Restructure: descriptive tier and experiment names, paper/manuscript

- paper/pnas -> paper/manuscript (venue-neutral)
- configs/layer1 -> configs/inheritance, src/knowledge -> src/inheritance
  (imported as `inheritance`), make layer1 -> make inheritance; layer2 alias dropped
- inheritance and trained-network bundles named after the manuscript figure
  they feed (fig2_grounding_sweep, figS3_rebaselining, ...), or descriptively
  where they feed none; configs keep their `experiment:` value so parquet
  hashes are unchanged, only output.dir moves
- figure scripts, SI figure sources, notebooks, REPRODUCING.md, README and the
  SI Methods/tables updated; make clean no longer deletes tracked manifests;
  reproduce.sh hashes the s{seed}/ layouts too

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y64o8FKP7rCuXzC48pxpMm

paper/arxiv/md2tex.py

@@ -19,11 +19,11 @@ OUT = Path(__file__).resolve().parent / "body.tex"
 
 # Figure references in the text -> (graphics file under figs/, caption).
 FIGURES = {
-    "results/E14/E14.png": ("figs/E14.pdf",
+    "results/figS13_mating_breadth/E14.png": ("figs/E14.pdf",
         "Mating systems (E14): the best mate-pool breadth shrinks as skills get more entangled. "
         "(A) best fitness peaks at intermediate breadth on rugged landscapes; (B) the population mean "
         "is monotonically favoured by promiscuity; (C) diversity is monotonically destroyed by it."),
-    "results/E12/E12.png": ("figs/E12.pdf",
+    "results/fig5_speciation_bdm/E12.png": ("figs/E12.pdf",
         "Model speciation, analytic (E12): hybrid fitness vs divergence traces compatible $\\rightarrow$ "
         "outbreeding depression $\\rightarrow$ inviability; the isolation cliff arrives earlier the "
         "denser the incompatibilities (epistasis), and damage grows super-linearly (the Orr--Turelli "

paper/blueprint.md

@@ -240,7 +240,7 @@ dynamics:
 metrics:
   kl_floor: 1.0e-9
 output:
-  dir: results/E2/
+  dir: results/fig2_grounding_sweep/
   save_per_generation: true
 ```
 

paper/manuscript/build.py

@@ -5,7 +5,7 @@ Adapted from paper/arxiv/md2tex.py (same Markdown subset + pipe tables), with on
 codename-free panels re-plotted from the committed artifacts). Run make_figs.py before building when
 results change.
 
-Usage:  python paper/pnas/build.py && (cd paper/pnas && tectonic main.tex)
+Usage:  python paper/manuscript/build.py && (cd paper/manuscript && tectonic main.tex)
 """
 
 from __future__ import annotations
@@ -27,7 +27,7 @@ from si_figures import SI_FIGURES  # noqa: E402  (supplementary figures, numbere
 
 # figure name -> (single publication PDF from make_figs.py, caption)
 FIGURES: dict[str, tuple[list[str], str]] = {
-    "fig1": (["paper/pnas/figs/fig1a.pdf", "paper/pnas/figs/fig1b.pdf"],
+    "fig1": (["paper/manuscript/figs/fig1a.pdf", "paper/manuscript/figs/fig1b.pdf"],
         "A map of the study. (A) Each row is a biological mechanism the paper borrows, each column a level "
         "of realism at which it is tested: an inheritance model (an exact simulation of knowledge "
         "transmission, green), trained neural networks measured against exact oracles (blue), and language "
@@ -42,7 +42,7 @@ FIGURES: dict[str, tuple[list[str], str]] = {
         "time, which is what population genetics describes. Dots are capabilities: the rare one (gold) is "
         "lost under single-parent inheritance, reassembled by merging complementary parents, and re-supplied "
         "by grounding."),
-    "fig2": (["paper/pnas/figs/fig2.pdf"],
+    "fig2": (["paper/manuscript/figs/fig2.pdf"],
         "How much real data stops model collapse. (A) An image-generating network (a variational "
         "autoencoder) is trained on handwritten digits, then a fresh copy is trained only on the digits the "
         "previous one drew, for fifteen generations, with no real data added. Each row is a later generation "
@@ -58,7 +58,7 @@ FIGURES: dict[str, tuple[list[str], str]] = {
         "data's diversity is kept, about 0.05 (bootstrap CI shaded). The hollow point at $g = 0$ has not yet "
         "reached its equilibrium of zero. The trained image model needed about twice this fraction, because "
         "a trained network is not the exact copier the simulation assumes (Fig.~S2)."),
-    "fig4": (["paper/pnas/figs/fig4.pdf"],
+    "fig4": (["paper/manuscript/figs/fig4.pdf"],
         "A population of language models over six generations. (A) The set-up. Three lineages start "
         "from one frozen 1.5-billion-parameter base (Qwen2.5-1.5B). Each generation, every lineage learns "
         "one new skill from a public dataset by continuing to train its parent's adapter (300 new examples "
@@ -82,7 +82,7 @@ FIGURES: dict[str, tuple[list[str], str]] = {
         "agreement with the crowd instead of on the truth, collapses the population onto a confident but "
         "wrong consensus (D, F); removing recombination or diversity preservation strands it below the "
         "optimum (D) and drains diversity fastest (E). Each removal fails in its own way."),
-    "fig5": (["paper/pnas/figs/fig5.pdf"],
+    "fig5": (["paper/manuscript/figs/fig5.pdf"],
         "Model speciation: when two lineages can no longer merge. (A, B) Small image classifiers "
         "(multilayer perceptrons) forked from one trained base. Two networks that compute the same function "
         "can still differ in their weights, because hidden units can be renumbered and rescaled without "
@@ -101,7 +101,7 @@ FIGURES: dict[str, tuple[list[str], str]] = {
         "compatible, then crashes, sooner the denser the incompatibilities (E), and the probability of a "
         "non-viable hybrid rises with divergence (F). What breaks merging is conflicting conventions on "
         "shared machinery, not distance or specialisation as such."),
-    "fig3": (["paper/pnas/figs/fig3.pdf"],
+    "fig3": (["paper/manuscript/figs/fig3.pdf"],
         "Merging language-model specialists: when it helps, and predicting when it will hurt. All models "
         "are built from one frozen base (Qwen2.5) plus a LoRA adapter, a small set of extra weights trained "
         "on one family of tasks (list puzzles, string puzzles or arithmetic); a verifier marks every answer "

paper/manuscript/build_lay_legends.py

@@ -1,10 +1,10 @@
 """Build the student-level figure guide (figure_legends_for_students.md -> .tex -> PDF).
 
 The Markdown is the source of truth. Lines of the form ``![](path.pdf)`` become figure includes
-(paths relative to paper/pnas/); ``## `` headings become unnumbered sections; everything else goes
+(paths relative to paper/manuscript/); ``## `` headings become unnumbered sections; everything else goes
 through build.py's inline() converter, so the same Markdown subset and unicode handling apply.
 
-Usage:  python paper/pnas/build_lay_legends.py && (cd paper/pnas && tectonic figure_legends_for_students.tex)
+Usage:  python paper/manuscript/build_lay_legends.py && (cd paper/manuscript && tectonic figure_legends_for_students.tex)
 """
 
 from __future__ import annotations

paper/manuscript/build_zotero_library.py

@@ -1,6 +1,6 @@
 """Build a Zotero-importable library from the manuscript's reference list.
 
-For each of the numbered references in paper/pnas/main.md: take the DOI printed in the entry when
+For each of the numbered references in paper/manuscript/main.md: take the DOI printed in the entry when
 there is one, otherwise ask Crossref for it by title (accepting only a high-scoring match whose title
 really is the same, checked by normalised comparison). Then fetch authoritative metadata for every
 resolved DOI by content negotiation against doi.org, which serves Crossref and DataCite alike, and
@@ -9,7 +9,7 @@ write the result as CSL-JSON plus RIS.
 Entries whose DOI cannot be resolved (pre-DOI literature, books, chapters) are reported and written
 from the manuscript's own metadata so nothing is silently dropped.
 
-Usage:  python paper/pnas/build_zotero_library.py
+Usage:  python paper/manuscript/build_zotero_library.py
 """
 
 from __future__ import annotations

paper/manuscript/figure_legends_for_students.md

@@ -93,7 +93,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S1. Collapse and rescue in three different kinds of network
 
-![](figs/si/s1_architectures.pdf)
+![](figs/si/s1_figS1_architectures.pdf)
 
 **What was done.** The same generational loop as Figure 2 (train a child only on its parent's output, with or without 5% real data) was run with three generators: an exact histogram (a simple frequency count, no neural network), a recurrent neural network (one that reads and writes sequences one token at a time), and a feed-forward network. Each had to learn a synthetic "universe" of 256 kinds of item whose true frequencies were known exactly, for 22 generations, five times over.
 
@@ -113,7 +113,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S3. Re-baselining a collapsed population locks in the damage
 
-![](figs/si/s3_E6.pdf)
+![](figs/si/s3_figS3_rebaselining.pdf)
 
 **What was done.** A tempting shortcut in practice is to declare a model's current output the new "ground truth" and stop keeping the original data. The simulation tests what that does. Two hundred generations in, and again at 300, the population's current frequencies are frozen as the new reference for grounding and the original truth is thrown away (it is kept only to measure against). Four arms: re-baseline while still healthy (green); re-baseline after collapse (red); the same, but only allowed when diversity is above 0.75 (blue); never re-baseline (grey).
 
@@ -123,7 +123,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S4. The full grounding sweep in the simulation
 
-![](figs/si/s4_E2.pdf)
+![](figs/si/s4_fig2_grounding_sweep.pdf)
 
 **What was done.** The complete version of the experiment summarised in Figure 2B: 1,000 knowledge items with a long tail of rare ones, 200 samples per generation, 500 generations, 100 lineages, and the fraction g of real data swept from 0 to 0.4.
 
@@ -133,7 +133,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S5. Real data protects only the topics it covers
 
-![](figs/si/s5_E3.pdf)
+![](figs/si/s5_figS5_aimed_grounding.pdf)
 
 **What was done.** The 1,000 items were divided into ten topics. The same total budget of real data was spent in two ways: spread evenly over all ten topics, or concentrated on a single topic that the experimenter wants to protect.
 
@@ -143,7 +143,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S6. Grounding in a trained recurrent network
 
-![](figs/si/s6_grounding.pdf)
+![](figs/si/s6_figS6_grounding_rnn.pdf)
 
 **What was done.** The grounding sweep of Figure S4 repeated in a trained recurrent network rather than the simulation: 256 kinds of item, 200 samples per generation, 30 generations, nine values of g from 0 to 0.2, eighteen repeats.
 
@@ -153,7 +153,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S7. Collapse and rescue on real handwritten digits, in numbers
 
-![](figs/si/s7_mnist_collapse.pdf)
+![](figs/si/s7_fig2_mnist_collapse.pdf)
 
 **What was done.** The experiment whose pictures are in Figure 2A, quantified. Thirty kinds of digit, a classifier reading the kind of each drawn digit with 98.5% accuracy, 6,000 drawings per generation, fifteen generations, four repeats, with 0% (red) or 10% (green) real digits mixed in.
 
@@ -163,7 +163,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S8. Averaging parents cancels the benefit of having several; keeping the best of each does not
 
-![](figs/si/s8_E4.pdf)
+![](figs/si/s8_figS8_multiparent_union.pdf)
 
 **What was done.** Several parent models each remember a random share of the rare items, and the experimenter controls how similar their shares are (from fully complementary to identical). A child is then built either by averaging the parents' output frequencies, or by keeping, for each item, the largest frequency any parent gives it (a *union*). The child then resamples, as every generation does, and the question is how many rare items survive in it.
