lauren shares the key points of yminsky's talk on improving codebases for Agents

Oct 26

1. 16:49

   Shao Meng

   Jane Street on How to Make a Codebase Better Suited for Agents

   In a Bug Bash talk, Jane Street's Yaron Minsky argued that after Agents make writing code cheaper, types, tests, and code review become more important, with the goal of amplifying human leverage and understanding. He advocates using types to express global invariants and using expect tests to embed behavioral output into source code to form a fast feedback loop, and believes Agents are good at turning red lights green but not at maintaining non-functional properties such as simplicity and scalability, which makes review even more critical now. He also mentioned that the cost structure of formal methods is changing, that he began shifting toward them about eight weeks ago, and that he plans to form a formal methods team.

   Expand summary
2. 15:54

   Lauren Tan

   yminsky on how to adapt codebases for agents

   lauren shared the key points of yminsky's talk on how to improve codebases for AI agents: types are one of the best sources of feedback, and using types to make illegal states unrepresentable can enforce global rules across the entire codebase; in addition, there are property-based testing, fuzzing, full-system deterministic simulation testing, and expect tests. The talk mentioned that formal methods used to be extremely costly: seL4's roughly 8,700 lines of C code took about 25 person-years, with 23 lines of proof per line of C code, but agents have changed the cost-benefit ratio.

   Expand summary