When the government passed the same-sex marriage act last year, the law had one curious wrinkle: while it would allow gay couples finally to marry, couples already in a civil partnership would have to wait if they wanted to "upgrade" to wedded status.
One reason for the strange delay in a policy that passed with the approval of all three party leaders was both mundane and common in Whitehall: the computer said no. The various government IT systems supposedly couldn't cope with the changed status yet, and so plans to convert civil partnerships to marriage would just have to wait (and are still waiting to this day).
The story – although contested in Whitehall – has become symbolic of the power the government's IT systems now have over its policies. One impact assessment for the same-sex marriage act noted the law change would require updates to computer systems at HM Revenue & Customs, the Department for Work and Pensions, Home Office, Ministry of Justice and Office for National Statistics.
No wonder, then, that it's often IT projects that are among the most high-profile failures in government contracting collapses, as has proved the case this week with the failed e-borders system, which has landed the taxpayer with a £224m bill for costs and damages to Raytheon, the company contracted to deliver the scheme before it was unceremoniously axed by Theresa May in the early years of the coalition.
The e-borders farrago joins many other public sector cautionary tales, including £100m squandered by the BBC on a video archives system that never materialised, a £56m Ministry of Justice back-office project cancelled after the department realised the Cabinet Office had a system doing the same thing, and more than £10bn sunk into the NHS's national programme for IT before it too was shelved. The list of reasons for failure quickly becomes familiar to those who follow these things, such as Tom Gash, the director of research at the Institute for Government.
"There are some patterns here we've seen in a lot of other failed projects," he says. "The one that stands out is doing a 'big bang', announcing it all at once."
Rather than announce these high-profile technical fixes, he continues, governments should leave themselves some flexibility: break out the different things a new system should do (known as the requirements), and commission them one by one, to build something up over time – and pick up early if it isn't working. Similarly, taking direct control over subcontracting, rather than parcelling everything out to one or two big players, lets departments see what is going wrong earlier, he says. By doing this, and hiring more civil servants with business experience, Gash believes, the public sector is slowly getting better at handling such projects – despite often being legally outgunned.
Still, across IT and other major contracting projects, there may be trouble ahead. The Major Projects Authority is a central body set up to monitor large-scale and risky government enterprises. It is tracking almost 200 schemes with a total value of almost £400bn. Of those, just 17 projects – with a value of less than 4% of the total monitored – have been given "green" status (defined as having "the lowest risks to success") by the authority.
A particular threat to many longterm projects is a change of government: when the coalition came into power, it cancelled swaths of contracting deals. Bodies such as the MPA are one solution to handling and running such projects through changes of government, though Gash notes "such bodies often [themselves] get canned or restructured" during changes of administration. Another factor is to build cross-party consensus for projects which will take more than one parliament to complete.
All of which cannot help but draw attention to the elephantine IT project in the room: universal credit. Universal credit is a proposal to combine complex and myriad benefits into one payment and is Iain Duncan Smith's flagship proposal. It requires joining up data from numerous departments and subdomains to work effectively and relies on a small number of major contractors.
It is mired in difficulties and has already led to some of the largest write-offs of IT spending of all government projects. A report published by the Commons public accounts committee in November last year found that much of the £425m spent on the proposals to date were likely to be written off, including at least £140m for IT systems [see footnote].
Pilots for the project were dramatically scaled back. Even those sympathetic to the scheme have wondered if the benefits will justify the costs: the benefits system is not solely complex due to the number of benefits, but people's lives – if people have children, leave their partners, get ill (or recover), or find work, their eligibility will still change in future. It's not clear how much even the perfect IT system could save. Such an unpromising recipe is enough to make even those like Gash, working hard to be diplomatic and reserve judgment, venture a sceptical conclusion or two. "Universal credit … has a lot of the ingredients of projects that have failed in the past," he says.
• This footnote was appended on 5 September 2014. The article says the Commons public accounts committee found in November last year that "much of the £425m spent on proposals for Universal Credit would need to be written off, including at least £140m for IT systems." To clarify, the committee found that write-offs were likely to amount to £140m. The Department for Work and Pensions has already written off £41.1m in IT related to Universal Credit, according to its accounts for 2012/13, and a further £91m has been spent on IT assets that will support service for the first five years only.
