offences, or offences where there is no discernible victim.
The second positive function of blaming, according to von Hirsch, is that it addresses the criminal. The criminal is conveyed the message ‘that he culpably has injured someone, and is disapproved of for having done so’. This holds the offender responsible for his or her actions. However, as we have mentioned earlier, even if we accept this, von Hirsch is still no closer to drawing the link between punishment and crime: it is still an open question why we should blame the offender. There is no intrinsic merit in telling people that they have done the wrong thing. Even if it is felt that there is some benefit in this, it merely justifies the conveying of such a message, not the further step of imposing an unpleasantness.
Von Hirsch, possibly in an attempt to diffuse the obvious counter that the link between punishment and crime can only be made by introducing consequentialist considerations, states that censure does not aim to change the criminal’s moral attitudes or elicit any particular response: censure ‘is not a technique for evoking specified sentiments’. He also thinks that censure provides the criminal with an opportunity to respond (for example, by making an effort at improved self-restraint or an acknowledgment of wrongdoing) and that he or she is expected to respond: if the criminal does not respond positively to censure this would ‘itself be grounds for criticizing him’. However, if the goals of punishment do not include the promotion of moral development, the source of such an expectation is unclear: it is as if it pops up from nowhere. Expectations, as opposed to hopes, are grounded in obligations, which in turn are derived from (voluntary or inadvertent) participation in goal-orientated practices or transactions. Obligations occur because they are necessary to facilitate the objectives of the relevant practice or transaction. There is an obligation (and hence an expectation) that each party to a contract will honour his or her promises, otherwise the purpose of the agreement will be defeated; government officials are expected to provide individuals who may be adversely affected by their decisions an opportunity to be heard on the matter, otherwise the objective of natural justice would be frustrated. More generally, we should not deliberately harm others, because this violates important moral goals. Von Hirsch’s suggestion that we do not censure in order to reform the criminal but that we expect such reform to result from censure is merely a sophistical attempt to introduce consequentialist considerations (attitudinal reform and behavioural change).
Von Hirsch goes on to advance two other reasons in support of his claim that censure does not aim to alter the values of wrongdoers and hence does not have a consequential foundation. First, he points out that there is no attempt to seek information about the wrongdoer’s ‘personality and outlook, so as to better foster the requisite attitudinal changes. ... The condemner’s role is not that of mentor or priest’. However, the fact that there is no positive concerted attempt to discover the particular mental processes of the criminal hardly means that there is no attempt to encourage attitudinal and behavioural reform. Advertising, political campaigns and pressure groups bear testimony to this. The reason that we do not attempt to find out the intricate workings of each individual whose behaviour we seek to influence is that we make certain assumptions about human behaviour. One is that people respond to reason, another is that they can be moved by emotive pressure, and another is that they try to avoid punishment.
Von Hirsch’s second basis for claiming that censure does not aim to achieve moral improvement is that blame is supposedly appropriate even where it is apparent that it will have no effect on the offender’s outlook: for example, where the offender is already repentant or is stubbornly defiant. However, von Hirsch provides no reason why blaming in such circumstances is appropriate, as opposed to just being a waste of time. Admittedly, censure is the ‘authoritative expression of the condemner’s ethical judgment’. However, as David Adams points out, if blaming is justified solely by the condemner’s desire to express it and is only ‘something we are strongly moved to do, without any further (instrumental) purpose’, this leaves us with the question of why ‘we are justified in building such responses [to crime] into the institution of criminal punishment.’
The third and last positive function of censure is that it supposedly addresses third parties and provides them with reasons for desistence from crime; blaming conveys the message that the relevant conduct is reprehensible, and should be eschewed. Von Hirsch concedes that this reason is partly consequential, but denies that it is purely consequential by asserting that ‘the censure embodied in the prescribed sanction serves to appeal to people’s sense of the conduct’s wrongfulness, as a reason for desistence’, as opposed to an attempt to portray that the conduct is wrong, ‘for those addressed (or many of them) may well understand that already’. At the core of von Hirsch’s argument is the paradigm consequentialist goal of desistence. Tinkering with the means employed to facilitate this aim cannot alter the fundamental character of the goal: whether one trains a dog with treats or a stick, in the end the (consequential) goal is the same — to have an obedient pet.
Thus, it emerges that what von Hirsch calls the ‘positive moral functions of blaming’ are simply the ‘positive consequences of blaming’.
Von Hirsch accepts that criminal sanctions are too severe to be justified by the need for censure alone. Censure on its own only justifies the expression of blame, not the further step of imposing hard treatment. To justify the need for hard treatment, von Hirsch expressly calls into play the goal of deterrence, which to him is a secondary purpose of punishment. He thinks that hard treatment provides prudential reasons for desistence, which supplement the normative reason supplied by censure.
The critical point here is that the concept doing all the hard work is the utilitarian flagship of deterrence. Von Hirsch attempts to distance himself from a core reliance on deterrence by stating that a primarily consequentialist account for punishment and blaming is inappropriate. He contends that it would permit neutral sanctions, which provide for hard treatment but no censure and, hence, fail to recognise the wrongfulness of criminal behaviour. Thus, the door would be left open for fining offenders instead of prescribing criminal responsibility and criminal sanctions to criminal conduct.
However, this fails to recognise the good utilitarian reasons that exist for ascribing moral blame to criminal behaviour. Empirical studies have revealed that normative issues are closely linked with compliance with the law. People do not merely obey the law because it is in their self-interest to do so, but also because they believe it is morally proper to do so.
Von Hirsch believes that it is ‘evident enough’ that punishment conveys blame or reprobation, but he accepts that the crucial issue regarding the justification of punishment is ‘why should there be a reprobative response to the core conduct with which the criminal law deals?’ He further accepts that punishments hurt. This being the case, to justify punishment on any ground one must invoke sound moral reasons. However, von Hirsch, and this is the most serious flaw in his theory, fails to address the issue of why his censuring account of punishment is morally justifiable (apart from the utilitarian reasons from which he seeks to divorce himself). He in fact ‘argues ... within, rather than for a desert-based or (broadly) retributive justification of punishment.’ Such an approach will not sway the unconverted and we must move on.
Thus, von Hirsch is unable to provide a rationale for censure and punishment which does not ultimately invoke consequential considerations. Despite this it is still understandable that his theory strikes a responsive chord with so many. His theory is not ostensibly consequentialist and urges the intuitively appealing principle of proportionality as the main sentencing consideration.
Duff advances another communicative theory of punishment. He believes that the main aim of punishment is ultimately to integrate offenders back into the community: punishment aims to induce repentance (the remorseful acceptance of guilt), self-reform, reparation (the repairing of a damaged relationship with the rest of the community by genuine recognition of the wrong) and finally, reconciliation. Punishment, he argues, is a means through which we engage in a punitive dialogue with the offender. Duff believes that criminal sanctions serve to bring ‘the criminal to recognise the wrongfulness of her past conduct; to induce the kind of pain which flows from an understanding of the condemnation which they express’. The criminal is then reconciled with the community by expressing his or her repentant understanding through being punished.
Duff states that punishment also communicates to the rest of the community the wrongfulness of criminal conduct and to victims it represents an ‘authoritative disavowal of such conduct’. However, these purposes are merely subsidiary; punishment is essentially a means of engaging in a punitive dialogue with the offender ‘which aims to persuade (but not to coerce or manipulate) her to recognise and repent that wrong, and thus to restore her relationship with her victim and with the community’.
Thus, while Duff’s theory has many similarities with von Hirsch’s account of punishment, there are several points of divergence. Both theories emphasise the censuring aspect of punishment, but whereas von Hirsch claims that punishment merely aims to give offenders external (prudential) reasons for desistence, Duff is far more ambitious, claiming that through punishment we should aim to alter the moral sentiments of offenders. The other main difference between the theories is Duff’s insistence that punishment is not coercive.
The educative and cohesive function of punishment has received some judicial support. In R v Williscroft Adam and Crockett JJ cite Sir John Barry with approval:
The aims of punishment are often classified as retributive, preventive, deterrent, and reformative, but this classification is plainly an oversimplification. It ignores or leaves inarticulate, for example, other purposes which the criminal law serves by its solemn procedures as a teacher of minimal standards of morality and behaviour; as an agency for the expression of public indignation and condemnation; and as a force operating to produce cohesion in society.
The supposed non-coercive nature of punishment is central to Duff’s theory. He contends that punishment aims to communicate to the offender the condemnation of his or her conduct and seeks for the offender to
condemn himself, and to modify his future conduct accordingly; and thus to persuade him not merely to obey the law, but to accept its justified demands and judgments. Punishment, like moral blame, respects and addresses the criminal as a rational moral agent: it seeks his understanding and his assent; it aims to bring him to repent his crime, and to reform himself, by communicating to him the reasons which justify our condemnation of his conduct.
At first glance it may seem that Duff’s account is essentially consequentialist. By aiming to reintegrate criminals back into the community he concedes that his theory has a forward-looking purpose. However, he contends that his theory is retributivist for several reasons. For Duff, the purpose of punishment is ‘logically rather than contingently related to the “means” by which it is to be achieved’. Even though punishment aims at reforming and rehabilitating criminals, Duff asserts that this does not amount to a consequentialist justification, since the aim lies in its internal, not instrumental, ability to promote reform and rehabilitation, which can only be achieved by bringing the criminal to ‘recognise and suffer for her wrong-doing’. Duff also denies that his theory is consequentialist because although our reasons for maintaining a system of law and punishment relate to the prevention of crime and the good of the community, we do not merely use criminals as means to achieve these ends since ‘the law addresses citizens as rational moral agents, seeking their assent and understanding’. According to Duff, it follows that the pursuit of rehabilitation through punishment is justified even if there is no hope of reformation, since punishment in these circumstances still expresses a proper moral concern for the offender.
A significant shortcoming of Duff’s theory is that it lacks practical relevance. He admits that the sanctions which are typically imposed by our penal system are unlikely to achieve repentance, reform, reparation and reconciliation, and that the concept of punishment he propounds is only justified and appropriate in an ideal society and legal system.
In order for punishment to be justified, he believes that the criminal must be a responsible agent and accountable to those holding him or her to account. For Duff, this requires that the law that the offender has broken embodies the ‘values of a community in which she participates, and to whose values she is or should be committed. That is, she must be a citizen of whom it can properly be said that she has an obligation to obey the law because it is her law’. Duff goes on to claim that the law must be justified by regard to a common moral good, otherwise crimes which are not destructive of morality and punishment cannot aim to restore the offender to the common good and the community. This does not require each law to be morally sound, since ‘we can still owe it to our fellow-citizens to obey even an imperfect system of law, so long as it can be adequately ... justified to us in the relevant moral terms’. For offenders to be properly answerable to society it is also necessary that they have been treated as fellow citizens and accorded the concern and respect which is owed to all citizens.
Duff rightly points out that these preconditions do not exist in our social and legal system. He claims that our laws cannot be justified by reference to the common good because of our general failure to accord all citizens the concern and respect that they deserve. This, he believes, provides disadvantaged offenders with the strongest moral basis for resisting punishment: ‘not because their actions are justified, not because they ought to be excused, but because we lack the moral standing to condemn them’. Duff appreciates that his ideal is far short of the real world and he suspects that it may be so contrary to human nature that it may not be worth striving towards, since this may destroy our society entirely.
The telling point to emerge here is that in our community, according to Duff, punishment is not justified. Duff’s ideal setting is so far removed from the community in which we live that his theory could be disregarded on this basis alone, especially as Duff, while not an abolitionist, fails to advance an alternative account of what is an appropriate response to crime. However, Duff’s theory raises many interesting issues concerning punishment and it would be remiss prematurely to reject all aspects of his theory: occasionally, things that are worth doing, are worth half doing. Although one can hope that the criminal justice system and society in general may eventually match Duff’s ideals, we suggest that even in such an environment Duff’s theory is not persuasive. Duff fails to justify the purported non-coercive nature of punishment; he cannot explain the need for hard treatment and ultimately cannot draw the crucial normative link between crime and punishment.
A preliminary objection to Duff’s theory is his characterisation of punishment as being a communicative dialogue. He claims that punishment is communicative, as opposed to expressive, since it is a means by which the community responds to criminals as rational agents through a dialogue in a manner which respects them as fellow members of the community. However, this ignores the fact that a dialogue requires approximately equal parties in order that to facilitate a meaningful opportunity for a genuine exchange of views. To suggest that this occurs as part of the process of punishment totally misconceives a fundamental aspect of punishment: it is imposed not negotiated. Once a criminal is sentenced, there is nothing which he or she can possibly advance which will alter the sentence. Prior to sentencing, the only option for the offender is to plead for, rather than to request or demand, leniency. Given the weakness of the criminal’s position, punishment is the antithesis of a communicative institution (at least of the kind claimed by Duff). Now in itself, this does not totally rebut Duff’s theory, however, it is a noteworthy riposte because the description of punishment as a dialogue forms a central plank of his theory of punishment, which he describes as being a caring, almost altruistic practice.
Duff’s justificatory account for a system of criminal law and punishment is ultimately based on the Kantian principle of respect for autonomy. Duff assumes that a fundamental tenet of morality is that all people should be treated with concern and respect. From this it follows that we must take seriously the ‘respect and autonomy of those with whom we deal’.
