Writing in 1923, the prominent American judge and legal philosopher Learned Hand stated that “under our criminal procedure the accused has every advantage. While the prosecution is held rigidly to the charge, [the defendant] need not disclose the barest outline of his defense. He is immune from question or comment on his silence; he cannot be convicted when there is the least fair doubt in the minds of any one of the twelve [jurors]…. Our dangers do not lie in too little tenderness to the accused. Our procedure has been always haunted by the ghost of the innocent man convicted. It is an unreal dream.”

As a factual matter, Hand’s argument was completely wrong, for as we now know—thanks chiefly to DNA testing—hundreds of innocent people have been wrongly convicted of the most serious crimes in just the past few decades. It is bad enough that our vaunted criminal procedure utterly failed to prevent their convictions. But as Daniel Medwed shows convincingly in his new book, Barred: Why the Innocent Can’t Get Out of Prison, American criminal procedure more often serves to hide their innocence, even after they’ve been wrongly convicted.

Medwed’s excellent book—aimed at the general reader rather than the specialist—is a model of clarity and persuasiveness. In 12 short chapters, he describes how procedural barriers, ranging from largely limited pretrial discovery to highly deferential reviews on appeal, have been implemented by the courts in ways that severely hinder the proof of innocence at every stage of the judicial process.

Many of the examples Medwed gives of innocent defendants whose release from prison was delayed or even entirely blocked by procedural technicalities are truly disturbing, not least because of what they say about both prosecutors and judges. Take, for example, the case of Keith Edward Turner. In 1983, he was convicted in Texas of rape, largely on the basis of the victim’s identification of Turner as her assailant—an identification that was not made until several months after the crime was committed and that suffered from some of the shortcomings that have made mistaken eyewitness identification the single most common factor in wrongful convictions. At trial, Turner took the stand and testified that he was home watching Monday Night Football at the time of the crime. But on cross-examination the prosecutor asked him why, if that was the case, he hadn’t offered this alibi to the police when they first questioned him, instead of simply denying his involvement and otherwise choosing to remain silent at the time of his arrest. The question was a clear violation of Turner’s constitutional right to remain silent. But his defense lawyer initially failed to object, so the prosecutor repeated variations on the question several more times, to which the defense counsel made only belated and conclusory objections that the judge overruled. The jury convicted Turner, and he was sent to prison.

On appeal, the Texas Court of Criminal Appeals, while recognizing that the prosecutor’s questions were improper, nevertheless affirmed the conviction, holding that Turner’s claim of error had not been preserved because “trial counsel failed to make timely objections each time the appellant was questioned regarding his post-arrest silence” and that “trial counsel also failed to identify exactly what he was objecting to and to specify the grounds of his objections.” Nineteen years later, after Turner, largely acting as his own attorney, finally convinced the trial judge to have a DNA test done on the sample of the rapist’s semen obtained from the victim’s cervix shortly after the attack, the results conclusively proved that someone else had committed the crime. Turner was exonerated, but not before he had served many years in prison and his life had been made a shambles.

Medwed also examines the case of Dion Harrell, who was accused of raping a teenager in Long Beach, N.J. Once again, the prosecution’s main evidence was the victim’s belated identification of her attacker, coupled in this instance with the input of a forensic expert, who said that the blood found at the scene of the crime was of the same type as Harrell’s—and that this type was possessed by only 2 percent of the population. Although this assertion was grossly inaccurate (though not untypical of the testimony by so-called “forensic experts,” which has been shown to be a factor in another large group of wrongful convictions), Harrell’s defense attorney, rather than challenging either element of the prosecution’s case, relied mainly on Harrell’s own testimony that he’d been playing basketball with friends on the night of the crime and then had visited the home of another friend. Even though several witnesses corroborated his alibi, Harrell was convicted—and when he was finally released from prison, he found it difficult, as a registered sex offender, to get a job or a permanent residence and wound up living in homeless shelters.

In 2013, the Innocence Project took up Harrell’s case and demanded access to the biological evidence from the victim and the crime scene gathered by the police so that DNA tests could be performed. But by the time the Innocence Project had became involved, the US Supreme Court had already ruled that a defendant has no constitutional right to the use of DNA testing to prove his innocence, so the prosecutors refused to comply on the ground that Harrell was no longer in the state’s custody and therefore lacked “standing” to qualify for access to crime evidence under New Jersey law. Only after the Innocence Project began lobbying the New Jersey legislature to change this law (which eventually happened) did the prosecutors relent and allow Harrell’s lawyers access to the biological evidence. The DNA tests totally exonerated Harrell, and in 2015—more than 25 years after he’d been sent to prison—his conviction was overturned by the New Jersey courts. Harrell, by then a broken man, died a few years later.

