I propose Stoke's Law, which is that as the amount of data that the government collects grows, so will the number of people who are victims of crimes that were made possible by unauthorised access to government databases.[From Digital Identity Forum: A new law]
We never really settled on the shape of the Stoke's Law curve, leaving it as a square law (ie, the amount of crime goes up as the square of the amount of data collected) but I'm really beginning to wonder if this is steep enough. This is because, in the U.K. at least, civil servants and management consultants appear obsessed with data sharing, which of course makes the problem much worse. It's no surprise to see stories about the abuse of government databases appearing with apparently increasing frequency. For example, I read only last weekend of a case in which a civil servant was tapping into databases to pass a woman's details on to her violent ex-partner so that he could track her down. This wasn't for money — the civil servant was the new girlfriend of the violent man in question — but could have had a much more serious outcome than the kind of identity crime (ie, credit card fraud) that the government says is a priority with respect to the national ID card scheme.
As someone who believes that cock-up rather than conspiracy is the guiding principle of government IT, I have to say that corrupt civil servants passing on information to criminals is unlikely to be the biggest problem with the joined-up administration imagined by the designers of new public sector infrastructure:
Government records are notoriously inaccurate. If a person is wrongly listed in a database, the problems of that error are now amplified.[From Concurring Opinions]
When government databases were inaccurate and distinct, the errors were there but it was difficult for them to propagate. Now they will be able to zoom around at the speed of light.
This is the intractable problem with data sharing. Aside from general hopelessness when it comes to doing complicated things with computers, the government is hamstrung by the tension between security and (as they see it) efficient public administration. Thus they end up with crazy patches to the "system" as in the case of the Children's Index:
The database won't give you the address and phone number of every child: the government has said that people who are rich and/or famous will not have their children's addresses and phone numbers stored in the index[From Digital Identity Forum: Failsafe]
Now, none of this is to say that attempts at joined-up government are either wrong or doomed to failure. What I am saying is that for them to succeed, there must be a more sophisticated model of identity at their centre.
Sir James Crosby's report, commissioned two years ago by the Treasury, accuses the government of adopting an "uncoordinated" approach to the problem of identity assurance.[From ID card report criticises government's approach | Politics | guardian.co.uk]
He was, of course, correct. But a more co-ordinated approach to identity assurance does not mean creating a single identity and linking any and all public sector databases to that identity. There is no reason for your "health identity" to be the same as your "police identity" or your "waste disposal identity". If someone is a risk of life and limb because of the criminal dumping of waste, then the police should be able to apply for a warrant to obtain the police identity corresponding to a waste disposal identity but they should not be able to simply trawl around, because if they can, then the bad guys can too.
These opinions are my own (I think) and are presented solely in my capacity as an interested member of the general public [posted with ecto]
Dave, a cogent and piercing insight, as usual… thank you.
There’s another point about erroneous entries in databases, and I thing it’s analogous to Gresham’s Law, except that in this case, ‘bad data drives out good’. Or rather, it doesn’t drive it out, it just outnumbers it.
The reason I think this is that, if two organisations have databases, each with an entry for Dave Birch, and one of the entries is erroneous, what’s the most likely result should the two organisations share data? Probably, that rather than try to resolve the apparent conflict, they will simply store both, hoping that it will never actually matter.
In an ideal world (with a tiny number of records), one might actually do something to resolve each conflicting set of records – but as anyone knows who has ever tried to sync two slightly conflicting calendars, the problem very quickly becomes unmanageable, and the temptation is either to give up, or live with the duplicate records…