This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
Is the wing's speed penalty specific to the calm crossing, or does it also appear with gusts and a slow, lossy radio link? Is it the wing's own?
What was tested
The wing flew the 56-scenario storm test (storms plus a poor radio link) at three speeds, three layouts each; the eVTOL and slow-responding test craft flew it on one layout.
What was found then
The wing's collisions per craft-hour rose 2.5× from slowest to fastest, far below the calm crossing's 11×, though even at 8 m/s the storm rate was about seven times the calm rate. The eVTOL rose 11× from a very low base, the test craft 3.7×. Scenario types split: in a dense, disturbed grid of craft collisions fell with speed (eVTOL counts too small to read), in crowded airspace they rose. Directions matched across airframes; sizes did not. Separately, a counting bug in the project's flight log was fixed: 37 194 runs, not 6 537; 3.00 million craft-hours, about twice the old figure.
What we learned later
Reviewers ruled that matching directions don't show whether the pattern belongs to the scenario types or to the airframes; that is not settled.
research/reviews/auditor-c166.md (F-PRESETS-SCENARIO-ATTRIBUTION); research/design/tau_closebub_spec_fixes_c166.md
What changed
nothing in the simulator; the project's flight log (running totals of all runs) was corrected
Corrections retained in the record
Earlier wording: in storm and link conditions the wing already loses seven times the calm-crossing rate at 8 m/s, and that part — the link and the gusts — does not grow with cruise
Record correction: The next cycle's reviewers disputed this as a cause the rows do not identify; the lower storm-set ratio is recorded as a description only.
research/reviews/auditor-c164.md (F-TAU-SLOPE-OVERCLAIM); research/data/oos/tau_speed_suite.md (Registered)
Earlier wording: Every total quoted so far is roughly half the true figure
Record correction: Craft-hours were about half (1.48 vs 3.00 million), but runs were undercounted about 5.7-fold (6 537 files vs 37 194 runs).
research/reviews/researcher-c165.md item 1
Numbers and source references
- 56
- scenarios in the storm-and-bad-link setresearch/data/oos/tau_speed_suite.md
- nine; 504
- suite runs (3 speeds × 3 seeds); scenario records (9 × 56)research/reviews/researcher-c164.md item 1
- 2.5× (0.045 to 0.115)
- wing collisions per craft-hour, 22 vs 8 m/s, storm set (2.54×; 0.0452 → 0.1148)research/data/oos/tau_speed_suite.md
- 11×
- the same ratio on the calm crossing (11.3×)research/data/oos/tau_speed_suite.md
- about seven times
- wing storm rate vs calm-crossing rate at 8 m/s (0.0452 vs 0.0067)research/design/tau_stage_close.md §8.3
- 11× / 3.7×
- eVTOL / delayed-sideways collisions per craft-hour, 22 vs 8 m/s (11.22 / 3.72)research/data/oos/tau_speed_presets.md
- 37 194 (not 6 537)
- corrected run countresearch/reviews/researcher-c165.md item 1
- 3.00 million (vs 1.48 million)
- corrected craft-hoursresearch/reviews/researcher-c165.md item 1