Experiments / E1
Detection of up-but-slow links
Can deadlines detect an up-but-slow ISL without excessive false alarms, and where do they fail relative to local and conventional monitors?
Label: DERIVED
statistics and accounting
Label: MODELLED
network outcomes
Pre-registered hypothesis
At a pilot-selected operating point with clean-packet FPR at most 1%, ChronoRoute detects nominal/strong faults more often than LS-H and CRH-16, and has shorter censored detection delay than LS-D. Small faults inside the budget may remain undetected. LQ may beat ChronoRoute; this is a primary comparison, not an excluded case.
Frozen preregistration, identified and verified against this result’s manifest.
Verdict
Mixed, mostly not supported: at the frozen pilot rule (clean-packet false-positive rate 0.0041 to 0.00587 across cells, within the registered 0.01 limit), the hypothesis is not supported for the nominal rate cut (factor 0.3) and not supported for the added-delay family, supported for the strong rate cut (factor 0.2) and supported for ARQ loss (significantly more detections than both LS-H and CRH-16 in 10 of 24 nominal and strong cells; neither comparator has a slowness detector). LQ beats ChronoRoute significantly in 13 cells, and the LS-D comparison is not valid evidence.
- Operating point: rule S=1.5 u1 with k = 2 consecutive alarms, chosen on pilot seeds only and frozen; clean-packet false-positive rate 0.0041 to 0.00587 across cells, within the registered 0.01 limit.
- Rate cut, runs detected per shell (C5, C2): nominal 0.3 (first, middle, last third): C5 9/36, 0/36, 0/36; C2 0/36, 0/36, 0/36 — not supported. strong 0.2: C5 36/36; C2 36/36 — supported. small 0.6: C5 0/36; C2 0/36 — may stay undetected by hypothesis.
- Added delay, runs detected per shell (C5, C2): nominal 1 ms (first, middle, last third): C5 3/36, 0/36, 0/36; C2 0/36, 0/36, 0/36 — not supported. strong 4 ms: C5 2/36; C2 0/36 — not supported. small 0.25 ms: C5 0/36; C2 0/36 — may stay undetected by hypothesis.
- ARQ loss, runs detected per shell (C5, C2): nominal 0.1 (first, middle, last third): C5 27/36, 15/36, 15/36; C2 33/36, 14/36, 14/36 — supported. strong 0.3: C5 23/36; C2 23/36 — supported. small 0.01: C5 7/36; C2 7/36 — may stay undetected by hypothesis.
- Why delays are missed: S = 1.5 leaves several milliseconds of spare time per hop mid-path. Every unselected rule that detects the 4 ms delay in at least half the runs has a clean false-positive rate of at least C5 0.0282 and C2 0.0202, above the 1% limit.
- Null controls (rate factor 1, delay 0, ARQ loss 0): no detections in any cell.
- Detection against LS-H: 10 significant wins, 5 non-significant wins, 15 ties, 0 significant losses and 0 non-significant losses over 30 cells; CRH-16: 10 significant wins, 5 non-significant wins, 15 ties, 0 significant losses and 0 non-significant losses over 30 cells. These wins hold by construction: LS-H and CRH-16 detect no fault in any cell, because they have no slowness detector.
- Against E2E: 3 significant wins, 10 non-significant wins, 17 ties, 0 significant losses and 0 non-significant losses over 30 cells. E2E's clean false-positive rate is 0 in every cell, so this is not a matched-false-positive comparison.
- Against LQ, a practical local monitor (false-positive rate 0 to 0.00913): 0 significant wins, 2 non-significant wins, 12 ties, 13 significant losses and 3 non-significant losses over 30 cells; 26 registered material losses (detection and restricted delay). These losses are robust.
- Against LS-D: 2 significant wins, 2 non-significant wins, 2 ties, 17 significant losses and 7 non-significant losses over 30 cells; 35 registered material losses. They are kept as registered but are not valid evidence: LS-D measures the queue estimate plus the exact injected fault delay plus the expected ARQ delay from the configured loss probability, without noise, and re-advertises any change at the 1 s update interval. Its detection hardly depends on fault size: for 0.25, 1 and 4 ms of added delay it detects C5 20/36, 20/36, 19/36; C2 23/36, 23/36, 23/36. The registered LS-D delay comparison therefore cannot be judged.
- Registered material losses: 61 (30 detection, 31 restricted delay); point disadvantages: 82. All are in paired_contrasts.csv.
- 23 adjusted p-values are stored as 0. They are below the numerical resolution of the computation (9 from zero-variance paired differences, the rest from cancellation), not exactly zero.
Every number in this verdict is computed at build time from the verified summary.csv and the registered paired comparisons in paired_contrasts.csv (SHA-256 c44cf51fcbd9985204222fba0a7b419c2418ce3187bffd46f37b2ebfa639b6d5, checked against the manifest).
C5 orbital inputs are REAL public data; C2 orbits and positions are MODELLED. Topology, probes, faults and network behaviour are MODELLED. Starlink’s operational routing is UNKNOWN; Starlink does not use T-CRH.
Charts and data
Download CSV · Download manifest · Download paired contrasts
Source: results/E01/20261011T154940Z-confirmatory-resumed/summary.csv. SHA-256: bee06647b9f70f64c12ef2b74f918b6ebeb9a8a5f0e5dc91c0f251ec37aaafc2.
This first view shows the 7 rows of the initial selection, out of 1620. With scripting, the page loads the same summary.csv offered for download and checks its SHA-256 against the manifest before other selections are enabled.
