For most of the history of software development, the bottleneck between having an idea and having a working artifact was implementation. Good ideas waited months or years to be validated because building them required time, skill, and resources that were not always available. Coding agents compress this cycle dramatically. The bottleneck has moved from implementation to judgment — knowing what to build, why it matters, whether what was built is correct, and where the design will fail. This shift has profound implications for what skills are valuable, what kinds of organizations are competitive, and what gets built. Ideas that were previously too small to justify a full team and too complex to build alone are now the most accessible category of work.
(Agent frameworks are rational responses to an irrational constraint)