Improving Accuracy in Criminal Cases
Erik Lillquist *
The past eight years have been marked by a huge growth in concern about accuracy in our criminal justice system, particularly in death penalty cases. To a large extent, this concern has been motivated by anxiety over the possibility of executing innocent persons, although that has not been the only worry. The resulting unease has been shared not only by legal academics and death penalty opponents, but also by many other actors in the criminal justice system. Most notably, the possibility of wrongful convictions has led a number of states to form commissions and other bodies to study the problem and recommend changes, in order to avoid erroneous convictions in the future.
Unlike some commentators and jurists, I assume erroneous convictions must exist, and the number of such cases is not insignificant. The criminal justice system is a human enterpriseâ€”one in which we ask a group of twelve people to make a decision about the guilt or innocence of a particular individual. Like any other human epistemic practice, this process inevitably leads to mistakesâ€”both erroneous convictions and erroneous acquittals. If nothing else, the experiences in Illinois and elsewhere make it vividly clear that trial juries occasionally convict the wrong person. This realization should not be surprising. After all, if historians of science can readily show that every scientific theory of the last two millennia has proven to be wrong in some aspect, should we really be surprised that our criminal justice epistemic practices sometimes lead to inaccurate results?
Knowing that we will always convict the wrong person on occasion does not mean that we should cease attempting to improve the criminal justice system, and the death penalty system in particular. Despite repeated failures, scientists have not given up improving their theories, and there is no reason that those of us engaged with the criminal justice system should cease trying to improve it.
My goal in this article is to address which, if any, of the proposals advanced in recent years are likely to improve the performance of the criminal justice system, particularly in death penalty cases. In considering this question, though, my focus will differ from that taken by some commentators. I will assess not only whether a proposed change is likely to decrease the number of erroneous convictions, but also the proposal’s likely effect on erroneous acquittals. As I have argued at some length elsewhere, any honest account has to weigh both erroneous convictions and erroneous acquittals.
There are two proposals I will not address here. One is the suggestion that the standard of proof in death penalty cases be increased above the proof beyond a reasonable doubt standard. I have asserted elsewhere that such a change would be unwise, and readers interested in that issue are better served by reading the arguments I make there. I also will not address the proposal to end the death-qualification of juries at the guilt stage. As I and others have noted, present practice inevitably leads to juries that are inherently biased toward conviction, but there is no reason to rehash those concerns here.
This article proceeds in three parts. In Part I, I survey some major, recent proposals to improve the accuracy of the criminal justice system. In Part II, I briefly set forth a framework for addressing the epistemic worthiness of any particular proposal. Then, in Part III, I address which, if any, of the proposals are likely to have beneficial effects. Finally, in the Conclusion, I briefly note one often overlooked reformâ€”eliminating junk science from the courtroomâ€”that could go a long way toward improving accuracy.
My conclusions are that some, but by no means all, of the suggested reforms would make sense. But the point I wish to make here goes beyond the efficacy of one reform or another. Too often, in an effort to avoid erroneous convictions, advocates suggest solutions that will make it difficult to obtain a conviction at all. While such an approach will (usually) reduce the number of erroneous convictions, it actually may vastly increase the total number of errors that the system makes. If so, the reform is both unwise and unlikely to be adopted. It is only by focusing on improving the functioning of the system as a wholeâ€” and trials in particularâ€”that the problem of erroneous convictions may be solved.
I. Recent Proposals for Improving the
Criminal Justice System
It is well documented that the present concern about the inaccuracy of the criminal justice system was ignited in large part by the uncovering of numerous erroneous convictionsâ€”many in capital casesâ€”in Illinois. In particular, between 1977 and 1999, Illinois exonerated thirteen persons who had been condemned to death (while executing only twelve during that time period), leading then-Governor George Ryan to impose a moratorium on the death penalty in that state. The moratorium in Illinois was eventually followed by similar suspensions in Maryland and New Jersey. Despite pressure from the American Bar Association and other groups, however, there has been no nationwide spread of the suspension of the death penalty.
More influential was Governor Ryan’s creation of the Commission on Capital Punishment to study the death penalty system in Illinois. Numerous states have followed with commissions or study groups of their own, including Arizona, California, Connecticut, Maryland, New Jersey, and Wisconsin, although not all of these states have focused their groups on the death penalty. In addition, one state that does not have the death penalty, Massachusetts, formed a gubernatorial council to examine what procedures should be in place when, if ever, that state reintroduced the death penalty. At this point in time, some of these states have completed their reports, while others are still ongoing.
In addition, numerous organizations and commentators have issued reports and recommendations concerning how to prevent wrongful convictions. Among these groups have been the American Bar Association, the North Carolina Center on Actual Innocence, and the Innocence Commission for Virginia.
The various state commissions and other groups have made a blizzard of recommendations for changes to police, prosecutorial, and judicial procedures in death penalty cases. Some of the common proposals address concerns that are more systemic and less epistemic, at least in my view. Among these are the frequent calls to increase funding and training for capital defense counsel and to change appellate and post-conviction review procedures. In addition, many states, following the lead (in part) of the federal government, have undertaken to study the effect of race and other factors on the imposition of the death penalty. Finally, other states have investigated the costs associated with the death penalty.
My concern here, though, is with the proposals that have been targeted directly at eliminating erroneous convictions. Prominent among these suggestions are requirements that interrogation of suspects be audio taped or video recorded, line-up procedures be reformed, discovery procedures be liberalized, the defendant be allowed to select a new jury for the capital stage, a heightened standard of proof be used in death penalty cases, and the use of jailhouse informants be limited. Finally, some have called for the creation of innocence review commissions.
What unifies almost all of these proposals is a focus on improving the accuracy of the decision of the jury at the guilt stage of the process. As I noted in the Introduction, I have addressed two of these issues previouslyâ€”increasing the standard of proof and jury selectionâ€”and I will not repeat those arguments here. Of course, innocence commissions are not meant to improve the function of the jury; they are instead meant to review and perhaps alter the conclusions reached by juries. Nonetheless, I include them in this study because, unlike appeals and post-conviction review, innocence commissions are a mechanism designed only to improve the criminal justice system’s performance on innocence and guilt determinations; other legal issues unrelated to guilt are intended to play no part in their decisions. Thus, what hopefully unites these proposals is they are all justified by their proponents on the principal ground that they will, if enacted, improve accuracy.
II. How Do We Know Such Proposals Are Effective?
Just because each of these proposals would decrease the number of innocent persons convicted does not imply they are a good idea. After all, the best way to safeguard against wrongful convictions would be to eliminate convictions altogether. Obviously, no one actually suggests this is a legitimate solution to the problem because society recognizes that one important function of the criminal justice system is to convict the guilty.
This balance between ensuring the guilty are convicted and the innocent are acquitted frequently stymies efforts of reform. One concern of prosecutors, judges, and others is that while a particular change may avoid some wrongful convictions, it might do more harm than good because of the additional number of guilty persons who are acquitted as a result. Therefore, an important consideration in deciding whether to implement the reforms that have recently been advocated is whether they will, on balance, aid or harm the criminal justice system.
Of course, one important factor is whether the particular reform is an epistemic improvement on existing practices. For instance, moving from a system that randomly decides whether the defendant is guilty or innocent to a system based on evidence of the defendant’s guilt or innocence is an obvious improvementâ€”not only does an evidence-based system convict fewer innocent persons, but it also convicts more guilty persons.
Some of the reforms proposed to date are such improvements. Many, though, are not. For instance, increases in the standard of proof required for conviction clearly makes it harder to wrongfully convict, but also easier to wrongfully acquit. In such cases, knowing whether the proposal improves overall accuracy requires a balancing between the effect of the reform on both wrongful convictions and acquittals.
At first glance, such a balancing might seem hopeless. Our oft-stated goal is that defendants should be given the benefit of the doubt to a large degree; that is, the number of erroneous acquittals should outweigh the number of erroneous convictions, usually by a ratio of 10:1, although some commentators have suggested higher numbers. Accordingly, it would seem that for each reform that reduced the risk of wrongful conviction, but increased the number of erroneous acquittals, we need to know: (1) the proper error ratio, and (2) whether the proposal generates a ratio that is above or below that proper error ratio. Coming to agreement on the first issue may be impossible, and accurate information about the second may be quite hard to come by.
