Following on from a previous post of mine, Frank Patrick points out another reason why simple formulas have hidden dangers when it comes to calculating potential task durations:
When considering task estimates, we should simply face reality and recognize that estimates are best communicated in terms of ranges and best agreed to as such, and not as single point commitments. The idea of “accurate estimates,” over which too much time, effort, and angst is too often spent in project planning, needs to be set aside.
I’ve often thought that life would be easier (for project managers, at least) if we could get away with using the type of predictions that the Bank of England uses for monetary inflation (PDF link). The graphic is embedded in the PDF, unfortunately, but imagine a line graph where the end fans out into a range of possible values – the expanding ranges shaded by their probability, the darker the more likely (and closer to previous values).
This neatly illustrates both the potential range, and the potential likelihood, of the values. It eliminates the danger of focussing on a single value, but manages to convey the full range of possibilities. It’s an effective graphical way of saying, “there’s a 10% probability of X, but a 50% proability of Y”.
The problem with this approach is that Gantt charts don’t deal with probabilities – they deal with arbitaryly-selected values. And however much you caveat a Gantt chart with ranges and disclaimers, once you’ve produced a hard number, that’s the number that’s going to be quoted.
