> **Historical evidence extract.** Original study record; its proposed next steps may be superseded. See the [18 September reassessment](/kimonav/plan/) for the current plan. Source: `knav/research/2026-09-11-terminal-family-capacity/EXECUTION_REPORT.md`; SHA-256: `00357af7f42018442a737f5d4de777ed664f72748fd67255f83c6cd5f16efebc`.

**Method decision: fixed rule.** The best global fixed setting reaches both hindsight capacities in this inspected allocation; an adaptive selector has no demonstrated additional qualified capacity here.

Best global fixed capacity is **2**, one hindsight choice per program **2**, and one per program/seed **2** qualified successes. These counts concern fifteen available program/seed pairs, of which seven had no certified pre-family failure; eight are causally excluded from new starts. They are retrospective development capacity results, not deployed selectors or prospective generalization.

The tie-resolved global choice is **lead020**; all co-optimal settings are lead020. No-stride remains the simplest effective component family. Stride is retired from this active comparison. No predictor, new draw, ARDY/SONIC training, lateral tuning or final36 access occurred. The earlier replication and old left ablation remain unchanged.

| Setting | Qualified successes, all 15 capacity cells | Eligible measured task / quality / qualified passes | Eligible measured cells |
| --- | --- | --- | --- |
| keep | 0 | 0/7/0 | 7 |
| lead012 | 1 | 1/7/1 | 7 |
| lead020 | 2 | 2/7/2 | 7 |
| lead032 | 1 | 1/7/1 | 7 |

All original keep cases in this replication fail the task. Zero qualified regressions against keep therefore is not preservation evidence. Regressions against the already effective 0.20 setting are reported independently. Rejected construction followed by keep is a pipeline fallback, not an executed adaptation failure.

| Contrast | Evaluable measured pairs | Qualified rescues | Qualified regressions | Quality regressions |
| --- | --- | --- | --- | --- |
| keep__lead020 | 15 | 2 | 0 | 1 |
| keep__lead012 | 7 | 1 | 0 | 0 |
| keep__lead032 | 7 | 1 | 0 | 0 |
| lead020__lead012 | 7 | 0 | 1 | 0 |
| lead020__lead032 | 7 | 0 | 1 | 0 |

**All-program outcomes.** Entries are task / quality / conjunction. Causally excluded alternatives have a zero qualified-capacity upper bound under the fixed prefix contract, but no measured trajectory, margin or quality result is imputed. The original A/1701 infrastructure-unavailable pair remains excluded from the fifteen-pair denominator, with its historical failure preserved.

| Program/seed | keep | 0.12 | 0.20 | 0.32 |
| --- | --- | --- | --- | --- |
| A/1701 | Prior infrastructure unavailable | Prior infrastructure unavailable | Prior infrastructure unavailable | Prior infrastructure unavailable |
| A/1702 | F/P/F | P/P/P | P/P/P | P/P/P |
| B/1701 | F/P/F | F/P/F | F/P/F | F/P/F |
| B/1702 | F/P/F | F/P/F | F/P/F | F/P/F |
| C/1701 | F/P/F | Prefix ruled out; unmeasured | F/P/F | Prefix ruled out; unmeasured |
| C/1702 | F/P/F | Prefix ruled out; unmeasured | F/P/F | Prefix ruled out; unmeasured |
| D/1701 | F/F/F | Prefix ruled out; unmeasured | F/F/F | Prefix ruled out; unmeasured |
| D/1702 | F/F/F | Prefix ruled out; unmeasured | F/F/F | Prefix ruled out; unmeasured |
| E/1701 | F/P/F | F/P/F | F/P/F | F/P/F (keep fallback) |
| E/1702 | F/P/F | F/P/F | P/P/P | F/P/F (keep fallback) |
| F/1701 | F/P/F | Prefix ruled out; unmeasured | F/F/F | Prefix ruled out; unmeasured |
| F/1702 | F/P/F | F/P/F | F/P/F | F/P/F |
| G/1701 | F/P/F | F/P/F | F/P/F | F/P/F |
| G/1702 | F/P/F | Prefix ruled out; unmeasured | F/P/F | Prefix ruled out; unmeasured |
| H/1701 | F/F/F | Prefix ruled out; unmeasured | F/F/F | Prefix ruled out; unmeasured |
| H/1702 | F/F/F | Prefix ruled out; unmeasured | F/F/F | Prefix ruled out; unmeasured |

