Tuesday, 21 July 2009

Decoding the DNA database

The British government is currently trying to circumvent a demand by the European Court of Human Rights that it reduce the scope of its beloved DNA database. In May, it launched a "consultation exercise" (pdf) setting out its response to the Strasbourg ruling in S and Marper. The two plaintiffs in that case, who had been arrested but not convicted of any crime, successfully argued that for the police to retain their DNA profiles indefinitely breached their human rights. The court ruled that their DNA, and that of people in their position, be removed.

The government's minimal proposal is to delete DNA details - where the arrest was for a relatively minor offence - after six years (or twelve, for more serious offences). It's hard to see how this meets the requirements of the EHCR judgement. S and Marper themselves would not have had their profiles removed from the database if these rules had been in force - and yet the court upheld their claim on the grounds that to retain them was disproportionate.

In its attempt to justify the largest possible database, the government has resorted to some questionable science - claiming, for example, that people who are arrested and cleared are as likely to commit future crimes as those who are convicted. Ben Goldacre has convincingly demolished their case, as have professors Keith Soothill and Brian Francis. Goldacre calls the consultation paper - which draws on work by the criminologist Prof. Ken Pease (to whom we shall return) - "possibly the most unclear and badly presented piece of research I have ever seen in a professional environment". His Bad Science blog and Guardian column usually concentrate on journalistic exaggeration, scaremongering and cut'n'paste reproductions of press releases; for the Government to be guilty of such shoddy work, in an area of such importance to both crime prevention and civil liberties, is frankly alarming. If not entirely without precedent. It is a criminological dodgy dossier.

Some of the most perceptive criticisms of the Home Office's approach - albeit expressed in appropriately judicious language - were contained in a speech made last month by Mr Justice Beatson (pdf).

The judge points out that the Home Office proposals fail to address many of the ECHR's concerns - as they are legally bound to do. He notes, for example, that to take into account the ECHR's reasoning "would require an assessment of the legitimate needs of privacy and autonomy" - something which appears not to have been done, apparently a deliberate ommission. Nor does the consultation paper address the ECHR's criticisms of the effect on the presumption of innocence of blanket DNA retention. The principle that one is innocent until proven guilty is the jewel in the crown of English jurisprudence, one of our nation's greatest gifts to the world. Now it has to be imposed on the government by a court sitting in a foreign country. It's too depressing for words.

Justice Beatson also notes the contradiction between the report's claim to be based on evidence and the figures of 6 and 12 years, which it describes merely as "a commonsense approach". Most damningly, he comments that "It is not apparent from the Consultation Paper whether [Pease's] research has been subjected to the normal scientific peer review that is a prerequisite of academic respectability." The findings of Soothill and Francis come close to peer review, and they are fairly devastating. Their main criticism is based, like Goldacre's, on the document's questionable use of statistics. But they make some broader points, too. For example:

The notion of “arrest” is the main criterion used for action in the consultation document. While police arrests are not whimsical, they come at the beginning and not the end of the criminal justice process. Some people are disproportionately at risk of being taken into questioning by the police and being arrested. In contrast, a conviction is the outcome of evidence being tested in court. In fact, arrests are useful indicators of police action but not of guilt. Re-arrests are dangerous indicators and making arrests the pivotal criterion encourages the notion that we are moving towards becoming a police state.

Southill and Francis are also struck by a confusion "between two types of discourse - a discourse relating to crime and a discourse relating to criminals" and by "a remarkable lack of discussion about persistent criminals". This, though, is no accident, as I hope to show.

The Home Office admits that the ECHR ruling requires that there is a distinction between people convicted of an offence and those merely arrested. Yet their proposal - to retain all profiles, however obtained, for six years - makes no such distinction (except one of time). Legally innocent people will continue to be regarded, as Soothill and Francis put it, as "honorary criminals". And, reading the lengthy document, it becomes clear why: in their view, there is no distinction. Acquittal, or even a decision not to press charges, is irrelevant. They appear to believe that only habitual or potential criminals get arrested - or else that everyone is either a criminal or a criminal-in-waiting.

Many people who get themselves arrested, however, are not criminals in any meaningful sense of the term. Some may be perfectly innocent bystanders who get mixed up in an incident simply by being there. Others may be peaceful protesters, or public spirited citizens tackling antisocial behaviour (in such circumstances, the police's general practice is to arrest first, ask questions afterwards), or victims of mistaken identity or malicious allegations. They may be leakers of official documents, or people otherwise caught up in political shenanigans (Ruth Turner, Damian Green, Lord Levy). They may be like the Southport teenager arrested because he found a mobile phone and handed it in to the police. They find themselves on the database despite being no threat to anyone.

This isn't to deny a correlation between being on the database and propensity to commit crime. Obviously a high proportion of arrestees will be of dubious character. The large, though decreasing, number of people who manage to go through life without ever being arrested are more likely to be law-abiding, if perhaps somewhat timid. But for any given individual, having been arrested is not a marker for anything. This doesn't mean, of course, that there will not be cases of people arrested for a minor offence who turn out to be serial rapists, and whose sample, given on arrest, leads to their conviction on a serious charge. There have been such instances - in one case, a man was convicted because the police held a DNA sample from his sister. But such instances are (however welcome) statistical quirks. The vast majority of people whose sisters are arrested do not turn out to be rapists or murderers. Media concentration on such cases distorts public perception, perhaps deliberately so.

