INTERPRETATIONS OF REASONABLE DOUBT

THE ULTIMATE ACT OF DESPERATION: AVOIDING CLARIFICATION

The coexistence of several different definitions of BARD is not necessarily a bad thing. After all, we can define most complex notions in a variety of ways. What is troublesome is that most of these definitions fail to be characterizations of the same underlying idea. Believing firmly in or having an abiding conviction about someone's guilt is not the same thing as having no doubts for which you can give a reason. And neither of these is the same as a belief on which you would base important actions in your life. What we face are not different glosses of the same notion but different conceptions of the level of proof necessary to convict someone of a crime. To make matters worse, courts have faulted all these definitions as wrong, misleading, or unintelligible. Versions that some courts have found acceptable, even exemplary, have been dismissed by other courts as violating the constitutional rights of the accused.

This situation has prompted many appellate courts to inform trial judges that they should not define BARD for jurors in their instructions. At least ten states now insist on this. In some (e.g., Oklahoma), if a judge offers a jury any explanation of what reasonable doubt is, this is automatic grounds for reversing a conviction.³⁶

If this argument for avoiding definition will not work, there is a long line of appellate judges and legal scholars willing to find other reasons for not defining BARD. Hence, Justice Woodlock, U.S. district judge for Massachusetts, insisted that it is “a basic philosophical precept that the concept of reasonable doubt has an a priori existence in the minds of all jurors.”⁴⁰

In 1994, the Fourth Circuit Court held that, even when the jury asks the judge to define “reasonable doubt,” the judge may properly refuse to do so.⁴³

It is more than mildly curious that such a large body of legal opinion would favor a policy according to which juries must be told that conviction for a crime requires belief in guilt beyond a reasonable doubt although judges are to say nothing to clarify what reasonable doubt is, not even when the jury requests such clarification. The explicit rationale for discouraging a definition of BARD avoids admitting that the notion itself is deeply confused. The most common line is that BARD is “self-evident” or “self-defining,” and hence not in need of further commentary. Hence the Seventh Circuit has gone on record saying that the idea of reasonable doubt is so transparent that definitions of BARD, attempts—as they put it—to make “the clear more clear,” can only confuse.⁴⁵

We also have impressive, direct evidence that jurors are deeply confused about what the reasonable-doubt standard demands of them. In a study of some 600 Michigan jurors, Kramer and Koenig discovered that a quarter of jurors believed that “you have a reasonable doubt if you can see any possibility, no matter how slight, that the defendant is innocent.”⁴⁶

How can jurors rationally apply a standard of proof to the evidence when they have such divergent interpretations of that standard and when so many courts refuse to clarify the standard for them? If judges can tell jurors no more than that a guilty verdict requires belief in guilt beyond a reasonable doubt, we have reason to expect that almost every jury will contain jurors who are bringing inappropriate construals of that standard to bear. In such circumstances, failure to explain BARD to jurors is unconscionable. For such reasons, it is difficult to resist the conclusion that what really drives the growing reluctance to define BARD is not so much the purported truth that its meaning is clear to all as an abiding worry on the part of the judiciary that the persistence of conflicting definitions of BARD will bring to public notice the fact that no univocal sense can be attached to the phrase “beyond a reasonable doubt.” Rather than concede that fact and do something to remedy it, many courts—especially at the federal level—seem disposed to paper over the real differences that exist by pretending that the conception requires no further explication or clarification. A system founded on the idea that justice can be secured simply by having the judge, with a wink and a nod, tell jurors that “you know what reasonable doubt means” is in drastic need of repair.

There is doubtless another engine driving the growing disinclination of courts to instruct jurors in the subtleties of BARD. I refer to the fact that if a judge says something to the jury about BARD with which a higher court disagrees, and if the accused is subsequently convicted and the case is appealed, then the guilty verdict may be overturned. Better, apparently, to say nothing—leaving the problem unresolved, the guilty verdict in place, and the jury perplexed—than to say something that might trigger a successful appeal. A vacuous BARD instruction is not likely to be reversed, while one with content may well be. By the way, my favorite BARD misinstruction (which was properly deemed unacceptable by a higher court) comes from a New York state court. Here is how the judge instructed the jury about what a reasonable doubt is:

Where the standard of proof is concerned, our criminal justice system is blind, not in the traditional and commendable sense of being impartial but in the more literal sense of not knowing what it is doing or where it is going. The wheels of justice keep grinding, and juries keep producing verdicts, but the deliberate obscurity and obfuscation surrounding the standard for conviction does little to inspire confidence in the fairness of the system.

