AI, Programming, Entropy

Entropy, or the tendency for things to collapse into disorder.

Software development is only sustainable if it fights against entropy. See: the Pragmatic Programmer's section on broken windows. This is a lost idea in the age of AI coding tools.

A common fallacy by leadership: more is better. What defines better in its true form? Processes and orientation towards the fight against entropy.

Humans reach towards all things that reduce energy consumption. What is easy will be taken without the presence of a counterweight. The lack of energy spent towards resisting entropy only leads to disorder.

AI coding tools are the easy tools which reduce energy consumption. The reliance on AI coding tools without a counterweight: decayed codebases, failing systems, upset stakeholders. This is the physics of human nature and systemic pressure.

"More is better." How can we produce more? Through AI, resulting in higher velocity. Points, or the measurement of effort, were never designed to represent chaos. This is the failure of reporting metrics.

What gates more? Relaxed processes and decay of the programmer's orientation. In this paradigm, both code review and craft are structurally necessitated to suffer. This is the mathematics of 'more'.

A core contradiction in modern software development: we extend job offers to those who fight against entropy, but we reward 'more' in practice. We screen for excellence but reward mediocrity.

An additional issue on the use of AI tools: the lack of construction and therefore the lack of expertise on what was built. How can you efficiently maintain when even the founding engineer doesn't have a coherent mental model?

"More is better" in the age of AI software development is incapable of fighting entropy. All that follow this path are doomed.