Friday, October 14, 2005

Why are people such bad estimators?

Why are people such bad estimators?: "Painters say 5 days, it takes 8. Designers say 2 weeks, it takes 4. Programmers say 3 months, it takes 5. Government says it will take $X and it takes $XX. It’s been a very rare occasion (in my life, at least) that estimates have been anywhere near accurate. Even people in business for 20 years seem to have a bad time making accurate estimates. I realize an estimate is just a guess, but there’s a lot of bad guessing going around. What is it about people that make us such bad estimators?"

I think this idea has been around a long time- here's some top reasons why we're bad:
1) Humans are too optimistic.
2) While it may be possible to estimate how long it would take to do something if everything things went right, it's often difficult to estimate with error, interruptions, etc. that you don't have good data on.
3) What type of date is being given? What is a 5 day estimate? Is it the earliest date something could be possibly finished? Is the 50-50 date? The 90% date? You are talking about something that is stochastic, and would occur in a distribution.
4) The future does not always work in the way that the past did

And what to do about it:
1) Stop estimating
2) Adjust estimates based on progress (running features in a software project)
3) Present estimates as a probability distribution.