Infinite or undefined results; retained, not replaced
- o_q / C2 / rate-0.3-middle / as run / 2: estimate undefined; interval undefined to undefined
Data table
Showing 1–7 of 7 matching rows. The download contains every original row and column.
| const | condition | scheme | rule | selected | k | n | complete | event_mean | event_lo | event_hi | fpr | fpr_lo | fpr_hi | restricted_mean_s | restricted_lo | restricted_hi | lost_at_fault | effective_p99_inf_flows |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| C2 | rate-0.3-middle | chronoroute | S=1.5 u1 | True | 2 | 36 | True | 0.0 | 0.0 | 0.0 | 0.00413280544557894 | 0.003264455634658795 | 0.005069629275274431 | 4.0 | 4.0 | 4.0 | 0 | 0 |
| C2 | rate-0.3-middle | crh16 | as run | True | 2 | 36 | True | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 4.0 | 4.0 | 4.0 | 0 | 0 |
| C2 | rate-0.3-middle | e2e | as run | True | 2 | 36 | True | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 4.0 | 4.0 | 4.0 | 0 | 0 |
| C2 | rate-0.3-middle | lq | as run | True | 2 | 36 | True | 1.0 | 1.0 | 1.0 | 0.0 | 0.0 | 0.0 | 1.3675388035 | 1.3553703907003556 | 1.3797072162996444 | 0 | 0 |
| C2 | rate-0.3-middle | ls_d | as run | True | 2 | 36 | True | 0.6111111111111112 | 0.4438257405985373 | 0.778396481623685 | undefined | undefined | undefined | 1.6560312404166668 | 1.0120991678479843 | 2.2999633129853496 | 0 | 0 |
| C2 | rate-0.3-middle | ls_h | as run | True | 2 | 36 | True | 0.0 | 0.0 | 0.0 | undefined | undefined | undefined | 4.0 | 4.0 | 4.0 | 0 | 0 |
| C2 | rate-0.3-middle | o_q | as run | True | 2 | 36 | True | undefined | undefined | undefined | undefined | undefined | undefined | undefined | undefined | undefined | 0 | 0 |
Where ChronoRoute loses
Every listed point disadvantage is retained, including privileged-oracle advantages. Where a registered paired comparison exists for the same cell, its paired Student-t 95% interval and Holm-adjusted p-value are quoted; otherwise marginal intervals alone do not establish a significant loss. Ties in the predicted stuck-loss reduction are listed as unsupported advantages.
Rows against LS-D are kept as registered, but LS-D reads the injected delay without measurement noise and advertises any change, so they are not valid evidence about slowness detection.
Rows against LQ are robust: LQ is a practical local monitor using only local queue and ARQ state.
False-positive rows against E2E, LQ and CRH-16 compare different operating points; E2E runs at a clean false-positive rate of 0.
| Cell | Metric | Policy | ChronoRoute | Comparator | Comparator value | Interpretation | Source rows |
|---|---|---|---|---|---|---|---|
| C2 / arq-0-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 12, 488 |
| C2 / arq-0-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 12, 542 |
| C2 / arq-0-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 12, 596 |
| C2 / arq-0.01-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 39, 491 |
| C2 / arq-0.01-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 39, 545 |
| C2 / arq-0.01-middle | event_mean | chronoroute | 0.19444444444444445 | lq | 0.3333333333333333 | Not significant after adjustment; paired difference -0.139 [-0.258, -0.0202] (Student-t 95%), Holm-adjusted p 1 | 39, 599 |
| C2 / arq-0.01-middle | restricted_mean_s | chronoroute | 3.469621274916667 | lq | 2.9731989320277776 | Not significant after adjustment; paired difference 0.496 [0.149, 0.844] (Student-t 95%), Holm-adjusted p 0.533 | 39, 599 |
| C2 / arq-0.01-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0007639684689377366 | Point disadvantage; no registered paired contrast for this metric | 39, 599 |
| C2 / arq-0.01-middle | event_mean | chronoroute | 0.19444444444444445 | ls_d | 0.6666666666666666 | Registered material loss; paired difference -0.472 [-0.662, -0.283] (Student-t 95%), Holm-adjusted p 0.0015 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 39, 653 |
| C2 / arq-0.01-middle | restricted_mean_s | chronoroute | 3.469621274916667 | ls_d | 1.5191717230555555 | Registered material loss; paired difference 1.95 [1.28, 2.62] (Student-t 95%), Holm-adjusted p 0.000121 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 39, 653 |
| C2 / arq-0.1-first | fpr | chronoroute | 0.0040969377126588435 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 66, 494 |
| C2 / arq-0.1-first | fpr | chronoroute | 0.0040969377126588435 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 66, 548 |
| C2 / arq-0.1-first | event_mean | chronoroute | 0.9166666666666666 | lq | 1.0 | Not significant after adjustment; paired difference -0.0833 [-0.178, 0.0115] (Student-t 95%), Holm-adjusted p 1 | 66, 602 |
| C2 / arq-0.1-first | restricted_mean_s | chronoroute | 0.9565635304722223 | lq | 0.1820156676111111 | Not significant after adjustment; paired difference 0.775 [0.354, 1.2] (Student-t 95%), Holm-adjusted p 0.0587 | 66, 602 |
| C2 / arq-0.1-last | fpr | chronoroute | 0.004132087919719435 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 93, 497 |
| C2 / arq-0.1-last | fpr | chronoroute | 0.004132087919719435 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 93, 551 |
| C2 / arq-0.1-last | event_mean | chronoroute | 0.3888888888888889 | lq | 1.0 | Registered material loss; paired difference -0.611 [-0.778, -0.444] (Student-t 95%), Holm-adjusted p 0.00000143 | 93, 605 |
| C2 / arq-0.1-last | restricted_mean_s | chronoroute | 3.0245349839999998 | lq | 0.3206317333333334 | Registered material loss; paired difference 2.7 [2.27, 3.14] (Student-t 95%), Holm-adjusted p 1.48e-12 | 93, 605 |
| C2 / arq-0.1-last | fpr | chronoroute | 0.004132087919719435 | lq | 0.0034723427896801973 | Point disadvantage; no registered paired contrast for this metric | 93, 605 |