TAKING A STEP BACK FROM THE ABYSS

Most critics familiar with the many twists and turns in the recent handling of BARD by the courts seek the remedy for the problem in a newly crafted set of jury instructions, instructions that will in plain language explain to jurors what the idea of (beyond a) reasonable doubt means. I have to part ways with my fellow critics at this point because I believe that the proper moral to draw from this sordid story is that no canonical notion of reasonable doubt is on the table.

The sad fact is that we have been expecting one doctrine to do multiple tasks for us. To begin with, we have wanted a way to stress to jurors that the burden of proof in criminal proceedings falls on the state, not the defendant. Then we have wanted a way to warn jurors that they must not let exaggerated, hyperbolic doubts stand in the way of conviction. We have wanted to make clear that guilty verdicts in criminal trials must depend on much higher levels of proof than those associated with civil actions or practical life. We have wanted to impress on jurors just how much is at stake and how somber must be the decision when they decide to send someone to prison, depriving her of her liberty and blackening her good name. Finally, we have wanted to insure—insofar as possible—uniformity of standards, making sure that every criminal verdict hews to the same bar for conviction. On most of these scores, BARD is failing.

It is time to try to diagnose the causes of that failure. It will probably not have escaped notice that almost all the familiar glosses on BARD (as moral certainty, as firm or abiding conviction, as belief on which you would base important actions, etc.) define it in terms of the target mental state of the juror. He or she must be “firmly convinced,” “almost certain,” “fully persuaded,” with a “satisfied conscience” about guilt before a vote for conviction is indicated. It is no accident that almost all the proposed definitions focus on the subjective state of the juror. On several occasions, the Supreme Court has underscored its belief that the right way to characterize the bar for conviction is in terms of the “subjective state of mind” that jurors should be in if they are to condemn or acquit the accused.⁵⁰

On reflection, this focus on the juror's mental state is more than a little curious. Suppose that we tried to teach young scientists how to judge when a theory was acceptable by telling them what their mental state should be before they should accept a theory? Such advice would be seen as missing the point, for what matters is not their mental state per se, but how they came to that state. You become a scientist not by learning the state of mind you should be in before accepting a theory but by learning how to evaluate evidence and its bearing on theory. Likewise, doctors do not decide whether they have the right diagnosis of a puzzling disease by scrutinizing their own mental states but rather by reviewing the clues and symptoms that they have to hand and seeing whether those factors strongly support the diagnosis, according to accepted rules of theory evaluation. Pharmacologists do not decide whether a proposed new drug is really safe and effective by reaching a certain predefined mental attitude. On the contrary, they do so by hewing to accepted principles for designing and assessing clinical trials of new therapies. Imagine saying to a mathematician that he knows he has an important result when he believes it without the slightest hesitation! What establishes a mathematical theorem as a theorem is the robustness of its proof, not the confidence of its discoverer.

Even in the law, at least the civil law, we see the same pattern. When John sues Sally for breach of contract, the members of the jury are not told that they can find for John only if they are in a particular mental state. Instead, they are told to look to the evidence and see whether the evidence makes John's claims more powerful, more probable, or better supported than Sally's. When civil law borders on to criminal law (for instance, if Bill sues Bob on the grounds that Bob committed arson on Bill's warehouse), a new standard comes into play. It insists that a finding in favor of the plaintiff requires “clear and convincing evidence” that Bob did what Bill alleges. This bar, like the preponderance-of-evidence rule, characterizes what is necessary for a jury's finding in terms of the kind of evidence brought into play. These two legal standards of proof stand in sharp contrast to BARD, which refers to evidence, if at all, only obliquely and which focuses instead on the juror's state of mind.

