Monday, August 08, 2005
Relationship between Effort and Schedule
We are often asked for estimates, even more often for "quick estimates". Managing estimates is not trivial, and it starts by realising that you have to manage estimates :-), that is, that an estimate is not a static promise. It doesn't really matter if you used a formal approach like function points and COCOMO or you just concocted some numbers based on your experience: you may need to review your estimates later, when you obtain new knowledge, requirements change, etc. This is somehow recognized by agile methods, that in many cases simply avoid the project estimation part, and concentrate on planning only the next iteration / user story.
Anyway, our estimates are often given in man/hour or man/weeks, that is, as an effort. We translate this effort in a schedule by assigning resources to the project. This is reasonable but somehow simplistic: it assumes that the effort is fixed, and the schedule is adaptable. We know that's not true: we can't compress schedule indefinitely by just adding resources. The effort is not constant: it is a function of the schedule as well!
Phillip G. Armour elaborates on the nonlinear relationship between effort and schedule in a very nice (and short :-) article in Communications of ACM, June 2004: "Real Work, Necessary Friction, Optional Chaos". Worth reading :-).
Anyway, our estimates are often given in man/hour or man/weeks, that is, as an effort. We translate this effort in a schedule by assigning resources to the project. This is reasonable but somehow simplistic: it assumes that the effort is fixed, and the schedule is adaptable. We know that's not true: we can't compress schedule indefinitely by just adding resources. The effort is not constant: it is a function of the schedule as well!
Phillip G. Armour elaborates on the nonlinear relationship between effort and schedule in a very nice (and short :-) article in Communications of ACM, June 2004: "Real Work, Necessary Friction, Optional Chaos". Worth reading :-).
