Thank you for inviting me here to discuss the American experience with the death penalty. I will focus today on the meaning and nature of errors in the imposition of capital punishment. Before turning to my main topic, though, I want to situate it in the larger debate over capital punishment. That debate has both utilitarian and normative components.
The utilitarian arguments for the death penalty are that it deters and that it responds to the understandable desire for retribution of those affected by homicides (capital punishment in the U.S. today is in fact only imposed for murder, although other capital crimes are on the books). The significance of deterrence is unclear. Statistical studies of capital punishment provide at most only weak support for deterrence as a consequence of executions, but the studies and underlying data are themselves quite problematic. (2) The consequences of capital punishment cannot be studied in anything remotely resembling controlled conditions; double-blind studies, for example, are a practical if not literal impossibility. Moreover, the most that could come from carefully controlled studies would be the conclusion that the death penalty as administered does not (or does) deter. But, the death penalty in the United States is limited to a very small set of all homicides. Since the mid-seventies, less than 300 people have been executed in the United States. Capital punishment remains a freakishly rare punishment. Even if, as administered, capital punishment has minimal deterrence effects, that may just as well be an argument for its increased rather than its decreased use. It is, accordingly, difficult to know what to make of the arguments from deterrence.
A second utilitarian consideration is the significance of capital punishment for the sense of justice of the secondary victims of homicides, such as the family and friends of the deceased. For example, an important, though perhaps neglected, aspect of punishment is the legitimation and vindication of personal feelings of vengeance. Officially sanctioned punishment channels the understandable emotions of the victims away from personal acts of retribution, and thus may help to preserve the peace. Even if the victims would not act on their impulses, punishment of wrongdoers may assist victims to cope with their feelings and to reestablish personal equilibria, just as it may on a larger scale respond to a community's sense of justice. Again, though, the extent to which capital punishment has these effects is difficult to say; more difficult still to appraise is their significance.
Arrayed against these possible benefits of capital punishment are significant costs, including the cost of errors, which is the main focus of this paper. Before turning to it, however, other costs of capital punishment should be noted:
1. The first cost to consider is the direct financial implications of capital punishment. Capital cases are costly to try, involve protracted appellate and post-conviction relief processes, and generally do not result in executions. The North Carolina Administrative Office of the Courts commissioned a study of the North Carolina experience with the death penalty:
One conclusion is that the extra costs to the North Carolina public of adjudicating a cse capitally through to execution, as compared with a noncapital adjudication that results in conviction for first degree murder and a 20-year prison term is about $329,000, substantially more than the savings in prison costs, which we estimate to be $166,000. We note that a complete account must also include the extra costs of cases that were adjudicated capitally but did not result in the execution of the defendant. All told, the extra cost per death penalty imposed is over a quarter million dollars, and per execution exceeds $2 million. (3)
2. There is good reason to believe that the existence of capital punishment has distorting effects on both substantive and procedural law. Many judges are hostile to the death penalty and harbor justifiable concerns about the possibility of error. For both reasons, legal rules may be subtly modified to decrease the incidence of executions, but in many instances those now modified rules cannot be limited to capital cases.
3. Efforts to limit capital punishment has generated rather explicit lawless behavior in the lower courts. Perhaps the best example is the effort by the United States Court of Appeals for the Ninth Circuit in the Robert Alton Harris case to undermine the legitimate authority of the Supreme Court of the United States. (4)
4. Many believe that the death penalty is discriminatorily applied, although there is not good evidence that this is so. Blacks are overrepresented on death row, but poverty may be the real explanation. This most sophisticated examination of discrimination and the death penalty to date found only that the color of the victim, not the color of the perpetrator, had a significant effect on the outcome. Still, the belief of discrimination exists, which is a cost in its own right.
How these possible costs and benefits compare is very difficult to say. Consequently, the debate in the United States over the death penalty tends to be normative, and I quite frankly do not have a view on the moral issue. I spent a fair amount of my time over a decade assisting in the defense of capital cases, primarily through work on appeals at all levels of courts in the United States, from local intermediate appellate courts to the Supreme Court of the United States. I did so because of the intuition that each human being is the end product of the infinite development of the universe and in a small way brings light into the darkness through the development of consciousness and cognition. The thought of deliberately cutting short the glow of any individual light, no matter how small, no matter how distorted, was, and is, repulsive to me.
