Anthony Sampson: Gullible governments and crafty salesmen always manage to create computer chaos

People who commission new systems rarely think seriously about what they want to, and can, achieve

Why do so many government computer systems still end up in extravagant fiascos, with huge escalation of costs and without achieving their purpose? This week we learnt that the Child Support Agency had spent £456m on a system that had completely failed to provide the payments - desperately needed by single mothers - it was set up for, a state of affairs which led to the resignation of the chief executive, Doug Smith.

Why do so many government computer systems still end up in extravagant fiascos, with huge escalation of costs and without achieving their purpose? This week we learnt that the Child Support Agency had spent £456m on a system that had completely failed to provide the payments - desperately needed by single mothers - it was set up for, a state of affairs which led to the resignation of the chief executive, Doug Smith.

But we have heard it all so many times before. It is only the latest of a long series of computer scandals stretching back more than 40 years. First, an ambitious scheme is announced which promises to rationalise a whole department, whether for passports, taxes or benefits; then a few years later the scheme turns out to be causing backlogs, cock-ups and overspending, and sometimes has to be abandoned altogether.

Why can't governments after all this time effectively supervise this computer-mania? The obvious answer is that they have always been outgunned by the giant multinational companies. The computer salesmen knew how to exploit the ignorance of governments everywhere, and to play on the insecurity of civil servants. And they knew how to make their clients dependent on their expertise.

The huge EDS company, which set up the system for the Child Support Agency, was first founded by the American master salesman Ross Perot (who later campaigned to be the president of the US); now it provides half the British Government's computer services. "It's raison d'être is to make itself indispensable to its client," said the editor of Computer Weekly, Tony Collins.

The dependence of governments was understandable in the days when computer systems appeared incomprehensible and mysterious. But now computers are part of everyday life and every manager has a laptop. Yet the clients seem as gullible and vulnerable as ever to the salesmen.

It is not just governments and civil servants who are responsible for huge blunders. The Treasury assures me that the overspending on government computers is no greater proportionately than the overruns in private companies. And today there are plenty of comparable cock-ups in the private sector.

Sainsbury's supermarkets, which depend heavily on data banks and bar-codes, are currently in the midst of computer chaos after installing a new system, which appears to be emptying their shelves instead of filling them. And Penguin books are now hard to find in bookshops because of computer failures.

But governments are rightly most criticised because they commission the biggest contracts of all, and they should be accountable to the taxpayers. After 40 years they should have established a formidable regulatory force, with independent expertise. Yet they still appear to be the biggest suckers of all.

Why have they learnt so little? Civil servants who have watched these failures all seem to agree about the cause: the people who commission the projects have not thought seriously about what they want to, and can, achieve. They look to a new computer to solve problems and answer questions they have not properly asked themselves.

Computer companies, with all their jargon of wizardry and specialisation, have always led their clients to believe that they can bypass the normal mental processes, and take the burden of thinking away from them. Clients are always looking for "state of the art" systems that promise new breakthroughs. But major programmes, even if carefully thought through, always involve trial and error to overcome teething problems; and an existing system which has been tried and tested has the huge advantage of having learnt from previous mistakes. To develop a new program which involves catering for thousands of individual people is full of special hazards: for individuals always behave with exasperating independence - changing their addresses, spouses, partners or even their names.

The computerising of the Child Support Agency - with all the complications of single mothers, missing fathers and wayward children - provided a field day for computer-chaos, as we now know from the bizarre evidence in this week's hearings of the Work and Pensions Select Committee. It involved 2,000 changes before it could be implemented, and it came up against all the nightmares of "error messages", malfunctions and shutdowns that drive staff to distraction and technological paranoia.

And now there is scope for far greater chaos in the bigger area of the National Health Service as it tries to implement the huge new system planned to cover all patients in the NHS, to provide millions of medical histories and to allow them to make medical appointments for themselves.

For ambitious planners it offers a computer utopia, building up a comprehensive national database and bypassing all the human contacts and errors with hospital staff, and promising individual choice and direct access to a centralised system, according to the best New Labour principles.

But more than any other computer project it will come up against all the complications of human failings and idiosyncrasies, as patients change their minds, move between different doctors, hospitals and regions; records are lost, and symptoms and diagnoses are mistaken.

And it provides the prospect of huge escalation of costs: the first estimate of £6bn will certainly be rapidly exceeded, and the maintenance and additions will multiply that sum, with new opportunities for overruns and revisions.

In medicine and hospitals, more than in any other field, the human factor and contacts are all important; computers can never be a real substitute for the basic understanding between patients and doctors in their own communities.

But all computer operators work on the frontier between fallible and unpredictable human beings and an inhuman machine - a kind of caricature of an impersonal bureaucracy - that is full of perils. It cannot be completely delegated to huge corporations which naturally wish to extend their technology, and to maximise their profits.

Governments have to be sure that they can rely on their own departments to have first-class experts with long experience, but also independent motivation and careers, to enable them to confront corporate salesmen and technocrats on equal terms.

They must be able to question and supervise rigorously each new ambitious computer project commissioned in Whitehall, to make sure that it is not misconceived and mishandled, like the CSA fiasco; that its managers have really thought through what they want to achieve, and can relate their schemes to ordinary human beings.

Two months ago a new head of government information technology was appointed: Ian Whatmore, formerly the British head of the American company Accenture (formerly the management consultancy affiliate of Arthur Andersen, the ill-fated accounting firm). Whatmore is reputedly the highest paid civil servant in the country, but he will need to justify his pay, by showing in high profile that he is constantly vigilant against the blunders of government computing.

Above all, he must go back to the central warning of all computer programs: the principle of "Gigo" - garbage in, garbage out.