If respect for autonomy is the guiding moral virtue, then Duff must explain how an apparently coercive institution such as punishment promotes, rather than violates, individual autonomy. Ostensibly, it is difficult to conceive of anything which is more likely to be destructive of personal autonomy than a practice which involves the deliberate and systematic imposition of hardship by the state. This argument is even more persuasive when we consider the type of penalties that Duff advocates. Sanctions aimed at an offender’s internal moral reform are arguably akin to brainwashing and, hence, pose an even more serious threat to an offender’s autonomy than those which aim to provide prudential reasons for desistence. Duff is alert to these criticisms and responds by stating first that punishment aims to induce repentance and, secondly, that reform does not have a coercive character since offenders have a right to be punished, and moreover they want to be punished. These claims are now considered separately.
Duff contends that an offender
has a right to be punished: a right to be punished rather than be subjected to some kind of manipulation or preventive treatment which would not address him as a rational agent; a right to be punished rather than be ignored or dismissed, since punishment expresses a proper response to his crime as the wrong-doing of a responsible moral agent, and a proper concern for his moral well-being as a fellow-member of the community. We owe it to him, as well as to those he has injured, to condemn his crime and to try to bring him to repentance and reform.
He also notes that:
Through punishment, we accord the criminal the respect which is his due as a responsible moral agent — we do not simply try to manipulate or coerce him into conformity. He may therefore claim a right to be blamed, or even punished, for his wrongdoing; a right to be treated, respected and cared for as a moral agent.
This ‘rights’ characterisation of punishment misconceives the nature of rights. Despite the complexities involved regarding exactly what it means to have a right, at minimum a right entails a plus; a benefit to the agent. To mistreat someone and to impose unpleasantness upon him or her is the direct opposite of a right. As Quinton notes, ‘[i]t is an odd sort of right whose holders would strenuously resist its recognition.’
Duff’s response to this might be to urge us to take a broader view of the effect of punishment on the offender. According to Duff, punishment ultimately helps offenders by improving their moral outlook and providing a means for them to be admitted once again as worthy members of the community. Thus, it could be argued that although offenders may not realise the advantages which stem from punishment at the time of apprehension or sentencing, in time they will come to be grateful for the fate imposed upon them, and it is in this long-term outlook that the benefit lies.
Duff proposes that despite the seemingly unpleasant nature of punishment, in the long run it is neither something which aims to harm offenders nor something they should seek to avoid, but is instead something they should embrace. This is the most pervasive — but ultimately indefensible — aspect of Duff’s account of punishment. This proposition is totally at odds with the surface nature of punishment as being an evil imposed upon wrongdoers. Duff asks us to ignore the overt character of punishment as being a practice which harms offenders, and to look deeper into its purpose: the imposition of short-term detriment will, in the end, benefit the offender. Now, of some practices which involve short-term harm, such an analysis is plausible and is epitomised by the ‘no pain, no gain’ slogan. The athlete who flogs his or her body for hours on end does it with a view to attaining superior fitness, and the factory worker who slaves away tirelessly performing meaningless and degrading work does so in a bid to achieve financial security. Both endure relatively minor hardships in a bid to attain distant desirable objectives. Despite the harsh day-to-day pursuits of such people, we are willing to accept that they are nevertheless engaging in practices which are, ultimately, beneficial to them.
There appear to be two central features of activities which may be regarded as being in an agent’s interest, despite their immediate harshness. It is difficult to conceive of punishment as possessing either of these features. The first characteristic of such practices is that they are not inherently ‘seriously’ harmful. Although a particular venture might seem harmful initially, it will still be justifiable if it is in the agent’s long-term interest. For example, the overall health benefits of a fitness regime mean that its immediate drawbacks, such as tiredness and cost, are acceptable. Despite their initial unpleasantness, criminal sanctions are not necessarily too damaging to be justified in the long term. For example, certain forms of gentler sanctions, such as community-based orders, may be regarded as being in the long-term interests of offenders. However, harsher treatments, such as imprisonment, may be so painful and oppressive as to outweigh any possible future advantages.
A second characteristic of activities which may be justified by their positive long-term effects is that they are voluntarily undertaken. We attach enormous weight to personal choice and it is for this reason that we do not generally attempt to dissuade others from engaging in practices we might consider misguided or futile, such as marrying the ‘wrong partner’, playing computer games, reading romance novels or studying pure mathematics. Given that offenders are forced by the machinery of the state to endure criminal sanctions, it is implausible to assert that they choose, in any sense of the word, such treatment, no matter how strongly others may think that they should for the sake of improving their moral standpoint. This is a point strongly challenged by Duff.
Not only does Duff believe that criminals have a right to be punished, but he also goes on to make the claim that they want to exercise this right: they want to be punished, no matter how determinedly wicked they appear to be. On Duff’s account, the criminal really wills his or her own good and therefore desires punishment as a means ‘of restoring him to the good’. Even in relation to criminals who expressly declare that they desire no moral or behavioural reformation, Duff maintains that they desire moral improvement because they should consider it an important goal. In this way Duff reconstructs punishment as an exercise of an offender’s autonomy, as opposed to a deprivation of it. Underlying Duff’s claim that criminals want to be punished is the Kantian proposition that rational agents universalise their principles of conduct. Thus, by breaking the law offenders consent to being punished. However, this argument is circular: it assumes that there is a necessary link between breaking the law and being punished.
Further, given the lengthy steps that most offenders take to avoid detection and apprehension, Duff’s claim is obviously counter-intuitive. He appears to be mindful of this and responds by stating that he is not making a claim about the relative strengths of the criminal’s desires, but rather an assertion that the criminal should care most about his or her moral wellbeing: ‘want’ in this context does not concern a ‘factual or empirical claim about the relative strength of ... desires, [but] a moral claim about what ... [the offender] should desire’. Duff thereby claims that punishment is not coercive, since it aims to persuade the criminal to repent the crime and to accept and consent to the punishment as penance as a means of reintegrating him or her into the community.
This account is unsatisfactory for several reasons. Firstly, if we accept Duff’s reasoning, there is no end to the types of interference that could be justified on the basis that ‘we think it is good for them’. It would open the door for paternalistic intervention, real or feigned, into every area of one’s life.
Secondly, at the theoretical level, Duff misunderstands what it is to have a desire or want. Desires are the states of mind, personal stimuli that move us to act. As is alluded to by Duff, it may be the case that one may want something and yet make no endeavour to attain it. This is most commonly the case because our long-term wants are overpowered by more pressing immediate inconsistent desires. That a person eats cake because it is a special occasion or a particularly appetising cake does not mean that he or she lacks the desire to lose weight. It is another thing to assert, however, that one can desire a thing because it is felt that it would be in one’s moral interest to do so. Desires do not work like that. As Hume observed several centuries ago, desires are not driven by reason.
Also, even if desires are directed by reason, Duff still must overcome the considerable obstacle of convincing criminals that they should take better care of their moral health and that punishment is a good medicine for this. Given the measures offenders take to avoid punishment and the rate of recidivism, the overwhelming weight of empirical evidence is stacked against Duff. There is no evidence that offenders desire moral education or that punishment generally leads to reform or feelings of guilt:
The idea that punishment produces a feeling of guilt in the offender is ... not confirmed by experience. Its effect is more often than not just the opposite: the offender becomes hardened in resistance to the demands of the community. Alternatively, it may have a wholly destructive effect and bring about a general demoralization.
Duff’s obvious counter here is to emphasise that he is making a claim about what offenders should desire, not what they actually want. This, however, constitutes perhaps the most serious flaw in Duff’s theory. At the core of his thesis is an important personal (as opposed to a collective principle, such as utilitarianism) moral virtue: autonomy. This involves people being able (at least to some degree) to do as they want. Thus, to rely on the principle of autonomy to justify punishment, Duff must present his account of punishment in a way which, at least to some degree, involves offenders doing as they want. This he cannot do. Instead, he invokes a collective or depersonalised notion of what others want the relevant agents (wrongdoers) to want and then attempts to portray this as respecting the autonomous wishes of wrongdoers. This, however, makes his account unintelligible: it is meaningless to refer to autonomy in a manner which is totally divorced from personal wants. At its highest, Duff’s justification for punishment is not personal autonomy, but whatever ‘virtue’ it is that prescribes that people ought to do what others think is in their interest. Thus, dressing up autonomy in terms of respecting the wishes that people should have advances Duff’s cause no further.
Further, it is extremely doubtful that punishment could serve as a vehicle for improving the moral values of offenders. Coercive measures can at best only produce prudential reasons to modify behaviour, rather than internal ones. This is the reason that people are taught ethics, rather than having it beaten into them. In the end the concept of forced moral development involves an internal contradiction — virtually all moral theories advocate tolerance and freedom of thought and expression as fundamental values.
To buttress his view that punishment is a right reposed in offenders and a process which respects their autonomy, Duff draws an analogy with the right of self-defence. He states that:
[J]ust as I believe that I may properly kill an assailant to protect my or others’ lives against his wrongful attack, while still respecting and responding to him as an autonomous agent, I also believe that the coercive imposition of punishment on a criminal can be consistent with a respect for her autonomy.
However, this analogy, rather than advancing Duff’s argument, highlights its failing. Killing an assailant has nothing to do with respecting his or her autonomy. Rather, it simply demonstrates the fact that no right is absolute and, more particularly, that in certain circumstances a person may by his or her appalling behaviour forfeit even the most fundamental right of all. This is not to say that the assailant is normally not worthy of our concern and respect, but rather that in these circumstances (when attacking another) he or she abrogates his or her entitlement to be treated with such regard. Likewise, where an offender seriously violates the interests of others, his or her interests can be violated to about the same extent as he or she has infringed upon the interests of others. Just as the right of self-defence is not framed in terms of the entitlements of the aggressor, neither is, nor should be, the practice of punishment.
The reason that Duff portrays punishment as something that criminals desire is obvious: otherwise, punishment can only be a coercive institution and, hence, by punishing offenders it must follow that we are violating their wishes. In a bid to circumvent this conclusion, Duff uses the term ‘desire’ so loosely as to deprive it of any relevant content and meaning. Thus, Duff cannot avoid the conclusion that punishment is coercive and therefore a violation of the offender’s autonomy. At this point his theory becomes vacuous. His fundamental moral premise concerns the inviolability of autonomy, yet in order to protect the autonomy of the rest of the community he is advocating a practice which directly impinges on the autonomy of offenders. So long as the activation of punishment depends upon a finding of criminal guilt and occurs against a background of state force, no matter how kind the stated goals of punishment may appear to be, there is little question that what is being done to the offender is ultimately for our, sake not his or hers.
Duff also has difficulty in justifying the need for hard treatment. As stated above, Duff believes that the essential purpose of punishment is ‘to bring the criminal to recognise the wrongfulness of her past conduct; to induce the kind of pain which flows from an understanding of the condemnation which they express’, and ultimately restore the criminal to the community. To achieve this, Duff claims that the ‘pain which expressive punishments aim to inflict or induce — the pain which wrongdoers deserve to suffer — must be mediated by the criminal’s own understanding of the condemnation which they express’.
However, if the aim of punishment is to reform offenders by communicating the wrongfulness of their conduct, it is difficult to understand why punishment needs to be in the form of hard treatment (such as imprisonment and fines): the verdict of the court already expresses the requisite condemnation. The process of subjecting offenders to trial, conviction and the associated moral blame would in most cases surely suffice to focus the attention of offenders on the wrongfulness of their behaviour. Even if this is inadequate, there are other (less draconian) ways to grasp their attention: ‘[o]ne can shout, ring a bell, or ask him to stand at attention.’ These may yet prove easier to ignore than the pain of hard treatment, but this does not mean that hard treatment is more effective at inducing reform.
Duff’s response is that mere formal condemnation does not go far enough: only hard treatment can induce repentance, reform, reparation and reconciliation. Repentance and self-reform, he asserts, are unlikely to be induced by merely formal condemnation, due to our fallible nature and reluctance to face up to our moral failings: ‘[p]unishment ... aims ... to induce and strengthen ... repentance, and to make possible its expression [by accepting and undergoing punishment].’ It is induced by confronting the criminal with the consequences of his or her actions and strengthened by keeping his or her attention on the wrong. Undergoing hard treatment, Duff continues, also achieves moral reparation, signifies repentant recognition of wrongdoing, and assures us of the offender’s respect for the victim and for the moral values by which they are bound. Reconciliation with the community, Duff claims, also requires more than simply accepting formal condemnation. Unlike purely symbolic punishment or the stigma of a conviction, hard treatment cannot be readily ignored.
Punishment can thus be portrayed as a secular form of penance: as a way of inducing, strengthening and expressing penance ... More precisely, it must be burdensome, if it is to effectively communicate the censure which the wrongdoing deserves; ... the pain involved in the repentant awareness of guilt.
In summary, there seem to be three separate points that Duff makes here. First, in order to promote repentance and self-reform, (compulsory) hard treatment is necessary because human nature is such that we would not willingly undergo it. Secondly, undergoing hard treatment provides a means through which repentance can be expressed: it is a way of signifying regret for previous misdeeds. Finally, the community will only accept tangible evidence of repentance before allowing the offender back into the community.
The first point, however, actually undermines Duff’s theory in several ways. It emphasises our earlier criticism that offenders do not in any meaningful sense desire punishment. Further, it entails that there are at least two sorts of offenders who should not be subjected to criminal sanctions: the already repentant and those who are beyond saving. However, this is rejected by Duff. He refuses to accept that a simple apology from the genuinely repentant is enough, despite the fact that they have already undergone the necessary internal reform. So far as the utterly defiant are concerned, he thinks that ‘[w]e owe it to every moral agent to treat him as one who can be brought to reform and redeem himself — to keep trying’; we should not give up on them since we have a duty to treat them with concern and respect and accordingly to impose punishment aimed to restore them to the community. This determination to punish at all costs leads to the suspicion that Duff’s real justification lies elsewhere. In the absence of any other alternative justification for punishment, the only place to turn is intrinsic retributivism.
