The End of Liberty

Abstract

Theorists treat liberty as a great equalizer. We can’t easily distribute equal welfare, but we can purport to distribute equal liberty. In fact, however, nothing about “equal liberty” is meaningfully equal. To demonstrate, I turn not to familiar cases of distributing positive goods but to the distribution of a negative good, namely carceral punishment. Many theorists believe we should impose proportional punishment by depriving offenders of liberty in proportion to their blameworthiness. In this manner, equally blameworthy offenders are said to receive equal punishment when incarcerated for the same period of time. Equal periods of incarceration do not yield equal punishments, however, because liberty cannot serve as the great equalizer theorists hope for. Pretending it can prevents us from justifying the full harms of punishment or leads to such counterintuitive results that it makes proportional punishment an unattractive goal.

1 Introduction

Not all Americans have the same liberties. Most must pay for their own food, shelter, and healthcare. For one group, however, these essentials are provided at state expense. Members of this group also have access to government facilities that you are prohibited from entering, at least without special permission. What did they do to receive these freedoms? They committed serious crimes.

Prisoners have legal rights to food, shelter, and healthcare. You have no such rights. They also have some liberty to move about their cells and prison grounds. You, however, have no right to walk around a prison, absent special permission. Of course, the food, shelter, and healthcare prisoners receive are often quite deplorable, and the liberties they do have are rather unappealing. What we will see, however, is that even though many believe punishment should impose a deprivation of liberty proportional to offenders’ blameworthiness, there are many reasons to doubt that proportional deprivation of liberty should serve as a fundamental feature of criminal justice.

As the liberties unique to inmates reveal, not all liberties are the same in magnitude or quality. While many moral and political theories speak of equal liberty, liberty cannot be accurately measured. Even if it could, it’s not the amount of liberty that matters but its value, and that varies in ways theorists love to ignore. The details of measurement are complicated, inconvenient, and raise doubts about the appeal of “equal liberty” in political theory generally and of “proportional deprivation of liberty” in punishment theory.

The concept of liberty is surely not useless, but it may be better understood as just a shorthand, non-foundational way of speaking about a cluster of more fundamental matters. I will gesture at doubts about the central role liberty plays in moral, legal, and political theory but will defend a more limited claim. Namely, many retributivists believe that punishment should deprive offenders of liberty proportional to their blameworthiness (or culpability or crime seriousness or some such). While retributivists vary in their willingness to punish below what is proportional, they are nearly unanimous that it is impermissible to purposely or knowingly punish in excess of desert.¹ I will argue that any account of proportional punishment based on liberty deprivation will face tremendous obstacles and will likely be incapable of justifying common punishment practices such as incarceration.

1.1 Part I: Retributivism and Proportional Deprivation of Liberty

Many theories have been lumped together under the label “retributivism.” Some retributivists claim offenders deserve to suffer or be punished in proportion to their blameworthiness. Some claim it is intrinsically valuable when wrongdoers suffer or are punished for their wrongdoing.² At a minimum, “it is widely agreed that, for a justification of punishment to qualify as retributivist, a wrongdoer’s negative desert must play some significant role in the justificatory account.”³

Desert and proportionality are closely linked. Many believe that the more blameworthy a person is, the greater his negative desert. And as one is gradually punished, it seems, his negative desert is gradually reduced. So implicitly or explicitly, proportionality is central to retributivism. Antony Duff states that some principle of proportionality is “intrinsic to any version of retributivism.”⁴ Reflecting the widespread appeal of retributive proportionality, the almost-half-century-old Model Penal Code was amended several years ago to make proportionality the centerpiece of its sentencing philosophy.⁵

I will focus on problems for proportionality primarily in the context of incarceration. Incarceration presents the most interesting challenge for punishment theorists; it is a common practice (unlike the death penalty), and unlike fines and other burdens we regularly use in civil contexts, incarceration raises especially serious justificatory challenges. Of course, some retributivists may think incarceration unjust in its present form (and perhaps in any form), but the discussion will be broad enough to apply to carceral systems more humane than those commonly criticized in the United States.

• A. Familiar Problems with Retributive Proportionality

Retributive proportionality has three principal parts. It has a measure of blameworthiness, a measure of punishment severity, and a method of anchoring one to the other. Turning first to blameworthiness, I use the term as something of a placeholder. A retributivist might instead speak of wrongdoing, crime seriousness, or something similar. The differences are unlikely to matter for our purposes. What we’re talking about is the input to the proportionality calculation.⁶

Perhaps the biggest challenge in assessing blameworthiness concerns the debate over moral luck. Some theorists believe that blameworthiness is merely a function of the risks of harm an offender culpably creates.⁷ Others believe blameworthiness includes not only an offender’s culpability but also the harm the offender actually caused, if any.⁸

The differences can be dramatic. If we imagine two people engaging in identical conduct except that in the first case, the crime is consummated (for example, the bullet hits and kills someone) and in the second, the crime is merely attempted (for example, the bullet is deflected by a bird that happens to fly by), the first shooter could be convicted of murder while the second could, at most, be convicted of attempted murder, even though all that differs is the fortuity of the bird’s path. Though positive law clearly takes resultant luck into account, there is much debate over whether blame ought to align with the actual results of our actions or merely the risks we culpably create.