Putting the point a bit more philosophically, we can note that, in most of the areas of life where inquiry after the truth is at stake, the usual kind of advice offered to insure that such inquiry is rational involves specifying the kinds of evidence, or tests, or proofs necessary to justify you in having a well-founded belief. Such advice explains (for instance) when an hypothesis is well- or ill-supported by the evidence or when it is proper to infer a certain conclusion from given premises. During the last two centuries, Anglo-American criminal law has generally sought to avoid saying anything about the kind of evidence needed for a conviction. (Contrast this with European, Talmudic, or Church law, which for hundreds of years specified precisely how much proof or evidence was needed to convict.)

Perhaps one reason why characterizations of BARD seem so vacuous or unsatisfactory is that they studiously avoid saying to juries anything about the structure of proof or about the case the prosecution must present in order for it to be a powerful indicator of the guilt of the accused. Instead, those characterizations focus single-mindedly on the mental state of the juror. It is true that every jury is enjoined to attend carefully to the evidence and to weigh it, but even the most general features of that process are regarded as a black box, something that happens behind the closed doors of the jury room, as if that territory ought not be invaded by a judge telling the jury what a compelling case looks like. Instead, the judge in effect simply says “Open your mind to all the evidence and then see if you are fully persuaded (or firmly convinced, etc.) at the end of the trial.” What this formula ignores, as we have noted before, is that persuasion is supposed to be a process of reasoning through the evidence. Judges do not even hint to jurors as to how they should do that. Worse, it is not even clear that jurists believe that jurors should arrive at their verdict by reasoning their way to it. If that appraisal seems harsh, remember that the highest court has repeatedly insisted that a doubt can legitimately block conviction even if jurors can give no reasons for it. David Hume famously argued that belief may simply be an “animal instinct,” beyond the control of one's critical faculties of reason and rational evaluation. By refusing to say anything to jurors about how they to reach their verdict and by exempting them from the obligation that one has in every other branch of inquiry to explain to one's peers the basis for one's conclusions, the Supreme Court seems tacitly to concur.

This would be forgivable if we could suppose that all, or even most, competent jurors already knew how to engage in the ampliative reasoning (as Peirce called it) that leads you from the disparate, purported facts of the case presented as evidence to a conclusion about the guilt of the accused. If everyone were implanted with a kind of “natural” disposition to reason correctly about complex matters of fact, such information would be superfluous. But there is nothing innate about making correct inferences from confused bodies of protoevidence. It took scientists more or less two millennia before they figured out how to test their theories correctly. The logic governing clinical trials of drugs emerged only in the twentieth century, after eons of the use of medications on only the flimsiest empirical basis. In the law itself, we have learned much about what distinguishes a powerful case for the prosecution from a weak one. I submit that the bar for conviction would be better defined in terms of the features of the case needed to convict rather than in terms of the jurors' inner states of mind, especially since the latter—if not disciplined by certain guidelines about the appropriate logical connections between evidence and verdict—are apt to be ill-founded, prejudicial, and irrational, however powerfully they may lead to a firm belief in guilt.

There is a different way of approaching the same point. Whatever else BARD might mean, it insists that juries should not convict the accused if they have a rational doubt about his guilt. Subjective criteria of the sort that appellate courts favor get at only half of this demand. Basically, they tell the juror that he must have no doubts. But precisely because the criteria they offer are purely subjective, they cannot address the question about the rationality of the jurors' confidence or lack of confidence in the guilt of the accused. That discrimination between rational and irrational doubts can never be resolved if one remains focused exclusively at the level of the jurors' degree of conviction. What distinguishes a rational doubt from an irrational one is that the former reacts to a weakness in the case offered by the prosecution, while the latter does not.

Let us remember that in a trial by jury, the jurors are supposed to be “the finders of fact.” They are the ones who have to decide whether a crime was committed and whether the defendant committed it. The prosecution will generally lay out a “theory” about the crime, that is, a narrative of supposed events in which the defendant participated. The prosecution, if it is savvy, will adduce evidence in the form of testimony or documents or physical evidence for every key part of its theory. The aim of presenting the evidence is to corroborate the prosecution's theory of the crime. For its part, the defense may (but is not obliged to) present an alternative theory of the events associated with the crime. At a minimum, the defense will seek to identify weak points or implausibilities in the prosecution's narrative. This critique may involve challenging the evidence or testimony or it may focus on the large inferential leaps required to move from the evidence presented to the thesis that the defendant is guilty.