The task is not so hopeless, however, because the approach detailed above is misguided. Professor Larry Laudan has recently argued that the only proper place to effectuate the benefit of the doubt is in the standard of proof. Giving the benefit of the doubt to the defendant in other contexts has two primary vices. First, it gives the defendant too large a benefit, because the defendant has already received the benefit of the doubt in the standard of proof. Giving the defendant the benefit of the doubt in other evidentiary practices only makes it harder to convict the defendant and, therefore, may have the perverse effect of encouraging a decrease in the standard of proof required to convict.
Second, giving the benefit of the doubt to the defendant on evidentiary issues is unfair for some defendants because such a benefit only aids defendants who are subject to the particular evidentiary practice. For instance, if we make it harder to admit expert evidence by prosecutors but not by defendants, this will undoubtedly make it harder to erroneously convict only those defendants against whom expert testimony is sought. It provides no benefit to defendants in cases where there is no expert evidence sought by the prosecution.
Assuming that Laudan is right about this, and I believe he is, then the only inquiry is whether the practice will lead to more, or less, erroneous verdicts in total. This inquiry is not simple, but strikes me as less complicated than the one I noted above. At a minimum, we should be able to get some rough idea about whether the proposal may be unwise based on this metric.
There is one other complication, though. Changes in practices have upstream effects; they may alter not only how juries decide cases, but also how police investigate and prosecutors decide to pursue various crimes. For instance, assume a requirement that no hearsay evidence be admittedâ€”even under a hearsay exemption or exceptionâ€”was adopted because it was shown that the absence of hearsay would improve accuracy across the board in criminal trials. Nonetheless, the rule might be undesirable if the effect was that each trial became longer. Unless the amount of resources expended on the trial system increased proportionatelyâ€”a rather optimistic assumption, I believeâ€”each trial would become more expensive, leading ultimately to fewer trials. If we assume, as many do, that fewer trials means less accuracy in the system, then the effect of this change might not be to increase overall accuracy. Thus, in determining whether a proposal makes epistemic sense, we have to be mindful not only how it affects trials, but how it affects the accuracy of the system altogether.
III. Do the Proposed Reforms Work?
The preceding Parts discussed both kinds of reforms that have been suggested to create more accuracy in criminal trialsâ€”particularly in death penalty casesâ€”and the sort of standards that might apply in deciding whether these reforms truly improve accuracy. In this Part, I turn to the critical issue of whether any of the proposed reforms work. My conclusion is that many of the reforms are positive changes; they not only avoid unjust convictions, but unjust acquittals as well. However, not all of the proposed reforms are positive. Indeed, I will start by considering one change that, while superficially appealing, may have some negative long-term consequences.
I will also address the issue of implementation. In several cases, the difficulty is not with the reform itself, but the means of enforcing commitment to the new norm. In particular, will an exclusionary remedy apply, or will the courts just use limiting instructions? As I will note, this is a particularly thorny problem in several areas.
A. Innocence Review Commissions
Innocence review commissions and related institutions seem, on their face, to be utterly uncontroversial. Their role is to take a second look at the evidence against a defendant and determine, presumably based upon additional information, if the defendant is in fact innocent of the crime. North Carolina has recently become the first state to implement such a commission, the North Carolina Innocence Inquiry Commission (“N.C. Innocence Commission”). This commission is empowered to investigate claims of factual innocence as to persons presently incarcerated in North Carolina, provided those persons waive their “procedural safeguards and privileges” to any claims of innocence. If the N.C. Innocence Commission determines that “there is sufficient evidence of factual innocence to merit judicial review,” the Commission is to request that the North Carolina Chief Justice appoint a special three-judge panel to conduct a hearing. If that panel unanimously agrees that the claim of actual innocence is proven “by clear and convincing evidence,” the person shall be declared “innocent of the charges.”
England takes a similar approach. Under English law, the Criminal Cases Review Commission has the power to refer cases of possibly erroneous conviction to the Court of Appeal if the Commission believes “there is a real possibility that the conviction . . . would not be upheld.” The Court of Appeal has been given the power to either reverse a conviction or order a new trial in a case where the evidence was legally sufficient, but the court believes the guilty verdict to have been “unsafe and unsatisfactory.” The court can exercise this power based either on a review of the cold record or based on additional evidence. My colleague, Professor Michael Risinger, has recently suggested that American courts should either adopt or be granted similar authority.
Such proposals have an obviously intuitive appeal. By focusing only on claims of actual, factual innocence, they are narrowly tailored to free only those who are truly innocent. In other words, unlike many other procedural reforms, they would seem to pose little to no risk of freeing the guilty. The net result is an increase in accuracy; no guilty persons are freed, but innocent persons are.
There is, however, a lurking problem with such programs. They are, by nature, a one-way ratchet. Their function is to lower the number of erroneously convicted innocent persons, not to address the problem of erroneous acquittals. Of course, this would not be a problem if we could confidently assert that innocence commissions and the like will have no effect on erroneous acquittals. But that is unlikely to be true.
One concern about innocence commissions is the possibility that they might release a guilty person. Just as we imagine that the results of trials must sometimes be wrong, so too must we assume that innocence commissions will err at some point. Nonetheless, I believe this problem is fairly trivial. In both North Carolina and England there is a high burden placed upon the convict to demonstrate his innocence. For instance, in North Carolina, the panel of judges must be unanimously convinced “by clear and convincing evidence” of the defendant’s innocence. Similarly, under English law, the Court of Appeal must determine that the verdict is “unsafe.” While this term may sound like an easy one for a convict to meet, as interpreted by the English courts, it has generally been quite difficult. Because of this high burden, it is likely that far more erroneous than accurate convictions will be set aside by innocence commissions.
There are other, more legitimate, reasons to worry about the effects of innocence commissions. The first is the possible impact on existing criminal cases. As I have noted elsewhere, there is little question the proof beyond a reasonable doubt standard is a floating standard, not a fixed rule. Among the things that are likely to influence how much certainty a set of jurors requires before convicting the defendant are the costs and benefits of acquittals and convictions. From a juror’s perspective, the existence of innocence commissions may serve to lower the costs of erroneous convictions. As a result, jurors may lower the amount of proof they require. After all, if one of the reasons for the high standard of proof in criminal cases is that the costs of an erroneous conviction are so high, anything that lowers the costs of erroneous convictions may well lower the standard of proof at trial.
Furthermore, there are reasons to worry that this trade-off between the availability of innocence commissions and a decreased standard of proof at trial will make things worse, not better, for innocent defendants. The additional innocent defendants convicted under the “new” standard of proof will often be unable to gain relief from an innocence commission. The result is an increase, not a decrease, in the number of erroneous convictions.
An example here may help. As a thought experiment, imagine that jurors presently require 90% certainty before convicting a defendant, but that after the imposition of an innocence commission, that percentage will be lowered to 80% (because the existence of an innocence commission makes them less worried about the possibility of an innocent defendant being executed). Now, when the likelihood of a defendant’s guilt appears between 80% and 90% certain, he will be convicted instead of acquitted. But as I have noted, for the court to reverse the conviction after referral from the commission, there will have to be substantial doubt about the now-convicted person’s guilt. It will have to appear to the court that the certainty of the convict’s guilt is, say, less than 50%. Are all of those convicted under the new standard likely to be picked up by this? Or more important, are the total number of resulting exonerations likely to be greater than the increased number of convictions? I think the best that can be said is that, to date, the answer is unclear.
The second reason to worry about the creation of such commissions concerns resources for the criminal justice system more generally. It is common to assume that such resources are generally fixed. The creation of innocence commissions necessarily shifts those resources from other sources to the commissions themselves. If it is true that there are only fixed resources available to the system, some judges and lawyers will be made unavailable to the rest of the criminal justice system. The most likely result of this development is a decrease in accuracy. For instance, both the North Carolina and English systems require judges to ultimately decide whether or not to release the convict. The creation of the innocence commission therefore increases the judges’ workloads. Assuming that the number of judges is not increased to accommodate the creation of the commissions, judges can respond in one of two ways: (1) by increasing the total amount of work they do (I believe this is unlikely), or (2) by decreasing the amount of time they spend on other cases (much more likely). If even one judge chooses the second option, the investment of less time in those other cases will result, we can imagine, in less accuracy in deciding those cases (assuming, of course, that the investment of time is positively correlated with accuracy).