There are many other issues associated with measuring blameworthiness.⁹ For example, how do we rank the relative seriousness of a crime with a very culpable intentional mental state but a low risk of harm to a crime with a less culpable mental state but high risk of harm? In other words, retributivists must not only decide how harm factors into blameworthiness but also decide how different mental states connected with likely harms affect blameworthiness as well.

The most widely recognized challenge of proportionality is to anchor some amount of blameworthiness to some amount of punishment severity.¹⁰ No one has found a satisfying method. Paul Robinson and others claim that we have remarkably consistent cross-cultural intuitions about the relative seriousness of many traditional crimes.¹¹ But this observation, even if true and morally relevant, is of strikingly little help given the remarkably inconsistent cross-cultural intuitions about amounts of punishment in absolute terms.¹² Errors in the ordinality of punishment likely concern inmates far less than errors of cardinality. A series of punishments may be ordinally proportional, but that provides little comfort to inmates punished far too long in absolute terms.

My own scholarship has focused on problems connected to punishment severity—the presumed output of proportionality analysis. According to what we can call “suffering retributivism,” offenders should suffer in an amount proportional to their blameworthiness.¹³ My critique of suffering retributivism asks us to imagine that Sensitive and Insensitive are equally blameworthy offenders sentenced to equal terms of incarceration under identical prison conditions.¹⁴ The only relevant difference is that Sensitive finds incarceration unbearably upsetting (though, we could add, not in ways that rise to the level of recognized mental illness) while Insensitive dislikes incarceration but is able to cope, make friends, explore hobbies, and learn useful skills. Because they are equally blameworthy, suffering retributivism suggests that something has gone wrong: Sensitive and Insensitive should suffer, but are not suffering, the same amount.

While we could try to change our practices to sentence Sensitive and Insensitive in ways that reflect their unique experiences, doing so leads to surprisingly counterintuitive results. For example, even though wealthy celebrities like Paris Hilton will likely be more sensitive to prison conditions, few think they should spend less time incarcerated (or have swankier prison conditions) than ordinary people when they commit crimes of equal blameworthiness. While there is much more to say on the topic,¹⁵ contemplating variation in the experience of punishment raises questions about whether we truly approve of proportional punishment as understood by suffering retributivists. It also raises the possibility that instead of focusing on suffering, retributivists should understand punishment severity in the form of something else, such as liberty deprivation.

• B. Understanding Punishment Severity as a Liberty Deprivation

According to John Rawls, under the proper conditions, someone is punished “whenever he is legally deprived of some of the normal rights of a citizen.”¹⁶ Ken Simons writes that “[w]hen the state imposes criminal sanctions, it deprives the offender of property or liberty.”¹⁷ Robert George takes retributivists to hold that “a criminal may justly be deprived of liberty commensurate with the liberty he wrongfully seized in breaking the law.”¹⁸ Similarly, David Gray states that, “[f]or the most part, retributivists argue that offenders who perpetrate serious crimes deserve severe deprivations of their liberty, and those who commit less serious crimes deserve less severe deprivations of their liberty.”¹⁹

Scholars are not always clear whether they are speaking of liberty deprivation as a method of punishment (for example, incarceration falls under the category of liberty-depriving punishments while flogging falls under the category of pain-inducing punishments) or whether they are speaking of liberty deprivation as the currency of punishment (meaning the yardstick by which we ought to measure the severity of an instance of punishment). I believe that many go beyond merely describing a method of punishment to claim that we morally ought to deprive offenders of their liberty in amounts proportional to their wrongdoing. In any event, this is the view that I will address.

Scholars can also be unclear about precisely what they mean by liberty. I will read them as speaking of some form of negative liberty (a right against interference with one’s options).²⁰ Punishment is one way of interfering with options people would otherwise have. So while no one to my knowledge has developed a detailed normative theory of proportional negative liberty deprivation, my task is to show that likely versions of such views face substantial challenges.

• C. Some Considerations in Favor of Liberty-Deprivation Retributivism

There are several reasons one might initially be attracted to the view that punishment severity ought to be measured as a liberty deprivation. First, liberty is closely connected to broader justifications of state coercion. Political philosophers might argue that we have no rights in a state of nature. The state gives us certain rights, and those rights come with corresponding obligations to obey the criminal law. When one violates a criminal law, one loses some rights of citizenship. The claim under consideration is that the state grants us liberties in the first place, so it is fitting that some violations of a social contract result in the loss of some of those liberties.

