Test Desiderata
A third take on Kent Beck’s Test Desiderata (2019) and Emily Bache’s 2.0 reorganization (2025). This is a position, not a neutral catalog: it keeps what the field independently converged on, renames where a rename is an improvement, and, where a property leans on a contested mechanism, names the neutral goal it serves. The reasoning lives in the position essay.
Eleven properties: Fast · Run in Any Order · Readable · Deterministic · Diagnosable · Behavioral · Structure-Insensitive · Automated Checks · Writable · Minimal Data · Composable.
The eleven are not independent: some reinforce each other, some pull against each other, and no context lets you maximize them all at once. That structure is its own note, Desiderata in Tension.
One macro goal: Predict Success in Production. “Inspiring” is that goal’s outcome, not a property; there is no Inspiring KPI.
The Automated Checks rename concedes a context-driven distinction, Testing vs. Checking: what a suite automates is checking; deciding which checks are worth running is testing, and that doesn’t automate.
This list covers one dimension of testing: properties. Categories (unit/integration/E2E), mechanics (seams, doubles), suite shapes (pyramid, trophy), and team processes are separate axes the field names elsewhere.
Sources
- Kent Beck, Test Desiderata (2019).
- Emily Bache, Test Desiderata 2.0 (2025).