Is that database fit for what purpose?

A recent FIPR alert reporting the result of an FOI on the accuracy of DVLA files adds grist to the mill of debate over the database society. Over 10% of vehicle records and over 25% of driver records contain errors, mainly because we drivers do not tell DVLA about our changes. Most other government departments could make the same complaint – because we will not make the effort unless we have a good reason to do so.       

The basic principles of information management have to be understood and appreciated by those taking policy decisions which assume that centralised databases will be fit for purpose. For example – that data must be entered and/or verified by those who need it to be accurate and are in a position to know whether it is. The reasons for such basic principles to be followed are rarely understood, let alone taken into account. In consequence many public sector databases are so full or errors that they are not fit for purpose – if the purpose is timely and accurate decision-taking on matters that are important to the individual. 

This can, and all too often does, lead to bad decisions at every level – from avoidable death and suffering to equally avoidable cost and waste (including because those with decision responsibility have lost faith in the likely accuracy of the records put before them).  

Do take a look at “The Truth is Out There” the latest Audit Commission study.

Then listen to the recordings of the workshop organised by the Audit Commission and the Information Society Alliance on the current state of public sector information sources and the issues arising. [a summary report will be available shortly].

Then be very afraid!

You can almost forget about data protection. It is data inaccuracy, (from random error, through systemic distortion to outright fraud) that is the bigger menace to the database society. 

We are beginning to learn from our colleagues in the former communist bloc that there is no virtue in telling the truth if false answer would do equally well. The more personas you have, the less likely it is that the Stasi will come for the “real” you. That applies even more so in the on-line world. Why on earth do you want your electronic identities to be linked by people you have never met, let alone have any reason to trust? But equally, how do you teach a naive teenager who has been protected from physical risk the limits of on-line trust?

The Information Society Alliance (EURIM) study on Identity Governance will be “interesting”. The first step will be to “educate” those who believe in the “self-evident truth” that we need a holistic (albeit perhaps federated) approach to electronic identity that the majority of mankind does not share that belief. The second is to demonstrate the kaleidoscope of approaches that are already in use – working in some contexts while they fail in others.

Hopefully we will then be able to home in on the principles to be enshrined in governance frameworks that will encourage inter-operability across those systems that we have chosen to trust – while recognising that we may well tell different “stories” to those we trust, as well as those we do not.


FOI Request on DVLA database accuracy statistics

1. Vehicle database

As at 30th September 2009 there were 34,395,400 licensed vehicles on the DVLA database.

An estimated 88.60% of vehicle records are correct in every respect.

An estimated 11.4% of vehicle records contain errors.

The most recent statistic on the percentage of vehicle records where the registered keeper of a vehicle cannot be traced from the details held on the DVLA record is 4.3%.

An estimated 2.7% of vehicle records contain errors that originated within DVLA.

An estimated 1.18% of vehicle records contain incorrect information submitted by registered vehicle keepers.

An estimated 7.52% of vehicle records contain incorrect information because the registered vehicle keeper has failed to update DVLA about a change of details.

2. Driver database

As at 14th December 2009 there were 43,958,637 Driver records on the DVLA database.

An estimated 73.82% of driver records are correct in every respect.

An estimated 26.18% contain errors.

The most recent statistic on the percentage of driver records where the driver cannot be traced from the details held on the DVLA record is 20.62%.

An estimated 2% of driver records contain errors where the error originated within DVLA.

An estimated 3.56% of driver records contain errors that originate from incorrect information submitted by the driver, and 20.62% stem from the driver’s failure to update DVLA with a change of details.