Brooke Masters: Apart from whether, as a human-rights issue, we ought to have the death penalty, what's the problem with the way the death penalty is being administered in the United States--both in terms of racial and economic bias and in terms of sheer mistake? Perhaps Professor Liebman might want to tell us about his work.

Panelists discuss whether the errors that arise in the administration of the death penalty in the United States are symbolic of systemic disorder or whether the death penalty presents a unique set of circumstances. (22:09 min)

James Liebman: This is an important point, and it ties into things that we talked about before. I think that the issue in the United States is now becoming one of public policy, not of law. The issue has been taken away from the courts, which have really controlled it for decades, and people are saying, "We're for the death penalty, by and large, but we recognize that it's a government system, and we want the system to work. We want it to serve some purposes."

So the question is: Does it serve those purposes? And that's where I think a lot of people, even death penalty supporters, think a lot of problems are emerging. Colleagues and I at Columbia have done a study of death penalty cases during 23 years following the reinstitution of the death penalty, after the moratorium that existed in this country in the '60s and '70s. We found that 68 percent of the death sentences imposed in this country during that period were found to be so legally flawed that they couldn't be carried out and had to be overturned. Most of those legal errors are the kinds that do affect the reliability of the outcome in some way or another. There had to be a finding of an effect on reliability, or a finding of no violation there. As a result of all of that error, it takes an average now of 12 or 13 years to get from death sentence to execution. Add all this up, and it means that you have many people getting sentenced to death, very few people getting executed, and in the meantime there's a huge amount of delay. The average outcome is that a person sentenced to die is going to end up coming off death row because his death sentence is found by the courts to be invalid. And that raises two big questions. First, if there's that much error in the system, how can we be sure we're catching all of the errors? And if we're not catching all of the errors, so that some serious errors that undermine accuracy slip through the net, is it probable that people who are innocent are being executed, or that people who may have committed a crime but are legally ineligible for the death penalty are being executed by mistake?

We know the courts fail to catch mistakes of this sort, because there are a number of people who have been found to be innocent before they got executed but after the courts said it was fine to execute them. In these cases, it was not the courts that caught the problem but the accidental intervention of a journalist or a moviemaker or some undergraduate students that led to the discovery that the person was innocent. So one big problem is just error and bad outcomes leading to the execution of the wrong person.

The second problem, which I think is more widespread and may be more important to many people, is that this is a system that doesn't work. It's essentially a system designed to make mistakes and then try to catch and cure them. It's an expensive machine for making mistakes and trying to fix them. Yet, who in their right mind would design a production scheme or a machine to make errors and then try to correct them? Because that in essence is how our system works today. A lot of people, whatever their views on the death penalty, are reaching the conclusion that it is not a valid way for the government to be operating.

Brooke Masters: Professor Stone, you've done a lot of these cases. What kinds of mistakes do get made?

Randolph N. Stone: In Illinois, there's a moratorium on the death penalty imposed by the governor after about a dozen individuals who had been sentenced to death were later discovered to be innocent. The most recent case before the moratorium was a gentleman who was two or three days away from being executed and the defense lawyers were focusing on his mental abilities, because his IQ was allegedly very low. As Professor Liebman pointed out, a group of journalism students went out and discovered that someone else had actually committed the crime, and they interviewed that individual, who ultimately confessed. The errors that led to the innocent man being convicted-- inadequate defense resources and alleged police misconduct--are tolerated by the system.

Moreover, prosecutorial misconduct regarding withholding exculpatory evidence exists, because, as we all know, or at least we should know, in the trial setting it's a very competitive enterprise. Once

	Discussion
	In your opinion, should the American capital-punishment system be abolished, even if DNA evidence or other scientific evidence can prove that a defendant is guilty? One argument for a moratorium on the death penalty is that the judicial system is not always confident that every defendant is, in fact, guilty of the crime they are charged with. If DNA evidence could definitively prove a defendant's guilt, thereby improving sentencing accuracy, would you still oppose the death penalty?

people get involved in the trial, the prosecution wants to win. And in most cases the crime typically is often very heinous, a very brutal crime. The prosecutors become convinced that the person is guilty, and then they cut corners, withhold evidence, don't produce exculpatory evidence, hide reports. Finally, people make mistakes. The identification testimony of witnesses is often given a much greater weight than it should be, and so we have a series of cases where people, victims, misidentify an individual. Later, DNA evidence or other scientific evidence proves that the defendant could not have committed the crime. So, in sum, you're talking about defense incompetence based on resources in many instances, not having the resources to hire investigators, etc. You're talking about prosecutorial misconduct, you're talking about police misconduct or incompetence in some instances, and you're talking about judicial indifference in many of these cases where the system encourages judges to focus on case processing and efficiency as opposed to fairness and justice. All of those circumstances contribute to the error that occurs in capital cases, and it's merely symbolic of the way our criminal-justice system functions. The good thing about this spotlight on capital cases is that hopefully it will have some impact on the rest of the criminal-justice system and some seepage in reforming the way our system operates.

