We review 76 practitioner sources to understand cognitive debt in AI-assisted development: how gaps in developers’ understanding arise, the consequences for teams, and the practices proposed to prevent, detect, and mitigate them.
We investigate the factors that affect developer experience and the strategies individuals and teams use to improve it. The resulting framework helps organisations understand what makes a productive and effective work environment for their developers.
In a large-scale study, we investigate how code reviews are done at Microsoft, which benefits they provide, and which challenges developers face. We also identify best practices for reviewers, authors, and organisations, along with needs for better tools.
Code reviews support knowledge sharing and code quality, but require a substantial time commitment. We examine review practices and how teams can get more value from the time they spend reviewing.
How can we reduce test suite size and execution time without sacrificing code quality? We describe an approach to safely removing test cases and reducing the cost of testing at Microsoft.
Through a large-scale study, we investigate which code review comments are useful and which create additional work for the author. The findings help reviewers improve the value of their feedback.
We describe what we learned from building and deploying a code review analytics platform at Microsoft, including the challenges of creating a platform that developers adopt in their daily work.