This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
Speed-matching caused the delay, so it was switched off, with own dodging kept at a fifth. That cost some safety, so the craft then announced its next seconds of flight as a moving reserved corridor that other craft leave.
What was tested
Runs at 2× with speed-matching off or at 5 %, then three corridor shapes.
What was found then
With speed-matching off the craft was on time, but ordinary losses rose from 82 to 91 and the emergency craft was lost once in five runs. A 5 % share was no cheap middle ground, so zero was chosen. The 10-second wide corridor kept it on time, with 78 ordinary losses and no emergency craft lost, a gain within noise. The narrow 10-second corridor lost the most (101), read then as width, not warning time, buying safety. On the layered network ordinary losses did not differ.
What we learned later
The 'no cheap middle setting' reading was narrowed: the three settings behaved roughly in a straight line, shares between 0 and 5 % were not tried, and 5 % met the timing target on the layered network.
research/design/prio_align0_record.md (P11)
'Width, not warning time, buys safety' was softened to a guess: three versions on one test case cannot separate a width effect from scatter.
research/design/prio_corridor_record.md
What changed
Switches for the speed-matching share and the reserved corridor were added, off by default. The 'calm set' of emergency rules (speed-matching off, own dodging at a fifth, 10-second wide corridor) was carried forward, not adopted.
Corrections retained in the record
Earlier wording: 20 more half-hour runs ... With its flocking off and its dodge kept at a fifth
Record correction: Of the 20 runs, 10 had flocking off and 10 had a 5 % flocking share.
research/design/prio_align0_record.md (Arms)
Numbers and source references
- 20 / 30
- runs in the two cyclesresearch/design/prio_align0_record.md; research/design/prio_corridor_record.md
- 2×
- emergency craft speedresearch/design/prio_align0_record.md
- five
- seeds per versionresearch/design/prio_align0_record.md
- a fifth
- own dodge strength keptresearch/design/prio_align0_record.md
- 5 %
- partial flocking share testedresearch/design/prio_align0_record.md
- 1.040 / 1.013
- lap ratio with flocking off, grid / latticeresearch/design/prio_align0_record.md (P9)
- 91 vs 82
- ordinary craft lost, flocking off vs flocking at a fifthresearch/design/prio_align0_record.md (P10)
- 1 of 5
- runs in which the emergency craft was lost with flocking offresearch/design/prio_align0_record.md
- 5 s / 10 s
- tube look-ahead timesresearch/design/prio_corridor_record.md
- 1.047
- grid lap ratio with the 10 s wide tuberesearch/design/prio_corridor_record.md (P13)
- 78
- ordinary craft lost with the 10 s wide tuberesearch/design/prio_corridor_record.md
- 0
- emergency craft lost with the 10 s wide tuberesearch/design/prio_corridor_record.md (P13)
- 101
- ordinary craft lost with the 10 s narrow tuberesearch/design/prio_corridor_record.md