A second reason potentially speaks to those who claim that retributivism is barbaric.²¹ They say that suffering retributivists inappropriately respond to wrongdoing by causing additional suffering. Liberty-deprivation retributivists, by contrast, aim not to cause suffering but to remove liberties of citizenship, even if they know they will cause suffering as a side effect. Anti-retributivists might argue that liberty deprivation is also an inappropriate response to wrongdoing because liberty deprivation will surely cause tremendous suffering. From a public relations standpoint, however, liberty-deprivation retributivism has a less sadistic ring to it.

Third, analyzing punishment severity in terms of liberty deprivation may better match our intuitions about stoical inmates who cope unusually well in prison. Suppose such inmates meditate for hours each day and gradually come to terms with their prior wrongdoing. Though their daily life is far from comfortable, they have developed an overarching sense of tranquility in the face of life’s challenges and make progress along spiritual paths to greater fulfillment and enlightenment.

Stoical inmates seem to present a challenge to suffering retributivists. They could say stoical inmates are not properly punished because they are suffering too little. But if their routines are rehabilitative, they seem like model prisoners. Perhaps suffering retributivists can identify suffering in inmates’ lack of ordinary comforts, even if they experience incarceration as a net benefit (though it’s hard to see how they are receiving the treatment they deserve for what may have been brutal crimes if they find the treatment enlightening and fulfilling).

If suffering retributivists claim to identify suffering in the deprivation of life’s ordinary comforts, what about the rare, but still quite real inmates who are happier in prison than outside because they get reliable food, companionship, and healthcare? Their enjoyment of incarceration may be neither rehabilitative nor spiritually redemptive. Must they be subjected to even harsher conditions to extract suffering? Whether or not suffering retributivists can give satisfying answers to these questions, liberty-deprivation retributivists can largely sidestep them by recognizing that, no matter what inmates experience, they are still subject to serious deprivations of liberty.

Finally, as mentioned, liberty-deprivation retributivism has been proffered as a response to problems raised by variation in the subjective experience of punishment as illustrated by the Sensitive and Insensitive example.²² If we understand punishment severity in terms of liberty deprivations, then Sensitive and Insensitive are arguably punished proportionally because they are both deprived of equal liberty when they are equally restricted in their freedom of motion (and other rights) for the same period of time under identical conditions.

1.2 Part II: Problems with Liberty-Deprivation Retributivism

At first glance, at least, liberty-deprivation retributivism fits some of our punishment intuitions, avoids some problems afflicting suffering retributivism, and may have the public relations advantage of appearing less barbaric. In this part, I argue that, despite these superficial advantages, liberty-deprivation retributivism is unworkable and counterintuitive.

• A. Measuring Liberty

The view that liberty deprivation should be proportional to blameworthiness implies that there are amounts of liberty. Yet measuring liberty is not just difficult; it may be impossible. Do we have more liberty in our freedom of expression or our freedom to associate with others? Do we have more liberty to picket outside the White House or display controversial art near the local library? Without some rough measurement of liberty (“libertrons” perhaps), it is doubtful that we can make liberty deprivation proportional.

Sometimes we can compare relative magnitudes of liberties. Suppose citizens are required to provide one year of (a) military service, (b) community outreach, or (c) research in medical facilities. Then the state takes away the option to participate in medical research. We can now say that citizens have less liberty than they had before. Similarly, when a new option is added, they have more liberty than before. So long as one set of liberties is a subset of another, we know which is larger.²³ Beyond that, however, there is often little we can say about quantities of liberty.

Defenders of proportional punishment may be comfortable with the idea that, despite widespread disagreement about how to measure crime seriousness and punishment severity and how to anchor the two, these matters largely boil down to intuitions. They can’t tell us precisely how serious some crime was, but they have an intuition that a particular punishment “seems about right” or “is not clearly disproportional” or some such.

It is more plausible, however, that these intuitions speak to amounts of suffering rather than amounts of liberty deprivation. I simply have no strong intuitions about the size of freedoms of expression relative to sexual freedoms relative to freedoms to choose one’s profession and so on. Under the best circumstances, proportional punishment is a hazy notion that is hard to operationalize. When proportional punishment is rooted in liberty deprivation, it imports not only the general difficulties associated with proportionality and anchoring but special intuition-resistant difficulties of measuring amounts of liberty.

Another problem with measuring liberty is that amounts of liberty may vary based on our capacities. A jurisdiction that deprives inmates of the right to abortion surely deprives many women of liberty, but does it deprive men? If liberty is understood as interference with an option people would otherwise have, then (cisgender) men do not have the option to abort and cannot be deprived of the liberty to do so. If all other liberties are the same, young women in this jurisdiction would generally be deprived of more liberty than young men when incarcerated for the same time under the same conditions. Hence, we cannot determine the extent to which people are deprived of liberty without a careful assessment of their capacities.