 
@@ -173,7 +173,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S9. Many complementary parents can produce an offspring better than any of them
 
-![](figs/si/s9_E8.pdf)
+![](figs/si/s9_figS9_specialist_superparent.pdf)
 
 **What was done.** A capability is modelled as a string of twelve yes/no positions (a *genotype* of twelve *loci*), and fitness is the number of positions that are right. Each parent is a specialist: confident and correct on the positions it has mastered, unsure elsewhere, and no parent has mastered them all. Offspring are built from 2 to 12 parents either by averaging or by taking, position by position, the answer of the parent most confident about it.
 
@@ -183,7 +183,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S10. When skills are entangled, blind recombination harms the offspring
 
-![](figs/si/s10_E9.pdf)
+![](figs/si/s10_figS10_rugged_landscapes.pdf)
 
 **What was done.** The same twelve-position genotypes, now on a *rugged* landscape (Kauffman's NK model), in which the value of a position depends on what its neighbours hold, with a knob K from 0 (positions independent) to 8 (highly entangled). Parents are local optima found by hill-climbing, the model of a trained specialist. Offspring are made by recombining them at rates from 0 (copy a parent) to 0.5 (free shuffling).
 
@@ -193,7 +193,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S11. Directed sex: choosing and screening offspring rescues recombination
 
-![](figs/si/s11_E10.pdf)
+![](figs/si/s11_figS11_directed_recombination.pdf)
 
 **What was done.** Biology is stuck with two random parents and no preview of the offspring. A model population is not: it can pick complementary parents, breed many candidate offspring, test them, keep the fittest and repeat. On the rugged landscapes of Figure S10 three strategies are compared: the best single parent (grey), random recombination (blue) and this *directed* recombination (red, five rounds).
 
@@ -203,7 +203,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S12. Selecting for the best destroys diversity; rewarding novelty preserves it
 
-![](figs/si/s12_E5.pdf)
+![](figs/si/s12_figS12_quality_diversity.pdf)
 
 **What was done.** Each generation, the simulation now *selects* which items to keep, all arms receiving the same grounding. Three rules: no selection; *greedy*, keeping the items of highest true probability; and *quality-diversity*, which rewards an item for being rare as well as good, with a knob (alpha) for how much rarity counts.
 
@@ -213,7 +213,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 ## Figure S13. Who should mate with whom: mating breadth on rugged landscapes
 
-![](figs/si/s13_E14.pdf)
+![](figs/si/s13_figS13_mating_breadth.pdf)
 
 **What was done.** Forty-eight simulated agents sit on a ring. When an offspring is made, its second parent is drawn from a neighbourhood whose width is the knob: narrow (mating only with neighbours, like an isolated village) to the whole ring (anyone can mate with anyone). An offspring replaces the agent at its position only if it is fitter. Ruggedness K is swept from 0 to 10.
 

paper/manuscript/figure_legends_for_students.tex

@@ -105,7 +105,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S1. Collapse and rescue in three different kinds of network}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s1_architectures.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s1_figS1_architectures.pdf}\end{center}
 
 \textbf{What was done.} The same generational loop as Figure 2 (train a child only on its parent's output, with or without 5\% real data) was run with three generators: an exact histogram (a simple frequency count, no neural network), a recurrent neural network (one that reads and writes sequences one token at a time), and a feed-forward network. Each had to learn a synthetic ``universe'' of 256 kinds of item whose true frequencies were known exactly, for 22 generations, five times over.
 
@@ -125,7 +125,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S3. Re-baselining a collapsed population locks in the damage}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s3_E6.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s3_figS3_rebaselining.pdf}\end{center}
 
 \textbf{What was done.} A tempting shortcut in practice is to declare a model's current output the new ``ground truth'' and stop keeping the original data. The simulation tests what that does. Two hundred generations in, and again at 300, the population's current frequencies are frozen as the new reference for grounding and the original truth is thrown away (it is kept only to measure against). Four arms: re-baseline while still healthy (green); re-baseline after collapse (red); the same, but only allowed when diversity is above 0.75 (blue); never re-baseline (grey).
 
@@ -135,7 +135,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S4. The full grounding sweep in the simulation}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s4_E2.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s4_fig2_grounding_sweep.pdf}\end{center}
 
 \textbf{What was done.} The complete version of the experiment summarised in Figure 2B: 1,000 knowledge items with a long tail of rare ones, 200 samples per generation, 500 generations, 100 lineages, and the fraction g of real data swept from 0 to 0.4.
 
@@ -145,7 +145,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S5. Real data protects only the topics it covers}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s5_E3.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s5_figS5_aimed_grounding.pdf}\end{center}
 
 \textbf{What was done.} The 1,000 items were divided into ten topics. The same total budget of real data was spent in two ways: spread evenly over all ten topics, or concentrated on a single topic that the experimenter wants to protect.
 
@@ -155,7 +155,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S6. Grounding in a trained recurrent network}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s6_grounding.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s6_figS6_grounding_rnn.pdf}\end{center}
 
 \textbf{What was done.} The grounding sweep of Figure S4 repeated in a trained recurrent network rather than the simulation: 256 kinds of item, 200 samples per generation, 30 generations, nine values of g from 0 to 0.2, eighteen repeats.
 
@@ -165,7 +165,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S7. Collapse and rescue on real handwritten digits, in numbers}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s7_mnist_collapse.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s7_fig2_mnist_collapse.pdf}\end{center}
 
 \textbf{What was done.} The experiment whose pictures are in Figure 2A, quantified. Thirty kinds of digit, a classifier reading the kind of each drawn digit with 98.5\% accuracy, 6,000 drawings per generation, fifteen generations, four repeats, with 0\% (red) or 10\% (green) real digits mixed in.
 
@@ -175,7 +175,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S8. Averaging parents cancels the benefit of having several; keeping the best of each does not}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s8_E4.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s8_figS8_multiparent_union.pdf}\end{center}
 
 \textbf{What was done.} Several parent models each remember a random share of the rare items, and the experimenter controls how similar their shares are (from fully complementary to identical). A child is then built either by averaging the parents' output frequencies, or by keeping, for each item, the largest frequency any parent gives it (a \emph{union}). The child then resamples, as every generation does, and the question is how many rare items survive in it.
 
@@ -185,7 +185,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S9. Many complementary parents can produce an offspring better than any of them}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s9_E8.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s9_figS9_specialist_superparent.pdf}\end{center}
 
 \textbf{What was done.} A capability is modelled as a string of twelve yes/no positions (a \emph{genotype} of twelve \emph{loci}), and fitness is the number of positions that are right. Each parent is a specialist: confident and correct on the positions it has mastered, unsure elsewhere, and no parent has mastered them all. Offspring are built from 2 to 12 parents either by averaging or by taking, position by position, the answer of the parent most confident about it.
 
@@ -195,7 +195,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S10. When skills are entangled, blind recombination harms the offspring}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s10_E9.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s10_figS10_rugged_landscapes.pdf}\end{center}
 
 \textbf{What was done.} The same twelve-position genotypes, now on a \emph{rugged} landscape (Kauffman's NK model), in which the value of a position depends on what its neighbours hold, with a knob K from 0 (positions independent) to 8 (highly entangled). Parents are local optima found by hill-climbing, the model of a trained specialist. Offspring are made by recombining them at rates from 0 (copy a parent) to 0.5 (free shuffling).
 
@@ -205,7 +205,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S11. Directed sex: choosing and screening offspring rescues recombination}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s11_E10.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s11_figS11_directed_recombination.pdf}\end{center}
 
 \textbf{What was done.} Biology is stuck with two random parents and no preview of the offspring. A model population is not: it can pick complementary parents, breed many candidate offspring, test them, keep the fittest and repeat. On the rugged landscapes of Figure S10 three strategies are compared: the best single parent (grey), random recombination (blue) and this \emph{directed} recombination (red, five rounds).
 
@@ -215,7 +215,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S12. Selecting for the best destroys diversity; rewarding novelty preserves it}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s12_E5.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s12_figS12_quality_diversity.pdf}\end{center}
 
 \textbf{What was done.} Each generation, the simulation now \emph{selects} which items to keep, all arms receiving the same grounding. Three rules: no selection; \emph{greedy}, keeping the items of highest true probability; and \emph{quality-diversity}, which rewards an item for being rare as well as good, with a knob (alpha) for how much rarity counts.
 
@@ -225,7 +225,7 @@ The supplementary figures are the experiments behind the main text that either r
 
 \section*{Figure S13. Who should mate with whom: mating breadth on rugged landscapes}
 
-\begin{center}\includegraphics[width=\textwidth]{figs/si/s13_E14.pdf}\end{center}
+\begin{center}\includegraphics[width=\textwidth]{figs/si/s13_figS13_mating_breadth.pdf}\end{center}
 
 \textbf{What was done.} Forty-eight simulated agents sit on a ring. When an offspring is made, its second parent is drawn from a neighbourhood whose width is the knob: narrow (mating only with neighbours, like an isolated village) to the whole ring (anyone can mate with anyone). An offspring replaces the agent at its position only if it is fitter. Ruggedness K is swept from 0 to 10.
 

paper/manuscript/make_figs.py

@@ -8,7 +8,7 @@ Figs. 3 and 4 open with a schematic panel explaining the set-up, so a figure is
 caption. The per-experiment figures under results/ remain the exploratory versions; these are the
 manuscript's.
 
-Usage:  python paper/pnas/make_figs.py
+Usage:  python paper/manuscript/make_figs.py
 """
 
 from __future__ import annotations
@@ -26,7 +26,7 @@ import os
 os.chdir(ROOT)                                            # load_bundle uses repo-relative paths
 from _figlib import load_bundle, load_seed_bundles, mean_ci  # noqa: E402
 
-OUT = ROOT / "paper" / "pnas" / "figs"
+OUT = ROOT / "paper" / "manuscript" / "figs"
 
 plt.rcParams.update({
     "font.size": 8, "axes.labelsize": 8.5, "legend.fontsize": 7, "legend.frameon": False,
@@ -348,11 +348,11 @@ def fig1b():
 
 # ---------------------------------------------------------------- fig 2: grounding + MNIST
 def fig2():
-    from knowledge.analysis import critical_grounding, reduce_to_stationary
-    from knowledge.metrics import heterozygosity
-    from knowledge.truth import make_true_distribution
+    from inheritance.analysis import critical_grounding, reduce_to_stationary
+    from inheritance.metrics import heterozygosity
+    from inheritance.truth import make_true_distribution
 
-    df, cfg = load_bundle("results/E2")
+    df, cfg = load_bundle("results/fig2_grounding_sweep")
     n = cfg["dynamics"]["n"]
     td = make_true_distribution(cfg["truth"]["K"], 1, "zipf", cfg["truth"]["tail_frac"],
                                 cfg["truth"]["zipf_s"], 0, tail_threshold=cfg["truth"]["tail_threshold"])
@@ -391,7 +391,7 @@ def fig2():
 
     ax = axes[0]
     from PIL import Image
-    im = np.asarray(Image.open("results/mnist_collapse/mnist_montage.png"))
+    im = np.asarray(Image.open("results/fig2_mnist_collapse/mnist_montage.png"))
     # Strip the baked-in title band and left label margin (raster text is unreadable at panel
     # size); measured on the committed montage: boxes span y >= 69, x >= 75, row centres below.
     top, left = 60, 68
@@ -424,7 +424,7 @@ def fig4():
     import pandas as pd
     from matplotlib.patches import FancyArrowPatch, FancyBboxPatch
 
-    df, _ = load_bundle("results/E11")
+    df, _ = load_bundle("results/fig4_society_ablation")
     arms = [("full", "#2ca02c", "full system"),
             ("no_sex", "#ff7f0e", "no recombination"),
             ("no_diversity", "#9467bd", "no diversity preservation"),
@@ -582,7 +582,7 @@ def fig5():
     fig, axes = plt.subplots(2, 3, figsize=(11.4, 8.4), gridspec_kw={"wspace": 0.45, "hspace": 0.75})
     fig.subplots_adjust(top=0.9)
 
-    bdm, _ = load_bundle("results/E12")
+    bdm, _ = load_bundle("results/fig5_speciation_bdm")
     rhos = sorted(bdm["rho"].unique())
     colors = plt.cm.viridis(np.linspace(0.15, 0.85, len(rhos)))
 

paper/manuscript/refs/README.md

@@ -1,6 +1,6 @@
 # Zotero library for the manuscript's references
 
-`Evolution of sex for AI.ris` — all 80 references from `paper/pnas/main.md`, with metadata
+`Evolution of sex for AI.ris` — all 80 references from `paper/manuscript/main.md`, with metadata
 fetched from the publisher of record rather than retyped.
