This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
On dense tests with wings that cannot stall, the two anti-braking switches seemed to make the worst close calls deeper.
What was tested
We checked every very deep close call (over 90 % of the way into the safety bubble), then re-flew the same 144 runs with craft that reach a lane end waiting for a clear spot on the start line.
What was found then
All 14 deep close calls with the switches, and all 4 without, happened when a craft was put back at a lane start. With the fix there are none in 144 runs, and no difference between switch settings; the fix itself barely changed losses. For wings that cannot stall, the switches' only remaining cost is their extra 40–90 m of drift off the lane, which is real. The switches are still rejected; the fix stays a test setting.
What we learned later
Under the fix, the adoption rule rejects 10 of 12 comparisons, not 11; the dense 3 m/s wing passes against both baselines. No new default was written.
research/reviews/brief-c46.md
What changed
Nothing.
Corrections retained in the record
Earlier wording: Every deep near-miss disappears — zero on 144 runs, against fourteen before
Record correction: 14 with switches plus 4 without: 18 before
research/reviews/brief-c45.md
Earlier wording: aircraft lost unchanged on every cell, collision rates unchanged
Record correction: on the calm 3 m/s wing with both switches the hold resolved slightly lower losses (−4 per run) and collision rate
research/reviews/brief-c45.md
Numbers and source references
- 0.9
- severity threshold for a very deep near-missresearch/reviews/brief-c44.md
- 14 of 14 / 4 of 4
- deep near-misses that were re-entries, with / without switchesresearch/reviews/brief-c44.md
- 0 of 144
- runs with a deep near-miss under the re-entry fixresearch/reviews/brief-c45.md
- −4 [−8, −1]
- craft lost per run, calm 3 m/s wing with both switches, fix vs noneresearch/reviews/brief-c45.md
- 40–90 m
- extra off-lane drift with the switchesresearch/reviews/brief-c43.md