In software development, or cooking barbecue, when we estimate our small errors will net out – more often than not it will take longer than planned, sometimes less. I rarely remember anyone’s estimate really improving materially the more time they spent on it.
But spending time working on estimation will ALWAYS cause a project to take longer by at least the time you spend obsessing over adding precision to your estimates, simply because you’re taking time estimating and not doing.
It’s therefore much more important to estimate quickly and frequently, assuming and accepting that there will be error, but not obsessing over false precision. It’s more important to prioritise assymetrical work – work that has a high potential benefit for a low investment. Once again, the calculation of the ROI should be done quickly – say as t-shirt sizes (small, medium, large etc) rather than as pound (or dollar) values with decimal point precision.
By limiting the work in progress, keeping the backlog short, the prioritisation focused and the requirement for approval to a minimum, we can deliver the most value in the shortest amount of time.
Simple?