UK Border Agency flip-flops on asylum seeker DNA tests

Refusal

Rather than undermining their case, I would a refugee to be very wary of DNA testing.

Alan Johnson shows he dangers of DNA profiling

Once again, Alan Johnson, the new Home Secretary that replaced Jacqui Smith 'stamps' his authority on a bad decision of Jacqui Smiths.

But it shows the problem with DNA profiles of innocent being stored, sooner or later someone will think they can match particular markers to particular crimes, then will use those markets as a predictor for those crimes.

Weak thinking like Alan's is the first step down that route, and the later steps become far easier.

It will be a science paper: "T82764 marker combination correlates approx to Theft".

Alan Johnson: "We'll give T82764 marker tests to the police to aid in policing"

Police: "We have a theft, lets check the T82764 people first to see if we have a case against any"

Alan Johnson Mk2: "Our analysis shows that yes indeed, the link between T82764 & theft is far stronger than previously thought, so we'll make the presence of T82764 admissible as evidence in court for theft prosecutions"

Alan Johnson Mk3: "We are getting many more theft crimes prosecuted, but strangely the number of thefts is not going down, we need to lock up all T82764s from birth to prevent crimes".

People like this are a threat to a free society.

History is written by the winners

As I commented previously, the Home Office has orders to make this work. The narrative for Nature, etc will be concocted afterwards.

An African has a DNA marker

So an African says he's from Somalia, but his DNA tests shows a marker, this marker has been found to be in 95% of the population of a particular valley, but only in 20% of normal Africans. Geneticists have concluded that the people's of that valley don't travel widely because the marker is so concentrated there.

What is the probability of that Asylum seeker being from the valley?

Well lets see, population of Africa is 680million, say 5000 people live in the valley, so (5000*0.95) / ( 680000000*.2) = 4750 / 136000000 = 0.000035

i.e. the test tells you nothing.

I created at petition on the number10 site in the hope of ending this.

Link to petition http://petitions.number10.gov.uk/BorderAgency-DNA/

Link to original article http://blogs.sciencemag.org/scienceinsider/2009/09/border-agencys.html that alerted me to this

Link to Reddit post where we can hopefully raise the profile of this issue http://www.reddit.com/r/science/comments/9t9kr/petition_to_stop_the_border_agencys_dna_testing/

Link to Digg post http://digg.com/general_sciences/Scientists_Decry_Flawed_and_Horrifying_Nationality_Tests

Madness

Just because the whole premise of the test is complete bollocks, does not mean that we should not use it to give brown foreigners a hard time. In fact it fits in perfectly with the rest of our logical and humane immigration policy.

They can't cancle it now...

I already spent the kickback from my golf buddy who "won" the contract.

@Anonymous Coward - Hippocratic oath...

"Medical procedures are supposed to be undertaken for the benefit of the patient - not the benefit of the state."

Well said mate - but that fight is already lost, good and truly, in almost every area where govt and medicine overlap. The vast majority of today's doctors are in it for the money and the (considerable) perks - the few with a conscience (especially over their astronomical earnings) don't number enough to matter.

The BMA? It's a mutual admiration society that rarely punishes other than those who affect the profession's public image...

David X. Li & Gaussian copula function

http://www.wired.com/techbiz/it/magazine/17-03/wp_quant?currentPage=all

I think this also illustrates the problem. You have a method that determines a correlation between one thing and another, however there is a slight feedback: If the equation says that X correlated to Y, then the action taken on that statistic, causes X to more closely correlated to Y, which feedback into a stronger correlation from the equation, round and round and round.

The feedback causes the correlation to fly off into fantasy land.

This is what happened in the financial collapse brought on by the Gaussian copula function. The equation told them they were safe, but it was fantasy.

So NuLabour start with a correlation and it is unreliable, and they intent to use it anyway. And it shows that any idea that DNA profiles of innocent people will not be used to separate people into groups by their genes, is totally wrong.

That is step 1 in Eugenics.

don't worry

We all know the govt is committed to policy based evidence.

Id

I can understand them taking a biometric id so that if they are refused and booted out they can be identified when they invariably return but this stuff seems like pseudo-science at best.