"More than 75 percent of those convictions rested, in significant part, on positive but false eyewitness identification evidence," the task force concluded in its 2012 report.
The most celebrated Connecticut case was that of James Calvin Tillman. He spent more than 18 years in prison for rape (based on eyewitness testimony) before DNA testing exonerated him. The General Assembly gave him $5 million in compensation.
"It's finally been exposed as a less valuable form of evidence," veteran New Haven defense attorney Hugh Keefe says of eyewitness statements.
Keefe says that, when he first started practicing law (in 1967), eyewitness testimony was considered second in importance only to a confession. "Today the science is pretty clear that eyewitness testimony in most cases is not very reliable," he says.
The task force Borden and Cameron served on called for changes in various police procedures when dealing with eyewitnesses. The legislature approved every one of the recommended reforms.
One of the key changes is what's called "blind" or "double blind" identifications.
In the past, an investigator on a case would often show a witness photos of many suspects all at once. Studies have shown that procedure results in lots of incorrect IDs, in part because an investigator's body language can give a witness hints of which picture to choose.
In a double-blind ID session, which is now the recommended system in Connecticut, an officer who doesn't know who the prime suspect in a case is shows a witness photos in a sequence. (Live "line-ups" in where a witness views actual people is only rarely used in this state, but reforms were also made in those procedures.)
If, as is often the case in small-town police departments, the only cop available is one who does know the identity of the prime suspect, another "blind" system can be used. In that situation, the photos of possible suspects are placed in envelopes, shuffled and handed to the witness to look at one at a time, while the investigator sits across the table where he or she can't see what photo is being looked at.
Borden says law enforcement officials who helped the task force were in favor of finding better ways to improve on eyewitness-identification accuracy.
The reason, Borden says, is that cops know that a wrong ID in a case "means the real bad guy is still out there to do more crimes."
And there's a good chance that's the case in the Vega killing. In 2010, state Superior Court Judge Stanley T. Fuger conducted new hearings on the convictions of Gould and Taylor, based largely on the fact that Stiles recanted her testimony.
There was no physical evidence that Taylor and Gould had killed Vega: no fingerprints, no DNA evidence. More than $1,800 was found in Vega's pockets despite the fact that prosecutors claimed the motive for the crime was robbery.
(Taylor and Gould admitted they were out on the streets that night in 1993, pretending to be drug dealers and ripping off people who came to buy dope. But they've always insisted they were nowhere near Vega's bodega when the killing took place.)
Gould and Taylor's defense attorneys insisted that the real killer was Vega's son, Carlos DeLeon, a convicted sex offender. They accused him of embezzling tens of thousands of dollars from the bodega's bank account, and shooting his father in the head to avoid discovery.
Judge Fuger concluded that a "manifest injustice" had been done to Gould and Taylor and ordered them freed. The prosecution appealed and the state Supreme Court ordered a second retrial on the grounds the two men had failed to "prove" they were innocent.
Taylor, suffering from colon cancer, was allowed to go home to die. Gould was returned to prison. A second judge, Samuel J. Sferrazza, denied Gould's petition for a new trial, and that ruling is now being appealed.
Officials with Connecticut's Innocence Project, which works to get wrongfully convicted people freed, say the court rulings surrounding the Vega killing will make their job even tougher. Now, according to the state Supreme Court, even having a critical witness recant testimony won't be enough to overturn a conviction.
O'Donnell's lawyer says he will also appeal his client's conviction.
Cameron warns that O'Donnell's guilty verdict is "a devastating blow" to Gould's chances of ever getting out of prison. A prosecutor can now point to that trial as proof that the key eyewitness was tampered with and bribed. "It's going to be very difficult for George Gould now," Cameron says.
Cameron's voice is tinged with disgust when he talks about the attitude of state prosecutors and judges in these cases, where the only basis for finding Gould and O'Donnell guilty is a single, unreliable, flip-flopping eyewitness.
"I have reached the unavoidable conclusion... they will do whatever it takes to defend a conviction," he says. "I think the system will defend a conviction come hell or high water."