**Hindsight capacity, not selection performance.** The objective is qualified-success count; the frozen tie order is keep,0.20,0.12,0.32. Every co-optimal set is shown so ties cannot be mistaken for a need for seed-dependent action. Unknown runtime outcomes would be bounded rather than scored as physical zeros.

| Scope | Lower capacity | Upper capacity |
| --- | --- | --- |
| Global fixed | 2 | 2 |
| Per program | 2 | 2 |
| Per program/seed | 2 | 2 |

| Program | Shared-choice capacity | Tie-resolved choice | All co-optimal settings |
| --- | --- | --- | --- |
| A | 1 | lead020 | lead020, lead012, lead032 |
| B | 0 | keep | keep, lead020, lead012, lead032 |
| C | 0 | keep | keep, lead020, lead012, lead032 |
| D | 0 | keep | keep, lead020, lead012, lead032 |
| E | 1 | lead020 | lead020 |
| F | 0 | keep | keep, lead020, lead012, lead032 |
| G | 0 | keep | keep, lead020, lead012, lead032 |
| H | 0 | keep | keep, lead020, lead012, lead032 |

| Program/seed | Capacity | Tie-resolved choice | All co-optimal settings |
| --- | --- | --- | --- |
| A-1702 | 1 | lead020 | lead020, lead012, lead032 |
| B-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| B-1702 | 0 | keep | keep, lead020, lead012, lead032 |
| C-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| C-1702 | 0 | keep | keep, lead020, lead012, lead032 |
| D-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| D-1702 | 0 | keep | keep, lead020, lead012, lead032 |
| E-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| E-1702 | 1 | lead020 | lead020 |
| F-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| F-1702 | 0 | keep | keep, lead020, lead012, lead032 |
| G-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| G-1702 | 0 | keep | keep, lead020, lead012, lead032 |
| H-1701 | 0 | keep | keep, lead020, lead012, lead032 |
| H-1702 | 0 | keep | keep, lead020, lead012, lead032 |

**Individual physical results.** Position accepts a nonnegative margin; STOP requires a strictly positive margin. These are the original sampled evaluations, not a continuous-time guarantee. Signed errors use the final task heading; quality endpoint error is relative to each delivered reference at the last captured sample. Aggregate quality gates are not assigned invented instantaneous onset times.

