From: Charles Hixson (charleshixsn@earthlink.net)
Date: Wed Jul 31 2002 - 10:19:19 MDT
Charlie Stross wrote:
>...
>
>Coming at this a bit late -- the UK has a legal system derived from the
>same roots as the US system, so it's probably fair to extrapolate from
>the UK to the US. Over the past decade, a criminal review body has been
>subjecting evidence relating to previous murder convictions to DNA
>testing in an attempt to reduce the proportion of contentious decisions.
>The false positive error rate in capital convictions between 1945 and
>1964 (when capital punishment in the UK was suspended) appears to have
>been a steady 10%. Thus, roughly 10% of those executed in the UK over
>that period were innocents convicted in error.
>
>Since 1970, the recidivism rate for murderers released after a life
>sentence was served ran at roughly 1.5%.
>
>So, it seems logical that if the UK had retained the death penalty, it
>would have executed ten times as many innocents as would have been saved
>from murderers.
>
>(If we accept that the British and American judicial systems are comparable
>in efficiency, then this is a very strong case that capital punishment is
>worse than useless as a means of protecting the public from recidivist
>murderers.)
>
>
>-- Charlie
>
>
Quibble: This argument is approximately correct, but it assumes that the DNA evidence is 100% accurate. Unfortunately, this isn't true. Aside from lab problems, I believe that the current protocol only involves testing the DNA at 10 locations for 10 alleles (I could be wrong here, but similar arguemtents can be made with differing numbers). Each of those alleles occurs in the population in a small number differing forms, with different frequencies, so the possibility of calculation of a false positive is a bit tricky. Now if the person was convicted without reference to DNA evicence, then this is a quite reasonable approach. If, on the other hand, DNA evidence was used to convict him, then all a retest can do is confirm that the lab didn't make an error, it can't correct for a false positive. (False negatives, OTOH, would have to assume that the evidence was mishandled, and that never happens: ref. OJ Simpson [He was probably guilty, but the police should be fired and jailed for tampering with the e
vidence.])
So, if well handled, the DNA evidence can, as currently used, *usually* isolate one man out of 100,000 or so. But not one man out of 1,000,000 Quite good, especially if used in conjunction with other evidence. But note the usually. Relatives are examples of people who are likely to share all of the alleles. So if choosing among relatives, it is a much less good test. (How good? I haven't seen any statistics.) And inbred populations would also be likely to share alleles. Etc. (I don't know which sites are choosen, or how long a fragment is checked. But it's short, as the limitation on checking was decided several years ago as a cost containment measure. It probably predates the database, and it's original use was just to confirm other evidence, so it didn't need to be too precise.)
-- -- Charles Hixson Gnu software that is free, The best is yet to be.
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:15:48 MST