 
 **To import:** Zotero → File → Import… → choose this file → keep "Place imported collections and

paper/manuscript/renumber_refs.py

@@ -1,14 +1,14 @@
 """Renumber the manuscript's references to first-appearance order (PNAS style).
 
-Reads paper/pnas/main.md, finds every parenthesised citation group in the text above
+Reads paper/manuscript/main.md, finds every parenthesised citation group in the text above
 "## References", derives the order in which references first appear, and rewrites the citation
 groups in main.md, si.md, and the figure captions in build.py, then reorders the reference list.
 Citation groups are parentheses containing only reference numbers, commas, en-dash ranges, an optional
 "cf. " prefix, or a prose prefix ending in a semicolon ("...; 11, 12"). Four-digit numbers (years)
 never match, and any number above the list length is reported and left alone.
 
-Usage:  python paper/pnas/renumber_refs.py            # dry run: mapping + per-file counts
-        python paper/pnas/renumber_refs.py --apply    # rewrite the three files in place
+Usage:  python paper/manuscript/renumber_refs.py            # dry run: mapping + per-file counts
+        python paper/manuscript/renumber_refs.py --apply    # rewrite the three files in place
 """
 
 from __future__ import annotations
@@ -18,7 +18,7 @@ import sys
 from pathlib import Path
 
 ROOT = Path(__file__).resolve().parents[2]
-MAIN, SI, BUILD = (ROOT / "paper" / "pnas" / n for n in ("main.md", "si.md", "build.py"))
+MAIN, SI, BUILD = (ROOT / "paper" / "manuscript" / n for n in ("main.md", "si.md", "build.py"))
 REF_HEADER = "## References"
 CIT = re.compile(
     r"\((?P<pre>[^()]*?;\s*)?(?P<cf>cf\.\s*)?"

paper/manuscript/si.md

@@ -217,20 +217,20 @@ and the item-wise maximum rises with it.
 |---|---|---|---|---|
 | Population collapse in the inheritance model is Wright–Fisher drift | Closed form; the diagnosis itself is due to prior work | Knowledge is a categorical distribution; refitting means resampling | Closed forms reproduced to <0.5% | Real learners add a signed, architecture-specific estimator bias (measured) |
 | Grounding behaves like immigration, and the critical real-data fraction is far below one | Closed form, plus the sign confirmed empirically | Fresh samples from a fixed, non-drifting truth | Exact `H_eq`; `g*≈0.048`; sign holds in RNN/MLP/VAE and on MNIST | Deepest tail unrescuable at feasible budgets (`m ∼ 1/p`); sharp threshold softens in trained nets |
-| "Merge, don't average" conservation | Exact **for the output-mean operator** | Rare-item regime; an oracle/verifier identifies the strongest source | E4 closed form + simulation; neural reproduction | Weight-averaging and routing are empirical cousins, not instances; budgets differ; bridge = the headroom rule |
-| Offspring exceed every parent (Fisher–Muller) | Interpretation + empirical | Complementary (decorrelated) parents; verifiable fitness | E8 (inheritance model); LoRA merges beat the best specialist overall in every seed at 0.5B (5 seeds) and 7B (3 seeds) | LLM tier: 3 lexically-distinct families |
-| Outbreeding depression on rugged landscapes; operator design rule | Biological-model result; hypothesis at LLM scale | NK epistasis stands in for skill entanglement | E9–E10; directed selection rescues | Not yet mapped onto a real task-entanglement measure |
-| Optimal mate-pool breadth shrinks with ruggedness | Biological-model result; hypothesis for merging populations | Ring population, local selection | E14 | Phenomenon known to island-model evolutionary computation; the contribution here is the mapping and the diversity/mean decomposition |
+| "Merge, don't average" conservation | Exact **for the output-mean operator** | Rare-item regime; an oracle/verifier identifies the strongest source | `figS8_multiparent_union` closed form + simulation; neural reproduction | Weight-averaging and routing are empirical cousins, not instances; budgets differ; bridge = the headroom rule |
+| Offspring exceed every parent (Fisher–Muller) | Interpretation + empirical | Complementary (decorrelated) parents; verifiable fitness | `figS9_specialist_superparent` (inheritance model); LoRA merges beat the best specialist overall in every seed at 0.5B (5 seeds) and 7B (3 seeds) | LLM tier: 3 lexically-distinct families |
+| Outbreeding depression on rugged landscapes; operator design rule | Biological-model result; hypothesis at LLM scale | NK epistasis stands in for skill entanglement | `figS10_rugged_landscapes`, `figS11_directed_recombination`; directed selection rescues | Not yet mapped onto a real task-entanglement measure |
+| Optimal mate-pool breadth shrinks with ruggedness | Biological-model result; hypothesis for merging populations | Ring population, local selection | `figS13_mating_breadth` | Phenomenon known to island-model evolutionary computation; the contribution here is the mapping and the diversity/mean decomposition |
 | Merge failure decomposes into a coordinate artefact plus a functional residual | Empirical at the trained-network and language-model tiers | Alignment enumerates the architecture's unit symmetries | Full-symmetry residual ≈ 0 for compatible parents versus ≈ the naive barrier under conflict; a cliff in hybrid fitness; function-specific breakdown at the LLM tier | Scoped to aligned linear interpolation; conflict floor is information-theoretic, not genetic |
-| Epistasis (not divergence) sets the cliff; snowball onset | Biological-model result; **hypothesis** at the neural tier | BDM incompatibility structure | E12 | Snowball count ≠ performance cliff without the effect-size link; neural test outstanding |
+| Epistasis (not divergence) sets the cliff; snowball onset | Biological-model result; **hypothesis** at the neural tier | BDM incompatibility structure | `fig5_speciation_bdm` | Snowball count ≠ performance cliff without the effect-size link; neural test outstanding |
 | Pre-merge functional disagreement predicts merge penalty | Empirical, within a controlled grid (0.5B, 13 conditions × 3 seeds) | Constructed conflict/overlap/duration axes; oracle-potential outcome (pre-registered; ordering sensitive to reference) | Clustered CIs exclude 0; held-out LOCO ρ≈0.4; selected geometry baselines ≈ 0 | Head-to-head predictor differences not individually significant; only selected baselines; generalisation to real task pairs open |
 | Confidence weighting improves rank prediction over raw disagreement | Not supported (pre-registered internal prediction) | — | Paired contrast over the same bootstrap resamples: Δ\|ρ\| = −0.021, CI [−0.130, +0.059] | The weighting does sharpen the conflict-versus-compatible level contrast, so it is not useless — only no better as a rank predictor |
 | The predictor improves budget-matched operator choice | **Open** | — | Soup-vs-route gap readout noise-dominated at 0.5B | The practical payoff; untested |
 | Emergent speciation without label conflict | Not observed (pre-registered) | Shared ancestry; compatible tasks; the divergences tested | Residual 0.000 to 6.4× base training; the merge rescues the specialists | Bounds the hypothesis; longer horizons/distribution shift/capacity pressure untested |
-| Grounding, recombination, and diversity preservation make complementary contributions | Biological-model result; hypothesis at LLM scale | Conformity stands in for self-consumption | E11 four-arm ablation; each arm fails in a distinct way | General joint necessity is not established; the language-model population (Fig. 4B–C) lacks differential reproduction between lineages |
+| Grounding, recombination, and diversity preservation make complementary contributions | Biological-model result; hypothesis at LLM scale | Conformity stands in for self-consumption | `fig4_society_ablation` four-arm ablation; each arm fails in a distinct way | General joint necessity is not established; the language-model population (Fig. 4B–C) lacks differential reproduction between lineages |
 | Obligate recombination collapses once partners carry conflicting conventions | Empirical (1.5B base, 3 lineages × 6 generations, 3 seeds) | Latin-square curriculum; replay present; linear merge; no culling of lineages | Best lineage 0.269 vs 0.796 never merging; onset at complementarity < 0.8; own-ancestor merge 0.663; three alternative mechanisms refuted (SI Text S3) | Six generations; one base; the arrival order of conflicting families is set by the curriculum |
 | A declinable merge reverts the population to asexual accumulation without advance knowledge of when to stop | Empirical (same population, plus two controls, 3 seeds each) | "Keep the parent" scored as one candidate on validation data | Fraction declined 0.44 → 1.00 across generations; finishes 0.792 vs 0.796 never merging. Forced stop after generation 2 finishes 0.793 (veto − stop3 per seed −0.008/−0.006/+0.011). Under a decorrelated curriculum (complementarity 0.00 → 0.70 → 0.00) declines still rise 0.44 → 0.89; pooled partial ρ(declined, complementarity \| generation) = −0.07, CI (−0.21, +0.09); partial ρ with generation +0.31 | The reduction-principle reading (declines track complementarity) is **not supported**; declines track generation, which here confounds adapter age, skill count and the arrival of conflicting conventions. Modifier set by evaluation, not evolved |
-| Recombination's net benefit across six generations is an early lead, not a final gain | Empirical (same population); consistent with the inheritance model's speed advantage (E7) | Every skill reaches every lineage by the curriculum regardless | +0.08 at generation 0; −0.005 at generation 5 (per-seed −0.03/+0.01/+0.01) | Replay present, so forgetting was not a live pressure; a curriculum that withholds skills from some lineages is untested |
+| Recombination's net benefit across six generations is an early lead, not a final gain | Empirical (same population); consistent with the inheritance model's speed advantage | Every skill reaches every lineage by the curriculum regardless | +0.08 at generation 0; −0.005 at generation 5 (per-seed −0.03/+0.01/+0.01) | Replay present, so forgetting was not a live pressure; a curriculum that withholds skills from some lineages is untested |
 
 ## SI Table S2: headline quantitative results
 
@@ -240,7 +240,7 @@ per-experiment tables and falsifier status in the per-experiment documentation).
