In my mid-twenties, I was mugged while driving cab in Rochester, New York.
Next morning, the RPD called me downtown to view a line-up
Although I could not positively identify anyone in that line-up, Chief Detective Fantigrossi leaned on me -- figuratively and literally - to positively identify one particular fellow who bore a general resemblance to my assailant.
When I told Fantigrossi that "I could never identify him under oath," he replied (and this is close to verbatim): "Oh, sure you can!"
Nearly ten years earlier, my Dad was a member of the longest-sitting Grand Jury in Rochester history. It was his jury's duty to rule on all cases arising from the Rochester Race Riots that began on July 24, 1964. http://en.wikipedia.org/wiki/Rochester_1964_race_riot
My father was a very bright, judicious man who took pains to weigh all sides of a story.
One evening, after a full day with the Grand Jury, Dad made this exasperated comment over dinner: "How can eye witnesses, with no apparent ax to grind, give antithetical testimony to the same event? It happens all the time!"
We now know from academic study that eye witnesses to staged robberies will positively identify the wrong perpetrator 30% of the time.
Whether or not Carlos de Luna was guilty of the Texas murder discussed in the following audio file, we now know that there was a superabundance of evidence, which - if presented at trial - would have introduced massive amounts of reasonable doubt.
The NPR blurb: "Columbia Law School professor James Liebman organized his students to re-investigate a murder case which led to the execution of Carlos De Luna. They found evidence that they believe proves that the state of Texas executed the wrong man." http://thestory.org/archive/The_Story_53012.mp3/view
I also recommend the following New Yorker article, "Trial by Fire," which probes the case of Cameron Todd Willingham who was executed for killing his two children by arson. http://www.newyorker.com/reporting/2009/09/07/090907fa_fact_grann