One remedy to the one-way ratchet problem would be the introduction of guilt review commissions. But the Double Jeopardy Clause, at least as interpreted by the Supreme Court of the United States, effectively forbids such commissions. Nonetheless, it is not a complete surprise that England, after eight years experience with the Criminal Cases Review Commission, took the radical step of rolling back its double jeopardy protections in serious criminal cases. In the Criminal Justice Act 2003, Parliament provided that a new trial may be had in serious criminal cases where there is “new and compelling evidence against the acquitted person.” The result, of course, is to ensure that more actually guilty people are convicted.
Of course, just like the innocence commissions, eliminating double jeopardy protections has some potential downsides. For instance, the existence of this possibility may make jurors less likely to convict. But perhaps, when taken in conjunction with the creation of the Criminal Cases Review Commission, this would only reassert the traditional standard of proof. Slightly more troubling is the requirement that the government petition the Court of Appeal before being allowed to re-indict, and that a conviction may not follow unless there is a retrial. Both of these things result in a shift of resources and a possible decrease in accuracy elsewhere in the criminal justice system.
On the whole, we cannot be sure whether the creation of innocence commissions will create more or less accuracy in our criminal justice system. And, to be clear, I have no fixed view myself. But I think it is worth noting that such proposals are far from the accuracy-enhancing panaceas as they are sometimes portrayed. And, over the long term, I believe that the increased willingness to review convictions will in turn lead to pressures in the United States to revisit acquittals as well.
B. Discovery Practices
As even a casual observer of American litigation is aware, there is a much smaller amount of pretrial discovery available in criminal cases than in civil cases. This disparity has led to a long-standing debate on whether discovery rules in criminal cases should move closer to the rules used in civil cases, and, over the past fifty years, the amount of discovery permitted in criminal cases has indeed increased. Nonetheless, the limitations on discovery have concerned observers of the criminal justice system, and it is no surprise that some proposals for improving the criminal justice system have focused on the possibility of further liberalizing criminal discovery.
The debate over the past fifty years has generally acknowledged that increasing the amount of information available to both sides prior to trial is likely to increase the accuracy in the outcome. Indeed, to the extent that Brady v. Maryland requires pretrial disclosure of information that is “favorable to the defense and material to guilt or punishment,” those disclosures are “perfectly aligned with accuracy interests in the trial context.” Thus, the opposition to extending discovery rights has not been to deny the accuracy-enhancing nature of exchanging information. Rather, opponents have worried that the disclosure of information by the government to the defense would lead to the creation of perjured testimony and witness intimidation. Other concerns have centered around the problem of the privilege against self-incrimination. So long as the Constitution gives the defendant the ability to refuse to testify, either before or at trial, there is a fundamental imbalance in requiring the government to tip its entire case to the defendant, while permitting the defendant to shield his version of events.
Resolving this debate is far too great a task for this article, so I want to make some assumptions before addressing it further. First, I want to assume that the perjury and intimidation issues are real, but trivial. That is, while it is certainly true that the defendant may use discovery as an opportunity to put together a false defense, or to “encourage” witnesses not to attend, I believe that most defendants can: (1) create a false story based on what they hear at trial anyway and, (2) have enough information before trial in most jurisdictions to intimidate most witnesses, if they so choose.
Second, and perhaps more importantly, the self-incrimination problem can be partially eliminated by conditioning the availability of discovery from the government on the defendant being willing to provide similar discovery to the prosecution. The Federal Rules of Criminal Procedure provide that if the defendant requests documents and objects or reports of examinations and tests from the government, and the government complies, the defendant must supply those same things to the government. Some states have gone even further. In New Jersey, which generally provides for open discovery, the defendant is obligated to turn over discovery unless he affirmatively opts-out. California appears to mandate discovery even if the defendant does not ask for it.
Even after limiting the debate in this way, there remains a fundamental problem with further liberalizing discovery practices in criminal cases. Like innocence commissions, expanded discovery runs the risk of reducing the number of erroneous convictions, while doing little or nothing to minimize erroneous acquittals. In all American jurisdictions, there remains one source of information that is off-limits: the defendant himself. Under existing constitutional doctrine, not only is the government unable to obtain the defendant’s statement without his consent, but the state cannot give the defendant an incentive to waive his privilege against self-incrimination through the use of a jury instruction that would permit the jury to draw a negative inference from the defendant’s failure to testify. Thus, beyond a certain point, liberalization of discovery rules will mainly increase information disclosed to the defendant, not information disclosed by the defendant.
If we assume that the disclosure of information by one side benefits the other side, then defendants obviously benefit more than the government. In an adversarial system, that should mean that liberalized discovery, even when it strives to be balanced, will almost inevitably increase the number of acquittals. Furthermore, unless the increased information only helps innocent defendants, that means we should see not only a decrease in erroneous convictions but also an increase in erroneous acquittals.
This may not be such a bad thing. As Professor John Douglass has noted, “the government enjoys substantial practical advantages over most defendants” in obtaining information prior to trial, particularly in serious cases. The government has the ability, at least prior to indictment, to obtain the sworn testimony of witnesses from whom it wants to hear by putting them in front of the grand jury. Furthermore, police officers and other law enforcement officials have the ability to contact and interview witnesses, and in many cases witnesses may be far more willing to talk to such agents than to representatives of the defendant. Even though the defendant may eventually gain access to the results of these interviews, either as part of discovery or at trial, what is important here is the difference in the government’s and the defendant’s access. Unlike the government, the defendant simply has no way in a typical criminal case of compelling a witness to talk to him or his representative prior to trial.
If it is true that the government presently has access to more pretrial investigative resources than the defendant, then perhaps the effect of liberalized discovery is just to “even-up” a presently imbalanced system. But even if present discovery and investigative techniques, taken as a whole, favor the government more than the defense, I do not believe this alone can justify liberalizing discovery practices. After all, the question is simply whether this reform will be a good or bad thing. On this metric, it seems clear that the effect of liberalized discovery is more erroneous acquittals and fewer erroneous convictions.
There is one other reason to worry about increased discovery: failures to disclose. Until now, I have assumed that liberalizing the rules of discovery will lead the parties to comply with the rules and actually exchange the required information. While that will undoubtedly happen in most cases, there will inevitably be cases in which one or both sides fails in its obligations. The impact of such failures, though, will differ greatly. When the prosecutor fails to turn over information that should have been disclosed, and the result is an erroneous conviction, the failure can always be remedied. That is not to say that it will be remedied. First, the defendant has to become aware that the information was hidden from him, and, second, even if he subsequently discovers the information, the defendant may only gain a new trial where “there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been” an acquittal. Undoubtedly, there will be many cases in which prosecutors cheat, innocent defendants are convicted, and no remedy is ever given. Still, at least some innocent defendants will presumably gain relief this way.
When the defendant cheats on discovery, however, no reversal is possible. Even if the government can show that the defendant knowingly withheld discovery, the Double Jeopardy Clause will bar a retrial on that charge. Criminal punishment for the failure to disclose, if it happens to be available, will not blunt the impact of the Double Jeopardy Clause, at least in a serious criminal case. For example, the punishment for obstruction of justice is likely to be far lower than that for carjacking. The result is that not only will liberalized discovery lead directly to more erroneous acquittals, but there will also be a similar indirect impactâ€”some erroneous convictions will be set aside, but no erroneous acquittals will be remedied.
We might hypothesize, though, that defense counsel are less likely to cheat than prosecutors when it comes to discovery obligations. The most likely way to enforce discovery obligations, outside of reversals, is through sanctions. If the violation is discovered during the trial, the judge can impose sanctions as part of the proceeding. If it is discovered afterwards, the authority to enforce the rules of discovery may shift to the state or local bar. To the extent that either type of proceeding serves as a deterrent, defense counsel may fear the results more than prosecutors. Prosecutors have both the advantage of representing the government and the impression that they are “the good guys;” defense counsel may fear that the unpopularity of their clients may spill over to them. This possibility may then give defense counsel greater reason to comply with discovery rules than prosecutors. Alternatively, it may be that prosecutors internalize their roles as the good guys, and therefore they are less likely to cheat.
