OverviewFilesCommitsIssuesDocsReleasesPeers

basal-agent commits / 43b15f0

basal-agent: DOWNGRADE chemoton C3 — it fails at matched realized size
C3 was scored PARTIAL by comparing at matched DIVISION THRESHOLD, with a caveat that realized mean size is ~half the threshold and that this "plausibly accounts for the shift". The caveat was wrong AND pointed the wrong way. ribozyme.c holds compartment size FIXED, so its axis already IS realized size. Comparing like with like: matched size | ribozyme | chemoton fis | diff 4 | 11.64 | 10.83 | -0.81 8 | 11.34 | 6.08 | -5.26 half-fall size (phi falls halfway from max to the 5.99 floor): ribozyme 11.7 | chemoton fission 4.6 | chemoton propagule 6.1 => chemoton's threshold is 2.52x TIGHTER than the decree's. So "three of four points land on ribozyme almost exactly" was an artifact of the matched-threshold comparison; at matched realized size only the size-4 point is close. Growth-coupling yields A corrector, a substantially weaker one, not THE corrector. Mechanistically unsurprising: ribozyme applies between-compartment selection fully every generation (sample parents by mean phi, reseed everything), while chemoton's acts only through slow accumulation of growth advantage against continuous per-unit death. WHAT SURVIVES: C2 (phi 5.99 -> 11.68 with no group-selection step anywhere) and C4 (phi-blind control flat in every row). The EMERGENCE result stands — compartment-level differential reproduction does not have to be supplied. Only the quantitative signature claim dies. Consistent with Test A, which survived its own matched-size check: propagule's half-fall (6.1) exceeds fission's (4.6), so the bottleneck does widen the working range — still well short of 11.7. Third correction of the day from a matched-size check: Test A survived, Test B's T4 was retracted, chemoton's C3 is downgraded. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
43b15f06410ac67875b77ac45b41190997ed351e
by plankatron <fleeky@gmail.com> · 2026-08-01T00:16:18.000Z

