Normative elements of parole risk.

INTRODUCTION

In prison systems employing discretionary parole release, prisoners are not granted automatic release into society at the end of an enumerated sentence of years. Instead, a parole board assesses whether eligible prisoners meet criteria for safe release. The upside to sentences terminating in parole release what I will call indeterminate sentences (1)--is obvious. First, states can use parole as a population safety valve without indiscriminately endangering public safety, since parole boards can release only those prisoners least likely to reoffend. Second, indeterminate sentencing gives prisoners incentives both to behave and to rehabilitate themselves, since misbehavior and untreated risk factors will prolong their stay in prison. Parole boards are ideally situated to make prerelease assessments of a prisoner's risk because they can do so shortly before release. A sentencing judge, on the other hand, can only hazard a guess at what risk an offender might present at the end of his or her term. Parole fits neatly within the wider move towards evidence-based sentencing, which involves the use of actuarial tools and risk assessments to guide officials in their decisions about sentencing and release.

Although California employs indeterminate sentencing for several types of crimes, the state's use of indeterminate sentencing is less a form of parole release than a form of parole retention: few prisoners are ever released. According to statute, parole shall "normally" be granted unless "consideration of the public safety requires a more lengthy period of incarceration," but this presumption of release is not observed in practice. (2) Just six parole-eligible murderers out of several thousand eligible were granted parole release during the tenure of Governor Gray Davis; by one recent estimate, each year the parole board finds only three percent of parole-eligible prisoners serving life sentences suitable for release, and only one percent are actually released after review by the full parole board and the governor. (3) Parole guidelines and expert assessments of risk are supposed to result in predictable and logical release decisions. In California, however, release decisions are predictably and illogically negative.

One reason for California's parsimonious release rate has to do with the fact that the parole board considers only the risks of release, not the benefits. Without analyzing the costs and benefits of release (or retention), however, we cannot assess whether release (or retention) is worth the cost. But even if we could calculate the costs and benefits more accurately, the parole board would still be faced with what I contend is a normative question: whether an offender should be released. How much risk is too much risk? How much cost is too much cost? Without clearly answering these questions ex ante, the state has backed into a policy commitment one individual release decision at a time.

In this Article, I argue that these normative questions are not ones that the parole board should answer. Parole board officials are experts at evaluating risk, but they are not experts at calculating how much society values (or devalues) those risks. Other bodies must set the policy, answering the "should" questions about how much risk reduction is worth what cost, leaving the parole board to implement these policies. So, while I will propose that the parole board be empowered to more accurately evaluate costs and benefits of release and retention, at the same time, l will propose that other bodies determine the policy questions of how much risk is too much.

I will propose two contradictory policies as means of exploring how these normative decisions might otherwise be reached. The first policy would involve a systematic, actuarial approach wherein legislation or regulation could more clearly set out the parameters of release and create mechanisms to ensure that these values drive parole release. Population or percentage targets, for example, would force individual parole decisions into a broader population context. These individual decisions would no longer result in unforeseen aggregate consequences; instead, the system as a whole would control the prison population more in line with system- and society-wide costs and benefits.

The second policy looks through the other end of the telescope, replacing actuarialism with a more normative, if less scientific, body: the jury. This would provide for greater normative legitimacy at the cost of transparency and systemic management, but it might also better reflect the idea that what constitutes a risk worth taking might be so situational that we can only be guided by standards. That is, the real issue in parole release might properly be framed as one of process, not outcome, and that, in the absence of social agreement on particular normative values, we might want to use the jury as a means of hashing out those values in each case as it comes along.

These suggestions are not meant as concrete policy proposals; after all, it would be foolish to suggest completely incompatible ideas. Rather, ! hope these suggestions frame different aspects of parole, illustrating how different goals might require different release (or retention) mechanisms. In some ways, each proposal tests a hypothesis about whether parole is inherently about risk or inherently about desert, or whether it is irreducibly about both.

The Article will proceed as follows. Part I explores ways in which we might better contextualize risk by looking at the costs and benefits associated with various outcomes. I conclude that California's failure to account for the benefits of parole release and the costs of parole denial results in denial decisions that do not accurately reflect social utility. If the costs of risk reduction are not part of the decision, then as long as an inmate poses any risk, he or she will be denied parole.

In Part II, I will explore the two contradictory proposals that seek to vest the normative aspect of parole release--who should be released--in the legislature or in a jury of citizens. Ultimately, these two proposals illustrate that parole mechanisms are tied to the purpose of parole, and that risk and desert in the parole context are not easily separated.

I. RISK ASSESSMENT REQUIRES COST-BENEFIT ASSESSMENT

It seems obvious that deciding whether to reduce risks without evaluating the costs and benefits of reducing those risks is illogical. After all, society has a finite amount of resources and an infinite number of uses for them. Subpart A looks at the California parole equation and points out that it does not account for costs and benefits of the release/retention calculation. Subpart B suggests ways in which the costs and benefits of parole might be more accurately accounted for.

