OUP UNCORRECTED PROOF

CHAPTER 2
Is Twenty-first Century Punishment
Post-desert?
M ATT M ATR AVER S

T here are grave dangers in looking forward and making predictions about the way in
which political theory and practice may develop. The obvious danger is simply that
one might be mistaken. Less obvious, but still damaging, is to give in to the temptation
to see the past and present as punctuated by decisive periodic breaks with a present fis-
sure heralding a new era. The world is a complicated place and things are seldom predict-
able or describable in neat periods. Nevertheless, these dangers should not mean that we
ignore broad shifts in emphasis or that we should be willfully blind to the evidence that
things have changed, or are changing. Theories and policies do change, and the capacity
of theory to respond to practice often depends on its ability to sense those changes and
to think them through in advance of the policies getting too tight a grip on the way we
act. With this in mind, there is good reason to consider the past and future of penal
theory and practice.
Looking back, there is a now well-established story in penal philosophy that has it
that the broadly consequentialist consensus of the postwar period was overturned in the
1970s by a retributivist revival. Although, as Michael Tonry makes clear in his introduc-
tory essay (Tonry 2011a), penal practice was much more complicated, the story does
have some plausibility when applied to the theoretical literature. In 1969 a survey of
justifications of punishment found that “there are no defenders [of traditional retributive
theory] writing in the usual places” (Honderich 1969, p. 148). Ten years later one would
have been able to say exactly the same about defenders of traditional consequentialism,
while retributivists—in no fewer than nine varieties (Cottingham 1979)—were com-
monplace. However, the decline in consequentialism and the revival of retributivism
does not map easily on to a story about desert. To see this, it is necessary to unpack the
conventional story.

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 30 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

I. DESERT

There is little doubt that the retributive revival in penal theory was accompanied by, and
could be thought to be part of, a general anticonsequentialism in the philosophical liter-
ature and a resurgence in the politics of desert. The decline of consequentialist moral
thinking can be overemphasized, but the publication of John Rawls’s A Theory of Justice
(1971), which provided a systematic alternative to utilitarianism, was undoubtedly the
dominant philosophical event of the period. Politically, in the United Kingdom and
United States, the end of the decade brought electoral success to right-of-center parties
led by Margaret Thatcher and Ronald Reagan. These leaders emphasized individual de-
sert and responsibility, and their success can be measured in part by the fact that even
when the political winds changed, their left-of-center successors went to extraordinary
lengths to include these notions packaged as a “third way” that synthesized “rights and
responsibilities” and insisted, for example, that welfare was “a hand up, not a hand out”
(Matravers 2007, pp. 5–11).
Thus there seems to be a neat, almost overdetermined, story that leads from a con-
sequentialist, welfarist heyday, dominant throughout the century until the 1970s, to a
revolution that encompassed penal theory (Kleinig 1973; von Hirsch and Committee
for the Study of Incarceration 1976); penal policy; legal, political, and moral theory
(Hart 1968; Rawls 1971); and political practice. There is indeed much to this story,
and many aspects of it are masterfully described in David Garland’s The Culture of
Control (2001). However, it is a mistake to think that there is a single story in which
consequentialism’s decline was accompanied, or caused, by a resurgence of the notion
of desert.
The mistake is an easy one to make—particularly for penal theorists—since, as we
have seen, over a relatively short period consequentialist penal theories (and some prac-
tices) declined, as did consequentialist theorizing more broadly (in the face of the
Rawls-led neo-Kantian revival), retributive theories increased, and the rhetoric of desert
became critical in political practice. Yet, this is not one story. To see this, consider just
how odd it would be to claim that desert is critical to the criticisms of consequentialism
and to the alternative neo-Kantian theory offered in Rawls’s A Theory of Justice.
Rawls explicitly denies that the notion of moral desert has any part to play in a theory
of distributive justice. “The principles of justice that regulate the basic structure,” he
writes, “do not mention moral desert, and there is no tendency for distributive shares to
correspond to it” (Rawls 1971, p. 311). Given this, it is clear that Rawlsian theory should
be included only in the “decline of consequentialism” part of the story and not in the
narrative of the rise of desert. However, my claim is that this is true of the majority of
retributive penal theories as well, and that these theories can, at best, only underwrite
something that looks much more akin to Rawls than to traditional desert-based retribu-
tivisms. To see this, consider first Rawls’s approach. Rawls’s aversion to desert is well
known, although I think sometimes misunderstood. One account has it that Rawls argues
that we are all equally nondeserving (or, for that matter, deserving) because by declaring
“morally arbitrary” everything that might differentiate us one from another, Rawls leaves
nothing—natural talent, willingness to make an effort, social status, etc.—that could play

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 31]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 31 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

the role of a desert basis and so legitimate anything other than equality of outcome (at
least initially). A better account, I think, has it that Rawls thinks desert irrelevant to
distributive justice. It is rejected as the foundation for justice because there is no sensible
way of conceiving of a relevant, legitimate desert basis and then translating that into dis-
tributive shares. In cruder terms, one reading has it that no features of human beings can
be attributed to them in a way that would legitimate treating one such being different
from any other (we are not responsible for all those things—our heights, talents, intelli-
gence, etc.—that enable us to achieve different things, so we do not deserve any differen-
tial reward or penalty for those achievements). The other, more plausible reading is that
whether or not we are responsible for our talents, etc., there is no legitimate way of trans-
lating natural differences into distributive outcomes. Either way, the conclusion is, as
Rawls puts it, that the common-sense tendency “to suppose that income and wealth, and
the good things in life generally, should be distributed according to moral desert” is
rejected (1971, p. 310).1
So the idea of a resurgent “desert theory” in theories of distributive justice—indeed,
I think, in moral theory generally—is not sustainable. The Kantianism that forced out,
and took the place of, the dominant consequentialist paradigm is one without Kant’s
metaphysics and one that has no place for a strong notion of desert. My claim, consid-
ered in the next section, is that the same is true of (most of) the penal theories that dis-
placed their consequentialist counterparts.

