In early April 1976, Potter Stewart, Lewis Powell, and John Paul Stevens met for lunch at the Monocle, a venerable Washington steakhouse, and decided the future of the American death penalty. The three U.S. Supreme Court justices were in a bind. Each harbored substantial misgivings about capital punishment, but each man — Stewart especially — also felt constrained by the issue’s peculiar constitutional history and by the tidal wave of public support that returned the death penalty to the Supreme Court just four years after a splintered court had declared it dead.
In 1972, Stewart had brokered a 5-4 decision holding that the death penalty as then practiced violated the Eighth Amendment’s ban against cruel and unusual punishment. The court’s ruling — in Furman v. Georgia — was a spectacular long shot. Just one term earlier, in 1971, the justices had upheld the constitutionality of the death penalty under the due process clause of the Fourteenth Amendment. Furman seemed headed in the same direction until Stewart struck an eleventh-hour deal with Justice Byron White, who’d been on the fence for most of the term. Stewart would abandon the moral statement against the death penalty that he’d intended to make and would instead say that the problem with capital punishment was excessive arbitrariness.
Arbitrariness thus became the dominant theme in the most splintered decision in Supreme Court history. Each justice in the Furman majority wrote his own solo opinion — meaning that he spoke only for himself. Each emphasized a different aspect of arbitrariness. Some focused on racism. Some focused on the failure of states to condemn only the “worst” criminals. Some focused on the infrequency with which the death penalty was employed. No one said precisely how much arbitrariness violated the Constitution. Surely Stewart understood, as his colleagues must have, that the focus on arbitrariness — as opposed to deeming capital punishment unconstitutional per se — left the door open for states to rewrite their laws. Nevertheless, the justices believed that, as Stewart told his clerks, “The death penalty in the United States was finished.”
That intuition couldn’t have been more wrong. Between the Furman decision and 1976, 35 states passed new death penalty statutes. Seven made the death penalty mandatory for murder. Others, including Georgia, instead attempted to make the process less “arbitrary” by requiring capital jurors to find “aggravating” factors, by separating capital trials into the guilt/innocence and sentencing phases we see today and by guaranteeing appellate review of all death sentences.
The political and legal momentum against Furman forced the justices to reconsider their positions. So, over two days of oral argument beginning on March 30, 1976, the justices evaluated the constitutionality of the various new state approaches, with Georgia’s new statute as the test case. The hearing had the feeling of a heavyweight prize fight, pitting against one another two of the great lawyers of their generation: Solicitor General Robert Bork, who, a decade later, would be nominated to the Supreme Court, and Anthony Amsterdam, principal architect of the NAACP Legal Defense Fund’s victorious strategy in Furman and widely acknowledged as the greatest civil rights lawyer of the 20th Century.
Two days later, on April 2, the justices met in conference to consider the new death penalty laws. Justices William Brennan and Thurgood Marshall said they would reject both the statutes that made a death penalty mandatory for murder as well as statutes that gave jurors the discretion to sentence defendants to death. Justice White and three of the Nixon appointees—Justices William Rehnquist, Harry Blackmun and Chief Justice Warren Burger—said they’d uphold both approaches. The case came down to Justices Stewart and Stevens and Nixon’s fourth appointee, Justice Lewis Powell.
At that Washington steakhouse, the troika, as they’d come to be known, decided to split the baby. They would reject the mandatory statutes, which they regarded as barbaric, but uphold the guided discretion approach. Together with the four Nixon appointees, they formed a 7-2 majority in Gregg v. Georgia, upholding Georgia’s new discretionary law, and, with Marshall and Brennan, a separate 5-4 majority rejecting the mandatory statutes. This Solomonic compromise created the bedrock principles of modern death penalty jurisprudence: that a non-arbitrary death penalty satisfies the Constitution and that the requirement of non-arbitrariness could be satisfied by Georgia’s approach.
