Tuesday, October 07, 2008

The Trouble with Data

William Heath alerts us to a 'terrific article by SA Mathieson in Government Computing in response to the earlier piece by Matthew Taylor about government use of data.'

"He points to limitations and unexpected side-effects of government driven by data. He argues that it ignores the human costs in something like abolishing the common travel area with Ireland, or of endemic workplace surveillance, and adds the dangers of new forms of discrimination eg against those unable to provide biometrics, or whose data is inaccurate...

The trouble with data by SA Mathieson

On a comparison of timetables, many of Britain’s train services are slower now than two decades ago. Part of the reason is that train companies are judged on punctuality, with fines if they miss targets, so they pad the timetables.

In September’s GC, Matthew Taylor, chief executive of the RSA, cogently made the case for increased government use of data. He sees it as a way for the government to govern more intelligently and to help those disadvantaged in society, such as in choice of school.

Gathering and analysing data is the dominant management technique of the age, standard practice for the management consultancies that have influenced and got much business, out of this government. Yet even in the private sector, managing by data has its drawbacks. In the public sector, these problems are significantly magnified.

The first, as with train companies, is that when data is used as the yardstick, those being judged start ‘teaching to the test’. Sats tests, taken at 11 and 14 mainly to gather data on the performance of schools, have a financial and administrative cost. But the opportunity cost, of training and examining all children for tests with little point to them when they could have been learning, is surely greater.

This is the main problem with managing government by data: you have to gather a lot of data on the governed.

The human costs of this are often ignored by government...

...data discrimination can hit two groups. There are those who do something wrong then suffer disproportionately. With the greatly increased use of criminal record checks in employment, a trivial offence long ago can narrow someone’s chances in life years after they have supposedly paid their debt to society.

Then there are those who suffer because their data is wrong, whether through error or fraud, and a greater reliance on data makes its fraudulent use much more attractive. The time taken sorting out the mess tends to be spent by the innocent data subject, not the organisation which fails to keep records properly.

It would be daft to say that government should stop using data. But the current government has tended to treat its gathering and analysis as a panacea. It is not."

There was another post at IdealGovernment that really tickled me recently and I never got round to blogging about it: the reverse Turing test.

"Dealing with the daily IdealGov spam dose it occurs to me we should have a reverse Turing test. If the original test is

a human judge engages in a natural language conversation with one human and one machine, each of which try to appear human; if the judge cannot reliably tell which is which, then the machine is said to pass the test

then our reverse Turing test is

a citizen or customer engages in a natural language conversation with one public servant and one machine, each of which adheres to prevailing policies and targets. If the citizen cannot reliably tell which is which, then the public servant is said to fail the test

We could include that in Sam’s UK Feedback/Bureaucracy Bingo service.

Wibbi public servants took pride and were rewarded for their humanity and not for their conformance to rules and targets? Wibbi we dropped the idea that personal service is something we can mechanise? "

Absolutely spot on!

No comments: