This is an earlier experiment. Read the amendments below before treating its findings as current.

The question

The other two slow airframes needed the same check. Do they do as well with full extra room and early reaction?

What was tested

All four combinations on the crossing (three layouts; five for the wing at 16 m/s) and the storm and standard mixed test sets on five sets of layouts. A 90 % crash-cut screen the researcher had added was reviewed and dropped before the wing's test-set results were read.

What was found then

Slow-responding test craft at 16 m/s: 19 crashes with full room against 18 under the new default (old controller 82); in storms, 622 against 661. At 22 m/s it looked worse (73 against 45). Wing at 16 m/s: full room kept 86 % of the crash cut (40 against 26 crashes), below the screen, and was worse at 8 m/s too. But the wing's test sets clearly favoured full room: 994 against 1 190 storm crashes, 438 against 592 mixed. The screen, lacking a physical basis, was retired; all four airframes qualified.

What we learned later

LATER / CYCLE 178

The slow-responding test craft's cost at 22 m/s did not repeat on two fresh layouts (40 against 53 crashes).

research/data/oos/fastrow_lagged_lateral_22.md

LATER / CYCLE 179

The wing's cost at 16 m/s did not repeat on two fresh layouts (13 against 15 crashes).

research/data/oos/fastrow_wing_16.md

LATER / CYCLE 185

The wing switched to the full extra room on the researcher's decision.

research/design/tau_wing_lagk_record.md

What changed

nothing

Corrections retained in the record

  • Earlier wording: So it is the airframe: three separate looks give 84, 89 and 86 %

    Record correction: There were two batches and their pool, not three independent looks, and they do not establish an airframe mechanism.

    research/reviews/auditor-c174.md (F-WING-AIRFRAME-OVERCLAIM)

  • Earlier wording: the 90 % line itself came from a multirotor reading of 90.3 %

    Record correction: The 90 % line was registered before any candidate row existed; it was not fitted to the multirotor's 0.903.

    research/design/tau_entry_rule.md

  • Earlier wording: the wing is the only one that keeps the smaller bubble

    Record correction: Superseded within the step: once the screen was retired, the wing qualified at cycle 177.

    research/design/tau_entry_rule.md; research/data/oos/wing_c5_gate.md

Numbers and source references
three; five
crossing seeds; wing 16 m/s extension and benchmark offsetsresearch/data/oos/lagged_lateral_2x2.md; research/data/oos/wing_16_ext.md
90 %
retained crash-cut screen, retired as an eligibility ruleresearch/design/tau_entry_rule.md
19 / 18 / 82
delayed-sideways crossing crashes at 16 m/s: full bubble / adopted / oldresearch/data/oos/lagged_lateral_2x2.md
622 / 661 / 896
delayed-sideways storm-set crashes: full bubble / adopted / old, five seedsresearch/data/oos/lagged_lateral_c5_gate.md
73 vs 45
delayed-sideways crossing crashes at 22 m/s, full bubble vs adopted, three seedsresearch/data/oos/lagged_lateral_2x2.md
86 %
wing: share of adopted crash cut kept at 16 m/s, five seedsresearch/data/oos/wing_16_ext.md
40 / 26 / 126
wing crossing crashes at 16 m/s: full bubble / adopted / old, five seedsresearch/data/oos/wing_16_ext.md
994 vs 1 190
wing storm-set crashes, full bubble vs adopted, five seedsresearch/data/oos/wing_c5_gate.md
438 vs 592
wing matrix crashes, full bubble vs adopted, five seedsresearch/data/oos/wing_c5_gate.md
Original structured record ↗