Check 3 of the pilot review, read as a fluidity question: where natural order sits, and whether it lasts (run 2026-10-14 16:01 Halifax).
After the fact and exploratory: a description of the pilot's saved results, not a reading fixed in plan.md and not a new experiment. No model was run.
A toy model: structure inside the model, not evidence about cells. Histories, not memory: differences are between intervention histories and the states they leave.
Read: the 12 untouched base runs (each matching the hash in pilot_runlog.json) and the pilot's 120 moment files (each matching the hash in pilot_results.json). Reproduced before reading on: untouched states at R from the base runs against the report's descriptions (coherence, slower half, faster half) (largest difference 3.3e-16); coherence from the saved states against the base runs' record at T and R (largest difference 1.4e-17); the moment files' untouched coherence against the base runs (largest difference 0.0e+00); the three contributions adding up to the group's coherence (largest difference 2.1e-17); the report's state descriptions at R (coherence, slower half, faster half; seven states) (largest difference 0.0e+00); the report's persistence over (R + 200, R + 500], medium and push (largest difference 0.0e+00).
Natural states: the untouched run at T and at R of every moment, 20 per start, both groups (480 group states). Moments with both pushes built for the routes: 119. Error bars and 'clear' (more than 3 bars) as in the report.

Answers
  1. Where natural order sits: more in the slower half than an even spread.
  2. The same relationship across routes: like the medium.
  3. Which order lasts: the slower half's order lasts longer.

1. Where natural order sits: the share of the extra natural coherence each part supplies (exact; they add up to 1)
  slower half: +0.810 +- 0.029 (+28.2 bars; 12 of 12 starts above zero)
  faster half: +0.181 +- 0.033 (+5.5 bars; 11 of 12 starts above zero)
  free turners: +0.008 +- 0.007 (+1.2 bars; 8 of 12 starts above zero)
  slower half minus the even-spread share (0.484): +0.326 +- 0.029 (+11.4 bars; 12 of 12 starts above zero)
  states at R alone: slower half +0.842 +- 0.038; faster half +0.142 +- 0.047; free turners +0.015 +- 0.014
  states at T alone: slower half +0.766 +- 0.039; faster half +0.233 +- 0.039; free turners +0.001 +- 0.007
  how a group's coherence is made up on average: slower half +0.853 +- 0.014; faster half +0.138 +- 0.013; free turners +0.009 +- 0.005
  average R_g 0.0251; its spread within starts (SD) 0.0128; random phases alone would give about 0.0140 for a group and 0.0201 for a half

2. The same relationship across routes: the slower half's share of the extra, from each half's own coherence
  natural: 0.928 +- 0.028; medium: 0.994 +- 0.005; push: 0.714 +- 0.024
  natural minus medium: -0.066 +- 0.027 (-2.5 bars); natural minus push: +0.213 +- 0.020 (+10.5 bars); medium minus push: +0.279 +- 0.022 (+12.5 bars)
  natural from states at R alone: 0.978 +- 0.025; at T alone: 0.870 +- 0.041
  per unit of extra group coherence (slower half, faster half):
    natural slopes: +1.526 +- 0.061, +0.119 +- 0.048
    medium: +2.150 +- 0.029, +0.013 +- 0.012 (group change +0.0145)
    push: +0.981 +- 0.008, +0.392 +- 0.045 (group change +0.0145)
  against untouched, separately (slower half's share; change in the slower half, in the faster half):
    medium raised: 0.994 +- 0.006; +0.0438 +- 0.0027 (+16.2 bars; 12 of 12 starts above zero); +0.0003 +- 0.0003 (+0.8 bars; 6 of 12 starts above zero)
    medium lowered: 0.993 +- 0.005; +0.0187 +- 0.0014 (+13.0 bars; 12 of 12 starts above zero); +0.0001 +- 0.0001 (+1.2 bars; 8 of 12 starts above zero)
    push raised: 0.589 +- 0.010; +0.0197 +- 0.0013 (+15.4 bars; 12 of 12 starts above zero); +0.0138 +- 0.0010 (+13.9 bars; 12 of 12 starts above zero)
    push lowered: 1.367 +- 0.136; +0.0088 +- 0.0007 (+12.4 bars; 12 of 12 starts above zero); -0.0024 +- 0.0008 (-3.1 bars; 2 of 12 starts above zero)
  medium over all 120 moments: 0.994 +- 0.005

