The future of the Labour government's G-Cloud plan and how much of it will survive the Cabinet Office spending...
and IT review will depend on whether the department has the determination to force through big changes - for challenging vested interests, for market-making, and for establishing a central authority over an unruly IT industry.
That much can be gleaned from reports, published this week by the Cabinet Office, produced by Phase 2 of the G-Cloud programme, containing the technical, commercial and philosophical blueprint of the plan in its most idealised form.
G-Cloud aims to bring the principles of the cloud to public sector IT through massive datacentre consolidation, sharing services and establishing a central "app store" of relevant software.
From the perspective of Phase 2, on which some 100 people from government and industry worked to produce a set of detailed proposals, there is little reason for compromise. Its aim is simply to gain the efficiencies promised by cloud technology. It would do this the cloud way: break the status quo by establishing an open technology infrastructure upon which an open market of suppliers can operate. It's a popular proposal for good reason.
Back in the real world, there is a handful of major IT companies who supply the IT that operates the incompatible computer silos that populate the public sector. It makes them - and costs taxpayers - a lot of money. But some observers fear the Cabinet Office likes the status quo and doesn't want to spend the billions of pounds G-Cloud would require.
This may be apparent in the G-Cloud Strategic Outline Case (SOC) - effectively the business case - completed last summer under the aegis of John Suffolk, the former government CIO, which until now has never seen the light of day.
The SOC describes how the Cabinet Office established the G-Cloud business case on numbers supplied by just six incumbent IT suppliers who critics say have a vested interest in the status quo. Over just two weeks between 30 November and 14 December 2009, BT, Cable & Wireless, CSC, Fujitsu, HP/EDS and IBM generated numbers to justify a scheme that could challenge their incumbency.
"Together these organisations comprise a significant percentage of government external datacentre ICT spend," said the SOC report.
The questionnaire, obtained by Computer Weekly, asked suppliers to provide, for each of 12 time periods, estimates of costs and benefits for each of 26 categories of ICT spend in each of six imagined G-Cloud scenarios - that's 1,872 numbers.
The six scenarios were the possible G-Cloud strategies being considered by the Cabinet Office. They ranged from simple rationalisation to the grand vision of a single, private G-Cloud for the whole of government, both with and without making use of the public cloud providers who present such a competitive threat to the surveyed suppliers, and with and without the migration of existing applications to the paradigm-breaking G-Cloud system.
The figures were unique among the outputs of the seven "workstreams" that comprised Phase 2 of the programme: the SOC was the only report not subject to peer review among the 100 people in the various work teams. The other reports were freely distributed and matured.
The Phase 2 business case for G-Cloud should be treated with deep scepticism, according to workstream contributors with the most to gain. It would indeed suit them: the only possible inference to draw from the numbers is they are not favourable enough - by multiples of 10, they say.
The SOC claimed expenditure of about £0.5bn would deliver savings of £1.5bn in the first four years. That's good enough for most.
Of those who contributed the numbers moreover, BT, Fujitsu, HP/EDS and IBM were among those who formed the 10-strong group of Intellect members who fleshed out the original G-Cloud proposal the Cabinet Office took to the CIO Council in 2009. But some members of Phase 2 told Computer Weekly the incumbents may have neutered the G-Cloud by making the numbers sound less attractive than they really are.
It is tempting to think of all such numbers as a best guess, though they are merely the best available. The only part of the G-Cloud proposal that could be formulated with anything approaching numerical certainty was datacentre consolidation, which is what the supplier estimates addressed. The other forecast benefits - such as procurement reform and the innovation engendered by opening the market up to SMEs - are inestimable.
But there is more to be drawn from the flimsy numbers. The report lists departmental estimates provided by the Office of Government Commerce (OGC) - now part of the Cabinet Office Efficiency and Reform Group running the G-Cloud programme.
The OGC reckons each department's potential share of net benefits derived from G-Cloud will be roughly proportional to its share of total government IT spend. So the NHS, which spends 37.76% of the government IT budget, would take 43.41% of G-Cloud benefits. Local authorities spend 18.36% and would take 19.58% of the gains, and so on.
