Poor people lack the financial resources to mount a vigorous and thorough defense. All too often, indigent defendants are represented by public defenders who themselves are underfunded and overworked. —Daniel Medwed, University of Utah law school

Judge Michael DiReda of Utah's Second District is convinced that Debra Brown didn't do it. He found her "factually innocent" in May 2011, after she served 17 years in prison. Attorney General Mark Shurtleff is convinced she did do it.

In dispute is the cold-blooded murder of Brown's former employer, shot in his sleep near Logan in 1993. Today, Brown is free, reunited with her children and grandchildren. While the Attorney General is not seeking to put her back behind bars, he is asking the Utah Supreme Court to overturn the finding of innocence, arguing that the trial judge erred in his ruling (see related Reply Brief PDF at left).

Since the 1989 advent of DNA analysis, 290 convicted Americans have been exonerated through biological evidence, according to the Innocence Project, the national clearing house for investigating and pursuing innocence claims. The most recent DNA exoneree is a Colorado man released this month.

But few cases actually have DNA evidence available. Most convictions are based on circumstantial evidence and eye witnesses. Proving errors in such cases is difficult, as they often come down to disputed reconstructions of fading memories.

False convictions are a particular concern for those lacking economic resources, according to Daniel Medwed at the University of Utah law school.

"Poor people lack the financial resources to mount a vigorous and thorough defense," Medwed said. "All too often, indigent defendants are represented by public defenders who themselves are underfunded and overworked."

In the absence of DNA, no one knows how many innocents are mistakenly languishing in prison. If the false conviction rate is .027 percent, as U.S. Supreme Court Justice Antonin Scalia suggested in 2006, most experts on all sides would view this as regrettable but inevitable. But if the rate is closer to 3.3 percent or 5 percent, as Michael Risinger of Seton Hall University has suggested, most would agree a civilized system would demand strong correctives.

Combatting 'urban myths'

With real numbers elusive, people imagine reality through stories. Even innocence advocates admit such stories are often distorted and counterproductive.

Joshua Marquis, district attorney of Clatsop County, Ore., is a feisty, outspoken and widely quoted defender of the law enforcement and prosecutors, whom he sees as unfairly maligned by news media and reform advocates.

Exhibit A for Marquis is an acclaimed off-Broadway play, "The Exonerated," which features actors playing six "exonerated" death row convicts. The play ran for two years in the early 2000s, became a made-for-TV movie with Mimi Rogers and has since toured college campuses around the country.

In Marquis' view, the errors in the play typify public misunderstanding and distrust cultivated by activists and sympathetic news media.

Marquis points out that at least three of the six convicts were never exonerated at all. One of these, Robert Earl Hayes, had his conviction overturned over a procedural error and was acquitted at retrial. But he was quickly rearrested and convicted for a different rape-murder with facts strikingly similar to his earlier conviction.

In a law review article, Marquis documented the active role played by another of the six "exonerees" in killing two Florida police officers in cold blood at a highway rest stop. Marquis notes that Sunny Jacobs was, in fact, never exonerated for her part in the killings. After her conviction was vacated for procedural errors, she avoided a new trial by pleading guilty to two second-degree murder charges in a complex plea that allowed her to walk with time served.

"For a human endeavor we have an awfully good track record," Marquis said. "That doesn't mean everything is great, but I do get frustrated with the urban myths promulgated by the Innocence Project and found in news media and television shows."

Did she or didn't she?

Marquis' frustration is shared by many on the prosecutor side, who feel that the vital efforts of law enforcement and prosecutors are being undermined. Among these is Scott Reed, an assistant Utah attorney general working on the Debra Brown case. Reed believes the original verdict was correct and gets frustrated at local media portrayals that slight the state's case in favor of the human-interest angle of the grandmother-set-free.

Reed thinks the case is strong. Not even the defense disputes that Brown had both motive and opportunity. She had forged the victim's checks to a total of $3,600, had a key to the front door, and — as a housekeeper — had intimate knowledge of his home, including where he kept his financial records and gun, according to the attorney general's brief. Most crucially, incriminating checks and financial records had disappeared from the crime scene.

Much therefore hinged on alibi, time of death and a couple of witnesses. On time of death, the medical examiner was of little help, placing the time as likely 9 p.m. on Saturday night but allowing it could have been as early as 9 p.m. Friday or as late as the wee hours of Sunday.

