>> Certainly the wrongly convicted constitute an immeasurably small
>> percentage.
>>
>
>> No. Anywhere from 2 to 10% in the US, which is small, but hardly
>> insignificant, let alone immeasurably small.
>
>Is it 2% or 10%.
>That's a awfully wide variance.
>What evidence do you have that it is anywhere near 10%?
>Have you actually met the people who go before a judge with a public
>defender to plea bargain a case?
>If .5% were not guilty of what they are charged with I'd be shocked.
It is a wide variance. That's because there's no data. The only data that exists is that from DNA exonerations, but, of course, this data is woefully incomplete. Nobody believes that everybody who has been exonerated by DNA to date represents the sum total of wrongfully convicted people. Cases in which there is DNA evidence--sexual assault and some murders--are the exception, not the rule. Additionally, most DNA evidence from older crimes has been destroyed. (The reason Dallas County, Texas has so many DNA exonerations is because that county had an unusual policy of preserving biological evidence in the 1980s and 1990s. Most counties and states did not.) It is therefore impossible to know what percentage of people in prison are wrongfully convicted. However, what data there is suggests a lot of wrongful convictions (of course, still nowhere near "most" prisoners). This is because of what we learn from those DNA exonerations (e.g., that the most widely relied upon evidence in the criminal law--eyewitness identifications--is the least reliable form of evidence).
My experience confirms this. I do post-conviction work. I've had about 10 long-term clients in the past five years whose cases I have investigated. Of those, four were innocent. All dirt poor. Of those four, only one had any connection to the offense whatsoever (his mere presence). In that case, the guy who actually committed the murder--and who testified *for the State* in the trial--eventually committed another murder. While in prison for that murder, he admitted to committing both crimes. One case was caused by police and prosecutorial misconduct in conjunction with the true perpetrator having falsely implicated the client to spare the actual second participant, to whom he was close. He randomly picked the client out of a hat while he was being interrogated by police, who, due to the nature of the crime, knew there had to be a second participant and refused to take no for an answer. (The client had an alibi, which was disregarded by the jury, undoubtedly because he and his alibis were black). The last two were caused by egregious police and prosecutorial misconduct. Note that none of these cases even involved erroneous eyewitness identification, the most common cause of wrongful conviction. I recognize this experience encompasses an extremely small sample size, but note I do not claim a 40% error rate on the basis of it. The short of it is that adversarial criminal jury trials, at least where the resources available to the defendant pale in comparison to the State, are horrible means to accurately evaluate historical events.
While the cases I have deeply worked have been few, I have reviewed many, many more cases at a more superficial level, e.g., read transcripts and court pleadings. It is impossible to determine if a person is wrongfully convicted without undertaking a time-consuming (and expensive) investigation (something that is rarely done by the defense lawyer before trial), but many of these cases on their face raise suspicions and present many avenues of potentially fruitful investigation.
So my estimate of the rate of wrongful convictions is just that. But the kind of work I do places me in the best position of anybody to make this estimate. See also http://papers.ssrn.com/sol3/papers.cfm?abstract_id=%20931454.