"Dead Paul" wrote in message
news:g0e31h$cq3$1@news.datemas.de...
> I'm unsure if the British system is equally flawed but it's an angle
> which I had not previously considered.
> This account is relating what's currently happening in the USA.
>
> <<
> Like fingerprints, DNA are very powerful and scientifically sound
> evidence, when used to connect a known suspect to evidence found at the
> scene of the crime. Jurors are easily persuaded to accept the DNA link for
> someone who had already been suspected of a crime scene when told the odds
> against a false identification are 1 in millions or billions.
>
> But DNA is far less certain when you compare one sample against all of the
> profiles in the database typically known as one-to-many. In that case the
> chances that a match between a DNA sample -- especially an incomplete one
> -- and a person in a DNA database could nab an innocent person has
> different math. Very different math.
>
> So if you have a probability of 1 in 1.1 million chance of people having a
> certain sequence of DNA markers and you have a database of 550,000 people,
> you have a 50% chance of making a match. That's great, if you know that
> the perpetrator is in that database. But what it also means is that as you
> start testing DNA profiles against more and more people, the chances that
> you will match an innocent person to a DNA profile from a crime scene gets
> higher.
>>>
> http://blog.wired.com/27bstroke6/2008/05/feds-to-collect.html
>
I've been banging on about this for years. It's called signal to noise
ratio. The more data you hold, the less useful it is.
Ironically, governments desire to get us all DNA fingerprinted could lead to
the technique of DNA analysis to be made useless.
|