| C2 / arq-0.1-last | event_mean | chronoroute | 0.3888888888888889 | ls_d | 0.6388888888888888 | Not significant after adjustment; paired difference -0.25 [-0.511, 0.0105] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 93, 659 |
| C2 / arq-0.1-last | restricted_mean_s | chronoroute | 3.0245349839999998 | ls_d | 1.6750626075833335 | Not significant after adjustment; paired difference 1.35 [0.493, 2.21] (Student-t 95%), Holm-adjusted p 0.255 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 93, 659 |
| C2 / arq-0.1-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 120, 500 |
| C2 / arq-0.1-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 120, 554 |
| C2 / arq-0.1-middle | event_mean | chronoroute | 0.3888888888888889 | lq | 1.0 | Registered material loss; paired difference -0.611 [-0.778, -0.444] (Student-t 95%), Holm-adjusted p 0.00000143 | 120, 608 |
| C2 / arq-0.1-middle | restricted_mean_s | chronoroute | 2.7202257176111115 | lq | 0.2955811941944444 | Registered material loss; paired difference 2.42 [1.87, 2.98] (Student-t 95%), Holm-adjusted p 2.23e-8 | 120, 608 |
| C2 / arq-0.1-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.003785116505191513 | Point disadvantage; no registered paired contrast for this metric | 120, 608 |
| C2 / arq-0.1-middle | event_mean | chronoroute | 0.3888888888888889 | ls_d | 0.6666666666666666 | Not significant after adjustment; paired difference -0.278 [-0.486, -0.0698] (Student-t 95%), Holm-adjusted p 0.866 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 120, 662 |
| C2 / arq-0.1-middle | restricted_mean_s | chronoroute | 2.7202257176111115 | ls_d | 1.5191717230555555 | Not significant after adjustment; paired difference 1.2 [0.437, 1.96] (Student-t 95%), Holm-adjusted p 0.256 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 120, 662 |
| C2 / arq-0.3-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 147, 503 |
| C2 / arq-0.3-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 147, 557 |
| C2 / arq-0.3-middle | event_mean | chronoroute | 0.6388888888888888 | lq | 1.0 | Registered material loss; paired difference -0.361 [-0.526, -0.196] (Student-t 95%), Holm-adjusted p 0.00831 | 147, 611 |
| C2 / arq-0.3-middle | restricted_mean_s | chronoroute | 1.7659008346944445 | lq | 0.04863549058333332 | Registered material loss; paired difference 1.72 [1.11, 2.32] (Student-t 95%), Holm-adjusted p 0.000177 | 147, 611 |
| C2 / arq-0.3-middle | event_mean | chronoroute | 0.6388888888888888 | ls_d | 0.6666666666666666 | Not significant after adjustment; paired difference -0.0278 [-0.217, 0.162] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 147, 665 |
| C2 / arq-0.3-middle | restricted_mean_s | chronoroute | 1.7659008346944445 | ls_d | 1.5191717230555555 | Not significant after adjustment; paired difference 0.247 [-0.455, 0.948] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 147, 665 |
| C2 / delay-0-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 174, 506 |
| C2 / delay-0-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 174, 560 |
| C2 / delay-0-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 174, 614 |
| C2 / delay-1000000-first | fpr | chronoroute | 0.0040969377126588435 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 201, 509 |
| C2 / delay-1000000-first | fpr | chronoroute | 0.0040969377126588435 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 201, 563 |
| C2 / delay-1000000-first | fpr | chronoroute | 0.0040969377126588435 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 201, 617 |
| C2 / delay-1000000-first | event_mean | chronoroute | 0.0 | ls_d | 0.4722222222222222 | Registered material loss; paired difference -0.472 [-0.644, -0.301] (Student-t 95%), Holm-adjusted p 0.000311 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 201, 671 |
| C2 / delay-1000000-first | restricted_mean_s | chronoroute | 4.0 | ls_d | 2.1947984257499997 | Registered material loss; paired difference 1.81 [1.15, 2.46] (Student-t 95%), Holm-adjusted p 0.000295 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 201, 671 |
| C2 / delay-1000000-last | fpr | chronoroute | 0.004132087919719435 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 228, 512 |
| C2 / delay-1000000-last | fpr | chronoroute | 0.004132087919719435 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 228, 566 |
| C2 / delay-1000000-last | fpr | chronoroute | 0.004132087919719435 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 228, 620 |
| C2 / delay-1000000-last | event_mean | chronoroute | 0.0 | ls_d | 0.5555555555555556 | Registered material loss; paired difference -0.556 [-0.726, -0.385] (Student-t 95%), Holm-adjusted p 0.0000147 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 228, 674 |
| C2 / delay-1000000-last | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.8413703749166668 | Registered material loss; paired difference 2.16 [1.5, 2.82] (Student-t 95%), Holm-adjusted p 0.0000142 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 228, 674 |
| C2 / delay-1000000-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 255, 515 |
| C2 / delay-1000000-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 255, 569 |
| C2 / delay-1000000-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 255, 623 |
| C2 / delay-1000000-middle | event_mean | chronoroute | 0.0 | ls_d | 0.6388888888888888 | Registered material loss; paired difference -0.639 [-0.804, -0.474] (Student-t 95%), Holm-adjusted p 4.04e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 255, 677 |
| C2 / delay-1000000-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.5438155981944446 | Registered material loss; paired difference 2.46 [1.82, 3.09] (Student-t 95%), Holm-adjusted p 4.15e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 255, 677 |
| C2 / delay-250000-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 282, 518 |
| C2 / delay-250000-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 282, 572 |
| C2 / delay-250000-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 282, 626 |
| C2 / delay-250000-middle | event_mean | chronoroute | 0.0 | ls_d | 0.6388888888888888 | Registered material loss; paired difference -0.639 [-0.804, -0.474] (Student-t 95%), Holm-adjusted p 4.04e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 282, 680 |
| C2 / delay-250000-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.5438155981944446 | Registered material loss; paired difference 2.46 [1.82, 3.09] (Student-t 95%), Holm-adjusted p 4.15e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 282, 680 |
| C2 / delay-4000000-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 309, 521 |
| C2 / delay-4000000-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 309, 575 |
| C2 / delay-4000000-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 309, 629 |
| C2 / delay-4000000-middle | event_mean | chronoroute | 0.0 | ls_d | 0.6388888888888888 | Registered material loss; paired difference -0.639 [-0.804, -0.474] (Student-t 95%), Holm-adjusted p 4.04e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 309, 683 |
| C2 / delay-4000000-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.5438155981944446 | Registered material loss; paired difference 2.46 [1.82, 3.09] (Student-t 95%), Holm-adjusted p 4.15e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 309, 683 |
| C2 / rate-0.2-middle | fpr | chronoroute | 0.004112214671545565 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 336, 524 |
| C2 / rate-0.2-middle | fpr | chronoroute | 0.004112214671545565 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 336, 578 |
| C2 / rate-0.2-middle | restricted_mean_s | chronoroute | 0.06522071149999999 | lq | 0.01697295752777778 | Significant after adjustment, below the material threshold; paired difference 0.0482 [0.0439, 0.0526] (Student-t 95%), Holm-adjusted p < numerical resolution | 336, 632 |
| C2 / rate-0.2-middle | fpr | chronoroute | 0.004112214671545565 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 336, 632 |
| C2 / rate-0.3-first | fpr | chronoroute | 0.004097649060666042 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 363, 527 |
| C2 / rate-0.3-first | fpr | chronoroute | 0.004097649060666042 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 363, 581 |
| C2 / rate-0.3-first | event_mean | chronoroute | 0.0 | lq | 1.0 | Registered material loss; paired difference -1 [-1, -1] (Student-t 95%), Holm-adjusted p < numerical resolution (zero-variance branch) | 363, 635 |
| C2 / rate-0.3-first | restricted_mean_s | chronoroute | 4.0 | lq | 1.3692997175555557 | Registered material loss; paired difference 2.63 [2.62, 2.64] (Student-t 95%), Holm-adjusted p < numerical resolution | 363, 635 |
| C2 / rate-0.3-first | fpr | chronoroute | 0.004097649060666042 | lq | 3.472463365511494e-05 | Point disadvantage; no registered paired contrast for this metric | 363, 635 |
| C2 / rate-0.3-first | event_mean | chronoroute | 0.0 | ls_d | 0.6111111111111112 | Registered material loss; paired difference -0.611 [-0.778, -0.444] (Student-t 95%), Holm-adjusted p 0.00000143 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 363, 689 |
| C2 / rate-0.3-first | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.989890764666667 | Registered material loss; paired difference 2.01 [1.39, 2.63] (Student-t 95%), Holm-adjusted p 0.0000148 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 363, 689 |
| C2 / rate-0.3-last | fpr | chronoroute | 0.00413280544557894 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 390, 530 |
| C2 / rate-0.3-last | fpr | chronoroute | 0.00413280544557894 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 390, 584 |
| C2 / rate-0.3-last | event_mean | chronoroute | 0.0 | lq | 1.0 | Registered material loss; paired difference -1 [-1, -1] (Student-t 95%), Holm-adjusted p < numerical resolution (zero-variance branch) | 390, 638 |
| C2 / rate-0.3-last | restricted_mean_s | chronoroute | 4.0 | lq | 1.367952981777778 | Registered material loss; paired difference 2.63 [2.62, 2.64] (Student-t 95%), Holm-adjusted p < numerical resolution | 390, 638 |
| C2 / rate-0.3-last | fpr | chronoroute | 0.00413280544557894 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 390, 638 |
| C2 / rate-0.3-last | event_mean | chronoroute | 0.0 | ls_d | 0.6666666666666666 | Registered material loss; paired difference -0.667 [-0.828, -0.505] (Student-t 95%), Holm-adjusted p 9.84e-8 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 390, 692 |
| C2 / rate-0.3-last | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.6156081411666667 | Registered material loss; paired difference 2.38 [1.77, 3] (Student-t 95%), Holm-adjusted p 4.15e-7 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 390, 692 |
| C2 / rate-0.3-middle | fpr | chronoroute | 0.00413280544557894 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 417, 533 |
| C2 / rate-0.3-middle | fpr | chronoroute | 0.00413280544557894 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 417, 587 |
| C2 / rate-0.3-middle | event_mean | chronoroute | 0.0 | lq | 1.0 | Registered material loss; paired difference -1 [-1, -1] (Student-t 95%), Holm-adjusted p < numerical resolution (zero-variance branch) | 417, 641 |
| C2 / rate-0.3-middle | restricted_mean_s | chronoroute | 4.0 | lq | 1.3675388035 | Registered material loss; paired difference 2.63 [2.62, 2.64] (Student-t 95%), Holm-adjusted p < numerical resolution | 417, 641 |
| C2 / rate-0.3-middle | fpr | chronoroute | 0.00413280544557894 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 417, 641 |
