An archive like this one is built entirely on convictions. The number beside a killer's name, the dates, the sentence at the end, every bit of it comes from what a court decided happened. Which means when a court gets it wrong, the archive inherits the mistake and files it alongside everything else.
Twelve entries here record exactly that. They fall into two groups, and the difference between them matters more than it might first appear. In four cases, the person the entry is about was convicted of murders they did not commit. In the other eight, the entry is about a real killer, and somebody else went to prison, or to the death chamber, for what he did.
Start with Kathleen Folbigg, because her case shows how respectable a wrongful conviction can look. Australia convicted her in 2003 of smothering three of her infant children and the manslaughter of a fourth. The case against her was essentially arithmetic: four deaths in one family, therefore no accident. Twenty years later, genetic evidence indicated the deaths most likely were natural after all. She was pardoned in June 2023, and that December her convictions were quashed. She had served the full twenty years.
Billy Glaze is harder to sit with. He was convicted in 1989 of murdering three Native American women in Minneapolis, on the strength of witnesses who reported him making racist remarks. No physical evidence tied him to any of it. DNA testing eventually excluded him from all three killings and, in one, pointed to a different man entirely. Minnesota refused him relief anyway. He died in prison weeks after the ruling, twenty-eight years in.
Frances Newton's case never resolved at all, and Texas executed her in 2005. Her trial lawyer, later disbarred, called no witnesses in mitigation. A second gun of the same calibre went untested. The residue found on her clothing was equally consistent with fertiliser. The ballistics came from a police laboratory shut down for reinvestigation in 2002. A reprieve in 2004 bought testing that settled nothing either way, and the doubts are still sitting there.
Masumi Hayashi has been on death row in Japan since 2002, convicted of a mass poisoning on circumstantial evidence and a trace-element analysis that scientists have been arguing about ever since. She has never confessed. If you want to know why critics object to Japan's willingness to hand down capital sentences on circumstantial cases, hers is the case they point to.
The other eight entries show you the same failure from the other end. Somebody was convicted. The man who actually did it stayed out, and kept going.
Huugjilt was eighteen years old. He reported an attack on a woman in a public toilet in Hohhot, Inner Mongolia, in 1996, and was arrested for it, convicted, and shot sixty-one days after the killing, swept up in a national anti-crime campaign that was processing cases at speed. Zhao Zhihong confessed to the murder in 2005. The authorities sat on it for nine years. Huugjilt was declared innocent in December 2014, and twenty-seven officials were disciplined or investigated over what had been done to him.
Nie Shubin was twenty. China shot him in 1995 for the rape and murder of Kang Juhua, ten months after his arrest. A decade later Wang Shujin confessed to that killing in some detail, and the courts simply declined to accept it. They convicted Wang of three other murders and executed him in 2013. In 2016 the Supreme People's Court finally quashed Nie's conviction and declared him innocent. Nobody stands convicted of Kang Juhua's murder today.
Frank Lee Smith spent fourteen years on Florida's death row on the strength of a single eyewitness identification. He asked for DNA testing. He was refused while he was alive, and died of cancer in January 2000, eleven months before the test was finally run. Jerry Frank Townsend, a man with the intellectual capacity of a young child, had confessed under police questioning in 1979 and been convicted of six murders. The testing cleared them both and matched Eddie Lee Mosley, who was found incompetent to stand trial and has never been convicted of killing anyone.
Rolando Cruz and Alejandro Hernandez were both sentenced to death for the murder of Jeanine Nicarico in Illinois. Brian Dugan confessed to it in 1985. DuPage County prosecutors then spent twenty years defending the convictions instead of revisiting them. Cruz's was overturned three times; at his third trial, a lieutenant admitted under oath that a key statement had never actually happened. Seven officers and prosecutors were indicted over the case. All seven were acquitted. Cruz and Hernandez walked out in 1995, and the wreckage of the case is a large part of why Illinois eventually abolished the death penalty.
Anthony Capozzi, a Buffalo man with schizophrenia, lost twenty-two years to two bike path rapes he had nothing to do with. He was convicted on victim identifications, and then refused parole again and again because he wouldn't admit to offences he hadn't committed. DNA identified Altemio Sanchez in 2007, who had spent those same twenty-two years attacking women in the same city. Capozzi was exonerated and compensated.
Kevin Green served seventeen years for attacking his pregnant wife and killing their unborn child. The evidence was her identification of him, given after she had sustained severe head injuries. Gerald Parker admitted the attack in 1996, once DNA had connected him to a series of killings.
Claude McCollum, who has an intellectual disability, was convicted in 2006 of a murder at Lansing Community College. What convicted him was a statement in which he allowed that he might have done it. When Matthew Macon was arrested, someone went back through the case and found surveillance footage placing McCollum somewhere else at the time. The footage had been sitting there the whole way through. His conviction was vacated in 2007, the charge dismissed in 2008.
Roosevelt Dedmond came closest to the version of this story that never gets corrected. He had been charged with murdering his own wife, the first of Lee Roy Martin's four victims in Gaffney, South Carolina, and was waiting to go to trial. He would very probably have been convicted, had Martin not carried on killing and then rung a local newspaper to tell them where two more bodies could be found. Dedmond was released without ever facing a jury.
Read all twelve together and a pattern shows up that is worth naming.
Seven of them came apart because of physical testing, testing that either arrived decades late or was refused while it could still have done some good. Three came apart because the real killer confessed, which is not a safeguard in any meaningful sense. It's luck. One came apart because someone finally looked at evidence that had been in the file all along. And in one, Frances Newton's, nothing came at all.
What almost none of them came apart because of is the appeals process. Cruz was convicted three times over. Smith was refused his test and died waiting. Glaze was excluded by DNA and stayed convicted until the day he died in prison. Huugjilt waited eighteen years past his own execution. These convictions did not unravel because the system noticed its own mistakes. They unravelled because something outside the system forced the issue, usually far too late to matter.
And look at who these people were. Schizophrenia. An intellectual disability. The mental capacity of a young child. Severe head injuries. Eighteen years old. Twenty years old. A mother whose babies had died. Their circumstances made them convictable, and a conviction that is easy to obtain is the one that gets obtained.
Which is the reason these twelve sit in an archive of killers rather than in a file marked administrative error. A wrongful conviction isn't a footnote to serial murder, it's part of the same story. In eight of these cases, it's the reason the killing carried on.