The task of the jury is to evaluate the prosecution's theory or story, deciding whether it points unambiguously to the conclusion that the defendant is guilty. In making that decision, jurors must decide whether the evidence presented is powerful enough to justify them in concluding that the defendant committed the crime. The principal question is not whether the jurors, individually and collectively, are convinced by the prosecution. The issue is whether the evidence they have seen and heard should be convincing in terms of the level of support it offers to the prosecution's hypothesis that the defendant is guilty.

The key point is that the question: How strongly does the evidence support the theory of guilt? is an objective question about logical relations between events, not merely or primarily a question about the subjective state of the jurors' minds. The issue at trial should not be whether, as a contingent matter of fact, the twelve jurors are all fully convinced of the guilt of the accused. Instead, it should be: Does this evidence strongly support the theory that the defendant is guilty? Putting it differently: Would a rational and sober-minded person confronted with this evidence find that it made a compelling case for guilt? For jurors to be able to answer such questions, it is of little or no avail to tell them that they must be firmly convinced or have an abiding conviction of guilt. Rather, they must be able to say to themselves and to one another precisely why the evidence to hand does or does not point unambiguously to the guilt of the accused. Doing so requires them to make a whole series of decisions about whether the evidence supports each key claim of the prosecution's theory to the requisite degree.

What the jurors may well be unclear about is how to tell whether the evidence provides powerful or only mild support for the prosecution. For instance, is one credible witness to a crime sufficient to convict? Is circumstantial evidence potentially as powerful as eyewitness testimony? Are some sorts of physical evidence more damning than others? And so on. Given the practical demands of the criminal justice system, the judge cannot give a mini-course in inductive inference to every empaneled jury. Nevertheless, I think that some things could be said briefly that would provide models or paradigms of powerful forms of argument. Armed with these paradigmatic examples, jurors could go to the jury room ready to struggle with the question of whether the case in hand has the sorts of features that make up powerful proof.

Such a course of action is not wholly without precedent, although the current Supreme Court would surely fault it. Consider, for instance, what the presiding judge said to the jury in the famous 1882 case of the assassination of President Garfield:

A more compelling example of the judge instructing the jury about the logic of the decision they face comes from the way in which, until very recent times, a case resting on circumstantial evidence was explained to jurors. From early on in American criminal trials, it was common for judges to define when it was reasonable to accept circumstantial evidence as establishing guilt beyond a reasonable doubt. Indeed, for almost a century, it was a common-law rule in cases involving circumstantial evidence for the judge to give a brief explanation as to when such evidence could be taken as probative. Typically, this instruction would inform the jurors that a case based on circumstantial evidence was powerful enough to convict only if “it ruled out every reasonable hypothesis except that of the guilt of the defendant.” This instruction goes directly to the question of the nature of the evidence and avoids dalliance with characterizing how the jurors should feel about the matter. It tells jurors that they should not convict unless the evidence before them has refuted every version of the events in the case that would be exculpatory for the defendant. It says to the jurors: Figure out whether the facts established by the prosecution preclude any hypothesis you can think of that would leave the defendant innocent. If they do, you should convict. Otherwise, you must acquit.