 | Result | Setting / n | Outcome definition | Headline |
 |---|---|---|---|
 | Closed-form validation | Inheritance model; standing tests | Simulated vs closed-form H-decay, immigration equilibrium, multi-parent union | Agreement < 0.5% |
-| Grounding retention | Inheritance model (E2); 100 lineages per grounding level | Fraction of equilibrium diversity retained at grounding `g` (operational threshold) | `g ≈ 0.05` retains ≥95% in the tested setting; smooth in `g` |
+| Grounding retention | Inheritance model (`fig2_grounding_sweep`); 100 lineages per grounding level | Fraction of equilibrium diversity retained at grounding `g` (operational threshold) | `g ≈ 0.05` retains ≥95% in the tested setting; smooth in `g` |
 | MNIST collapse & rescue | Conv-VAE, 4 replicates; frozen oracle (98.5% mode acc.) | Mode support / forward-KL over generations | Dry: 30→1 modes; 10% grounding: 30/30 held |
 | Fisher–Muller in LLMs | 5 seeds (0.5B) and 3 seeds (7B), fixed tests | Merged vs best-specialist accuracy (overall; worst family); ±: 95% CI over seeds | 0.5B ties 0.647±0.027 vs 0.592±0.009; 7B soup 0.873±0.004 vs 0.807±0.038 (soup − best +0.066±0.036, 3/3 seeds) |
 | Union vs blend (headroom) | 3 seeds (0.5B hard); 3 seeds (7B hard) | Paired per-seed ordering, routing vs weight-average | 0.5B: routing > blend in 3/3 seeds, one catastrophic blend failure avoided. 7B: routing 0.503±0.007 vs soup 0.408±0.021 (+0.094±0.015, 3/3); soup vs best specialist +0.001±0.041 (the seed-1 'soup below best parent' did not replicate). Directed − soup +0.073±0.031 (3/3) |
@@ -309,20 +309,19 @@ Replicate counts, and why each is what it is:
 
 | Experiment | Replicates | Reasoning |
 |---|---|---|
-| E1, E2, E3, E5, E6 | 100 lineages | Long horizons (400–600 generations) with drift-dominated variance; 100 lineages put the CI on stationary diversity well inside the effect being resolved |
-| E4 | 200 | Outcomes are per-item binary retentions, the highest-variance quantity in the paper |
-| E7 | 20 | Trajectory contrast (sexual vs asexual adaptation speed), large and monotone |
-| E8 | 40 | The vertical claim; the headline separation, so the most replicated of the genotype experiments |
-| E9, E10 | 24 | Landscape sweeps where each point aggregates 200 offspring internally |
-| E11 | 12 | Four-arm ablation over 80 generations; arms separate by margins far exceeding the CI |
-| E12, E12_nk | 15 | Each point already averages 500 (E12) or 200 (E12_nk) offspring |
-| E14 | 20 | Breadth × ruggedness grid, 60 generations per cell |
-| kernel_sharpen, kernel_smooth | 24 | Two-parameter kernel fits against neural reference endpoints |
-| bridge | 60 | The harness gate: must detect *any* departure from the inheritance model, so the most replicated neural run |
-| grounding | 18 | Nine-point grounding sweep with per-generation network retraining |
-| collapse, architectures | 5 | Sign-level demonstrations across architectures; each lineage retrains a network 22–25 times |
+| `fig2_grounding_sweep`, `figS5_aimed_grounding`, `figS12_quality_diversity`, `figS3_rebaselining` | 100 lineages | Long horizons (400–600 generations) with drift-dominated variance; 100 lineages put the CI on stationary diversity well inside the effect being resolved |
+| `figS8_multiparent_union` | 200 | Outcomes are per-item binary retentions, the highest-variance quantity in the paper |
+| `figS9_specialist_superparent` | 40 | The vertical claim; the headline separation, so the most replicated of the genotype experiments |
+| `figS10_rugged_landscapes`, `figS11_directed_recombination` | 24 | Landscape sweeps where each point aggregates 200 offspring internally |
+| `fig4_society_ablation` | 12 | Four-arm ablation over 80 generations; arms separate by margins far exceeding the CI |
+| `fig5_speciation_bdm` | 15 | Each point already averages 500 offspring |
+| `figS13_mating_breadth` | 20 | Breadth × ruggedness grid, 60 generations per cell |
+| `figS2_kernel_sharpen`, `figS2_kernel_smooth` | 24 | Two-parameter kernel fits against neural reference endpoints |
+| `bridge` | 60 | The harness gate: must detect *any* departure from the inheritance model, so the most replicated neural run |
+| `figS6_grounding_rnn` | 18 | Nine-point grounding sweep with per-generation network retraining |
+| `collapse`, `figS1_architectures` | 5 | Sign-level demonstrations across architectures; each lineage retrains a network 22–25 times |
 | recombination | 8 | Operator contrast in trained weights |
-| mnist_collapse | 4 | 15 generations × a conv-VAE retrained from scratch each generation; the contrast (30 modes vs 1) is categorical |
+| `fig2_mnist_collapse` | 4 | 15 generations × a conv-VAE retrained from scratch each generation; the contrast (30 modes vs 1) is categorical |
 | speciation_real, _cliff | 3 | Barrier decomposition; the quantity is a near-deterministic function of the training condition (residual 0.001 vs 0.497) |
 | speciation_real_emergent | 4 | A null: replicates are spent on longer divergence horizons rather than more repeats |
 | llm_merge_seeds | 5 training seeds | The Fisher–Muller signature, the most-replicated language-model claim |
@@ -351,30 +350,33 @@ population size in the Wright–Fisher correspondence and the distillation sampl
 reading. Horizons of 400–600 generations were chosen so that ungrounded lineages reach fixation and
 grounded ones reach stationarity within the run, which the trajectories confirm.
 
-*Sweeps.* E2 sweeps grounding `g ∈ {0, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.4}`; E3 contrasts uniform
-against region-matched grounding allocation; E4 crosses parent count `K_T ∈ {1,2,3,5}` with parent
-correlation `ρ ∈ {0, 0.25, 0.5, 0.75, 1}` and `g ∈ {0, 0.02, 0.05}`; E5 crosses selection mode
-(none / greedy / quality-diversity) with novelty weight; E6 compares four re-minting arms.
+*Sweeps.* The grounding sweep (`fig2_grounding_sweep`) sweeps `g ∈ {0, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.4}`;
+the aimed-grounding experiment (`figS5_aimed_grounding`) contrasts uniform against region-matched
+grounding allocation; the multi-parent union experiment (`figS8_multiparent_union`) crosses parent count
+`K_T ∈ {1,2,3,5}` with parent correlation `ρ ∈ {0, 0.25, 0.5, 0.75, 1}` and `g ∈ {0, 0.02, 0.05}`; the
+selection experiment (`figS12_quality_diversity`) crosses selection mode (none / greedy /
+quality-diversity) with novelty weight; the re-baselining experiment (`figS3_rebaselining`) compares
+four re-minting arms.
 
-*The correlated-parent construction (E4).* Parent correlation is constructed directly rather than
+*The correlated-parent construction (`figS8_multiparent_union`).* Parent correlation is constructed directly rather than
 obtained by tuning drift, so that `ρ` is not confounded with `n`, `m`, tail size, or generation
 count. For each tail item a shared switch `z ~ Bern(ρ)`, a shared retention `s ~ Bern(q)`, and
 per-parent `u⁽ᵏ⁾ ~ Bern(q)` give parent `k` retention `s` if `z` else `u⁽ᵏ⁾`. This yields exact
 marginal retention `q` and exact pairwise correlation `ρ`, and is exchangeable, so `ρ` is a single
 scalar knob.
 
-*Multi-locus experiments (E7–E11, E14).* Genotypes are `L = 12` biallelic loci (4096 genotypes —
+*Multi-locus experiments* (`figS9_specialist_superparent`, `figS10_rugged_landscapes`,
+`figS11_directed_recombination`, `fig4_society_ablation`, `figS13_mating_breadth`). Genotypes are `L = 12` biallelic loci (4096 genotypes —
 effectively open-ended relative to the population sizes used), with fitness either additive or a
-Kauffman NK landscape whose interaction count `K` tunes ruggedness from 0 to 10. E9 and E10 breed
-from `n_parents = 6` local optima into populations of 200 offspring; E10 additionally screens
-offspring and iterates (5 rounds, keeping 8). E11 runs a population of `N = 60` agents for 80
+Kauffman NK landscape whose interaction count `K` tunes ruggedness from 0 to 10. The landscape and directed-recombination experiments breed
+from `n_parents = 6` local optima into populations of 200 offspring; the directed one additionally
+screens offspring and iterates (5 rounds, keeping 8). The society ablation runs a population of `N = 60` agents for 80
 generations at ruggedness `K = 8`, with mutation `μ = 0.03`, 120 offspring per generation, and
-selection weighting true fitness against consensus conformity at `g = 0.85`. E14 sweeps mate-pool
+selection weighting true fitness against consensus conformity at `g = 0.85`. The mating-breadth experiment sweeps mate-pool
 breadth on a ring of `N = 48` against ruggedness.
 
-*Speciation (E12).* `L = 20` loci, incompatibility density `ρ ∈ {0.1, 0.25, 0.5}`, parental
-divergence swept 0–20 substitutions, 500 offspring per cell at recombination rate 0.5. E12_nk repeats
-the question on NK landscapes (`L = 16`, `K` 0–10, 40 parent pairs, 200 offspring).
+*Speciation (`fig5_speciation_bdm`).* `L = 20` loci, incompatibility density `ρ ∈ {0.1, 0.25, 0.5}`,
+parental divergence swept 0–20 substitutions, 500 offspring per cell at recombination rate 0.5.
 
 *Validation.* Three closed forms are asserted as standing tests to within 0.5%: neutral
 heterozygosity decay `E[H_t] = H_0(1 − 1/n)^t`, the exact immigration–drift equilibrium, and the

paper/manuscript/si.tex

@@ -1,5 +1,5 @@
 % Supplementary Information — readable single-column build (tectonic/XeLaTeX). Content is generated from si.md
-% by `python paper/pnas/build.py si`; journal-template reflow happens at submission.
+% by `python paper/manuscript/build.py si`; journal-template reflow happens at submission.