In reality, I imagine this debate can be side-stepped, because the greatest problem with discovery obligations will be a much more informal form of cheating. In many cases, the rules require the parties to disclose things such as “written or recorded statements of witnesses” or “written statements, if any, including any memoranda reporting or summarizing the oral statements, made by any witnesses.” We can easily imagine that prosecutors and defense counsel know about the existence of such rules when they interview witnesses, and that they will sometimes, either consciously or unconsciously, strive to keep information damaging to their cases from being written or recorded anywhere. What the lawyers have heard, but not recorded on paper or tape, cannot be turned over to the other side. The result is the careful creation of documents that do nothing to undermine a party’s case. Such evasions probably pervade the system on both sides.
A final (and under-explored) concern is the potential drain on resources. A truly liberalized version of discovery might abolish the privilege against self-incrimination (or make its waiver a condition of discovery), and then permit the parties the sort of wide-ranging pretrial discovery now seen in civil litigation. The danger of such a system is that it would require a vast increase in the amount of resources spent on criminal litigation if it were to be implemented. As others have noted, court-appointed counsel tend to have too many cases, sometimes hundreds of cases, assigned to a single attorney. Adding civil-like, pretrial discovery to the work of such attorneys would require the appointment of more public defenders, which in turn would probably necessitate the hiring of more prosecutors.
The reality, as I have noted, is that resources are likely to stay fixed, and that true discovery would occur in only a handful of cases. In particular, it would occur in serious cases and in cases where defendants hire their own lawyers. Limiting discovery to the most significant cases would probably be a good thing; it would mean that resources are placed in cases where the stakes are the highest for both sides. But allowing the availability of discovery to turn on the wealth of the client seems like a bad development. The criminal justice system is already pervaded by concerns of racism and sexism. As it is most likely to be implemented in practice, it would appear to be partial justice at best.
While full-scale discovery therefore may be a bad idea, limited liberalization to an open-file system seems harmless at worst, and potentially a real benefit. In an open-file system, such as that used in New Jersey, the government and the defense are required to make available to the other side most of the information they have collected on the case. While such a system undoubtedly increases the ability of defendants to gain acquittals, I suspect that, at worst, it simply maintains the overall accuracy of the systemâ€”for every erroneous conviction eliminated, there will be another erroneous acquittal. At best, the information provided will only help innocent defendants, leading solely to an increase in accurate acquittals. Because there is no real downside, further liberalizing discovery appears to be a positive change.
C. Jailhouse Informants
The use of testimony by informants and similar witnesses has long been a serious concern, even before the present controversy over wrongful convictions. Of particular notoriety was the case of Leslie Vernon White. Mr. White was the subject of a Los Angeles Times series and a 60 Minutes segment in the late 1980s. In the latter, he demonstrated for a national television audience how easy it was for him to manufacture testimony against other inmates in the Los Angeles County Jail, thereby leading to leniency for him.
The use of the testimony of both accomplices and jailhouse informants presents an obvious dilemma. On the one hand, such people often either (in the case of accomplices) have first-hand knowledge of the crime or (in the case of jailhouse informants) have witnessed a confession by the defendant. Such information, if true, is highly probative. On the other hand, accomplices and jailhouse informants, because they are often subject to criminal charges themselves, will not testify unless they are granted leniency. However, if leniency is offered, they have an obvious incentive to either exaggerate the conduct or statements of the defendant or, if they are sufficiently sophisticated, to wholly fabricate their testimony against the accused. If the criminal justice system uses such testimony, it assumes a serious risk that the testimony is false, thereby leading to erroneous convictions. However, if the system forgoes such testimony completely, it risks many erroneous acquittals because of the absence of very probative evidence.
The solution to this dilemma has proven to be elusive. The problem dates back at least to the early eighteenth century in England, where the government took two steps to help bring about the prosecution and conviction of serious offenses. First, because England at the time lacked professional prosecutors, individuals were required to spend their own time and effort to bring cases. Many persons were reluctant to do so, particularly for property crimes. To give individuals an incentive to pursue such cases, the Crown offered rewards for the apprehension and successful prosecution of criminals. Jurors, though, were often reluctant to convict in these cases “because of their suspicion of prosecutors’ and witnesses’ motives.” Second, identifying those who had engaged in serious property crimes, particularly robberies and burglaries, was often quite difficult. Because many such criminals worked in groups, one way to obtain such identifications was to have one member of the group testify against the others. In exchange for their testimony, so-called “Crown Witnesses” received leniency in their own cases, frequently a pardon. Again, though, jurors and judges often distrusted such testimony.
English officials found it hard to come up with a solution to the problem. Indeed, as Professor John Langbein has noted, the English judiciary developed a whole set of rules governing both accomplice testimony and alleged confessions. In the case of accomplice testimony, the rule, as it emerged in the first half of the eighteenth century, abolished the use of accomplice testimony unless it was corroborated. But by the middle of that century, the judges in turn rejected that rule and instead left the issue in the hands of juries. As for confessions, the English Justices during the eighteenth century developed a rule that a confession was given in the hope of leniency or fear of prosecution was not admissible. But no general bar to confession evidence emerged, as confessions had long been seen as important evidence.
Just as the judiciary could not bar the use of prosecution rewards and Crown Witnesses in eighteenth century England, eliminating the use of jailhouse informants and other witnesses who testify in return for leniency seems impractical in twenty-first century America. In one recent case, United States v. Singleton, the defendant challenged the use of such testimony as a violation of 18 U.S.C. § 201(c)(2), which provides that it is a crime to promise “anything of value to any person, for or because of the testimony under oath or affirmation given or to be given by such person as a witness upon a trial.” The district court denied the defendant’s motion, but on appeal to the Court of Appeals for the Tenth Circuit, the panel reversed, holding that where a witness testified in exchange for leniency, the prosecutor had violated § 201(c)(2), and therefore the testimony was inadmissible. The Tenth Circuit, though, reheard the case en banc and ultimately reversed the panel, holding that § 201(c)(2) did not apply to government prosecutors and their grants of leniency. The ultimate outcome in Singleton is a vivid illustration of the present unwillingness to eliminate the use of such testimony.
This unwillingness to bar such testimony suggests that the prevailing view (among the judiciary, at least) is that, on the whole, such evidence is accuracy-enhancing. While there are of course other possibilities, the most likely explanation for why English justices rejected the complete bar of uncorroborated accomplice testimony and why American judges rejected the bar of testimony in exchange for leniency is that the number of erroneous convictions that would be averted is vastly outweighed by the number of erroneous acquittals that would result. In other words, moving away from our current admissibility rule, it seems, would almost assuredly be accuracy-decreasing.
If this is right, it suggests there are presently only limited ways to deal with the problem of jailhouse informants. One possibility that sometimes is discussed is a corroboration requirement. The suggestion is that a jailhouse informant should only be able to testify if there is other evidence that will independently identify the defendant as the perpetrator of the crime. When set forth as a complete bar to the admission of uncorroborated jailhouse testimony, the concern is that the rule will lead to more errors than it will prevent. As I note above, eighteenth century English judges eventually abandoned the bar on uncorroborated accomplice testimony. One reason appears to have been a concern that the corroboration rule was too biased in favor of acquitting defendants. Of course, the accomplice corroboration rule lives on in some American jurisdictions. As presently enforced though, it appears to be only a weak constraint, because it applies only in the narrow circumstances where it is the accomplice’s testimony alone that ties the defendant to the crime. Accomplice testimony remains admissible in any case where there is some corroboration of the evidence. Drawing this line may make some sense; after all, drawing a line that forbids conviction in a case where there is absolutely no other corroboration that the defendant is guilty may result in a greater decrease in erroneous convictions than an increase in the number of erroneous acquittals. The set of cases involving completely uncorroborated accomplice testimony may be particularly weak.
As applied to jailhouse informants, such a corroboration rule could work in a similar way; it would bar a prosecution when the only evidence against the defendant was that of the informant. Such a rule would probably only have a limited effect, however, for most cases likely will produce some other evidence against the defendant. Indeed, the very fact that such testimony occurs when the defendant is in jail suggests that the government already has some evidence against the defendant. Left undisturbed by such a rule would be a large number of cases in which the testimony of jailhouse informants is just one of several factors leading to an erroneous conviction.