The Gregg decision revived the American death penalty. It also began a social experiment. Underlying Gregg is an empirical proposition: legal standards would make capital jury decisions more predictable. “Let’s have them be as guided and as rational as they can be,” Stewart told his law clerk Ron Stern in 1976. Yet in five years of archival research and interviews for my book A Wild Justice, I found not a shred of evidence that any of the justices considered social science data. Certainly none is cited in the opinion. The most striking features of the compromise Stewart, Stevens and Powell embraced were the speed with which it was reached, the absence of supporting empirical evidence, and the three men’s unquestioning faith in the power of law and the state and local officials sworn to carry it out.
Forty years later, the data are in on the court’s grand compromise. How one interprets the results may depend on what’s being asked. If the essential question today is whether the death penalty is still being applied arbitrarily, the answer couldn’t be clearer. Arbitrariness is rampant. But, on the occasion of Gregg’s ruby anniversary, let’s ask a more refined question, which more directly honors the case’s peculiar history: Is arbitrariness less of a problem than it was before the Supreme Court got involved in 1972? In other words, has Gregg worked?
The answer is a conclusive, resounding no. Whether one interprets the Furman decision to have been about — individually or collectively — excessive racism, a failure to identify the “worst of the worst” among murderers, the death penalty’s sporadic use, or simple geographical randomness, the “guided discretion” statutes endorsed in Gregg haven’t remotely fulfilled their promise. Randomness has not been reduced and in many respects has grown substantially worse.
Almost all of the justices in Furman noted the low percentage of death-penalty eligible murders that resulted in death sentences. They estimated the rate to be between 15% and 20%. From this statistic, the justices drew different conclusions. Brennan and Marshall cited it as evidence that the death penalty had been rejected by contemporary standards of decency (though the truth is they would have opposed the death penalty regardless). White said an infrequently-used death penalty couldn’t adequately deter crime. His idiosyncratic opinion seemed to invite mandatory laws.
A third view seemed to get at something important about what a constitutional death penalty might look like. Not even the most ardent supporters of the death penalty believe that all murderers should be executed. Somewhere a line has to be drawn, and it should be drawn in such a way that juries regularly accept the penalty’s use. For example, if a state restricted its use of the death penalty to mass murderers it presumably would generate a high sentencing rate. As the state broadened its law to include less aggravated kinds of murder its sentencing rate would decline as jurors (or judges) rejected borderline capital charges.
Implicit in Furman was the premise that states had for decades defined the universe of death-eligible murders too widely. Implicit in Gregg was the premise that guiding jury discretion would create some balance between death-eligible cases and actual death penalties. But 40 years of statistics tell us that the death penalty is even rarer than it was before.
In gross terms, U.S. executions have been trending downward for some time. Annual executions peaked at 197 in 1935, hovered between 50 and 100 per year during the 1950s, fell further after the NAACP Legal Defense Fund took on the death penalty in the mid-1960s, and temporarily ended with Furman. Post-Gregg, executions peaked at 98 in 1999. They’ve been declining fairly steadily since. Thirty-five people were executed in 2014. Last year, states killed 28 people, the lowest total since 1991. Death sentences have been dropping too—from more than 300 annually in 1995 and 1996 to 73 last year.
Over this period, death-sentencing rates — meaning the percentage of murders eligible for the death penalty who are actually sentenced to death — have experienced a parallel decline. Almost every state-level study has identified a rate either at the low end of, or substantially below, Furman’s 15%-20% threshold. In California, the nation’s largest producer of death sentences, the most comprehensive statewide studies have identified a sentencing rate between 4.6% and 5.5%. In a review of 34 years of Connecticut death penalty cases, Yale’s John Donohue found a sentencing rate of 4.4%. A study of all Colorado murder convictions between 1999 and 2010 revealed a rate of 0.56%.
