Home Office builds 'harm model' to track fear, threats
How high do press clippings figure in it?
Much of the UK Home Office's recent legislation has been driven by the 'fear agenda', which is to some extent articulated by previous Home Secretary David Blunkett here (although he thinks he's denying it). A curious interview with Sir Stephen Lander in the Independent, however, reveals that the agenda is to an extent being set by what the press currently happens to be most exercised about, and that (rather more interestingly) the Home Office is developing a "harm model" intended to measure the extent of threats and to prioritise them.
In our short acquaintance with Sir Stephen we have grown to love him dearly; his lack of media savvy is a heartening boon for freedom of information, and should be savoured in that brief period before his bosses twig and get a bag over him. His manifest joy when Dimbleby mentioned wiretaps should really have alerted them in November. This time around, the Indy tells us, Home Office researchers have been going through about 30 newspapers, broadsheet, tabloid and regional, covering five years. They've then worked out the importance of organised crime issues by number of column inches and number of stories. Organised immigration crime came first, with drugs second.
Yes, we know, this is potty, because great swathes of these stories are likely to have been driven by Home Office initiatives, press releases, ministers' speeches and comments, and so on. Which is of course one of the key flaws in Blunkett's pitch that fear of crime, terror, and the unknown (code for foreigners) should drive government action even in the absence of data justifying such fears. As a rough and ready measure of this The Register searched homeoffice.gov.uk for documents including all the words "home office immigration crime press release" (613 hits); "drugs crime press release" (728 hits); and "press release" on its own, which yielded 2,225 hits. This certainly shows the Home Office doesn't talk about drugs and immigration crime all the time, but it sure as hell talks about them a lot.
Blunkett's previous statements have however left little doubt that tabloid-fueled fear has been a primary driver of government security policy and legislation, so it's scarcely surprising to learn that the Home Office has teams of people trying to measure its level. It is however surprising that the head of 'Britain's FBI' sees nothing wrong with formalising this by admitting it in public, and the "harm model" sounds like something we'd all like to know lots more about. How much IT is involved? Is it intended to be predictive? How much is the new Home Office Oracle system involved? And, given that many of the threats that are alleged to be being tackled by, say, ID cards are precisely those the Government hasn't been able to quantify in public, does the harm model include any more enlightening data?
There are quite a few possible FOIA requests there, we'd judge. According to Lander the model "basically articulates the harm that is caused to the UK under a number of headings - the rewards taken and made by the criminal; the social and economical harm to the UK; the institutional harm - corruption for example and illegal immigration - and tries to put a cost [on them].
"It also brings into play judgements about the degree of public concern and they have a proxy for this, which is the amount of column inches in the press. Which is not quite right, but is probably as good as you will get. It is pretty rough and ready but it is asking the right questions. It is asking not, what is the incidence of something, but what is its impact."
If we parse that a tad we can suggest that the model could in theory provide a measurement of damage caused by crime, but that in practice that measure is likely to be substantially compromised by the relatively arbitrary nature of the data that goes into it. What, for example, does organised crime cost the UK per annum? £20 billion or £40 billion? Or what does benefit fraud cost, £2 billion, £7 billion, or something else entirely? These are all numbers the Government has used, and they really mean that the Government just plain doesn't know. It's not clear whether or not the model takes account of more precise data the Government has (e.g. British Crime Survey or the Office of National Statistics immigration and asylum data), and if so what importance is placed on that data.
In addition to the arbitrariness of the input data and level chosen, the weighting of the components in the model is going to be decidedly non-scientific. Given the information we've got so far (which amounts to there being something in there, and approximately what it's supposed to do) it's probably pitching it a bit strong to compare it directly to madder US schemes like TIA, but it does appear to share the faulty premise that you can measure everything, and once you've done so you'll actually learn useful stuff.
Back in the clippings department, one of the most troubling things about the model is the fact that the amount of space papers devote to specific issues is something that can be measured accurately (never mind that the meaning you derive may not have value in itself). We therefore have a model where public opinion as measured by newspaper coverage is quantifiable, 'firm' data, being looked at by a Government which already obsesses about public opinion, and which will seize on any 'data' that justifies its own fears. In an election year. Yum.
In addition to the revelation of the harm model, Lander had a couple more snippets on terrorism (which SOCA doesn't actually do, but never mind). On Osama, he tells us: Al-Qa'ida and the linked groups are much more difficult [than the IRA] because of their differing nationalities. You have to work hard to get to the first base, which is who is who. We are on the first phase - identifying who is who." If you don't know who the people are, then it seems plausible to suppose you're not in much of a position to identify the organisational structure yet. In a speech in November, meanwhile, Eliza Manningham-Buller, who succeeded Lander as head of MI5, said: "'Al-Qaida' has become shorthand for other terrorist groups or networks that, inspired by al-Qaida's successes, and in imitation of it, are now planning attacks against western interests."
Which suggests that "identifying who is who" is an infinitely-extensible exercise. ®
Sponsored: Fast data protection ROI?