AI coding reignites testing debate: bigger test units are fundamentals, not dead

mattpocockuk · x · 2026-09-23

Reacting to the "unit tests are dead" take, Matt Pocock argues the real issue is defining a unit as the smallest possible piece. Testing larger units has always been good practice — it makes tests less sensitive to structural changes. Nate Berkopec prefers heavy e2e/black-box/golden-master testing when AI writes the tests, dropping to lower levels only when needed.

Original post →

More from coding & agent

coding & agent channel →