| Cell | Position max / margin (m) | Peak time (s) | STOP peak / margin (m/s) | Heading max (deg) | Final along / lateral (m) | First sampled failure |
| --- | --- | --- | --- | --- | --- | --- |
| A-1702-keep | 0.077735 / 0.122265 | 1.48 | 0.103195 / -0.003195 | 4.779 | -0.071939 / 0.012400 | {'channel': 'strict_STOP_speed', 'time_s': 4.32} |
| A-1702-lead012 | 0.088950 / 0.111050 | 4.24 | 0.086712 / 0.013288 | 4.779 | -0.080583 / 0.009239 | None |
| A-1702-lead020 | 0.114014 / 0.085986 | 4.94 | 0.065897 / 0.034103 | 4.779 | -0.113682 / 0.006484 | None |
| A-1702-lead032 | 0.161604 / 0.038396 | 4.64 | 0.040898 / 0.059102 | 4.779 | -0.155784 / 0.013556 | None |
| B-1701-keep | 0.133640 / 0.066360 | 1.06 | 0.150453 / -0.050453 | 3.610 | -0.062630 / 0.027889 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1701-lead012 | 0.133640 / 0.066360 | 1.06 | 0.182496 / -0.082496 | 3.535 | -0.070319 / 0.025684 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1701-lead020 | 0.133640 / 0.066360 | 1.06 | 0.154652 / -0.054652 | 3.344 | -0.070061 / 0.016645 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1701-lead032 | 0.134047 / 0.065953 | 4.48 | 0.146330 / -0.046330 | 3.748 | -0.131404 / 0.015532 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1702-keep | 0.083405 / 0.116595 | 3.84 | 0.149406 / -0.049406 | 4.846 | 0.056147 / 0.027556 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1702-lead012 | 0.082237 / 0.117763 | 3.82 | 0.147383 / -0.047383 | 4.846 | 0.039737 / 0.024575 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1702-lead020 | 0.126923 / 0.073077 | 4.00 | 0.230033 / -0.130033 | 4.846 | 0.084360 / 0.021863 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| B-1702-lead032 | 0.080765 / 0.119235 | 3.96 | 0.197438 / -0.097438 | 4.846 | 0.044998 / 0.000511 | {'channel': 'strict_STOP_speed', 'time_s': 3.6} |
| C-1701-keep | 0.213510 / -0.013510 | 4.52 | 0.316998 / -0.216998 | 4.432 | -0.057458 / -0.135311 | {'channel': 'position', 'time_s': 4.34} |
| C-1701-lead020 | 0.213510 / -0.013510 | 4.52 | 0.220559 / -0.120559 | 4.432 | -0.117004 / -0.123321 | {'channel': 'position', 'time_s': 4.34} |
| C-1702-keep | 0.297392 / -0.097392 | 6.12 | 0.278090 / -0.178090 | 5.112 | -0.153031 / -0.177150 | {'channel': 'position', 'time_s': 4.26} |
| C-1702-lead020 | 0.298017 / -0.098017 | 6.12 | 0.182268 / -0.082268 | 5.112 | -0.192682 / -0.181838 | {'channel': 'position', 'time_s': 4.26} |
| CONTROL-1701-keep | 0.158159 / 0.041841 | 2.16 | 0.065293 / 0.034707 | 5.531 | -0.139469 / -0.029427 | None |
| CONTROL-1701-lead020 | 0.144061 / 0.055939 | 2.16 | 0.057793 / 0.042207 | 5.531 | -0.138386 / -0.029862 | None |
| D-1701-keep | 0.382037 / -0.182037 | 4.64 | 0.203514 / -0.103514 | 4.920 | 0.060107 / 0.320646 | {'channel': 'self_contact', 'time_s': 0.02} |
| D-1701-lead020 | 0.382037 / -0.182037 | 4.64 | 0.128325 / -0.028325 | 4.920 | -0.151216 / 0.274537 | {'channel': 'self_contact', 'time_s': 0.02} |
| D-1702-keep | 0.290566 / -0.090566 | 5.84 | 0.215557 / -0.115557 | 4.613 | 0.048231 / 0.143703 | {'channel': 'position', 'time_s': 4.66} |
| D-1702-lead020 | 0.290566 / -0.090566 | 5.84 | 0.139877 / -0.039877 | 4.613 | -0.151023 / 0.121740 | {'channel': 'position', 'time_s': 4.66} |
