Agile Estimation - Learning from traditional Project Management
I have long wondered why Agile (well, read Scrum) planning is so simplistic.
Yes, having the team estimate the tasks and then agree to them is, frankly, a world away from the handed-down-from-on-high estimation found in most companies, big and small. Even where estimation is initially done by the teams, I have found a number of times that the estimates are either massaged, or are part of a in-all-but-name waterfall approach.
Even a little more sophistication can lead us to estimates that are dramatically more useful outside the team, and that tends to lead to teams getting left alone with a reputation for hitting thier promises.
So I suggest that all Planning Poker estimates are provided with PERT three-way estimation points, and then thyat is
Planning poker is based on Wide-Band Delphi which came out of RAND Corporation via Barry Boehm.
(Boehm: Know why you are building the software, know your architecture is upto it and make sure the initial release carries the promise.)
http://csse.usc.edu/csse/TECHRPTS/2007/usc-csse-2007-715/usc-csse-2007-715.pdf (Incremental commitment - a very Agile approach, basically using risk based analysis to judge if those paying will get what they are paying for - at each stage they can incrmentally commit a little more money. Or as Ward Cunningham puts it "Can leave with software representing the amount of money committed to date"
E = (O + 4N + P)/6 stdev = (pessimistic – optimistic) / 6 Combining estiamtes (chances of EPIC / backlog)
The combined mean is the sum of the two old means. 32.5 + 32.5 = 65 minutes. The combined stdev is the square root of the sum of the squares of the two old stdevs. SQRT(7.5^2 + 7.5^2) = 10.6. The combined optimistic estimate is the mean minus ( 3 * stdev). Optimistic = 65 – (3 * 10.6) = 33 minutes. The combined pessimistic estimate is the mean plus (3 * stdev). Pessimistic = 65 + (3 * 10.6) = 97 minutes. Our combined PERT estimate is 33/65/97 with a mean of 65 minutes and a stdev of 10.6 minutes.