3. Which order lasts: how much of each part's contribution at t is still there in the group's coherence later (coefficients, the three parts together)
  (t+0, t+100]: slower half +0.380 +- 0.030 (+12.8 bars; 12 of 12 starts above zero); faster half +0.056 +- 0.049 (+1.1 bars; 8 of 12 starts above zero); free turners -0.064 +- 0.190 (-0.3 bars; 7 of 12 starts above zero); slower minus faster +0.324 +- 0.039 (+8.3 bars; 12 of 12 starts above zero)
  (t+100, t+200]: slower half +0.260 +- 0.033 (+7.9 bars; 12 of 12 starts above zero); faster half +0.005 +- 0.042 (+0.1 bars; 7 of 12 starts above zero); free turners +0.064 +- 0.171 (+0.4 bars; 6 of 12 starts above zero); slower minus faster +0.255 +- 0.040 (+6.3 bars; 12 of 12 starts above zero)
  (t+200, t+300]: slower half +0.165 +- 0.038 (+4.3 bars; 10 of 12 starts above zero); faster half +0.060 +- 0.081 (+0.7 bars; 5 of 12 starts above zero); free turners +0.113 +- 0.132 (+0.9 bars; 9 of 12 starts above zero); slower minus faster +0.104 +- 0.071 (+1.5 bars; 9 of 12 starts above zero)
  (t+300, t+400]: slower half +0.073 +- 0.042 (+1.8 bars; 8 of 12 starts above zero); faster half +0.026 +- 0.048 (+0.5 bars; 3 of 12 starts above zero); free turners -0.100 +- 0.076 (-1.3 bars; 4 of 12 starts above zero); slower minus faster +0.047 +- 0.051 (+0.9 bars; 7 of 12 starts above zero)
  (t+400, t+500]: slower half +0.027 +- 0.037 (+0.7 bars; 7 of 12 starts above zero); faster half +0.001 +- 0.045 (+0.0 bars; 5 of 12 starts above zero); free turners -0.142 +- 0.114 (-1.2 bars; 7 of 12 starts above zero); slower minus faster +0.026 +- 0.043 (+0.6 bars; 6 of 12 starts above zero)
  blocks where the slower half's contribution is still clearly there: (t+0, t+100], (t+100, t+200], (t+200, t+300]
  share of a coherence excess still there (natural: slope of later coherence on R_g; medium and push: share of their raised-lowered difference at R):
    (0, 100] after t or R: natural +0.319 +- 0.034; medium +0.877 +- 0.014; push +0.676 +- 0.013
    (100, 200] after t or R: natural +0.213 +- 0.030; medium +0.646 +- 0.018; push +0.538 +- 0.031
    (200, 300] after t or R: natural +0.139 +- 0.037; medium +0.468 +- 0.023; push +0.385 +- 0.035
    (300, 400] after t or R: natural +0.059 +- 0.036; medium +0.342 +- 0.020; push +0.287 +- 0.035
    (400, 500] after t or R: natural +0.020 +- 0.034; medium +0.239 +- 0.024; push +0.223 +- 0.033
    (200, 500] after t or R: natural +0.072 +- 0.033; medium +0.350 +- 0.018; push +0.298 +- 0.031 (the report's persistence)

Limits
  The natural states at T and R are 130 to 200 units apart within a start, so they are not independent; the error bars treat the start as the unit.
  Natural slopes compare moments with each other; the routes compare arms at the same moment. They agree in form (extra in each half per unit of extra group coherence), not in design.
  A half's own coherence responds only weakly to small amounts of order near the random-phase floor, the same way for every route. Question 2 shares that limit; questions 1 and 3 use the exact contributions instead.
  This describes where order sits and how long it lasts. It does not test what carries the growth difference.

Compute: 17.0 seconds, in the foreground.