As it obviously should. Unfortunately, for whatever reason, appellate courts are usually VERY reluctant to say that a previous trial was flawed, even in cases where most outside observers agree that it was obvoiusly and egregiously so. The case I usually cite is Rodney Reed:

There is no physical evidence tying him to the crime scene, and DNA evidence from the crime scene that rules him out does NOT rule out arresting officer Jimmy Fennell, whom the victim had coincidentally dumped and who, according to one witness, publicly argued with the victim two hours before her death. That is only one witness, of course, but is one more witness than can place Reed anywhere near the crime scene. That ought to be enough for reasonable doubt right there, and the prosecutions failure to inform the defence of the crime scene DNA evidence during discovery ought to be enough for a new trial--but an appellate court disagreed (after initially refusing to even hear the appeal because it was filed one day after the appeal deadline; prior to that another judge had upheld the ruling made by HER OWN MOTHER in Reeds original trial.) To top it all off, one of the arresting officer/ex-fiances police academy classmates testified that if he caught someone cheating on him he would strangle them to death with a belt to avoid leaving fingerprint evidence; the victim, his ex-fiance who was having sex with someone else, was, of course, strangled with a belt.

The obvious tragedy in executing innocent people is that they are wrongfully killed, but there is another often overlooked yet equally great threat to society: Executing an innocent person for a serious crime "solves" and closes the case--ending the search for a criminal who remains at large, unsuspected and dangerously violent. The actual killer could, for example, take a job with a different police force, respond to a domestic violence call, then kidnap and rape the woman present. Which, in Jimmy Fennells case, is not so hypothetical. When we execute an innocent man for a heinous crime, it means society remains at risk until the person responsible commits ANOTHER such crime and, hopefully, is at last caught and convicted of it.

The Austin Chronicle also reports that Reeds case was presented to a FEDERAL Appeals Court in January, to which I can only say, "thank God," because, as a federal appellate judge once stated IN HIS OFFICIAL RULING, "the only justice in the Texas Dept. of Criminal Justice is in the name." Odds are realistically still not good though because, once again, appellate courts are, for whatever reason, usually disinclined to say that their colleagues on a lower court screwed up so badly that it could cost an innocent man his life. It is rather perverse; the worse the error and the more serious its consequences, the more likely appeals coursts seem to protect their fellow judges rather than wrongfully convicted people.