II. WAS IT EVER ABOUT “JUST DESERTS”?

Tim Scanlon characterizes what he calls “the Desert Thesis” as follows: “the idea that
when a person has done something that is morally wrong it is morally better that he or
she should suffer some loss in consequence” (1998, p. 274).2 Narrowed to the field of
punishment, I take it that the relevant thesis is that a person who has committed a (legit-
imate) criminal wrong deserves to suffer some loss, and it is the function of the system
of punishment to impose that loss for the wrong done. That is, the—or, at least, a—
function of the system of punishment is to ensure that the suffering that is (prejusti-
cially) deserved by a given offender for a given act is imposed on the offender.
Once desert is characterized in this way, it is not at all clear that there are many gen-
uine desert theorists among those who would identify themselves, or be identified by
others, as such. Of course, retributivists come in a variety of forms and the role of desert
may be subtly different in each. However, our interest here is not in narrow differences
between retributive arguments, but is rather in the place of desert in the overall ap-
proach. Thus it is possible to restrict the analysis to fewer, broader, and more abstract
forms of the argument.3 The retributive accounts briefly considered below are Michael
Moore’s intuitionist theory and so-called fair play theories. After that, the essay takes a
longer look at communicative accounts and at mixed theories. The conclusion is that
where desert has an independent and important role in the argument, the retributive
theory is either implausible (Moore) or incomplete (communicative theories). In cases
where the retributive theory fares better, desert is not central.

[ 32 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 32 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

A. Michael Moore’s Intuitionist Account

According to Moore, the retributive theory that he defends “is the view that we ought
to punish offenders because, and only because, they deserve to be punished. Punish-
ment is justified, for a retributivist, solely by the fact that those receiving it deserve it”
(Moore 1987, 1993, p. 15; see also Moore 1997, chap. 2–3). Clearly this is a desert-
based view. Since my main concern in this essay is not to evaluate retributive the-
ories, but to establish that the retributive revival was not primarily a revival in desert
thinking, Moore stands as a counterexample. However, for all the sophistication of
his account, Moore’s theory has not established itself in the mainstream. This is
because it depends on a combination of a very demanding, if idiosyncratic, moral
realism, and a thesis that our moral intuitions offer a good guide to the moral truth
that offenders deserve to suffer (for Moore’s moral realism, see Moore 1982, 1992). For
reasons given elsewhere, I find this account implausible (Matravers 2000, pp. 81–86),
but whether it is or not, it has played only a very minor role in the revival of retri-
butive punishment theory (perhaps because of the metaphysical theory on which
it relies).

B. Fair Play Theory

Fair play—or benefit and burden—theories enjoyed a brief period of popularity in the
retributive revival of the 1960s and 1970s. The core of the argument is that, given a just
initial distribution of benefits and burdens in society, a criminal offense disturbs this
equilibrium and needs to be rectified. It does so because the criminal free rides on the
willingness of others to constrain the pursuit of their interests in accordance with the
law (Morris 1968; Murphy 1973). Again, the purpose here is not to consider the merits
or otherwise of the account. That being said, taken as a complete account of punish-
ment, few have found it compelling, and its critics include some of its early propo-
nents. The problem, as Tonry puts it, is that “gaining an unfair advantage” by free riding
is not “an adequate or even plausible characterization of the wrongfulness of many of-
fenses” (for criticisms of fair play theory, see von Hirsch 1985, 1990; Duff 1986, 2001;
Dolinko 1991; Matravers 2000; Tonry 2011b, p. 109; for a defense, see Dagger 1993).
Putting that to one side, what of desert? Of course, in some sense the free rider
“deserves” punishment. However, the kind of desert being invoked here is not the
prejusticial desert of the desert thesis. There is not some appropriate level of suffering
deserved by the offender that it is the job of the system of punishment to ensure that
he gets. Rather, what the offender deserves is whatever loss (or suffering) is dictated
by the system of justice that will restore the balance of benefits and burdens. Desert
here is determined by the overall account of the balance of benefits and burdens; in
Rawlsian terms (further discussed below), the offender and the wider society of
which he has taken advantage have legitimate expectations, not desert claims, that
need to be met.

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 33]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 33 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