A. California Parole Boards Consider Risks, Not Costs and Benefits

California regulations require that parole boards evaluate whether an offender "will pose an unreasonable risk of danger to society if released from prison." (4) The board is to consider "[a]ll relevant, reliable information," but the enumerated examples of this information in the code of regulations--such as the offender's social history, criminal history, and commitment offense--are centered entirely around the individual prisoner. (5) There is no suggestion that the board look explicitly at whether this individual prisoner poses a greater or lesser risk than other prisoners. Officials are also not directed to look at the costs and benefits of continued incarceration; they are only directed to evaluate the risks of release. Even though the governing statute indicates that release should be presumptively granted (6)--which might be seen, in some ways, as a judgment that the average case will not present unreasonable danger--the presumption does no work, and few eligible prisoners are released. (7)

Without considering the benefits of granting parole, there is no incentive for parole boards to vote in favor of release. The retention cost borne by the parole board is zero: parole boards are not rewarded for any of the social benefits of release and are not penalized for the social cost of continued incarceration. As long as the parole board's retention cost is zero, any potential release risk greater than zero will justify retention. At the same time, release potentially imposes significant reputational and political costs to the board. Parole board commissioners are appointed by the governor, and released parolees who kill are not good news for gubernatorial incumbents. The long-term costs of a swollen prison system are, of course, borne primarily by the governor's successors, particularly in states with term limits. (8)

On the other hand, California recently took a step towards aligning the costs of retention with social costs and benefits, passing a statute enabling the board to grant "medical parole" for prisoners with huge medical bills. (9) The legislation's sponsor, Mark Leno, specifically identified the opportunity cost of keeping seriously ill inmates in prison, declaring that he "would rather keep 100 school teachers employed than continue to waste millions of taxpayer dollars on incarcerating 10 severely incapacitated inmates." (10) But even this tentative step applies only to extremely costly inmates who pose almost no public safety risk: only a prisoner who is "permanently medically incapacitated with a medical condition that renders him or her permanently unable to perform activities of basic daily living, and results in the prisoner requiring 24-hour care" is eligible for release, and even these incapacitated prisoners must nevertheless "not reasonably pose a threat to public safety." (11) The new legislation cannot be seen as part of a larger discussion about the cost of risk reduction; it simply accounts for cases where the cost is prohibitive and the risk is infinitesimal. The truth is that Leno's analysis applies more broadly: more money for prisons always means less money for other things. The difference is that discussing the incarceration cost of prisoners who can walk or feed themselves is, politically, too hot to handle. (12)

My initial suggestion, then, is a modest and obvious one: that the parole board consider the net costs of imprisonment versus release as a factor in parole eligibility decisions. The costs of imprisonment in California run approximately $49,000 a year. (13) This figure does not include other costs of incarceration, such as the value of a prisoner's liberty and the potential value a returning prisoner might contribute to the economy, to his family, or to his community. Using cost to inform criminal justice decision-making is not as novel as it might seem. Missouri presents its judges with information on how much a given sentence costs, and provides comparable figures for alternative sentences (e.g., prison versus probation). (14) This move has been criticized as overly "mathematical," (15) but it acknowledges, at its root, that cost is part of the calculation about whether punishment is worth it.

B. Problems With Calculating Risks, Costs, and Benefits

Including the costs of imprisonment in a cost-benefit analysis is an obvious step, but these costs are only one part of the equation. To more accurately assess the costs and benefits of various sentencing choices, we must account for the fact that there are a number of possible outcomes, and that some are more likely and/or more costly than others. Tort law prices the cost of risk most famously through Judge Learned Hand's Carroll Towing formula: "if the probability be called P; the injury, L; and the burden, B; liability depends upon whether B is less than L multiplied by P: i.e., whether B [is] less than PL." (16) In parole decisions, then, we would want to multiply the probability (P) of reoffense by the loss (L) caused by that offense, and determine whether the result was greater than the burden (B) of continued incarceration.

Filling in values for Hand's variables is more complicated than it seems, however, given that there are a multitude of possible outcomes. There is no standard definition of recidivism: the term can refer to criminal activity, rearrest, or reconviction. Simply knowing that the risk of recidivism is seven in ten says nothing without knowing how recidivism is defined. An ex-offender might be considered a nonrecidivist if he or she merely avoids conviction, merely avoids arrest, or only if he or she desists from any criminal activity. The underlying type of crime is, typically, not disaggregated from recidivism statistics, which makes the calculation of probability and loss more difficult. Just as there's a difference between a car with a failed brake light and one with an explosive gas tank, so too is there a difference between reoffending by smoking pot and reoffending by killing someone. The problem is that discussions about recidivism tend not to acknowledge that potential recidivist offenses might be less serious than the offense that landed the offender in prison. Sex offender recidivism statistics, for example, can measure more than recidivism for future sex offenses: gross figures for recidivism might include other, nonsexual offenses. (17) The ultimate point is that the kind of reoffending matters. We should take care to be sure that the sentenced offense doesn't serve as an anchor for the conversation about risk, and if the "recidivism" rate includes lesser offenses, we should disaggregate those risks from the risk of more serious offenses. (18)

Assuming that we could isolate the risks of various outcomes from one another, we would then have to assign each outcome a loss value. It is relatively easy to assign a cost to property crimes--one might start with the value of the stolen item, and perhaps include things like the deadweight loss from increased insurance premiums or a decrease in commercial activity in areas with high theft rates--but it is much more difficult to assign values to violent crimes. How can one value the avoidance of rape--by what someone would pay to avoid it, by the costs of treatment, or by estimating the value of psychiatric torment? Does the cost of murder include just the cost of the victim's foregone lifetime earnings, or does it include the suffering of the victim's family and the community's loss of security? Pricing crime is difficult. Markets use price to convey value, but there is no efficient market for rape and murder avoidance, and the "goods" involved are not standardized. There is, to put it mildly, a lack of consensus on the value of human life and well-being.

