CategoryevidenceManualCEvidencedocumented
ProblemAn enormous corpus of uneven quality, much of it design walkthroughs presented as results.
TechniqueAsk, in order: does it name a limit it hit, is there a number that could be wrong, does it show runtime output or a design, and does the author benefit if you believe it. Separate the technique from the outcome: the method is often accurate because it is derived from the documentation, while the result is not evidence.
Why it worksHonest hands-on accounts always name a limit; a survey of nineteen autonomous-company write-ups found none carrying a verifiable receipt and only one containing real end-to-end execution with a measured result.
When to useEvery secondary source.