C. Punishment as Communication 4

Perhaps the most important and long-lasting of the retributive theories that emerged in
the last third of the twentieth century was the communicative account of punishment. In
its most sophisticated form, punishment aims, and is justified by the need, to convey
censure. Moral wrongdoing deserves censure, and where a society has declared some
behavior to be wrong, then censure is “owed” to the offender as “an honest response to
his crime,” to his victims “as an expression of concern for their wronged status,” and to
“the whole society, whose values the law claims to embody” (Duff 1998, p. 50).
There is a clear desert claim here: moral wrongdoing deserves censure (and its legal
extension is that, following criminal wrongdoing, it is the job of punishment to inflict the
deserved censure on the offender). Moreover—and perhaps one reason for the account’s
attractiveness and longevity—this claim does not seem to rely on any odd metaphysics
or other mysterious ingredient. As Duff puts it, “whatever puzzles there might be about
the general idea that crimes ‘deserve’ punishment . . . there is surely nothing puzzling
about the idea that wrongdoing deserves censure” (1998, p. 50). However, this is not the
desert thesis, which is about the deserved nature of an imposed loss. In short, the desert
thesis as applied to punishment needs to accommodate deserved hard treatment. As
Duff notes, while censure can be conveyed by hard treatment, it need not be. Thus the
censure theory faces the “familiar task . . . to explain and justify the role of hard treat-
ment” (Duff 1998, p. 51).
Although Duff believes that hard treatment can be intrinsically linked to censure, few
others are persuaded. Of course, being censured might itself be unpleasant, but there is
no reason to believe that it has to be so. Thus some additional argument is needed.
According to Duff, hard treatment is intrinsic to the account because censure needs to be
forcefully expressed; because sometimes words are insufficient to express remorse or
repentance; and because the offender needs to undertake some form of suffering to show
to her community that she is serious, and thus to achieve reconciliation. Discussion of
these claims would take us beyond the purpose of the argument here (for criticisms, see
Matravers 2011a). In relation to the argument here, the point is that, while censure may
be deserved, hard treatment (at best) merely provides the vehicle of transmission for the
censure and the offender’s response to that censure. More plausibly, censure is deserved,
but hard treatment must find some other justification. This—that censure is only one
element of a complete account of punishment—is the argument that has had the most
purchase in the literature, which means that unless the theoretical resources upon which
a revised account draws are also desert based, then the censure theory cannot be said to
underpin a revival in the centrality of desert-based theorizing.

D. Mixed Theories

The mixed theory on which I want to concentrate is one that does indeed aim to supple-
ment the censure-based account with an independent justification of hard treatment.
However, it is worth adding a brief word about H. L. A. Hart’s account, both because it

[ 34 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 34 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

remains influential and because Hart is sometimes cited in the orthodox story of the
decline of consequentialism and the rise of desert.
Hart’s work certainly fits the period. Punishment and Responsibility was published in
1968 and offers a broadly liberal account of the subject matter (on occasions in explicit
opposition to the consequentialism of the English penal theorist Barbara Wootton) (see
Hart 1968, chap. 7, 8). Yet, of course, Hart believed the overall purpose—the general
justifying aim—of punishment to be consequentialist (Hart 1968, chap. 1). And while it
is true that once the system is in place its operation is limited by desert side constraints,
these are not expressions of the desert thesis.5 Rather, these side constraints capture a
liberal model of the proper relation between the state and its citizens. As a citizen, it
should be up to me whether I put myself in the realm of punishment, and having done so,
I should be treated as a “person” and not as something “alterable, predictable, curable or
manipulable” (Hart 1968, p. 183). Hart, then, cannot be invoked in defense of the desert
thesis, although his work (like Rawls’s) was undoubtedly important in setting the tone of
the post-consequentialist era.
Hart’s is probably the most famous mixed theory among justifications of punish-
ment. However, like most philosophical theories, its claim to have directly influenced
policy is at best moot. That is not true of the mixed theory on which I want to focus in
the rest of this section. For more than a quarter of a century, Andrew von Hirsch has
championed “proportionality” in sentencing with considerable success in terms of both
theory and practice. A bumper sticker for the account is that “the punishment must fit
the crime,” which can be understood in at least two ways. Confusion between these is
probably the single most likely source of what I argue is the conflation of the rise of
retributivism with the rise of desert.
In one interpretation, there is a desert thesis account of the idea that the punishment
must fit the crime. This is that there is some preestablished quantum of suffering that is
appropriate, that “fits” the crime, and that it is the job of the system of punishment to
inflict on the offender. However, that is not what is meant by proportionality as champi-
oned by von Hirsch (and others). Proportionality in sentencing is primarily a matter of
the relations within a scheme of penalties, not of the anchoring of that scheme. It requires
two things. Ordinal proportionality requires that similarly culpable “persons convicted
of crimes of comparable gravity should receive punishments of comparable severity.”
Cardinal proportionality concerns “the overall magnitude and anchoring points of a
penalty scale” (von Hirsch 1990).
It is the proportionality interpretation of the punishment fitting the crime that has
been championed by retributivists like von Hirsch and Andrew Ashworth, so the ques-
tion arises whether proportionality is an expression of the desert thesis. The requirement
of ordinal proportionality has nothing to do with the desert thesis. It merely requires that
if offender A commits an offense O with no mitigating or aggravating circumstances and
receives a punishment of severity P, then offender B, who commits a similar offense in
similar circumstances should also receive a punishment of severity P. Similarly, offender
C, who culpably commits an offense that is twice as serious as that committed by of-
fenders A and B, should receive a punishment that is twice as severe as that handed out
to A and B. Note how easy it is to describe this in terms of “desert.” If, once the scheme is
established, A and B are punished by P, it makes sense to say that C deserves 2P. But the

