Cognitive biases in software development
summary
Researchers catalog over 200 biases and the talk gives concrete examples and fixes: numeric anchoring skews estimates—use planning poker or model-based estimation (see Troy McGinnis / focusedobjective.com) instead; availability and fixation cause bad searches and deletes—use role-onboarding docs and retrospectives; a plane crash (United 173) prompts aviation crew resource management now used for incident response; teams rewrite auth instead of buying it (not invented here) and keep throwing money at failing projects due to the sunk cost fallacy—use kickoff risk decks, cost models and empowered teams to decide.
AI-written summary · about
chapters (23)
- 0:00Introduction
- 1:51Cognitive Bias Codex
- 3:45Too much information
- 5:07Anchoring
- 5:38Anchoring example
- 9:47System designers
- 10:42Availability bias
- 14:40Plane crash story
- 21:46Not enough meaning
- 23:07Not Invented Here
- 26:08Extrinsic Incentives
- 28:31Backfire Effect
- 30:23hindsight bias
- 31:20kickoff deck
- 32:54Black Swan
- 34:49Act fast
- 35:54Overconfidence bias
- 39:44Dunning Kruger effect
- 42:14Overconfidence
- 44:24Appeal to novelty
- 47:52Sun cost fallacy
- 52:48Negative bias
- 58:50Conclusion