| E-1701-keep | 0.164186 / 0.035814 | 6.92 | 0.289477 / -0.189477 | 5.263 | -0.110394 / -0.048138 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| E-1701-lead012 | 0.204452 / -0.004452 | 7.98 | 0.137726 / -0.037726 | 5.263 | -0.197844 / -0.046369 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| E-1701-lead020 | 0.218564 / -0.018564 | 7.98 | 0.111331 / -0.011331 | 5.263 | -0.211273 / -0.051216 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| E-1701-lead032 | 0.164186 / 0.035814 | 6.92 | 0.289477 / -0.189477 | 5.263 | -0.110394 / -0.048138 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| E-1702-keep | 0.135359 / 0.064641 | 8.04 | 0.197658 / -0.097658 | 6.321 | -0.113574 / -0.072777 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| E-1702-lead012 | 0.180157 / 0.019843 | 8.06 | 0.113937 / -0.013937 | 6.321 | -0.170757 / -0.057355 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| E-1702-lead020 | 0.194792 / 0.005208 | 8.06 | 0.096547 / 0.003453 | 6.321 | -0.181809 / -0.069298 | None |
| E-1702-lead032 | 0.135359 / 0.064641 | 8.04 | 0.197658 / -0.097658 | 6.321 | -0.113574 / -0.072777 | {'channel': 'strict_STOP_speed', 'time_s': 7.08} |
| F-1701-keep | 0.256481 / -0.056481 | 7.94 | 0.165935 / -0.065935 | 7.373 | -0.229567 / 0.038402 | {'channel': 'position', 'time_s': 5.88} |
| F-1701-lead020 | 0.378531 / -0.178531 | 9.06 | 0.155451 / -0.055451 | 7.373 | -0.377784 / 0.023427 | {'channel': 'position', 'time_s': 5.88} |
| F-1702-keep | 0.176207 / 0.023793 | 6.34 | 0.127737 / -0.027737 | 8.895 | -0.142487 / 0.053180 | {'channel': 'strict_STOP_speed', 'time_s': 8.08} |
| F-1702-lead012 | 0.215029 / -0.015029 | 9.04 | 0.084271 / 0.015729 | 8.895 | -0.211656 / 0.036531 | {'channel': 'position', 'time_s': 8.36} |
| F-1702-lead020 | 0.238477 / -0.038477 | 8.98 | 0.083020 / 0.016980 | 8.895 | -0.235262 / 0.034664 | {'channel': 'position', 'time_s': 7.92} |
| F-1702-lead032 | 0.282210 / -0.082210 | 8.94 | 0.115365 / -0.015365 | 8.895 | -0.279596 / 0.029380 | {'channel': 'position', 'time_s': 7.86} |
| G-1701-keep | 0.149295 / 0.050705 | 1.76 | 0.121174 / -0.021174 | 3.781 | -0.005494 / -0.025868 | {'channel': 'strict_STOP_speed', 'time_s': 4.84} |
| G-1701-lead012 | 0.149295 / 0.050705 | 1.76 | 0.163779 / -0.063779 | 3.781 | -0.012185 / -0.022956 | {'channel': 'strict_STOP_speed', 'time_s': 4.84} |
| G-1701-lead020 | 0.149295 / 0.050705 | 1.76 | 0.146060 / -0.046060 | 3.781 | -0.016037 / -0.027980 | {'channel': 'strict_STOP_speed', 'time_s': 4.84} |
| G-1701-lead032 | 0.149295 / 0.050705 | 1.76 | 0.103345 / -0.003345 | 3.781 | 0.003466 / -0.030536 | {'channel': 'strict_STOP_speed', 'time_s': 4.9} |
| G-1702-keep | 0.206857 / -0.006857 | 1.80 | 0.093004 / 0.006996 | 3.867 | -0.065766 / -0.104421 | {'channel': 'position', 'time_s': 1.68} |
| G-1702-lead020 | 0.206857 / -0.006857 | 1.80 | 0.107224 / -0.007224 | 3.867 | -0.080049 / -0.106957 | {'channel': 'position', 'time_s': 1.68} |
| H-1701-keep | 0.400834 / -0.200834 | 6.76 | 0.106002 / -0.006002 | 3.879 | -0.370555 / 0.082290 | {'channel': 'position', 'time_s': 2.94} |
| H-1701-lead020 | 0.401621 / -0.201621 | 7.80 | 0.157344 / -0.057344 | 3.879 | -0.385625 / 0.085765 | {'channel': 'position', 'time_s': 2.94} |
| H-1702-keep | 0.438603 / -0.238603 | 6.72 | 0.078079 / 0.021921 | 5.802 | -0.383855 / 0.181100 | {'channel': 'position', 'time_s': 3.18} |
| H-1702-lead020 | 0.431315 / -0.231315 | 6.68 | 0.100016 / -0.000016 | 5.802 | -0.377624 / 0.177056 | {'channel': 'position', 'time_s': 3.18} |