An alternative approach would be to have the judge, through jury instructions, note the generally untrustworthy nature of such testimony. Because this is a compromise, it might seem attractive, but as a matter of trying to achieve accuracy it seems misguided. The effect of such instructions is simply to give less weight to such testimony in general. It does nothing to help the jury sort out those informants who are truthful from those who are not. In effect, such an instruction just works as a watered-down version of a complete bar; it would probably eliminate a few erroneous convictions, but only at the cost of increasing to a greater degree the number of erroneous acquittals. Proposals that simply increase the standard of proof in some subset of casesâ€”which is precisely what a jury instruction like this one would doâ€”only serve to shift the distribution of errors and not improve the functioning of the system.
Another possible way to reform jailhouse testimonyâ€”one which might aim at actually increasing overall accuracyâ€”would be to require disclosure of the rewards that the informant may receive for truthful testimony. Defense counsel can, and do, already inquire about what is promised to witnesses as a valid form of impeachment evidence. Thus, the thrust of such proposals (I assume) is to require that the details of the leniency be set forth in advance. But as Professor Daniel Richman has noted, prosecutors frequently have valid reasons to be somewhat opaque as to the promises they are making. If they guarantee the defendant a particular break in advance, prosecutors run the risk that the defendant will either recant or fail to give his “best” testimony. In other words, the witness might not recant, but he may testify in a way that is less favorable to the government than the version initially proffered in exchange for the sentencing reward. As Professor Richman notes, the government also has less savory reasons for preferring vagueness in the agreement: “the expectation is that jurors will be less likely to be put off by the â€˜deal’ than if the agreement set out a precise discount.” Nonetheless, the fact that the government attorney has valid worries about the accuracy of the testimony she will receive in the absence of vagueness means that requiring her to agree in advance to the precise details of leniency is unlikely to increase accuracy.
An additional piece of information that should be disclosed is the informant’s testimonial history. As the case of Leslie Vernon White illustrates, the most doubt-inducing informants are those who have repeatedly been the recipients of confessions by other inmates. In a sense, the logic here is akin to the doctrine of chances in evidence; after a certain number of occasions, it becomes completely implausible that it is mere coincidence that the informant has been the recipient of a confession. It is, rather, more likely that the informant is doing something to help create those confessions. Therefore, disclosing such information and allowing defense counsel to use such information should hopefully have the effect of reducing the probative value of such testimony in the very case in which it is the least reliableâ€”where the informant has testified before.
Jailhouse informants are here to stay. That is because they are a source of valuable information against defendants, a source that most people appear to believe is accurate more often than not. Of course, such informants can and do lead to wrongful convictions, and there are reforms that show promise in limiting the harmful effects of such testimony. Thus, requiring at least some minimal corroboration of the informant’s testimony, disclosing any previous occasions in which the informant testified, and divulging the terms, if any, that have been agreed to for the witness’s testimony are all welcome changes. But we should also recognize that these are marginal reforms in an area that is ripe to create wrongful convictions, and that this is a problem that needs further study and innovation.
D. Recorded Testimony
Perhaps the least controversial proposal is that police interrogations of witnesses be recorded, preferably through the use of video, although alternatively through the use of audio recording technology. The benefits of such an approach are straightforward; recording allows jurors, judges, prosecutors, and defense counsel all to have access to exactly what the interrogator and interogee said during the interrogation, and to have better information about the conditions under which the statements were made.
Two objections to this improvement that can be disposed of rapidly are cost and training. Admittedly, the purchase of recording equipment costs money, and I have noted already that in a world of fixed resources, additional expenditures in any one area will take away resources from another place in the criminal justice system. In addition, use of recording devices will be a new technique for law enforcement, and therefore require training. This too will mean a shift in resources.
Ten or twenty years ago, such concerns may have had some merit. But today, when many cell phones come with video recording capabilities, it seems anachronistic to worry too much about the costs of either purchasing recording equipment or the training to use it. A typical digital camcorder can be purchased for less than $400 (no doubt less when purchased in bulk), and if police officers can be trusted to operate radar devices, then they should be able to operate these devices as well.
Another concern with the rise of video recording devices may be that jurors will come to place too much weight on the tape and too little importance on testimony about things alleged to have occurred both before and after the tape rolled. Judges have long been aware of the possibility that computer animations and similar video evidence could pose a danger under Rule 403 of the Federal Rules of Evidence because it could “mislead juries.” In other words, the jury would believe it to be more persuasive than it actually is. There is some reason to believe that the use of videotape confessions can actually be similarly distorting.
In addition, police officers can always use torture and other techniques on a witness prior to the videotape starting. Even if the witness’s subsequent interview is taped, the recording may not reveal the results of the torture. For instance, there are numerous techniques that law enforcement officers were known to have used in the twentieth century that would leave no mark on a defendant. And even if the defendant shows signs of physical abuse, it may be unclear whether the bruises and cuts are from torture or are the result of the defendant’s own actions, such as an attempt to escape or to resist arrest. Thus, the recording of interrogations is far from a guarantee that a confession was, in any meaningful sense, free and voluntary.
But even if there is a danger that jurors will rely too heavily upon the videotaped confession and too little on testimony about events surrounding the confession, it seems unlikely that the result would be less accurate than the system we have now. The presentation of all evidence at a trial is a manipulated event. It is commonplace for lawyers to prepare their witnesses in advance of testifying, not just as to the substance of the testimony, but as to the manner of presenting it. Videotape at least has the advantage of capturing events close in time to when the police and the witness first came into contact, and it provides a record that is at least somewhat more difficult to distort than a police officer’s statements about the alleged confession of the defendant.
After all, even if police officers first torture a suspect and then turn on the camera, the images may at least force the officers to concoct a story to explain the injuries. A written confession, accompanied only by the defendant’s denial and his claim that he was tortured, may not require any meaningful response. Furthermore, having a video record of the interrogation may cut down on false claims of improper techniques during the interrogation itself, possibly resulting in fewer erroneous acquittals. Thus, even with its potentially distorting effects, recording of interrogations appears likely to improve accuracy, not harm it.
The question that then arises is how confessions that are not recorded should be treated. One possibility would be exclusion. Even if such a proposal might have some effectâ€”a possibility I address in connection with eyewitness testimony in the next sectionâ€”it is less plausible here. There is little doubt that there will be confessions that are bound to be given under circumstances where recording is difficult or impossible, such as when the confession is made at the scene or where it is made in a casual conversation. It is not hard to imagine that, in many cases, such a confession would nonetheless be quite reliable and therefore excluding the confession would radically decrease accuracy. Unless we are prepared to go to a system in which all confessions are elicited in only formal settingsâ€”a possibility that I think is quite unlikelyâ€”such a system would not work.
The other alternative, which has been suggested by several groups, is to give jurors an instruction that unrecorded confessions are less reliable. As noted in the last section, the problem with jury instructions is they tell the jury only to give less weight to such evidence, without necessarily aiding the jury in determining which unrecorded confessions are true and which are false. To the extent the instruction aids the jury in appropriately down-grading the potential reliability of such unrecorded confessions (and such confessions are, I am assuming here, less likely to be accurate than recorded ones), that is acceptable. The danger is that the instruction will make jurors too reluctant to credit an unrecorded confession. However, because such instructions probably have little impact on the jurors’ actual decision-making process, this risk is minimal. In other words, enforcing the recording requirement through the use of a jury instruction would be less accuracy than what we have now.
E. Line-Ups and Eyewitness Testimony
A final area of criminal practice that seems ripe for change is the process of eyewitness identification of suspects. As has been repeatedly noted, erroneous identifications of defendants as the perpetrator of crimes have constituted a large proportion of the erroneous convictions that have been uncovered in recent years. Accordingly, any improvement made in this area would seem to be welcome.
A number of different proposals have been made to improve the process. Some are relatively small and uncontroversial. For instance, commentators are essentially unanimous in suggesting a line-up, photo array, or similar process be “double-blinded;” that is, the administrator of the process should not know which person in the line-up is the suspect. The basic idea is that when the administrator knows which person the police suspect, the administrator may either consciously or subconsciously signal that person to the witness. Having a double-blind line-up eliminates this possibility and would seem guaranteed to improve accuracy in eyewitness identification. Because the additional costs of such a procedure are certain to be low, this reform clearly should be adopted.