Unfortunately (in my view), this type of instruction vanished when the Supreme Court decided, and rightly so on other grounds, that circumstantial evidence was on an evidential par with eyewitness testimony and hence required no special instruction in its own right. In Holland v. U.S., the defendant had requested the customary circumstantial evidence instruction, which the judge refused to provide to the jury. On appeal, the Supreme Court responded:

The trouble, of course, is that the Supreme Court has repeatedly held that judges need not and probably ought not “instruct jurors on the standards for reasonable doubt.” In my view, the Holland Court missed a splendid opportunity to take the traditional instruction about the character of the inference that jurors should make about cases involving circumstantial evidence (what we might call “the no-other-reasonable-hypothesis rule”) and generalize it to all criminal cases. Instead, the court has opted for the idea that the mere mention of BARD, construed as describing the subjective state of the jurors' mind, is adequate guidance for jurors.⁵³

It would be well to recall at this point that for almost two hundred years in Anglo-American law, the notion of belief beyond a reasonable doubt was identified with moral certainty. Although such certainty surely alludes to a subjective state of mind, those thinkers who were most influential in articulating the idea never left it at that. Philosophers from John Wilkins and Robert Boyle in the seventeenth century through William Whewell in the nineteenth century always identified the kinds of circumstances in which a morally certain judgment would be appropriate. Specifically, they held that you properly reach a morally certain conclusion when and only when you have several independent lines of evidence, no one of which establishes the conclusion with certainty but each of which points with likelihood in the same direction. It is important to stress that both the independence and the multiplicity of lines of evidence were required before you were entitled to hold that a belief was morally certain. A single piece of evidence (for instance, one eyewitness alleging that an event occurred or one incriminating piece of physical evidence of guilt) was never sufficient for moral certainty. Rather, what was required were two and preferably more strong indicators of guilt. Their independence was likewise important. Two witnesses likely to be in cahoots with one another would not count as two. Neither would two distinct bloodstains on the accused's knife count for more than one since we suppose that they are not independent.

The idea behind the independence requirement is that you “triangulate” in on a morally certain conclusion by discovering several distinct lines of inquiry or analysis that point to the same conclusion. Arguably, it was a rudimentary version of this intuition that provided the rationale for the famous two-witnesses-or-a-confession rule in Roman and biblical law. One witness to a crime would not be sufficient to convict, for such testimony would simply be gainsaid by the defendant's denial of guilt. But if several, noncolluding witnesses attested to the same thing, it became increasingly difficult to doubt it. Similarly, where circumstantial evidence is concerned, theorists of moral certainty stressed that such certainty could emerge only when a series of independent lines of evidence converged on the same conclusion. This kind of “proof” is by no means unique to the law. As both Boyle and Whewell argued persuasively, powerful theories in the natural sciences likewise exhibit this feature.⁵⁴

The salient point is that moral certainty or proof beyond a reasonable doubt, if we identify the one with the other, arises only in certain special situations. On this analysis, conviction of a defendant BARD requires that the prosecution's theory of the crime must enjoy support of this powerful type. Instead of instructing jurors as to what mental state they should have in order to vote to convict, courts would do better to inform jurors in a general way about the epistemic or logical features that a case must exhibit before it can properly be regarded as putting the accused's guilt beyond reasonable doubt.

I do not mean to suggest—I hasten to add—that the evaluation of a complex case can be reduced to a mechanical algorithm or simplistic rule. Most of them patently cannot. Inevitably, jurors must bring their own judgment to bear. But as things currently stand, jurors are provided with no examples of inferences that achieve BARD or moral certainty–examples that they could use as a baseline for determining whether the case in hand was powerful enough to warrant conviction. They are simply charged with determining guilt beyond a reasonable doubt when they generally have only the foggiest notion of what that requires.

I have belabored the question of the standard for conviction at such length because the ambiguities surrounding it may be the single largest impediment to insuring that the criminal justice system acts so as to minimize error and maximize truth. “Beyond a reasonable doubt” has become a mantra rather than the well-defined standard of proof that it once was. Divested of its entire substantive content by a series of well-meaning but ill-advised judicial findings, it now serves at best as an admonition to jurors not to take their tasks lightly. What it does not do in its present form is to provide the slightest clue to jurors as to what a convincing proof of guilt should look like.

IS BARD A PROOF FOR ALL SEASONS?

