GAO needs to reinforce Agile concepts in IT project planning
The Government Accountability Office (GAO) survey on “IT Cost Estimation” in eight US Government Agencies takes pretty much a traditional view that Government IT projects often fail because their plans are not detailed enough.
The main ‘take away’ from the GAO report is that whole-life costing is the Achilles heel of many Government business cases. The Patents system, for example, whilst having adequate plans for the $130m Agile development costs, had ommitted $58m of running costs from the whole-of-life business case.
A good point – but: will the GAO’s calls for Departments to spend more time (and money) drawing up more and more detailed plans provide more reliable forecasts of performance? I am not convinced…
Despite the GAO being satisfied with the IT cost estimating procedures at the Department of Defense they noted that the DoD Tactical Mission Command programme still did not have credible plans. Two of the 18 programmes (Patents and Health Data at VA) both claim to have adopted an Agile approach, but the recent problems at the latter just show the importance of taking note of the GAO’s advice to take a wider view of a programme than just the technical development costs in isolation. The one isolated ‘perfect’ programme they found was the $1bn CANES programme, which despite having met the GAO criteria for “comprehensive, well-documented, accurate, and credible plans” has still encountered many set-backs and changes in direction.
My suggestions for improvement:
- Do not mix up the need for overall estimation for a business case, with the fetishisation of over-planning in detail which often results in ‘waterfall’ style project failures
- Deliver fast and early, and measure the ‘velocity’ of development to continually test the business case. If actual progress is slow, do not blame the estimation techniques, but spend time focussing on more acheivable objectives
- Work out what products you want first, rather than focussing on the tasks. Please, GAO, ask planners in the USA to start using ‘Product Breakdown Structures’ for planning, not ‘Work Breakdown Structures’ !
(Thanks to Jim Greene at QSM (Quantitative Software Management) for jogging me into action to comment on this GAO report!)