This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
The adoption record named two storm-test cases where the new default kept craft closer than the full-room version. Were they real? And what is the stage's answer?
What was tested
Multirotor and eVTOL storm tests (storms plus a poor radio link) were re-flown on two more sets of layouts, then the multirotor on five, and the stage-close record was written.
What was found then
The eVTOL's extra closeness faded into the range counted as the same (1.28×). The multirotor's held (1.57×; 1.49× over five sets). Its collisions were 42–83 % of the old controller's depending on layout (58 % overall), with more time close on every set: a real but variable gain. Stage close: the adopted rule halves or better the losses at every speed on every slow airframe, but losses still grow with speed (wing 3 to 47 per run), so the literal target was missed.
What we learned later
The extra closeness in storms was traced to the smaller extra room; early reaction carries the cut in crashes.
research/data/oos/mr_bubble_only.md
What changed
nothing
Corrections retained in the record
Earlier wording: As first filed (c151), the close record quoted 'inside the new law' speed profiles
Record correction: Those profiles were the full-bubble variant's from c134, not the adopted law's; fixed at c152 within this step.
research/reviews/auditor-c152.md (F-TAU-STAGECLOSE-SUMMARY); research/design/tau_stage_close.md §2
Numbers and source references
- 1.28×
- eVTOL storm-set deep exposure vs full bubble, three offsets pooledresearch/data/oos/tau_sevcell.md
- 1.57×
- multirotor storm-set deep exposure vs full bubble, three offsets pooledresearch/data/oos/tau_sevcell.md
- 1.49×
- the same, five offsets pooledresearch/data/oos/tau_mroffsets.md
- 42–83 %
- multirotor storm-set collisions, new vs old controller, per offsetresearch/data/oos/tau_mroffsets.md
- 58 %
- the same, pooled over five offsets (190 vs 330)research/data/oos/tau_mroffsets.md
- 3 / 10 / 47
- wing lost per run under the new law at 8 / 16 / 22 m/sresearch/design/tau_stage_close.md §2