Autonomous agents cannot escape Brooks's Law

Adding more engineers to a delayed project makes it later, and dropping autonomous agents into the same codebase creates the exact same trap. Writing about Frederick Brooks's 1975 book The Mythical Man-Month, j11g notes that the coordination problem has not vanished with AI. Without tight task boundaries, multiple agents assigned to a single repository will happily overwrite each other's code in a high-speed loop.

Brooks's central point was that partitioning software requires clean interfaces, and sequential dependencies cannot be compressed simply by adding workers. Generating syntax is now cheap, but verifying it remains strictly sequential. Producing pull requests from several agents against loose requirements creates conflicting assumptions that a human engineer still has to untangle.

The operational bottleneck has shifted from code generation to system architecture and verification. If a team lacks deterministic test suites and strict task boundaries, deploying fleets of agents may only accelerate the accumulation of technical debt. Dense code generation saves time only when the surrounding engineering is already airtight.