Since magnitudes of liberty depend on capacities, those magnitudes are at least as hard to measure as are capacities. With enough time, energy, technology, and money, many post-menopausal women could become pregnant. If a prison system allows women to use artificial reproductive services at their own expense but deprives them of the right to abort, does the abortion restriction limit the liberty of all post-menopausal women who could possibly become pregnant? Somehow measurements of liberty have to take into account notoriously difficult measurements of capacities.

The prior examples involve liberty deprivations, though they are arguably not punitive. Perhaps we should only deem liberty deprivations punitive when they are intended as condemnation for past bad behavior. But speaking of common punitive deprivations of liberty doesn’t help much. Recall the point made in the introduction. Inmates both lose liberties (freedoms of expression, association, movement, and so on) and gain liberties (access to free food, shelter, and healthcare, as well as freedom to be present in particular cells and areas of the prison grounds that are off-limits to non-inmates).

Commonsense suggests that inmates lose net liberties, but what grounds this determination? (Some might describe access to prison food, shelter, and healthcare as positive liberties, but the point can be made by focusing only on the negative liberty of being free to move around a correctional facility in accordance with the rules governing inmates.) Suppose at some point in history, a British offender was forced to move to Australia as punishment for a crime. At the same time, a person native to Australia committed a crime and was banished and forced to sail away from Australia. If they’ve essentially switched liberties, it’s hard to see how both offenders have been punished by a deprivation of liberty.

Perhaps blameworthiness isn’t supposed to be proportional to liberty deprivation but to the aversiveness of liberty deprivation.²⁴ That could explain how both the Brit and the Australian could be punished even if they merely switch liberties. After all, they value particular liberties differently. Similarly, depriving felons of voting rights and rights to bear arms is a deep concern to some and inconsequential to others. Losing the right to abortion would devastate some female prisoners but matter little to those who would never abort in the first place.

Focusing on the aversiveness of liberty deprivation helps explain which liberties society tends to use as punitive deprivations. We could punish offenders by prohibiting them from attending the opera. While doing so would deprive offenders of liberty, it wouldn’t be much of a punishment to those who hate opera,²⁵ and so it’s likely not a good general method of punishment. What seems to be doing the relevant work here is aversiveness of harsh treatment rather than liberty deprivation.

Liberty deprivation has other counterintuitive features that general aversiveness does not. If offenders were deprived of the right to attend operas, that seems equivalent to being deprived of the right to see La Traviata, Madama Butterfly, Die Zauberflöte, and so on. If we focus on deprivation of liberty, it seems that we can deprive inmates of more liberty by creating new operas. And since incarceration consists of many liberty deprivations (including the right to attend operas), whenever a new opera is created, all prison sentences become just a bit more severe than they were before the opera was created. On the other hand, if we focus on the aversiveness of liberty deprivation, we avoid this counterintuitive result: the creation of new operas has little impact on those who detest operas of any sort.

Notice that turning to the aversiveness of liberty avoids some counterintuitive results but does so by relying on subjective variation. And as we will discuss in Part III, if the currency of punishment is the aversiveness of liberty deprivation, all the problems raised by Sensitive and Insensitive quickly return. Hence, measuring liberty itself is practically impossible and may be especially resistant to intuitive assessment. Measuring the aversiveness of liberty is more practical but seems to destroy the original impetus for turning from suffering retributivism to liberty-deprivation retributivism.

• B. The Dual-Measurement Nature of Punishment

A further problem with proportional liberty deprivation is that the magnitude of a deprivation requires not one but two difficult or impossible measurements. It requires a measurement of an offender’s liberty in a baseline condition and a measurement in a punished condition. The deprivation depends on the change between the two. For example, inmates have less freedom of expression than ordinary people, but they still have some. To measure their deprivation, we compare their freedom of expression while incarcerated to their baseline level.²⁶ Some might understand the baseline level as a pre-punishment level (the offender’s liberties before his sentence began), and some might understand it in counterfactual terms (the offender’s liberties had he not been punished).²⁷ But either way, liberty is measured as a change from a baseline.

Once we recognize these baselines, however, many problems arise that I have discussed elsewhere in more detail.²⁸ For example, if people commit a crime while in long-term medical isolation or immigration detention, their baseline condition is already quite liberty restricted. To punish them with severity equal to that of ordinary inmates, we would have to deprive them of greater liberty (or the same amount of liberty for a longer duration) in order to extract an equal deprivation. Such a result is counterintuitive to most people.

