DNA Registries

[Yoda]

Considering the number of people falsely accused of crimes and the heightened security nesseccary today, would a worldwide (or at least UN) DNA Registry be acceptable?

Should everyone be required to submit a sample, or should samples be mandatory at birth, with it being optional for everyone else?

Feel free to debate/discuss.

[Wyrm]

Unless the DNA profiling is damn good, any search through the registry will have a large number of people who match the same profile, just by chance alone. If the false positive rate for a DNA comparison is one in one million, then if the whole world is in that registry (~6 billion), then there will be ~6000 people in that registry who match that profile through chance alone, along with the one person to whom the DNA profile belongs. A random person picked out of this pool would have a 6000/6001 chance of being accused wrongly of a crime.

(It's not just DNA registries; any large registry based on a comparison that is prone to false positives will have similar problems.)

Of course, performing ~6 billion DNA profiles with enough resolution to have a fighting chance of picking out the one unique individual in that registry, as well as performing one each time you want a comparision, is not going to be cheap.

Given that any test prone to false positives is going to be defeated by obscenely large numbers, extra steps are obviously necessary to assure guilt or identity, by cutting down on the pool of people who are likely to be guilty or have authorization. In the authorization example, a simple security card is already enormously effective at cutting down the number of individuals who can access a high-security facility. Coupled with biometrics such as the DNA profile and/or fingerprinting, security can be pretty tight.

And if the range of possible people who are authorized is that restricted, why compare against everyone in the world?

As for crime and punishment, the principle is the same, although the ways to cut down the pool of likely suspects are quite different.

I say "no" to the worldwide DNA registry. There are far better ways to assure identity and guilt.

[Invictus ChiKen]

Unless the DNA profiling is damn good, any search through the registry will have a large number of people who match the same profile, just by chance alone. If the false positive rate for a DNA comparison is one in one million, then if the whole world is in that registry (~6 billion), then there will be ~6000 people in that registry who match that profile through chance alone, along with the one person to whom the DNA profile belongs. A random person picked out of this pool would have a 6000/6001 chance of being accused wrongly of a crime.

[nitpick]Wouldn't process of elemination counter this? Say some of those 6000 not being in the same hemisphere for example?[/nitpick]

[Yoda]

[nitpick]Wouldn't process of elemination counter this? Say some of those 6000 not being in the same hemisphere for example?[/nitpick]

That would definately cut down significantly on the suspects. Whith prope methodology the results could be narrowed down even more.

As an example: A black, 6'0 man breaks into a corner store somewhere in the U.S. While searching for the money, he rubs against a counter, the man gets into his car and drives away.

Using these criteria alone, you are likely to come up with thousands of suspects. If you factor in geographical and sociological information however, the number of suspects will decrease greatly.

[SeeingRed]

Using these criteria alone, you are likely to come up with thousands of suspects. If you factor in geographical and sociological information however, the number of suspects will decrease greatly.

But still not enough to be useful in suspect identification the way fingerprinting is now. Unfortunately, the state of the art of DNA sequencing and matching is not good enough yet to have it be a useful tool on scales as large as this.

Privacy issues would also factor into the consideration of this proposal in a big way. I'd say no to this proposal as well.

[Seggybop]

For what it's worth, from Wikipedia:

STRs do not suffer from such subjectivity and provide much better powers of discrimination, for unrelated individuals (of the order of 1 in 10^29 if using a full profile) It should be noted that figures of this magnitude are not considered to be statistically suportable by scientists in the UK, for unrelated individuals with full matching DNA profiles a match probability of 1 in a billion (one thousand million) is considered statistically supportable (Since 1998 the DNA profiling system supported by The National DNA Database in the UK is the SGM+ DNA profiling system which includes 10 STR regions and a sex indicating test, this test updated the SGM DNA profiling system on which the National DNA Database was founded in 1995. The SGM system included 6 out of the 10 STR regions used in the SGM+ system and the same sex indicating test, however the discriminating power of the SGM system was only considered to be supportable at 1 in a million) . However, with any DNA technique, the cautious juror should not convict on genetic fingerprint evidence alone if other factors raise doubt. Contamination with other evidence (secondary transfer) is a key source of incorrect DNA profiles and raising doubts as to whether a sample has been adulterated is a favorite defense technique. More rarely, Chimerism is one such instance where the lack of a genetic match may unfairly exclude a suspect.

[Wyrm]

[nitpick]Wouldn't process of elemination counter this? Say some of those 6000 not being in the same hemisphere for example?[/nitpick]

Yes, it would. But the users need to be aware that this one, single piece of evidence matches all 6001 of these people. If the database spits out 6001 names, then this should be clear enough. However, DNA is considered teh goald standerd by the idiots populating our jury pools, and the prosecutor's falacy is a distinct hazard.

(Also, this 1:1e6 only applies for the test that really does have an error rate that good in the real world. Introducing human error and contamination possiblities degrades performance, and not all DNA tests are created equal, so in reality this false positive rate may be much higher. There is also the possibility of a non-zero false negative rate, again from contamination or human error.)

DNA evidence is usually slam-dunk because the pool of availible possible guilty people is very small.

Now, I'm going to retract some of my practical objection to a world DNA register just a little bit. If it is ONLY used as part of a comprehensive criminal program, where the police gather a list of likely suspects before they go to the DNA evidence, whereupon they get the requested DNA profiles and compare them to their unknown, then I see no problem with the world DNA database from this angle.

This, of course, still leaves legal issues. There might be a decision theoretical problem so solve, but I'll get back to you when I work through it.