Pause for a moment to note that these are only sentencing rates. Between 1973 and 2013, only 16.1% of people sentenced to die were ultimately executed. In other words, the chance of being executed — among defendants sentenced to die — is only about one in six. The probability of receiving the death penalty in the U.S. is miniscule. The Centers for Disease Control reported 16,121 homicides in 2013, while states conducted 39 executions — an execution rate of approximately .24%.
Why sentencing rates are so low is no secret. Rather than define a narrow, clear universe of the most aggravated crimes — such as mass murder — states have moved in the opposite direction and included more and more kinds of homicides as death eligible. The Supreme Court has countenanced this, saying that any aggravating factor is constitutional so long as it doesn’t include everyone. For example, the court upheld Idaho’s aggravating factor that made someone death eligible who “exhibited utter disregard for human life,” presumably drawing a distinction between murderers who show “utter” and, say, “mere” disregard for human life. The court has similarly upheld a “cold, calculated” aggravating factor and yet another for murders that are “especially heinous, cruel or depraved,” although it’s hard to imagine a murder that isn’t cold or cruel or heinous.
These are the worst examples, though, and reasonable arguments can be made in favor of many aggravating factors. The devil is in how they work together. Consider this: some states — superficially reasonably — make a murderer death eligible if the victim is very young. Other states make it an aggravating factor to kill an elderly person. Some states include both. Each aggravating factor can be defended individually. The young have more life ahead of them. The elderly are less able to defend themselves. But by this logic, why not include murders of the infirm, as many states do, or pregnant women, as some states do, or, for that matter, all parents, on the reasoning that they have more responsibility?
Collectively, aggravating factors fail to accomplish Furman’s fundamental goal of limiting the circumstances in which the penalty may be applied. A California study found that 87% of murders are potentially eligible for the death penalty under the state’s definitions. In Colorado, the rate is 91.1%.
Some abolitionists fear that pushing the court on this point would lead to another death penalty reset, and that states would respond, as they did after Furman, by yet again rewriting their laws. What would happen is impossible to know, of course. Perhaps the court would see that 40 years of history shows that the political temptation to protect every category of murder victims is irresistible — once a state legislature has decided to make police killers death eligible, how, for example, does it refuse corrections officers or firefighters? Perhaps the court would conclude that writing a meaningfully nuanced death penalty law is therefore impossible. Or, perhaps, we’d live another generation with slightly narrower but still grossly imperfect statutes. The only thing that can be said conclusively is that states have failed to engage in the kind of “limiting” exercise Furman intended to require.
Prior to Furman, the American death penalty was largely a Southern phenomenon. Between 1930 and 1967, of the 3,859 executions carried out in the U.S., 2,306 —59.7%— occurred in the South. It was his concern with the racist application of the death penalty in the South that motivated Justice Arthur Goldberg and his law clerk Alan Dershowitz to take on capital punishment in the early 1960s, even though no one in the court and almost no one in the scholarly community believed it to be unconstitutional.
Since Gregg, the American death penalty has become almost exclusively a Southern phenomenon. Of the 1,429 executions conducted in the U.S. since 1976, 1,163 —81.3%— have been carried out by Southern states. The three most active death penalty states — Texas, Oklahoma and Virginia — have been responsible for 52.9% of executions since 1976. Texas, by itself, accounts for 37.6% of the total. Of the 28 executions carried out in 2015, 13 occurred in Texas. Of the nine carried out so far this year, Texas has conducted five. It makes no sense to speak of an American death penalty. There is the South and everywhere else.
Over the past few years, better data have facilitated study of how the death penalty functions at smaller geographical units than the state. The picture that has emerged from this disaggregated data is about as random as could be imagined. In 2012, just 62 of the 3,143 counties in the U.S., or 1.9%, accounted for all American executions. Since 1976, only 15% of counties have sentenced someone to die who was eventually executed. The top 15 most active death penalty counties have been responsible for more than 30% of all executions. Generally speaking, rural areas use the death penalty more than cities. For example, the odds of receiving the death penalty in Cook County are approximately 83% lower than for killing someone in rural Illinois.
