Thursday 5 September 2013

Now UC IT

The National Audit Office (NAO) have published their report on Universal Credit (UC). UC is the Department for Work and Pensions (DWP) initiative to rescue benefit claimants from the poverty trap created by the UK's inept welfare system. The idea is to rescue them by making work pay.

Universal Credit: early progress is 60 pages long. 60 pages which document the unrelenting and expensive failure of DWP to get to grips with UC. There is a summary for you kindly prepared by Tony Collins – Will Universal Credit ever work? – NAO report.

By 31 March 2013, DWP had spent £425 million on UC. £425 million spent by intelligent and experienced public servants and there is nothing to show for it.

Accenture have picked up £125 million of that money, IBM £75 million, Hewlett-Packard (HP) £58 million and BT £16 million. That accounts for £274 million. £274 million spent with intelligent and experienced software engineers and there is nothing to show for it.

Is it the politicians' fault (Iain Duncan Smith, the Secretary of State at DWP, and his junior ministers)? Is it the officials' fault (Robert Devereux, Permanent Secretary at DWP, and his staff)? Is it the contractors' and consultants' fault? Yes. In each case.

How on earth can such a catastrophic failure happen? It's happened before, please see for example It's all John's fault. The lessons never seem to be learnt.

It's time to stop this nonsense. DWP have "pressed the reset button" apparently and are taking time out to think. About time, too.

The thinking so far centres on the software engineering methods being used. DWP, it is said, failed to use "agile" methods. Appendix Seven of the NAO report, beginning on p.53, provides a handy cribsheet on agile v. traditional software engineering.

This may be a cul-de-sac. After all, no engineering methodology in history has ever recommended spending £425 million before thinking what it is you're trying to achieve. Also, there is no guarantee that agile methodologies would avoid the same problem.

To the extent that "agile" means anything in Whitehall, it means the Government Digital Service (GDS). GDS are great advocates of agile, they claim to be successful exponents of agile and they want to see central and local government become 100 percent agile.

They're getting their message across.

Howard Shiplee, the man in charge of UC for the past 100 days, says in his Telegraph article Universal Credit: The First 100 days:
As the Secretary of State outlined in July, we are working with the new Government Digital Service (GDS) to explore an enhanced IT programme that would offer more flexibility and security to benefit claimants. We’re planning to take the best of the existing system and make improvements using GDS support.
Why?

The BBC and the Guardian give GDS great publicity, please see GDS PR blitz. So do the Times, please see Toe-curling: GDS PR Blitz.

Why?

The Design Museum declared GDS's only product to date, GOV.UK, to be Design of the Year 2013. The Design and Art Direction charity created a new category this year especially to be able to give GOV.UK a prestigious D&AD award.

Why?

The answer in each case is, presumably, competent public relations. An attractive brand is being created. But is there any substance there? What skills of GDS will stop the next £425 million from being wasted?

According to five IT professors, none.

Martyn Thomas gave evidence to the Public Administration Select Committee to the effect that GDS are wasting their time with agile software engineering, please see Digital-by-default, an open letter to the House of Commons Science and Technology Committee (para.13).

That's one professor.

The other four – Alan W Brown, John A McDermid, Ian Sommerville and Rob Witty – reviewed GDS's Government Digital Strategy and were entirely unimpressed. "Simplistic and highly risky", they said about agile, please see Four professors review the Government Digital Strategy.

Just because GDS's staff are an alternative to the hopeless staff of DWP, Accenture, IBM, HP and BT doesn't mean that they're any better.

D&AD, the Design Museum, the Times, the Guardian, the BBC, Howard Shiplee and the NAO would all do well to consider the expert views of the five professors before assuming that GDS is the answer. In the meantime, for the sake of the £425 million lighter taxpayer, and everyone caught in the poverty trap, another reset button should be pressed. On GDS.

----------

Updated 21 October 2013
  1. House of Commons oral evidence taken before the Public Accounts Committee, Universal Credit, Wednesday 11 September 2013
  2. Welfare fiasco chief 'to resign'

Updated 14.4.16

In the 2½ years since the post above was written:
  • GDS's all-agile system written for DEFRA's Basic Payment Scheme failed, leaving farmers to apply for their EU Common Agriculture Policy subventions using pencil and paper.
  • Iain Duncan Smith has resigned.
  • Robert Devereux hasn't. And he has become Sir Robert Devereux KCB.
  • DWP have fought against Freedom of Information requests to publish the 2011 and 2012 Universal Credit (UC) risk register, issues register and Major Projects Authority (MPA) assessment. They have finally lost that fight.
  • The MPA have become the Infrastructure and Projects Authority.
  • Some of the documents now disclosed suggest that ministers and officials at DWP did, indeed, mislead everyone about the progress being made on UC. Cyber security arrangements were inadequate, the system would have been open to fraud, there was no precedent for agile being used at the scale of UC and DWP didn't even have a plan for the transition from the existing benefits schemes to UC.
UC is utterly benighted.

As to GOV.UK Verify (RIP), another fairly major infrastructure project where Whitehall keep telling us that there is only good news, indeed the system is meant to go live this month, it's decision time some time in the next 16 days, what do the MPA have to say about cyber security and the use of agile?

Nothing.

The MPA, sitting in the Cabinet Office, haven't assessed the Cabinet Office's GOV.UK Verify (RIP), even though it's meant to provide 60 million people in the UK with an on-line ID, using which we are meant to be able to transact with government.

Risk level? Unmeasured. Could be high. Could be low. The MPA don't know and presumably don't care.

No comments:

Post a Comment