This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
The rules survived a 2-second link. Where, across link delay and emergency-craft speed, do they break?
What was tested
The grid with link delays up to 2 seconds and the emergency craft at 1.5× to 3× traffic speed. Then fresh layouts on the failing case and two neighbours.
What was found then
The rules passed 11 of 12 cases: others lost 0.79 to 1.11 times as many as without priority, the craft was on time, and with no link the rules cost nothing. The one miss (1-second link, 2×) lost the emergency craft 3 of 5 times. On fresh layouts that case and a clean neighbour both read 2 losses against 1. Pooled, at 2× under a link the emergency craft was lost 7 of 25 times against 3 of 25; at 1.5× and 3× no difference. Read as a weak spot at 2× under any link delay.
What we learned later
'Any link delay' was withdrawn: the 7 against 3 adds up several sampled cases, and two of the five layout sets showed no excess. The pooled count had first been mis-added as 7 against 2.
research/design/prio_stage_close.md §2, §6
The no-link cases had been flown without wind, while the link cases had a 5 m/s wind. In that wind with no link, the rules were worse for the others at 1.5× and 3×. 'Free with no link' and '12 of 12' were withdrawn.
research/design/prio_wind_record.md
What changed
nothing
Numbers and source references
- 110 + 30
- new and reused runs on the mapresearch/design/prio_boundary_record.md
- 0.5 / 1 / 2 s
- link delays sampledresearch/design/prio_boundary_record.md
- 11 of 12
- map cells passing the full adoption testresearch/design/prio_boundary_record.md
- 0.79–1.11×
- ordinary losses, package vs without priorityresearch/design/prio_boundary_record.md
- 0.93–1.08×
- emergency craft lap ratio on the mapresearch/design/prio_stage_close.md §2
- 3 of 5
- emergency craft lost on the 1 s / 2× cellresearch/design/prio_boundary_record.md
- 30 / five
- fresh runs / fresh seeds per cellresearch/design/prio_boundary_fresh_record.md
- 2 vs 1
- fresh-seed losses on the 1 s and 2 s cells at 2×research/design/prio_stage_close.md §2b(iv)
- 7 of 25 vs 3 of 25
- emergency craft lost at 2× under a link, package vs without priorityresearch/design/prio_stage_close.md §2b(iv)