A question we must face now is whether, supposing jurors could be made to understand just how demanding the BARD standard is meant to be, it should be the appropriate standard to use across the board in criminal cases. At the end of the eighteenth century, when BARD was introduced in Britain and the United States, the justice system was much less nuanced than it now is. All felonies implied the same punishment (death) and all were judged to the same standard. (Misdemeanors were another matter altogether.) As a part of the penal reforms of the nineteenth century, legislatures and courts accepted the notion that the punishment should fit the crime. The range of crimes that were potentially capital shrank drastically and it continues to do so in our time. If the U.S. has not followed England's example of eliminating the death penalty altogether, it is not very far from it in practice if not yet in theory. Of the roughly 10,000 persons convicted of murder every year in the U.S., about 260 receive the death sentence and, after the process of appeals and pardons, fewer than one quarter of those will actually be executed.⁵⁵

If punishments have become drastically less severe, there has been another sea change in the justice system that is likewise relevant. I refer to the introduction in the nineteenth century of a system allowing for the appeal and reversal of convictions if appellate courts find that serious errors occurred during the trial. When BARD was introduced, the principal argument for making it such a tough standard to satisfy depended on (1) the very high price associated with the punishments for convictions; and (2) the virtual irreversibility of convictions. Mistakenly convicting someone of a felony in those times extracted a very high price. It was natural under such circumstances to set the bar for conviction extremely high for all crimes.

It is appropriate to ask whether changed circumstances ought not alter the calculation we do when comparing the relative costs of false acquittals and false convictions. Given that (statistically speaking) the death penalty has vanished, and given that guilty verdicts are routinely vetted by higher courts, does it remain plausible that it is better that ten (or twenty or ninety-nine, depending upon which eighteenth-century authority you cite) guilty persons go free than that one innocent person is convicted? The question takes on particular force when we are dealing with felonies that have quite mild punishments associated with them.

Clearly, the costs of false convictions for most crimes are now a great deal less than they once were. Since there is elaborate appellate machinery for catching and canceling the sometimes-deadly effects of many such convictions, an incorrect conviction is no longer the guarantee of a very nasty fate that it once was. Despite this sea change in the openness of convictions to review, we are still hostage to the clichés of an earlier era when justice was both harsh and swift and where the jury was the final arbiter of guilt. We continue to mouth slogans about the egregiously high costs of false convictions—slogans that once made a lot of sense—even though the introduction of appellate machinery and of a graded scale of punishments deprives those slogans of much of their conceptual and emotive rationale. A conviction in a court of law is no longer the irrevocable decision by society that it once was. Moreover, as we just saw, a felony conviction no longer implies the gallows except in very rare cases. This change to the once-irrevocable character of convictions has, I suspect, yet to be factored into the question of how we think about the relative costs of false convictions and false acquittals.

Does it make sense to have a one-size-fits-all standard of proof when the costs associated with an erroneous finding of guilt are as low as they now are in many cases? It is common knowledge that a great many crimes now carry punishments that are little more than fines. Others, including felonies, can involve nothing more than probation or relatively brief times of incarceration. Are the costs of mistaken guilty verdicts in such circumstances so steep that we still want (say) ten or a hundred guilty defendants to be acquitted for every innocent defendant convicted? One's hunches about the relative costs of mistakes, perhaps crystal clear in the case of capital crimes, become clouded when it we turn to much milder punishments. Such differences invite the suggestion that the standard for conviction, instead of being the same for every crime from homicide to shoplifting, should—as the punishment does—vary with the severity of the crime. After all, many crimes now carry punishments less harsh than one's potential liability in civil cases, where the standard of proof is simply “more probable than not.” In a civil court, for instance, you can be sued for your life savings if you are shown to be probably liable for harm to some third party. Civil proceedings can—on the strength of a proof that is merely more probable than not—deny someone their parental rights, commit them indefinitely to a mental institution, and deprive them of their citizenship. Are these fates less severe than being found guilty of felony drunk driving and facing a year's probation? If they are not, what is the sense in holding to such different standards of proof in the two cases?

It is also worth asking whether it is an efficient use of money and other resources to require the state to mount the same sort of proof to send an embezzler to jail for six months as it invests to convict a serial killer. Part of the problem, of course, is that it is not easy to see how to construct a graded scale of standards of conviction that might be mapped on to a scale of criminal severity. Essentially, the U.S. court system recognizes only three degrees of proof: BARD, proof by a preponderance of the evidence, and proof by clear and convincing evidence. If courts were disposed to think of standards of proof as probabilities (which they are not), you could imagine at least in principle having a graded series of probabilities associated with increasingly serious punishments, perhaps reserving BARD for only the most serious.