There are other problems with making cost-benefit analysis work in the parole context. The nature of parole failures makes risk calculation particularly difficult. A released prisoner who commits a heinous crime presents an easy picture of cause, effect, and the means to avoid future harms, and this picture is more easily grasped and correspondingly more heavily weighed than problems whose causes are more complicated. (19) Overpopulated prisons are also a failure, but because the effects of not releasing relatively safe prisoners are more complicated and harder to grasp, people are less likely to see retention as a problem, even if the harms to the state are ultimately greater than those posed by parolees. (20)

It isn't entirely clear, of course, that the analysis in tort law is really that similar to that of parole risk assessment. Part of the justification for risk analysis in tort law is that it incentivizes the reduction of risk at the socially optimal cost. Products are made by repeat players who can continue to refine their manufacturing processes to the point where the costs of avoidance are in equilibrium with the costs to society. Getting the pricing right will necessarily align individual incentives with social benefits, leading to safer products at exactly the price society is willing to pay for them. In the prison context, however, each "product" is a unique person who is not a repeat player, and the "manufacturers"--the parole board and prison officials--do not profit from reduced costs. It is not clear that getting risk right will result in a greater degree of utility for everyone, absent efficient markets for rehabilitated prisoners.

Tort law is also retrospective, while parole evaluations are prospective. An evaluation of negligence and the harm of that negligence is at least somewhat concrete. That is, while it might be difficult to assess the cost of the harm--the value of lost wages, etc.--we have an idea of who suffered the harm and who should pay for the harm. In the parole context, however, we are not only dealing with the difficulty of assessing the values of various outcomes, we're doing so without knowing what the various outcomes might be, who might suffer, or how much they might suffer. We are dealing only with probabilities, not results. Finally, criminal law tends to account for costs in a moral sense, not an economic one. Although estimates of the social cost of crime tend to use mechanisms similar to those used in tort (foregone wages, property damage, and the like), criminal victim restitution statutes typically assign a value based on the offense, not the economic characteristics of the victim. They are more like fines and less like compensation.

All of these factors might nevertheless be accommodated in the current system by simply folding them into the concept of whether release is "reasonable." But part of the problem is that standards are enforced mostly through appeals courts giving guidance to decision-making bodies (in most cases, trial courts, but also bodies like parole boards and administrative tribunals). Standards gain traction primarily when appellate courts discipline tribunals that do not follow them. United States v. Booker established that federal criminal sentences have to be reasonable, but federal appeals courts have helped define what is or is not a reasonable sentence by accepting or rejecting various sentences. (21) Currently, though, state review of parole board decisions is deferential in the extreme: state courts review only for "some evidence" (22) and cannot order release. (23) Moreover, prisoners denied parole are not entitled to publicly financed appellate counsel, so a prisoner with a strong case might nevertheless not have the resources to make that case in state court.

II. NORMATIVE ELEMENTS OF PAROLE RISK

Cost-benefit analysis can help isolate the costs and benefits of a given choice, but they cannot answer one key question: whether the costs are worth the benefits. What drives decisions about crime is never solely cost-benefit analysis. There is, invariably, a normative component to it as well. In the parole context, we can know levels of risk and their cost, but the question we need to answer is whether these costs are worth the benefits. This is a separate question from that about the accuracy of actuarial forecasts, and it is also separate from the statutory and procedural mechanisms states used to make release easier or more difficult, (24) neither of which I address in this Article.

An acceptable level of risk would mean, somewhat tautologically, that we accept the fact that some parolees we choose to release will reoffend. That is, if a parole board predicted ex ante that one in fifty offenders would commit another crime, (25) subsequent evidence that one in fifty released offenders committed a crime would not be treated as evidence of systematic failure. The resulting crime would, instead, be the price we paid--and were willing to pay--given the costs and benefits associated with release.

This is not a conversation society is having. We tend to discuss the crime level as something that should, optimally, be zero, and we do so without insisting that the rate of illiteracy or traffic accidents should also be zero. Public safety is not an optional budget item; other kinds of risks are. (26) Each crime committed by a parolee is seen as both avoidable and worth avoiding. These crimes are, of course, avoidable in one sense: if no one is released on parole, no parolees will commit crimes. The problem with this line of thinking is that it avoids the costs of nonrelease--the financial and human costs of incarceration--and fails to account for the marginal crime-reduction benefits in sufficient detail. Put another way, eliminating crime is probably impossible and certainly prohibitively expensive when it concerns nonparolee populations. With parole populations, however, we can bar release, avoiding any risk of reoffense while avoiding the larger conversation about whether this practice is worth it.