| C2 / rate-0.3-middle | event_mean | chronoroute | 0.0 | ls_d | 0.6111111111111112 | Registered material loss; paired difference -0.611 [-0.778, -0.444] (Student-t 95%), Holm-adjusted p 0.00000143 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 417, 695 |
| C2 / rate-0.3-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.6560312404166668 | Registered material loss; paired difference 2.34 [1.7, 2.99] (Student-t 95%), Holm-adjusted p 0.00000144 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 417, 695 |
| C2 / rate-0.6-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 444, 536 |
| C2 / rate-0.6-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 444, 590 |
| C2 / rate-0.6-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 444, 644 |
| C2 / rate-1-middle | fpr | chronoroute | 0.004132374900163212 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 471, 539 |
| C2 / rate-1-middle | fpr | chronoroute | 0.004132374900163212 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 471, 593 |
| C2 / rate-1-middle | fpr | chronoroute | 0.004132374900163212 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 471, 647 |
| C5 / arq-0-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 822, 1298 |
| C5 / arq-0-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 822, 1352 |
| C5 / arq-0-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 822, 1406 |
| C5 / arq-0.01-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 849, 1301 |
| C5 / arq-0.01-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 849, 1355 |
| C5 / arq-0.01-middle | event_mean | chronoroute | 0.19444444444444445 | lq | 0.6111111111111112 | Registered material loss; paired difference -0.417 [-0.586, -0.247] (Student-t 95%), Holm-adjusted p 0.00178 | 849, 1409 |
| C5 / arq-0.01-middle | restricted_mean_s | chronoroute | 3.4906426665 | lq | 2.376601897611111 | Registered material loss; paired difference 1.11 [0.621, 1.61] (Student-t 95%), Holm-adjusted p 0.00538 | 849, 1409 |
| C5 / arq-0.01-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0008333622695232473 | Point disadvantage; no registered paired contrast for this metric | 849, 1409 |
| C5 / arq-0.01-middle | event_mean | chronoroute | 0.19444444444444445 | ls_d | 0.6111111111111112 | Registered material loss; paired difference -0.417 [-0.604, -0.229] (Student-t 95%), Holm-adjusted p 0.00696 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 849, 1463 |
| C5 / arq-0.01-middle | restricted_mean_s | chronoroute | 3.4906426665 | ls_d | 1.8617553129166664 | Registered material loss; paired difference 1.63 [0.965, 2.29] (Student-t 95%), Holm-adjusted p 0.00176 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 849, 1463 |
| C5 / arq-0.1-first | fpr | chronoroute | 0.005833333333333334 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 876, 1304 |
| C5 / arq-0.1-first | fpr | chronoroute | 0.005833333333333334 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 876, 1358 |
| C5 / arq-0.1-first | event_mean | chronoroute | 0.75 | lq | 0.8611111111111112 | Not significant after adjustment; paired difference -0.111 [-0.219, -0.00327] (Student-t 95%), Holm-adjusted p 1 | 876, 1412 |
| C5 / arq-0.1-first | restricted_mean_s | chronoroute | 1.4977735351666668 | lq | 0.9634321836666666 | Not significant after adjustment; paired difference 0.534 [0.171, 0.898] (Student-t 95%), Holm-adjusted p 0.436 | 876, 1412 |
| C5 / arq-0.1-first | fpr | chronoroute | 0.005833333333333334 | lq | 0.0040625 | Point disadvantage; no registered paired contrast for this metric | 876, 1412 |
| C5 / arq-0.1-last | fpr | chronoroute | 0.00583373845405931 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 903, 1307 |
| C5 / arq-0.1-last | fpr | chronoroute | 0.00583373845405931 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 903, 1361 |
| C5 / arq-0.1-last | event_mean | chronoroute | 0.4166666666666667 | lq | 0.8888888888888888 | Registered material loss; paired difference -0.472 [-0.644, -0.301] (Student-t 95%), Holm-adjusted p 0.000311 | 903, 1415 |
| C5 / arq-0.1-last | restricted_mean_s | chronoroute | 2.813100250083333 | lq | 0.7313199386388889 | Registered material loss; paired difference 2.08 [1.52, 2.64] (Student-t 95%), Holm-adjusted p 8.45e-7 | 903, 1415 |
| C5 / arq-0.1-last | fpr | chronoroute | 0.00583373845405931 | lq | 0.002777970692409195 | Point disadvantage; no registered paired contrast for this metric | 903, 1415 |
| C5 / arq-0.1-last | event_mean | chronoroute | 0.4166666666666667 | ls_d | 0.5833333333333334 | Not significant after adjustment; paired difference -0.167 [-0.416, 0.0826] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 903, 1469 |
| C5 / arq-0.1-last | restricted_mean_s | chronoroute | 2.813100250083333 | ls_d | 2.2330287874166665 | Not significant after adjustment; paired difference 0.58 [-0.162, 1.32] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 903, 1469 |
| C5 / arq-0.1-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 930, 1310 |
| C5 / arq-0.1-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 930, 1364 |
| C5 / arq-0.1-middle | event_mean | chronoroute | 0.4166666666666667 | lq | 0.8611111111111112 | Registered material loss; paired difference -0.444 [-0.615, -0.274] (Student-t 95%), Holm-adjusted p 0.000757 | 930, 1418 |
| C5 / arq-0.1-middle | restricted_mean_s | chronoroute | 2.565014645 | lq | 0.7787164580833332 | Registered material loss; paired difference 1.79 [1.19, 2.38] (Student-t 95%), Holm-adjusted p 0.0000718 | 930, 1418 |
| C5 / arq-0.1-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.003819577068648217 | Point disadvantage; no registered paired contrast for this metric | 930, 1418 |
| C5 / arq-0.1-middle | event_mean | chronoroute | 0.4166666666666667 | ls_d | 0.6111111111111112 | Not significant after adjustment; paired difference -0.194 [-0.421, 0.0317] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 930, 1472 |