Another similarly uncontroversial suggestion is an increase in the number and similarity of “foils”â€”other persons in the line-up or photo array who are not suspected of the crime. Having a small number of foils makes it more likely that a guess by the witness will end up identifying the suspect. Making sure that the foils bear a close resemblance to the suspect ensures the witness is identifying the suspect because he is the person the witness believes committed the crime, not just because he is the only one in the group who resembles the perpetrator. Again, the marginal costs seem to be slight, and because the only effect would seem to be to increase the accuracy of the identification, the reform should be welcome.
Finally, it is frequently suggested that witnesses be instructed that the perpetrator may or may not be among the persons presented. This instruction has the virtue of being true; no one can know for sure in advance whether the actual perpetrator is in the line-up or spread, and there is no reason for the witness to make any contrary assumption. Assuming that the witness listens to and understands the instruction, it should only enhance accuracy, and again, the cost is minimal. This is yet another reform that should pose no difficulties.
Other potential reforms are more controversial. One common suggestion is that photo arrays and line-ups be done sequentially. In other words, the witness should be shown only one person at a time, rather than all of the individuals or pictures at once. But as the ABA’s report notes, “[t]here is a growing dissenting view among some well-respected social scientists that the research has not proceeded far enough to determine under what conditions, if any, a sequential lineup is to be preferred to a simultaneous lineup.” Sequential line-ups and spreads, therefore, do not seem guaranteed to improve accuracy, and it appears to be too early to decide whether such a switch is wise.
As an alternative to a switch in procedures, the ABA suggests that police departments and prosecutors should participate in field studies, comparing the use of simultaneous and sequential line-ups and photo spreads. As an accuracy-enhancing measure, such experiments almost assuredly improve accuracy over the long-run. Particularly because it may be that the two systems are superior to one another under different circumstances, further investigation to determine the appropriate use of each system would improve eyewitness identification. In the short-run, there would be a loss in accuracy because one of the two systems being used would have more errors than the other system. But since we start from a point where we do not know which system is betterâ€”and apparently we have little evidence to even make an educated guessâ€”this short-term cost is clearly outweighed by the long-term gains.
Nonetheless, it is worth noting that the ABA’s report, like the Illinois statutory requirement on which it is based, sends a potentially disturbing message to defendants: we know that we are using an inferior method to produce eyewitness identifications (although we are not sure it is the one we used on you), but this heightened risk of error for you (perhaps) is acceptable, because in the long run, we will have a more accurate system. Defendants are being used as an end to a means, something that deontologists often find to be impermissible.
If this does not strike a reader as disturbing, then imagine the following possibility. Assume, even if you do not believe, that the question of whether the death penalty works as a deterrent is unsettledâ€”there is evidence both ways. However, with further study, we might be able to resolve the conflict one way or the other. So, an organization suggests that we execute a lot more people to find out whether the deterrent effects are real. I am quite certain most people would find this sort of use of defendants to be morally abhorrent, even if, in the long-run, we saved lives as a result.
Of course, my analogy is imperfect, and I am not claiming that the two situations are exactly the same. I do think, however, that the example helps illustrate the idea that we are usually uncomfortable with consciously undertaking experiments in the area of criminal law, particularly in capital cases. This is precisely what the ABA report and the Illinois legislature have both suggested is appropriate, however, when it comes to eyewitness identifications.
Another approach to the problem of eyewitness testimony has been to suggest that juries be given special instructions discussing the problems with such testimony, or that expert witnesses be allowed to testify as to the infirmities of eyewitness testimony. The difficulty with proposals such as these, as I have noted above, is that they do nothing to improve the accuracy of the testimony itself. At best, they might improve the jury’s ability to sort out accurate and inaccurate eyewitness identifications. To the extent this is the goal, expert evidenceâ€”designed to educate the jury on the conditions that make eyewitness testimony more reliable or less reliableâ€”is much more likely to be effective than jury instructions. And jury instructions are likely to do little more than simply increase the standard of proof in eyewitness identification cases across the board. Again, changes like this are usually not accuracy-enhancing.
A more controversial proposition would be to eliminate show-up identifications, where the witness is shown only a single individual and asked if he is the perpetrator. As others have noted, “although show-ups are inherently more suggestive . . . , under some circumstances the use of show-ups is appropriate,” particularly where the witness is shown the individual soon after the crime and the individual has been apprehended soon after the crime and near the scene. The reality remains that show-ups and similar procedures may, in some circumstances, be more accurate than other techniques, if only because the witness is being asked to make the identification immediately after the crime. Eliminating them would not seem to improve accuracy.
In addition, there is the question about what to do with identifications that violate the reforms I have suggested are clearly warranted. An approach adopted elsewhere in American constitutional criminal procedure law has been to exclude the results of searches and interrogations that violated the rules set down by the Supreme Court. At some very general level, exclusionary rules have some effect. The existence of an exclusionary rule undoubtedly increases the incentives of the police to abide by constitutional constraints, even if only indirectly. So the imposition of new eyewitness identification rules, coupled with an exclusionary rule, would seem to make it more likely that police departments would abide by the rules.
The difficulty comes after the initial reforms have been implemented, and the question remains whether a particular identification that is out of compliance with the normal routine should nonetheless be admissible. Assuming there is still some “truth-value” to the identificationâ€”that is, that the identification, despite the violation, still really does make it more likely that the defendant is guiltyâ€”the question becomes whether the loss of accuracy from this particular identification is outweighed by the marginal incentive effect created by exclusion. In this more limited context there is good reason to suspect the answer is no.
As with recording confessions, enforcing changes to the methods of eyewitness identification is therefore problematic. Jury instructions are of limited help and exclusionary rules are too broad. Perhaps the best hope is educationâ€”finding methods such that law enforcement personnel want the techniques. In other words, success may need to come from changes in law enforcement culture, rather than from legislative and judicial edicts.
Reforming the American criminal process is not only desirable, it is necessary. The significant number of erroneous convictions that have been unearthed over the past decade, particularly because of the increasing availability of DNA testing, has made it clear that our criminal justice system makes mistakesâ€”too many mistakes for us to be comfortable.
But reforming the criminal justice system also requires us to be smart and wise. Avoiding erroneous convictions cannot come at any cost. Instead, it needs to be done with the understanding that process changes have collateral consequences, particularly in a world of fixed resources. Most importantly, we need to ensure we do not inadvertently increase the number of erroneous acquittals. On almost any account of the criminal justice system, we have powerful reasons to punish those who are actually guilty of crimes. That is, after all, the ultimate goal of the system.
If accuracy is our ultimate goal, then many of the reforms that I have discussed should be beneficial to the system. Recording confessions is likely to make them more reliable; changes that are proven to be improvements to eyewitness identifications are to be applauded. However, a change like an innocence commission, which serves only to ensure that the innocent are not convicted, may have deleterious side-effects, particularly if the inadvertent consequence is to make it easier to wrongly convict someone in the first place.
There is a final note I wish to sound. One reform I have not discussed, because it has not received much attention from the various commissions, is the need to eliminate junk science in the criminal courtroom. As others have noted, there is a large amount of supposed scientific evidence that is admitted in American courtrooms, even though the scientific basis for such testimony is, at best, quite weak, notwithstanding the wonderful results achieved on CSI. In recent years, academics have noted the serious limitations of handwriting expertise evidence, hair evidence, lead bullet analysis, and bitemark analysis, just to mention some egregious examples. Clearly, admitting such evidence when it has no scientific warrant can only lead to erroneous convictions (and in some much rarer cases, could lead to erroneous acquittals, where the analysis suggests the lack of a match). If it is true that the evidence has no proven science behind itâ€”and this is true about a disturbingly large percentage of non-DNA forensic evidenceâ€”then there is no reason it should be admitted in the first place, because it can only lead to inaccurate outcomes.
Although some commentators have suggested that the expert evidence rules under Daubert and Kumho Tire represent an opportunity to exclude such unreliable evidence, I remain unconvinced that the solution to this problem is likely to come from courts, mainly for political reasons. Reform commissions, however, are well-positioned to suggest and even help implement necessary changes to the rules surrounding the admissibility of junk science in the courtroom. Unlike courts, commissions can call before them scientific experts and undertake investigations into the scientific basis (or lack thereof) for various forms of forensic evidence. And because such groups often consist of members or former members from a broad spectrum of the criminal justice communityâ€”legislators, judges, prosecutors, defense counsel, and victims’ groupsâ€”their conclusions that certain types of evidence should be inadmissible are more likely to be given deference by the creators of evidence rules than the decisions of one or a small number of judges.