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 35]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 35 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

desert claim here is relative to the scheme, not to some ideal of suffering that needs to be
imposed in response to the offense. It is, in Rawlsian terms, an “entitlement.” C is “enti-
tled” or has a “legitimate expectation” to receive 2P, given the scheme and the treatment
of A and B. C’s punishment “fits the crime” in accordance with the scheme, as, of course,
does A’s and B’s, and again, according to the scheme, these punishments are what is
deserved, but none of this has the slightest thing to do with the desert thesis.
The overall magnitude and anchoring points of the scale, its cardinal status, could of
course be fixed by considerations drawn from the desert thesis. If it is given—or can be
intuited as a moral fact in Moore’s sense—that the suffering appropriate for the least se-
rious crime on the scale is a nominal monetary fine and that appropriate for the most
serious is, say, death, then the scale would be most appropriately constructed in accor-
dance with these requirements (ordinal proportionality would only be required if the
moral truth about deserved suffering turned out to meet that condition). However, this
is not the position that von Hirsch takes in any of his writing.
Rather, von Hirsch has contemplated a number of ways of fixing the penalty scale,
derived from asking questions such as “what is available?” and “what is conventional?”
(see, e.g., von Hirsch 1985, p. 159). More recently, in particular in his work with Andrew
Ashworth (von Hirsch and Ashworth 2005), he has pursued a mixed theory in which
censure plays a leading retributive role and hard treatment acts as a prudential supple-
ment aimed to aid citizens—who are neither fully saints nor fully sinners—in their
resisting the temptation to commit crime. In this case, then, one might say that censure
is deserved (in response to some offense) and it is the job of the criminal justice system
to deliver that censure. Hard treatment is deserved, too, but only in an “entitlement”
sense. The offender may expect hard treatment both as a means of expressing censure
and to deter him and others, but the degree of hard treatment is dependent on a scheme
of penalties that has the reduction of future crime at its core.

E. Retributivism

In short, with the notable exception of Michael Moore, the mainstream revival in retrib-
utivism since the 1970s has not been a revival in the desert thesis. The slogan “the pun-
ishment must fit the crime” is part of contemporary retributivism, but its association
with traditional notions of desert is inappropriate. Retributivists, of course, may wel-
come this conclusion. They may do so for two reasons. First, the traditional desert thesis
is defensible only by invoking some pretty robust metaphysical commitments (such as
can be found in Kant, Hegel, and Moore), and such commitments are not only out of
fashion philosophically, but are widely regarded by liberals as an inappropriate basis on
which to ground public policy in pluralistic societies (see Rawls 1999; famously, John
Rawls described his theory as “Political not Metaphysical”). Second, it is often taken to
be a cruel irony that what began for many as a liberal, left-of-center call for fairness in
sentencing and an end to arbitrary punishments was hijacked by right-of-center politi-
cians pushing for harsher punishments. If what I have said is correct, then there was no
such capture of the “just deserts” movement, because that movement was only about
proportional justice, not about desert. Reagan, Thatcher, and the theorists who followed

[ 36 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 36 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

them corrupted, rather than hijacked, the retributive revival. One example will suffice:
California legislators may believe that those who commit three felony offenses deserve
indefinite detention under the “three-strikes” law, but this clearly has nothing to do with
retributivism understood as proportionality.
It may be thought that the argument has reached a dead end. This essay has relied
on a particular, traditional, understanding of deserved punishment—the desert
thesis—that, it turns out, few other theorists or theories share, and few would wish to
share. Retributivists are broadly about the communication of censure and the pro-
portional use of hard treatment; but that is what they said they were about, so what
gains are made by pointing it out (other than clarifying why politically motivated
rhetoric about desert has little to do with retributivism)? The gains, I think, lie in
clarifying what needs to be done when confronted by recent and not so recent devel-
opments in penal practice. By that I do not mean that we are able to respond to the
corrupt version of retributivism that has given us severe mandatory sentences for
many crimes and three-strikes legislation, but also that if we are to respond to the
challenge of thinking about recent therapeutic or restorative practices, we need to
know what the issue is. If the above argument is correct, then the issue is not one of
reconciling those practices to desert, or conceptualizing a post-desert world, but
rather, or so I will argue below, it is one of thinking about the requirements of liberal
justice as a whole.

III. PROPORTIONALITY

For most retributivists, then, the justification of systems of penal hard treatment is not
that they exist in order to give an earthly form to some kind of “celestial mechanics” in
which wrong actions deserve “an equal and opposite reaction” in the form of imposed
suffering (Cohen 1939, p. 279). Rather, they are a mechanism of social order needed
because living together on shared territory in conditions of moderate scarcity is difficult
and throws up all manner of coordination problems.6 However, not just any means to
social order are acceptable. For retributivists (at least for those considered for the rest of
this essay), the principle of proportionality stands independently and thus dictates at
least a significant part of the system of punishment.
It is important to be clear: proportionality does not provide the ultimate rationale for
having a system of punishment (as against not having one). It is not that there is some
proportionate suffering that must be imposed on wrongdoers so that a system of punish-
ment is required to fulfill this demand. It is that in designing or critiquing a system of
punishment—one that is to be or has been created for some other reason—the demands
of proportionality must be respected. Although not the desert thesis, this would still be
a substantive demand, and one that could underpin criticism of much recent penal prac-
tice (disproportionate sentencing, therapeutic justice, etc.).
Although I have a great deal of sympathy with the proportionality thesis, for reasons
given in the next section, I believe it should not be used too quickly. It is not at all ob-
vious that proportionality is an independent principle that should automatically be
deployed as a trump to defeat other approaches to crime management and reduction.

