Continuous Experimentation
summary
Argues that lead time falls from months to days or even hours, citing Comcast moving from 12–18 month releases to three releases per week. Recommends serverless stacks — AWS Lambda plus managed services like Auth0 and Firebase — to spin up low‑cost, ephemeral experiments, run A/Bs at scale, and roll back quickly. Describes organizational shifts: DevOps culture, blameless post‑mortems, cost visibility via short‑lived sub‑accounts, and spreading product ownership so experiments are cheap, frequent, and informative.
AI-written summary · about
chapters (11)
- 0:00Intro
- 3:00What is the point of software engineering?
- 4:36From Feature Factories to Product Labs
- 13:38Minimize initial deployment time
- 14:36Safe to fail 8. Budgetary safety
- 20:502-Reduce infrastructure cost Costs based on actual usage, not any commitment Costs scale up from, and down to, zero, automatically .Removes cost-saving incentive to co-locate features
- 23:30If you're only using Serverless for cost savings you're missing a big trick
- 24:05IV - Continuous delivery, continuous value
- 30:32Product ownership is a team's responsibility, not an individual's
- 37:21Data and Security Safety • Make it part of a team's regular work ("shift left")
- 39:40How will you transform your organization from a feature factory to a product laboratory?
related talks
59mThe continuous culture
28mUnleashing Innovation by Unblocking your Developers
48mMoving Fast at Scale
29mThe Experimental Enterprise
41mTechnology mix: Around the serverless in 180 days
25mInnovation and Iteration — Understanding how we improve
45m"Moving Fast at Scale"
44mApplying principles of chaos engineering to serverless