Moreover, variation in baselines due to detention or medical isolation are just two dramatic examples of a very common phenomenon. All people vary in their baseline liberties to access and use real and personal property. When we incarcerate, we deprive offenders at least temporarily of virtually all of their real and personal property. This means that the amount inmates are deprived of liberty depends on how much money and property they had in their baseline conditions.

Though this result seems inescapable for supporters of proportional liberty deprivation, most think we punish the very rich and the very poor the same when they are incarcerated for the same period of time. In fact, however, doing so deprives the rich of more liberty than the poor because the rich had more liberty to begin with. In other words, when we consider the dual-measurement nature of punishment,²⁹ we return to a variation of the Paris Hilton problem that reliance on liberty deprivation was supposed to help us avoid.

The dual-measurement nature of punishment also means that if we take proportional liberty deprivation seriously, the severity of imprisonment can once again change based on conditions outside of prison. For example, suppose that soon before her nineteenth birthday, Brenda commits a crime and deserves a maximum liberty deprivation of one-year in some particular prison. Just before sentencing, however, her country institutes a military draft requiring all nineteen-year-olds to serve for one year. People in that age group have now lost substantial freedom because they must follow the military’s rigid requirements of where to sleep, when to wake, what to eat, and what risks one must take with one’s own life. If we assume Brenda’s baseline condition includes a year of military service which she will now avoid because of her prison time, it would now be permissible (and perhaps obligatory) for proportional liberty deprivation retributivists to sentence Brenda to significantly greater than one year in prison to make her liberty deprivation proportional. This is so even though they would have deemed it impermissible overpunishment to sentence her to greater than one year of incarceration absent the draft. Another implication for those who use a counterfactual baseline is that all the nineteen-year-olds already in prison when the draft is instituted would suddenly have much less severe sentences. Not all will find such conclusions objectionable. But our punishment practices do not overtly take baseline liberties into account, perhaps because doing so seems so troubling in Paris Hilton-style cases.

The dual-measurement nature of punishment reveals particularly deep flaws in our intuitions about proportionality when we compare our approach to fines relative to incarceration. When we fine offenders, we implicitly accept dual-measurement. A fine represents a decrease in assets from a baseline state to a punished state. The amount of the loss is supposed to be proportional to the offender’s blameworthiness. By contrast, when it comes to incarceration, our intuitions do not readily line up with the dual-measurement approach. People think that wealthy Jeff Bezos would be equally deprived of liberty when placed in the same carceral conditions as an equally blameworthy but penniless counterpart. In fact, however, Bezos would be deprived of far more liberty, including the liberty to walk around his estates and to use his extensive personal property.

The dual-measurement method is the more viable liberty deprivation candidate because it has the potential to justify the full measure of harms that punishment inflicts understood as setbacks to liberty interests.³⁰ But this method of measurement would lead to carceral practices most consider quite disproportional. And we can’t have it both ways. Either the dual-measurement method used for fines leads to proportional punishment or the single-measurement method we in fact use for incarceration does. To the extent our intuitions seem to prefer inconsistent methods, those intuitions are suspect.

• C. The Awareness Requirement

A person can be deprived of liberty without knowing it. Consider the “unknowing confinee” who has been convicted and sentenced to a month of home confinement. His punitive confinement has already begun, but he mistakenly believes it begins a few months hence.³¹ Nevertheless, he stays home frantically creating a website for a demanding client. Had the unknowing confinee tried to leave, armed guards outside his house would have prevented him. But blissfully unaware, he serves his sentence without realizing it until soon after it ends. (If it matters to you, you can assume the rest of the world knew he was confined so that some sort of condemnation was communicated to the public.) Has the unknowing confinee been retributively punished?

Liberty-deprivation retributivists, as I’ve described them so far, seem compelled to say “yes.” The confinee has, in fact, had his liberty restricted. Yet it is hard to see how he received the punishment he deserves on account of his wrongdoing when his life, as he experiences it, is entirely unaffected by the punishment. By contrast, suffering retributivists can simply deny that the confinee was punished, and I think they better capture most people’s intuitions.

Liberty-deprivation retributivists could insist that offenders deserve to lose liberty but add a condition that offenders be aware of the deprivation. The reason for adding the requirement, however, seems rooted in the negative experience of losing one’s liberty. After all, awareness that one’s liberties were deprived in the past does not seem to count as punishment. And if the unknowing confinee were to discover his confinement with one day remaining on his sentence, we’d presumably only count that one day as punishment, suggesting that negative experiences are far more relevant than liberty deprivation is.