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 37]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 37 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

A. Proportionality and Fairness

The attractions of proportionality to those who consider fairness a value, and to liberals
more generally, ought to be clear. Once the system of justice is in place, people are enti-
tled to certain things, to be treated in certain ways. Penal hard treatment must respect
those entitlements. Proportional sentencing—the claim that “the penal sanction should
fairly reflect the . . . harmfulness and culpability of the actor’s conduct” (von Hirsch and
Ashworth 2005, p. 4)—treats people fairly both in the narrow sense of treating like cases
alike (and unlike cases differently) and, its proponents claim, in the sense of treating
people as agents who are entitled to a certain kind of respect.
The first of those claims looks to be uncontroversial: similarly harmful and cul-
pable offenders will receive similar punishments and those whose harmfulness or cul-
pability is different will receive different punishments. The second is not quite so
apparent, but rests on the belief that citizens are entitled to a certain form of equal
respect. This not only means that they are entitled to be treated alike (when relevantly
alike), but also that they should not, in Hart’s words (1968, p. 183), be treated as if
“alterable, predictable, curable or manipulable.” In short, the state should appeal to
our capacities as reasoning agents, and not merely threaten or manipulate. Thus pro-
ponents of proportionality claim the anchoring points of the scale of penalties have to
be such as to respect citizens as persons. To threaten citizens with death for a minor
traffic offense might reduce violations of traffic laws, but it would hardly be to treat
citizen drivers as agents.
Although attractive, I am not convinced that these arguments are sufficient to estab-
lish proportionality as a independent side constraint on permissible systems of punish-
ment. Consider them in reverse order.
The second argument is that respect for persons as agents requires an overall an-
choring of the penalty scheme such that threatened hard treatment is not so severe as to
fail to recognize our status as reasoning beings. This means that were a penalty scheme to
be proposed that was very severe—perhaps in response to some consequentialist argu-
ment that, for example, conviction rates are so low that general deterrence can only be
achieved by increased penalties—the principle of proportionality would rule it out. The
argument that is deployed in support of this position is the “drowning out” objection.
That is, if the sentencing scheme is very severe, the moral appeal of the law will be lost
and citizens will think only in prudential terms. This is to control citizens by threats
rather than to offer them moral reasons for action (von Hirsch 1990; Duff 1998; von
Hirsch and Ashworth 2005).
The problem with this is that it just does not seem very plausible once one considers
the ways in which citizens actually reason. Consider someone considering parking ille-
gally. Presumably many people park illegally for short periods without thinking too
much about the moral wrong that may be involved. Now, consider what would happen if
the state imposed a severe penalty for this offense (say, the confiscation of one’s car). In
such circumstances, presumably most people would pause and think the risk was not
worth it. Should they then feel that they have been treated as less than an agent? That
seems wildly overdramatic. All that has happened is that the state has changed the out-
come of one’s prudential reasoning.

[ 38 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 38 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

For more serious offenses, the situation is difficult, but no less damaging for the
drowning out thesis. Imagine the penalty for murdering one’s spouse is to be choked and
then burnt at the stake. Would that make most married citizens think any less about the
moral reasons not to murder their husbands or wives? Surely not, since most people,
most of the time, do not think about the reasons they have not to commit murder. For
most people, the reasons not to do so are inert, since there is never an occasion in which
they need to figure in their mental life.
What of those people who are sorely tempted to murder their spouses? In such cases
it does not seem to me at all plausible that the absence of overwhelming prudential rea-
sons not to do so would help them to focus on the moral reasons not to do so. That point
has passed. In short, for the core criminal offenses, we are not—in Andrew von Hirsch’s
terms—neither saints nor sinners, but something in between (von Hirsch 1990). We are
actually saints or sinners (in the relevant senses) for whom the threat is either inert or
(we hope) sufficient.
Thus there is no independent principle of cardinal proportionality linked to a notion
of the respect we are owed as persons that can limit the system of punishment or speak
in favor of a reduction in sentencing levels. Of course, there are many other reasons why
we should limit punishments. For example, the risk of wrongly falling foul of the law and
the need to respect Bentham’s (1970, p. 168) injunction that penalties should not en-
courage wrongdoers to greater wrongdoing, but these are not arguments that can under-
pin proportionality as an independent principle.
What of the first, seemingly more powerful argument that ordinal proportionality
ensures equality; that like cases are treated alike? This has great appeal to liberals, for
whom equality is a foundational value. Thus, clearly a system of penalties that distin-
guishes between persons on the basis of skin color and imposes greater penalties on
black-skinned offenders than on white-skinned ones would be a paradigm instance of
injustice. By extension, a system that penalizes the kinds of drug use associated with one
community more severely than similarly harmful kinds of drug use associated with an-
other is unjust.
However, the argument becomes more complicated once one considers other, less
arbitrary rationales for different treatment (i.e., we need to be sure that the problem with
the above examples is that they fail to respect ordinal proportionality rather than that
they are based on arbitrary—and offensive—distinctions).
Assume that there are good public policy reasons for the state to wish to crack
down on a particular kind of offense; say, the state is very worried about the influx of a
certain gang culture and decides to issue sentencing guidelines that make gang-related
crime automatically subject to an extra tariff. Thus two offenders who are equally cul-
pable and have committed offenses involving equal harm may receive different pen-
alties as a result of one being in a gang and the other not. Is this a violation of one’s
status as an equal?
I think that is at least arguable. Of course, their treatment is unequal, but the respect
we are owed as citizens is, as Dworkin (1978) has usefully phrased it, not a matter of
equal treatment, but of “treatment as an equal.” Dworkin famously argued that it might
be the case that the state has an interest in developing African American professionals
(doctors and lawyers) and thus quotas for graduate school places in those disciplines

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 39]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 39 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

could be compatible with treatment as an equal (since the reason for differentiating
between applicants was not their skin color, but their ability to develop into socially
useful role models). Similarly, to treat gang members differently from otherwise iden-
tical offenders seems less like treating blacks differently from whites and more like
treating trainee brain surgeons differently from equally hard-working trainee beauti-
cians. The brain surgeon is entitled to expect greater rewards not because social policy
aims to reward the clever, but because rewarding the clever in this case serves a useful
social policy.
The point of this section is not to deny that proportionality has an important
role in our thinking about punishment. It does and will continue to do so. However,
it is not an independent principle. Rather, it is one of many considerations that
must be taken into account when we devise a system of criminal justice and punish-
ment. It may often be one of the most important considerations, and violations of
proportionality will require special justification, but it is not a liberal trump card
that can be played without further need for justification in proposing or criticizing
a penal system.