| C5 / arq-0.1-middle | restricted_mean_s | chronoroute | 2.565014645 | ls_d | 1.8617553129166664 | Not significant after adjustment; paired difference 0.703 [-0.104, 1.51] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 930, 1472 |
| C5 / arq-0.3-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 957, 1313 |
| C5 / arq-0.3-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 957, 1367 |
| C5 / arq-0.3-middle | event_mean | chronoroute | 0.6388888888888888 | lq | 1.0 | Registered material loss; paired difference -0.361 [-0.526, -0.196] (Student-t 95%), Holm-adjusted p 0.00831 | 957, 1421 |
| C5 / arq-0.3-middle | restricted_mean_s | chronoroute | 1.7488938587777776 | lq | 0.06890149611111111 | Registered material loss; paired difference 1.68 [1.08, 2.28] (Student-t 95%), Holm-adjusted p 0.000246 | 957, 1421 |
| C5 / delay-0-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 984, 1316 |
| C5 / delay-0-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 984, 1370 |
| C5 / delay-0-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 984, 1424 |
| C5 / delay-1000000-first | fpr | chronoroute | 0.005833333333333334 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1011, 1319 |
| C5 / delay-1000000-first | fpr | chronoroute | 0.005833333333333334 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1011, 1373 |
| C5 / delay-1000000-first | fpr | chronoroute | 0.005833333333333334 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1011, 1427 |
| C5 / delay-1000000-first | event_mean | chronoroute | 0.08333333333333333 | ls_d | 0.5555555555555556 | Registered material loss; paired difference -0.472 [-0.678, -0.266] (Student-t 95%), Holm-adjusted p 0.00459 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1011, 1481 |
| C5 / delay-1000000-first | restricted_mean_s | chronoroute | 3.8123646889166665 | ls_d | 1.8338578987222225 | Registered material loss; paired difference 1.98 [1.24, 2.72] (Student-t 95%), Holm-adjusted p 0.000476 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1011, 1481 |
| C5 / delay-1000000-last | fpr | chronoroute | 0.00583373845405931 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1038, 1322 |
| C5 / delay-1000000-last | fpr | chronoroute | 0.00583373845405931 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1038, 1376 |
| C5 / delay-1000000-last | fpr | chronoroute | 0.00583373845405931 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1038, 1430 |
| C5 / delay-1000000-last | event_mean | chronoroute | 0.0 | ls_d | 0.4444444444444444 | Registered material loss; paired difference -0.444 [-0.615, -0.274] (Student-t 95%), Holm-adjusted p 0.000757 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1038, 1484 |
| C5 / delay-1000000-last | restricted_mean_s | chronoroute | 4.0 | ls_d | 2.5420230462499998 | Registered material loss; paired difference 1.46 [0.833, 2.08] (Student-t 95%), Holm-adjusted p 0.00354 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1038, 1484 |
| C5 / delay-1000000-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1065, 1325 |
| C5 / delay-1000000-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1065, 1379 |
| C5 / delay-1000000-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1065, 1433 |
| C5 / delay-1000000-middle | event_mean | chronoroute | 0.0 | ls_d | 0.5555555555555556 | Registered material loss; paired difference -0.556 [-0.726, -0.385] (Student-t 95%), Holm-adjusted p 0.0000147 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1065, 1487 |
| C5 / delay-1000000-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.9280966396388892 | Registered material loss; paired difference 2.07 [1.41, 2.73] (Student-t 95%), Holm-adjusted p 0.0000264 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1065, 1487 |
| C5 / delay-250000-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1092, 1328 |
| C5 / delay-250000-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1092, 1382 |
| C5 / delay-250000-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1092, 1436 |
| C5 / delay-250000-middle | event_mean | chronoroute | 0.0 | ls_d | 0.5555555555555556 | Registered material loss; paired difference -0.556 [-0.726, -0.385] (Student-t 95%), Holm-adjusted p 0.0000147 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1092, 1490 |
| C5 / delay-250000-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.9280966396388892 | Registered material loss; paired difference 2.07 [1.41, 2.73] (Student-t 95%), Holm-adjusted p 0.0000264 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1092, 1490 |
| C5 / delay-4000000-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1119, 1331 |
| C5 / delay-4000000-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1119, 1385 |
| C5 / delay-4000000-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1119, 1439 |
| C5 / delay-4000000-middle | event_mean | chronoroute | 0.05555555555555555 | ls_d | 0.5277777777777778 | Registered material loss; paired difference -0.472 [-0.644, -0.301] (Student-t 95%), Holm-adjusted p 0.000311 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1119, 1493 |
| C5 / delay-4000000-middle | restricted_mean_s | chronoroute | 3.8067143592777777 | ls_d | 1.9536741829444446 | Registered material loss; paired difference 1.85 [1.19, 2.51] (Student-t 95%), Holm-adjusted p 0.000207 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1119, 1493 |
| C5 / rate-0.2-middle | fpr | chronoroute | 0.005819423633132383 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1146, 1334 |
| C5 / rate-0.2-middle | fpr | chronoroute | 0.005819423633132383 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1146, 1388 |