Normative discussions tend to result in more questions than answers, and they inevitably point to the ways in which risk and desert are intertwined. On a pure risk basis, we might want to release the safest prisoners, but we might feel uncomfortable letting out a safe prisoner who committed a heinous crime. We might feel more sympathy towards inmates who have done the most to reduce their risks, but still release those who pose the least risk, even if they did nothing but sit on their hands while in prison. (27)

In any of these evaluations, case law already treats repeat offending as both an actuarial and a moral factor. (28) We treat people who have already offended--those in prison--differently from those who have not yet offended, even if the risks they pose are similar. (29) Recidivism is, in some ways, a hybrid, not just a sign of risk and not just a sign of moral failing. We mix and match risk and retribution when it comes to various detention and crime policies. (30) Questions about risk are invariably influenced by questions about desert, and it is hard to draw clear and clean lines. (31)

Given that there are value judgments implicit in risk decisions, the next question should be who is authorized to make them. California regulations require the parole board to determine whether, in its judgment, "the prisoner will pose an unreasonable risk of danger to society if released from prison." (32) California has vested parole board officials with the power to make these normative judgments for us about whether a given risk is worth the cost. This is a bad idea. A parole board is good at evaluating risk, but no one would argue that parole boards have great moral legitimacy, or are somehow skilled at policymaking. (33) Parole boards are selected on the basis of expertise about criminal risk and rehabilitation; (34) it is this expertise that justifies traditional judicial deference. (35) Indeed, the existence of parole board regulations indicates that the board's role has already been circumscribed. A parole board's job is to implement policy, not create it, but without clear policy guidance, board decisions will themselves effectively establish state policy.

As a means of exploring these issues, I will make two proposals for circumscribing parole board decision making, each from a diametrically opposed perspective. The first takes what might be considered an actuarial approach, where a given parole determination is even less individualized than it already is. Boards would be told to release a certain number or percentage of people; their job would only be to figure out who would be best to release throughout the system as a whole. The second approach takes a completely individualized approach. Here, release would explicitly include questions of desert, but the decision-making body would have the legitimacy to make those kinds of nonnative judgments. In this case, then, the factors which bear on parole release decisions might not be that different from what they are today. The only difference would be that a jury would make those decisions, not a parole board. Both of these are thought experiments--pointing out how parole partakes of both risk analysis and desert. I will address each in turn.

A. Parole Release by the Numbers

In this Subpart, I will consider the implications of setting, by regulation or legislation, a set percentage or population target of prisoners that the parole board would implement. The board would use its discretion and expertise to determine only which prisoners were safest to release, not how many should be released overall.

Moving in a purely actuarial direction would eliminate parole board policymaking. As I have shown, the current regime's individual, case-by-case model has resulted in a system where almost no one gets released. The de facto policy is no parole, even though this is not a policy that has been endorsed by the legislature or any other representative body. (36) Arguably, this policy hasn't even been endorsed by parole board members themselves; it is simply the result of individual parole suitability determinations.

If the legislative or executive branch set annual targets of parole release, the board would then simply find the safest parole-eligible prisoners to release in order to meet those targets. The parole board would no longer need to justify what it considers reasonable risks of danger. It would bear some of the "costs" of non-release in the individual case--knowing that another prisoner might pose greater risks but need to be released in order to meet that term's population target. With a target, the parole board would evaluate whether a given prisoner was safer than others, not whether he or she was, in some absolute sense, safe.

A prison system which knows a given percentage of offenders would have to be let out might shift resources towards decreasing risks among its prison population. That is, if the parole board's hands were tied and it had to release a given percentage of eligible prisoners, any blame might shift towards the institution as a whole. The question might then change from "why did the parole board let this person out?" to "why, given that twenty percent of the people eligible for parole were going to be released, did prison officials not do more to make sure that at least twenty percent of prisoners were ready for release?"

The California prison system as a whole has been ordered by a panel of three federal judges to reduce overcrowding; given the lack of excess capacity in the system, this order, recently upheld by the U.S. Supreme Court, requires releasing prisoners. (37) The overcrowding order can be criticized on the grounds that judges have less democratic legitimacy than a state's elected officials, but the same is certainly true of parole commissioners. (38) If the legislature or executive were to set population targets, however, the targets would have greater legitimacy.

There is, ultimately, nothing inherently right or wrong about a given target for release. It could be set at any level--including zero. But if elected officials were to set this number, it would at least have greater transparency and legitimacy. We might choose to continue to incarcerate all parole-eligible prisoners, but we would at least be doing so openly. We could continue on our current course to bankrupt the state university system, but we would at least be freely choosing to do so via policies established by our representatives. We would still be driving off a cliff, but at least we would be doing so with our eyes open. In other words, the process by which we reach the result is important. We could continue on our present path, but it would be tied to a managerial, systemic decision, not an amalgamation of individual parole release decisions.

In short, I am not making a normative argument that we should have a given policy, or that we should release a given number of prisoners. Even if parole boards have little normative authority, they certainly have more than law professors do. I am, however, suggesting that we should think more clearly about what our decisions might mean, both in terms of the resources they use and the purposes they serve. If we are really serious about parole as a measure of risk, we need to give that decision some teeth. We need to force some consideration of costs into the system, whether it's via a population or percentage target, or simply through a requirement that prisoners with a given risk score be released.

This systemic approach might discard what's best about indeterminate sentencing, however: that it is unique to every sentenced individual. Indeterminate sentencing is nothing if not an acknowledgement that one sentence does not fit all; offenders are going to be ready to reenter society at different times. So perhaps each individual offender is so unique that offenders as a whole can't be aggregated into a system-wide numerical goal. I consider the implications of this view in the following Subpart.

