Truth in Capital Punishment

One of the bigger crimlaw news items these days is a report from the Columbia School of Law that claims Texas executed an innocent man in 1989. According to an AFP wire story by Chantal Valery:

The report, entitled “Los Tocayos Carlos: Anatomy of a Wrongful Execution,” traces the facts surrounding the February 1983 murder of Wanda Lopez, a single mother who was stabbed in the gas station where she worked in a quiet corner of the Texas coastal city of Corpus Christi.

Forty minutes after the crime Carlos DeLuna was arrested not far from the gas station.

He was identified by only one eyewitness who saw a Hispanic male running from the gas station. But DeLuna had just shaved and was wearing a white dress shirt — unlike the killer, who an eyewitness said had a mustache and was wearing a grey flannel shirt.

DeLuna denied killing Lopez and instead identified Carlos Hernandez as the killer, but it appears police didn’t give his claim much credence. Prosecutors even accused DeLuna of making up Carlos Hernandez. DeLuna was convicted and sentenced to death, and six years later he was executed.

The Columbia report identifies an actual person named Carlos Hernandez who was in the area at the time. He looked a lot like Carlos DeLuna, and police knew who he was. He was eventually imprisoned for murdering another woman. He died there of natural causes, but not before repeatedly confessing to the murder for which DeLuna was executed.

I haven’t read the actual report, but from what I’m hearing, it makes a pretty good argument. I think Texas has probably executed an innocent man. He might not be the only one, if you believe the forensic argument that Cameron Todd Willingham didn’t commit the murder for which he was executed in 2004.

Regarding the DeLuna case, Houston criminal defense lawyer Mark Bennett says we should shout it from the rooftops:

The position of most adherents of the death penalty is that there are enough procedural safeguards built into the system that nobody has ever been executed for a crime he did not commit, and that the probability that someone factually innocent could be executed is so small that it does not merit chucking the penalty altogether.

It should be noted at the outset that the dissent does not discuss a single case–not one–in which it is clear that a person was executed for a crime he did not commit. If such an event had occurred in recent years, we would not have to hunt for it; the innocent’s name would be shouted from the rooftops by the abolition lobby.

Kansas v. Marsh (Scalia, J., concurring).

Prepare for that to change. Shout it from the rooftops: Carlos DeLuna, executed in Texas for the 1983 murder of Wanda Lopez.

Well, why not? I guess. I’m not really surprised by this news, though. Although some death penalty proponents have their heads in the sand, if we continue to have capital punishment, execution of an innocent person is inevitable.

About six years ago I got into an argument with a prosecutor who blogged that “the error rate for carried-out executions remains zero.” At the time, neither the Willingham nor DeLuna cases had come to light. Even if we ignore those cases, however, the prosecutor’s statement was misleading on two counts.

First, nobody knows the ultimate truth about the errors of our system of capital punishment. It’s not as if God comes down after the execution and tells us whether we got it right. We’ll never be able to rule out the excution of an innocent person with perfect certainty. The best we can do is talk about the rate of discovered errors.

There have been a number of cases where death row inmates had their sentences — or even their convictions — overturned because some error was discovered in the process that put them there. Proponents of the death penaltly say this is an example of the system working and correcting its mistakes. That’s true enough, the system corrected a mistake, but that doesn’t mean the system has corrected all its mistakes.

Imposition of the death penalty is a long road that starts with the sentence and ends with the execution. In between, there are a bunch of corrective measures, appeals and reviews, each of which reduces the chance of an erroneous execution. It’s a mistake to think that no innocent people have reached the end of the road just because we don’t know their names.

In response to a couple of comments along those lines, the prosecutor responded,

Nice try, but the issue is not what statistical probabilities tell us, since they after all are only educated guesses. The point I have made is that the known error rate for completed executions stands at zero. That is fact, not conjecture: there has not been a single case of an innocent person being executed who was thereafter shown to be factually innocent. It’s as simple as that, and all the statistical gymnastics in the world do not change that actuality.

This brings us to the second reason it’s misleading to look at the error rate for carried out executions. The prosecutor dismissed “conjecture” about the error rate, but that ignores the purpose of our argument. We’re talking about capital punishment because we are trying to make a policy decision: Should we continue to execute criminals? And if so, under what circumstances?

This is necessarily a decision about our future, and when it comes to our future, we don’t have the facts yet. Even if we could prove, with absolute and perfect certainty, that no innocent person has been executed, that doesn’t prove we won’t execute one tomorrow. When it comes to the future, conjecture is the all we’ve got.

We’ve got to make decisions about the future of our capital punishment system, and that requires us to understand how our capital punishment system behaves, for which the best evidence is its performance in the past. But the past is only a sample of the system’s behavior, not a complete description. We use past performance data to constrain statistical models that predict its future.

Suppose you flip a coin five times, and it comes up heads all five times. Would you be willing to declare that the coin is a perfect head-flipping coin? Probably not. What if you got 10 heads in a row? That should make you more confident, but is it enough? How many times in a row would the coin have to come up heads before you would be willing to bet something valuable on it? Something like the life of an innocent American.

