This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
The work had no finish line, and the human asked how long the stage would run. The map was condensed into one table for the reviewers.
What was tested
Writing only. The table gave, per factor, where safety breaks, why, how sure we are and what was withdrawn. It was checked four times; the reviewers were asked three times (cycles 115–117) whether to close.
What was found then
The first draft revived withdrawn claims, such as losses rising 'in a straight line' with lag and heading copying costing the ideal craft 'no craft', and it mixed units. 12 corrections over four checks were logged in the table. Each time one reviewer voted to close and the other to continue. After fixes both ranked the same experiment next: speed had only been flown on the ideal craft (instant response), before the look-ahead existed, so the wing's speed test came next.
What changed
Nothing.
Numbers and source references
- 12
- corrections to the summary table over four auditsresearch/envelope.md:245
- three
- close-out requests, cycles 115–117research/envelope.md:65