The UK papers are again full of stories about project overruns in time and cost (the Grey report on the MOD). The majority opinion believes the process to be broken and the problems endemic.
A recent study (by Flyvbjerk et al) found that time and cost estimate errors in government procurement were skewed to underestimates and hence generate overrun in execution. The conclusion was this was deliberate, that estimates were known to be optimistic and that it suited the ‘system’ or Establishment to so underestimate. It was more acceptable to underestimate and overrun rather than face the true estimate.
When I was taught to estimate I learnt that I should develop from experience my personal error factor. Once I have created my best estimate of a job I should apply my factor to create a usable estimate. Most people underestimate and their factor is greater than one.
The MOD factor appears to be 1.4 for cost (40% historical overruns) and around 1.5 for time (five years late on projects). I can see the Establishment struggling with such factors being applied to project costs and timescales and then having to make decisions… It is far more palatable to accept the estimate, approve the projects and live with the consequences of overrunning years later – well ‘they’ got it wrong! I feel the study has its point.
To fix the problem implies changing the principles of the Establishment and without a radical approach that is not going to happen. In the commercial world organisations with such poor performance go bankrupt and the market need is filled with newer, forward-looking companies who can get it right.
A starting point would be for a (commercial) fast-track procurement organisation to be provided with (modest) procurement budgets, their remit being to buy solutions to threats being faced by our forces in the current asymmetric conflicts. The organisation would use extreme RAD approaches e.g. SCRUM, to meet specific problems with specific solutions in months rather than years.
Could it be done? Historically, yes – think Barnes Wallis.
Comments