Since that proposal would probably go nowhere—given the courts' (probably correct) aversion to thinking of standards of proof as degrees of probability—a more realistic option would be to use the intermediate standard of proof by “clear and convincing evidence” for lesser crimes, by which I mean not only misdemeanors but also the less serious felonies. This standard, like BARD, is defendant-friendly in that it requires the state to establish much more than the bare probability of guilt, and it has the added virtue of presumably being intelligible to juries and familiar to judges, since civil trials already make frequent use of it. It is, moreover, fully compatible with the much-prized principle of the presumption of innocence. Such a proposal would require the Supreme Court to reverse its famous opinion in In re Wingate that all criminal defendants are entitled to be tried according to the BARD standard, but (as we have already seen) there are powerful, independent reasons for believing that the framers of the Constitution never contemplated laying down a specific standard for criminal conviction, let alone contemplated using BARD, not the least of which is that BARD was not even entrenched in the common law when the Constitution was adopted.

An ideal case in point of the anomalies produced by an omnibus standard of proof is offered by In re Wingate itself. In this case, a juvenile of twelve was facing a hearing to determine whether, having committed larceny, he should be sent to reformatory for eighteen months. According to then prevailing New York state law, decisions in juvenile hearings could be settled using the preponderance-of-the-evidence standard. The costs associated with erroneous incarceration under these circumstances were relatively modest. Unlike a normal conviction for theft, the juvenile court's finding would not form part of the defendant's criminal record. His incarceration would not involve a loss of any citizenship rights or privileges. Because the juvenile hearing was conducted in private, there would be no otiose public stigma associated with its outcome. All this notwithstanding, the Supreme Court declared the New York provision for settling such matters by a preponderance of the evidence unconstitutional. In essence, the Court denied the relevancy of a calculation of the costs of error to a decision about the standard of proof.

This is more than mildly curious since, for more than a century, the Supreme Court itself had justified a standard as tough as BARD precisely on the grounds that the costs of a mistaken conviction are so enormous that the bar for conviction must be set extremely high. Here, confronted by a case in which the costs were no more severe than those imposed in many civil trials, the Court went out of its way to insist nonetheless on BARD's application across the entire spectrum of punishments, saying that due process required as much. With respect, I confess to finding this ruling unintelligible. If the motive for a tough version of BARD is the importance of avoiding mistakes that can exact very high costs, and no one denies that, then there can be no rationale for the insistence on BARD in the adjudication of cases in which the costs of error are comparatively modest. The same twelve-year-old whom the Court protects from a reformatory with BARD can be sent to a mental institution or denied parental visitations or taken from his parents and put into foster care or an orphanage on the strength of the preponderance of the evidence. The attractiveness of the clear-and-convincing-evidence criterion is that, being intermediate between BARD and a preponderance of the evidence, it seems tailor-made for cases of this sort.

Against my suggestion it can be argued, not that it is wrongheaded to link severity of punishment with standard of proof, but rather that it is superfluous. We have already seen that BARD is generally left quite vague by the courts, usually deliberately so. That means, of course, that jurors have plenty of room for discretion in deciding just how demanding BARD is in any particular case. Several scholars, most especially Erik Lillquist, have argued that jurors will adjust the standard of proof to the circumstances of the case and of the defendant. Making a virtue out of the necessities imposed by the ambiguities of BARD, Lillquist holds that: “The use of a flexible standard allows the decision maker [that is, the jury] to apply the level of certainty that is most appropriate to a particular case.”⁵⁶

While intriguing, these studies do not show that jurors exploit the ambiguities in BARD so as to vary the standard of proof with the punishment for a crime. On the contrary, as Lillquist concedes, leaving capital cases aside, “there is also reason to believe that the severity of the punishment has no actual effect on jurors' decision to convict or acquit.”⁵⁸

THE UNCERTAIN REACH OF THE BENEFIT OF THE DOUBT

If it is plausible to suggest that BARD is not an appropriate standard for judging all crimes, it is equally important to stress that BARD, or something like it, is appropriate for serious crimes. Curiously, despite the Supreme Court's recent fetish for BARD, it has allowed BARD's role, even in serious cases, to atrophy. The result has been that many crucial decisions about a defendant's fate hinge on much weaker standards of proof. These weaker standards are especially clear in the case of so-called “affirmative” or “excuse” defenses.