Changed files (2) +55 -8

SYNTHESIS.md +14 -8
……@@ -694,12 +694,15 @@ the control, with uptake constant per unit and blind to φ, stays at or below ne
694694division-plus-assortment alone does nothing, and the growth→catalysis coupling does the work.
695695**C2 passed** — φ 5.99 → 11.68.
696696
697**C3, the key test, is PARTIAL by its own pre-registered bar.** Three of four points land on the
698decreed version almost exactly (11.68/11.64, 11.45/11.34, 5.86/5.73), but division size 16 gives
699**7.65** where the decree gave 5.90 — a soft shoulder rather than a cliff, scoring small 2/2, large
7001/2. Caveat offered as a caveat, not a rescue: compartments here die back continuously, so realized
701mean size is roughly *half* the threshold (1.87 … 12.25), which puts "division size 16" at ~6-unit
702compartments, inside ribozyme's working range. The verdict stays PARTIAL.
697**C3, the key test — scored PARTIAL, then DOWNGRADED to failed.** At matched division *threshold*
698three of four points appeared to land on the decreed version almost exactly (11.68/11.64,
69911.45/11.34, 5.86/5.73). That comparison was wrong: `ribozyme.c` holds compartment size fixed, so its
700axis already IS realized size. At **matched realized size** chemoton gives 10.83 vs 11.64 at size 4
701but **6.08 vs 11.34 at size 8** (−5.26), and its half-fall size is **4.6 against the decree's 11.7 —
7022.52× tighter**. The caveat originally offered (realized size ≈ half the threshold, so the shift is
703explained) pointed the *wrong way*: at matched size chemoton is much worse, not better.
704Mechanistically sensible — ribozyme applies between-compartment selection fully every generation,
705while chemoton's acts only through slow growth advantage against continuous death.
703706
704707**This is the first time in the arc that compartment-level differential reproduction arose instead
705708of being supplied**, and it reproduces the decree's *signature* — crossing at small division sizes,
……@@ -859,8 +862,11 @@ transfer to 1→2 on its own — only Law 2 does.
859862- **Isolation is not compartmentalization** (`byproduct.c`) — supplied total isolation achieves
860863 nothing (5.65) without differential reproduction of the compartments (10.78). ✓
861864- **Compartment-level selection need not be supplied** (`chemoton.c`) — coupling growth to the
862 chemistry's own output generates it with no fitness-proportional sampling anywhere, reproducing the
863 decreed version's size-threshold signature (PARTIAL: small 2/2, large 1/2). ✓
865 chemistry's own output generates it with **no fitness-proportional sampling anywhere** (φ 5.99 →
866 11.68; φ-blind control flat in every row). ✓ **But it does not reproduce the decreed version
867 quantitatively**: at matched *realized* size chemoton's threshold is **2.52× tighter** (half-fall
868 size 4.6 vs ribozyme's 11.7, and −5.26 φ at size 8). C3 downgraded from PARTIAL to **failed at
869 matched size**; growth-coupling yields *a* corrector, a weaker one, not *the* corrector.
864870- ~~Optimal assortment is a trade-off, not a maximum~~ — **RETRACTED same day.** Test B's reversal was
865871 a size artifact: propagule mode shrinks realized compartment size and RAF size scales with it. At
866872 matched realized size the two modes differ by −0.13, and RAF tracks size in 8/8 cells regardless of
chemoton.md +41 -0
……@@ -119,3 +119,44 @@ demographic artifact of the reproduction mode.
119119
120120Caveat: the printed small/large crossing counts have wrong denominators after the sweep was widened
121121to 9 sizes; the matched-size table is the actual test and is unaffected.
122
123---
124
125## ⚠️ C3 DOWNGRADED — it fails at matched realized size
126
127C3 was scored PARTIAL by comparing at matched **division threshold**, with a caveat that realized
128mean size is ~half the threshold and that this "plausibly accounts for the shift". **The caveat was
129wrong, and it pointed the wrong way.** `ribozyme.c` holds compartment size *fixed*, so its size axis
130already IS realized size. Comparing like with like:
131
132| matched realized size | ribozyme (decreed) | chemoton fission | diff | chemoton propagule | diff |
133|---|---|---|---|---|---|
134| 4 | 11.64 | 10.83 | −0.81 | 11.36 | −0.28 |
135| 8 | 11.34 | **6.08** | **−5.26** | (out of range) | |
136
137**Half-fall size** — the compartment size at which φ falls halfway from its maximum to the 5.99
138neutral floor:
139
140| | half-fall size |
141|---|---|
142| ribozyme (decreed selection) | **11.7** |
143| chemoton, fission | **4.6** |
144| chemoton, propagule | 6.1 |
145
146**Chemoton's threshold sits 2.52× tighter than the decree's.** So growth-coupling produces *a*
147corrector, but not *the same* corrector — a substantially weaker one. The claim that "three of four
148points land on ribozyme almost exactly" was an artifact of comparing at matched division threshold;
149at matched realized size only the size-4 point is close.
150
151Mechanistically this is unsurprising in hindsight: `ribozyme.c` applies between-compartment selection
152directly and completely every generation (sample parents ∝ mean φ, reseed everything), whereas
153chemoton's selection acts only through slow accumulation of growth advantage against continuous
154per-unit death. The selection differential per generation is far smaller.
155
156> **What survives:** C2 (φ rises from 5.99 to 11.68 with no group-selection step) and C4 (the
157> φ-blind control flat in every row). The **emergence** result stands — compartment-level
158> differential reproduction does not have to be supplied. What does **not** survive is the claim that
159> it quantitatively reproduces the decreed version's signature.
160
161Consistent with Test A: propagule's half-fall (6.1) exceeds fission's (4.6), so the bottleneck does
162widen the working range — still well short of the decree's 11.7.