B. Individuation Requires Individuals: The Parole Jury Proposal

My second proposal starts with this basic insight: that each offender, and each indeterminate sentencing decision, is unique. I will acknowledge that release decisions have wiggle room; they do not admit of error-free solutions with agreed-on goals. Instead, parole release decisions are messy. For every rule, one can come up with an exception. Desert entwines itself in the factors that make up parole release decisions, and desert is considered not in the abstract--whether this kind of person deserves release--but in the concrete--whether this particular person deserves release.

The assertion that every case is unique contradicts the very notion of actuarial risk--the idea that situations average out across relevant cohorts. Again, I suggest this proposal, which is inconsistent with and orthogonal to the prior suggestion, simply to illuminate what each policy choice might mean. My job is not to prescribe, but to diagnose. Social goals need to be set by society; I am merely attempting to analyze the meaning, impact, and significance of these potential choices.

A position that every offender is unique does not mean that decisions cannot be made; it simply means that rather than attempting the frustrating task of answering questions about release for all people at all times, in an individual case we need only find a particular answer for a particular set of facts. As unsystematic as that might seem, we already have a mechanism for sorting through messy issues which combine facts and judgments, a mechanism that applies rules but which also deals with exceptions to those rules. That mechanism is the jury.

A jury can deal with facts, but it can also deal with norms. It has the legitimacy to consider not only "did" or "didn't" but also "should" or "shouldn't." And even though juries are imperfect representatives of the "conscience of the community," (39) they are at least more representative than the appointed members of the parole board. Having juries decide questions of parole release would acknowledge that parole is something where we can only do so much: we can talk about it, we can look at the evidence of risk, and then we make a decision that we feel is right.

My proposal is not intended as model policy, but more as a thought experiment about the limits of actuarialism. (40) This jury might be more akin to a grand jury than a civil or criminal jury. Grand juries sit for a while and make similar probabilistic assessments (e.g., whether there is "probable cause" that the defendant committed the crime and therefore that society and the defendant can incur the cost of a criminal trial). The parole jury might sit in on multiple cases, hearing the first few as nonvoting observers before sitting in judgment for a series of release hearings. They could and should be presented with information about risk and about proportionality of sentences. They could determine whether releasing an individual would be "worth it," with all the vague, value- and policy-laden implications that phrase entails.

Again, this mechanism does not have a particular valence to it--indeed, juries might decide that fewer people deserve to get out than do at present. Perhaps what I see as overly conservative parole board decisions are, in fact, liberal compared to what a jury of one's peers might decide. But the use of a parole jury would transparently acknowledge that some of the questions relevant to parole are about policies and values. A parole board making up policies about desert does not have legitimacy, but a jury can and does now consider questions of desert in individual cases. A parole board--or a series of regulations--can never pin down exactly what makes someone deserve more or less time in prison, but a jury can take those factors, hash it out during deliberation, and come up with a decision that we can accept. In short, a jury has the freedom and the legitimacy to openly make decisions based on factors that are not actuarial.

In fact, I do not think it is a foregone conclusion that juries would be unwilling or unable to consider arguments from the prisoner's point of view, and we could certainly provide them information to aid in their decisions. We could ask them to think in terms of whether it is more important to keep this person in prison for another period of years or whether we should use that money to hire another teacher, fix another road, hire another police officer, or lower taxes. But my predictions are, again, not the point. The choice of a parole jury is not about outcomes, but about the legitimacy, purpose, and meaning of the process of parole suitability.

Getting people to serve on parole juries might have several other benefits as well--again, benefits that are unrelated to their effects on prison populations. First, it might finally shed some light on what goes on behind prison walls. Prisons have increasingly moved toward rural areas of the country; most prisoners are literally uncounted in the census and other kinds of government statistics. Sending ordinary citizens inside prisons might do nothing more than put prisons and prisoners--on the radar a bit more, and might minimize the fact that prisoners today are typically "out of sight, out of mind." Second, the possibility of release through a parole jury might make release itself more meaningful. It would be a way of formally recognizing that an individual's debt to society has been paid. This would mean that the offender was not only accepted as a risk worth taking, but accepted as one who has been deemed to have paid his or her debt to society. Too often civil, postrelease sanctions present the opposite message about just how "out" an offender is, by continuing to stigmatize him or her. (41) An official approval by a body of one's peers might solemnize the end of prison and indicate an official passage from prison back into society.

But even appeals to words like "society" and "community" raise important, and material, questions about how those terms are defined. From what body would a jury venire be taken? Assuming parole is about reintegration, it might make sense to take a jury venire from the community to which the offender would return. If parole is about evaluating just punishment, however, it would make more sense to take the jury from the place where the crime was committed. (42) Changing the process also begs questions about whether parole is, or should be, adversarial. Is the point of parole to have advocates on one side or another speaking about the costs of release or the costs of retention, and, if not, how should information about those costs be developed? Should the victim's family provide most of the information, and should the prisoner's family counter it'? Should the district attorney stand in for the victim and the public defender for the family? It's unclear what the answers to these questions should be, because parole isn't exactly like a trial and it isn't exactly unlike it, either.