 \ifdefined\XeTeXversion\else\ifdefined\pdfoutput\pdfoutput=1\fi\fi
 \documentclass[11pt]{article}
 

paper/manuscript/si_body.tex

@@ -68,20 +68,20 @@ Both statements are confirmed by simulation in Fig. S8, where mean-mixture survi
 Claim & Status & Key assumptions & Evidence & Known limits \\ \hline
 Population collapse in the inheritance model is Wright--Fisher drift & Closed form; the diagnosis itself is due to prior work & Knowledge is a categorical distribution; refitting means resampling & Closed forms reproduced to <0.5\% & Real learners add a signed, architecture-specific estimator bias (measured) \\[3pt]
 Grounding behaves like immigration, and the critical real-data fraction is far below one & Closed form, plus the sign confirmed empirically & Fresh samples from a fixed, non-drifting truth & Exact \texttt{H\_eq}; \texttt{g*\(\approx\)0.048}; sign holds in RNN/MLP/VAE and on MNIST & Deepest tail unrescuable at feasible budgets (\texttt{m \(\sim\) 1/p}); sharp threshold softens in trained nets \\[3pt]
-``Merge, don't average'' conservation & Exact \textbf{for the output-mean operator} & Rare-item regime; an oracle/verifier identifies the strongest source & E4 closed form + simulation; neural reproduction & Weight-averaging and routing are empirical cousins, not instances; budgets differ; bridge = the headroom rule \\[3pt]
-Offspring exceed every parent (Fisher--Muller) & Interpretation + empirical & Complementary (decorrelated) parents; verifiable fitness & E8 (inheritance model); LoRA merges beat the best specialist overall in every seed at 0.5B (5 seeds) and 7B (3 seeds) & LLM tier: 3 lexically-distinct families \\[3pt]
-Outbreeding depression on rugged landscapes; operator design rule & Biological-model result; hypothesis at LLM scale & NK epistasis stands in for skill entanglement & E9--E10; directed selection rescues & Not yet mapped onto a real task-entanglement measure \\[3pt]
-Optimal mate-pool breadth shrinks with ruggedness & Biological-model result; hypothesis for merging populations & Ring population, local selection & E14 & Phenomenon known to island-model evolutionary computation; the contribution here is the mapping and the diversity/mean decomposition \\[3pt]
+``Merge, don't average'' conservation & Exact \textbf{for the output-mean operator} & Rare-item regime; an oracle/verifier identifies the strongest source & \texttt{figS8\_multiparent\_union} closed form + simulation; neural reproduction & Weight-averaging and routing are empirical cousins, not instances; budgets differ; bridge = the headroom rule \\[3pt]
+Offspring exceed every parent (Fisher--Muller) & Interpretation + empirical & Complementary (decorrelated) parents; verifiable fitness & \texttt{figS9\_specialist\_superparent} (inheritance model); LoRA merges beat the best specialist overall in every seed at 0.5B (5 seeds) and 7B (3 seeds) & LLM tier: 3 lexically-distinct families \\[3pt]
+Outbreeding depression on rugged landscapes; operator design rule & Biological-model result; hypothesis at LLM scale & NK epistasis stands in for skill entanglement & \texttt{figS10\_rugged\_landscapes}, \texttt{figS11\_directed\_recombination}; directed selection rescues & Not yet mapped onto a real task-entanglement measure \\[3pt]
+Optimal mate-pool breadth shrinks with ruggedness & Biological-model result; hypothesis for merging populations & Ring population, local selection & \texttt{figS13\_mating\_breadth} & Phenomenon known to island-model evolutionary computation; the contribution here is the mapping and the diversity/mean decomposition \\[3pt]
 Merge failure decomposes into a coordinate artefact plus a functional residual & Empirical at the trained-network and language-model tiers & Alignment enumerates the architecture's unit symmetries & Full-symmetry residual \(\approx\) 0 for compatible parents versus \(\approx\) the naive barrier under conflict; a cliff in hybrid fitness; function-specific breakdown at the LLM tier & Scoped to aligned linear interpolation; conflict floor is information-theoretic, not genetic \\[3pt]
-Epistasis (not divergence) sets the cliff; snowball onset & Biological-model result; \textbf{hypothesis} at the neural tier & BDM incompatibility structure & E12 & Snowball count \(\neq\) performance cliff without the effect-size link; neural test outstanding \\[3pt]
+Epistasis (not divergence) sets the cliff; snowball onset & Biological-model result; \textbf{hypothesis} at the neural tier & BDM incompatibility structure & \texttt{fig5\_speciation\_bdm} & Snowball count \(\neq\) performance cliff without the effect-size link; neural test outstanding \\[3pt]
 Pre-merge functional disagreement predicts merge penalty & Empirical, within a controlled grid (0.5B, 13 conditions \(\times\) 3 seeds) & Constructed conflict/overlap/duration axes; oracle-potential outcome (pre-registered; ordering sensitive to reference) & Clustered CIs exclude 0; held-out LOCO \(\rho\)\(\approx\)0.4; selected geometry baselines \(\approx\) 0 & Head-to-head predictor differences not individually significant; only selected baselines; generalisation to real task pairs open \\[3pt]
 Confidence weighting improves rank prediction over raw disagreement & Not supported (pre-registered internal prediction) & --- & Paired contrast over the same bootstrap resamples: \(\Delta\)\textbackslash{} & \(\rho\)\textbackslash{} \\[3pt]
 The predictor improves budget-matched operator choice & \textbf{Open} & --- & Soup-vs-route gap readout noise-dominated at 0.5B & The practical payoff; untested \\[3pt]
 Emergent speciation without label conflict & Not observed (pre-registered) & Shared ancestry; compatible tasks; the divergences tested & Residual 0.000 to 6.4\(\times\) base training; the merge rescues the specialists & Bounds the hypothesis; longer horizons/distribution shift/capacity pressure untested \\[3pt]
-Grounding, recombination, and diversity preservation make complementary contributions & Biological-model result; hypothesis at LLM scale & Conformity stands in for self-consumption & E11 four-arm ablation; each arm fails in a distinct way & General joint necessity is not established; the language-model population (Fig. 4B--C) lacks differential reproduction between lineages \\[3pt]
+Grounding, recombination, and diversity preservation make complementary contributions & Biological-model result; hypothesis at LLM scale & Conformity stands in for self-consumption & \texttt{fig4\_society\_ablation} four-arm ablation; each arm fails in a distinct way & General joint necessity is not established; the language-model population (Fig. 4B--C) lacks differential reproduction between lineages \\[3pt]
 Obligate recombination collapses once partners carry conflicting conventions & Empirical (1.5B base, 3 lineages \(\times\) 6 generations, 3 seeds) & Latin-square curriculum; replay present; linear merge; no culling of lineages & Best lineage 0.269 vs 0.796 never merging; onset at complementarity < 0.8; own-ancestor merge 0.663; three alternative mechanisms refuted (SI Text S3) & Six generations; one base; the arrival order of conflicting families is set by the curriculum \\[3pt]
 A declinable merge reverts the population to asexual accumulation without advance knowledge of when to stop & Empirical (same population, plus two controls, 3 seeds each) & ``Keep the parent'' scored as one candidate on validation data & Fraction declined 0.44 \(\rightarrow\) 1.00 across generations; finishes 0.792 vs 0.796 never merging. Forced stop after generation 2 finishes 0.793 (veto \(-\) stop3 per seed \(-\)0.008/\(-\)0.006/+0.011). Under a decorrelated curriculum (complementarity 0.00 \(\rightarrow\) 0.70 \(\rightarrow\) 0.00) declines still rise 0.44 \(\rightarrow\) 0.89; pooled partial \(\rho\)(declined, complementarity \textbackslash{} & generation) = \(-\)0.07, CI (\(-\)0.21, +0.09); partial \(\rho\) with generation +0.31 \\[3pt]
-Recombination's net benefit across six generations is an early lead, not a final gain & Empirical (same population); consistent with the inheritance model's speed advantage (E7) & Every skill reaches every lineage by the curriculum regardless & +0.08 at generation 0; \(-\)0.005 at generation 5 (per-seed \(-\)0.03/+0.01/+0.01) & Replay present, so forgetting was not a live pressure; a curriculum that withholds skills from some lineages is untested \\[3pt]
+Recombination's net benefit across six generations is an early lead, not a final gain & Empirical (same population); consistent with the inheritance model's speed advantage & Every skill reaches every lineage by the curriculum regardless & +0.08 at generation 0; \(-\)0.005 at generation 5 (per-seed \(-\)0.03/+0.01/+0.01) & Replay present, so forgetting was not a live pressure; a curriculum that withholds skills from some lineages is untested \\[3pt]
 \hline\end{tabular}\end{center}\medskip
 
 \section*{SI Table S2: headline quantitative results}
@@ -93,7 +93,7 @@ Headline quantitative results with sample sizes, uncertainty, and outcome defini
 \hline
 Result & Setting / n & Outcome definition & Headline \\ \hline
 Closed-form validation & Inheritance model; standing tests & Simulated vs closed-form H-decay, immigration equilibrium, multi-parent union & Agreement < 0.5\% \\[3pt]
-Grounding retention & Inheritance model (E2); 100 lineages per grounding level & Fraction of equilibrium diversity retained at grounding \texttt{g} (operational threshold) & \texttt{g \(\approx\) 0.05} retains \(\geq\)95\% in the tested setting; smooth in \texttt{g} \\[3pt]
+Grounding retention & Inheritance model (\texttt{fig2\_grounding\_sweep}); 100 lineages per grounding level & Fraction of equilibrium diversity retained at grounding \texttt{g} (operational threshold) & \texttt{g \(\approx\) 0.05} retains \(\geq\)95\% in the tested setting; smooth in \texttt{g} \\[3pt]
 MNIST collapse \& rescue & Conv-VAE, 4 replicates; frozen oracle (98.5\% mode acc.) & Mode support / forward-KL over generations & Dry: 30\(\rightarrow\)1 modes; 10\% grounding: 30/30 held \\[3pt]
 Fisher--Muller in LLMs & 5 seeds (0.5B) and 3 seeds (7B), fixed tests & Merged vs best-specialist accuracy (overall; worst family); \(\pm\): 95\% CI over seeds & 0.5B ties 0.647\(\pm\)0.027 vs 0.592\(\pm\)0.009; 7B soup 0.873\(\pm\)0.004 vs 0.807\(\pm\)0.038 (soup \(-\) best +0.066\(\pm\)0.036, 3/3 seeds) \\[3pt]
 Union vs blend (headroom) & 3 seeds (0.5B hard); 3 seeds (7B hard) & Paired per-seed ordering, routing vs weight-average & 0.5B: routing > blend in 3/3 seeds, one catastrophic blend failure avoided. 7B: routing 0.503\(\pm\)0.007 vs soup 0.408\(\pm\)0.021 (+0.094\(\pm\)0.015, 3/3); soup vs best specialist +0.001\(\pm\)0.041 (the seed-1 'soup below best parent' did not replicate). Directed \(-\) soup +0.073\(\pm\)0.031 (3/3) \\[3pt]
@@ -141,20 +141,19 @@ Replicate counts, and why each is what it is:
 \begin{tabular}{p{0.307\textwidth} p{0.307\textwidth} p{0.307\textwidth}}
 \hline
 Experiment & Replicates & Reasoning \\ \hline
-E1, E2, E3, E5, E6 & 100 lineages & Long horizons (400--600 generations) with drift-dominated variance; 100 lineages put the CI on stationary diversity well inside the effect being resolved \\[3pt]
-E4 & 200 & Outcomes are per-item binary retentions, the highest-variance quantity in the paper \\[3pt]
-E7 & 20 & Trajectory contrast (sexual vs asexual adaptation speed), large and monotone \\[3pt]
-E8 & 40 & The vertical claim; the headline separation, so the most replicated of the genotype experiments \\[3pt]
-E9, E10 & 24 & Landscape sweeps where each point aggregates 200 offspring internally \\[3pt]
-E11 & 12 & Four-arm ablation over 80 generations; arms separate by margins far exceeding the CI \\[3pt]
-E12, E12\_nk & 15 & Each point already averages 500 (E12) or 200 (E12\_nk) offspring \\[3pt]
-E14 & 20 & Breadth \(\times\) ruggedness grid, 60 generations per cell \\[3pt]
-kernel\_sharpen, kernel\_smooth & 24 & Two-parameter kernel fits against neural reference endpoints \\[3pt]
-bridge & 60 & The harness gate: must detect \emph{any} departure from the inheritance model, so the most replicated neural run \\[3pt]
-grounding & 18 & Nine-point grounding sweep with per-generation network retraining \\[3pt]
-collapse, architectures & 5 & Sign-level demonstrations across architectures; each lineage retrains a network 22--25 times \\[3pt]
+\texttt{fig2\_grounding\_sweep}, \texttt{figS5\_aimed\_grounding}, \texttt{figS12\_quality\_diversity}, \texttt{figS3\_rebaselining} & 100 lineages & Long horizons (400--600 generations) with drift-dominated variance; 100 lineages put the CI on stationary diversity well inside the effect being resolved \\[3pt]
+\texttt{figS8\_multiparent\_union} & 200 & Outcomes are per-item binary retentions, the highest-variance quantity in the paper \\[3pt]
+\texttt{figS9\_specialist\_superparent} & 40 & The vertical claim; the headline separation, so the most replicated of the genotype experiments \\[3pt]
+\texttt{figS10\_rugged\_landscapes}, \texttt{figS11\_directed\_recombination} & 24 & Landscape sweeps where each point aggregates 200 offspring internally \\[3pt]
+\texttt{fig4\_society\_ablation} & 12 & Four-arm ablation over 80 generations; arms separate by margins far exceeding the CI \\[3pt]
+\texttt{fig5\_speciation\_bdm} & 15 & Each point already averages 500 offspring \\[3pt]
+\texttt{figS13\_mating\_breadth} & 20 & Breadth \(\times\) ruggedness grid, 60 generations per cell \\[3pt]
+\texttt{figS2\_kernel\_sharpen}, \texttt{figS2\_kernel\_smooth} & 24 & Two-parameter kernel fits against neural reference endpoints \\[3pt]
+\texttt{bridge} & 60 & The harness gate: must detect \emph{any} departure from the inheritance model, so the most replicated neural run \\[3pt]
+\texttt{figS6\_grounding\_rnn} & 18 & Nine-point grounding sweep with per-generation network retraining \\[3pt]
+\texttt{collapse}, \texttt{figS1\_architectures} & 5 & Sign-level demonstrations across architectures; each lineage retrains a network 22--25 times \\[3pt]
 recombination & 8 & Operator contrast in trained weights \\[3pt]
-mnist\_collapse & 4 & 15 generations \(\times\) a conv-VAE retrained from scratch each generation; the contrast (30 modes vs 1) is categorical \\[3pt]
+\texttt{fig2\_mnist\_collapse} & 4 & 15 generations \(\times\) a conv-VAE retrained from scratch each generation; the contrast (30 modes vs 1) is categorical \\[3pt]
 speciation\_real, \_cliff & 3 & Barrier decomposition; the quantity is a near-deterministic function of the training condition (residual 0.001 vs 0.497) \\[3pt]
 speciation\_real\_emergent & 4 & A null: replicates are spent on longer divergence horizons rather than more repeats \\[3pt]
 llm\_merge\_seeds & 5 training seeds & The Fisher--Muller signature, the most-replicated language-model claim \\[3pt]
@@ -175,13 +174,13 @@ Knowledge is a distribution over \texttt{K} discrete items; reality is a fixed Z
 
 \emph{Parameter choices.} \texttt{K = 500}--\texttt{1000} with \texttt{zipf\_s = 1.1} and half the items designated tail: large enough that the rare tail contains hundreds of items (so tail statistics are not dominated by a handful of them) and small enough to sweep densely. \texttt{n = 100}--\texttt{200} sets drift strength; it is the population size in the Wright--Fisher correspondence and the distillation sample size in the AI reading. Horizons of 400--600 generations were chosen so that ungrounded lineages reach fixation and grounded ones reach stationarity within the run, which the trajectories confirm.