But as I assisted on cases, my initial revulsion at the thought of an intentional and deliberate execution was met with a revulsion emanating from the records of the trials I had to study. The records of some capital trials in the United States are documentaries of astonishing mayhem, complete indifference to the lives and welfare of other human beings, no matter how helpless or vulnerable, and torture of such cruelty that one feels that whoever committed these acts is not a human, and thus not deserving of being so treated. In a case in just tried in Chicago, a woman with three children was pregnant with a child fathered by a man who was the cousin of the defendant. The defendant, Jacqueline Annette Williams, wanted the child. The defendant, the father of the child (her cousin), and another man invaded the home of the pregnant woman. They murdered her, and either before or after her death removed the fetus from her womb with a dull knife. The deceased's ten year old daughter was stabbed to death. Her seven year old son was taken from the home, later made to drink iodine, was strangled with a cord and ultimately stabbed in the throat and killed. The third child, a one year old, was left unharmed at the scene. The defendant was recently sentenced to death, and it is difficult for me to say that she does not deserve to die. Chicago Tribune, Saturday, March 28, 1998, p. 1.
But the records of these cases always contain more than just the evidence of unthinkable cruelty. Few people, maybe none, are thoroughly and unwaveringly evil. And often, maybe always, their backgrounds are explanatory to some extent of their behavior. Thus, always for me, after this initial revulsion at the acts committed, invariably the metaphor of the light returned, and I was, and am, left with these incompatible thoughts in my mind. I thus concluded that, without regard to the morality of capital punishment, the State should execute no one without the most stringent processes, without, in other and somewhat crude but nonetheless expressive words, doing it right.
There are many components to doing it right, but I will only address one of them today--eliminating the risk of error. The risk of error looms over the capital punishment process in the United States with remarkable force, a force just barely insufficient to put a stop to the entire process, notwithstanding the obvious approval of capital punishment in our Federal Constitution. However, it is not obvious why this is so. Life and death decisions are made in every society everyday, and these decisions usually affect innocent, and often helpless, individuals. We know that somewhere around 30,000 people a year will die on our nation's highways, yet we keep building roads. Many of these will die as a consequence of the consumption of alcohol, yet we do not forbid it. Medical delivery decisions determine who will live and who will die. Allocations of scarce research dollars determine what diseases science will tame, and which will be put aside for now, again with real effects on innocent people. The idea that lives of truly innocent people cannot be sacrificed for a political conception of the greater good, at least in some meanings of those terms, is disproved beyond dispute by the lives that we actually live.
Nonetheless, ask yourself the question posed by Williams James, in The Will to Believe and other Essays in Popular Philosophy (1979, Harvard University Press, at 144). Imagine a world in which all human aspirations could be realized, "and then . . . imagine that this world is offered to us at the price of one lost soul at the farthest edge of the universe suffering eternal, intense, lonely pain." R.A Putnam, Perceiving Facts and Values, 73 Philosophy 5 (1998). Would we accept the bargain? James thinks not, that we would recognize "how hideous a thing would be its enjoyment when deliberately accepted as the fruit of such a bargain."
I think he is right, but I think our answer to his question stands in stark relief to our actual practices. What distinguishes the two, if anything does, is the personal identity of the sacrificed victim. In most of our social practices (but note, not all, such as medical triage), the victims are unidentified and not singled out personally for "hideous" treatment. The innocent convicted of a capital crimes looks to many to be more like James' victim, although it does depend on whether you look at the matter ex post or ex ante. Still, to many, the possibility of error, the possibility of executing an innocent person strikes many as being an example of James' Faustian bargain, of balancing the account books of the world on the back on an innocent person condemned to suffer and die for the benefit of the rest of us.
But, how does one reduce the probability of an innocent person being executed? This is the central problem. I will discuss here the answers given to this question by the Supreme Court of the United States to demonstrate just one proposition--that the Court, and all of its individual members--has had great difficulty in providing cogent answers to that question. I then provide an explanation of why that is so. In brief, the answer is because the Court is attempting to apply the normal approach of appellate decision making, which involves rules and deduction, to a situation that is not amenable to such an approach and calls instead for judgment. The awkwardness in the Court's numerous opinions all stem from the incompatibility between its methodology and the problem it is addressing. I will now explain.
To many, the U.S. Supreme Court's death penalty jurisprudence contains an apparently inconsistent set of decisions concerning the role of discretion in sentencing murderers to death. In the standard version of this story, the Supreme Court called a halt to executions because unguided discretion was resulting in capricious applications of the death sentence (5), subsequently reinstituted capital sentencing based on statutes guiding the exercise of discretion through the use of aggravating factors (6), and most recently has, in the view of one of its members, Justice Scalia, come full circle to only approving death sentences if the decision maker has virtually unfettered discretion to grant mercy, thus constitutionalizing the very problem of capriciousness that led to constitutional rulings in opposition to capital punishment. The case prompting Scalia's rather scathing assessment is Walton v. Arizona (7).