Brooke Masters: Do you think capital punishment is symbolic of a problem that is larger than just what happens in death-penalty cases?

Joseph Hoffman: On that point, I agree that the death penalty is the best and worst of our criminal-justice system. There's no doubt that the same kinds of problems that have come to light in capital cases can also happen and do also happen in other kinds of cases. It's in that sense that the reforms that people are discussing in the capital-case context maybe ought to be considered for other kinds of cases as well. But I want to circle back to the points that have just been made by Jim and by Randolph, because I want to present what may be a slightly different way of looking at the same situation.

There is no question in anyone's mind, no matter what their view is about capital punishment, that when a prosecutor misbehaves, hides evidence, when a defense attorney, as has happened on occasion, falls asleep during a capital trial, when a prosecutor makes a decision to seek the death penalty on bases other than the heinousness of the crime, on bases having to do with illegitimate factors or illegitimate motives, there's no question that those kinds of things are just plain wrong and ought to be prevented if it's possible to do that. It will be difficult in a system of human discretionary decision making, but we ought to be doing everything we can to prevent those things from happening, and we ought to be doing everything we can to identify those cases after the fact and correct them if we have not prevented them from happening initially. The problem is trying to figure out the depth of this problem. The kinds of errors that go to the substantive justice of the trial and the sentence undermine our confidence in the two key questions: Is this really the person that did this crime? And, if so, is this a person who deserves to die for the crime that they committed? Those are the two questions that we all ought to care about deeply and passionately, no matter what our view is about capital punishment. We've simply got to leave no stone unturned to try to fix any action, any misbehavior, any error that undermines our confidence in those two questions and the answers to those two questions.

What we don't yet know--and I know, Jim, that your study is going into another phase where we may get more information very soon about this--is of the 68 percent of the cases that got reversed during the period of this study, how many of those are cases where we really do have serious doubts about whether this person committed this crime, and about whether they deserve to die? What we do know is that the 68 percent were all reversed because of some kind of procedural irregularity. Some of those procedural irregularities are the kind that Randolph was just talking about, that we all ought to be staying awake at night worrying about. But many of the procedural irregularities accounted for in that pool of reversed cases are not that kind of error.

Brooke Masters: What's a not very important error?

Joseph Hoffman: A jury instruction error, where the language of the jury instruction may have misled a juror into thinking something about the way they were supposed to process this evidence compared with the way we'd like them to process the evidence. We would in a capital case want to correct that error, because it has some potential impact on the reliability of the outcome. But if you had 100 cases where the judge gave that instruction, maybe only five or six of them would be cases where the jury would be actually affected by it. We don't know. We can't know.

James Liebman: I strongly disagree on that point. I'll put my marker in here so we can come back to it.

Joseph Hoffman: I understand, and my guess is we'll continue to disagree about that point. There is no doubt that in capital cases the procedural rules that the courts have created were created in large measure because everybody recognized what the stakes were. We have a term called "super due process" that those of us in the legal business talk about in capital cases, and what it means is that we create rules that don't apply to other kinds of cases. They are special rules that, if they're violated, can lead to the reversal of a sentence or even a conviction. Those rules were made for good reasons, but it doesn't mean that every violation of such a rule means that we really would be in doubt about the result that was reached in that case. Would some of those results be questioned? Yes. Would all of them be questioned just because of that procedural error? No. We don't know how many of those errors--within that class of reversals, within that class of errors, as it were--are ones that really undermine our confidence that we got the right guy and that they deserve to die.

James Liebman: Let me offer some perspective on this. First of all, almost all violations that result in reversal require some kind of showing of prejudice, and for the most part--for most of the big sources of error--you have to show not that the error may have had an effect, but that there is reasonable probability that it did affect the outcome. For example, you can have a lawyer who slept through the entire trial, but that doesn't mean you have the case reversed. In addition, you have to show a reasonable probability that the outcome would have been different if the lawyer had been awake.

The same thing is true with prosecutorial misconduct, and with jury instructions. There has to be a reasonable probability that the jurors did the wrong thing because they got the wrong instruction. Most of these errors undermine reliability. Second, we found error rates above 50 percent in every single one of the 28 states we looked at save one, so you're finding error all across the board. Third, in all but one of the 23 years we examined, we found a 50 percent error rate. Fourth, we looked at what happened to these cases when the error was discovered and the case was sent back for retrial. Are the errors serious enough so that when you cure the error the result changes? We found that in 82 percent of those cases (state post-conviction cases) where we have the data, the result after you cured the error was very different indeed: a sentence less than death. In 7 percent of those cases, the defendant was acquitted. So we think there is a lot of evidence, even based on what we know, that these errors do go to the heart of the matter, that they do cause people to be condemned who by all accounts shouldn't be. They are errors that give you pause. This doesn't mean that the person in each one of those cases was innocent, but it does mean that there often is a defect in the case that creates a real doubt about either guilt or whether the death penalty was deserved. And when you have that over and over again, across 23 years, across 28 states, at such high rates, it means that something is wrong with the way the system is working.