The second point made by Duff lacks empirical support. The high rate of recidivism amongst those who have experienced hard treatment suggests that hard treatment is more likely to cause anger, frustration and a regression in one’s moral health rather than repentance and reform:
Deprivations often have the effect of focusing one’s attention on one’s own suffering. Being subject to hard treatment is likely to shift the agent’s attention from the nature of her previous wrongdoing to the nature of her current hardships. Hard treatment might be as likely to deflect the agent’s attention from her wrongdoing as focus her attention on it!
It is also questionable whether compulsorily-imposed sanctions provide a suitable avenue for expressing repentance, since there is no method for distinguishing between genuine repentance and expedient compliance. Undergoing a compulsory unpleasantness under threat of force normally has nothing to do with the acceptance or expression of repentance. There is nothing inherent in the process of undergoing punishment which requires internal reform; all sentences are completed merely through some means of physical compliance, for example, by paying money or being incarcerated for the required period. Even if sanctions were developed that required some expression of moral development, this would be superfluous. Given the strong incentive to avoid the sentencing process, one still could not distinguish between real and expedient declarations of moral wellbeing.
Duff attempts to strengthen his claim that punishment provides a means for signifying repentance by drawing an analogy with penance. He contends that punishment is akin to compulsory penance because it expresses a concern for the offender, who through punishment is given an opportunity to atone for the injury his or her crime has caused the community. Hard treatment makes the communicative aim more effective by forcing the criminal’s attention to the consequences of the crime, and it should also act as penance which the criminal wills for himself or herself.
This, however, ignores the fact that penance, unlike punishment, is assumed voluntarily. As Nigel Walker points out, penance must
be preceded by confession and contrition, and must be undertaken voluntarily, or at least in obedience to the instruction of the confessor. ... [In contrast] many offenders never admit their guilt, and ... many who do are unrepentant; or ... do not enter prison voluntarily or with uncoerced obedience.
In relation to the third point, Duff cannot provide a coherent explanation of how hard treatment achieves reparation, why reparation is necessary for reconciliation and, further still, why it is a precondition for reconciliation. Duff believes that the criminal owes the victim a material sign of repentant recognition — ‘[s]ome assurance that [the criminal] now disown[s] the hostility, indifference or contempt toward [the victim] which [the] wrong-doing displayed’ — and that formal condemnation cannot provide this. The critical issue is why a sincere apology cannot provide the necessary reparation — especially since Duff concedes that the type of reparation he is concerned with is moral, not material in nature. The only answer seems to be that the victim or the community demand more pain; otherwise the desire for revenge remains unsatisfied. Thus, once again Duff is cornered into intrinsic retributivism.
Despite these criticisms of the theories of Duff and von Hirsch, there is much to be said for the central plank of their respective theories: that censure is a desirable goal of punishment. It is simply that they fail to justify this goal. As noted earlier, the utilitarian has a simple answer: the censuring of criminal acts stigmatises criminal conduct, reinforces moral disapproval of such conduct and thereby makes it less likely that such behaviour will be engaged in. Duff and von Hirsch disagree with such an approach because they claim that it fails to explain why it is right to achieve such benefits by censuring offenders. For the utilitarian the answer is obvious: it will increase net happiness. The further step of imposing hard treatment is needed to provide an additional prudential deterrent not to engage in criminal conduct.
The unfair advantage theory of punishment is the view that offenders should be punished because they have taken unfair advantage of their victims and the community in general. The theory maintains that the criminal law confers benefits on all persons by prohibiting certain harmful acts and that these benefits can only be enjoyed if all people exercise self-restraint and do not infringe the criminal law. Law-breakers enjoy the benefits conferred by the law, but renounce the obligations (burdens) observed by the rest of the community. They deserve to be punished because by offending they have taken unfair advantage of the restraints observed by the rest of the community. Punishment, so the theory goes, restores the fair balance of benefits and burdens which is disturbed by crime.
This theory has been subjected to several criticisms, some of which are so damaging that the theory has been discarded by some of its most influential proponents. However, the unfair advantage theory again merits consideration due to a recent attempted resuscitation by George Sher, who tries to confront some of the more persuasive attacks that have been levelled against the theory.
One of these criticisms is that there is no workable definition of ‘unfair advantage’ which aptly captures the wrongfulness of crime: the exact nature of the benefit that offenders obtain from crime and the corresponding burden imposed on victims is not clear. It has been claimed that the benefit obtained by offenders is found in the renunciation of the burdens of self-restraint: ‘[a] person who violates the rules has something others have — the benefits of the system — but by renouncing what others have assumed, the burdens of self-restraint, he has acquired an unfair advantage.’
This version of the unfair advantage theory has been attacked on the basis that it entails that the harshness of the sanction should be commensurate with the strength of the desire to commit the offence, thereby leading to the perverse view that tax offenders should be punished more than murderers. Further, it is apparent that the repugnance of crime, apart from some forms of property offences, rarely has anything to do with obtaining a material or tangible benefit. Sometimes the exact opposite is true; the outcome of the offence (‘the benefit’) is the last thing which the offender wanted. Culpable driving is a good example. Offenders in such circumstances do not desire to kill and would be as pleased as the rest of the community if the results of their actions could be undone. Thus, the theory has difficulty in dealing with offences where the degree of freedom obtained by the offender has no correlation with the disadvantage incurred by the victim.
In a bid to overcome such problems, Sher suggests that the benefit derived by an offender consists of the extra freedom gained by the offender and that the benefit is measured by the strength of the constraint violated (not the strength of desire indulged). For Sher, the relevant constraint is moral rather than legal. In order to correct the balance that the offender has disrupted by his or her criminal behaviour, the victim (or the community acting on the victim’s behalf) must be freed from a constraint which is normally applicable and impose an unpleasantness on the wrongdoer which is about the same magnitude as the constraint violated by the offender.
Sher believes that the relevant constraints are moral rather than legal because our obligation to obey each law is the same and, accordingly, on a law-based criterion of benefits and burdens, every offender, from the shoplifter to the murderer, gains an equal amount of freedom and thus deserves the same amount of punishment. On the other hand, he contends that moral constraints are not equal, because some acts are more wrongful than others; hence, different wrongdoers will gain different amounts of freedoms: ‘the amount of freedom gained will be proportional to the wrongness of the act performed’.
However, this distinction between legal and moral norms is dubious. In relation to any norm, the degree to which compliance is obligatory depends not only upon the source of the norm but also upon its content. This is especially so in relation to the duty to obey the law. The strength of the obligation depends primarily on the nature of the law and the consequence of its non-observance. The fact that the source of a proscription is the law does not mean that we are equally duty bound to obey all prohibitions which share this origin. It is for this reason that many of us are normally vigilant in the observance of laws which prohibit the harming of others, but are at times almost indifferent to abiding by regulatory laws, such as those relating to parking or registration of pets. Sher believes that the strength of moral rules varies because it is accepted that one moral obligation can override another conflicting moral obligation. However, the same is also true in the case of legal obligations. Where laws appear to conflict it is accepted that the more important law prevails (or more accurately, obedience to the more important law will be a defence to breaching the lesser one). The legal duty to take one’s very sick child to the hospital will be a defence with respect to the speeding ticket received during the trip. Accordingly, it is not clear that Sher improves the theory by invoking a moral as opposed to a legal criterion of benefit.
At the core of the unfair advantage theory is the notion that benefits and burdens should be distributed equally. However, the most damaging criticism that has been levelled at the theory is that there is no true sense in which there is an equal distribution of benefits and burdens and therefore, it is argued, the theory is unable to explain coherently how offenders who have been previously wronged should be treated. If fairness requires an equal distribution of benefits and burdens, then offenders who have been previously wronged should have offence credits and be permitted enough transgressions to even the scales. Thus, the unfair advantage theory of punishment would not appear to have relevance in the real world. As Jeffrie Murphy has pointed out, there are enormous inequalities of income and opportunity in every society, and most offenders come from the most disadvantaged sectors of the community and cannot be said to derive any meaningful benefit from the rules of society. Given that the distribution of benefits and burdens is unfair in the first place, punishment will not restore any sense of fairness.
Sher attempts to circumvent this criticism by arguing that the relevant principle of fairness governs the distribution of benefits and burdens only between pairs of individuals as opposed to groups or individuals vis-à-vis the community. Thus, it follows that where A punches B and B punches A in return, we have a draw and punishment is not warranted. However, if B punches A because C has punched B or because B has previously been unfairly deprived of a benefit by the community (which is equivalent to freedom indulged in by punching another person), then B must be punished, even though from B’s perspective he or she has been subject to precisely the same amount of benefits and burdens. Sher claims that the benefit gained by B from punching A cannot be offset by the burden of being punched by another. It is on this basis that Sher approaches another interesting variant of the above situation. It has been argued that where
A assaults B, B assaults C, and C assaults A ... [i]n each case, an equivalent harm is inflicted on the victim. So we have three distinct acts of equal wrongdoing which proceed in a nice circle, balancing the benefits from moral restraint that each person receives. If the balancing of benefits and burdens is crucial to punishment, there should be no punishment.
Sher claims that in this situation all three deserve to be punished. Thus, focusing on the above instance of B assaulting C, the fact that B was previously assaulted by A has no bearing on the unfairness of B assaulting C because, according to Sher, the principle of fairness applies to pairs of individuals only. This unfairness will only be restored when C, or an agent acting on C’s behalf, punishes B.
Sher accepts that this narrow application of the principle of fairness, in so far as it governs the distribution of benefits and burdens associated with compliance with the law, may not be correct. He argues instead that a collective principle may be more accurate, since the benefits of the legal system require the cooperation of many. However, in the case of the distribution of benefits and burdens stemming from moral compliance, Sher advances two main reasons why the principle is not collective: firstly, the ‘benefits of moral behavior are not mainly collective’ and, secondly, ‘morality itself is standardly thought to be a set of rules that govern each person’s treatment of each particular other person’.
Sher’s argument misrepresents a fundamental aspect of morality. Broadly, morality consists of the principles that dictate how serious conflict should be resolved. This is not, however, confined to conflict between individuals. Typically, the conflict is between the interests of an individual and a large number of people and, in fact, the weightiest moral norms generally relate to conduct which affects large numbers of people. The harm caused by dumping toxic waste in the river or chopping down the rainforest is not confined to a single or particular individual, yet there is no question that such acts are almost invariably morally reprehensible. Whatever else moral rules are, they must surely have universal application. This means not only that the same rule applies where the conduct may affect person A or B, but also where it may affect A and B.
It is also wrong to claim that the benefits of morality are not collective. To assert otherwise is to underrate the pivotal role of morality in fostering and maintaining a harmonious and secure society. Even individuals who may not have benefited directly from the observance of moral rules by others have their lives enriched by the knowledge that moral norms are generally observed. Given that moral judgments can be universalised, this should give each individual confidence about the patterns of behaviour which he or she can expect from others. This is why, in a community in which each citizen abides by the same moral dictates, each individual would feel free to walk at night, ask a stranger the time and carry money in a handbag or wallet.
Thus, there seems to be no coherent basis upon which it can be contended that the scope of the principle of fairness should be confined to pairs of individuals. A bid to do so is merely an attempt to isolate the relevant balance of benefits and burdens in a manner which preserves the credibility of the unfair advantage theory of punishment. However, fairness is not so arbitrary.
Another curious feature of Sher’s theory is that he maintains that the victim is the ideal person who should inflict the punishment. As a result of the risk of reprisal inherent in this, he ultimately concedes that the state should act on the victim’s behalf. While this situation is not perfect, Sher maintains that ‘it represents a step in the right direction ... because it symbolically lifts the moral restraints on the victim, and ... treats the wrongdoer in a normally impermissible way’. This suggested role for the victim leads to suspicion that underlying the justification for punishment is the desire for personal revenge. If what is really important is that burdens and benefits are distributed equally, it should not matter how and by whom this distribution is achieved. Sher concedes that his account ‘does legitimate a form of retaliation’ and regards this ‘not as criticism of [his] account but rather as evidence that retaliation has received bad press’. This merely fuels the suspicion that the unfair advantage theory is being used to cloak the desire for revenge in the language of fairness; in which case the balancing of benefits and burdens is irrelevant to Sher’s theory.
Further, while it may be fair to ensure that distribution of benefits and burdens is not violated by immoral behaviour, this does not necessarily justify punishment. For it is also desirable that people should not be subjected to the type of unpleasantness that is constituted by punishment. When these two virtues clash, Sher weighs the former more heavily, but no reason is given for this. If it transpired that punishment increased the crime rate, this would further upset the fair distribution, yet Sher would still be compelled to punish, making his theory self-defeating.
Accordingly, Sher’s attempt to revive the unfair advantage theory should be rejected on the basis that the theory still succumbs to the decisive objections which have been levelled at it.
Retributivism has been the dominant theory of punishment for the past two or three decades. This is despite considerable definitional problems associated with the concept and a lack of convergence toward any particular retributive theory. The enormous array of contemporary retributive theories makes it impossible to examine exhaustively each theory. This paper has focused on what we consider to be the four most influential retributive theories.
