This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
The stage closed with the emergency rules, their limits, and a choice for the owner: make the link-set rules what 'priority on' means, or leave everything off.
What was tested
No new runs. After the owner chose the rules, the five settings behind 'priority on' were set to them. Runs without an emergency craft were checked to be identical to before, and one map run was flown again and matched in all 81 output fields.
What was found then
By its own rule the stage did not earn adoption. Only 5 of 8 test cases qualified, and the link-set rules were never flown on those eight cases. At 2× under a link the emergency craft was lost more often (7 against 3 in 25 runs). The close recorded a partial description, not a proven safe range. The owner overrode the rule and chose the emergency rules.
What we learned later
The claim that the rules cost the others nothing on every test case rested partly on no-wind runs; in a 5 m/s wind with no link they were worse at 1.5× and 3×.
research/design/prio_stage_close.md §0 [c199]
The absolute grid results describe a fixed-wing craft that dives far below its planned height. Holding it level did not show the 2× loss excess to be merely an effect of diving into the ground (c207).
research/design/prio_stage_close.md §0b
The owner replaced this default with the simple emergency craft, which has right of way only.
research/design/prio_default_simple_record.md
What changed
From c198, 'priority on' meant the link-set rules: speed-matching off, full-strength own dodging, and a 10-second reserved corridor with a radius of 2.5 standard spacing distances. Replaced by the simple emergency craft (right of way only) at c221.
Numbers and source references
- five
- settings changed behind the priority switchresearch/design/prio_stage_close.md §5
- 5 of 8
- family cells eligible (rule needed 6)research/design/prio_stage_close.md §3
- 81 of 81
- fields reproduced by the re-flown map runresearch/design/prio_stage_close.md §5
- 7 of 25 vs 3 of 25
- 2× link aggregate, package vs without priorityresearch/design/prio_stage_close.md §5
- 640
- new runs in stage 3research/design/prio_stage_close.md