Randolph N. Stone: Let me jump in just briefly. I think the two questions you pose are good ones. One is: Is the person who is on trial, or charged, the person who committed the crime? Certainly that's a question that we want to know. But I think we also want to know: Did the person receive a fair trial? Regardless of whether or not they actually committed the crime, there are some errors that shock the conscience so much that the person is deprived of a fair trial. For example, when a prosecutor excludes all racial minorities from a jury, even though the person committed the crime, the fairness in the process has been so violated that that person should receive a new trial. In Chicago, we had a series of cases where the police tortured people to confess to crimes. Now, some of those people actually committed the crime, but the fact that the fairness of the process was so violated by this police conduct entitled those people to a new trial, and I don't think you would disagree with that.

Joseph Hoffman: Of course not. I don't think any reasonable person would disagree that when a serious procedural error deprives someone of a fair trial--for example, in the exact examples that you gave--the legal system should remedy that. I was only challenging the claim that 68 percent of death penalty cases were erroneous in a sense that undermined the substantive justice of that verdict. We could probably go for hours on this point, Jim, but I continue to believe that, subject to enhancements and further stages of the study, the study that's been done thus far does not document that 68 percent of the cases were reversed because there was serious substantive doubt. Let me just give you an example.

In the past 25 years of Supreme Court litigation, we've all seen examples of these kinds of cases where you read the facts and you read what happened in the trial court, and a reasonable person would be left with some question about whether the right result was reached. Under our system of justice--and I think this has been a mistake; I mean, this has been a problem with our system of justice--typically our Supreme Court has not felt that it had the power or the right or the moral authority to simply reach in and say, "I disagree with this result. I think you, the jury, reached the wrong result in this case. This person either was innocent or, alternatively, they were guilty, but this is not a case that deserves a death sentence." Our Supreme Court, for a variety of reasons, has felt essentially disabled from making that kind of direct ruling that would result in overturning that case and that case alone. Instead, repeatedly, it has looked to ask, "Well, why did that result come out wrong? There must have been some procedural flaw in that trial or in that sentencing procedure. Something must have gone wrong procedurally that would have led to that result." Of course, in that particular case you could say that I think the procedural flaw led to this bad result. But when the Supreme Court announces its new procedural rule, you've got to do it differently. Because this flaw led to a bad result, hundreds of other cases, historically--this is changing now--hundreds of other death penalty cases would also have to be reversed on that same ground. Because the same procedural flaw would have existed in those other cases, and the result in those cases may not have been in serious doubt. We don't know in how many of those cases the result would be in doubt or in how many they wouldn't have been in doubt.

James Liebman: There's another statistic that's important here, not from our study, that is very stable and worth looking at. Essentially from the beginning of the reinstatement of the death penalty, in 1973, till today, it has been the case, year in and year out, that for about every seven people who are executed there's one person on death row who is exonerated--or at least about whom there is no evidence of guilt. Year in and year out, it just kind of works out that there's one exonerated death-row inmate for every seven or so executed. True, it's not as bad in Illinois, where essentially you have a one-to-one ratio of people being found innocent to people being executed. But even a one-to-seven, one-to-eight ratio is reason to be concerned about how we can be generating that much error at that serious level. And innocent people who are sentenced to die are just the tip of the iceberg. You have to think about all the people who are guilty of something but not guilty of a crime for which the law makes the death penalty even permissible who get the death penalty and are very likely executed.

William Schabas: I do have another point here, aside from quarreling about the statistics, and that is: If you want to justify the death penalty--and we're not getting into the arguments about retribution and deterrence and all the classic arguments--you have to have a system of imposing the death penalty that's functional, that works more or less reliably. What the study appears to show--and no one seems to be quarreling with it--is that the whole system is dysfunctional in a rather profound way. Maybe it's one that you can start to correct and see about correcting, but it seems to be terribly dysfunctional. So what possible justification can you have for it?

We're terrified at the idea of executing an innocent person, but we should be terrified at the idea of even sending an innocent person to jail. But you can correct that, of course. You can go back on it. But in terms of a public justification, it doesn't fill any retributive goal if it's so haphazard that it's like a shotgun. What was the expression the judge used? It's like being struck by lightning.

James Liebman: Let me give some statistics on that point, because I think it's really important. In the United States between 1973 and 1995, in those 23 years, the states imposed thousands of death sentences. Among those death sentences, only 5 percent were carried out. In no year during that period were more than 2.5 percent of the people on death row executed. In most years it was between .5 percent and 1.5 percent of death row. So, from the standpoint of death-penalty supporters, this is a system that is not functioning.

Most people who support the death penalty assume, I think, that if we give somebody the death penalty, he or she probably deserves it, and that the sentence will most likely be carried out and that it will occur within a reasonable time. But whatever else is true, that's not happening. And the main reason it's not happening is that there in fact is so much error in the system that we have to examine every case carefully, which leads us to find yet more error, which slows the process down. So the problem is not only that we are risking bad outcomes but that we have a bad process, whatever the outcomes are. And the two problems are related and mutually reinforcing.