Examined closely, each of the theories fails to justify the link between crime and punishment. Von Hirsch’s theory is incoherent because, ultimately, he relies on consequential considerations to justify key elements of his theory. This being so, he offers no reason why we should not adopt a purely consequential theory. Duff’s theory mischaracterises the nature of punishment, while Sher’s attempted revival of the unfair advantage theory is ultimately unable to circumvent the deep problems which have previously beset the theory. Intrinsic retributivism is the most coherent retributive theory, which is somewhat ironic given that it is the least popular theory amongst contemporary adherents. Ultimately, it too is unconvincing because the desire for revenge is an inappropriate moral ideal. However, because of its internal coherence, intrinsic retributivism provides the firmest platform of all retributive theories from which to launch a counterattack. Its future prosperity is likely to turn on future developments in the metaethical domain.
The sheer number and breadth of retributive theories means that it is premature to assert positively that no retributive theory of punishment can possibly justify punishment. However, given the breadth and popularity of the retributive theories that have been examined in this paper, enough has been said in order to redirect the search for a coherent justification for punishment. The most promising path is the one leading towards utilitarianism. Without adverting to consequentialist considerations, it is impossible to justify the link between crime and punishment.
[*] BA, LLB (Hons), LLM (Monash); Lecturer, School of Law, Deakin University.
[†] LLB (Hons), LLM (London), PhD (Monash); Senior Lecturer, Faculty of Law, Monash University. We wish to thank the anonymous referees for their detailed and insightful comments on an earlier version of this paper.
 For an overview of the academic and social trends in punishment, see Antony Duff and David Garland, ‘Introduction: Thinking about Punishment’ in Antony Duff and David Garland (eds), A Reader on Punishment (1994) 1, 8–16; Andrew von Hirsch, Past or Future Crimes: Deservedness and Dangerousness in the Sentencing of Criminals (1985) ch 1; Nigel Walker, Why Punish? (1991) 3–18; Andrew Ashworth, Sentencing and Criminal Justice (2nd ed, 1995) 69–72.
 The revival of retributivism is due largely to the work of Andrew von Hirsch, particularly Doing Justice: The Choice of Punishments (1976); von Hirsch, Past or Future Crimes, above n 1. In the United States, the just deserts model was responsible for the move away from wide discretionary sentencing powers to laws aimed at promoting greater certainty and consistency in sentencing, such as the Minnesota Criminal Code, 609 MINN STAT ss 107, 109 (1963). Legislation in Washington and Oregon also expressly adopts a just deserts-based philosophy: Ashworth, Sentencing and Criminal Justice, above n 1, 330–2. For a Scandinavian approach, see, eg, Nils Jareborg, ‘The Swedish Sentencing Reform’ in Chris Clarkson and Rod Morgan (eds), The Politics of Sentencing Reform (1995) 95; Tapio Lappi-Seppala, ‘Regulating the Prison Population: Experiences from a Long-Term Policy in Finland’ (Paper presented at the ‘Beyond Prisons’ Symposium, Kingston, Canada, 17 March 1988) <
http://www.csc-scc.ca/text/forum/bprisons/english/fine.htm> (copy on file with author).
 The unprincipled nature of sentencing practice has led to what Ashworth labels a ‘cafeteria system’ of sentencing, which permits sentencers to pick and choose a rationale which seems appropriate at the time: Ashworth, Sentencing and Criminal Justice, above n 1, 331. This is made significantly easier by the large number of discrete factors that the courts have identified as being relevant to sentencing. Two separate studies, about 20 years ago, determined that there were many such factors: see Joanna Shapland, Between Conviction and Sentence (1981) 55; Roger Douglas, Guilty, Your Worship: A Study of Victoria’s Magistrates’ Courts (Occasional Monograph No 1, La Trobe University Legal Studies Department, 1980). Douglas identified 292 relevant sentencing factors in a study of Victorian Magistrates’ Courts: Douglas, in this footnote. The results of such studies were noted in Pavlic v The Queen  TASSC 96; (1995) 5 Tas R 186, 202. Slicer J stated that ‘it is impossible to allocate to each relevant factor a mathematical value and from that, extrapolate a sum which determines the appropriate penalty.’ For the need for reform in this area, see Mirko Bagaric, ‘Sentencing: The Road to Nowhere’  SydLawRw 23; (1999) 21 Sydney Law Review 597.
 For an overview of many of the theories, see C L Ten, Crime, Guilt and Punishment (1987)
38–65; John Cottingham, ‘Varieties of Retribution’ (1979) 29 Philosophical Quarterly 238; Ted Honderich, Punishment: The Supposed Justifications (revised ed, 1984) 211.
 See Honderich, above n 4; David Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’ (1997) 16 Law and Philosophy 507, 507.
 See, eg, Antony Duff and Andrew von Hirsch, ‘Responsibility, Retribution and the “Voluntary”: A Response to Williams’ (1997) 56 Cambridge Law Journal 103, 107.
 Jamie Anderson, ‘Reciprocity as a Justification for Retributivism’ (1997) 16 Criminal Justice Ethics 13, 14.
 Von Hirsch, Past or Future Crimes, above n 1, 10.
 See Andrew von Hirsch, ‘The Politics of “Just Deserts”’ (1990) 32 Canadian Journal of Criminology 397, 397. See also Andrew von Hirsch, ‘Censure and Proportionality’ in Antony Duff and David Garland (eds), A Reader on Punishment (1994) 115, 125, where von Hirsch states: ‘[w]ere penalties ordered in severity inconsistently with the comparative seriousness of the crime, the less reprehensible conduct would, undeservedly, receive the greater reprobation’.
 It provides an upper and lower ceiling in relation to the amount of punishment that should be inflicted for any particular offence: Richard Fox, ‘The Meaning of Proportionality in Sentencing’  MelbULawRw 1; (1994) 19 Melbourne University Law Review 489, 495.
 For example, Bentham argued that proportionality has a secure utilitarian foundation and that, in fact, it has a central role in a utilitarian theory of punishment. Bentham asserted, as a secondary principle, that crimes should be punished in proportion to the harm done to the life and security of others in society. If crimes are to be committed, it is preferable that offenders commit less serious rather than more serious ones. Therefore, sanctions should be graduated commensurate with the seriousness of the offence so that those disposed to crime will opt for less serious offences. Absent proportionality, potential offenders would not be deterred from committing serious offences any more than minor ones and, hence, would just as readily commit them: Jeremy Bentham, An Introduction to the Principles of Morals and Legislation (first published 1789, J H Burns and H L A Hart (eds), 1970 ed) 178–88 (‘Principles of Morals and Legislation’). This argument has been persuasively criticised by von Hirsch, who points out that there is no evidence that offenders make comparisons regarding the level of punishment for various offences: von Hirsch, Past or Future Crimes, above n 1, 32. For a utilitarian argument in favour of proportionality, see Mirko Bagaric, ‘Suspended Sentences and Preventive Sentences: Illusory Evils and Disproportionate Punishments’  UNSWLawJl 6; (1999) 22 University of New South Wales Law Journal 535.
 Anderson, above n 7, 13. Some retributivists, such as Immanuel Kant and Georg Hegel, make the additional claim that punishment of wrongdoers is not only just, it is obligatory. However, it has been argued that even if punishing the guilty is an intrinsic good, there is still no moral duty to bring about such a state of affairs: see Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 518–22. Dolinko claims that there are two propositions which are defining of retributivism. He refers to them as the ‘intrinsic good claim’ and the ‘desert claim’. These claims assert that the guilty receiving their just deserts is an intrinsic good, and that punishment is justified solely by the fact that those upon whom it is inflicted deserve it. The first claim here really embodies two propositions: one, that punishing the guilty is an intrinsic good; the other, that the level of punishment should be in accordance with one’s just deserts. It should be noted that the other supposed central plank of retributivism, the desert claim (as stated by Dolinko), adds nothing to the intrinsic good claim. In fact, rather than being a defining aspect of retributivism, it appears to be at most a logical consequence of the intrinsic good claim: that is, the guilty deserve to suffer because the guilty receiving their just deserts is an intrinsic good. In any event, Dolinko argues that while these claims are characteristic features of retributivism, they are ultimately incompatible because the intrinsic good claim suggests that bringing about good is what justifies punishment, thereby invoking consequential considerations, while the desert claim provides that extraneous consequences cannot be relevant to the justification of punishment: at 516. However, this is only so if one adopts a consequential interpretation of the intrinsic good claim and (as Dolinko concedes) this is a claim which most retributivists, at least ostensibly, reject.
 See Honderich, above n 4, 212; John Kleinig, Punishment and Desert (1973) 67; D J B Hawkin, ‘Punishment and Moral Responsibility’ in Stanley Grupp (ed), Theories of Punishment (1971) 13, where he asserts that at the pre-reflective level it seems to be assumed that a guilty act deserves punishment. Intrinsic retributivism is considered at length in the discussion below.
 See generally von Hirsch, Past or Future Crimes, above n 1. This theory is considered below in Part VIII.
 Antony Duff, Trials and Punishments (1986). Duff’s theory is considered at length below in Part IX.
 Herbert Morris, ‘Persons and Punishment’ in Jeffrie Murphy (ed), Punishment and Rehabilitation (1973) 40; Jeffrie Murphy, Retribution, Justice and Therapy (1979) 82–115; Wojciech Sadurski, Giving Desert Its Due (1985) ch 8; John Finnis, Natural Law and Natural Rights (1980) 263–4. Recently, Sher has made a spirited attempt to revive the unfair advantage theory. This is considered below in Part X.
 However, it has been claimed that retributivism could be formulated as a consequentialist theory: Michael Moore, ‘Justifying Retributivism’ (1993) 27 Israel Law Review 15. For a convincing criticism of this, see Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 510, who argues that a non-deontological account of retributivism evinces a poor understanding of retributivism and that consequentialist retributivism is not a coherent form of retributivism. Dolinko must be accurate here. One of the key criticisms of many retributive theories is that logically they invoke consequential considerations, hence they do not provide a viable alternative to utilitarianism: see, eg, Ten, Crime, Guilt and Punishment, above n 4, 38–65, and the discussion below. If retributivism expressly invokes consequential considerations, then arguably it has nothing left to present as a viable alternative to utilitarianism.
 Duff, Trials and Punishments, above n 15, 7.
 It is also claimed that future-orientated considerations — the defendant’s need for treatment, his or her likelihood of offending again, the deterrent effect of his or her punishment on others — have no role in determining the comparative severity of penalties: von Hirsch, Past or Future Crimes, above n 1, 10.
 The retributive traits we advert to here are similar to those in the definition of retributivism adopted by Dolinko, who defines retributivism as any theory that ‘explains either the rational justification of punishment, or its moral justification, or both, by appealing to the notion that criminals deserve punishment rather than to the consequentialist claim that punishing offenders yields better results than not punishing them’: David Dolinko, ‘Some Thoughts about Retributivism’ (1991) 101 Ethics 537, 541–2. It should be noted, however, that retributivism does not adopt any particular moral theory.
 Ten, Crime, Guilt and Punishment, above n 4, 4. But see the discussion on the dominion theory of punishment, below Part V(B).
 Apart from hedonistic act utilitarianism, several other utilitarian theories have been advanced. Ideal utilitarianism is the theory that in addition to happiness there are other intrinsic goods such as knowledge, love and beauty (see George Moore, Principia Ethica (1903)) and, accordingly, we should also attempt to maximise these virtues. Ideal utilitarianism is unstable and ultimately collapses into hedonistic utilitarianism. It is true that we generally pursue virtues such as love, beauty and knowledge, but we do not do so for their own sake. Rather, we seek them because they generally tend to generate pleasure. To the extent that we desire other things such as money, power, virtue or fame, it is only because they are generally a means to happiness. This, however, does not change the derivative attraction of such virtues: see John Stuart Mill’s 1859 essay, ‘Utilitarianism’ in Mary Warnock (ed), Utilitarianism, On Liberty, Essay on Bentham Together with Selected Writings of Jeremy Bentham and John Austin (1969 ed) 251, 290 and D D Raphael, Moral Philosophy (1981) 34–43.
The most recent substitution of note is to define utility in terms of preference or desire satisfaction. The corresponding theory is called preference utilitarianism. Preference utilitarianism does not have the same degree of self-evident appeal as hedonistic utilitarianism. For example, it is unclear why we should seek to maximise desires which make people unhappy. Further, it is impossible to know which act will maximise desire satisfaction, given the overwhelming number of desires which will invariably need to be considered in any particular case. Also, it may be argued that our ultimate fundamental desire is generally, if not always, to be happy. Hence, preference utilitarianism, too, collapses into hedonistic utilitarianism. Preference utilitarianism is outlined in Richard Hare, Moral Thinking: Its Levels, Methods and Point (1981) and Peter Singer, Practical Ethics (2nd ed, 1993) 14. Singer states here that if happiness is defined broadly enough to include achieving what one desires (as we believe is the case), then there is no conflict between hedonistic and preference utilitarianism: at 14.
