Table of Contents Introduction I. Execution Scarcity A. Decoupling Death Sentences from Executions B. Institutional Coordination and Political Will C. Scarcity and Arbitrariness II. Sorting Dissensus A. Upstream Sorting B. Downstream Dissensus 1. Sorting dissensus and law 2. Sorting dissensus and norms III. Evaluating Sorting Criteria A. Blame B. Incapacitation C. Deterrence D. Vindication Surplus and Revenge Utilitarianism E. Accuracy IV. Institutional Design A. Design Principles 1. Centralization 2. Side-constrained rulemaking 3. Separated selection powers B. Nonarbitrariness and Legitimacy 1. Nonarbitrariness (equality) 2. Legitimacy Conclusion Introduction
There are two American death penalties. The first is the process by which a suspected perpetrator is sentenced to death: a homicide, law enforcement's pursuit of a suspect, a jury trial before peers, and a capital sentence expressing an aggrieved community's judgment for a heinous crime. This process inevitably involves arbitrary decisionmaking, but professional communities and legal institutions have developed norms and laws responsive to that problem. At least theoretically, prosecutors capitally charge only perpetrators of the most heinous crimes, (1) and juries levy death sentences on only the most deserving offenders. (2) In producing death sentences, U.S. jurisdictions (3) expend considerable resources sorting the "worst of the worst" from the "worst of the really bad."
The second American death penalty--and the topic of this Article--is the process by which inmates sentenced to death are actually selected and sequenced for execution, a process I refer to as "execution selection." U.S. jurisdictions maintain no pretense of avoiding arbitrary decisionmaking when they construct their execution queues. There is no federal law on execution selection, and state law usually governs only ministerial authority to seek and set execution dates. (4) Norms and legal rules tell us how the bell will toll, but almost nothing about when or for whom. There are virtually no constraints on execution selection, inviting a raft of questions about the role of chance and arbitrary decisionmaking in the most visible acts of state killing. As a descriptive matter, why is the process permitted to remain so under-regulated? As a normative matter, should execution selection even be analyzed as "punishment"? And how should institutions respond?
To analyze these questions, I conceptualize the modern American death penalty as a sequence of selection phases. After a homicide, law enforcement uses arrests to select suspects (5) and then uses the charging process to select certain arrestees for capital prosecution. (6) Judges, juries, and lawyers then select capitally prosecuted defendants for death sentences by determining guilt and punishment. (7) There is virtually no literature on the final phase, execution selection, for many of the same reasons that norms and laws fail to constrain it. (8) Execution selection is a term in desperate need of coining because almost nobody even thinks of it as a thing. The visible parts of American capital punishment are the crime, the arrest, the trial, and the ritualized drama preceding the execution itself. (9)
Important clues explaining the arbitrary construction of the execution queue come from literature about when society tolerates--and even nurtures-random decisionmaking. (10) At the final selection phase, the differences among eligible inmates are either incommensurable or, although commensurable, too small to be reliably ordered using available sorting criteria. (11) Phrased a bit differently, execution selection remains insufficiently constrained because, among other things, institutions cannot agree on sorting tools and, in any event, the available sorting tools are too crude to produce a reliable ordering of execution priority. By imposing some basic rules on the institutional design of execution selection, however, U.S. jurisdictions can suppress the arbitrariness that flourishes in the void.
