Devil's Advocate: The big deal of government buying small

The issue of government procurement of IT is hardly a new one. But, asks Martin Brampton, is the current approach entirely the wrong way around?

The government is being questioned over its plans to build a lot of big databases. Notably, David Blunkett's pet scheme, the national ID card, needs a database of the entire population of the UK. Is it really wise, though, to ask government to integrate all its schemes involving databases?

There are, of course, the concerns about government's aim to become Big Brother. It has just become public knowledge that the Oyster card, the key to regular travel around London, does not merely validate the holder's entitlement to travel, it also records journeys made. Combine this with your tax record, health record, email activities, mobile phone and credit card usage and you have a pretty formidable array of information.

Giving all this to a government that is already going down the road of locking people up merely on the basis of suspicion is surely enough to alarm even the least paranoid - especially when it is combined with a tendency to remove government decisions from the scrutiny of the law courts. These trends concern me but this week I wanted to look at something different.

Are the simplest questions about effectiveness really being tackled? Is government spending money sensibly on the creation of databases or would it be better to combine several projects into one single super project?

Computing seems to think it knows the answer and is campaigning for projects to be coordinated. Yet it is highly improbable that would lead to greater efficiency. The individual projects are already large and it is common knowledge that big, complex projects are disproportionately expensive. Building a system for multiple government departments - each jealously guarding its independence and promoting its own goals - is a recipe for disaster.

Might it not be better to go quite the other way? Are there ways to build multiple government databases, if they are really needed, at lower cost? Then it will be easier to ask simple, fundamental questions about the projects and the way in which government goes about buying technology.

One of the most interesting questions is whether government actually has a sensible procurement policy. Just because a database is large, it does not mean that the software is necessarily complex. Clearly it has to be capable of scaling up but the question of what has to be done may still be quite simple.

And anyone who has worked in software development knows that pretty complex systems can be built by a dozen people in a year or less. As the size of team grows beyond that, it becomes increasingly inefficient. When team members are added at the last minute to speed things up, the result is frequently the exact opposite.

So then we start to wonder why it is that government only deals with a handful of giant service providers. There are countless smaller organisations up and down the country that have the skills to build new software, many of them highly resourceful and experienced. But government procurement finds itself unable to deal with most of them.

Peter Gershon is justifiably proud of having increased the proportion of SME suppliers in the OGC approved lists. Yet he has to admit that the SME suppliers in the lists are only a tiny fraction of all the SME technology companies. Most find that government procurement policies work against them.

Ironically, it is often the principle of open and fair competition that keeps them out. Currently, procurement is so difficult and expensive that buyers are pushed towards framework agreements with the big suppliers. Small companies do not have the resources to compete for large-scale framework agreements and are then ruled out of the small projects they could tackle well.

When it comes to the big projects, the tendering process is beyond the capabilities of all but the biggest suppliers. Rather than roll together already large projects, it might be better for everyone if government were looking at ways to develop many smaller developments, using a wider range of providers at a more local level. The inefficiencies involved in overlapping systems might well be outweighed by the efficiency of each individual system.