Walton claimed that the Arizona statute under which he was convicted was unconstitutional in its requirement that the defendant establish mitigating circumstances by a preponderance of the evidence. The Court rejected the claim, with a plurality of the Court concluding that the Constitution forbids excluding from consideration "any particular type of mitigating evidence," but does not forbid a state "from specifying how mitigating circumstances are to be proved." (8) The Constitution requires that all mitigating evidence be admitted, but does not control how juries are instructed to process it.
The central argument of the majority has many difficulties. The most salient problem with the plurality's opinion stems from a misguided faith in the conventional preponderance of the evidence rule. Presumably the goal in capital cases is to have the sentencer engage with whatever mitigating factors are present. A preponderance of the evidence rule is an extremely crude tool to accomplish that purpose. Suppose that a defendant advances three grounds for mitigation, each of which is established to a .25 probability, and thus not to be considered under Arizona's rule. The probability that at least one of these factors is true, assuming they are independent, is 1-.753=.58. (9) As more mitigating factors are advanced, or as they are proved to a higher probability, this point is exacerbated. For example, if three independent mitigating factors are each proved to a probability of .4, the probability that at least one of them is true is .78, yet the sentencer must decide as though there were no ground for mitigation. The plurality does not explain, and indeed it could not rationally defend, how a .78 probability that at least one of three mitigating factors is present differs from a .78 probability that a particular mitigating factor is present.
This is not the objection raised by the dissenters. Brennan (10) and Blackmun (11) argue that death must be imposed "fairly and with reasonable consistency or not at all," but rather obviously consistency will not be advanced by increasing discretionary sentencing, except for the tautology that all such sentences will be discretionary. (12) Discretionary sentencing forces the decision away from objective standards that permit judgments of similarity, and thus judgments of consistency and reliability, to be made and directs it toward the unique matrix of background and experience possessed by the individual decision maker. This does not mean that individuated decision making is inconsistent or unreliable; it means that those concepts are not applicable just to the extent that the concern is to judge the uniqueness of some person or event by reference to the unique background of a decision maker.
The tension between individuated decision making and categorical reasoning is also at the heart of Justice Stevens' separate opinion. His argument in essence is that by restricting the scope of capital punishment through the requirement that aggravating factors be found, the risk of arbitrariness is sufficiently reduced to permit individuated decisions within the remaining class. (13) According to Stevens the Supreme Court's earlier decisions limiting discretion were "a function of the size of the class of convicted persons who are eligible for the death penalty. . . However, the size of the class may be narrowed to reduce arbitrariness, even if a jury is then given complete discretion to show mercy when evaluating the individual characteristics of the few individuals who have been found death eligible." (14)
Justice Steven's argument is quite curious. To be sure, it is true that as the size of the death eligible group is reduced, the risk of the absolute number of cases involving jury arbitrariness in imposing death is decreased, where "arbitrariness" means an inappropriate death sentence imposed. However, this does not equate with either reduced arbitrariness in the system as a whole or with a reduced proportion of arbitrary decisions by juries. Reducing the size of the death eligible class will almost surely reduce the number of arbitrary death sentences imposed, but it will just as surely increase the number of arbitrary life sentences imposed (again, whatever "arbitrary" might mean). Similarly, the effect on proportions will be determined by the proportions in the excluded and included classes, and by how the jury decides the cases. Accordingly, Stevens' opinion reduces to defending the present practices for the reason that the absolute number of arbitrary death sentences will be reduced, but it is not obvious why that matters so much. Suppose, for example, that pre-Furman, there were 1,000 death penalties, 100 of which "deserved" a life sentence instead, and post-Furman there were 100 death sentences, 10 of which deserved a life sentence instead. Shifting from pre- to post-Furman does not change the proportion of arbitrary death sentences to nonarbitrary ones, and results in 90 "correct" life sentences at the expense of 810 "incorrect" life sentences. In a system in which errors cannot be eliminated, distinguishing between these two scenarios is not simple, a task Stevens did not even attempt.