 
-\emph{Sweeps.} E2 sweeps grounding \texttt{g \(\in\) {0, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.4}}; E3 contrasts uniform against region-matched grounding allocation; E4 crosses parent count \texttt{K\_T \(\in\) {1,2,3,5}} with parent correlation \texttt{\(\rho\) \(\in\) {0, 0.25, 0.5, 0.75, 1}} and \texttt{g \(\in\) {0, 0.02, 0.05}}; E5 crosses selection mode (none / greedy / quality-diversity) with novelty weight; E6 compares four re-minting arms.
+\emph{Sweeps.} The grounding sweep (\texttt{fig2\_grounding\_sweep}) sweeps \texttt{g \(\in\) {0, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.4}}; the aimed-grounding experiment (\texttt{figS5\_aimed\_grounding}) contrasts uniform against region-matched grounding allocation; the multi-parent union experiment (\texttt{figS8\_multiparent\_union}) crosses parent count \texttt{K\_T \(\in\) {1,2,3,5}} with parent correlation \texttt{\(\rho\) \(\in\) {0, 0.25, 0.5, 0.75, 1}} and \texttt{g \(\in\) {0, 0.02, 0.05}}; the selection experiment (\texttt{figS12\_quality\_diversity}) crosses selection mode (none / greedy / quality-diversity) with novelty weight; the re-baselining experiment (\texttt{figS3\_rebaselining}) compares four re-minting arms.
 
-\emph{The correlated-parent construction (E4).} Parent correlation is constructed directly rather than obtained by tuning drift, so that \texttt{\(\rho\)} is not confounded with \texttt{n}, \texttt{m}, tail size, or generation count. For each tail item a shared switch \texttt{z \textasciitilde{} Bern(\(\rho\))}, a shared retention \texttt{s \textasciitilde{} Bern(q)}, and per-parent \texttt{u\(^{(k)}\) \textasciitilde{} Bern(q)} give parent \texttt{k} retention \texttt{s} if \texttt{z} else \texttt{u\(^{(k)}\)}. This yields exact marginal retention \texttt{q} and exact pairwise correlation \texttt{\(\rho\)}, and is exchangeable, so \texttt{\(\rho\)} is a single scalar knob.
+\emph{The correlated-parent construction (\texttt{figS8\_multiparent\_union}).} Parent correlation is constructed directly rather than obtained by tuning drift, so that \texttt{\(\rho\)} is not confounded with \texttt{n}, \texttt{m}, tail size, or generation count. For each tail item a shared switch \texttt{z \textasciitilde{} Bern(\(\rho\))}, a shared retention \texttt{s \textasciitilde{} Bern(q)}, and per-parent \texttt{u\(^{(k)}\) \textasciitilde{} Bern(q)} give parent \texttt{k} retention \texttt{s} if \texttt{z} else \texttt{u\(^{(k)}\)}. This yields exact marginal retention \texttt{q} and exact pairwise correlation \texttt{\(\rho\)}, and is exchangeable, so \texttt{\(\rho\)} is a single scalar knob.
 
-\emph{Multi-locus experiments (E7--E11, E14).} Genotypes are \texttt{L = 12} biallelic loci (4096 genotypes --- effectively open-ended relative to the population sizes used), with fitness either additive or a Kauffman NK landscape whose interaction count \texttt{K} tunes ruggedness from 0 to 10. E9 and E10 breed from \texttt{n\_parents = 6} local optima into populations of 200 offspring; E10 additionally screens offspring and iterates (5 rounds, keeping 8). E11 runs a population of \texttt{N = 60} agents for 80 generations at ruggedness \texttt{K = 8}, with mutation \texttt{\(\mu\) = 0.03}, 120 offspring per generation, and selection weighting true fitness against consensus conformity at \texttt{g = 0.85}. E14 sweeps mate-pool breadth on a ring of \texttt{N = 48} against ruggedness.
+\emph{Multi-locus experiments} (\texttt{figS9\_specialist\_superparent}, \texttt{figS10\_rugged\_landscapes}, \texttt{figS11\_directed\_recombination}, \texttt{fig4\_society\_ablation}, \texttt{figS13\_mating\_breadth}). Genotypes are \texttt{L = 12} biallelic loci (4096 genotypes --- effectively open-ended relative to the population sizes used), with fitness either additive or a Kauffman NK landscape whose interaction count \texttt{K} tunes ruggedness from 0 to 10. The landscape and directed-recombination experiments breed from \texttt{n\_parents = 6} local optima into populations of 200 offspring; the directed one additionally screens offspring and iterates (5 rounds, keeping 8). The society ablation runs a population of \texttt{N = 60} agents for 80 generations at ruggedness \texttt{K = 8}, with mutation \texttt{\(\mu\) = 0.03}, 120 offspring per generation, and selection weighting true fitness against consensus conformity at \texttt{g = 0.85}. The mating-breadth experiment sweeps mate-pool breadth on a ring of \texttt{N = 48} against ruggedness.
 
-\emph{Speciation (E12).} \texttt{L = 20} loci, incompatibility density \texttt{\(\rho\) \(\in\) {0.1, 0.25, 0.5}}, parental divergence swept 0--20 substitutions, 500 offspring per cell at recombination rate 0.5. E12\_nk repeats the question on NK landscapes (\texttt{L = 16}, \texttt{K} 0--10, 40 parent pairs, 200 offspring).
+\emph{Speciation (\texttt{fig5\_speciation\_bdm}).} \texttt{L = 20} loci, incompatibility density \texttt{\(\rho\) \(\in\) {0.1, 0.25, 0.5}}, parental divergence swept 0--20 substitutions, 500 offspring per cell at recombination rate 0.5.
 
 \emph{Validation.} Three closed forms are asserted as standing tests to within 0.5\%: neutral heterozygosity decay \texttt{E[H\_t] = H\_0(1 \(-\) 1/n)\textasciicircum{}t}, the exact immigration--drift equilibrium, and the multi-parent union formula. These run in CI alongside the correctness tests. If they fail, the science is wrong rather than merely the code.
 
@@ -236,7 +235,7 @@ Output of \texttt{figures/stats\_llm\_epistasis.py} (clustered CIs, paired predi
 Sixteen figures are cited from the main text by number. Each is the per-experiment figure regenerated from the committed results artifact (\texttt{figures/plot\_*.py}), reproduced here without re-plotting, so panel titles still carry the experiment's working name. Five of them are inheritance-model results with no real-model counterpart in this paper, reported here because each reproduces an established result: blending versus union retention (Fig. S8), the Fisher--Muller super-parent (Fig. S9), outbreeding depression on rugged landscapes (Fig. S10), directed recombination (Fig. S11), and the mate-pool breadth optimum (Fig. S13).
