Software estimation – embracing nuance and controlled chaos
summary
Estimation starts from the admission that everyone is bad at predicting the future and improves by naming uncertainty: use cone of uncertainty ranges not single points, separate effort vs duration, and avoid unwarranted precision. Practical tactics include count and compute, decomposition, historical data, expert judgment and proxy estimates like t-shirt sizing; defend estimates against peer pressure, communicate ranges that expose risk, and treat “done” as a negotiated outcome rather than an implicit promise.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:45What is an estimate?
- 5:11Estimates Include A Range
- 8:49Estimation Error
- 17:52Techniques To Help
- 23:50Remember...
related talks
01h 00mEstimation: The Art of the SWAG
41mSo you think you can estimate? The lost art of useful estimates
47mCreating efficient, accurate, software estimations
11mThe key to solving the conundrum of software estimation
40mEstimations: from guessing to weighing in grams
58mThe State of NoEstimates
03h 00mBeyond Estimates (Estimates and NoEstimates ) - Let's Explore the Possibilities
30mSoftware Estimation Made Easy
