What can a local action write?
The same swap command enters an older body less, and the disturbance travels less far through its future.
A fixed local write loses leverage as development proceeds. Whole-state feedback can pull a mature body toward chosen shapes, although restricted observers find competitive routes and the comparative edge reverses after release. At the same time, a specimen-or-genotype-specific causal passport is already detectable across the first age interval we tested.
The same swap command enters an older body less, and the disturbance travels less far through its future.
Yes while active. But whole-state access has no resolved unique edge over limited observers, replay, and sham.
Across ages, the reaction pattern becomes re-identifiable to the same specimen or genotype beyond a fixed passive summary.
Across nineteen ages, three intervention seeds, and 7,296 treated futures, older bodies admitted less immediate shape change and propagated less downstream displacement. This is not a metaphorical stiffness score; it is the measured response of this local-swap actuator.
Native frames from one illustrative trajectory. They show the body whose intervention physics we sample; the estimates below aggregate 64 organisms in 16 frozen groups.
By age 780 the matching procedure selected a mean dose of 0.540, up from 0.054 at age 60, while realized +4 entry still fell.
Late minus early downstream D4 distance; 95% interval [−0.0108, −0.0068], all 16 group contrasts in the same direction.
Geometric late/early dose ratio; interval [8.52, 10.40], 16/16 groups positive.
Fixed-command D4 change per realized matter L1; factor 0.080, 16/16 groups lower late.
The approximately matched arm is informative because it bends in the same direction, but it is not an exact equal-entry experiment: only 51.5% of its 3,648 records landed within 25% of the target. The post-hoc common-support analysis keeps the late +240 reduction inside the overlap region; it strengthens the impedance reading without retroactively turning the matching into something it was not.
Each cell is an observed group-mean D4 distance, not a schematic. Color is log-scaled so the smallest late-age responses remain visible.
Both mature target-level effects were positive with intervals above zero; averaging across both targets, all sixteen family means were positive. The more surprising result is architectural: once the frozen sham comparator is restored, whole-state access does not separate from the strongest restricted, replay, or sham route.
The first row asks whether whole-state control moves the mature body relative to doing nothing. The next two ask whether it beats every predeclared comparator and gains more leverage with age. The final row was held out until the controller-release comparison.
Whole versus sham; interval [+3.417, +5.027], 16/16 families positive.
Whole versus the per-instance maximum of partition, local, replay, and sham; interval [−0.676, +1.292].
Comparative edge after release; interval [−11.250, −2.798], 12/16 families negative.
Intermediate release points are post-hoc descriptive. The +128 endpoint is the frozen held-out result.
Whole versus partition agrees on the first action 75/192 times; whole versus local agrees 63/192 times. Sequence disagreement shows nonredundant policy inputs, not an information-theoretic quantity.
Source 18, family f37, age 780, expansive target: whole achievement +1.696, corrected active advantage +1.114, release advantage −7.196. Native 128×128 simulator PNGs; selected after aggregate results only to make the branch geometry visible.
The response vector contains 3,504 features across signed native and orthogonal probes, three horizons, scalar summaries, and visible-field structure. Residualization removes only the linear component predicted by 17 predefined passive visible and hidden summaries.
Interval [0.186, 0.271]; 17/18 groups positive.
Frozen 99% identity-permutation bound 9.7%; chance is 1.4%.
q48→q64 interval [0.191, 0.277], all 18 groups positive even as pooled top-1 retrieval crowds.
The q48→q64 pooled top-1 drop is not a loss of individuality: the same-specimen margin remains positive in all eighteen groups. The source-bound split analysis below resolves the apparent collapse as address reorganization and crowding in pooled visible-field space.
Aligned donor composition produces a positive donor-axis projection at q24 and q48. Phase-scrambling the same hidden-energy budget does too, and the paired aligned-minus-scrambled contrast crosses zero at both ages. Hidden composition matters here; its donor arrangement is not identified as the carrier of the passport.
q24 near minus scrambled is −0.025 [−0.054, +0.005]. q48 is +0.023 [−0.009, +0.058].
Left canvas: visible matter, channel 0 + channel 1. Right canvas: channel-0 fraction in blue-to-coral, brightness weighted by visible matter. These are deterministic renders from the frozen f32 state patches, not simulator screenshots.
537f431663a945bf81ddf03dc782c6cd2bde8333334e8ba4022b44d41ba54979This panel opens the completed response surface without running new physics. Every slice was chosen after the primary result was known, so it maps where the signature lives rather than confirming another primary.