The stories Medwed tells in his book are damning and heartbreaking. They also point to serious flaws in our legal system. It is perhaps not surprising that prosecutors are resistant, for the most part, to evidence of innocence that surfaces following a conviction, since it is hard for advocates, however well-intentioned, to admit that they have made a mistake of such magnitude. But more troubling is the tendency of judges, in the interests of “finality,” to gloss over such evidence, even when coupled with clear trial errors, such as those that occurred in Turner’s case. What happens, as Medwed demonstrates repeatedly, is that judges tend to interpret otherwise established doctrines in such a way as to make it very difficult for convicted defendants to get a chance to prove their innocence, even with the aid of newly developed techniques like DNA testing.

In Turner’s case, as I mentioned earlier, the appellate court refused to find error even in the clearly improper questioning of the defendant by the prosecutor, on the ground that the defendant’s counsel had failed to adequately object. Of course, the failure of a trial counsel to make timely and adequate objections is a well-recognized ground for denying appellate challenges. Medwed’s point, however, is that judges tend to carry such doctrines to an extreme in cases involving very serious crimes, a tendency that seems suspicious and that Medwed attributes to conscious and unconscious biases on the part of judges against overturning the convictions of people in such cases.

A good example of what Medwed means by this is his description of the doctrine known as “harmless error,” by which an appellate court can determine that even obvious errors made at a defendant’s trial were “harmless” in the context of the overall case, because the evidence of the defendant’s guilt was “overwhelming.” In other words, the appellate court rules that even if the wrongly admitted evidence had been excluded, the cold record presented to the judges satisfies them “beyond a reasonable doubt” that the jury would still have convicted the defendant.

As Medwed points out, while some form of the harmless-error doctrine is undoubtedly necessary to prevent even the most trivial errors from occasioning a whole new trial, the hypothetical and artificial exercise in which appellate judges engage in deciding whether a given error is “harmless” can easily be affected by the aforementioned biases. Why would this be the case? To begin with, most trial judges in state criminal courts are former prosecutors themselves, and perhaps this in itself can bias them against accusations of prosecutorial error. Also, in most states, these judges are elected and face a potential backlash if they are seen as freeing a convicted murderer or rapist on a “technicality.”

I would go farther and suggest that when an awful crime has been committed, there is an innate human desire to see the perpetrator caught and punished, and this instinctively biases judges—as it does juries and, indeed, everyday citizens—against the reversal of such convictions in emotionally fraught cases involving crimes like murder and rape. The practical result is that all sorts of errors are swept under the rug through the application of judicial thought experiments like the harmless-error doctrine. And the further result is that many innocent people get convicted through the introduction of improper evidence that should never have been allowed at trial.

In any event, as Medwed demonstrates, this tendency of judges to interpret the law a way that makes it very difficult for wrongly convicted defendants to prove their innocence pervades the judiciary, reaching even to the Supreme Court. This was on full display in the Supreme Court’s 2009 ruling that there is no constitutional right to post-conviction DNA testing, even though, in cases where DNA has been recovered from the crime scene, it is usually the most important evidence by far of guilt or innocence.

Fortunately, most (though not all) states have now responded to this decision with legislative actions that require providing access to DNA evidence. But what about the cases where, long after all the appeals have been exhausted, the defendant is finally able to provide strong proof of his innocence not through DNA evidence (which, after all, is not available in the majority of crime investigations) but through other forms of evidence, such as recantations by eyewitnesses, confessions from other suspects, perjury convictions of government witnesses, and so forth?