A further distinction is made between act utilitarianism and rule utilitarianism. Act utilitarianism is simply the theory that the correctness of an action is judged according to the degree of utility it promotes. Rule utilitarianism is the theory that the rightness of an act is assessed by reference to its compliance with rules established to maximise utility. For the rule utilitarian the principle of utility is used as a guide for the rules we should follow, as distinct to the particular actions we should perform. As a result of the difficulty in performing the utilitarian calculus necessary to determine which of a number of options we should choose, it is claimed that a set of rules guiding us in our decisions would be more likely to achieve the desired goal. The main problem with rule utilitarianism is that it is inevitable that in complying with the rules there will be occasions when happiness will not be maximised. To refuse to break the rule in such circumstances constitutes ‘rule-worship’: see John Smart, ‘An Outline of a System of Utilitarian Ethics’ in John Smart and Bernard Williams (eds), Utilitarianism: For and Against (1973) 3, 10. It is no answer that in most cases it is beneficial to comply with the rule, otherwise we are putting the rule above its justification. If we do break the rule, we are still being guided by the ultimate principle: act utilitarianism.
 For an overview of the literature on deterrence, see James Wilson, ‘Penalties and Opportunities’ in Antony Duff and David Garland (eds), A Reader on Punishment (1994) 177, where he argues that the main factor relevant to deterrence is not the penalty level, but rather the perceived probability of apprehension. This does not necessarily diminish the importance of punishment. The likelihood of being caught is only undesirable because of the accompanying realisation that punishment may follow. This is discussed further below.
 Anthony Bottoms, ‘An Introduction to the Coming Crisis’ in Anthony Bottoms and Ronald Preston (eds), The Coming Penal Crisis: A Criminological and Theological Exploration (1980) 1.
 Robert Martinson, ‘What Works? — Questions and Answers about Prison Reform’ (1974) 35 Public Interest 22, 22. Following research conducted between 1960 and 1974, Martinson initially noted that empirical studies had not established that any rehabilitative programs had worked in reducing recidivism. Martinson, however, softened his position several years later, concluding that some types of rehabilitation programs, particularly probation parole, may be effective and that generally ‘no treatment ... is inherently either substantially helpful or harmful. The critical factor seems to be the conditions under which the program is delivered’: Robert Martinson, ‘New Findings, New Views: A Note of Caution Regarding Sentencing Reform’ (1979) 7 Hofstra Law Review 243, 254 (emphasis in original).
 See, eg, Stanley Cohen, Visions of Social Control (1985).
 Given the difficulties in defining the factors that are relevant to proportionality, ‘approximate’ just deserts is possibly the most that can be hoped for. For a discussion regarding the considerations relevant to proportionality, see Andrew von Hirsch and Nils Jareborg, ‘Gauging Criminal Harm: A Living-Standard Analysis’ (1991) 11 Oxford Journal of Legal Studies 1. They state that there are several steps involved in gauging the seriousness of an offence. The first involves an appraisal of the types of interests which the paradigm instance of an offence violates or threatens to infringe upon. They identify four basic types of interests. In order from most to least important, they are: physical integrity; material support and amenity (such as nutrition and shelter); freedom from humiliating or degrading treatment; privacy and autonomy. Next it is necessary to determine the effect that violating the relevant interests typically has on the victim. Finally, it is necessary to consider the offender’s culpability and the remoteness of the harm: at 19.
 John Mabbott, ‘Punishment’ in Stanley Grupp (ed), Theories of Punishment (1971) 41, 44–5.
 John Rawls, A Theory of Justice (1972).
 Rights talk includes the abundance of declarations, charters, bills and the like, such as the Universal Declaration of Human Rights, GA Res 217A, 3 UN GAOR (183rd plen mtg), UN Doc A/Res/217A (1948); the International Covenant of Economic, Social and Cultural Rights, opened for signature 19 December 1966, 993 UNTS 3, 6 ILM 360 (entered into force 3 January 1976); and the European Convention for the Protection of Human Rights and Fundamental Freedoms, opened for signature 4 November 1950, 213 UNTS 221 (entered into force 3 September 1953), which seek to spell out certain rights. There were numerous declarations, and the like, of rights prior to the Second World War, such as the Declaration of Independence of the United States (1776) and the French Déclaration des Droits de l’Homme et du Citoyen (1789) [trans: Declaration of the Rights of Man and of the Citizen], however it is only in relatively modern times that such documents have gained widespread appeal, recognition and force.
 L W Sumner, The Moral Foundation of Rights (1987) 1.
 H L A Hart, Essays in Jurisprudence and Philosophy (1983) 196–7.
 For a historical account of how punishing the innocent has accounted for the rejection of utilitarianism, see F Rosen, ‘Utilitarianism and the Punishment of the Innocent: The Origins of False Doctrine’ (1997) 9 Utilitas 23.
 Kleinig, above n 13, 67.
 Von Hirsch, Doing Justice, above n 2, 69–71.
 See, eg, Mirko Bagaric, ‘In Defence of a Utilitarian Theory of Punishment: Punishing the Innocent and the Compatibility of Utilitarianism and Rights’ (1999) 24 Australian Journal of Legal Philosophy 95; Mirko Bagaric, ‘Incapacitation, Deterrence and Rehabilitation: Flawed Ideals or Appropriate Sentencing Goals’ (2000) 24 Criminal Law Journal 21, 37–8.
 See below nn 38, 39 and 47 and accompanying text.
 John Monahan, ‘The Prediction of Violent Behaviour: Toward a Second Generation of Theory and Policy’ (1984) 141 American Journal of Psychiatry 10, 11. Another study revealed a false positive rate of about 65 percent: see Harry Kozol, ‘Dangerousness in Society and Law’ (1982) 13 University of Toledo Law Review 241, 267. For an overview of the literature in this area, see Bagaric, ‘Suspended Sentences and Preventive Sentences’, above n 11, 535.
 Greenwood claimed that it was possible to identify high risk robbers and burglars by identifying seven supposed risk factors (similar prior convictions; incarceration for over a year in the previous two years; convictions at a young age; time served in a juvenile facility; use of drugs in the past two years; drug use as a juvenile; and employed for less than a year in the last two years) and, hence, significantly reduce the number of such offences by increasing the prison terms for the high risk offenders: Peter Greenwood, Selective Incapacitation: Report Prepared for the National Institute of Justice (1982) xv–xvi, 50. However, it seems that the technique used was flawed. A re-examination of the original data resulted in less promising results: see Alfred Blumstein, Jacqueline Cohen and Carol Visher, Criminal Careers and ‘Career Criminals’ (1986); Andrew von Hirsch, ‘Selective Incapacitation: Some Doubts’ in Andrew von Hirsch and Andrew Ashworth (eds), Principled Sentencing (1992) 158, 159–60.
 On the basis of the definition in the Criminal Justice Act 1985 (NZ) s 2. This essentially relates to crimes of serious violence, such as manslaughter, wounding and robbery.
 Mark Brown, ‘Serious Violence and Dilemmas of Sentencing: A Comparison of Three Incapacitation Policies’ (1998) Criminal Law Review 710.
 Ibid 714.
 Ibid 715.
 Franklin Zimring and Gordon Hawkins, Incapacitation: Penal Confinement and the Restraint of Crime (1995) 86.
 Jacqueline Cohen, ‘The Incapacitative Effect of Imprisonment: A Critical Review of the Literature’ in Alfred Blumstein, Jacqueline Cohen and Daniel Nagin (eds), Deterrence and Incapacitation: Estimating the Effects of Criminal Sanctions on Crime Rates (1978) 187, 188.
 Kevin Howells and Andrew Day, ‘The Rehabilitation of Offenders: International Perspectives Applied to Australian Correctional Systems’ (1999) 112 Trends and Issues in Crime and Criminal Justice 1.
 Ibid 3.
 Michael King, ‘Missouri’s Enlightened Sentencing Project’ (2000) 6(1) Hearsay 6; Darrin Farrant, ‘Meditation: A Serene but Effective Path to Criminal Enlightenment’, The Age (Melbourne), 16 November 1999, 6; Enlightened Sentencing Project, Enlightened Sentencing Project, Brochure (1999) (copy on file with author).
 Andrew von Hirsch and Lisa Maher, ‘Should Penal Rehabilitationism Be Revived?’ in Andrew von Hirsch and Andrew Ashworth (eds), Principled Sentencing (1992) 41, 42.
 For a discussion of the vast array of definitions of punishment that have been advanced over the ages, see Bagaric, ‘Suspended Sentences and Preventive Sentences’, above n 11, 535.
 The distinction between punishment and a social service is employed by von Hirsch and Maher: above n 50.
 Howells and Day, above n 47.
 Ibid 4–5.
 For example, in Victoria community-based orders or intensive corrections orders (which both have a large counselling/rehabilitative component) cannot be imposed without the consent of the offender: Sentencing Act 1991 (Vic) ss 20, 36.
 This is a point alluded to by Herbert Packer over a quarter of a century ago. He believed that the task of creating a society where the advent of crime was significantly reduced is beyond the compass of the practice of punishment or the criminal law; it would require us to ‘remake society itself’. Further, the process of rectifying inadequate social conditions is not one which can be accomplished in the context of processes that are devoted to apprehending, trying and sentencing offenders: Herbert Packer, ‘Theories of Punishment and Correction: What Is the Function of Prison?’ in Leonard Orland (ed), Justice, Punishment, Treatment (1973) 183, 189.
 Jeremy Bentham, ‘The Principles of Penal Law’ in John Bowring, The Works of Jeremy Bentham (1843) 396 (emphasis added). For Bentham, specific deterrence and rehabilitation were second order objectives after general deterrence.
 See also Daniel Nagin, ‘General Deterrence: A Review of the Empirical Evidence’ in Alfred Blumstein, Jacqueline Cohen and Daniel Nagin (eds), Deterrence and Incapacitation: Estimating the Effects of Criminal Sanctions on Crime Rates (1978) 95, 95, where Nagin adverts to the difficulty in distinguishing between rehabilitation and specific deterrence.
 See David Farrington, ‘Age and Crime’ in Michael Tonry and Norval Morris (eds), Crime and Justice: An Annual Review of Research  189; Daniel Nagin, David Farrington and Terrie Moffitt, ‘Life-Course Trajectories of Different Types of Offenders’ (1995) 33 Criminology 111.
 Panel on Research on Deterrent and Incapacitative Effects, ‘Incapacitation’ in Alfred Blumstein, Jacqueline Cohen and Daniel Nagin (eds), Deterrence and Incapacitation: Estimating the Effects of Criminal Sanctions on Crime Rates (1978) 66.
 Ibid 7.
 Franklin Zimring and Gordon Hawkins, Deterrence: The Legal Threat in Crime Control (1973) 29 (citations omitted). For more recent discussion of the failure of punishment to act as a marginal deterrent, see Lisa Stolzenberg and Stewart D’Alessio, ‘“Three Strikes and You’re Out”: The Impact of California’s New Mandatory Sentencing Law on Serious Crime Rates’ (1997) 43 Crime and Delinquency 457. See also Albert Reiss and Jeffrey Roth (eds), Panel on Understanding and Control of Violent Behaviour (1993) 6–7, 293–4; Daniel Nagin, ‘Criminal Deterrence Research at the Outset of the Twenty-First Century’ (1998) 23 Crime and Justice 1.
 Roderick Broadhurst and Nini Loh, ‘Selective Incapacitation and the Phantom of Deterrence’ in Richard Harding (ed), Repeat Juvenile Offenders: The Failure of Selective Incapacitation in Western Australia (2nd ed, 1995) 55, 71. The number of actual police pursuits was not an accurate indicator because, following the introduction of the Act, police pursuit practices changed, resulting in many pursuits being called off or not being commenced in the first place.
 The discussion regarding the events of the strike comes from Kerry Milte and Thomas Weber, Police in Australia (1977) 287–92.
 Andrew Ashworth, ‘Deterrence’ in Andrew von Hirsch and Andrew Ashworth (eds), Principled Sentencing (1998) 44, 51 refers to the Liverpool strike, and the Danish experience is discussed in Nigel Walker, Sentencing: Theory, Law and Practice (1985) 85.
 See Thomas Mathiesen, ‘General Prevention as Communication’ in Antony Duff and David Garland (eds), A Reader on Punishment (1994) 221, 231.
 Kenneth Wolpin, ‘An Economic Analysis of Crime and Punishment in England and Wales, 1894–1967’ (1978) 86 Journal of Political Economy 815, 837–8.
 See discussion in the text accompanying nn 64–66 and accompanying text. See also Bagaric, ‘Incapacitation, Deterrence and Rehabilitation’, above n 36, 37–8.
 For a discussion of the compatibility of utilitarianism and proportionality, see Bagaric, ‘Suspended Sentences and Preventive Sentences’, above n 11, 560–2.
 Duff gives two other types of examples of what he terms punishing the innocent: punishing people for strict liability offences and punishing someone more severely than is commensurate with the seriousness of the offence: Duff, Trials and Punishments, above n 15, 154–5. These situations will not be addressed in this paper. It is questionable whether such people are innocent, and in any event these situations do not represent the paradigm case of punishing the innocent.
 H J McCloskey, Meta-Ethics and Normative Ethics (1969) 180–1. A similar example to McCloskey’s is provided in E F Carritt, Ethical and Political Thinking (1947) 65.
 See, eg, Timothy Sprigge, ‘A Utilitarian Reply to Dr McCloskey’ (1965) 8 Inquiry 264, 272.