If our focus is on obtaining accurate convictions and acquittals, then eliminating worthless evidence should be among our highest priorities. Our present focus on eliminating wrongful convictions is a wonderful opportunity to clean-up the world of forensic evidence. Hopefully, it is one that future commissions will take seriously.
* Professor of Law and Director, Institute of Law, Science and Technology, Seton Hall Law School. B.A. and B.S., 1989, Stanford University; J.D., 1995, University of Virginia School of Law. I would like to thank Heather Taylor for reading a prior draft of this article and Randall Samson for his valuable research assistance.
. See, e.g., Kansas v. Marsh, 126 S. Ct. 2516, 2532â€“39 (2006) (Scalia, J., concurring); Joshua Marquis, The Myth of Innocence, 95 J. Crim. L. & Criminology, 501, 521 (2005); Joshua Marquis, Op-Ed., The Innocent and the Shammed, N.Y. Times, Jan. 26, 2006, at A23.
. Michael Risinger, Convicting the Innocent: An Empirically Justified Wrongful Conviction Rate, 97 J. Crim. L. & Crimin. (forthcoming 2007) (manuscript at 6â€“15, available at http://ssrn.com/abstract=931454) (suggesting an error rate of 3.3%).
. One possibility, of course, is that we have constructed our criminal justice system in such a way that it only makes erroneous acquittals, not convictions. As I have noted elsewhere, and will note again here, we aim to have the number of erroneous acquittals outweigh the number of erroneous convictions. See Erik Lillquist, Recasting Reasonable Doubt: Decision Theory and the Virtues of Variability, 36 U.C. Davis L. Rev. 85, 89â€“90 (2002). But it by no means follows that we have actually eliminated erroneous convictions, and the extant evidence strongly suggests that such erroneous convictions exist.
Nationwide, 25% of the conclusive DNA tests performed at the request of the police exonerate the suspect. There is no reason to suppose that the processes of police investigation are any better in cases that do not involve tissue samples that can be tested by the DNA technique. The same factors that implicate the innocent in rape and murder cases where DNA testing is often possibleâ€”misidentification, poor defense work, prosecutorial misconduct, informant perjury, and false confessionsâ€”are at work in other cases too. Our trial process is being asked to negate far more false accusations than criminal justice professionals previously believed.
Donald A. Dripps, About Guilt and Innocence: The Origins, Development, and Future of Constitutional Criminal Procedure, at xiii (2003).
. Id. at 10. The idea of a moratorium did not originate with Governor Ryan. For instance, the American Bar Association adopted a resolution in February 1997, calling upon all death penalty states to impose a moratorium until the states had put into place procedures consistent with prior ABA policies. Report with Recommendations, 107 A.B.A. Sec. Individual Rts. & Resps. 1 (1997).
. S. 1161, 2001 Gen. Assem., Reg. Sess. § 4 (Conn. 2001) (establishing the Connecticut Commission on the Death Penalty); see also State of Conn., Commission on the Death Penalty, Study Pursuant to Public Act No. 01-151 of the Imposition of the Death Penalty, Jan. 8, 2003, available at http://www.opm.state.ct.us/pdpd1/CDP/DCP_ Final_Report-Jan2003.doc [hereinafter Connecticut Report].
. See Raymond Paternoster et al., An Empirical Analysis of Maryland’s Death Sentencing System with Respect to the Influence of Race and Legal Jurisdiction 2â€“4 (2003) available at http://www.newsdesk.umd.edu/pdf/finalrep.pdf.
. See Mass. Governor’s Council on Capital Punishment, Final Report 3â€“4 (2004), available at http://www.lawlib.state.ma.us/docs/5-3-04governorsreportcapitalpun ishment.pdf [hereinafter Massachusetts Report].
. See Innocence Comm’n for Va., A Vision for Justice: Report and Recom-mendations Regarding Wrongful Convictions in the Commonwealth of Virginia, at xv (2005), available at http://www.thejusticeproject.org/press/reports/pdfs/17241.pdf.
. Of course, improving the performance of defense counsel and even prosecutors may improve accuracy, and states cite this as one of the reasons to improve attorney competency in capital cases. See, e.g., Connecticut Report, supra note 17, at 35 (“[H]igh quality legal representation in capital cases . . . is vital to protecting against wrongful convictions, avoiding death sentences when death is not the appropriate punishment in an individual case, and insuring that no innocent person is ever executed.”). I do not address this issue here for two reasons. First, accuracy in determining guilt and innocence is not the only reason to improve prosecutorial and defense performance. Also important is the effect of counsel deciding which guilty defendants are sentenced to death. This function vastly complicates the need for better counsel. Second, even when we view only the issue of defense counsel performance on guilt and innocence decisions, the effects of better counsel on accuracy is unclear. After all, the role of good defense counsel is to obtain an acquittal regardless of the actual merits. So while better defense counsel almost certainly decreases the chances of an erroneous conviction, it also increases the risk of erroneous acquittals. The ratio between these is critical, as I describe below, and I think far too complicated to discuss in this paper. See infra text accompanying notes 42â€“44.
. See, e.g., ABA Report, supra note 24, at 11; Connecticut Report, supra note 17, at 61; Cal. Comm’n on the Fair Admin. of Justice, Report and Recommendations Regarding False Confessions 4â€“7 (2006), available at http://www.ccfaj.org/documents/ reports/false/official/falconfrep.pdf; see also Massachusetts Report, supra note 22, at 19â€“20 (recommending special jury instruction stating that in-custody statements should be audio or video-recorded and that the lack of such a recording should be considered in determining reliability).
. See, e.g., Cal. Comm’n on the Fair Admin. of Justice, Report and Recom-mendations Regarding Eye Witness Identification Procedures 5â€“6 (2006), available at http://www.ccfaj.org/documents/reports/eyewitness/official/eyewitnessidrep.pdf; Con- necticut Report, supra note 17, at 61â€“62; Avery Task Force, Eyewitness Identifica-tion Procedure Recommendations 2â€“7 (2005), http://www.legis.state.wi.us/assembly/ asm84/news/janaveryrecommend_dd.pdf; ABA Report, supra note 24, at 23â€“26.
. See, e.g., ABA Report, supra note 24, at 63; Cal. Comm’n on the Fair Admin. of Justice, Report and Recommendations Regarding Informant Testimony 8â€“9 (2006), available at http://www.ccfaj.org/documents/reports/jailhouse/official/Official%20Report. pdf.
. Obviously, one might object that such consequentialist concerns are inappropriate in this context. I believe such arguments are fallacious and I have dealt with this topic elsewhere. See Lillquist, supra note 8, at 62â€“63.
. There is a further complication that I will not discuss here. Determining the number of errors, even assuming that we strive for only a 1:1 ratio, may be harder than it initially seems. That is because it is far from clear that there is presently an even number of actually guilty defendants and actually innocent defendants who go to trial. If, instead, it turns out that far more guilty defendants go to trial than innocent ones, then changes in trial procedures will benefit guilty defendants far more than innocent defendants; the opposite is true if there are far more innocent defendants at trial than guilty defendants. See Erik Lillquist, The Puzzling Return of Jury Sentencing: Misgivings About Apprendi, 82 N.C.L. Rev. 621, 671â€“72 (2004). Because we have no completely reliable information as to which assumption is correct, I will assume the numbers are the same. I am aware this is inconsistent with the assumption I have made elsewhere, but for present purposes, this assumption eases exposition and, I believe, little turns on it.
. As a point of reference, from 1995 to February 2007, the Court of Appeal has heard 310 cases referred by the criminal cases Review Commission. Of those, the Court of Appeal has reversed the conviction in 218. See Criminal Cases Review Commission, Case Statistics, http://www.ccrc.gov.uk/cases/case_44.htm (last visited Apr. 10, 2007). So the question in England would be whether, over that ten-year time period, there were an additional 20 innocent defendants convicted per year as a result of the decreased standard of proof, as jurors became aware that any mistake they made could be set aside later by the Review Commission and the courts.