But OGC numbers can be unreliable. The unit can be reduced to asking suppliers to give cost/benefit data for their own services. The Operational Efficiency Programme reported in 2009 that it was fruitless trying to benchmark IT spend because there were no standard ways of measuring anything. Sir Peter Gershon said the same on founding OGC in 1999.
Suffolk repeated the complaint before he stood down last year. As if to illustrate the point, he put it alongside the flaky numbers in the G-Cloud SOC. The common standards necessary to make G-Cloud operate would do more than free the time civil servants usually waste managing disparate, un-interoperable systems, said Suffolk in the report.
"Standardisation will also enable ICT departments to benchmark IT spend across the public sector in an easier and more meaningful way," it said.
The call for common standards is repeated over and again in the workstream reports. G-Cloud wouldn't work without them. They are a "pre-requisite", one of the primary project requirements, and would be needed for everything from individual data items, through application interfaces, and on to quality of service commitments.
The questions the Cabinet Office must answer when it publishes its new IT strategy in March regarding these standards are how, and how much. Phase 2 did not conclude whether common standards meant open standards or IT industry standards. It may make a difference.
Suffolk's team, which was a fairly broad bunch, concluded that if business communities were left to define their own standards, they would not likely be defined in common. Expensive public sector IT silos would be sustained by their inability to get onto the universal G-Cloud network and, as is the idealised plan, exchange IT resources and applications as one might exchange data on the internet.
"The conclusion is that the business community governance option will not provide sufficient commonality across ICT infrastructure, or sufficient value for moneyIt therefore will not be carried forward for evaluation," they said.
The idea floated in its place was for a pan-government standards body that would provide by diktat the glue necessary to hold G-Cloud together. Suppliers and local authorities alike would be told what standards to use. Perhaps because this proposal sounds so unfeasible when technology standards evolve as quickly as the innovation that creates them, the idea that government will leave G-Cloud to operate by whatever happens to be the dominant industry standard has not faded.
That leaves the threat that dominant vendors will continue to use their control of standards to sustain their market position at the expense of customers such as the UK government. Phase 2 considered the nightmare scenario that G-Cloud might lock government into the technology provided by whichever supplier happens to build it. There are open cloud standards, but not industry cloud standards because competing dominant suppliers prefer their own.
This is the most important question facing the G-Cloud planners. It is more important than the matter of whether the cloud is public or private, though that question gets more air because vendors obsess about it more.
David Chan, director of the Centre for Information Leadership at City University London, says he argued the case against industry standards with Suffolk, who thought they could be elected by market forces.
The market would not have created the internet, he says: "The World Wide Web, the IP protocol, the DNS domain name system: those haven't been defined for the cloud. To say the marketplace would have resolved those is ridiculous," said Chan.
The G-Cloud is, not coincidentally, being made possible by the installation of the Public Sector Network, a free-forming network of government networks, like the internet, which uses open internet technology by necessity.
The internet has been a success because it is governed in an open forum comprising of any party, from vendors though customers, architects through policymakers, with an interest in formulating the technical standards by which the network of disparate networks operates. It has no higher aim. ICANN, its governing body, operates under US-government edict. It is the model of governance for the systems interoperability the G-Cloud seeks to exploit.
The system of G-Cloud governance that is chosen could do more to water down the G-Cloud proposals than any lack of investment by the Cabinet Office.
|Department||Overall annual IT budget (£000)||% of total IT cost||Potential share of net G-cloud benefits|
|Department of Health||4,178,780||37.76||43.41|
|Ministry of Defence||1,391,000||12.57||8.67|
|HM Revenue and Customs||713,733||6.45||7.15|
|Department for Work and Pensions||603,000||5.45||5.51|
|The Scottish Government||574,250||5.19||3.79|
|Ministry of Justice||289,375||2.61||1.85|
|Department for Business||117,902||1.07||1.26|
|Foreign & Commonwealth Office||127,511||1.15||0.92|
Source: G-Cloud Strategic Outline Business Case, Cabinet Office, 2010