An independent source-bound implementation recovered the primary passport and transplant quantities without numerical drift.
q12→q24 residual margin against only the other three members of each frozen quartet; interval [0.143, 0.245], 17/18 groups positive. Within-quartet top-1 is 59.7% against 25% chance.
Visible-field features pass the frozen gate; 384 scalar features alone do not. The scalar margin interval [-0.008, 0.082] crosses zero.
Each bar is the q12→q24 residual same-specimen margin. Isolated t8 and t32 slices do not pass the frozen gate; t128 does.
Native and orthogonal probes, and positive and negative signs, each pass when isolated. Their similar margins point to a distributed reaction structure.
All four anchors retain positive mean margins; A03 alone misses its own permutation gate. The signature is broad, but not spatially uniform.
Across 432 hidden-response features, q12→q24 top-1 is 45.8% and within-quartet top-1 is 87.5%. This still cannot separate specimen identity from genotype because every specimen has a unique genotype.
Pooled visible top-1 falls below its 8.3% null bound, yet the strict within-quartet margin is 0.236, within-quartet top-1 is 56.9%, and hidden top-1 remains 36.1%. This is pooled-visible address reorganization and crowding, not loss of individuality.
Aligned-near minus phase-scrambled donor-axis projection at q24 and q48. Both intervals cross zero. Near-donor host-response distance exceeds self re-encode by 40.23 at q24 and 21.64 at q48, showing disruption without donor-passport specificity.
The body becomes harder to write, remains actively steerable, loses that comparative edge after release, and carries a re-identifiable response signature. Those facts coexist because they describe different axes.
Older bodies admit less macro shape change per command and per realized microscopic write.
It reaches both target families relative to sham while restricted observers find competitive routes.
The held-out comparative advantage reverses after the controller leaves.
Onset is left-censored, and the signature is specimen-or-genotype-specific under this cohort design.
Aligned donor structure does not beat its phase-scrambled energy-matched control.
Hashes establish the exact local artifacts used to build this report. They do not independently validate the science; the nearby labels preserve frozen, corrected, held-out, and post-hoc status.
| Role | Artifact | SHA-256 | Status | Scope |
|---|---|---|---|---|
| Frozen impedance | evidence-source/gard-compositional-causal-emergence-v1/developmental-causal-impedance-v1/results.json | c1ed7f40cfe27c71101bdded5786905e718ba2b12bb48aa6ced7a5b824377b81 | complete_group_bootstrapped_causal_impedance_analysis | 64 organisms / 16 groups / 7,296 treated futures |
| Post-hoc impedance | evidence-source/gard-compositional-causal-emergence-v1/developmental-causal-impedance-v1/exploration.json | c33c34b4e60a742ca3f0b390aacc48c0216a67bf460a47e2c7d654d671375ca7 | post_hoc_source_bound_exploration | matching support, microscopic write access, common-support sensitivity |
| Superseded tournament output | evidence-source/gard-compositional-causal-emergence-v1/developmental-whole-control-tournament-v1/results.json | 25e2802b7a73e4117466b5226e6e52aa3def11140fff866f2ac6fe8bb6a5e9fb | preserved; original joint decision invalid | analysis omitted the predeclared sham comparator |
| Protocol-conformant correction | evidence-source/gard-compositional-causal-emergence-v1/developmental-whole-control-tournament-v1/corrected-results.json | 4b30ab79f8df950d34d506b9727760681bd74d4380a1adefed8fe0ac8605eeeb | complete_source_bound_frozen_analysis_implementation_correction | 16 families / 192 policy-target-age instances / +128 held out |
| Frozen passport + transplant | evidence-source/gard-compositional-causal-emergence-v1/birth-of-causal-individuality-v1/results.json | 1d9ff7e65115ea4ac80a0e2e4d1171625b556e1497401eb067ddea3e8d1e17c3 | complete_causal_passport_and_identity_transplant_analysis | 72 specimens / 18 groups / 7,344 passport + 9,792 transplant branches |
| Post-hoc passport map | evidence-source/gard-compositional-causal-emergence-v1/birth-of-causal-individuality-v1/exploration.json | 30936fc81d2527c7458de1f0095f2fd828badfe820389369fe7ac8f421c1eb88 | post_hoc_source_bound_exploration | source-bound sensitivities only; no new physics |
The complete machine-readable image inventory, hashes, source fields, and rendering rules live beside this report in source-manifest.json. Every PNG is embedded as a data URI; every chart is inline SVG; the q48 composition panels decode the four bound f32 state patches in-page.
The passport analysis was independently reimplemented against the frozen receipts with maximum absolute difference 0. The source-bound exploration repeats that audit result and keeps its additional slices explicitly post-hoc.