Let me try to make this a little more complete with a very rough mathemetical model. Suppose we can approximate the decision to execute someone — from identification of a suspect to throwing the final switch — as a linear series of steps. If there are N steps in the capital punishment system, and at each step i there is a probability pi (where 0 <= pi <= 1) that an innocent person will slip past, then the probability pexecution of an innocent person slipping past all the steps and being executed is given by:

pexecution = p1 × p2 × p3 ×…× pN

Simple math tells us that The value of pexecution can only be zero if at least one of the terms pi is zero. That is, unless one step in the process is perfect, the whole process is necessarily imperfect. If any death penalty proponent really thinks the process is perfect, I invite them to identify the one perfect step so we can save a lot of time and money by dispensing with the rest of the justice system.

If that’s too abstract, then imagine our current policy extended into the distant future. Assume we execute about 50 people year, and assume we do it for the next million years. Do you really believe we can execute 50 million people without even one of them being innocent? No? Great, then we agree the death penalty isn’t perfect. Now we’re just arguing about the number.

Elsewhere in his concurrence on Kansas v. Marsh, Justice Scalia cites Oregon District Attorney Joshua Marquis’s absurd calculation of a false felony conviction rate of only 0.027%. However, assuming that DeLuna and Willingham were both innocent (as seems likely to me), and knowing that there have been 1295 executions since the Supreme Court decided it was constitional (again), we can set a lower limit on pexecution of (2/1295=) 0.15444%, more than five times higher than Scalia’s number.

The wrongful execution rate of 0.15444% is a lower limit because we can’t be sure there weren’t more innocent people executed. We’ve discovered two of them, but the total number depends on the probability of discovery pdiscovery of a wrongful execution, and right now all we can say is

pexecution × pdiscovery = 0.15444%

The smaller the probability of discovery, the more innocent people must have been executed for us to discover two of them.

In the end, Scalia did get it right:

Like other human institutions, courts and juries are not perfect. One cannot have a system of criminal punishment without accepting the possibility that someone will be punished mistakenly.

That may sound callous, but it has the advantage of being true.

5 Responses to Truth in Capital Punishment

  1. Interesting post, Mark… I haven’t looked into these cases yet, but I won’t be surprised if, like the Coleman case, it turns out there is less there than meets the eye. The desire of advocacy groups often is out in front of their care about facts.

    My only point about zero shown cases of error is that despite the sometimes near-hysteria rhetoric, the facts don’t support the fear-mongering.

    I’ve always acknowledged that the criminal justice system being composed of fallible human beings, errors may indeed occur. The “system” itself acknowledges that we cannot attain complete certitude when it requires, even in death cases, proof only beyond a “reasonable” doubt, not proof to a metaphysical certitude or beyond “any” doubt.

    We live in an imperfect world, and while we struggle mightily to ensure fair trials, occasionally a wrongful conviction may occur. I just haven’t seen conclusive evidence that this has happened in a death case, and that’s no surprise, since these cases are generally litigated more carefully, and certainly are more scrutinized on appeal, than other cases.

    Only about .05% of murders are punished by imposition of the death penalty. What this tells me is that prosecutors are very selective about which cases merit the extra work and scrutiny for a death sentence; judges and juries are very selective about which cases they believe merit imposition of the death sentence the prosecutor is asking them to impose; and appellate courts are very serious about scrutinizing, and many times reversing, death sentences.

    In sum, could there theoretically be a wrongful death sentence carried out? Yes. But my reply is “so what?” Everyone’s always known that the possibility exists. But so long as there are crimes that call for it (e.g., the Connecticut home invasion a while back), I believe it should be one weapon in the community’s arsenal to address certain heinous crimes.

    Some I’m sure believe we should execute no one because a mistake might be made. Everyone’s free to have that opinion, it’s just not one shared by the majority of our citizens or by the 6,000 year history of our Western jurisprudence.

  2. Tom, I haven’t looked into the details of either case, in part because I wouldn’t know what to look for, but based on what I’ve read about them, they sound like the real thing to me.

    “So what?” is definitely a more honest answer than refusing to believe there have been — or ever will be — wrongful death sentences. While we all would like that number to be zero, the cost of making it zero may be too high, especially in terms of letting too many guilty people continue to commit crimes.

    I think the 6000 year history of Western jurisprudence is more of a dire warning than a good template for an effective and moral system of justice. Perhaps we haven’t executed any innocent people in the last few decades, but we certainly used to, often for things that aren’t even considered crimes today such as heresy, or for witchcraft, which it isn’t possible for anyone to commit.

  3. Some of the people executed of heresy were indeed guilty of it.

    As for Tom’s comment that “believe we should execute no one because a mistake might be made”, the reason that is not popular is because we tolerate other things that kill innocent people a lot more often.

    If we take the one known case of innocent people being executed, which was two people in 1927, and couple it with these cases from 1989 and 2004, on average innocent people get executed once every twenty years in the U.S.. By sharp contrast, since 1996 I can name six cases of innocent people wrongfully shot to death by police. that averages to about once every three years. (And yes, we can adopt rules of engagement to make such events rare, such as forbidding police to use their weapons unless and until the belligerent uses force first.) Then of course there are the thousands of innocents killed by the atomic bombs.

    To execute a thousand innocent people would take about twenty thousand years.

  4. Michael, you’re right, innocent people die all the time doing things like commercial fishing, coal mining, logging, and construction work. Heck, farming is still a dangerous line of work. On family farms, that can mean children trampled by livestock. Or consider the thousands of innocents killed each year in auto accidents. Yet, as a society, we consider the benefits of all these things to be worth the cost in innocent lives. That’s also a decision we have to make regarding the death penalty: Is it worth the cost?

Leave a reply