This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
Does traffic density change things? And how large is the loss problem on one case (1-second link, emergency craft at 2× traffic speed)?
What was tested
Still air with 300 and 900 craft (the 600-craft case reused), and ten fresh layouts per wind direction on the 1-second case.
What was found then
Across density the rules held: at 300 craft the others lost half as many as without priority, at 600 and 900 about the same, and the emergency craft was never lost and stayed on time. On the 1-second case, in the first wind direction, it was lost in 4 of 10 runs against 0 of 10 without priority (9 against 1 over 20 runs). The uncertainty was wide (−26 to +77 percentage points). This sized a cost of the owner's choice. Best candidate cause: the announced corridor being one second (32 m) out of date.
What we learned later
The out-of-date corridor candidate was withdrawn, since the code already draws the corridor from where others hear the craft. Most of these losses were collisions near the ground after the craft dived.
research/design/prio_1s_census.md
With the craft held at its planned height, the same layouts read 1 loss against 1; at c207 this was judged not to show that the losses came purely from diving into the ground.
research/design/prio_level_hold_record.md
At the stage close, the size of the 1-second loss excess in wind was still listed as unknown.
research/design/prio_stage4_close.md (Open)
What changed
nothing
Numbers and source references
- 30 / 40
- runsresearch/design/prio_density_record.md; research/design/prio_stripe_record.md
- ×0.50 / ×0.93 / ×0.97
- others' losses, package vs without priority at 300 / 600 / 900 craftresearch/design/prio_density_record.md
- 8 / 80 / 238
- losses with no emergency craft at 300 / 600 / 900 craftresearch/design/prio_density_record.md
- 4 of 10 vs 0 of 10
- emergency craft lost, 1 s link 2×, first wind directionresearch/design/prio_stripe_record.md
- 9 vs 1 in 20
- same cell over three seed setsresearch/design/prio_stripe_record.md
- −26 to +77 points
- conservative 95 % interval on the loss differenceresearch/design/prio_stripe_record.md
- 32 m
- staleness of a 1 s old tube (candidate at the time)research/design/prio_stripe_record.md