이 에피소드에 관해
Dr Michaela Greiler has spent over a decade researching code review, first at Microsoft Research and the Visual Studio team, then as Head of Research at DX, and now through her own research and consultancy work. In this episode she talks to Rosie Sherry about what agentic development is doing to one of software's most central practices.
Michaela names two failure modes she's seeing in teams: code review exploitation, where verification work gets shifted onto a peer reviewer who never had the mental model in the first place, and code review surrender, where reviewers simply can't keep up with the volume and things get waved through. She walks through SCOPE, the model she's developing, which stands for Staged Code Oversight with Proportional Escalation, and explains why she thinks code review needs to start in the planning phase rather than at the pull request.
Michaela makes the case that reviewing test code, long the lowest priority on most teams, now matters more than ever, because agent-generated tests can signal safety where there is none.