Medwed offers as an example the case of Leonel Herrera, who was convicted and sentenced to death in 1982 for the murder of two law enforcement officers. But some years later, before the execution was scheduled to take place, Herrera’s defense team—which had always believed that the real culprit was Leonel’s brother Raul (one of the original suspects)—obtained an affidavit from Raul’s attorney stating that his client had confessed to committing the murders (which the lawyer was able to provide because he was no longer bound by attorney-client privilege after Raul’s death), along with an affidavit from Raul’s son saying that he had witnessed his father killing the two officers. Leonel then brought a habeas petition that in 1993 finally made it to the Supreme Court, only to be denied. While a bare majority of the court declared that “a truly persuasive demonstration of ‘actual innocence’ made after trial” might make it unconstitutional to execute a defendant, the bar for such proof of actual innocence must be “extraordinarily high,” a standard that the majority ruled had not been met in this case. Four months later, Herrera was executed.

Although it is not mentioned in Medwed’s book, I should include here that in 2002, as a federal district judge, I extrapolated from the Supreme Court’s Herrera decision in a opinion in United States v. Quinones that, since DNA testing had by then established the post-conviction innocence of many so-called “death eligible” defendants to the highest possible standard of proof, the federal death penalty must itself be considered unconstitutional, since it deprived executed defendants of the opportunity to prove their innocence through such testing. To no one’s surprise, the Court of Appeals disagreed with my interpretation of Herrera and promptly overruled my decision.

Medwed examines this litany of injustice in heart-wrenching detail, providing numerous examples of innocent people who were convicted of murder or rape and spent many years in prison before being exonerated and released (albeit not without deep psychological scars that they bore for the rest of their often difficult lives). But when it comes to fixing this problem, Medwed is somewhat pessimistic that judicial attitudes are likely to change enough to enable these innocents to get prompt relief in the courts. So what about the other players in the criminal justice system? The most powerful of these, Medwed argues, are not the judges but the prosecutors, for it is their unwillingness to face the possibility that they have convicted the wrong persons that presents the biggest barrier to prompt exoneration. Although, as Medwed notes, the increasing public awareness of wrongful convictions has led some progressive prosecutors to independently reexamine some prior cases—and even, in a few offices, to create entire units dedicated to such reexaminations—most prosecutors have not gone down this route. Medwed argues that it is hardly a “natural” path for them to take, not only because most American prosecutors are political animals highly sensitive to the public’s understandable concerns about reducing crime, but also because it is very difficult for advocates who have spent a great deal of time and energy convicting someone of a serious crime to consider the possibility that they have made a dreadful mistake.

Another potential path to reform is through legislation, especially since, as I noted earlier, most states have now passed laws permitting post-conviction DNA testing. But Medwed points out that these laws often require the defendant to establish various forms of “standing” that are difficult to meet. Moreover, state legislators are notoriously sensitive to the issue of violent crime. Most of the aforementioned laws were passed when crime rates were historically low. But if they begin to rise again (as they have in some cities), the chances of further legislative reform seem low.

Medwed therefore argues that the best solution is the creation, either on the national or, at the least, the state level, of permanent independent commissions whose job would be to reexamine prior convictions whenever there is a more than frivolous showing of actual innocence. The United Kingdom has created just such a commission, with what Medwed describes as very positive results, and at least one US state (North Carolina) has created a somewhat similar agency. But I am a bit less optimistic than Medwed that this solution will be widely adopted. It reminds me of the final recommendation in the 2009 report by the National Academy of Sciences, which exposed the many flaws in most “forensic science” apart from DNA testing. Even though the report was written by a distinguished group of scientists and other experts, and even though defective forensic science has been shown to be a major cause of wrongful convictions, the report’s proposed solution—the creation of a national agency that would assess each forensic science and provide scientific protocols for its application and improvement—has never gone anywhere, primarily because of opposition by politicians, prosecutors, and police. The proposed agency that Medwed describes—one that any politician could easily denounce as designed to “cut ‘em loose”—would likely face even greater opposition from such constituencies.

So what is the solution? I am afraid the best one I can offer consists of educating Americans about the ever-growing number of specific and awful examples of wrongful convictions in the hope that, over time, this trickles up into demands for larger reforms. This would include challenging the practices of prosecutors and judges, perhaps leading both to roll back many of the barriers to a post-conviction proof of innocence. As even Learned Hand acknowledged, our courts and criminal justice system have always been “haunted” by the “ghost” of the innocent person wrongly convicted. But this is not some “unreal dream”—as Medwed shows, it is a very real nightmare. For this reason, I very much hope that his book will reach a wide audience, not least among the judiciary, for it is an excellent step toward educating the general public on how our criminal procedure, for all its pretensions, serves more often to keep wrongly convicted persons in prison than to set them free.