IV. PUNISHMENT IN THEORY

I have argued that neither the desert thesis nor the demands of proportionality neces-
sarily dictate the shape of a legitimate and just criminal justice system. However, I have
not said very much that is positive, that is, about how we might think about such a
system. This is critical if we are to be able to respond—as I indicated we should—to
developments in penal practice.
Of course, it is not possible here to offer a complete argument in defense of a
system of criminal justice. Rather, I want to say something general about how we
should go about constructing such an argument and something more particular
about how it might inform our responses to at least some recent changes in the
penal landscape.
Underlying the argument so far has been the Rawlsian thought that the only relevant
notion of deserved hard treatment (but not deserved censure) is one that “presupposes
the existence of the cooperation scheme” (Rawls 1971, p. 103). That is, once a just
scheme is in place, it gives rise to legitimate expectations (e.g., if one does not break the
law, then one will not be subject to punishment). In thinking about penal hard treat-
ment, then, we have to think about the overall just scheme and the legitimate expecta-
tions it creates. That is best done, I believe (but cannot defend that belief here), by
considering what agents would agree to in some suitably constructed hypothetical
choosing situation (see Matravers 2000; for a more Rawlsian take on the social contract
and its application to punishment, see Matravers 2011b, 2011c).
The task of giving an adequate account of punishment is, of course, familiar, and such
accounts are invariably controversial. This is not the place to try to develop another.
Rather, I want to say something about how we might evaluate some recent examples of
penal practice and how, in doing so, we might better prepare for whatever is next as we
enter the second decade of the new century.

[ 40 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 40 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

A. Coercive Treatment

Consider the example of coercive treatment; say, the order that an offender undertake
anger management therapy on pain of some further penalty. Coercive treatment was, of
course, one of the targets in the revival of retributivism, and for some retributivists, the
recent advance of so-called therapeutic jurisprudence (as championed by Wexler 1995,
2008; Winick 1997) represents a return to the dark days before the revolution.
Can coercive treatments of this kind be given a rationale that accords with the
demands of liberal egalitarian justice? On the face of it, things are not promising.
The coercive nature of directed treatments speaks against compatibility with free-
dom, the treatment element against autonomy, and the fact that different offenders
may receive different punishment against proportionality. However, appearances
may be deceptive.
Consider persons located in a suitably modified Rawlsian original position choosing
principles of penal justice. They do not know, of course, whether they will be disposed to
aggression in the “real world,” but they will know general facts about that world, such as
that there is a need for social order, assurance, and so on. They must then choose to
respond to aggression, but there is reason for them to argue that the response can be
moderated by the needs of social policy. There is no independent standard of entitle-
ments that the people in the original position must translate into their principles of jus-
tice. Rather, what citizens will be entitled to is itself determined by the principles of
justice. Thus they may be able to endorse coercive treatment models by reasoning over
each of the potential problems.
Take treatment first. The clearest case would presumably be something akin (although
not identical) to quarantine, but quarantine potentially avoids autonomy problems by
conceiving of the agent as a mere “carrier.” It is the virus (or whatever) that is quaran-
tined; the agent’s being coerced is merely an unfortunate by-product. A better analogy
might come from distributive justice. It is held by some that an agent with unchosen
expensive tastes should be compensated for the loss of welfare that results from these
tastes unless, for example, he would choose to keep those tastes even were he able to take
an otherwise harmless pill that would rid him of them. Or, imagine someone who needs
very expensive treatment for depression. We may think, other things being equal, that
she is entitled to such treatment at public expense, but not if we discover that there is an
easy, harmless cure for her type of depression that she refuses to take.
Putting to one side issues of identification, false positives, and so on, an agent with
anger management problems who (possibly as a result of those problems) regularly falls
afoul of the law might be thought to be in a similar situation. It is compatible with his
freedom for others to offer treatment, with the clear understanding that if he refuses that
treatment he must then pick up the full costs of his behavior (which may in this case
mean something like a severe penalty the next time he breaks the law). Establishing ex-
actly what that means is difficult to determine without a full account of the theory of
punishment a given contract would generate, but nevertheless the reasoning here is nei-
ther unusual nor incompatible with the liberal requirement to treat people as equals. The
intuition is a standard one: Why should anyone else bear the costs of his behavior if he
refuses to do what he can to change that behavior?