The case of the unknowing confinee is, of course, highly stylized. But its real-world equivalent is not. Incarceration reduces many liberties of which inmates are likely unaware. Suppose a jurisdiction intentionally reduces inmates’ freedom of movement as well as their freedom of expression. A particular inmate never seeks to use any of the prohibited means of expression and for some reason never finds out about them. The logic of the awareness requirement suggests that this inmate is underpunished relative to similar inmates who are aware of their reduced freedoms of expression.³² Yet few would think the inmate deserves additional liberty deprivation because he was unaware of particular liberties of which he was deprived. Why require awareness that some punishment is underway but not require awareness of all the punishment underway?

1.3 Part III: Objections and Replies

• A. Measuring the Value of Liberty

As already suggested, the most promising path for liberty-deprivation retributivists is to focus not on proportional deprivation of liberty but proportional deprivation of the value of liberty. Doing so would resolve many of the problems noted earlier: even though inmates gain some liberties as they lose others, the value of liberties they gain is likely quite small relative to the value of liberties they lose. Similarly, the unknowing confinee isn’t punished merely because he lost liberty to leave his home. He lost liberty, but given his lifestyle choices, he did not lose options of value to him. Even if he valued the option to leave his home, he was unaware that he lacked that option so couldn’t experience its disvalue.

The problem with understanding proportionality as a deprivation of the value of lost liberty is that it brings back problems raised by the subjective experience of punishment. When Sensitive and Insensitive spend equal time incarcerated under identical conditions, they are still punished by different amounts because they value their liberties differently. If we adjust their sentences or conditions of confinement to take such differences into account, we will typically incarcerate wealthy celebrities for less time or under more luxurious conditions than ordinary people who are equally blameworthy. If the point of focusing on liberty was to avoid these problems, the approach has failed.

A pluralist might recognize both suffering and liberty deprivation (and more) as contributing to the proper measurement of punishment severity.³³ But doing so invites all the problems of suffering retributivism and liberty-deprivation retributivism as well as additional problems raised by the need to combine the two.

One could argue that while we care about the value of lost liberty, such valuations should be objective. The problem is a justification of punishment must justify the harms we purposely, knowingly, or foreseeably cause. These harms vary both because inmates experience confinement in different ways and because they have different baselines. Even if an offender’s desert is adequate to justify some objective or average amount of harm, that doesn’t make it adequate to justify the greater harm a sensitive, high-baseline offender actually suffers. In tort law, we do not compensate false imprisonment with an objective formula based on the number of hours or days falsely imprisoned. We try to determine the amount of harm suffered by the individual plaintiff.

Similarly, if we ignore the ways sensitive inmates suffer, their negative desert will fail to justify causing their above-average harms. Punishing them would knowingly be in excess of desert whenever we declined to hear evidence that they are particularly sensitive. We could gear amounts of punishment to the most sensitive offenders with the highest baselines, but such offenders may be one or more orders of magnitude more sensitive to punishment than average. We might have to punish violent rapists on the order of weeks or months rather than years, amounts so meager that they: (1) could hardly be deemed proportional by conventional standards, (2) would likely be unsatisfactory to the masses, and (3) may not justify the resource costs of a criminal justice system.

• B. Punishment Must Be Intentional

Punishment is typically understood as an intentional infliction. In H.L.A. Hart’s influential description, the “standard or central case” of criminal punishment “must be intentionally administered by human beings other than the offender.”³⁴ So, one might argue, it doesn’t matter whether Sensitive and Insensitive have very different experiences of punishment. They are punished the same provided that their punishers had the same intentions.³⁵ Similarly, one might argue, so long as a punisher intends to deprive an offender in medical isolation and an ordinary offender to the same degree, they both receive the same punishment.³⁶ The claim is that we can avoid some of the problems proportionality creates by focusing on the intentions of punishers.

This objection reflects perhaps the biggest confusion afflicting theorists on the topic of proportionality. Namely, we use the term “punishment” to mean two very different things. Sometimes we ask whether an act constitutes punishment. For example, is it punishment to incarcerate a violent, threatening person who has not yet committed a crime? (Hart would say “no,” because the treatment is not for an offense against legal rules, just the risk of such an offense.) Here we are asking about the “act-sense” of punishment. In legal contexts, the act-sense of punishment may determine whether some treatment violates due process or is unconstitutionally “cruel and unusual.”

Applying the act-sense of punishment to proportionality, however, yields some odd results. First, it conflicts with the common view among moral philosophers that intentions are irrelevant to moral permissibility. Many believe intentions are irrelevant to permissibility by contemplating scenarios like the following: assume it is morally permissible to divert an out-of-control trolley about to kill five people even though the train will kill one person on the diversionary track as an unfortunate side effect. If so, it also seems permissible to divert the trolley not to save the five but simply because one wants to watch the death of the person on the side track. If the conduct is permissible in the first scenario, the fact that the person in the second scenario diverts the trolley with a bad intention may make him blameworthy but does not affect the moral permissibility of the action.