These proposals, then, are not meant as model legislation. My intention, instead, is to use them to reveal questions that exist--but which are hidden--by our present use of the parole board. Our current system does not answer these questions so much as hide them. We have thoughtlessly delegated their discussion to the parole board. Whether or not we choose to deal with these questions, they are not going anywhere. A jury is one way for society to take on the responsibility for grappling with these issues ourselves. We can learn from the experience. We can be humbled by it. And we can make the outcomes of these decisions more meaningful, because we are the ones who have made them,

CONCLUSION

The problems with parole release could easily be solved by simply doing away with indeterminate sentencing, but indeterminate sentencing has a great upside. We can let people out when they're ready to go, and we need not let people out before they're ready. Determinate sentencing certainly ensures that people serve the minimum, but it treats them all the same. Under determinate sentencing, a prisoner might walk out of solitary confinement with fully untreated antisocial behaviors and onto the street, or she might he fully rehabilitated and stay in prison until her time has been served. Indeterminate sentencing recognizes the fact that individuals are going to react differently to prison, and that they have different risk levels.

Indeterminate sentencing necessarily involves some discretion, and that discretion in California has resulted in almost every prisoner being treated identically: as a risk not worth taking. Because any individual can be someone who turns into a parole nightmare, and because there is nothing an individual parole board member gains from successful releases, there is no incentive for release. Put another way, no one cheers when an ex-offender becomes a taxpayer. The only time we hear about parolees is when they've done something wrong. Part of my aim in writing this Article, then, has been to start an honest conversation about risk, because it's clear that we tend to treat all criminal risk as worthy of reduction: any amount of money is worth spending to reduce crime risks. We don't feel the same way about subsidized school lunches, the foster care system, mental health care, or subsidized drug treatment, even though failing to spend money on those issues generates huge social costs and needless suffering.

But risk analysis can ultimately only inform our decisions, not make them for us. While the calculation of the costs and benefits associated with risks can be improved, in the end, decisions based on risks involve value judgments. I have sought to highlight the actuarial and normative components of parole release through two contradictory proposals, each of which focuses on one element of parole release. It is my expectation that neither one feels complete not just because of their lack of specificity, but because the ideas behind them leave something important out of what we think about parole. My hope, then, is to raise questions about what parole is, and to sketch out the ways in which the purposes of parole, whatever we decide they are, determine the tools of parole. I do not provide those answers in this Article; I think answers that do not come as a result of social deliberation will be as illegitimate as those presently delegated to the parole board. Whatever our answers are, however, they are surely too important to delegate to the parole board or to a law professor without a clearer understanding of--or justification of--that delegation. No matter the risks involved in having this discussion, the benefits surely outweigh the costs.

(1.) For a discussion about other uses of the term indeterminate, see W. David Ball, Heinous, Atrocious, and Cruel: Apprendi. Indeterminate Sentencing, and the Meaning of Punishment, 109 COLUM. L. REV. 893. 906-09 (2009). This analysis draws heavily on Steven L. Chanenson, The Next Era of Sentencing Reform, 54 EMORY L.J. 377. 382-83 (2005).

(2.) CAL. PENAL CODE [section] 3041(a)-(b) (West 2010).

(3.) Sasha Abramsky, Barred.for Life, SF WKLY., Aug. 15, 2007, at 16, available at http://www.sfweekly.com/2007-08-15/news/barred-for-life/. These figures might be on the increase after In re Lawrence, 190 P.3d 535, 555 (Cal. 2008), which prohibited the parole board from relying exclusively on the commitment offense to deny parole.

(4.) CAL. CODE REGS. tit. 15, [section] 2281 (a) (2010).

(5.) Id. [section] 2281(b).

(6.) This presumption takes the form that the board "shall set a release date unless" it determines that, in the interest of public safety, release cannot be granted. CAL. PENAL CODE [section] 3041(b) (West 2010).

(7.) For a detailed discussion of just how few prisoners are released, see Ball, supra note 1, at 918-19.

(8.) The victims of parolees' crimes will usually be louder and better organized than the rest of us who enjoy the tax benefits of having fewer prisoners. As a result, the parole board, so long as it is at least indirectly politically accountable, will systematically under-release prisoners according to political criteria that are divorced from an actual cost-benefit calculation.

(9.) See Act of Sept. 28, 2010, [section] 2. 2010 Cal. Legis. Serv. ch. 405 (West).

(10.) Press Release, Senator Mark Leno, Assembly Approves Medical Parole Bill Authored by Senator Leno (Aug. 30, 2010), available at http://dist03.casen.govoffice.com/index.asp?Type=B_PR&SEC={FODFD1A5-1C7B-4F09- 9F09C48A423D1072}&DE=(4A9039EB-D8D1-492D-B3E5-20C515BFFCAB}.

(11.) Act of Sept. 28, 2010, [section] 2, 2010 Cal. Legis. Serv. ch. 405 (West).

(12.) See, e.g., W. David Ball, Time for a Cost Benefit Analysis on Crime, DALLY JOURNAL, Oct. 15, 2010, at 5.