Scalia concurs in the result, but announces that the Court's death penalty jurisprudence is so unjustifiable that he no longer will be bound by selected parts of it. His opening assertion of the illogic of his colleagues not only sets the tone for the remainder of his opinion but also demonstrates clearly the nature and limits of his methodology: "The ultimate choice in capital sentencing . . . is a unitary one--the choice between death and imprisonment. One cannot have discretion whether to select the one yet lack discretion to select the other." (15) The obvious explanation of this remarkable passage is that it rests upon the proposition that the only manner in which an issue may be analyzed is syllogistically with binary choices. A person is to live or die; a decision maker either has or lacks discretion. Every decision making process, maybe every process in Scalia's view, must be governed by a single principle entailing binary choices, no matter how complex or segmented that process may be. In essence, Scalia has conflated the effect of a decision with the process leading to it.
However, there is no principle, logical or legal, forbidding the use of differing intellectual tools depending upon the task. The real question is how could it be otherwise. Different parts of a process can be, and in some instances must be, structured in different ways. It is perfectly sensible to say that certain facts condition eligibility to some benefit or disability and that, after the minimum requirements are met, discretion is to be exercised, a point instantiated constantly in everyday life. Scalia's opinion suggests that he is blinded by the seductive power of deduction. Several passages in his opinion confirm this hypothesis. He says, for example, that "This second doctrine [of mercy]--counterdoctrine would be a better word--has completely exploded whatever coherence the notion of 'guided discretion' once had." (16) This overlooks that the two doctrines may be designed to do different things. He asserts as proof of the illogic of his colleagues that "Our cases proudly announce that the Constitution effectively prohibits the States from excluding from the sentencing decision any aspect of a defendant's character or record or any circumstance surrounding the crime: [for example] that the defendant had a poor and deprived childhood, or that he had a rich and spoiled childhood," (17) his point being that these both cannot be mitigating. But situated in real human beings, they could be, not in and of themselves, of course, but connected to other aspects of defendants' lives. Only if categories have to be carved out in advance would Scalia be right, but he has not explained why that is required.
Scalia's superficially appealing call for rationality and standards is a call for decision by rule. Fact X either is or is not mitigating, just like a decision maker must or must not have discretion. There is, in Scalia's cosmology, no judgment to exercise, only facts to be found and rules to be applied. If, however, mitigation is not designed to achieve rationality and predictability, if instead it is designed to implement judgment, then Scalia's argument is in error, and the Court's scheme may be rescued. We thus get to the heart of the matter: Judgment cannot be captured in rules; if it could be, judgment would not be required. Thus, the real question that cases like Walton pose is whether a justification other than one resting primarily on "predictability," "reliability," or "accuracy" can be constructed for the Court's bifurcated approach to capital sentencing, and I think it can. Doing so requires approaching the issue of capital sentencing free from the assumption that all legal questions require decision making by a priori rules, and examining instead the conditions under which a priori rules further a desirable goal at an acceptable cost.
Rough judgments of the relative culpability of acts have long been integral to the criminal law. If it is not illogical to use deliberation or premeditation to distinguish first degree murder from second degree murder, and the fact that most legislatures have done so indicates that the distinction captures a widespread intuition, why is it illogical to use specific aggravating factors to distinguish capital murder from first degree murder and why would we not think that this distinction most likely captures another intuition? The answer is obvious: it is not illogical at all to make the distinction, and it probably does capture a widely held intuition. Aggravating circumstances do not serve simply to reduce the size of the death eligible class, as Stevens suggests, nor do they serve simply as an undifferentiated brake on discretion, as Scalia suggests. (18) Rather, they eliminate from the death eligible group the cases for which death would be inappropriate. Thus, the proper vocabulary to criticize them with is precisely the vocabulary of rationality. Do the aggravating circumstances capture reasonably well understandable moral judgments? Are they seriously overinclusive, so that the risk of an inappropriate death sentence is not reduced? These are the questions to ask. Thus, attention must turn to how the mitigation function differs, for if it does not, Scalia's complaint of illogic is correct.
The difference emerges with the realization that the set of death eligible defendants is relatively homogeneous and that differentiation within a fairly homogeneous set requires the exercise of judgment. If the group of death eligible defendants deserve to die in some constitutionally acceptable sense, then that set is fairly homogeneous, which means that distinctions between them cannot be made by categorical rule. But if distinctions cannot be made by rule, they can only be made through the exercise of judgment. They can only be made, in other words, if the decision maker is free from rules and allowed to consider whatever is advanced. The role of mitigation is best understood as implementing just such a scheme. It is designed to permit fine, not gross, distinctions among death eligible individuals to be made. The counterintuitive nature of this point stems, I think, from the fact that the fine distinction yields a gross difference--literally the difference between life and death. But merely because the consequences are great does not entail that the justifications for the distinction are obvious or rough.
