Sound reasoning weighs both the quantity and the quality of connected evidence before accepting a claim, not just its plausibility.
accept if confidence = f(evidence_quality, count) > skepticism_threshold
- Claims in network — assertions being evaluated against connected evidence nodes.
- Evidence connectivity — how many pieces of evidence back (or contradict) each claim.
- Evidence quality — reliability of the evidence sources feeding into each claim.
- Skepticism threshold — confidence bar a claim must clear before being accepted rather than flagged.
This is the exact reasoning structure behind fact-checking workflows — a claim is only accepted once enough high-quality evidence clears a stated bar.