Creating efficient, accurate, software estimations
summary
Panel argues that reliable estimates come from specifics not slogans: teams use Monte Carlo simulation and historical throughput to give probabilistic forecasts like 80% probability dates, replace time-per-item with counting work items, and build explicit buffers for known interruptions (example: a two‑week manufacturing shutdown for Chinese New Year). Practical takeaways include giving ranges not single dates, distinguishing estimates from deadlines, negotiating scope to meet fixed dates, and accepting that a promised two days vs 14 days lesson is normal and fixable by better process.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 3:45Why you need buffers
- 5:23Is it worth it
- 6:30Deadlines vs estimates
- 8:38Deadlines and deadlines
- 11:12Scope and priorities
- 12:47Time and priorities
- 14:11Planning fallacy
- 17:23Who is this for
- 19:03unknowns
- 23:04buffers
- 26:42anything to add
- 27:50building trust
- 30:15deadline vs estimation
- 32:07estimation when it is not clear
- 33:25estimating when required to use a technology you are not familiar with
- 35:35no estimates
- 38:43finegrained vs detailed estimates
- 40:48Delivery of the future
- 41:58Predictability
related talks
30mSoftware Estimation Made Easy
29mSoftware estimation – embracing nuance and controlled chaos
45mHow to not shoot yourself in the foot with estimation
48mNoEstimates
41mSo you think you can estimate? The lost art of useful estimates
11mThe key to solving the conundrum of software estimation
07mEstimates are BS
46mEstimates and complexity: a tale in 42 and a half minutes