 
 \begin{figure}[p]\centering  % s1
-\includegraphics[width=\textwidth]{figs/si/s1_architectures.pdf}
+\includegraphics[width=\textwidth]{figs/si/s1_figS1_architectures.pdf}
 \caption{Collapse, and its arrest by real data, in three kinds of generator. The generational loop of Fig.~2 (train a child only on its parent's output, with or without 5\% real data) is run with an exact frequency count (a histogram, no network), a recurrent network and a feed-forward network, on a synthetic universe of 256 knowledge items whose true frequencies are known exactly; 200 samples per generation, 22 generations, 5 replicates. (A) Distance from the true distribution (forward KL divergence, which grows the more of the truth a model fails to cover) against generation: solid lines, with no real data, climb in every architecture; dashed lines, with 5\% real data, stay low. (B) The same distance at the end of the run (error bars over replicates): real data lowers it in all three. (C) The fraction of rare items still alive at the end: real data raises it in all three. The histogram's bars in C are small because a frequency count drops a rare item outright once it is unseen, whereas the networks keep some alive by smoothing (the subject of Fig.~S2). A variational autoencoder was excluded because it failed the generation-0 fidelity check on this task.}\label{s1}
 \end{figure}
 
@@ -246,57 +245,57 @@ Sixteen figures are cited from the main text by number. Each is the per-experime
 \end{figure}
 
 \begin{figure}[p]\centering  % s3
-\includegraphics[width=\textwidth]{figs/si/s3_E6.pdf}
+\includegraphics[width=\textwidth]{figs/si/s3_figS3_rebaselining.pdf}
 \caption{Re-baselining a collapsed population locks in its losses. A tempting shortcut is to declare a model's current output the new reference and discard the original data. In the inheritance model (500 items, 200 samples per generation, 400 generations, 100 replicates) the population's current frequencies are frozen as the new grounding reference at generations 150 and 300 (dotted verticals) and the original truth is kept only for measurement. Four arms: re-baseline while still diverse, under generous real data (green); re-baseline after collapse, under starved real data (red); the same starvation with re-baselining allowed only while heterozygosity is above 0.75 (blue); never re-baseline (grey). (A) Distance from the original truth against generation (bands over replicates): the red arm steps up at each re-baselining and never returns; the healthy arm shows small steps; the gated and never arms coincide. (B) Heterozygosity, with the gate's threshold dashed: the gated arm never re-baselines because it stays below the line. Once rare knowledge is gone from every copy it cannot be rebuilt (Muller's ratchet); a diversity gate prevents the shortcut from making the loss permanent.}\label{s3}
 \end{figure}
 
 \begin{figure}[p]\centering  % s4
-\includegraphics[width=\textwidth]{figs/si/s4_E2.pdf}
+\includegraphics[width=\textwidth]{figs/si/s4_fig2_grounding_sweep.pdf}
 \caption{The full real-data sweep in the inheritance model (the experiment summarised in Fig.~2B). 1,000 knowledge items with a long tail of rare ones, 200 samples per generation, 500 generations, 100 lineages; each generation also receives $m$ fresh real samples, so the real-data share is $g = m/(n+m)$, swept from 0 to 0.4. (A) Heterozygosity against generation, one line per $g$: with no real data it declines steadily; with any real data it levels off. (B) The level it settles at against $g$ (points, simulation) with the exact prediction (dashed) and the real data's own diversity (dotted); the red line marks $g^* = 0.048$ (95\% CI 0.047--0.050), where 95\% of the real data's diversity is kept. The hollow point at $g = 0$ has not converged (its equilibrium is zero). (C) The fraction of the rare tail retained, counted by items (red) and by their share of the truth (purple): both rise with $g$ but stay below 0.1 even at $g = 0.4$. (D) Survival by band of rarity, from the rarest (band 0) to the least rare (band 3): the rarest recover last. Overall diversity is cheap to protect; a rare item persists only once about one real example of it arrives per generation, so protecting it costs about one over its frequency in real samples.}\label{s4}
 \end{figure}
 
 \begin{figure}[p]\centering  % s5
-\includegraphics[width=\textwidth]{figs/si/s5_E3.pdf}
+\includegraphics[width=\textwidth]{figs/si/s5_figS5_aimed_grounding.pdf}
 \caption{Real data protects only the topics it covers. The 1,000 items are divided into ten topics (regions) and the same total budget of real data is spent either evenly over all ten or concentrated on one topic the experimenter wants to protect; 200 samples per generation, 400 generations, 100 replicates. (A) The fraction of that topic's rare items still alive against generation, with real data aimed at it (blue) or spread evenly (red), bands 95\% CI: aimed grounding holds about half the topic's rare items, spread grounding lets it fall to about 0.07. (B) Survival per topic at the end, same colours, the protected topic marked by the dotted line: aimed grounding protects its topic and leaves the others with no surviving rare items; spread grounding gives every topic the same low survival. Per-topic heterozygosity is confounded by how much of the truth each topic carries, so rare-item survival is the readout. A fixed budget of real data should be aimed at the knowledge one wants to keep.}\label{s5}
 \end{figure}
 
 \begin{figure}[p]\centering  % s6
-\includegraphics[width=\textwidth]{figs/si/s6_grounding.pdf}
+\includegraphics[width=\textwidth]{figs/si/s6_figS6_grounding_rnn.pdf}
 \caption{The real-data response in a trained recurrent network. The sweep of Fig.~S4 repeated in a recurrent generator rather than the exact simulation: 256 items, 200 samples per generation, 30 generations, $g$ swept over nine values from 0 to 0.2, 18 replicates. (A) Distance from the truth (forward KL divergence) against generation, one line per $g$: more real data suppresses the climb. (B) The final distance against $g$ (error bars 95\% CI), falling steadily from 2.08 with no real data to 0.75 at $g = 0.2$ (paired $t = 3.3$ at $g = 0.2$). (C) The fraction of the achievable improvement each $g$ buys: half of it arrives by $g = 0.040$ (red line; bootstrap 95\% CI 0.004--0.116 shaded), close to the simulation's $g^* = 0.048$ (black dashed), but the full improvement needs $g$ near 0.19. (D) Three ways of measuring collapse on one 0--1 scale: heterozygosity relative to the truth (blue) is flat near 0.8; the count of surviving rare items (orange) rises and falls with no pattern; the divergence-based recovery (green) rises cleanly. The direction of the effect matches the simulation, the threshold softens, and counting surviving items is the wrong ruler for a smoothing network, which keeps inventing rare items that are not in the truth; distance from the truth is the measure used for such networks.}\label{s6}
 \end{figure}
 
 \begin{figure}[p]\centering  % s7
-\includegraphics[width=\textwidth]{figs/si/s7_mnist_collapse.pdf}
+\includegraphics[width=\textwidth]{figs/si/s7_fig2_mnist_collapse.pdf}
 \caption{Collapse and rescue on real handwritten digits, in numbers (the experiment whose drawings are in Fig.~2A). A convolutional variational autoencoder is retrained from scratch each generation on the previous generation's drawings plus a fraction $g$ of real MNIST digits; the 30 kinds of digit (digit $\times$ stroke thickness, resampled to a long tail with about 18 rare kinds) are read out by a frozen classifier plus a thickness measure at 98.5\% accuracy. Two arms, $g = 0$ (red) and $g = 0.1$ (green); 6,000 drawings per generation, 15 generations, 4 replicates, bands 95\% CI. (A) Distance from the truth rises from about 0.5 to about 18 with no real data and stays near the floor with 10\%. (B) The number of distinct kinds still drawn falls from 30 to about 1 with no real data; with 10\% all 30 survive (dotted line). (C) The share of the rare kinds still alive falls to 0.06 with no real data; with 10\% all of it is kept. (D) Heterozygosity falls to zero with no real data and stays near 0.9 with 10\% (the truth's value dotted). Everything the simulation predicts appears on real images with an independent judge; the dose of real data needed is about twice the simulation's, for the reason shown in Fig.~S2.}\label{s7}
 \end{figure}
 
 \begin{figure}[p]\centering  % s8
-\includegraphics[width=\textwidth]{figs/si/s8_E4.pdf}
+\includegraphics[width=\textwidth]{figs/si/s8_figS8_multiparent_union.pdf}
 \caption{Averaging several parents cancels the benefit of having several; keeping each parent's strongest contribution does not. In the inheritance model (500 items) $K_T$ parents each remember a random share of the rare items, with the similarity of their shares controlled directly by a correlation $\rho$ (0 fully complementary, 1 identical); $K_T \in \{1, 2, 3, 5\}$, $\rho \in \{0, 0.25, 0.5, 0.75, 1\}$, 200 replicates. A child is built either by averaging the parents' output frequencies or by keeping, for each item, the largest frequency any parent gives it (a union), and then resamples as every generation does. (A) The fraction of the rare tail held by at least one parent against $\rho$, one curve per $K_T$: points are simulation, lines an exact formula, and they match. (B) The fraction that survives in the child: solid lines (union) rise with more and less similar parents; dashed lines (averaging) stay flat near 0.08 whatever the number of parents. (C) The same at $\rho = 0$ against the number of parents (error bars 95\% CI). Averaging dilutes each rare item by the number of parents, which exactly cancels the gain of having more parents to draw on (blending inheritance, the scheme Jenkin showed would swamp rare variants); the union realises the gain, and needs a judge to say which parent holds each item.}\label{s8}
 \end{figure}
 
 \begin{figure}[p]\centering  % s9
-\includegraphics[width=\textwidth]{figs/si/s9_E8.pdf}
+\includegraphics[width=\textwidth]{figs/si/s9_figS9_specialist_superparent.pdf}
 \caption{Many complementary parents can produce an offspring better than any of them. A capability is a string of twelve yes/no positions (a genotype of twelve loci) and fitness is the number of correct positions; each parent is a specialist, confident and correct (0.9) on the positions it has mastered and unsure (0.45) elsewhere, and no parent has mastered them all. Which positions a parent masters is drawn so that the number of parents $K_T$ and their correlation $\rho$ are independent knobs; the deployed capability is the fitness of the most probable genotype; 40 replicates, error bars 95\% CI. (A) Capability against the number of parents when parents master different positions ($\rho = 0$): position-wise recombination (red) reaches the perfect score of 12 with eight parents; the best single parent (grey) sits near 8.7; the average of the parents (blue) reaches about 11.6 at twelve parents. (B) Recombination against the number of parents at $\rho \in \{0, 0.5, 1\}$: complementary parents climb to the optimum, identical parents stay flat near 6. This is the Fisher--Muller effect, unbounded because a model population is not limited to two parents; Fig.~3B is its counterpart in language models.}\label{s9}
 \end{figure}
 
 \begin{figure}[p]\centering  % s10
-\includegraphics[width=\textwidth]{figs/si/s10_E9.pdf}
+\includegraphics[width=\textwidth]{figs/si/s10_figS10_rugged_landscapes.pdf}
 \caption{When skills are entangled, blind recombination harms the offspring. The twelve-position genotypes now sit on a rugged landscape (Kauffman's NK model) in which a position's value depends on its neighbours, with ruggedness $K$ from 0 (positions independent) to 8 (highly entangled). Parents are local optima found by hill-climbing, the model of a trained specialist; offspring are made from them at recombination rates from 0 (copy a parent) to 0.5 (free shuffling); 24 replicate landscapes, 200 offspring per point. (A) Mean offspring fitness minus the best parent against recombination rate, one curve per $K$: on a smooth landscape the difference is zero; as $K$ grows the curves fall, more steeply at higher rates, to about $-0.23$ at $K = 8$ under free recombination. (B) The fitness of the best offspring in a brood (parental level dotted): on rugged landscapes it peaks at an intermediate rate and falls back toward the parents under free shuffling. This is outbreeding depression; the optimal amount of recombination shrinks as skills become more entangled.}\label{s10}
 \end{figure}
 
 \begin{figure}[p]\centering  % s11
-\includegraphics[width=\textwidth]{figs/si/s11_E10.pdf}
+\includegraphics[width=\textwidth]{figs/si/s11_figS11_directed_recombination.pdf}
 \caption{Choosing mates and screening offspring rescues recombination on rugged landscapes. On the landscapes of Fig.~S10 three strategies are compared, all reported as deployed fitness in $[0, 1]$; 24 replicate landscapes, error bars 95\% CI: the best single parent (grey); random recombination, as in biology (blue: random parents, free recombination, offspring taken as they come); and directed recombination, which a model population can do and a living one cannot (red: complementary parents chosen, many offspring generated at rate 0.2, the fittest kept, for five rounds). (A) Capability against ruggedness $K$ with the global optimum dotted: random recombination falls from 0.66 at $K = 2$ to 0.51 at $K = 10$; directed recombination tracks the best parent and the optimum at every $K$. (B) The same as a difference from the best parent: directed stays at or above zero throughout; random falls to about $-0.2$. In language models this is ``breed many merges, keep the best'' (Table~S2).}\label{s11}
 \end{figure}
 
 \begin{figure}[p]\centering  % s12
-\includegraphics[width=\textwidth]{figs/si/s12_E5.pdf}
+\includegraphics[width=\textwidth]{figs/si/s12_figS12_quality_diversity.pdf}
 \caption{Selecting for the best destroys diversity; rewarding novelty preserves it. Each generation of the inheritance model (500 items, 200 samples per generation, 400 generations, 100 replicates, the same real data in every arm) now selects which items to keep, under three rules: no selection; greedy, keeping the items of highest true probability; and quality-diversity, which rewards an item for being rare as well as good, weighting item $i$ by $f_i p_i^{-\alpha}$ with $\alpha \in \{0.5, 1, 2\}$. (A) Heterozygosity against generation: greedy (red) collapses within a few generations to about 0.01; quality-diversity at $\alpha = 1$ (orange) and $\alpha = 2$ (blue) and no selection (green) hold a plateau above 0.85. (B) The settled heterozygosity against $\alpha$ (orange), with greedy (red dashed) and no selection (green dashed) as references: it rises from about 0.48 at $\alpha = 0.5$ to about 0.88 at $\alpha = 2$. (C) The number of distinct items alive at the end: about 9 under greedy, 14 to 28 under quality-diversity, about 32 with no selection. Chasing the best outputs is a directional pressure on top of drift; diversity has to be an objective in its own right, which is the diversity-preservation ingredient of Fig.~4D--F.}\label{s12}
 \end{figure}
 
 \begin{figure}[p]\centering  % s13
-\includegraphics[width=\textwidth]{figs/si/s13_E14.pdf}
+\includegraphics[width=\textwidth]{figs/si/s13_figS13_mating_breadth.pdf}
 \caption{Who should mate with whom: the best mating breadth narrows as skills become more entangled. Forty-eight agents carrying twelve-locus genotypes sit on a ring and evolve for 60 generations on an NK landscape of ruggedness $K \in \{0, 3, 6, 10\}$; an offspring's second parent is drawn from a neighbourhood of half-width $\approx bN/2$, so the breadth $b$ runs from mating only with neighbours ($b = 0.03$) to mating with anyone ($b = 1$), and an offspring replaces the agent at its position only if fitter (mutation 0.003, crossover rate 0.5, 20 replicates, error bars 95\% CI, breadth on a logarithmic axis). (A) The best fitness reached, relative to the optimum, against breadth, per $K$: on a smooth landscape every breadth reaches the optimum; at $K = 3$ the best breadth is 0.6, at $K = 6$ and 10 it is 0.35, and mating with everyone falls below it. (B) The population's mean fitness rises with breadth at every $K > 0$. (C) Standing diversity (mean pairwise Hamming distance) falls with breadth, fastest on rugged landscapes. Wide mixing spreads a good variant fast but homogenises the population, so on entangled problems it loses the ability to explore several solutions in parallel (Wright's argument for structured populations).}\label{s13}
 \end{figure}
 

paper/manuscript/si_figures.py

@@ -8,7 +8,7 @@ reading.