When someone is charged with a crime, his defense may consist in offering an alibi, pleading diminished responsibility, or insisting that he was acting in self-defense. Any one of these excuses, if true, should lead to an acquittal. The question at issue is: Who carries the burden of proof with respect to these defenses? Speaking in general terms, American criminal law permits the judge to instruct the jury—in cases where such defenses arise—that acquittal requires that the defendant must show “by a preponderance of the evidence” the truth of his excuse. This policy is, in my view, a flagrant violation of both the spirit and the letter of BARD.

If, for instance, defendant is charged with the robbery of a jewelry store at 1 PM on June 3 and he offers testimony from those claiming that he was in another city at the appointed hour, the testimony in question, the alibi, should only have to be strong enough to raise a reasonable doubt in jurors' minds about the prosecution's claim that the defendant was the felon. The demand that the credibility of the alibi must reach the civil standard of proof before the jury can credit it is plainly inconsistent with the claim that the prosecution must prove defendant's guilt beyond a reasonable doubt.

What is curious here is not so much that the defendant propounding an affirmative defense is obliged to assume an evidential burden—that seems fair enough, since the prosecution has (we are supposing) already established what it needs for a conviction. Leaving the alibi case to one side, the defendant is attempting to block a conviction not by disputing the claims against her but by putting them in a larger, exculpatory context. It is reasonable that she should have to do something to make her claim of self-defense or insanity plausible. What grates on one's epistemic sensibilities slightly, however, is the level or quantum of proof that the defendant's case must achieve. In general, a defendant is supposed to be acquitted unless the evidence of her guilt reaches the level of BARD. You would have thought, therefore, that all that was required of an affirmative defense was that it should establish some modest probability of its truth, enough to raise a reasonable doubt about its denial. Instead, state legislatures are at liberty to insist (and many do) that affirmative defenses must be more probable than not before a jury is instructed to use them to acquit in an otherwise airtight case for the guilt of the defendant.

In re Wingate supposedly established that every criminal defendant was entitled to a not-guilty verdict unless every element of the state's case against him was proved BARD. The epistemology of affirmative defenses seems to violate that doctrine. A juror cannot say to himself, “Although the accused clearly committed the crime with which she is charged, there are some grounds to suspect that she acted in self-defense.” (Well, the juror can say it but he will be instructed in many jurisdictions that such a belief is not an adequate basis to acquit.) Acquittal, he will be told, requires that he believe the self-defense hypothesis to be more likely than not. The tensions here pose an imminent threat of an invalid verdict, in the sense of a verdict incompatible with BARD. If a juror has reasonable suspicions that the defendant acted in self-defense, then that juror cannot be said to believe in the guilt of the defendant BARD, since guilt already presupposes that there was no element of exculpatory justification in the act of the defendant. Nonetheless, the juror will be instructed that he must convict unless he positively believes (instead of merely suspects) that the defendant acted in self-defense.

So long as legislators are left free to specify a standard for affirmative defenses that is weaker than BARD, it will be possible to bypass Wingate's insistence that defendants have a constitutional right to have their guilt proved beyond a reasonable doubt. I have earlier expressed doubts as to whether BARD is not too demanding a standard for relatively minor crimes. Nevertheless, those doubts do not extend to the question of whether, for a crime for which the BARD standard is appropriate, that standard should be scrupulously followed.

Similar worries arise with respect to the appellate stage in judicial proceedings. When, for instance, someone convicted of a crime appeals the conviction on the grounds that the jurors were wrongly instructed by the judge about some key point of law, the appellate court requires the appellee to show “a reasonable likelihood” that his conviction was due to the improper instruction.⁵⁹