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 41]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 41 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

Of course, this example immediately raises autonomy worries, since I am speaking of

the offender as being in need of treatment—as ill or defective—rather than as an auton-
omous chooser. I am less worried about this in theory, although how much we have to
worry in practice depends on the proposed intervention (there is a significant difference
between courses to teach information technology skills, cognitive behavioral therapy
(CBT), and an intervention such as chemical castration). Many forms of intervention
can make our lives better and better enable us to cope with the world and others. Recog-
nizing that, and responding with offers to help, seems to me unproblematic. Except in
very extreme cases, I think the offer must be one that can be refused. But as indicated
above, in many instances it is a liberal principle that one should not be able to refuse
while passing the continuing costs of the problem on to others.
Finally, what of the fairness (or proportionality) worry? Could those participating in
the social contract endorse different punishments for the same offenses depending on,
say, the potential dangerousness of the offender? One way around this might be to sepa-
rate the punishment tariff from the dangerousness tariff and argue that in some circum-
stances it would be rational for the contractors to agree, in effect, to quarantine the
dangerous. In circumstances (which are far from obtaining) in which we could identify
the dangerous, there would not seem to be anything to bar this move.

V. PUNISHMENT IN PRACTICE

The above offers a very brief introduction to how one might think about penal policies
in the abstract. Having rejected the desert thesis, I have argued that “an entitlement to
proportionality in sentencing” is not an independent principle, but is—as with all
“entitlements”—determined by the wider theory of justice. For this reason, policy pro-
posals have to be thought through; they cannot simply be rejected as incompatible with
desert or proportionality. The result, of course, cannot be known in advance. It may be
that therapy, mandatory sentences, sentencing guidelines that are not strictly propor-
tional, and so on are all incompatible with a liberal theory of punishment. I doubt it, but
my case here does not rest on whether they are or are not. Rather, it rests on whether
these things can be dismissed on the basis of being incompatible with retributivism
understood in terms of the desert thesis or proportionality (they cannot). In thinking
about the future of punishment, we must not be complacent. The revival of retributivism
swept away many terrible practices, but it did not leave us with a coherent theory of
punishment with which to judge future policies.
Finally, what has been said above relates to how we might, as penal philosophers,
think about policy proposals. However, in responding to such policies we must be pre-
pared to accept contingent facts that may play no part in ideal theory. That is, even if it
were true that a form of preventive detention or a mandatory minimum sentence for
some offense were compatible with the liberal requirement to treat citizens as equals,
that does not mean that such policies are justified, all things considered. It may well be
that the background conditions of distributive injustice, or the capricious nature of those
who police (in the widest sense) the criminal justice system, mean that greater injustice
would be done by departing from proportionality than by sticking with it.

[ 42 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 42 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

In short, and to answer the question posed in the title, twenty-first century punish-
ment will be post-desert, but that is not a recent change. Our best theories of punish-
ment (and much else) are post-desert, and have been since long before the retributive
revival at the end of the twentieth century. That revival was not about desert, but was
about sweeping away many practices that resulted in actual injustices. It has left us with a
principle of proportionality that is important, but insufficient. The actual injustices that
occur as a result of our penal systems are still many, and in many places are increasing.
What is needed in the future is not only to think through what are the legitimate entitle-
ments and expectations of liberal citizens in relation to their criminal justice system, but
also to confront the problems that arise in the nonideal world in which we live. That is
probably best done by considering distributive and retributive justice together in the
hope that the resulting thoughts will influence those who make and apply the law.

NOTES
1. Nevertheless, the difference between the readings is important because the former would
seem to generalize to retributive justice (if, because we are not responsible for our natural
starting points, we do not deserve different treatment on the basis of those differences,
then that would seem to apply to all questions of desert), whereas the latter does not (since
it may be that there are differences between distributive and retributive justice that mean
natural inequalities can be legitimately translated into inequalities in outcome in the one
but not the other). It is worth pausing to note just how radical is Rawls’s rejection of desert
(on this, and on why it may explain the difficulty Rawlsians confronted in addressing
political questions at the end of the twentieth century, see Scheffler [2001, chap. 1] and
Matravers [2004]). For a general discussion of the relationship of distributive and retrib-
utive justice in Rawls’s theory, see Scheffler (2001) and Matravers (2011b, 2011c).
2. It is worth noting that Scanlon—a significant moral philosopher in the Rawlsian, neo-
Kantian mold—regards the desert thesis as “morally indefensible,” which is yet further
evidence that the assault on consequentialism in the philosophical literature was not led
by theorists committed to restoring desert to a central place in moral thinking.
3. As noted above, Cottingham identified nine forms of retributive thinking. Further distinc-
tions can be found in Walker (1999) and Tonry (2011b, pp. 108–9) (see also Cottingham
1979).
4. This subsection draws extensively from Matravers (2011a).
5. As John Gardner puts it, “the only Hart-approved reason in favour of punishing the guilty
(or anyone else) is the reason given by punishment’s general justifying aim, viz. that future
wrongdoing is thereby reduced” (2008, p. xxv).
6. The exceptions are Moore and Duff, who believe that the system of punishment gives form
to the need for moral criticism in response to wrongdoing.

REFERENCES
Bentham, Jeremy. 1970. “An Introduction to the Principles of Morals and Legislation.” In The
Collected Works of Jeremy Bentham, ed. Fred Rosen and Philip Schofield. Oxford: Oxford
University Press.
Cohen, Morris R. 1939. “A Critique of Kant’s Philosophy of Law.” In The Heritage of Kant, ed.
G. T. Whitney and D. F. Bowers. Princeton, NJ: Princeton University Press.
Cottingham, John. 1979. “Varieties of Retribution.” Philosophical Quarterly 29:238–46.
Dagger, Richard. 1993. “Playing Fair with Punishment.” Ethics 103:473–88.

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 43]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 43 August 3, 2011 1:00 PM

OUP UNCORRECTED PROOF

Dolinko, David. 1991. “Some Thoughts About Retributivism.” Ethics 101(3):537–59.

