And although the NAO's often excellent reports highlight important lessons - which have been learned at a high cost to taxpayers - projects in government still fail to meet expectations for the same old reasons. That is one justification for an investigation by the work and pensions committee into why IT projects go wrong. It wants to know, for example, if systems at the Child Support Agency have a future and whether an IT-led modernisation at the Department for Work and Pensions - one of the biggest such undertakings in the world - is likely to succeed.
MPs on the committee have good reason to be concerned. At the work and pensions subcommittee last week several of those who gave oral evidence, including representatives of suppliers, pointed to things that are wrong with government IT. Computer Weekly has been reporting on the common factors in IT-related project failures for more than a decade, to no avail.
The main answer, we believe, is to strengthen accountability. In the US there is legislation, the Clinger-Cohen Act, which sets out the broad principles of how projects should be run and how central departments will account to Congress for their progress, or otherwise. Deviations from standards and procedures, costs and criteria for measuring their success must be reported.
It is not a perfect system. But it has led to improvements, according to the General Accounting Office, the US equivalent of the NAO. And it has strengthened accountability.
We cannot expect legislation to talismanically stop IT failures in government. But if a new law were allied with open Gateway reviews this should increase accountability immeasurably. And it is accountability that is key to avoiding failures.
When the House of Commons debated the work of the Public Accounts Committee last month it was revealed that Italy has the lowest rate of IT-related project failures. Its equivalent of the UK's Office of Government Commerce has the power to halt or curtail major projects, as has the US president's director of management and budget.
In the UK, the power to halt or curtail projects lies with the departments that have the most emotional and financial equity in schemes, and therefore the strongest reasons not to halt or curtail one that is failing.
To counter this potential conflict of interest, Gateway reviews on risky projects are carried out by independent and well-respected individuals, often those with a proven record of success in delivering IT systems. But their reports become the confidential property of the project's senior responsible owner.
The SRO is usually the business head of the department, who "owns" the project on behalf of its potential end-users. As the review belongs to the SRO, he or she is not compelled to show the result of a review even to the department's IT director, any of the computer staff or the suppliers, and certainly not to the potential end-users, Parliament, the public or the media.
Yet some NAO reports have referred in detail to Gateway review findings. The publication of parts of these reviews by the NAO has had no adverse impact. No one has been named, fired or had his or her careers harmed. To our knowledge no department has refused to be honest with Gateway reviewers because of the NAO's reports.
It is always easy to support the existing state of affairs out of timidity or simply deference to the cultured opinion of the day. Computer Weekly could easily collect influential support for Gateway reviews to be kept entirely confidential, but that would not stop disasters in government IT. We understand why Gateway reviewers do not want their reports published. Many chief executives in the private sector would rather not have their decisions scrutinised by shareholders and auditors. Some ministers would prefer to make decisions without having to worry about general elections.
Yet departmental heads have the power to commit to projects that cost billions of pounds without adequate accountability to taxpayers, boards of directors, the media or Parliament.
No departmental head has ever suffered any harm to their career as a result of an IT disaster. Nor will they if Gateway reviews are published. In practice they cannot be fired, and certainly not for doing their duty. But if the reviews were published they could be held accountable for going ahead with a project that was fundamentally flawed. That may make them think twice. Legislation that enshrines good practice - an integral part of Computer Weekly's campaign - would give a framework to make them further accountable.
The main argument against publishing Gateway reviews is that departmental executives would not be honest or frank with reviewers if they knew reviews were to be published. This argument does not hold water. The NAO carries out dozens of investigations into major government projects every year and publishes its reports - yet it does not complain that it is unable to get the truth.
Extracting the truth may be difficult, of course, but Gateway reviewers are doing such an important job - the future of millions or even billions of pounds may depend to some extent on their reports - they cannot, and should not, expect an easy life.
The cost argument
We better understand the cost argument against publishing Gateway reviews. Many reviews take place in three days. In this time reviewers cannot be thoroughly investigative. Were the reports open to the scrutiny of all, the reviewers might feel the need to be more investigative, which would cost more. But Gateway reviews are carried out only when there is a medium or high risk of failure. Many of the projects reviewed cost £100m or more. Surely it is better to pay a little more for a longer investigation which leads to a doomed project being halted, than save money with a short one which endorses a defective plan?
In the end, the arguments against publishing Gateway reviews can be boiled down to a single unpalatable assumption: that reports must be kept confidential because otherwise we could not trust the reviewers to get to the truth, and we could not trust those questioned to be honest and frank.
If we accept this, we accept that the system of government is septic - that the civil service is so unprincipled that it will lie or omit the truth when it knows its practices and information are to be scrutinised by Parliament, departmental end-users, taxpayers and the media.
On the other hand, if we think - as Computer Weekly does - that those who serve the country in central departments are committed public servants who will always do their jobs to the best of their abilities, and be open, frank and truthful in all professional circumstances, what has anyone to fear from publishing Gateway reviews?