Does the mutation of Brooks's Law-that adding AI agents to a project with unclear requirements simply produces the wrong answer faster-accurately reflect modern development?
Is Brooks's Law Still Relevant in the Age of AI Agents?
In his 1975 classic, The Mythical Man-Month, computer scientist Frederick P. Brooks Jr. introduced "Brooks's Law," which famously states that adding manpower to a late software project makes it even later. This concept is often illustrated by the "Baby Problem," where adding more people to a sequential task cannot speed up the outcome. As Large Language Models (LLMs) and autonomous AI agents enter the software engineering workflow, some wonder if this law is now obsolete due to the lack of human onboarding time and social friction. However, the law may have simply mutated. While agents do not require coffee breaks, the overhead has shifted from managing people to managing agent orchestration, catching hallucinations, and reviewing automated code. The core danger remains: without crystal-clear requirements, adding more agents might not speed up progress, but will instead lead to the wrong answer being generated at much higher speeds.
Options
- Yes, the "Baby Problem" and orchestration overhead still apply.
- No, the removal of human onboarding time breaks the law.
- The challenge has simply shifted from managing people to managing agents.
- It depends entirely on the clarity of the project requirements.