Writing tests and loss

Testing in software development is like calculating loss in machine learning. Writing good tests that are not overfitting to the specific reality but describing the problem domain is under-rated.

LLMs can enable us to re-think Test-Driven Development (TDD). A testing suite that integrates an LLM which helps you to start with a set of well fitting tests and then iteratively refine as you go along would be great. Software development starts with a conversation between the developer and the domain.

Write tests until fear is transformed into boredom

― Kent Beck, Test-Driven Development: By Example