The New Scientist this week is running a series of short articles on how to make the world a better place. One of the suggestions is to legalise drugs – I’ve blogged about why this is a good idea before (and Ben Goldacre has a nice account of why we haven’t done it already). Another is to learn to love genetic engineering – and again, I’m all for that.

But one of the suggestions is that the police should have access to a universal DNA database – and this is not something for which I’m cheering. It would not make the world a better place – or, at least, it’s by no means certain that it would, and it might actually be detrimental. (Coincidentally, the NS article is published on the day that the man who discovered genetic fingerprinting has come out against keeping the DNA of the innocent, iterating a point he’s made before.)

The article points out that

DNA profiling is a powerful forensic technique. It has led to the conviction of many criminals who would otherwise have eluded justice, and the freeing of many innocent people who had been wrongly convicted.

and continues:

One possible answer [to problems of criminal detection] – not yet adopted anywhere in the world – is to record everybody’s profiles at birth or on entry to the country. A universal DNA database of this kind would not solve all crimes: in many cases, no DNA sample is available, or would be irrelevant because the alleged perpetrator’s identity is not in question. But it would ensure a match could be found for most crime-scene DNA samples, while also ensuring people arrested but not convicted are treated no differently to the rest of the population. There is no other way to achieve this.

Well, it might be true that a universal DNA database would help in solving some crimes – note that it’s only some – but there’s still an open question about the propriety of the tool, and whether it’s a monkey-wrench to crack a monkey-nut.

In the first place, it’s worth remembering that having a DNA database wouldn’t do anything to reduce the crime rate, for reasons that I’ve explained elsewhere. So the better world wouldn’t be one in which you have less of a chance of being a victim of crime. In other words, wed be dealing with a world in which there’s no noticeable difference to the rate of offending, but in which there is – perhaps – an elevated chance of getting caught. Note, though, that it is only perhaps an elevated chance – and even then, a lot depends on the crime in question being the sort for which DNA evidence makes any odds either way.

Of course, there are some crimes where the perpetrator might well leave DNA evidence: things like rape and other kinds of physical assault are the obvious ones here. However, whether that’s going to be what secures the conviction, if conviction there is, is another matter entirely.  Take rape, for example: often what skewers a prosecution here is not the question of whether or not Smith and Jones had sex, but the question of whether that sex was consensual.  That’s much harder to prove one way or the other, and it’s a question that DNA evidence is going to be precisely no help in answering – and I can see situations in which a rapist would, having been tracked down generically, happily admit to having had sex with the accuser, but being reasonably confident that this left (a) a favourable impression of cooperation witht the jury and (b) the hard question unanswered.  Moreover, many, if not most crime, scenes are places where there’s going to be DNA from several people knocking about (perhaps less so with rape cases, but the general point stands).  So a person’s being picked up by a DNA scan in no way tells us that he ought to be convicted.  It doesn’t even tell us that he ought to stand trial.  In fact, it might not even tell us that he was at the scene of the crime, since hairs and skin cells get all over the place.  So whatever benefits there are to having a DNA database, there’s also this risk to balance.

Moreover, the article gives the impression that a database would help the innocent.  It wouldn’t, just because innocent people would be flagged up alongside the guilty.  And because juries tend to believe DNA evidence to be cast-iron, an innocent person’d likely as not find it harder to prove his innocence.

Of course, proving innocence is only strictly necessary at appeal – at trial, it’s guilt that has to be shown – and that’s another problem.  A DNA database doesn’t assume guilt, but it assumes potential guilt, and that’s almost as bad.  It means in practice that the person whose cells are found has to be able to account for their presence at the crime scene.  And there’s a moral hazard implicit in our being treated as suspects ab initio anyway.  And that, I think, is a serious moral argument against a DNA database: it’s corrosive of a trusting relationship between state and citizen.  And a world in which that trust is corroded is, as far as I can see, a worse world.

Yes: a DNA database might generate desirable outcomes – but it doesn’t follow from that that it’d be desirable.  There’re too many big questions about a database generating undesirable outcomes to take that for granted.