| Cell | Sliding ratio | Slip mean / p95 (m/s) | Self-contact peak (N) | Endpoint quality error / margin (m) | Quality rejection |
| --- | --- | --- | --- | --- | --- |
| A-1702-keep | 0.118932 | 0.050887 / 0.338131 | 62.766 | 0.074920 / 0.275080 | None |
| A-1702-lead012 | 0.111111 | 0.048550 / 0.326401 | 62.766 | 0.078281 / 0.271719 | None |
| A-1702-lead020 | 0.108173 | 0.049548 / 0.326431 | 62.766 | 0.102487 / 0.247513 | None |
| A-1702-lead032 | 0.103118 | 0.046458 / 0.298620 | 62.766 | 0.123490 / 0.226510 | None |
| B-1701-keep | 0.127976 | 0.057058 / 0.327360 | 198.365 | 0.066419 / 0.283581 | None |
| B-1701-lead012 | 0.126126 | 0.056086 / 0.335367 | 198.365 | 0.071016 / 0.278984 | None |
| B-1701-lead020 | 0.128358 | 0.059752 / 0.368027 | 198.365 | 0.061065 / 0.288935 | None |
| B-1701-lead032 | 0.123894 | 0.053813 / 0.308491 | 198.365 | 0.100929 / 0.249071 | None |
| B-1702-keep | 0.126126 | 0.055460 / 0.349294 | 251.441 | 0.065397 / 0.284603 | None |
| B-1702-lead012 | 0.126506 | 0.055981 / 0.349373 | 251.441 | 0.048928 / 0.301072 | None |
| B-1702-lead020 | 0.136364 | 0.059911 / 0.361024 | 251.441 | 0.096110 / 0.253890 | None |
| B-1702-lead032 | 0.140719 | 0.060462 / 0.385807 | 251.441 | 0.075321 / 0.274679 | None |
| C-1701-keep | 0.114478 | 0.042992 / 0.244975 | 97.386 | 0.129656 / 0.220344 | None |
| C-1701-lead020 | 0.110924 | 0.040114 / 0.218139 | 97.386 | 0.159517 / 0.190483 | None |
| C-1702-keep | 0.117840 | 0.049081 / 0.336441 | 101.092 | 0.217823 / 0.132177 | None |
| C-1702-lead020 | 0.112561 | 0.045766 / 0.308462 | 101.092 | 0.254595 / 0.095405 | None |
| CONTROL-1701-keep | 0.029605 | 0.017146 / 0.031954 | 26.160 | 0.140881 / 0.209119 | None |
| CONTROL-1701-lead020 | 0.029605 | 0.017310 / 0.031954 | 26.160 | 0.131801 / 0.218199 | None |
| D-1701-keep | 0.109677 | 0.053222 / 0.338226 | 351.910 | 0.334234 / 0.015766 | reference_path_tracking_error, excessive_self_contact |
| D-1701-lead020 | 0.105678 | 0.052098 / 0.339596 | 351.910 | 0.307719 / 0.042281 | reference_path_tracking_error, excessive_self_contact |
| D-1702-keep | 0.114618 | 0.061405 / 0.374009 | 248.348 | 0.161644 / 0.188356 | reference_path_tracking_error |
| D-1702-lead020 | 0.112561 | 0.060774 / 0.375097 | 248.348 | 0.184387 / 0.165613 | reference_path_tracking_error |
| E-1701-keep | 0.096121 | 0.041217 / 0.211474 | 94.121 | 0.119477 / 0.230523 | None |
| E-1701-lead012 | 0.091820 | 0.040262 / 0.213048 | 94.121 | 0.197505 / 0.152495 | None |
| E-1701-lead020 | 0.091820 | 0.040342 / 0.199453 | 94.121 | 0.202304 / 0.147696 | None |
| E-1701-lead032 | 0.096121 | 0.041217 / 0.211474 | 94.121 | 0.119477 / 0.230523 | None |
| E-1702-keep | 0.116776 | 0.045928 / 0.244896 | 140.633 | 0.133594 / 0.216406 | None |
| E-1702-lead012 | 0.112745 | 0.044636 / 0.240415 | 140.633 | 0.174961 / 0.175039 | None |
| E-1702-lead020 | 0.112930 | 0.044965 / 0.240467 | 140.633 | 0.180295 / 0.169705 | None |
| E-1702-lead032 | 0.116776 | 0.045928 / 0.244896 | 140.633 | 0.133594 / 0.216406 | None |
| F-1701-keep | 0.098748 | 0.034414 / 0.208761 | 207.088 | 0.234774 / 0.115226 | None |
| F-1701-lead020 | 0.087791 | 0.032600 / 0.207937 | 207.088 | 0.363550 / -0.013550 | reference_endpoint_tracking_error |
| F-1702-keep | 0.104225 | 0.038928 / 0.247602 | 95.047 | 0.154490 / 0.195510 | None |
| F-1702-lead012 | 0.101955 | 0.038473 / 0.250832 | 95.711 | 0.209330 / 0.140670 | None |
| F-1702-lead020 | 0.103064 | 0.039699 / 0.271636 | 74.812 | 0.222689 / 0.127311 | None |
| F-1702-lead032 | 0.097493 | 0.036658 / 0.242216 | 74.812 | 0.242568 / 0.107432 | None |
| G-1701-keep | 0.131687 | 0.049674 / 0.317290 | 188.096 | 0.033267 / 0.316733 | None |
| G-1701-lead012 | 0.132231 | 0.049779 / 0.318232 | 188.096 | 0.024748 / 0.325252 | None |
| G-1701-lead020 | 0.134298 | 0.049997 / 0.318232 | 188.096 | 0.028441 / 0.321559 | None |
| G-1701-lead032 | 0.134298 | 0.051204 / 0.318232 | 188.096 | 0.046335 / 0.303665 | None |
| G-1702-keep | 0.147705 | 0.058563 / 0.379026 | 78.826 | 0.132701 / 0.217299 | None |
| G-1702-lead020 | 0.144578 | 0.058019 / 0.380733 | 78.826 | 0.126955 / 0.223045 | None |
| H-1701-keep | 0.099723 | 0.037165 / 0.225825 | 219.747 | 0.381537 / -0.031537 | reference_endpoint_tracking_error |
| H-1701-lead020 | 0.105263 | 0.040672 / 0.241212 | 219.747 | 0.381090 / -0.031090 | reference_endpoint_tracking_error |
| H-1702-keep | 0.108911 | 0.037921 / 0.203333 | 160.958 | 0.427202 / -0.077202 | reference_endpoint_tracking_error |
| H-1702-lead020 | 0.110638 | 0.038330 / 0.203903 | 160.958 | 0.404192 / -0.054192 | reference_endpoint_tracking_error |