Duff, R. A. 1986. Trials & Punishments. Cambridge Studies in Philosophy. Cambridge: Cam-
bridge University Press.
———. 1998. “Punishment, Communication, and Community.” In Punishment and Political
Theory, ed. Matt Matravers. Oxford: Hart.
———. 2001. Punishment, Communication, and Community. Studies in Crime and Public Policy.
New York: Oxford University Press.
Dworkin, Ronald. 1978. Taking Rights Seriously: With a Reply to Critics. London: Duckworth.
Gardner, John. 2008. “Introduction.” In H. L. A. Hart, Punishment and Responsibility: Essays in
the Philosophy of Law, 2nd ed. Oxford: Oxford University Press.
Garland, David. 2001. The Culture of Control: Crime and Social Order in Contemporary Society.
Oxford: Oxford University Press.
Hart, H. L. A. 1968. Punishment and Responsibility: Essays in the Philosophy of Law. Oxford:
Oxford University Press.
Honderich, Ted. 1969. Punishment: The Supposed Justifications. London: Hutchinson.
Kleinig, John. 1973. Punishment and Desert. The Hague: Martinus Nijhoff.
Matravers, Matt. 2000. Justice and Punishment: The Rationale of Coercion. Oxford: Oxford Uni-
versity Press.
———. 2004. “Philosophy as Politics: Some Guesses as to the Future of Political Philosophy.”
In What Philosophy Is, ed. Havi Carel and David Gamez. London: Continuum.
———. 2007. Responsibility and Justice. Cambridge: Polity Press.
———. 2011a. “Duff on Hard Treatment.” In Crime, Punishment, and Responsibility: The Juris-
prudence of Antony Duff, ed. Rowan Cruft, Matthew H. Kramer, and Mark R. Reiff.
Oxford: Oxford University Press.
———. 2011b. “Mad, Bad, or Faulty? Desert in Distributive and Retributive Justice.” In Respon-
sibility and Distributive Justice, ed. Carl Knight and Zofia Stemplowska. Oxford: Oxford
University Press.
———. 2011c. “Political Theory and the Criminal Law.” In Philosophical Foundations of the
Criminal Law, ed. R. Antony Duff and Stuart Green. New York: Oxford University
Press.
Moore, Michael. 1982. “Moral Reality.” Wisconsin Law Review 1982:1061–156.
———. 1987. “The Moral Worth of Retribution.” In Responsibility, Character and the Emotions,
ed. Ferdinand Schoemann. New York: Cambridge University Press.
———. 1992. “Moral Reality Revisited.” Michigan Law Review 90:2424–533.
———. 1993. “Justifying Retributivism.” Israel Law Review 27:15–49.
———. 1997. Placing Blame: A General Theory of the Criminal Law. Oxford: Oxford University
Press.
Morris, Herbert. 1968. “Persons and Punishment.” Monist 52:475–501.
Murphy, Jeffrie. 1973. “Marxism and Retribution.” Philosophy and Public Affairs 2:217–43.
Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press.
———. 1999. “Justice as Fairness: Political not Metaphysical.” In Collected Papers, ed. John
Rawls and Samuel Richard Freeman. Cambridge, MA: Harvard University Press.
Scanlon, Thomas M. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University
Press.
Scheffler, Samuel. 2001. Boundaries and Allegiances: Problems of Justice and Responsibility in Lib-
eral Thought. Oxford: Oxford University Press.
Tonry, Michael. 2011a. “Can Twenty-first Century Punishment Policies Be Justified in Prin-
ciple?” In Retributivism Has a Past. Has It a Future?, ed. Michael Tonry. New York: Oxford
University Press.
———, ed. 2011b. Why Punish? How Much?: A Reader on Punishment. New York: Oxford
University Press.

[ 44 ] Retributivism Has a Past

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 44 August 3, 2011 1:00 PM

 OUP UNCORRECTED PROOF

von Hirsch, Andrew. 1985. Past or Future Crimes: Deservedness and Dangerousness in the Sen-
tencing of Criminals. Crime, Law, and Deviance Series. New Brunswick, NJ: Rutgers
University Press.
———. 1990. “Proportionality in the Philosophy of Punishment: From ‘Why Punish?’ to ‘How
Much?’” Criminal Law Forum 1:259–90.
von Hirsch, Andrew. 1976. Doing Justice: The Choice of Punishments: Report of the Committee for
the Study of Incarceration. New York: Hill and Wang.
von Hirsch, Andrew, and Andrew Ashworth. 2005. Proportionate Sentencing: Exploring the Prin-
ciples. Oxford Monographs on Criminal Law and Justice. New York: Oxford University
Press.
Walker, Nigel. 1999. Aggravation, Mitigation, and Mercy in English Criminal Justice. London:
Blackstone.
Wexler, David B. 1995. “Reflections on the Scope of Therapeutic Jurisprudence.” Psychology,
Public Policy, and Law 1(1):230–36.
———. 2008. Rehabilitating Lawyers: Principles of Therapeutic Jurisprudence for Criminal Law
Practice. Durham, NC: Carolina Academic Press.
Winick, Bruce. 1997. “The Jurisprudence of Therapeutic Jurisprudence.” Psychology, Public
Policy, and Law 3(1):184–206.

I S T W E N T Y-F I R S T C E N T U RY P U N I S H M E N T P O S T-D E S E RT ? [ 45]

TONRY_RETRIBUTIVISM-Chapter 02-PageProof 45 August 3, 2011 1:00 PM