This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
Do the two switches that stop rules braking the wing make it overshoot crossings? And would one switch alone do for the 3 m/s wing?
What was tested
We mapped collision sites in 72 paired runs, tried the heading switch alone on the 3 m/s wing, and recorded where each run's deepest close call happens.
What was found then
The test set in advance for collisions bunching at crossings did not trigger, so overshooting is not ruled out; in one dense case the share near crossings rose slightly. For the 3 m/s wing the heading switch alone cuts slow time only partly (4.0 to 2.3 %), the other switch barely, and both together to 0.3 %: it needs the pair. Surviving stalling wings with both switches spend 27–35 % more time in close calls. A sign that the deepest close calls lean toward crossings was inconclusive.
What we learned later
The tool that located the deepest close call was faulty (wrong craft in 29 of 48 runs); the crossing figure was taken back.
research/reviews/brief-c39.md
What changed
Nothing.
Numbers and source references
- 72
- matched runs in the collision-site checkresearch/reviews/brief-c37.md
- 48
- runs with the deepest-near-miss recordresearch/reviews/brief-c38.md
- 32.6 → 37.3 / 35.6 %
- collisions near crossings, dense stalling wing, no / one / both switchesresearch/reviews/brief-c37.md
- 4.0 → 2.3 %
- 3 m/s wing floor time, heading switch alone, calmresearch/reviews/brief-c38.md
- 0.3 %
- same with both switchesresearch/reviews/brief-c38.md
- +26.8 % / +35.1 %
- per-survivor time too close, 3 m/s stalling wing, calm / doubleresearch/reviews/brief-c38.md
- 25 % vs 15 %
- deepest near-misses within 100 m of a crossing vs time spent thereresearch/reviews/brief-c38.md