This answers Scalia's complaint. Mitigation is not designed to implement reliability or predictability, and the defenders of mitigation should forgo that vocabulary. It is designed to permit judgment to be exercised. Judge or juror is to consider whatever is advanced by the defendant to see if in the context of the defendant's life story, as seen through the lens of the decision maker's life story, there is an understandable and suitably powerful, even though subtle, reason to extend mercy. Neither the defendant's life story nor the intellectual resources brought to its evaluation by the decision maker will be capable of capture in rules, however. That is not the nature of this task.
Clarifying the nature of mitigation has a number of subsidiary benefits. It demonstrates further why Scalia's ridiculing of the apparently inconsistent mitigating factors is at best beside the point. Any particular fact is of very little consequence standing alone. The web of facts is what matters. A person from a poor and educationally deprived background who has transcended it and begun to make a success of life, but who kills to advance economic opportunities, may have no claim on our sympathies. A person from a rich and spoiled background may also be from one of little parental involvement in child rearing, and may suffer from recurring bouts of depression and drug use, and thus whose acts may be thought less reprehensible than that of our upwardly mobile killer.
Clarifying the nature of mitigation also clarifies one of the wooden aspects of the debate occurring in the Court. That debate proceeds as though the relevant "facts" were just what Scalia ridicules--such matters as a deprived or spoiled background--but this misses a crucial point. Those are not themselves "facts" in the sense of legally significant conclusions; rather, those "facts" are evidence of the legally significant conclusion, which is whether to grant mercy or not. If they were the legally operative "facts," Scalia's scathing attack on his colleagues would have some merit. For then these "facts" would begin to look very much like other operative legal facts, such as intent in the definition of homicide. Here Scalia would be right if he said "intent either is or is not an element of homicide." But he would be wrong if he said, "X's testimony about Y's behavior either does or does not establish intent." The legally operative fact in death hearings is whether to mitigate--whether the defendant's life history provides an adequate excuse to escape execution. Such a matter as the deprived nature of the defendant's background is evidence from which that conclusion may be inferred, but it will be inferred in light of all the evidence presented on the issue. (19)
If evidence of intent is analogous to evidence that mercy is in order, and an inference of intent is analogous to a conclusion that mercy should be extended, then it makes little sense to talk of burdens of proof with respect to the evidence of mitigation. This again is an analytical error of all the participants to this debate on the Court. By contrast, it makes perfect sense to ask whether the sentencer has been convinced of the justification for mercy by some standard of proof, and precisely that amount of "consistency" can be imported into the process. The justification for mercy, though, will emerge from all the evidence adduced and will not be a function of any discrete "fact" like a deprived or spoiled background.
And of course different sentencers will see the issue in different ways, which I suspect is the lurking unspoken problem. The evidence presented at a sentencing hearing will not bear its implications on its face; it will have to be intepreted by the sentencer. (20) Because of differing life experiences, one juror may find a defendant's background mitigating whereas another may find it aggravating. This is not proof of "randomness" in decision making, as Scalia would have it. (21) Again, it is simply the consequence of judgment, of human decision making encompassing too many variables to be reduced to rules.
There are, then, two related issues. The first is the nature of the problem. Mercy involves a judgment that runs over the entire range of a person's life. As I have tried to show, it is not possible to reduce such a judgment to rules. That means, and this is the second problem, that authority to decide must be allocated to some institution. It is the allocation of this power to juries, but one informed by some mistrust of them, that fuels the current debate about the procedures of the death penalty in the United States today. The cases of the Court taken as a whole are to be commended for intuiting that some things must be left to judgment. They are to be criticized for not seeing that same point clearly enough. The real issue that is lost in the misapplication of the Court's standard deductive, rule-bound approach is whether mercy should play a role in sentencing decisions. The arguments between the various members of the Court over this or that rule or its coherence are not only misguided but virtually beside the point. The real question is whether to accept a discretionary regime that merely allocates authority to decide to some person or institution and more or less live with the results, or to eliminate mercy through mandatory sentencing schemes. This is a deep and difficult question, but one I do not now have the time to address.
NotesParma, May 28, 1998
1. John Henry Wigmore Professor of Law, Northwestern
University. This lecture is based on a lecture given at the Sorbonne, April
28, 1998, which in turn was drawn from Allen, Evidence, Inference, Rules,
and Judgment in Constitutional Adjudication: The Intriguing Case of Walton
v. Arizona, 81 J. Crim. L. & Crim. 727 (1991).