Some of the quirks are bizarre. Murder defendants in Maryland’s Baltimore County, which has a comparatively low crime rate, are 23 times more likely to receive the death penalty than murder defendants in neighboring Baltimore City, which has a much higher crime rate. Geographic luck plays such an extensive role, it doesn’t make sense to speak of the death penalty system as a “system.” Even in Texas, only a handful of the state’s 254 counties seek capital punishment with any regularity. Harris County is single-handedly responsible for 280 of the state’s approximately 1,000 post-Furman death sentences.
Geographic arbitrariness is interrelated with racism. In a study of murder convictions in Alameda County, Steven Shatz and Terry Dalton found the district attorney was substantially more likely to seek the death penalty in the overwhelmingly-white southern half of the county than in the north, where more than 30% of residents are African-American — mirroring the race-of-victim racism noted in the overwhelming majority of death penalty studies. The closer one looks, the bleaker the picture. A system serious about creating consistency would require a statewide authority —say, a panel of experts— to make uniform death penalty recommendations, but none does, and the Supreme Court has not said one is required.
Probably the most basic concept in the court’s concern over arbitrariness is that the death penalty should be reserved for the most culpable offenders. Reasonable minds may differ on the morality of the death penalty, but we have a shared understanding that if it does exist, it should be reserved for the worst of the worst.
Some arbitrariness is an inevitable product of federalism. Many highly “aggravated” murders are exempt from the death penalty because they take place in an abolitionist state. Many less “aggravated” murders result in death penalties because they occur in active capital states. Given federalism, the Gregg court must have meant that states are supposed to make internally consistent decisions about who gets the death penalty. Some states may choose to prioritize protecting cops, for example, other states would choose to prioritize protecting the very young or very old. But whatever the state’s internal calibration of culpability may be, it needs to be applied consistently. Even judged by this more modest goal, Gregg has been a failure.
In his study of capital sentencing in Connecticut, Yale Law School’s John Donohue identified 205 potentially death eligible murders in the state between 1973 and 2007. Among these, death sentences were imposed in twelve cases, nine of which were upheld on appeal. For each defendant, Donohue calculated an egregiousness score by coding for factors that influence jurors, such as the extent of the victim’s suffering, the defendant’s motive, the number of victims, and mitigating factors, such as mental illness or impairment.
When he compared the egregiousness of the nine defendants sentenced to die with the other 196 cases, only one of the nine was among the “worst of the worst,” which he defined as being within the 15% most egregious cases. The other eight death sentences couldn’t be justified. As many as 170 of the defendants who’d been spared had committed more aggravated murders than the defendants who’d been sentenced to die.
Another study by Drake Law School’s David McCord used a “depravity point calculator” to review media accounts of murders, and again found no consistency. Among murderers who didn’t receive death sentences, 159 were more depraved than many of those who were sentenced to die. The question that we’ve asked together here — whether Furman and Gregg made the administration of the death penalty better, or at least less awful — makes no sense in this context. The system operates almost completely irrationally.
Again, it’s not difficult to identify why. The Supreme Court does not require guidelines for prosecutors in deciding whether to charge a defendant with the death penalty and no states have any. Nor does the court require—and again most states do not give—any guidance to jurors beyond the simple instruction that they need to find the presence of an aggravating factor. Jury decisions are effectively unreviewable, and abundant research shows that jurors show little consistency in how they evaluate evidence and are often swayed by irrelevant factors, such as race. Access to, and the quality of, mitigation experts varies widely, though these professionals play an essential role in a defendant’s case.
Most damningly and damagingly, the court does not require that capital attorneys have training in the death penalty. To the contrary, the bar has been set about as low as it could imaginably be set and, not surprisingly, the quality of representation varies widely. Stephen Bright, head of the Southern Center for Human Rights, says what distinguishes the cases in which the death penalty is imposed from those in which it is not “is not the facts of the crime, but the quality of legal representation.”