I explain in Part I why execution selection happens at all--that is, why the inability of U.S. jurisdictions to execute every inmate on death row produces what one might call "execution scarcity." The first reason involves the separation of death sentences and executions into legally and temporally distinct events. (12) Executions do not immediately follow death sentences because an increasingly protracted legal process must run its course. Still, even if executions are substantially delayed, something else must explain why so few are ultimately carried out. The second cause of execution scarcity involves a coordination problem and the absence of what Carol Steiker and Jordan Steiker have referred to as "political will." (13) Most states divide the responsibility for execution selection across multiple, frequently local, institutions. (14) Dispersed institutional responsibility elevates the level of coordination necessary to conduct executions at precisely the moment when reserves of political will to execute are running dry. In this perfect storm, newly condemned inmates will be only slightly more likely to die in an execution chamber than they will be to expire of natural causes or to kill themselves. (15)
If the state cannot execute everyone it sentences to death, it must make choices about which condemned inmates it should kill. In Part II, I shift my focus to a different question: Why are there no criteria for choosing? The major reason that execution selection remains substantively unconstrained is that there are few broadly shared intuitions about how executions should be prioritized. (16) What I call "sorting dissensus" persists in large part because most of the morally significant prioritizing--sorting by reference to offense conduct, blameworthiness, or future dangerousness--has already been performed by the upstream phases of the capital punishment sequence. (17) Sorting dissensus, as I use the term, simply describes a universe of sorting preferences that is highly differentiated.
Sorting dissensus frustrates the development of both legal rules and practice norms that might otherwise impose an order of execution priority. Moreover, the political economy of the execution phase compounds the natural effects of dissensus. Political beneficiaries of flexible execution selection laws tend to be incumbent political entities, so those stakeholders leave that flexibility intact. Norms of execution selection are slow to develop in environments where the decisions to which the norms would apply are infrequent, where the decisionmaking is opaque, where extreme differences in selection processes impair the transmission of occupational practices across jurisdictions, and where regular turnover at the top of local institutions wipes out predecessor practices.
Whereas Parts I and II are explanatory, Part III is a normative inquiry into sorting criteria that might be used to construct the execution queue--most of which turn out to be unworkable. For reasons I explain, blame-based criteria involving offense conduct and culpability fail to sort inmates in ways that are both desirable and workable, as do criteria based on deterrence, incapacitation, and what one might call "vindictive satisfaction." (18) The fairer criteria tend to be insufficiently precise, and the more precise criteria tend to be insufficiently fair. The one sorting criterion that both captures an important value and lends itself to sensible administration is accuracy. Executions involving inmates subject to convictions and sentences that may be wrongful should be deprioritized. More precisely, jurisdictions should deprioritize executions involving inmates whose cases are in certain procedural postures--inmates for whom available judicial proceedings are sufficiently likely to produce an authoritative legal declaration that a conviction or sentence was in error.
I devote Part IV to questions of institutional design, based on the normative premise that no set of constraints can entirely eliminate decisionmaking that is random, but that the system should not gratuitously facilitate decisionmaking that is arbitrary. First, U.S. jurisdictions should centralize execution selection; the appropriate role of local stakeholders is exhausted during the upstream selection phases, and local involvement in the execution phase does nothing but distort the process in ways that are arbitrary, opaque, and inefficient. Second, jurisdictions ought to use a process of informal (noticeand-comment) rulemaking, like in administrative law, to identify sorting criteria, subject to an accuracy-related side constraint. Third, jurisdictions should separate, on the one hand, the power to schedule executions, and, on the other, the power to determine the identity of those executed. Such selection structure promotes interests in equality and legitimacy that are compromised by the existing chaos in capitally active jurisdictions.
The risk of random and arbitrary treatment arises when human institutions must allocate burdens and benefits incapable of equal distribution. (19) If an insufficiently divisible burden or benefit is also scarce, then it is not possible for each eligible recipient to receive a pro rata share. (20) The United States, for example, has used randomization strategies to distribute the burdens of wartime military service and the benefits of immigration visas. (21) The modern execution belongs in this category. It is obviously indivisible insofar as its primary social burden is borne entirely by the person executed; there is no sharing (22) In this Part, I explain that executions are also scarce, and that such scarcity produces a decisionmaking climate vulnerable to random and arbitrary choices.
It is easiest to start with the top-line numbers. U.S. jurisdictions do not execute everyone they sentence to death. For example, as of mid-2018, there were 742...