. See, e.g., North Carolina v. Pearce, 395 U.S. 711, 717 (1969) (holding that the Double Jeopardy Clause protects against a second prosecution for the same offense after acquittal or conviction, and against multiple punishments for the same offense).
. Id. § 78. Serious offenses include, but are not limited to, murder, manslaughter, kidnapping, rape, some drug crimes, arson endangering life, war crimes, and crimes against humanity. See id. sched. 5.
. The liberalization of discovery rules dates back at least to Brady v. Maryland, in which the Court held it was a violation of the Due Process Clause for the government to suppress information favorable to the accused, 373 U.S. 83, 87 (1963).
. In recent years, the criminal justice system has taken steps to combat this problem. For instance, Federal Rule of Evidence 804(b)(6) was added in 1997 to permit the introduction of hearsay evidence in cases where the witness is unavailable because of wrongdoing by the opposing party. Similarly, the Supreme Court has asserted that wrongdoing on the part of a defendant can be a reason to allow the presentation by the government of testimonial statements that would otherwise be barred by the Confrontation Clause. See, e.g., Crawford v. Washington, 541 U.S. 36, 62 (2004) (recognizing the rule of forfeiture by wrongdoing as an exception to the Confrontation Clause).
. See Griffin v. California, 380 U.S. 609, 610â€“15 (1965); see also Portuondo v. Agard, 529 U.S. 61, 65 (2000) (declining to extend Griffin to prosecutor’s statements on defendant’s veracity based on defendant’s ability to tailor testimony after hearing other witnesses testify).
. Even if Professor Douglass is right about the present information-gathering imbalance in criminal cases, and I believe he is, the reality remains that the effect of liberalized discovery will mean relatively more information for the defendant than for the government.
. Compare 18 U.S.C. § 1510(a) (2000) (setting five years as the maximum prison term for obstruction of a criminal investigation), with 18 U.S.C. § 2119(1) (2000) (setting fifteen years as the maximum prison term for carjacking). I am not suggesting that a violation of a discovery order could lead to prosecution under § 1510. My point is only that criminal offenses of interfering with the proper functioning of courts generally are punished much less severely than is serious criminal conduct.
. See Adam M. Gershowitz, Raising the Burden of Proof: A Default Rule for Remedying the Under-Funding of Indigent Defense 10â€“11 (Working Paper Series, 2006), available at http://ssrn.com/abstract=930710.
. See Barry Scheck et al., Actual Innocence: When Justice Goes Wrong and How to Make It Right 165â€“66 (3d ed. 2003); ABA Report, supra note 24, at 65; Robert M. Bloom, Jailhouse Informants, Crim. Just., Spring 2003, at 20.
. For instance, in Willis v. State, the defendant was convicted of receiving a stolen air compressor upon the testimony of his alleged accomplice and the individual to whom they sold the air compressor, who admitted that his only conversations had been with the accomplice. 570 So. 2d 760, 760â€“61 (Ala. Crim. App. 1990). The court notably reversed for failure to give a proper accomplice instruction, not for insufficiency of the evidence, suggesting that a conviction upon such evidence, upon proper instruction, would have been permissible. Id. at 761â€“62.
. See ABA Report, supra note 24, at 74; see also Alexandra Natapoff, Snitching: The Institutional and Communal Consequences, 73 U. Cin. L. Rev. 645, 697â€“703 (2004) (calling for increased disclosure of information about informants pretrial).
. See generally Edward J. Imwinkelreid, An Evidentiary Paradox: Defending the Character Evidence Prohibition by Upholding a Non-Character Theory of Logical Relevance, The Doctrine of Chances, 40 U. Rich. L. Rev. 419 (2006) (discussing the legitimacy of the doctrine of chances).
. One possibility some commentators have noted is eliminating the use of police house confessions altogether, in exchange for a system in which the defendant would be questioned by a neutral magistrate. If the defendant refused to answer the questions, his silence could be used against him at any subsequent trial. See, e.g., Uviller, supra note 75, at 201â€“02. The benefit of such a system would be that any confession would be elicited in open court, with the presence of counsel, and therefore would be far more likely to be accurate. The defendant would have a powerful incentive to talk, because not talking would allow the jury to infer the defendant’s guilt. See id. at 202. And if a guilty defendant attempted to testify falsely, he would have to be careful, because information in his statement that turned out to be untrue would then be usable against him to show guilt. Such a change, though, would require not just a radical restructuring of interrogation tactics and the allocation of resources (the system would require more use of neutral magistrates), but it would also require overturning Griffin v. California, 380 U.S. 609 (1965). So even if we were convinced that such a system would be better, it strikes me as so unlikely that we need not consider it here.
. See Fred Galves, Where the Not-So-Wild Things Are: Computers in the Courtroom, the Federal Rules of Evidence, and the Need for Institutional Reform and More Judicial Acceptance, 13 Harv. J.L. & Tech. 161, 215â€“222 (2000).
. See, e.g., Steve Bogira, Courtroom 302: A Year Behind the Scenes in an American Criminal Courthouse 175â€“78 (2005) (noting use of electroshock, suffocation, telephone books and rubber hoses, among other interrogation techniques).
. See id. at 180 (describing case in which police at suppression hearing claimed (successfully, no less) that the defendant’s numerous lumps and bruises were the result of being tackled down a stairwell, during an escape attempt, notwithstanding a jail doctor’s testimony that the injuries were inconsistent with such a story).
. A slightly more narrow alternative would be to restrict the exclusionary rule to confessions made while in custody. Even here, though, the rule may be unrealistic. What should be done about confessions made on the way from the scene to the police station? What about those made to guards while in jail? What about confessions made when the recording equipment malfunctions?
Obviously, there are situations where this reform is impractical. In show-ups, which I discuss below, it is impossible to double-blind, because the witness is shown only one person. Additionally, there may be police forces that are so small that double-blinding is impossible, or there may be other situations in which double-blinding becomes practically impossible.
. The one condition in which I would imagine this is not true is if eyewitness identification testimony is simply very inaccurate even in the best of circumstances, and therefore an eyewitness is likely to perform better after receiving a signal from the administrator. But if this is so, it merely suggests the whole enterprise of eyewitness identification is so fraught with error that it should to be eliminated. For purposes of this paper, I leave that possibility aside.
. To give a sports analogy, imagine that you were assaulted by Bill Walton. You are then shown a line-up consisting of Dirk Nowitzki and Steve Nash. Since Nowitzki is the only seven-footer in the room, you might identify him, because you know it’s not Nash, who is only 6 feet, 3 inches tall.
. There is a limit, though, on the extent to which this sort of reform is beneficial. Showing an eyewitness 100 photos that almost, but not exactly, resemble the suspect, could have the effect of decreasing accuracy. But so long as the number of high-quality foils does not become excessive, this appears to be purely an accuracy-enhancing measure.
. See generally, John J. Donahue III & Justin Wolfers, Uses and Abuses of Empirical Evidence in the Death Penalty Debate, 58 Stan. L. Rev. 791 (2005) (assessing whether the death penalty saves lives).
. This could be the outcome regardless of the outcome of the experiment. If the experiment showed the death penalty worked as a deterrent, then policy-makers in the future could try to set the death penalty to minimize the number of lives lost. If the experiment showed the death penalty did not work, then policy-makers could reduce or eliminate the use of the death penalty, thereby saving lives. (I say reduce in recognition to those who believe that even in the absence of a deterrent effect, the death penalty is morally required. I take no position here on whether the death penalty should or should not be abolished.)
. But see Scheck et al., supra note 94, at 204â€“21; Press Release, Cal. Comm’n on the Fair Admin. of Justice, Focus Questions for Hearing on Forensic Science Issues (Jan. 10, 2007), http://www.ccfaj.org/documents/press/Press08.pdf.
. See Allen P. Wilkinson & Ronald M. Gerughty, Bite Mark Evidence: Its Admissibility Is Hard to Swallow, 12 W. St. U. L. Rev. 519, 537 (1985) (“â€˜Few attempts have previously been made to make any statistical assessment of reliability of bite marks and the acceptance of the method in a court of law does not necessarily prove its validity.’” (quoting D.K. Whittaker, Some Laboratory Studies on the Accuracy of Bite Mark Comparison, 25 Int. Dent. J. 166, 169â€“170 (1975))).