(13.) CAL. DEP'T OF CORR. & REHAB., FOURTH QUARTER 2008 FACTS AND FIGURES, available at http://www.cder.ca.gov/Adult_Operations/docs/Fourth_Quarter_2009 _Facts and Figures.pdf. These costs can be greater if an inmate spends time in solitary confinement. See Laura Sullivan, At Pelican Bay Prison. a Life in Solitary, NAT'L PUBLIC RADIO (July 26, 2006), http://www.npr.org/templates/story/story.php?storyId=5584254 (estimating the additional cost of solitary confinement in California at $50,000 per year): see also COMM'N ON SAFETY & ABUSE IN AMERICA'S PRISONS, CONFRONTING CONFINEMENT 54 (2006), available at http://www.prisoncommission.org/pdfs/Confronting_Confinement.pdf (finding costs of solitary confinement in Wisconsin are roughly double that of maximum security incarceration). Costs of parole vary with the level of supervision, but they are much less than the cost of imprisonment. In 1998, the California Legislative Analyst's Office estimated that the annual cost of parole per parolee was $2,159. CAL. LEGISLATIVE ANALYST'S OFFICE, ANALYSIS OF THE 1998-99 BUDGET BILL D-II (1998), available at http://www.lao.ca.gov/ analysis_1998/pdfs_an198/crim_justice_an198.pdf. The comparable cost figure for prisoners in 1996-1997 was $24,807. BUREAU OF STATE AUDITS, CALIFORNIA DEPARTMENT OF CORRECTIONS: THE COST OF INCARCERATING INMATES IN STATE-RUN PRISONS IS HIGHER THAN THE DEPARTMENT'S PUBLISHED COST I (1998), available at http://www.bsa.ca.gov/pdfs/reports/97125.pdf.

(14.) Monica Davey, Missouri Tells Judges Cost of Sentences, N.Y. TIMES, Sept. 19, 2010, at Al, available at http://www.nytimes.com/2010/09/19/us/19judges.html. Missouri's cost prediction system is available for anyone to try at https://www.courts.mo.gov/rs/. Since 2005. Missouri has also given its judges information on the median sentence of offenders charged with similar offenses who present similar scores on risk assessments. This information is not binding, but it is meant to provide judges with a sense of what would be a "normal" sentence, and what would be an extreme deviation up or down. See Michael A. Wolff, Missouri's Information-Based Discretionary Sentencing System, 4 OHIO ST. J. GRIM. L. 95, 98, 116 (2006).

(15.) Davey, supra note 14.

(16.) United States v. Carroll Towing Co., 159 F.2d 169, 173 (2d Cir. 1947).

(17.) For an accessible discussion of the scholarly research, see Carl Bialik, Underreporting Clouds Attempt to Count Repeat Sex Offenders, WALL ST. J., Jan. 25, 2008, at B1, available at http://online.wsj.com/article/SB120122376053515485.html.

(18.) But, of course, this further begs the question of what differentiates serious offenses from non-serious offenses. In the interest of organizational clarity, however, I will return to all of these implicitly normative questions about nominally objective assessments in Part II.

(19.) For a general discussion of risk assessment, including the ways in which newsworthiness distorts risk assessments, see CASS SUNSTEIN, RISK AND REASON: SAFETY, LAW, AND THE ENVIRONMENT ix-xii (2003). See also Sara Sun Beale, The News Media's Influence on Criminal Justice Policy: How Market-Driven News Promotes Punitiveness, 48 WM. & MARY L. REV. 397 (2006).

(20.) Even Supreme Court Justices are liable to conclude that any risk is worth avoiding. During oral argument in Coleman v. Schwarzenegger, discussed infra note 37, Justice Alito was dissatisfied with the idea that a prisoner who presents a seventeen percent risk of recidivism should be released:

   [T]he 17 percent figure goes exactly to my concern. This is going
   to have--it seems likely this is going to have an effect on public
   safety. And the experts can testify to whatever they want, but you
   know what? If this order goes into effect, we will see ... We will
   see, and the people of California will see: Are there more crimes
   or are there not?

Transcript of Oral Argument at 68-69, Brown v. Plata, No. 09-1233, 2011 WL 1936074 (May 23, 2011).

(21.) 543 U.S. 220, 261-62 (2005). Booker is also instructive because it shows how enumerating reasons for deviating from presumptive sentences might also reinforce certain standards.

(22.) See Ball, supra note 1, at 958-61.

(23.) See, e.g., In re Prather, 234 P.3d 541, 550-51 (Cal. 2010).

(24.) Such mechanisms include varying the presumption of parole release, the burden of proof, and the standard of proof in parole hearings.

(25.) Actual figures are much higher, but i present this low number to make a point: even a one in fifty risk might still be seen as unacceptable for reasons I will explore in this Article. As for the actual numbers, a 2008 report noted that

   California's recidivism rate as measured by the "return to prison
   rate" is 66 percent, compared to a 40 percent national average. At
   the end of three years, 66 percent of all California parolees had
   been returned to a California prison, 27 percent for a new criminal
   conviction and 39 percent for a technical or administrative
   violation, which can result from new crimes or violations of the
   conditions of parole. On any given day, six out of ten admissions
   to California prisons are returning parolees.

Ryken Grattet et al., Parole Violations and Revocations in California (Oct. 13, 2008) (unpublished report), available at http://www.ncjrs.gov/pdffiles1/nij/grants/224521.pdf. Changes have been made to parole revocation procedures, which might affect the rate of return going forward. For more information, see Why CDCR Developed a Parole Violation Decision Making Instrument (PVDMI), CAL. DEP'T OF CORR. & REHAB., http://www.cdcr.ca.gov/PVDMI/index.html (last visited Nov. 7, 2010).