 As an example, McCloskey’s hypothetical could be altered by providing that the town was an isolated one and that there is no opportunity for help to arrive before the riots occur. Also, the crime should be murder, not a rape, in which case there is one less person who could reveal the miscarriage of justice that has occurred, and thus the risk of a possible loss of respect and confidence in the law is not as significant. See also Ten, Crime, Guilt and Punishment, above n 4, 18; John Rawls, ‘Two Concepts of Rules’ (1955) 64 Philosophical Review 3. Brandt and Rawls have argued that the rule utilitarian is not necessarily committed to punishing the innocent: see, eg, Richard Brandt, Ethical Theory: The Problems of Normative and Critical Ethics (1959) 490–5; Rawls, in this footnote, 3. For a counter, see H J McCloskey, ‘A Non-Utilitarian Approach to Punishment’ (1965) 8 Inquiry 249; Ten, Crime, Guilt and Punishment, above n 4,
 Ten, Crime, Guilt and Punishment, above n 4, 18–25, draws a distinction between a fundamental moral principle (a principle which is not justified by reference to some further moral principle) and a secondary moral principle (which has to be justified by appeal to some further moral principle), and makes the point that fantastic examples play an important role in relation to the evaluation of fundamental moral principles. For the utilitarian, the wrongfulness of punishing the innocent is a secondary principle, deriving its justification from the sole utilitarian fundamental principle that it is wrong because it would cause net unhappiness; for others, the proscription against punishing the innocent is itself a fundamental moral principle. Accordingly, fantastic examples have a role in testing both this principle and the fundamental utilitarian principle. Fantastic examples also allow us to ascertain whether or not a principle is fundamental.
 The distinction we are making between intuitive moral judgments and those formed after due reflection is similar to that made by Hare between intuitive and critical levels of moral thinking: see Hare, above n 22, 25.
 Whilst this is not normally the case, ie, we normally like to think that we send our soldiers into situations with at least a fighting chance, there are countless reported instances of soldiers being ordered to go or remain in situations which can only be described as suicide missions. For those brave people who place themselves in such situations voluntarily, it is rather illuminating that the proscription against suicide disappears. They are heroes rather than bad people — they follow the dictates of utilitarianism.
 The classic deontological response to this and other examples, the doctrine of double effect, is discussed below at nn 86–93 and accompanying text.
 See Manuel Velasquez and Cynonia Rostankowski, Ethics: Theory and Practice (1985) 103–6; see also Allen Kurki, Operation Moonlight Sonata: The German Raid on Coventry (1995). A famous modern day example which comes closest to the dilemma of choosing whether to frame the innocent or tolerate massive abuses of rights followed the Rodney King beating in Los Angeles on 3 March 1991. The four policemen who beat King were acquitted under state criminal law of any offence regarding the incident. Riots ensued, resulting in widespread looting, damage to property and dozens of deaths. Shortly afterwards, the Federal Government announced the almost unprecedented step that the policemen, who one must remember were found not guilty of any offence, would be tried on federal civil rights charges relating to the incident. They were duly found guilty, despite the apparent double jeopardy involved. See, eg, Jewelle Gibbs, Race and Justice (1996); David Cole, No Equal Justice: Race and Class in the American Criminal Justice System (1999) 23. It seems clear that the Government’s motivation for bringing federal charges against the policemen was utilitarian.
 This has been used as an argument against a naturalistic view of morality. However, see Charles Pigden, ‘Naturalism’ in Peter Singer (ed), A Companion to Ethics (1991) 421, 422–6, where he points out that this phenomenon simply reflects the conservative nature of logic — you cannot get out of it what you do not put in.
 George Schedler, ‘Can Retributivists Support Legal Punishment?’ (1980) 63 Monist 185, 196. For this reason he concluded that retributivists simply cannot support the institution of punishment: at 196.
 This point is also made by Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 510; see also Douglas Husak, ‘Why Punish the Deserving?’ (1992) 26 Noûs 447, 450–1.
 Though perhaps no amount of procedural safeguards could ever fully prevent the conviction of at least some innocent people.
 In circumstances where there was no possibility that it was procured by threat or inducement or was otherwise involuntary.
 There may, however, be community support for increasing the level of safeguards to protect against wrongful conviction. A recent American study revealed that the vast majority of respondents supported a higher standard of certainty of guilt in cases involving the death penalty: David Weinstock and Gary Schwartz, ‘Executing the Innocent: Preventing the Ultimate Injustice’ (1998) 34 Criminal Law Bulletin 328, 339–40. However, given the finality and extreme nature of this form of punishment, it may well be that similar sentiments will not apply in relation to other forms of punishment.
 An attempt to attenuate the principle that people who are not blameworthy in any way should not be harmed by confining harm to legally imposed sanctions would appear indefensible.
 Duff, Trials and Punishments, above n 15, 172. The same point is made by Moore, ‘Justifying Retributivism’, above n 17, 20.
 It is also sometimes contended that a further condition is that the act must not be intrinsically bad: see H Tristram Engelhardt and J Kenny, ‘Principle of Double Effect’ in Baruch Brody and
H Tristram Engelhardt (eds), Bioethics: Readings and Cases (1987) 160. However, given that the doctrine is commonly applied to very grave cases involving things such as the killing of innocent people and it is on the basis of the doctrine itself that such acts are sought to be justified, it begs the question to make such a condition an internal part of the doctrine. See also Thomas Nagel, The View from Nowhere (1986) 179, whose formulation of the doctrine of double effect essentially accords with the above.
 Although there are also other justifications for the defence of self-defence.
 See, eg, Paul Mullen, ‘Euthanasia: An Impoverished Construction of Life and Death’ (1995) 3 Journal of Law and Medicine 121, 127.
  EWCA Crim 2;  1 WLR 1025, 1028. In Nedrick the appellant was convicted of murder after a child died in a house to which he had set fire. The House of Lords allowed his appeal and substituted a verdict of manslaughter.
  UKHL 2;  AC 55, 74. In Hyam the appellant was convicted of murder after setting fire to the house of the fiancée of a man with whom she had had a relationship. The House of Lords dismissed her appeal, holding that a person who, without intending to endanger life, did an act knowing that it was probable that grievous, in the sense of serious, bodily harm would result was guilty of murder. To explain this incongruity between foresight and intention, it has been suggested that where the motive is honourable there is room to distinguish between foresight and intention: David Lanham, ‘Euthanasia, Pain Killing, Murder and Manslaughter’ in John McKie (ed), Active Voluntary Euthanasia: The Current Issues (1994) 67, 73. However, this cannot be used to give a general account of the difference between that which is intended and foreseen, since this distinction is itself meant to be a test by which the moral status of an act can be evaluated. The doctrine would be redundant if the moral status of the act was clear from the outset.
  UKHL 5;  AC 789, 867. See also the Medical Treatment Act 1988 (Vic).
 See, eg, R v Cox (1992) 12 BMLR 38, 48; Auckland Area Health Board v A-G (NZ)  1 NZLR 235, 248; Re J (Wardship: Medical Treatment)  Fam 33, 46. The doctrine is the cornerstone of the Consent to Medical Treatment and Palliative Care Act 1995 (SA), which provides that where there is a conflict between proper control of symptoms and accelerating the timing of an inevitable death, then symptom control prevails. If a side effect of palliative care is death, this is deemed not to constitute death in law: see s 17(1)(c). This is consistent with the doctrine of double effect.
 Jonathon Glover, Causing Deaths and Saving Lives (1977) 88. It could be argued that the doctrine does not apply in this situation because of a lack of proportionality between the good and bad effects of the act. However, this could be answered by altering the example so that only one person was killed in the explosion, and the protest was against a brutal regime which had a history of bowing to such acts of aggression.
 See also Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 510–13, who considers Foot’s example of the wicked merchants selling oil they know to be poisonous and thereby killing innocent people: Philippa Foot, Virtues and Vices and Other Essays in Moral Philosophy (1978) 22.
 See also Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 510–13.
 Rachels argues that a person’s intention is not relevant to determining whether an act is right or wrong, but instead is relevant to assessing the character of the person who does it: James Rachels, The End of Life: Euthanasia and Morality (1986) 94–6. However, the difficulty with this is coherently to distinguish between the evaluation of the act and the agent; we normally judge people by their actions.
 J L Mackie, Ethics: Inventing Right and Wrong (1977) 161.
 Immanuel Kant, Groundwork for the Metaphysics of Morals (H J Paton trans, first published 1785, 1948 ed) 90–3.
 This conclusion is, however, by no means certain. A utilitarian could argue that earmarked deaths are worse than statistical ones because of the desensitisation that would follow if defined individuals were allowed to die. We seem to be built in such a way that when an identifiable individual is experiencing pain and suffering (or is in need of help) this impacts on us far more than when it is experienced by faceless strangers. Thus, in 1995, the Australian Government spent A$5.8 million rescuing French sailor Isabelle Autissier who was stranded while on a solo voyage around the world: Senate Foreign Affairs, Defence and Trade Legislation Committee, Consideration of Estimates: Parliamentary Debates, 27 February 1997, 24 (Senator Michael Forshaw), when the same money could have saved thousands of starving people around the world. Thus, unless the number of statistical deaths is significantly more than earmarked ones, a utilitarian may prefer to opt for bridge A.
 We thank Professor C L Ten for this example, which was provided during a discussion with one of the authors.
 See above Part IV(B)(2)(a).
 See Bagaric, ‘In Defence of a Utilitarian Theory of Punishment’, above n 36, 137–42.
 H L A Hart, Punishment and Responsibility (1968). In particular, see the first essay in this book, ‘Prolegomenon to the Principles of Punishment’. Hart claims that the principles of justice are not derivative from either retributivism or utilitarianism. Ten also urges a similar form of compromise theory. However, he believes that retributive considerations can also provide reasons justifying punishment: Ten, Crime, Guilt and Punishment, above n 4, 64–5.
 New South Wales Law Reform Commission, Sentencing, Discussion Paper No 33 (1996) 56.
 For a discussion regarding pluralistic or compromise theories of punishment, see Ten, Crime, Guilt and Punishment, above n 4, 71–81.
 Hart, Punishment and Responsibility, above n 104, 12.
 Although this is by no means beyond question. It has even been questioned whether hedonistic act utilitarianism adopts the one common virtue: Bentham believed so, but Mill believed that there were different levels of happiness. We believe that the latter is not defensible, since there is no way to rank different pleasures. See Mill, above n 22, 259–64; Bentham, Principles of Morals and Legislation, above n 11, 38–40, 42–5.
 This obviously rejects the Rawlsian ideal that a normative theory must be consistent with our considered moral judgments: see John Rawls, ‘Outline of a Decision Procedure in Ethics’ (1951) 60 Philosophical Review 117. Rather, this approach proposes a foundationalist concept of morality: but cf Bernard Williams, Ethics and the Limits of Philosophy (1985).
 Their theory is set out in John Braithwaite and Philip Pettit, Not Just Deserts: A Republican Theory of Criminal Justice (1990) 61–9 (‘Not Just Deserts’).
 They expand on the concept of dominion in Philip Pettit and John Braithwaite, ‘Not Just Deserts, Even in Sentencing’ (1993) 4 Current Issues in Criminal Justice 225.
 This is discussed at length below in Part VIII.
 For criticism of this and other aspects of their theory, see Andrew von Hirsch and Andrew Ashworth, ‘Not Just Deserts: A Response to Braithwaite and Pettit’ (1992) 12 Oxford Journal of Legal Studies 83.
 Although, as has been correctly pointed out, the reasons they advance for the view that their theory does not permit such an outcome are no more persuasive than those previously advanced by hedonistic utilitarians: see C L Ten, ‘Review Essay: Dominion as the Target of Criminal Justice’ (1991) 10 Criminal Justice Ethics 40, 45. See also Eric Ghosh, ‘Applying Pettit’s Republican Liberty to Criminal Justice and Judicial Decision-Making: The Need for Other Values Including Desert and a Suggestion that They Be Understood Consequentially’ (1999) 22 University of New South Wales Law Review 122, 135–8.
 See Bagaric, ‘In Defence of a Utilitarian Theory of Punishment’, above n 36, 138–42.
 See also Ten, ‘Review Essay’, above n 114, 44–5.
 Braithwaite and Pettit, Not Just Deserts, above n 110, 70, 173.
 Bentham, Principles of Morals and Legislation, above n 11, 283; see also Ten, ‘Review Essay’, above n 114, 44–5.
 Ronald Dworkin notes that judges cannot decide cases on policy grounds because they are not elected, and that to invoke policy grounds is to apply retroactive law, unlike a principled decision which merely applies existing rights and duties: Ronald Dworkin, Taking Rights Seriously (1977) 22.
 (1994) 71 A Crim R 459, 468 (Hunt CJ at CL) (footnote omitted). See also R v Fisher (1989) 40 A Crim R 442, 445 (Yeldham J); R v Lupoi (1984) 15 A Crim R 183, 190 (White J); R v Williscroft  VicRp 27;  VR 292, 300 (Adam and Crockett JJ), cf 303 (Starke J); R v Nichols (1991) 57 A Crim R 391, 395 (Lee AJ).
 R v Wheatley (1982) 4 Cr App R(S) 371, 374 (Kilner Brown J) (emphasis added). In England and Canada retribution has been accepted as an important sentencing consideration, but it is not clear what form of retribution has been adopted. See also R v Lyons  INSC 47;  2 SCR 309, 329; 44 DLR (4th) 193, 214 (Dickson CJ, Estey, McIntyre, Le Dain and La Forest JJ); R v Goltz  3 SCR 485; 67 CCC (3rd) 481.
 R v Collins (Unreported, Supreme Court of Victoria, Court of Appeal, Young CJ, Kaye and Tadgell JJ, 26 November 1984) 4 (Tadgell J).
 (1974) 60 Cr App R 74.
 See the discussion below in Part VII.
