- Published on
Conceptual integrity and counting lines of code
- Authors
- Name
- DeepHorus
On August 19, 2026, developer Simon Willison joined Claire Giordano on the Talking Postgres podcast to discuss AI's impact on software development. He argued that lines of code can be a valid productivity metric when using coding agents, contrary to popular opinion. The episode aired against a backdrop of rapid AI adoption in software engineering.
Traditionally, a productive engineer wrote about 200 lines of production-ready code per day, with many days yielding just 50 or 60. Willison called 200 lines an "incredibly good day." With AI agents, that figure can rise to a thousand lines, provided the code remains maintainable and tested. However, achieving such output requires deep skill and experience.
As a result, one engineer can now do far more work, which raises the question of why teams are needed. Willison explained that the limiting factor becomes cognitive capacity, not coding speed. No one can mentally manage 100 times more code, so distributing that load across a team remains essential.
The conversation also addressed conceptual integrity from The Mythical Man-Month. Willison said AI agents make it easy to add features quickly, causing software to grow in inconsistent ways. Claire equated this to the Winchester Mystery House, a 140-room mansion built continuously by Sarah Winchester due to a psychic's warning. The result is software that loses its internal coherence.
Willison pointed to discipline as the key issue. Previously, the time cost of building a feature—often a week—forced engineers to evaluate ideas carefully. With agents, a feature can be added in an hour, making impulsive additions much easier.
This shift suggests that while AI dramatically increases output, it also demands stronger oversight and curation. Senior engineers may move from writing code to guiding its evolution. As coding agents become standard, the industry must update its metrics and design principles to preserve software quality.
The podcast highlights a central tension in modern development: raw productivity vs. architectural coherence. Without deliberate safeguards, cheap feature creation could lead to digital monstrosities. Willison's insights offer a framework for balancing these forces.
