Evidence, uncertainty, and corrections

Evidence hierarchy

• Original platform terms, official documentation, and first-party datasets take priority for factual platform rules.
• Clearly visible photo evidence is separated from inferred identity, condition, and hidden-content hypotheses.
• Confirmed sold evidence and active asking prices are different evidence types and must be labeled accordingly.
• Worked examples are identified as illustrative and must not be presented as real auction outcomes.

AI-assisted analysis

AI helps identify visible objects, group repeated views, summarize condition clues, and produce structured estimates. It can hallucinate, misidentify an item, draw an inaccurate bounding box, or apply the wrong comparable. Users should treat every result as a reviewable draft rather than an authoritative appraisal.

Public explanatory content may be drafted or checked with AI, but publication responsibility remains with the UnitSift product team. AI is not listed as an author or source.

Commercial and platform independence

Paid UnitSift plans fund the product. Pricing does not change the valuation logic or permit stronger claims. Platform pages are independent guides; inclusion does not imply sponsorship, affiliation, endorsement, or privileged access.

Corrections and updates

When a material factual or methodological error is confirmed, the affected page should be corrected and its visible update date changed. Cosmetic edits do not require a new date. Platform rules should be rechecked against their primary source because fees, import behavior, and terms can change.

Corrections should preserve the distinction between a product limitation, a changed external rule, and an error in UnitSift content or analysis. The corrected claim and its evidence take priority over defending an earlier version.