It's a Noisy World Out There
summary
Research shows random variability — noise — often causes more error than bias: simple additive algorithms (a+b+c) outperform experts in Paul Meehl–style studies from 1954 and a 96‑case Oregon Research Institute cancer X‑ray study, fingerprints and estimates show huge inconsistency (estimates differing ~71%), and Louisiana judges give harsher sentences after LSU losses. Practical fixes include a noise audit, premortem (Gary Klein), structured interviews, and using an outside view baseline; reducing noise cuts overall error and also reduces bias.
AI-written summary · about
chapters (16)
- 0:00Intro
- 2:26How do we make decisions?
- 8:59Paul Meehl
- 13:16Oregon Research Institute
- 21:39What about algorithms?
- 23:43Noise
- 27:07Ask the right question
- 28:48Bias vs noise
- 35:21What to do?
- 40:35Hiring & interviews
- 44:55Estimations are bad
- 47:58Inside vs outside views
- 49:17Reducing noise reduces bias
- 51:01Don't eliminate intuition
- 51:30Summary
- 51:48Outro
related talks
31mUnderstanding Cognitive Bias Found In Judgement & Choice
11mHow to deal with your demons
31mUnderstanding Cognitive Bias Found In Judgement & Choice
01h 02mCognitive biases in software development
53mHow Not to Be Wrong in Software Engineering
40mOptimistic Security
01h 00mCognitive biases in software development
30mBehavioral Software Engineering