Full task/quality reports, every gate, coverage, STOP sample counts and reference/action divergence times are in analysis/result.json (local experiment artifact). The 62-cell CSV (local experiment artifact) retains all sixty main cells and two separate control cells.

**Measured histories and exposure.** Each plot shows original-task along/lateral error and measured horizontal speed through full STOP. Dotted lines mark actual consumed-reference exposure; dash-dot lines mark applied-action divergence. Initial contacts are retained. Reference XYZ alone is not interpreted as a directly consumed root-position command.

A histories (local experiment artifact)
B histories (local experiment artifact)
C histories (local experiment artifact)
D histories (local experiment artifact)
E histories (local experiment artifact)
F histories (local experiment artifact)
G histories (local experiment artifact)
H histories (local experiment artifact)

**Pre-choice observability audit.** The following values are from keep at the frozen S−1.70 s cutoff, before any member could act under the conservative S−1.68 s bound. No post-exposure or hold-entry state was used. Along/lateral task errors use simulator ground truth and are privileged diagnostics unless an estimator is supplied; numerical differences between seeds do not establish a reliable discriminator. No thresholds, classifier, predictor or feedback law were fitted. The offline full-preload experiment does not implement a runtime publication deadline.

| Program/seed | Cutoff / last sample (s) | Along / lateral error (m) | Speed (m/s) | Previous 0.20 s mean speed |
| --- | --- | --- | --- | --- |
| A/1702 | 2.62 / 2.62 | -0.044329 / 0.041634 | 0.459110 | 0.483193 |
| B/1701 | 1.90 / 1.90 | -0.096840 / -0.037748 | 0.776499 | 0.955351 |
| B/1702 | 1.90 / 1.90 | 0.028596 / -0.040606 | 0.831151 | 1.043772 |
| C/1701 | 5.30 / 5.30 | -0.067616 / -0.078970 | 0.587032 | 0.594707 |
| C/1702 | 5.30 / 5.30 | -0.166180 / -0.108380 | 0.474626 | 0.482391 |
| D/1701 | 5.86 / 5.86 | 0.013029 / 0.332445 | 0.642707 | 0.694065 |
| D/1702 | 5.86 / 5.86 | 0.000129 / 0.290101 | 0.607690 | 0.706183 |
| E/1701 | 5.38 / 5.38 | -0.135421 / -0.088369 | 0.467102 | 0.393657 |
| E/1702 | 5.38 / 5.38 | -0.059631 / -0.061581 | 0.628978 | 0.555618 |
| F/1701 | 6.38 / 6.38 | -0.221647 / 0.059173 | 0.722032 | 0.617264 |
| F/1702 | 6.38 / 6.38 | -0.138241 / 0.106402 | 0.637747 | 0.505185 |
| G/1701 | 3.14 / 3.14 | 0.030792 / -0.054030 | 0.712770 | 0.823780 |
| G/1702 | 3.14 / 3.14 | 0.014514 / -0.129904 | 0.805327 | 0.893074 |
| H/1701 | 5.06 / 5.06 | -0.273331 / 0.087999 | 0.416959 | 0.386354 |
| H/1702 | 5.06 / 5.06 | -0.277167 / 0.187415 | 0.265925 | 0.223546 |