From there, things get murkier. One witness claimed to have heard gunshots at 7 a.m. Saturday, when Brown lacked an alibi. But another witness claimed to have seen the victim alive at a local restaurant that afternoon. In the original trial, the defense chose not to call that witness, and it was this claim that became central in opening up and overturning the conviction.

"Why wasn't this 'new' witness called by the defense attorney?" Reed asked. "It seems to me that he concluded the guy couldn't do him any good." The defense attorney is not around to ask, having long since died. And the witness has no independent recollection of the events. Only the detective's interview notes verify the critical restaurant sighting.

Another witness claimed to have seen the victim Saturday evening, but that claim was disputed. Moreover, Reed said, Brown's alibi for Saturday night is subject to credibility questions.

Patterns of error

Such is the fuzziness of non-DNA innocence claims, in contrast to the clean lines of DNA evidence. But Brandon Garrett at the University of Virginia Law School sees a silver lining in the contrast. In his 2011 book "Convicting the Innocent," Garrett seized on the DNA exonerations to form a data set that aims to clarify the anatomy of error in non-DNA cases.

Because we have a body of cases we know came out wrong, Garrett reasoned, then once that anatomy is better understood, error-inducing behavior can be better combated wherever it occurs – without the need to prove, catalog and count non-DNA conviction errors. Since each DNA exoneration stemmed from a non-DNA case, Garret concluded that each error may be traceable to specific mistakes — and these mistakes ought to be correctable.

In short, Garrett sidesteps the question of how many errors occur and focuses instead on the patterns that cause them and systemic reforms that could prevent them.

In 76 percent of DNA exonerations, Garrett found that the eyewitness had identified the wrong suspect. He also noted that many of these identifications began hesitantly, but with positive feedback from law enforcement, by the time they got to trial the witness was 100 percent certain. A powerful illustration of this phenomenon is found in "Picking Cotton," a New York Times best-seller coauthored by a rape victim and the man she mistakenly identified who served 11 years in prison as a result.

Sixteen percent of DNA exonerations involved a false confession by the accused. Even more striking, Garrett found that these confessions often involved rich and striking detail that only someone familiar with the crime could have known. The detail stemmed from details discretely fed to the suspect during the interrogation, which then coalesce into a coherent whole toward the end. The easy solution to this, Garrett argues, is to require that all interrogations be recorded in full and require the judge to review the tape before admitting a confession in court. Only 11 states currently require that even some of interrogations be recorded.

"It's remarkable that so many states still don't do videotaping of entire interrogations," Garrett said.

Garrett also found invalid and exaggerated forensic work and testimony by forensic experts in 61 percent of the cases, including cases where dubious bite marks or hair analysis was asserted with undue confidence. Aside from the highly accurate DNA, Garrett said, "traditional forensics, most of the time, used unreliable techniques or exaggerated evidence on the stand." In these cases, he added, the defense did not recognize errors or did not have access to experts to counter them.

In 21 percent of the exonerations, jailhouse informants played a role, a problem Garrett believes could be mitigated by barring side deals for informants or requiring that they be disclosed to the defense.

"After exoneration, people think that cases against innocent people are weak," Garrett noted, "but at the time these cases seemed really strong. When evidence gets contaminated, the case can appear really strong, and the innocent may seem really guilty."

Unknown and unknowable

"American prosecutors have the toughest jobs, since we ask them to be zealous advocates but also ministers of justice," said Daniel Medwed. He quotes Samuel Gross of the University of Michigan Law School who said, "the true number of wrongful convictions is unknown and frustratingly unknowable."

This very uncertainty, innocence advocates maintain, is the best argument for process reforms designed to reduce the chances of error. And Medwed believes Garrett's analysis is groundbreaking because it provides the tools and knowledge to do so.

"Systemic reform aims to safeguard against good faith mistakes," Medwed argues. For example, Medwed suggested that internal review committees should evaluate testimony before charging a suspect in a single eyewitness case.

"We want to prevent error," Medwed said. "Our interests are aligned with those of prosecutors. Every time an innocent person is convicted, a guilty person is left on the street. We need to go beyond blame and work together to get better results."

Correction: The original version of this article stated that the Attorney General had filed for a new trial. In fact, the AG's office does not plan for a new trial nor to put Brown back in prison, but is only seeking to correct the record, in their view, by asking the Supreme Court to overturn the finding of innocence.