| C5 / rate-0.2-middle | restricted_mean_s | chronoroute | 0.0735032322777778 | lq | 0.01668116588888889 | Significant after adjustment, below the material threshold; paired difference 0.0568 [0.0499, 0.0638] (Student-t 95%), Holm-adjusted p < numerical resolution | 1146, 1442 |
| C5 / rate-0.2-middle | fpr | chronoroute | 0.005819423633132383 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1146, 1442 |
| C5 / rate-0.3-first | fpr | chronoroute | 0.005834548864346739 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1173, 1337 |
| C5 / rate-0.3-first | fpr | chronoroute | 0.005834548864346739 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1173, 1391 |
| C5 / rate-0.3-first | event_mean | chronoroute | 0.25 | lq | 1.0 | Registered material loss; paired difference -0.75 [-0.899, -0.601] (Student-t 95%), Holm-adjusted p 6.2e-10 | 1173, 1445 |
| C5 / rate-0.3-first | restricted_mean_s | chronoroute | 3.7816269789166665 | lq | 1.323009700611111 | Registered material loss; paired difference 2.46 [2.26, 2.65] (Student-t 95%), Holm-adjusted p < numerical resolution | 1173, 1445 |
| C5 / rate-0.3-first | fpr | chronoroute | 0.005834548864346739 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1173, 1445 |
| C5 / rate-0.3-first | event_mean | chronoroute | 0.25 | ls_d | 0.5555555555555556 | Not significant after adjustment; paired difference -0.306 [-0.559, -0.0521] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1173, 1499 |
| C5 / rate-0.3-first | restricted_mean_s | chronoroute | 3.7816269789166665 | ls_d | 1.9717415341944442 | Registered material loss; paired difference 1.81 [1.09, 2.53] (Student-t 95%), Holm-adjusted p 0.00115 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1173, 1499 |
| C5 / rate-0.3-last | fpr | chronoroute | 0.005834346240666782 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1200, 1340 |
| C5 / rate-0.3-last | fpr | chronoroute | 0.005834346240666782 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1200, 1394 |
| C5 / rate-0.3-last | event_mean | chronoroute | 0.0 | lq | 1.0 | Registered material loss; paired difference -1 [-1, -1] (Student-t 95%), Holm-adjusted p < numerical resolution (zero-variance branch) | 1200, 1448 |
| C5 / rate-0.3-last | restricted_mean_s | chronoroute | 4.0 | lq | 1.3232125755277775 | Registered material loss; paired difference 2.68 [2.65, 2.7] (Student-t 95%), Holm-adjusted p < numerical resolution | 1200, 1448 |
| C5 / rate-0.3-last | fpr | chronoroute | 0.005834346240666782 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1200, 1448 |
| C5 / rate-0.3-last | event_mean | chronoroute | 0.0 | ls_d | 0.5555555555555556 | Registered material loss; paired difference -0.556 [-0.726, -0.385] (Student-t 95%), Holm-adjusted p 0.0000147 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1200, 1502 |
| C5 / rate-0.3-last | restricted_mean_s | chronoroute | 4.0 | ls_d | 2.279200896888889 | Registered material loss; paired difference 1.72 [1.13, 2.31] (Student-t 95%), Holm-adjusted p 0.000122 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1200, 1502 |
| C5 / rate-0.3-middle | fpr | chronoroute | 0.0058688706764828446 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1227, 1343 |
| C5 / rate-0.3-middle | fpr | chronoroute | 0.0058688706764828446 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1227, 1397 |
| C5 / rate-0.3-middle | event_mean | chronoroute | 0.0 | lq | 1.0 | Registered material loss; paired difference -1 [-1, -1] (Student-t 95%), Holm-adjusted p < numerical resolution (zero-variance branch) | 1227, 1451 |
| C5 / rate-0.3-middle | restricted_mean_s | chronoroute | 4.0 | lq | 1.3231592685555558 | Registered material loss; paired difference 2.68 [2.65, 2.7] (Student-t 95%), Holm-adjusted p < numerical resolution | 1227, 1451 |
| C5 / rate-0.3-middle | fpr | chronoroute | 0.0058688706764828446 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1227, 1451 |
| C5 / rate-0.3-middle | event_mean | chronoroute | 0.0 | ls_d | 0.6111111111111112 | Registered material loss; paired difference -0.611 [-0.778, -0.444] (Student-t 95%), Holm-adjusted p 0.00000143 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1227, 1505 |
| C5 / rate-0.3-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 1.8599002083611111 | Registered material loss; paired difference 2.14 [1.5, 2.78] (Student-t 95%), Holm-adjusted p 0.00000764 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1227, 1505 |
| C5 / rate-0.6-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1254, 1346 |
| C5 / rate-0.6-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1254, 1400 |
| C5 / rate-0.6-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1254, 1454 |
| C5 / rate-0.6-middle | event_mean | chronoroute | 0.0 | ls_d | 0.027777777777777776 | Not significant after adjustment; paired difference -0.0278 [-0.0842, 0.0286] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1254, 1508 |
| C5 / rate-0.6-middle | restricted_mean_s | chronoroute | 4.0 | ls_d | 3.9755237781111115 | Not significant after adjustment; paired difference 0.0245 [-0.0252, 0.0742] (Student-t 95%), Holm-adjusted p 1 LS-D comparator flaw: it reads the injected delay without noise and advertises any change, so this is not valid evidence about slowness detection. | 1254, 1508 |
| C5 / rate-1-middle | fpr | chronoroute | 0.005868259314559534 | crh16 | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1281, 1349 |
| C5 / rate-1-middle | fpr | chronoroute | 0.005868259314559534 | e2e | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1281, 1403 |
| C5 / rate-1-middle | fpr | chronoroute | 0.005868259314559534 | lq | 0.0 | Point disadvantage; no registered paired contrast for this metric | 1281, 1457 |
Limitations
- Detection is counterfactual with common random numbers; loss before a checker can prevent an alarm.
