Feedback Loops in Sociotechnical Systems
summary
Claim: large-scale A/B work shows roughly one third of ideas have positive impact, one third neutral and one third harmful, so prioritize fast learning over clever engineering. Use OODA-style loops: observe with APMs, metrics and traces, orient with system design reviews and goal walls, decide with cheap validation like a marketing funnel, Wizard of Oz demos or CSV-backed prototypes, then act with quick CI/CD — aim for a 15 minute CI/CD pipeline — plus feature flags and targeted rollout to close feedback loops.
AI-written summary · about
chapters (8)
- 0:00Introduction by host
- 0:25Speaker introduction
- 1:09Defining feedback loops
- 3:20Building the right things
- 14:28Doing the right things
- 26:59Building things right
- 31:38Key takeaways and summary
- 32:03Q&A session
related talks
24mExpress Feedback Loops
32mThe Flywheel Effect of Feedback
17mDo you have a minute? A better approach to feedback culture
41mA Defence of Technical Excellence
55mKeynote: Experimenting in a Sociotechnical System in Software Engineering
40mRun Circles, Not Relays
51mA Defence of Technical Excellence
51mImproving Software Flow