Glen Alleman had a recent posting on sundry quotations about chance and uncertainty -- something we PMs deal with every day (no exceptions!). Here's the one I like -- and it justifies taking a course in statistics (See Khanacademy.org)
Glen's editorial on this: "all variables in project work are random numbers. Without knowing the underlying statistical process and the probabilistic outcomes, no credible forecast for future performance is possible. Deciding anything in the absence of probabilistic confidence is simple not possible"
Not so fast!
Of course, here's the issue: we can't know the things Glen wants us to know... we rarely -- if ever -- have the history or observations depth to develop the 'statistical process' or 'probabilistic outcomes' in the 'frequency' sense of probability (to wit: how often does it occur?)
And, the 'laws of chance' are immutable, not subject to management. To wit: you can't manage the probabilistic outcomes of fair die tossed repeatedly. Thus, we PMs don't deal with the laws of chance very often -- they are more like laws of physics insofar as our ability to influence them.
But, Glen's last point -- estimating without regard to uncertainty -- is valid: Not possible.
So, what's plan B? Enter Bayes Theorem, which broadly stated says we can:
- Start the estimate with really no knowledge (if we don't have any), then
- Systematically improve the estimate with even limited observations (too limited to be of use in a frequency definition of probability, but 'good enough' to improve an outright guess)
Check out these books I've written in the library at Square Peg Consulting