- Selected rules are frozen from the pilot. Other rule and event-setting sweeps remain available in the table.
- Chart intervals are marginal per-cell intervals. The verdict quotes the registered paired comparisons from paired_contrasts.csv: paired Student-t 95% intervals (primary) with Holm-adjusted p-values. A binary endpoint that is 0 or 1 in every run gives a zero-width interval.
- The rate-cut fault at factor 0.3 is the registered nominal magnitude. With fluid background traffic at 0.3 of nominal capacity, that cut puts the link at exactly full load, so the queue stays short and fits inside the S = 1.5 budgets.
- S = 1.5 gives each hop a deadline at 1.5 times the planned elapsed time, which leaves several milliseconds of spare time mid-path; 0.25 to 4 ms of added delay fits inside it. Looser rules detect more but exceed the registered 1% false-positive limit (rows with selected = False).
- LS-D is not a valid comparator for slowness detection in this release: its measurement is the queue estimate plus the exact injected fault delay plus the expected ARQ delay computed from the configured loss probability, with no measurement noise. With the registered 1 s update interval it re-advertises any nonzero change, so its 20% threshold never matters, and a run counts as detected unless a twin advert coincides. Its numbers stay as registered but are not evidence either way.
- LS-H and CRH-16 have no slowness detector, so ChronoRoute wins against them hold by construction. E2E runs at a clean false-positive rate of 0, so comparisons with it are not at matched false-positive rates. LQ uses only local queue and ARQ state; its advantages are robust.
- Pair 0 has only 2 ISLs, so its middle and last fault positions are the same link.
- O-Q is a privileged oracle. The local queue monitor is a practical comparator.
These pages use only summary.csv and manifest.json, plus paired_contrasts.csv for the verdict and the paired comparisons quoted above; each file is checked against the manifest SHA-256. They do not infer missing comparisons from figures, reports, raw traces or other result files.
How this was made
- Recorded producer
- E01 confirmatory campaign (1.0.0.dev0)
- Manifest commit
- b896bad3bbc13bd1f715d69ca1e8fc7e2b2c477d; dirty: false
- Run
- 530515825cdf3e1b
- Time
- 2026-10-11T15:49:42.730920Z to 2026-10-11T16:56:48.288356Z; 4025.557391 seconds of recorded host wall time (REAL)
- Snapshot
- 20261010T1358Z-celestrak; SHA-256: c37e70812155c2d1d9eeb3e96681303c00c9fca91287939b21bf5313ce83aecd
- Seeds
- 1000, 1001, 1002, 1003, 1004, 1005, 1006, 1007, 1008, 1009, 1010, 1011, 1012, 1013, 1014, 1015, 1016, 1017, 1018, 1019, 1020, 1021, 1022, 1023, 1024, 1025, 1026, 1027, 1028, 1029, 1030, 1031, 1032, 1033, 1034, 1035
- Command
venv/bin/python -B experiments/E01/confirm.py --pilot results/E01/20261011T113605Z-pilot --prerequisites results/E01/20261011T130817Z-calibration --e0 results/E00/20261011T100448Z --out results/E01/20261011T154940Z-confirmatory-resumed --workers 8 --resume-from results/E01/20261011T131015Z-confirmatory
- Config SHA-256
12850bacd4b79a5202deacfd6a2331098deab85b5772a6d74549bad8d6e8db9f- Labels
- fault_slow_link: MODELLED
- isl_topology_grid4: MODELLED
- orbital_elements: REAL
- routes_and_budgets: DERIVED
- traffic_probes: MODELLED
C2 override: C2 elements and positions MODELLED
- Manifest
chronoroute.manifest/1; 5271 recorded output files (85422193 bytes); 9 recorded input hashes. Download the full manifest (991950 bytes, SHA-256
80158cb7b3106183fe646caadc5f41d1667817823e12a4b29890975d78be82b9) for the configuration, versions, host and every input and output hash.summary.csv: 1399049 bytes, SHA-256bee06647b9f70f64c12ef2b74f918b6ebeb9a8a5f0e5dc91c0f251ec37aaafc2paired_contrasts.csv: 45577 bytes, SHA-256c44cf51fcbd9985204222fba0a7b419c2418ce3187bffd46f37b2ebfa639b6d5- Versions: chronoroute 1.0.0.dev0; matplotlib 3.11.1; networkx 3.6.1; numpy 2.5.3; python 3.14.4; sgp4 2.27
- Host (REAL): cpu_model: 13th Gen Intel(R) Core(TM) i9-13900H; logical_cores: 20; machine: x86_64; platform: Linux-6.18.33.2-microsoft-standard-WSL2-x86_64-with-glibc2.43; ram_bytes: 8131358720
For collected results, the manifest commit identifies the collector. Input hashes preserve the producer chain in the downloaded manifest; this page does not load those additional files.