 R v Sargeant (1974) 60 Cr App R 74, 77 (Lawton LJ). See also Channon v The Queen (1978) 33 FLR 433, 438–9, where Brennan J stated that ‘lex talionis may be an efficient law for protecting society from criminal recidivists, but it has no place in the administration of contemporary criminal justice’. See also R v Williscroft  VicRp 27;  VR 292, 300 where Adam and Crockett JJ stated that ‘retribution in the modern sense cannot be equated with the concept on which the lex talionis rested’.
  1 CCC 27, 45 (MacKay JA).
 Honderich, above n 4, 29. The satisfaction theory of retibutivism is persuasively criticised by Ten, Crime, Guilt and Punishment, above n 4, 51–2.
 R v Darby (1986) 8 Cr App R (S) 487, 490 (Lane CJ). See also R v Amituanai  QCA 80; (1995) 78 A Crim R 588, 596 (Thomas and White JJ); A-G (Vic) v Woolnough (Unreported, Supreme Court of Victoria, Court of Appeal, Starke ACJ, Crockett and McGarvie JJ, 4 June 1981).
 Channon v The Queen (1978) 33 FLR 433, 437 (Brennan J). See also R v Valentini  FCA 133; (1980) 48 FLR 416, 420 (Bowen CJ, Muirhead and Evatt JJ), where the court equated retribution with denunciation; R v McKenna (1992) 63 A Crim R 452, 461 (Seaman J); R v McGlynn  Qd R 526, 530 (Dunn J).
 Channon v The Queen (1978) 33 FLR 433.
 R v Dixon (1975) 22 ACTR 13.
 R v Williscroft  VicRp 27;  VR 292.
 (1991) 57 A Crim R 391, 395. See also R v Valentini  FCA 133; (1980) 48 FLR 416, 420 (Bowen CJ, Muirhead and Evatt JJ); R v McKenna (1992) 63 A Crim R 452, 461 (Seaman J); R v Prindable (1979) 23 ALR 665, 669 (Brennan, Gallop and Lockhart JJ).
 R v Hitchcock (1982) 4 Cr App R (S) 160, 161 (Lane CJ).
 R v Hay (1982) 4 Cr App R (S) 392, 394 (Lane CJ); see also R v Milne (1982) 4 Cr App R (S) 397. In Canada a similar view is expressed in R v Clayton (1983) 69 CCC (2nd) 81, 82 (Cory JA).
 R v Dole  VicRp 75;  VR 754, 768 (McInerey J). See also R v Causby  TASRp 9;  Tas R 54, 59–60 (Green CJ).
 Inkson v The Queen  TASSC 13; (1996) 6 Tas R 1, 16 (Underwood J).
 R v H (1980) A Crim R 53, 65 (Moffit P). Cf R v Miller  VicRp 60;  2 VR 348; Smith v Luker  NTSC 95; (1992) 111 FLR 99.
 R v McKenna (Unreported, New South Wales Court of Criminal Appeal, Lee AJ, 16 October 1992) 9. See also Walden v Hensler  HCA 54; (1987) 163 CLR 561.
 R v Williscroft  VicRp 27;  VR 292, 300 (Adam and Crocket JJ).
 R v Dixon (1975) 22 ACTR 13, 19 (Fox J).
 R v Collins (Unreported, Supreme Court of Victoria, Court of Appeal, Young CJ, Kaye and Tadgell JJ, 26 November 1984) 4.
 Australian Law Reform Commission, Sentencing, Report No 44 (1988) 88. The report does, however, attempt to distinguish retribution from a just deserts theory.
 New South Wales Law Reform Commission, above n 105, 46.
 Canadian Sentencing Commission, Sentencing Reform: A Canadian Approach (1987).
 While this is regrettably opinionated, apologies are tendered to adherents to those retributive theories which are not considered. For an excellent overview of many other retributive theories, see Ten, Crime, Guilt and Punishment, above n 4, 38–65; Honderich, above n 4, 211.
 Honderich, above n 4, 211.
 There is much uncertainty regarding the meaning of an intrinsic good. Perhaps the easiest way of understanding it is to contrast it with an instrumental good. An instrumental good is one which is a step towards the attainment of something we regard as good, whereas an intrinsic good is a good in itself, not because of the results it may lead to: see Robert Nozick, Philosophical Explanations (1981) 414; Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 521.
 Kleinig, above n 13. See also Hawkin, above n 13, 13, where he asserts that at the pre-reflective level it seems to be assumed that a guilty act deserves punishment.
 Packer, above n 56, 184. The intuitive appeal in the view that the guilty deserve to suffer is also discussed by Moore: Michael Moore, ‘The Moral Worth of Retribution’ in Ferdinand Schoeman (ed), Responsibility, Character and Emotions: New Essays in Moral Psychology (1987) 179.
 Duff, Trials and Punishments, above n 15, 198.
 Lawrence Davis, ‘They Deserve to Suffer’ (1972) 32 Analysis 136, 139.
 Ten, Crime, Guilt and Punishment, above n 4, 49; Gertrude Ezorsky, ‘The Ethics of Punishment’ in Gertrude Ezorsky (ed), Philosophical Perspectives on Punishment (1972) xxiv.
 Or put in another way, it could be claimed that earlier suffering credits are a matter relevant to mitigation rather than the threshold issue of whether or not punishment is deserved.
 See also Honderich, above n 4, 213–15, where he agrees that the view that the guilty deserve to suffer lacks self-evident appeal. This is in contrast to the view that there is intrinsic badness in suffering, which he claims is far more appealing.
 See also Honderich, above n 4, 212–13.
 See, eg, Nils Christie, Limits of Pain (1981); Thomas Mathiesen, Prison on Trial (1990); Herman Bianchi and Rene Van Swaaningen (eds), Abolitionism: Towards a Non-Regressive Approach to Crime (1986).
 Dolinko, ‘Retributivism, Consequentialism, and the Intrinsic Goodness of Punishment’, above n 5, 521.
 C L Ten, ‘Crime and Punishment’ in Peter Singer (ed), A Companion to Ethics (1991) 366, 369.
 A C Ewing, The Morality of Punishment with Some Suggestions for a General Theory of Ethics (1929) 18.
 Ten, ‘Crime and Punishment’, above n 160, 368.
 In fact, it has been argued that retributivism is a consequentialist theory: see text accompanying above n 97.
 Moral justification is mandatory for the practice of punishment, which involves the deliberate infliction of pain on others.
 See, eg, Ten, Crime, Guilt and Punishment, above n 4, 47. Ten argues that the shared belief that the guilty deserve to suffer may be based on a desire for revenge rather than justice.
 J L Mackie, ‘Morality and the Retributive Emotions’ (1982) 1 Criminal Justice Ethics 3, 3.
 Mackie, Ethics, above n 98, 8. See also Honderich, above n 4, 232.
 Eugen Dühring, Der Werth des Lebens: Eine philosophische Betrachtung (1865) [trans: The Value of Life: A Philosophical Study] ch VIII, cited in Robin Small, ‘Ressentement, Revenge, and Punishment: Origins of the Nietzschean Critique’ (1997) 9 Utilitas 39, 40–1.
 Small, above n 168, 42. Although, it has been noted that surely vengeance must have its limits:
The prisoner too is a member of society and no matter how atrocious and detestable his or her actions may have been, society is bound to limit what otherwise could become unrestrained vengeance amounting to hate, which of itself could threaten the overall well-being and peace of a community.
J F Nagle, ‘Punishment, Parliament and the People’ (1998) 10 Judicial Officers’ Bulletin 17, 17.
 John Elster, ‘Norms of Revenge’ (1990) 100 Ethics 862, 883. See also Small, above n 168, 45.
 Susan Jacoby, Wild Justice (1983).
 Hume’s powerful theory of human motivation distinguishes between two states of mind: beliefs and desires. Beliefs are copies or replicas of the way we believe the world to be. Desires are representations of how the world is to be; they are our wants, the states of mind that move us to act. On their own, beliefs can never provide a source of motivation, they are ‘perfectly inert, and can never either prevent or produce any action’: David Hume, A Treatise of Human Nature (first published 1738, 1978 reprint) 458. It is only our desires that can motivate us. Beliefs are mere replicas of the way we believe the world to be. We can assess beliefs for truth and falsehood — a true belief being one which is a copy of the way the world actually is. In order for an action to occur, we need a desire that prompts us to effect a certain change in the world and a belief informing us how this change can be achieved.
 The reason for this is that moral judgments appear to be inherently action-guiding: upon judging something right or good we are always moved or prompted to act in accordance with that judgment. In the motivational sense we are never totally indifferent to such judgments. This is ‘the practicality of moral judgments’ and is supposedly supported by the fact that it seems odd to assert that one genuinely views something to be right, but has no inclination to do it. Further, it would seem that a failure to act in accordance with a moral judgment always requires some explanation, such as ‘I had a stronger inconsistent need or want’. The theory of ‘internalism’ supports such a picture of morality. Externalism is the opposite view: it claims that moral judgments are not necessarily action-guiding. See also David Brink, Moral Realism and the Foundations of Ethics (1989) 29; Jeffrey Goldsworthy, ‘Externalism, Internalism and Moral Scepticism’ (1992) 70 Australasian Journal of Philosophy 40.
 See, eg, G J Warnock, Contemporary Moral Philosophy (revised ed, 1982) 24–6.
 Small, above n 168, 47.
 Friedrich Nietzsche, Daybreak: Thoughts on the Prejudices of Morality (first published 1881, R J Hollingdale trans, 1997 ed) 85.
 Nozick, above n 149, 366–88.
 Ten, Crime, Guilt and Punishment, above n 4, 43.
 Nozick, above n 149, 366. See also Walker, Why Punish?, above n 1, 4.
 Nozick, above n 149, 388.
 Ibid 367.
 Ten, Crime, Guilt and Punishment, above n 4, 43.
 Nozick, above n 149, 367.
 Ibid. See also Walker, Why Punish?, above n 1, 4, who states that punitive feelings are more disinterested than vengeful feelings.
 See Bagaric, ‘Sentencing: The Road to Nowhere’, above n 3, 597.
 In the system of punishment being considered here this other party is always obviously the state.
 Ten, Crime, Guilt and Punishment, above n 4, 48. It has been proposed that once the retributivist justifies the core claim, there is nothing about the theory which prevents the further assertion that punishment should be dispensed with in the most efficient and organised manner, and in a way which is most likely to avoid possible adverse consequences. To this end, the state is the obvious choice; thus, for practical reasons society has found it convenient to establish its own monopoly on revenge rather than leaving it to individuals: see Dühring, above n 168, 43. It is also contended that society is better at fixing the punishment that fits the crime, given the propensity for private punishment to elicit resentment and reprisals: John Locke, The Second Treatise of Civil Government (first published 1690, 1969 ed) 124–6.
 For example, a high portion of victims of crime do not even report the offence. It is very difficult to obtain accurate data concerning the number of crimes that are unreported to police, but for an overview of some of the relevant surveys, see Ashworth, Sentencing and Criminal Justice, above n 1, 15–17. It must, however, be noted that the reason for not reporting an offence is often not indifference to the outcome of the potential prosecution, but the belief that the police would not be able to do anything.
 Andrew von Hirsch, Censure and Sanctions (1993) 10.
 See also Packer, above n 56, 183, especially 185–6, where in a similar vein he argues that the only two ultimate purposes of punishment are the deserved infliction of suffering on wrongdoers and the prevention of crime.
 Von Hirsch, Past or Future Crimes, above n 1, 48.
 Von Hirsch, ‘Censure and Proportionality’, above n 9, 127.
 Von Hirsch, Past or Future Crimes, above n 1, 53.
 Von Hirsch, Censure and Sanctions, above n 189, 9. In this regard, von Hirsch expressly adopts Strawson’s account: P F Strawson, ‘Freedom and Resentment’ in P F Strawson (ed), Freedom, Resentment and Other Essays (1974) 1.
 Von Hirsch, Censure and Sanctions, above n 189, 10.
 See Ten, Crime, Guilt and Punishment, above n 4, 51. Honderich, above n 4, 43–4, in fact concludes that retributivism comes down to a claim that punishment is justified because it gives satisfaction to victims and to other people, and that criminals freely and responsibly commit offences. See also Narayan, who states that the interests of the victim do not provide a full justification for censure, because this still leaves unanswered the issue of why condemnation must be addressed to the wrongdoer: Uma Narayan, ‘Appropriate Responses and Preventive Benefits: Justifying Censure and Hard Treatment in Legal Punishment’ (1993) 13 Oxford Journal of Legal Studies 166, 171–3. Narayan thinks that censure is justified because criminals have an interest in being informed about the nature of critical moral judgments against them. Censure brings home to the offender his or her status as a wrongdoer in the eyes of society and thus that a moral response by him or her is appropriate.
 See also Ten, Crime, Guilt and Punishment, above n 4, 52.
 Von Hirsch, Censure and Sanctions, above n 189, 10.
 It could be argued that blame is still appropriate for the (already) repentant and the defiant because it reinforces the values of the repentant and, hopefully, wears away at the defiant. However, these considerations are both consequentialist.
 Von Hirsch, Censure and Sanctions, above n 189, 9.
 David Adams, ‘Fitting Punishment to Crime’ (1996) 15 Law and Philosophy 407, 410.
 Von Hirsch, Censure and Sanctions, above n 189, 11.
 Von Hirsch, Past or Future Crimes, above n 1, 57.
 Von Hirsch, ‘Censure and Proportionality’, above n 9.
 Von Hirsch, Censure and Sanctions, above n 189, 10 (emphasis added).
 Ibid 123–4.