Carol Steiker and and Jordan Steiker, of Harvard and University of Texas respectively, say that, whatever the text of Furman and Gregg may have been, the cases were really about racism. Indeed, race is a substantial part of what got and kept Arthur Goldberg and Alan Dershowitz involved in the issue over the course of their lives. It’s certainly what attracted NAACP Legal Defense Fund lawyers Tony Amsterdam and Michael Meltsner, who together did more to shape death penalty law than any other attorneys in American history. The question here, though, isn’t whether racism continues to affect capital sentencing — it clearly does — but, rather, whether Gregg helped.
The gross numbers are disquieting. Of the 1,429 executions conducted since Gregg, 494 defendants — 34.6% — have been African-American. By comparison, in the most recent census, 13.2% of respondents identified themselves as black or African-American. So, black men — only 16 women have been executed since 1976 — are overrepresented on death row by almost triple.
In 1980, a team of researchers from the University of Iowa led by David Baldus undertook a mathematically rigorous examination of the influence of race on death penalty decisions in Georgia between 1973 and 1979, coding for 230 potentially explanatory variables. They found that blacks who killed whites were 22 times more likely to receive the death penalty than whites who killed blacks. In only two of the 2,484 cases reviewed was a white defendant sentenced to die for killing a black person. This race-of-victim effect has been re-observed time and again, in different states and over different time periods.
Baldus’s study, widely recognized as the seminal one of its kind, became the basis for an Equal Protection Clause challenge to the death penalty, which the court rejected, 5-4, in 1987. Lewis Powell wrote the majority opinion and cast the decisive vote in McCleskey v. Kemp during what would be his last term in office. Just four years later, he told his biographer that he regretted his vote, would change it if he had the chance, and that he’d “come to think capital punishment should be abolished.”
Abolitionists generally focus on Furman and Gregg as the pivotal moments in this legal history, but it’s McCleskey that did the real damage. The death penalty is a statistically trivial phenomenon. Had McCleskey ruled the use of a punishment unconstitutional because of racism in its implementation, criminal justice in the United States would have been transformed. It didn’t, because of a single, almost-immediately-regretted vote, which ended the relevance of statistical evidence of racism under the Equal Protection Clause.
Others who have looked carefully at components of the system have found profound evidence of racism. Stanford Professor Jennifer Eberhardt, expanding upon Baldus’s data, showed that in cases involving a white victim, black defendants were more than twice as likely to receive a death sentence if they had a “stereotypically Black appearance” than if they were lighter skinned, had a narrowed nose and thinner lips. If you keep one fact in your hip pocket for dinner-table arguments about the death penalty, I’d say make it this: Since Gregg, 327 Americans have been executed for interracial murders. Of these 296 were for black defendants who killed white victims. Thirty-one —9.4%— were for whites who killed blacks.
Whatever they may have written, Stewart, Stevens and Powell’s true project in Gregg was to rationalize the American death penalty and make sentencing decisions turn on the severity of a defendant’s offense instead of random factors, such as where the crime occurred, or insidious factors, such as race. On the occasion of its 40th anniversary, we can deem that project a complete and dismal failure. Virtually all murders are death eligible. States have made no effort to define capital murder with any precision.
Given what we know about the history of Gregg, and the subsequent evolution of the deciding justices’ views, it’s clear that if the evidence presented here had been available to the justices at the end of their careers, the outcome of the 1976 cases would have been a 6-3 decision rejecting the death penalty as unconstitutional. At a time when the balance may shift on a Supreme Court that may already be poised to end the death penalty, this seems like an anniversary message worth repeating.
Evan Mandery, a professor at John Jay College of Criminal Justice, is the author, most recently, of “A Wild Justice: The Death and Resurrection of Capital Punishment in America."
An earlier version of this commentary mistakenly said that rural areas use the death penalty less, not more, than cities.