(26.) But see the recent discussion about police layoffs in Oakland, which does balance the costs of policing versus the benefits of other services, in Lauren Callahan, Oakhmd City Council Weighs Budget Cuts, Police Officer Layoffs, OAKLANDNORTH (June 16, 2010, 8:09 AM), http://oaklandnorth.net/2010/06/16/oakland-city-council-weighs-budget-cuts- police-officer-layoffs/.

(27.) Given the kinds of factors that go into risk assessment, one author has argued that risk assessment embeds racism. See Bernard E. Harcourt, Risk as a Proxy for Race (Univ. of Chi. Law Sch. John M. Olin Program in Law and Econ. Working Paper No. 535 (2d Series), 2010), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1677654.

(28.) See, e.g., United States v. Rodriquez, 553 U.S. 377, 385 (2008) (explaining that recidivism bears on both seriousness of offense and public safety).

(29.) Consider two people, a prisoner eligible for parole and someone not in prison. If we knew that the person not in prison had a seventy percent chance of committing a violent crime and the person inside prison had only a thirty percent chance of doing so, should we incarcerate the outsider and release the insider? Should we do so even before the outsider has committed a crime? What if the insider was incarcerated for murder--or for a nonviolent property offense?

(30.) I have discussed at length the ways in which offense seriousness is a proxy for risk in contexts such as bail and sex offender registries. See, e.g., W. David Ball, The Civil Case at the Heart of Criminal Procedure: In re Winship, Stigma, and the Civil-Criminal Division, 38 AM. J. CRIM. L. (forthcoming Apr. 2011) (manuscript at 57-59), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1628950&http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=1628950.

(31.) Cf. Jonathan Haidt, The Moral Emotions, in HANDBOOK OF AFFECTIVE SCIENCES 852, 852-53 (Richard J. Davidson et al. eds., 2003) (arguing that both reason and emotion guide human morality).

(32.) CAL. COOL REGS. tit. 15. [section] 2281(a) (2010).

(33.) For a fuller discussion see Ball, supra note 1, at 935. Parole commissioners are appointed by the Governor. CAL. PENAL CODE [section] 5075(b) (West 2010). Parole commissioners share their duties with deputy commissioners, who receive civil service appointments. See Deputy Commissioners, CAL. DEPARTMENT CORRECTIONS & REHABILITATION, http://www.cdcr.ca.gov/BOPH/deputy_commissioners.html (last visited Nov. 7, 2010). But the politically appointed commissioners have greater power than the civil-servant deputy commissioners. A maximum of one deputy commissioner can serve on each two-person parole suitability panel. CAL. PENAL CODE [section] 3041(a) (West 2010). In the event that votes split on the panel, the full Board of Parole Hearings meets en banc to consider the matter; at this session only commissioners may vote. Id. [section] 3041(e).

(34.) See, e.g., CAL. PENAL COnE [section] 5075.6(b) (West 2010) (establishing qualifications and training). "Insofar as practicable, commissioners and deputy commissioners shall have a varied interest in adult correction work, public safety, and shall have experience or education in the fields of corrections, sociology, law, law enforcement, medicine, mental health, or education." Id. [section] 5075.6(b)(1).

(35.) See. e.g., Greenholtz v. Inmates of the Neb. Penal & Corr. Complex, 442 U.S. 1, 9-16 (1979) (discussing parole): see also Bell v. Wolfish, 441 U.S. 520, 547-48 (1979) (discussing deference to prison administrators more generally).

(36.) of course, the legislature's inaction might be read as tacit acceptance of this de facto no-parole policy.

(37.) Coleman v. Schwarzenegger, No. CIV S-90-0520 LKK JFM P, 2010 WL 99000 (N.D. Cal. Jan. 12, 2010), aff'd sub nom. Brown v. Plata, No. 09-1233, 2011 WE 1936074 (May 23, 2011). The Coleman case was combined with Plata v. Schwarzenegger, No. C01-1351 TEH.

(38.) See supra note 33 (describing the appointment of parole commissioners).

(39.) Spaziano v. Florida, 468 U.S. 447, 487 (1984) (Stevens, J., dissenting). For a more in-depth discussion of the jury as moral agent, see Ball, supra note 1, at 923-26.

(40.) For an expanded discussion of how the jury should be more involved in all aspects of sentencing, see Jenia Iontcheva, Jury Sentencing as Democratic Practice, 89 VA. L. REV. 311 (2003).

(41.) See Ball, supra note 30, at 32-38.

(42.) I note also that juries from the community of release might be more willing to vote in favor of retention: they will directly bear the costs of release, but they will share the cost of continued incarceration with the rest of the state.

W. David Ball, Assistant Professor, Santa Clara School of Law. The genesis of the parole/reentry jury idea came from a very insightful question from Professor Paul Diller during discussion of a prior article: I am sorry I have only now, years later, gotten around to providing at least a preliminary answer. My thinking about the normative components of risk benefited greatly from the discussion in Brian J. Ostrom et al., Nat'l Ctr. for State Courts, Offender Risk Assessment in Virginia: A Three-Stage Evaluation (Sept. 12, 2002) (unpublished report), available at http://www.ncjrs.gov/pdffiles1/nij/grants/196815.pdf. Many thanks to those who provided feedback on drafts of this Article: Kyle Graham, Deep Gulasekaram, Sebastian Kaplan, Michelle Oberman, and David Yosifon.