**Separate success-conditioned control.** Gentle WALK 0.5 m at 0.25 m/s then STOP 1 s, seed1701, one archived original draw. The historical unchanged task and quality success selected this control; neither old outcome nor its initial receipt was assumed reusable. Both methods were paired under the frozen offline contract, with hold 2.00 s and braking 1.60 s for fixed 0.20. The control is not pooled into rescue prevalence or timing selection.

| Control arm | Task / quality / conjunction | Position margin (m) | STOP margin (m/s) | Quality reasons |
| --- | --- | --- | --- | --- |
| keep | P/P/P | 0.041841 | 0.034707 | None |
| lead020 | P/P/P | 0.055939 | 0.042207 | None |

One success-conditioned design/seed cannot establish broad preservation. Any failure to reproduce unchanged success remains visible; no replacement control or altered roll-in was used.

**Construction and accounting.** E/0.32 was rejected at native frame168 (6.72 s) with `terminal:root_height_slew_infeasible`, before requested hold6.76 s; requested braking was6.36 s. The hold pose itself passed projection, but the transition did not satisfy the unchanged height/slew constraints. This is a construction dependency, not a measured robot failure. Its two seed cells reuse unchanged keep and consume no simulator starts. No hold delay, gain change or new candidate was substituted.

| Construction | Admitted | Reason |
| --- | --- | --- |
| A/lead012 | True | None |
| A/lead032 | True | None |
| B/lead012 | True | None |
| B/lead032 | True | None |
| E/lead012 | True | None |
| E/lead032 | False | terminal:root_height_slew_infeasible |
| F/lead012 | True | None |
| F/lead032 | True | None |
| G/lead012 | True | None |
| G/lead032 | True | None |
| CONTROL/lead020 | True | None |

11 construction attempts used 39.405 worker seconds in total, under individual 30 s caps. 14 simulator starts were counted, including the separate control. The two E fallbacks reduced the approved maximum of 16 starts to 14; every other allocated start completed. Thirty old outcomes were validated and reused. Sixteen new-lead cells were causally excluded without construction/physics. No retry, hidden smoke run or generator call occurred.

New runtime-qualified captures: 14/14. Total recorded simulator worker time: 483.900 s. Shared qualification failure: None. Complete coverage among scored cells: True; termination events among scored cells: 0. Each alias remains a logical cell, not an independent run.

**Evidence boundary and stop rule.** This completed allocation evaluates a centered scheduling rule whose center and transition change with lead. It does not isolate time translation of a fixed pose sequence. The eight inherited failures cannot be rescued by this terminal family under the frozen prefix contract; they require earlier/interface changes for broader coverage. Successes and comparisons in these repeatedly inspected designs remain development evidence. No further A/E tuning or timing additions are justified by this allocation. Keep final36 sealed. Real-time delivery, continuous physics, new-program validation and deployment-quality dataset/training claims remain untested.

**Mechanism paragraph.** The no-stride centered terminal family exposes a progress-versus-settling tradeoff within the frozen motion stack. Across the fifteen evaluable development program/seed pairs, the best global fixed setting attains 2 qualified outcomes, while program-shared and seed-specific hindsight choices attain 2 and 2; eight pairs already violate full-task criteria before terminal influence is possible. E's 0.32 s construction fails the unchanged height/slew constraint and retains its keep fallback. These findings delimit terminal scheduling capacity and support the stated fixed rule scope, without establishing a deployed selector, broad preservation, or new-draw generalization.