It is eight years since the OECD called for “dolphins not whales” in a plea for smaller projects when they warned of the “hidden danger to e-Government” caused by over-large ICT projects, cautioning that they “should be avoided wherever possible”. Yet, it seems that the lessons have still not been learnt. The ‘goldfish memory’ of public sector ICT procurement persists along with the high-profile disasters.
The famous and often quoted studies into ICT project failures, such as the original 1995 Standish Group CHAOS report and later UK publications by the Public Accounts Committee, the National Audit Office and POST, reiterate the same basic findings: the problems are caused by unclear objectives, poor understanding of requirements, inadequate project management, insufficient user involvement and lack of top management support. All of these “people problems” are very important, to be sure, but far less attention was paid to the problem of project size, tacitly accepting that big is the only way to do public sector ICT.
There is however another way and other countries are showing us how it can and should be done. Research conducted into public sector ICT procurement in the developed world found that the country doing this best was Holland, where they are not afraid to break large projects down into smaller parts, despite the possible accusations of aggregation the UK seems paranoid about. The research also, not surprisingly, placed the UK at the bottom of the successful public sector ICT project league.
But, there’s a snag: splitting big projects into smaller ones and managing their implementation and interoperation requires government to be an intelligent customer – you can’t just hand off the whole problem to a supplier. The Dutch government values and retains in-house technical skills whereas in the UK they have been chronically undervalued and, assisted by outsourcing and civil service fragmentation into agencies, have relentlessly migrated to the private sector.
Rebuilding civil service ICT skills is a worthy, if long-term, goal but what can be done in the short term to alleviate the problem? High street retailers solved the problem of uneconomic over-large department stores by adopting the “shop within in a shop” concept allowing flexible response to changing market conditions. The Web, in a way, does the same by decoupling big infrastructure (Internet and the Web) from service delivery (Web applications) which can be developed, deployed and updated rapidly.
Instead of massive stove-pipe systems embedded into dedicated back-ends and complicated client-side applications, provide public sector staff with “Web terminals”, connecting to common backend servers and databases all procured and maintained as commodity infrastructure. All new functionality can then be developed as smaller projects: Web applications to run over that infrastructure. Development would be simpler, faster and less risky, supplier dependence would be greatly reduced and small, innovative companies would at last have a chance to compete with the big boys, assuming the government fixes the massive bureaucratic obstacles they face to getting public sector business.
It sounds like an uphill struggle but maybe the recession and falling government spending power will force small to become the order of the day. Now, that really would be beautiful.
Published on 27 October 2009 in Computer Weekly, under the title "Think small to avoid public sector IT disasters"