Slashdot Mirror
New DNA Test to Solve More Cases
Krishna Dagli writes From the BBC,"Tens of thousands of unsolved crimes could be cracked with a new forensic technique, it has been claimed.The Forensic Science Service (FSS) is piloting a computer-based analysis system which can interpret previously unintelligible DNA samples.It claims the technique is a world first which will boost its crime detection rates by more than 15%.The method is being tested by the West Yorkshire, South Yorkshire, Northumbria and Humberside police forces."
...it will hopefully free lots of people who have been falsely accused of crimes they didn't commit.
I'll form my OWN solar system! With blackjack! And hookers!
West Yorkshire. A large number of crimes have recently been solved in this slumbering community. Using a new forensic technique, crime investigators were able to implicate most of the current police force in what was previously reported to be "unsolved" or "mysterious" crimes. 'As our DNA evidence clearly shows, the whole police department was involved in the crimes and their cover up. The crimes were then classified as "unsolved" to cover up their tracks. We have never seen such a wide spread of corruption.' Unidentified sources claimed that the are similar investigation of the police force in 3 other communities.
But if your suspect hasnt been taken into custody as of yet? Why not just 'swab' the entire world population.
---- Booth was a patriot ----
He speaks English.
The Queen's English, that is.
Be careful who you kiss then... A good loving french kiss, and you're suddenly not the suspects' girlfriend, but the suspect himself!
It actually doesn't work out that way, but I can imagine a movie coming out shortly with that as the main storyline, you wait and see...
Two problems: 1)The system/process will be made mostly available to "solving" crimes, not freeing criminals; it's bad prioritization politically, existing criminals could swamp the system, and if a guilty criminal were released after a false negative and was a repeat offender, there'd be hell to pay. 2)While a "maybe a match" will certainly be grounds for the police getting warrants and such, a "maybe not a match" won't get a convicted criminal much.
I'd expect if anything for them to be very cautious about using this tool; DNA match evidence is widely perceived as completely reliable by juries, public, judges, etc...and a less-reliable matching will erode that confidence.
Please help metamoderate.
England, you fool. You know, the place where all the guys in the funny hats and white wigs invented, well, uhh...English. Now, any Brits here want to say the obligatory "Damn Yanks?"
Good. Cheap. Fast. Pick Two.
You mean the DNA tests for the past few decades havent been 100% ?!?
This sounds like Intel's marketing department.
"Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
Now, any Brits here want to say the obligatory "Damn Yanks?"
I thought only we Southerners said that? And trust me, you have never heard "Damn Yankess" until you've heard it from someone from Mississippi.
Fly me to the moon Let me sing among those stars Let me see what spring is like On jupiter and mars
Damn Yanks!
How do you face your accuser when it is a piece of software?
The ghosts of the many executed prisoners who are soon-to-be-exonerated applaud......sort of.
Never play chicken with a passive aggressive.
That's because it takes them about 4 1/2 minutes to drool (drawl?) it out their mouths.
DNA fingerprinting basically measures the number of repeat units at ~13 different locations in the genome. As you have 2 copies of your chromosomes you esentially get a unique 26 digit code. Looking at just one of the repeat locations, let's say the normal range is 8 to 12 repeats, so you can be an 8,8 8,9 8,10, 8,11 8,12 9,9 9,10, 9,11 9,12 10,10 10,11 10,12 11,11 11,12 or 12,12 (so there are 15 possibilities seen in the population, with generally similar frequencies for each). The chance of matching your sample with any randomly selected unknown will be 1 in 15, but if we go up to 13 different markers we have (1/15)^13 which gives a chance of any 2 UNRELATED individuals matching being about 1 in a quadrillion (more people than have ever lived, and likely ever will live). This means a match is a definite match, this doesn't mean evidence wasn't planted or some such conspiracy crap, but a match is a match, no chance of a collision like with 2 people having O- blood.
The "new" stuff here is that they have come up with software which will allow the system to extract 2 sets of "seial numbers" from one reaction. Like having 2 fingerprints on top of one another and seperating them to determine the swirls. They also are claiming a more sensitive technique which will allow for smaller or partially degraded samples to be tested, but this is probably just tweaking the experimental protocol.
This is no new test, just tweaks and algorithms.
Does this mean we will finally find out who "Jack the Ripper" was once and for all?
The /crime detection/ rate is up by 15%? Just great. How many headless corpses and savage beatings were we not noticing before this? Are they finally going to start ticketing that bastard who parks in the middle of four spaces in my complex or something?
Yahoo! Pipes are awesome. How awesome? http://pipes.yahoo.com/jesdynf/slashdot
And when it reaches 100% nobody is safe.
No thanks; far too busy beating my head off a brick wall in frustration at the police state we've ended up with.
Don't worry, I'll be removing all DNA traces from the wall afterwards...
We dont have the death penalty.
As has already been commented, it's not necessarilly valid to simply multiply out the individual probabilities as if all the elements were truly independent, since they may not be, and 15^13 is a very big number and an exceedingly bold claim to make. FTFA the chance of a random match between two people is stated as being about 1 in 1 billion, which is roughly what I've seen quoted before for the type of test currently used by the Forensic Science Service.
Now you throw in the effect of the Birthday Paradox, and what you'll discover is that in a population of 60 million people, such as the UK, you should expect to find roughly 3.6 million [*] people who share a profile with at least one other person. Or in other words, there's roughly a 6% chance that there's someone else in the UK with a profile that matches yours. Think about that - what it means is that a national/international database simply doesn't work, because it doesn't scale, even if the test is accurate enough to be (mostly) unique on the current database size.
[*] actually, compared to the birthday paradox, when the random match chance is very small and the size of the database is very large, it's very easy to calculate a good approximation of the expected number not unique. All you do is imagine picking two people at random from the population, the chance that they match must be 1 in one billion. Assume there are n profiles which are shared between two people, and it follows that:
i.e. with a population of 60e6 there will be 1.8e6 profiles shared between two people, for a total of 3.6e6 people. Discounting the number of triples and higher, of course, which isn't too important for this purpose because once you're starting to get triples "uniqueness " in the database has long since been lost!
To have a right to do a thing is not at all the same as to be right in doing it