This is an earlier experiment. Read the amendments below before treating its findings as current.

The question

Half the wing's slow time remained. Suspects: vertical speed-matching (craft copy neighbours' velocity) and the rule that turns craft onto their lane heading.

What was tested

Six layouts, calm and double traffic: vertical matching off; a switch stopping the heading rule braking a wing; both switches; then the 3 m/s wing.

What was found then

Vertical matching is not the problem, and it protects crowded wings (off: about 15 % more lost). Both switches together cut the 5 m/s wing's time at minimum speed to 3.5 % in calm, as low as with matching off, and the 3 m/s wing's from 4 % to 0.3 %. So the slowing comes from what the rules command. The price: the stalling wing loses 12–28 % fewer craft than with no switch, but survivors spend about 20 % more time in close calls than with the first switch alone.

What we learned later

LATER / CYCLE 37

For the 3 m/s wing, the heading switch alone gives only part of the cut; it needs both.

research/reviews/brief-c38.md

LATER / CYCLE 42

On wings that cannot stall by design, neither switch reduced craft lost; not adopted.

research/reviews/brief-c43.md

What changed

Nothing.

Corrections retained in the record

  • Earlier wording: about half as many aircraft lost to stalls (3 m/s)

    Record correction: losses fell 48 % calm but 27 % at double traffic

    research/reviews/brief-c36.md

  • Earlier wording: the crowded grid’s collision rate came out lower than with the first gate alone

    Record correction: 13.99 [12.73–15.35] vs 15.21 [13.81–16.71]: lower point estimate, intervals overlap

    research/reviews/brief-c35.md

Numbers and source references
6
seeds per configurationresearch/reviews/brief-c35.md
862 → 990
wing craft lost at double traffic, vertical matching off (+15 %)research/reviews/brief-c34.md
3.5 % / 4.7 %
5 m/s wing floor time, both switches, calm / doubleresearch/reviews/brief-c35.md
3.6 % / 7.1 %
floor time with the matching rule off, calm / doubleresearch/reviews/brief-c35.md
4.0 → 0.3 %
3 m/s wing floor time, calm, both switchesresearch/reviews/brief-c36.md
12–28 %
fewer stalling-wing craft lost vs no switch (2 982 → 2 149; 5 174 → 4 538)research/reviews/brief-c35.md
+20 %
per-survivor time too close, first half, both switches vs first switch aloneresearch/reviews/brief-c35.md
Original structured record ↗