The larger the database, moreover, the greater the chance that it will produce miscarriages of justice. There may be "false positives", especially if DNA found at a crime scene is fragmentary; or a person's DNA may be present at a crime scene quite legitimately (if it's a pub, they might have had a drink there) - but because they are already "on the database" they will be regarded with automatic suspicion.

But none of this matters to the Home Office, which has long since stopped thinking in terms of individuals and cares only about actuarial justice. The prevailing criminological doctrine is something known as Rational Choice Theory - a way of thinking about human behaviour borrowed from economics, and which asserts (to simplify) that people act in utilitarian ways to maximise their advantages. Other things being equal, the theory assumes, if you see a wallet lying on the ground, and there are no CCTV cameras watching, you will steal it. To quote a leading study of the subject, RCT "views all individuals, regardless of their social background, as capable of crime, and approaches its preventative measures without identifying any crime-prone group". Hence anyone who happens to be on the database is equally suspect.

The ultimate origins of RCT are found in the work of some post-Enlightenment thinkers, above all Jeremy Bentham, although its modern incarnation relies on the mathematical modelling made possible by modern computers. For most of the Twentieth century, such ideas were shunned in favour of "dispositional" theories of crime, which concentrated on the criminal: what made people criminals, and how they could be reformed and rehabilitated. RCT and its practical offshoot, Situational Crime Prevention, aren't interested in criminals (because everyone is or could be a criminal) but in crime as a kind of social hazard to be combatted via managerial techniques - a bit like swine flu.

This approach is not, of course, confined to the world of criminology. As Keith Haywood puts it (pdf),

One aspect of the ‘wider context in which crime occurs’ germane to discussions of RCT and SCP is the roll-out of neo-liberal forms of governance and control and the accompanying cultures of risk and resource management. Increasingly prevalent across the spectrum of social life, from primary education to health care, these managerial, ‘government-at-a-distance’ techniques frequently incorporate the principles of rational calculation and interest maximization in their analyses. The result is a totalizing language that speaks not of difference, contradiction or alternative moralities and subjectivities, but of a single, unquestioned rationality that sets itself up as the enemy of all forms of ‘irrationality’.

Haywood contends, unarguably in my opinion, that RCT is based on a "sociologically hollow narrative" about human behaviour.

You will not, I hope, be too astonished to discover that Home Office favourite Ken Pease, whose work forms the basis of the consultation paper on DNA retention, is a leading proponent of such methods. Haywood calls him "one of the high priests of the neo-classical revival". Elsewhere, Pease has argued that crime is best tackled by such techniques as "embracing crime-reductive design", promoting birth-control and even abortion (statistically, the products of unwanted pregnancies are more likely to grow up to become criminals). Rehabilitating offenders, by contrast, is relatively unimportant, because it "seems to be less profitable in reducing levels of crime, at least within the range of intensities of intervention which are routinely practicable" (No Through Road, 2006, pdf). Better that potential criminals are never born, he seems to believe, - although, of course, that would mean aborting the "innocent" along with the "guilty". But - hey - fewer crimes. Chilling.

Pease also quotes with approval the words of Prof Per-Olof H. Wikström:

The key suggested inhibiting mechanism is conceptualized as deterrence, defined as the perceived risk of intervention, and associated risk of sanction if acting unlawfully in pursuing a temptation or responding to a provocation. Deterrence occurs in response to monitoring. The potentially deterrent effect of monitoring is influenced by the individual’s executive functions, through the self-control exercised.

In other words, people will only be good if they are being watched. Human beings tend naturally towards evil.

In the Home Office document, Pease interestingly chooses to speculate about the possible deterrent effect of being on the database . He has, he writes, sought published work on this issue "but can find none. It seems important to remedy this." Research over many years, as he is well aware, has failed to show a strong deterrent effect even for capital punishment. But that's scarcely the point. If the purpose of the DNA database is to catch criminals, then it would be more effective - cheaper, faster and less prone to error - if it concentrated on the guilty, or those against whom there was at least some evidence. But that isn't, ultimately, the purpose of the database.

Pease's approach is one of misanthropic utilitarianism and hyperactive risk-assessment: like ubiquitous CCTV cameras, the DNA database is a technocratic device to monitor society, to reduce the scope of unobserved action, to squeeze out uncertainty. It follows that it should be as comprehensive as cost or public opinion or the law will allow. That the innocent should be included as well as the guilty is not an unfortunate side-effect, it is a necessity. Because they, too, will be deterred from committing crimes. And because there's no such thing as an innocent person, just someone who hasn't committed a crime yet. Or hasn't been caught. See "Independent Safeguarding Authority" for further details.

This pessimistic view of human nature is now firmly entrenched at the Home Office. From the vantage-point of Whitehall bean-counters, it is of little consequence who is guilty or innocent, whose privacy is being invaded or what potential for false-positives and other mistakes there may be. They have simply ceased to bother about individual civil liberties. As Tony Blair used to say, all that stuff belongs to another age. This is the 21st century. The computers have taken over.