 Ibid 126.
 See Tom Tyler, Why People Obey the Law (1990) 107, 175–6. Tyler’s 1984 study entailed surveying around 1500 Chicago residents about their contact with legal authorities. Tyler noted that normative issues are closely linked with compliance with the law.
 Von Hirsch, Censure and Sanctions, above n 189, 9.
 Ibid (emphasis omitted).
 Adams, above n 206, 408.
 Although, as has been noted, this is not a purely retributive virtue: see Bagaric, ‘Suspended Sentences and Preventive Sentences’, above n 11, 560–2.
 On Duff’s account, reparation refers to moral as opposed to material reparation: Antony Duff, ‘Punishment, Citizenship and Responsibility’ in Henry Tam (ed), Punishment, Excuses and Moral Development (1996) 17, 24.
 Ibid 22.
 Duff, Trials and Punishments, above n 15, 242.
 Ibid ch 9.
 Ibid 236.
 Duff, ‘Punishment, Citizenship and Responsibility’, above n 220, 28.
  VicRp 27;  VR 292.
  VicRp 27;  VR 292, 300 (emphasis added). Similarly, in R v Spence (1992) 78 CCC (3rd) 451, 454, the Court stated that denunciation ‘will reflect society’s view of the wrongness of the conduct, and persuade those who might be confused about what is right and wrong’.
 Duff, Trials and Punishments, above n 15, 72, 236, 238. For Duff, moral blame is essentially an attempt to engage with another in the search for a true understanding of the moral character of his or her actions, and the purpose of moral blame, as with punishment, is essentially an attempt to bring the wrongdoer to repentance and reform: at 238.
 Ibid 234.
 Ibid 70.
 Ibid 238.
 Ibid 70–1.
 Duff, ‘Punishment, Citizenship and Responsibility’, above n 220, 23.
 Duff, Trials and Punishments, above n 15, 267.
 Duff, ‘Punishment, Citizenship and Responsibility’, above n 220, 29.
 Duff, Trials and Punishments, above n 15, 292.
 Duff, ‘Punishment, Citizenship and Responsibility’, above n 220, 28–9.
 Ibid 23.
 Duff, Trials and Punishments, above n 15, 295.
 Indeed, more recently, Duff has stated that sanctions, especially non-custodial orders, may operate as penances: Antony Duff, ‘Punishment, Expression and Penance’ in Heike Jung and Heinz Müller-Dietz (eds), Recht und Moral: Beiträge zu einer Standortsbestimmung (1991) [trans: Law and Morality: Towards a Definition], as cited in von Hirsch, Censure and Sanctions, above n 189, 75.
 Duff, Trials and Punishments, above n 15, 238.
 Ibid 11. He does, however, concede that any workable legal system may include elements which may be contrary to that Kantian principle: at 11.
 Ibid 268.
 Ibid 263 (emphasis added).
 Ibid 70. The right to be punished is not a novel suggestion. It was a fundamental feature of the theory of punishment by Morris, who initially argued that the justification of punishment was to restore the proper balance of benefits and burdens which had been disrupted by the crime; but later argued for a paternalistic theory of punishment which relies on a concern for the offender’s moral welfare and for his or her relationship with the community to justify punishment: see Herbert Morris, ‘Persons and Punishment’ in Stanley Grupp (ed), Theories of Punishment (1971) 76; Herbert Morris, ‘A Paternalistic Theory of Punishment’ in Antony Duff and David Garland (eds), A Reader on Punishment (1994) 95. The right to be punished was also advanced by Hegel:
[T]he injury which falls on the criminal is not merely implicitly just — as just, it is eo ipso his implicit will, an embodiment of his freedom, his right; on the contrary, it is also a right established within the criminal himself; ie in his objectively embodied will, in his action. The reason for this is that his action is the action of a rational being and this implies that it is something universal and that by doing it the criminal has laid down a law which he has explicitly recognized in his action and under which in consequence he should be brought as under his right.
Georg Hegel, Hegel’s Philosophy of Right (first published 1821, 1942 ed, T M Knox trans) 100.
 Mirko Bagaric, ‘The Diminishing “Right of Silence”’ (1997) 19 Sydney Law Review 366, 372.
 Duff cannot answer this criticism by falling back on his distinctive concept of punishment, since he does not push for less severe sanctions than those which are presently employed and in fact he does not even rule out capital punishment: Duff, Trials and Punishments, above n 15, 294.
 Anthony Quinton, ‘On Punishment’ (1954) 14 Analysis 133, 136.
 Duff, Trials and Punishments, above n 15, 271–7.
 This is in line with the view adopted by Morris, ‘Persons and Punishment’, above n 245, 76.
 Duff, Trials and Punishments, above n 15, 271.
 Ibid 200–2.
 Ibid 270–1.
 Ibid 271.
 Ibid 275, 277.
 The philosophical problem regarding the weakness of the will is being ignored here due to its peripheral relevance to the discussion.
 As was outlined earlier, if internalism is correct, it is wrong to contend that one can accept that there is a moral reason for doing some act and yet have no inclination to perform the act: see above n 174 and accompanying text.
 See, eg, Bagaric, ‘Incapacitation, Deterrence and Rehabilitation’ above n 36, 29–30, where it is noted that recidivism is particularly high in relation to minor offences.
 Small, above n 168, 57, discussing Nietzsche, Zur Genealogie der Moral (1887) [trans: On the Genealogy of Morals]. The failure of punishment to reform also presents a problem to the utilitarian, since this is one of the traditional purposes with which some utilitarians have sought to justify punishment. However, this is not fatal to utilitarianism since there are many other purposes on which they can rely. Unlike Duff, the utilitarian can look beyond reformation and assert that, despite the overall harmful impact of punishment on the offender, it is justified on the basis of its other positive effects such as incapacitation and general deterrence.
 In circumstances where an agent does not have the capacity for rational thought and judgment, we often justify things we do to them on the basis of whether a rational person in the agent’s position would want those things, as opposed to what the agent actually wants. This cannot be so in the case of punishment since, as Duff acknowledges, punishment presupposes responsibility, which requires rationality and the capacity for rational thought: Duff, Trials and Punishments, above n 15, 224.
 Also, as Ten notes, externally imposed punishment will only achieve Duff’s justificatory aims if it is ultimately accepted by the offender as necessary for these purposes. However, Duff does not even attempt to provide empirical support for this: see C L Ten, ‘Positive Retributivism’ (1990) 7 Social Philosophy and Policy 194, 201. Indeed, there is no evidence to support the possibility of such internal education. There is much stronger evidence for the position of the utilitarian, who punishes in the hope that it will deter: Walker, Why Punish?, above n 1, 79.
 Even if punishment could alter moral views, Duff’s view has been criticised by von Hirsch on the basis that citizens have not entrusted their moral development to the state; hence, the state is not entitled to use its coercive powers to seek to induce moral sentiments: see von Hirsch, Censure and Sanctions, above n 189, 73.
 Duff, Trials and Punishments, above n 15, 273.
 This is similar to the distributive principle devised by Farrell, who states that ‘one must suffer if one’s decision to do wrong makes it necessary that someone must suffer and that sufferer must either be the wrongdoer or some innocent victim’: Daniel Farrell, ‘The Justification of General Deterrence’ (1985) 94 Philosophical Review 367, 373.
 This tension is not necessarily irreconcilable: Duff could contend that in a case such as this where one must choose between preserving the autonomy of the community or that of the criminal, the rights of the innocent should prevail. However, Duff has no way of sorting out how such conflicts should be resolved.
 See also Packer, above n 56, 188.
 Duff, Trials and Punishments, above n 15, 241–2.
 Ibid 241.
 See Ten, ‘Positive Retributivism’, above n 261, 200.
 See also Narayan, above n 198, 176–7.
 See Ten, ‘Positive Retributivism’, above n 261, 202.
 Duff, ‘Punishment, Citizenship and Responsibility’, above n 220, 24.
 See also Narayan, above n 198, 166–7.
 Duff, Trials and Punishments, above n 15, 266.
 Ibid 271.
 Narayan, above n 198, 177.
 The point is emphasised by Narayan, who states that if an offender’s bonds with society are broken by his or her offending, such bonds can only be restored by sincere repentance. Compulsory hard treatment is a poor vehicle for this, given that it is impossible to distinguish genuine repentance from that which is motivated purely by expedience: ibid.
 Duff, Trials and Punishments, above n 15, 255.
 Ibid 256.
 Ibid 260–1.
 Walker, Why Punish?, above n 1, 79. See also Narayan, above n 198, 177. Narayan makes a similar point: given that penance is compulsory, it is always precipitated by introspection, conscience and guilt, whereas punishment is not.
 Duff, ‘Punishment, Citizenship and Responsibility’, above n 220, 23.
 Ibid 24.
 Husak points out that a strong justification for hard treatment is necessary, given the unpleasantness inherent in hard treatment and other associated problems with it, such as the expense involved, inevitable errors in deciding who should be subject to it and difficulties in determining how much hard treatment is appropriate. This is something that Duff’s account does not offer: see Husak, above n 81, 180.
 Duff and von Hirsch, ‘Responsibility, Retribution and the “Voluntary”’, above n 6, 112. They also assert that it relies on unproven empirical claims: at 112.
 See also, Narayan, above n 198, 180, who states that the role that hard treatment (which according to her is only justifiable where it comports with a censuring framework) plays in preventing crime seems to explain why it is morally permissible despite the hardships it inflicts.
 See, eg, Morris, ‘Persons and Punishment’, above n 16; Murphy, Retribution, Justice and Therapy, above n 16, 82–115; Sadurski, above n 16, ch 8; Finnis, above n 16, 263–4.
 Morris and Murphy have both subsequently moved away from this theory: see Morris, ‘A Paternalistic Theory of Punishment’, above n 245; Jeffrie Murphy, ‘Retributivism, Moral Education and the Liberal State’ (1985) 4(1) Criminal Justice Ethics 3. Von Hirsch also previously endorsed this theory as a partial justification for punishment, but has resiled from it because it is vulnerable to unjust society objections and fails to provide guidance on how much punishment is deserved: von Hirsch, Past or Future Crimes, above n 1, ch 5; von Hirsch, Censure and Sanctions, above n 189, 7–8; von Hirsch, ‘The Politics of “Just Deserts”’, above n 9, 408.
 For a good overview of many of the significant criticisms of this theory, see Ten, Crime, Guilt and Punishment, above n 4, 52–65.
 Herbert Morris, On Guilt and Innocence (1976) 33.
 Richard Burgh, ‘Do the Guilty Deserve Punishment?’ (1982) 79 Journal of Philosophy 193, 209.
 For example, the wrongfulness of rape consists of taking unfair advantage of those who obey the law, rather than the offender’s obtaining a material or tangible benefit: Duff and Garland, ‘Introduction: Thinking about Punishment’, above n 1, 44.
 Walker also raises the problem of the theory in dealing with attempted offences. Here there is no benefit achieved, yet the impulse to punish still persists: Walker, Why Punish?, above n 1, 76.
 George Sher, Approximate Justice, Studies in Non-Ideal Theory (1997) 166–7. Davis also attempts to redefine the nature of the unfair advantage: he believes that the offender’s advantage is the price which would be paid for the behaviour: Michael Davis, To Make the Punishment Fit the Crime: Essays in the Theory of Criminal Justice (1983). For criticism of this, see von Hirsch, Censure and Sanctions, above n 189, 8–9.
 Sher, Approximate Justice, above n 296, 167.
 George Sher, Desert (1987) 82.
 The circumstances in which a legal duty is imposed are discussed in R v Russell  ArgusLawRp 98;  VLR 59; R v Instan (1893) 17 Cox CC 602; R v Nicholls (1874) 13 Cox CC 75; R v Gibbins and Proctor (1918) 13 Cr App R 134. Further, if punishment is reserved only for breaches of moral rules, the role of punishment would be unacceptably attenuated. Given the uncertain moral status of many acts it is unlikely that a sufficiently clear well-developed set of legal proscriptions could be constructed.
 Sher, Approximate Justice, above n 296, 166.
 Ten, ‘Crime and Punishment’, above n 160, 369.
 Jeffrie Murphy, ‘Marxism and Retribution’ in Antony Duff and David Garland (eds), A Reader on Punishment (1994) 44.
 Sher, Approximate Justice, above n 296, 175–6.
 Ten, ‘Positive Retributivism’, above n 261, 198.
 Sher, Approximate Justice, above n 296, 176–7.
 Ibid 176 (emphasis added).
 A judgment has universal application if the acceptance of it in a particular situation entails that one is logically committed to accepting the same judgment in all other situations, unless there is a relevant difference. Thus, to state that moral judgments are universal entails that whenever one judges a certain action or thing (situation) as having a particular moral status, then one is logically committed to the same judgment about any relevantly similar action or situation. It has been suggested that numerical differences are irrelevant. This refers to specific descriptions of the person, relation or situation. Thus, the fact that the judgment relates to a particular person (such as John Smith), place (Melbourne), or relation (John’s mother) is irrelevant. Also irrelevant are generic differences, such as tastes, preferences and desires: see Mackie, Ethics, above n 98, 83–102.
 See also Anderson, above n 7, 13, who argues that the benefit–burden theory of punishment (which he calls reciprocity retributivism) cannot justify punishment in a world where there is an unequal distribution of resources and opportunities.
 Sher, Approximate Justice, above n 296, 168.
 Ibid 179.
 Ten, ‘Crime and Punishment’, above n 160, 369.