Importantly, the objection that punishment only includes intentional inflictions not only treats intentions as relevant to the permissibility of punishment, it restricts the entire scope of punishment to intentional conduct. And by making intentions central to the permissibility of conduct commonly called punishment, the objection may conflict with the views of “perhaps a majority of contemporary moral philosophers.”³⁷

Perhaps the tendency to treat intentions as central to punishment should not surprise us because, as Husak argues, much of the criminal law treats intentions as relevant to permissibility.³⁸ The criminal law’s focus on punitive intent may even provide some ammunition against the conventional wisdom of moral philosophers today. But recall that we are not now considering a matter of criminal law. We are considering the moral permissibility of criminal punishment. The permissibility of a social practice is not itself a matter of criminal law. Hence the frequency with which the criminal law treats intentions as relevant to permissibility has no obvious bearing on questions about the moral justification of treating punitive intent as relevant to the permissibility of punitive acts.

Second, if we really restricted punishment to its intentional components, most real-world punishments would be indeterminate in magnitude. Suppose a judge sentences an offender to ten years’ incarceration. Now ask the judge: How much do you intend to restrict the offender’s liberty? How large do you intend his cell to be? Are you intentionally restricting his freedom of expression as punishment or is that merely a side effect of promoting prison safety and security? We are not likely to receive satisfying answers for many reasons—among them, these matters are largely controlled by prison bureaucrats rather than the judges who select sentences.

We could also ask judges: When an inmate you sentenced is crestfallen because he can no longer hold his infant daughter, is that suffering intended? What if his wife was pregnant with the child at sentencing but everyone was unaware? Did you anticipate he might have a child whose absence would later make him suffer? These questions are more difficult still because the intentions governing an offender’s sentence likely depend on an amalgam of the intentions of legislators, governors, prison bureaucrats, corrections officers, and so on.

Third, sentences may be not only indeterminate in magnitude but indeterminate in form. We have been speaking of liberty-deprivation and suffering as though the choice of metrics is a matter of theory. But if punishment depends on intentional inflictions, perhaps we can only make sense of punishments based on the form intended by the punisher. So some inmates are being made to suffer, some to be deprived of liberty, and some are in a strange superposition of suffering and liberty deprivation based on the varied views of the various people who set and administer their punishments. Of course, theorists can still have a view of what intentions punishers ought to have, but if they hope to speak to the real world, they would need a view of how to assess the magnitude of punishments composed of a cloudy haze of imprecise intentions.

Fourth, we don’t actually care whether inflictions are intentional or not when assessing punishment severity. Suppose Purp and Fore are equally blameworthy and sentenced to equal terms in identical prison conditions.³⁹ In fact, they are alike in all relevant respects except that all the judges, legislators, prison administrators, and so on in Purp’s jurisdiction intend all of the inflictions associated with prison while the same people in Fore’s jurisdiction merely intend to deprive Fore of his liberty of movement. For those in Fore’s jurisdiction, all the other limitations are mere side effects of his punishment. Under these circumstances, Purp is being punished far more severely than Fore because almost every bad thing that happens to Purp is punishment whereas the equally bad things happening to Fore were merely foreseen or considered highly likely. So if the intentionality objection is taken seriously, we could keep Fore incarcerated far longer than Purp and still deem their punishments equal and proportional.

The reason the Purp and Fore example strikes us as so absurd is that we don’t follow the “gift model” (it’s the thought that counts) when it comes to assessments of quantities of punishment. The views of laypeople seem to fetishize the passage of time almost at the expense of everything else. But a more sensible retributivist view would follow the “jellybean model.” An offender has a certain desert reflected by the size of a jellybean container. As the offender receives the inflictions commonly associated with our punishment practices, the container gradually fills with jellybeans over time. Maybe the red jellybeans reflect intentional inflictions, but there are blue jellybeans reflecting foreseen inflictions, and green jellybeans reflecting inflictions that were expected with various probabilities. Some inflictions that were very much unexpected and could happen to anyone, perhaps a construction accident at the prison, might get no jellybeans at all. But whatever the details, our intuitions about the sorts of inflictions that fill up the container (and thereby reduce and eventually eliminate an offender’s desert debt) include far more than just intentional inflictions.

Our practice of giving offenders credit for time served may offer the clearest evidence that we reduce desert debt by non-intentional inflictions. Almost everywhere, defendants are given credit for time spent in pretrial detention on a day-for-day basis against any incarceration they may face if subsequently convicted. So if we reduce offenders’ desert debt for inflictions pretrial that were not intentionally administered as punishment, we clearly should reduce their desert debt for inflictions during incarceration which are at least arguably part of punishment.⁴⁰ Whatever one thinks about the act-sense of punishment, proportionality relies on the “quantity-sense” of punishment that is more about the accrual of harms associated with our punishment practices than the set of acts that fall under the act-sense of punishment.

