They refer to £161m that the Home Office spent on the IT-based project, C-Nomis. The project aimed to provide a single database of offenders, to be accessed by prison and probation officers.
The NAO said: "We have not been able to ascertain precisely what this money spent on."The Home Office had not kept proper records on the C-Nomis project, said the authors of the report.
How is it possible, in the age-old machinery of government, for £161m to be spent without proper records being kept, such that auditors have to speculate on where the money went?A disinterested observer is entitled to ask whether the Home Office has opened the door on anarchistic practices and the risk of corruption.
We "believe", said the NAO, that the Home Office spent most of the £161m developing the software and testing it in prisons. But it does not know for certain.
The report's authors were unable to "determine the full value of the waste and inefficiencies associated with the failure of the C-Nomis project with certainty". This was because the National Offender Management Service's recording of costs was poor, it said.
The audit office found major decisions were taken without proper authority. There were no formal controls on changes to the contract and initially no penalties on the supplier for late delivery. There was a "vacuum of leadership".
There have been improvements since Jack Straw's Ministry of Justice took over the project in 2007.
But the NAO still finds some surprising lapses. Information on the expected cost and time-to-completion of C-Nomis was missing. This is the information one would expect to be depicted in lights on any large IT-based business change project or programme.
With characteristic reserve, the NAO says of the information that is missing: "Without this information the project manager cannot identify where money has been spent or what has, or is, to cost more than planned."
Even today the Ministry of Justice relies on "unvalidated cost figures".
Ministers gave good reports to Parliament on the state of C-Nomis in 2006 when the project was failing, so it is possible that Jack Straw will not be told the whole truth.
A technical report on the project was submitted by Accenture in September 2008. It said:"C-Nomis suffered from cost over-runs, scope changes, delays, restructuring and insufficient governance management lacked early warning of programme failure."
The project was due to cost £234m but the latest estimate is £513m for a scheme that has been reduced in scope. Accenture has warned that costs might rise further.
Press releases issued every month by Whitehall give the impression that the government machine runs as smoothly as Big Ben. The Office of Government Commerce does regular assessments on dozens of mission-critical IT projects and programmes in the public sector.The OGC has thousands of online pages of advice on project management.
But it slipped out in 2003 that some project planning in government was all but non-existent.
Peter Gershon, then Chief Executive of the OGC, said in a speech: "There are still far too many projects reviewed by Gateway teams where, frankly, project planning is little better than something on the back of a cigarette packet."
Has much changed?
Several NAO reports have recognised IT-related successes in government. But it's the failures that people remember, and it is especially worrying when the NAO cannot establish precisely how millions of pounds has been spent on an IT-based project.
It is a matter likely to be taken up by the Public Accounts Committee when it interviews civil servants on the C-Nomis project in the coming months.