Cognitive Bias and On-Call
summary
Surveys of ~200 SREs measure concrete biases: anchoring shifts time estimates when a 45s versus 30s prompt shows a sawtooth graph, ~30% fall for a Linda-style specificity trap, and probability questions show the familiar prospect-theory curve (risk-averse for high-probability gains, risk-seeking to avoid losses). Recommended mitigations are practical: consequence-free buttons, limited shifts, incident command to preserve System 2, dual visualizations, and automated draining; overall conclusion is clear — invest in self-healing/autoremediation to remove humans from costly System 1-driven fixes.
AI-written summary · about
chapters (9)
- 0:00The psychology of on-call
- 5:33System 1 and System 2
- 10:45On-call decision making
- 14:32Anchoring effects
- 21:16Availability and bias
- 26:35The Linda problem
- 30:48Prospect theory
- 37:42SRE survey results
- 52:03Mitigation and conclusions
related talks
01h 00mCognitive biases in software development
31mUnderstanding Cognitive Bias Found In Judgement & Choice
49mAgainst On-Call: A Polemic
25mMental Models for SREs
31mUnderstanding Cognitive Bias Found In Judgement & Choice
19mIts Time We Rethink On-Call Culture
01h 02mCognitive biases in software development
34mFearless SREs