spike

To validate a proof-of-concept, you run a spike: a throwaway feasibility experiment that tests a specific idea with minimal research and a targeted prototype, returning a structured verdict before you build production code.

A spike in engineering planning is a quick, isolated experiment used to decompose uncertainty into testable Given/When/Then questions. Use it to compare approaches or test feasibility before investing time in a full production build.

You compare technical approaches by running parallel spikes: building isolated throwaway prototypes for each option, testing edge cases, and returning a head-to-head winner with a clear recommendation for the real build.

Yes, you can test feasibility without production code by building throwaway prototypes that produce observable evidence like CLI output, minimal HTML, tiny server endpoints, or focused tests to avoid false positive conclusions.

The best way to de-risk a technical idea is to create a spike plan that researches just enough to choose an approach, builds a minimal prototype, and returns a VALIDATED, PARTIAL, or INVALIDATED verdict with recommendations.

Proof-of-concept tests return false positives when they lack observable evidence. Building throwaway prototypes with CLI output or focused tests ensures you avoid false 'it works' conclusions and accurately invalidate unfeasible ideas.