• C. Retributivism is About Punishment Not Punishment Practices

How can it ever be justified to intentionally harm someone because of their wrongdoing? This is the question some retributivists take themselves to be addressing, and it is an interesting question in its own right. But retributivists who purport to defend proportionality in the context of state punishment must go deeper. They ought to say what it means for a carceral sentence to be proportional. As I have argued, proportional liberty deprivation leads us in counterintuitive directions.

Focusing on intentional liberty deprivations fails to capture what incarceration is all about. Suppose one takes incarceration to impose intentional reductions in inmate’s freedom of motion. This could mean spending several hours a day in a small cell and several more on shared prison grounds. But if we hold liberty constant, one could describe a barren, boring, dangerous prison or a ritzy Four Seasons-style resort prison, so long as the resort “inmates” are required to stay on the premises and spend several hours a day in their luxurious, though small bedrooms. How could we possibly say that two equally blameworthy inmates in these different facilities are punished equally per unit time?

Moreover, when retributivists purport to justify punishment in the act-sense rather than in the quantity-sense, they fail to justify punishment practices such as incarceration. Incarceration involves inflictions by many people with varied beliefs and intentions. The harms that require justification go beyond just intentional actions. To the extent we know incarceration harms and risks harms to prisoners, we need a justification for causing these harms and risks, just as we need a justification for all significant harms and risks we intentionally or foreseeably cause.

Some retributivists might argue that they seek to justify punishment, not punishment practices such as incarceration. They might say the task of justifying punishment practices (which include unintentional harms) belongs to other aspects of moral or political theory. Doing so feels like a bait and switch, however, at least by retributivists who purport to speak to real-world issues of state punishment. In that realm, proportionality is often thought to inform sentencing practices and those practices frequently include incarceration. Any retributivist discussion of proportionality is of little use if it only speaks to intentional harms, for such harms arguably represent a relatively small fraction of the harms of incarceration. And such discussions should come with major warning signs because most people, as seen in the Purp and Fore example, treat a person’s desert debt as satisfied by all sorts of inflictions, even when they are not strictly intended.

2 Conclusion

As I write, the world is in the grip of a global pandemic that may lead to the death of millions and the loss of trillions of dollars in the value of the world economy. Yet the pandemic (or one like it) may have been started by a single person’s criminal violation of a law governing safe animal butchering practices.⁴¹ Such laws have been routinely violated, probably with little harm on most occasions, but with a tiny probability of catastrophic harm. If true, what is a proportional punishment for such a crime?

To me it seems, the answer doesn’t matter. We should determine appropriate punishments with reference to what will lead to the best consequences. The best consequences may sometimes require us to consider people’s perceptions of proportionality. Those perceptions can have very real effects on people’s behavior and even their well-being. From that perspective, perceptions of proportionality certainly matter—they matter even when they are unstable and inconsistent.

As for whether proportionality itself matters, there are good reasons to be skeptical. Whether we understand punishment severity in terms of bad subjective experiences, liberty deprivations, or a combination of considerations, measuring punishment severity is extraordinarily difficult and leads to counterintuitive results. Those more sensitive to punishment and those with better baselines (the wealthy, for example) would generally need better or shorter treatment than their equally blameworthy, ordinary counterparts to be punished with equal severity. On the flip side, those with worse baselines, such as those in medical isolation or immigration detention, would typically require longer incarceration or worse conditions in order to be punished with the same severity as their equally blameworthy, ordinary counterparts.

Proportionality’s greatest theoretical weakness—its inscrutability—may be its greatest rhetorical strength. Almost everyone can say they support proportional punishment because they have wildly different interpretations of what proportionality means, both in principle and as applied to particular cases. When it comes to factual issues that bear on how much punishment a person deserves, most retributivists accept a very demanding beyond a reasonable doubt standard that reflects the high Blackstonian cost of punishing the innocent. It is, therefore, puzzling that retributivists—who recognize a firm deontological prohibition on intentional or knowing overpunishment—would be satisfied with the low credence in proportionality that even its most ardent supporters can realistically have.⁴²

I have sought to put an end to liberty deprivation as the sole end of proportional punishment. Whether or not I’ve succeeded, I have not made the case that we ought to put an end to the end of liberty in moral, legal, and political theorizing more generally. At most, I address just one domain, and do so rather quickly at that. But there are some hints at broader problems for liberty, as I suspect theorists resort to liberty in the first place as a somewhat desperate measure to find some substrate that can easily be applied equally to different people. The problem is that precisely those aspects of liberty that make it versatile and convenient strip away important features of the liberty holder that affect the value of the liberty in question. We should resist the effort to rely on liberty as a convenient equalizer or measuring rod because efforts to identify quantities of liberty are in jeopardy in all domains.