 """
 
 SI_FIGURES: dict[str, tuple[str, str]] = {
-    "s1": ("results/architectures/architectures.pdf",
+    "s1": ("results/figS1_architectures/figS1_architectures.pdf",
         "Collapse, and its arrest by real data, in three kinds of generator. The generational loop of "
         "Fig.~2 (train a child only on its parent's output, with or without 5\\% real data) is run with an "
         "exact frequency count (a histogram, no network), a recurrent network and a feed-forward network, "
@@ -22,7 +22,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "rare item outright once it is unseen, whereas the networks keep some alive by smoothing (the "
         "subject of Fig.~S2). A variational autoencoder was excluded because it failed the generation-0 "
         "fidelity check on this task."),
-    "s2": ("results/kernel_sharpen/kernel.pdf",
+    "s2": ("results/figS2_kernel_sharpen/kernel.pdf",
         "Why trained networks deviate from the ideal copier, in opposite directions. The inheritance model "
         "assumes a child's frequencies are exactly those it sampled from its parent. Two knobs are added to "
         "that copying step: a smoothing knob (a small pull toward treating every item as possible; mutation "
@@ -40,7 +40,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "appear without being inherited) and the sharpening knob is positive frequency-dependent selection "
         "(the majority gains, nothing new appears); a trained network behaves as drift plus one of these two "
         "biases, set by its architecture."),
-    "s3": ("results/E6/E6.pdf",
+    "s3": ("results/figS3_rebaselining/figS3_rebaselining.pdf",
         "Re-baselining a collapsed population locks in its losses. A tempting shortcut is to declare a "
         "model's current output the new reference and discard the original data. In the inheritance model "
         "(500 items, 200 samples per generation, 400 generations, 100 replicates) the population's current "
@@ -54,7 +54,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "dashed: the gated arm never re-baselines because it stays below the line. Once rare knowledge is "
         "gone from every copy it cannot be rebuilt (Muller's ratchet); a diversity gate prevents the "
         "shortcut from making the loss permanent."),
-    "s4": ("results/E2/E2.pdf",
+    "s4": ("results/fig2_grounding_sweep/fig2_grounding_sweep.pdf",
         "The full real-data sweep in the inheritance model (the experiment summarised in Fig.~2B). 1,000 "
         "knowledge items with a long tail of rare ones, 200 samples per generation, 500 generations, 100 "
         "lineages; each generation also receives $m$ fresh real samples, so the real-data share is "
@@ -68,7 +68,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "of rarity, from the rarest (band 0) to the least rare (band 3): the rarest recover last. Overall "
         "diversity is cheap to protect; a rare item persists only once about one real example of it arrives "
         "per generation, so protecting it costs about one over its frequency in real samples."),
-    "s5": ("results/E3/E3.pdf",
+    "s5": ("results/figS5_aimed_grounding/figS5_aimed_grounding.pdf",
         "Real data protects only the topics it covers. The 1,000 items are divided into ten topics "
         "(regions) and the same total budget of real data is spent either evenly over all ten or "
         "concentrated on one topic the experimenter wants to protect; 200 samples per generation, 400 "
@@ -80,7 +80,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "grounding gives every topic the same low survival. Per-topic heterozygosity is confounded by how "
         "much of the truth each topic carries, so rare-item survival is the readout. A fixed budget of real "
         "data should be aimed at the knowledge one wants to keep."),
-    "s6": ("results/grounding/grounding.pdf",
+    "s6": ("results/figS6_grounding_rnn/figS6_grounding_rnn.pdf",
         "The real-data response in a trained recurrent network. The sweep of Fig.~S4 repeated in a "
         "recurrent generator rather than the exact simulation: 256 items, 200 samples per generation, 30 "
         "generations, $g$ swept over nine values from 0 to 0.2, 18 replicates. (A) Distance from the truth "
@@ -95,7 +95,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "direction of the effect matches the simulation, the threshold softens, and counting surviving items "
         "is the wrong ruler for a smoothing network, which keeps inventing rare items that are not in the "
         "truth; distance from the truth is the measure used for such networks."),
-    "s7": ("results/mnist_collapse/mnist_collapse.pdf",
+    "s7": ("results/fig2_mnist_collapse/fig2_mnist_collapse.pdf",
         "Collapse and rescue on real handwritten digits, in numbers (the experiment whose drawings are in "
         "Fig.~2A). A convolutional variational autoencoder is retrained from scratch each generation on the "
         "previous generation's drawings plus a fraction $g$ of real MNIST digits; the 30 kinds of digit "
@@ -109,7 +109,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "data and stays near 0.9 with 10\\% (the truth's value dotted). Everything the simulation predicts "
         "appears on real images with an independent judge; the dose of real data needed is about twice the "
         "simulation's, for the reason shown in Fig.~S2."),
-    "s8": ("results/E4/E4.pdf",
+    "s8": ("results/figS8_multiparent_union/figS8_multiparent_union.pdf",
         "Averaging several parents cancels the benefit of having several; keeping each parent's strongest "
         "contribution does not. In the inheritance model (500 items) $K_T$ parents each remember a random "
         "share of the rare items, with the similarity of their shares controlled directly by a correlation "
@@ -124,7 +124,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "dilutes each rare item by the number of parents, which exactly cancels the gain of having more "
         "parents to draw on (blending inheritance, the scheme Jenkin showed would swamp rare variants); the "
         "union realises the gain, and needs a judge to say which parent holds each item."),
-    "s9": ("results/E8/E8.pdf",
+    "s9": ("results/figS9_specialist_superparent/figS9_specialist_superparent.pdf",
         "Many complementary parents can produce an offspring better than any of them. A capability is a "
         "string of twelve yes/no positions (a genotype of twelve loci) and fitness is the number of correct "
         "positions; each parent is a specialist, confident and correct (0.9) on the positions it has "
@@ -138,7 +138,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "$\\rho \\in \\{0, 0.5, 1\\}$: complementary parents climb to the optimum, identical parents stay "
         "flat near 6. This is the Fisher--Muller effect, unbounded because a model population is not limited "
         "to two parents; Fig.~3B is its counterpart in language models."),
-    "s10": ("results/E9/E9.pdf",
+    "s10": ("results/figS10_rugged_landscapes/figS10_rugged_landscapes.pdf",
         "When skills are entangled, blind recombination harms the offspring. The twelve-position genotypes "
         "now sit on a rugged landscape (Kauffman's NK model) in which a position's value depends on its "
         "neighbours, with ruggedness $K$ from 0 (positions independent) to 8 (highly entangled). Parents are "
@@ -150,7 +150,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "the best offspring in a brood (parental level dotted): on rugged landscapes it peaks at an "
         "intermediate rate and falls back toward the parents under free shuffling. This is outbreeding "
         "depression; the optimal amount of recombination shrinks as skills become more entangled."),
-    "s11": ("results/E10/E10.pdf",
+    "s11": ("results/figS11_directed_recombination/figS11_directed_recombination.pdf",
         "Choosing mates and screening offspring rescues recombination on rugged landscapes. On the "
         "landscapes of Fig.~S10 three strategies are compared, all reported as deployed fitness in "
         "$[0, 1]$; 24 replicate landscapes, error bars 95\\% CI: the best single parent (grey); random "
@@ -162,7 +162,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "the optimum at every $K$. (B) The same as a difference from the best parent: directed stays at or "
         "above zero throughout; random falls to about $-0.2$. In language models this is ``breed many "
         "merges, keep the best'' (Table~S2)."),
-    "s12": ("results/E5/E5.pdf",
+    "s12": ("results/figS12_quality_diversity/figS12_quality_diversity.pdf",
         "Selecting for the best destroys diversity; rewarding novelty preserves it. Each generation of the "
         "inheritance model (500 items, 200 samples per generation, 400 generations, 100 replicates, the same "
         "real data in every arm) now selects which items to keep, under three rules: no selection; greedy, "
@@ -176,7 +176,7 @@ SI_FIGURES: dict[str, tuple[str, str]] = {
         "alive at the end: about 9 under greedy, 14 to 28 under quality-diversity, about 32 with no "
         "selection. Chasing the best outputs is a directional pressure on top of drift; diversity has to be "
         "an objective in its own right, which is the diversity-preservation ingredient of Fig.~4D--F."),
-    "s13": ("results/E14/E14.pdf",
+    "s13": ("results/figS13_mating_breadth/figS13_mating_breadth.pdf",
         "Who should mate with whom: the best mating breadth narrows as skills become more entangled. "
         "Forty-eight agents carrying twelve-locus genotypes sit on a ring and evolve for 60 generations on "
         "an NK landscape of ruggedness $K \\in \\{0, 3, 6, 10\\}$; an offspring's second parent is drawn "

paper/response-to-review-3.md

@@ -1,7 +1,7 @@
 # Response to the third review (of the PNAS-format draft)
 
 *All five priority fixes are made, plus the presentation items. The revised draft is
-`paper/pnas/main.md` (rebuilt PDF alongside); the long-form document and the results documentation
+`paper/manuscript/main.md` (rebuilt PDF alongside); the long-form document and the results documentation
 were corrected wherever they carried the same overstatements. Point-by-point:*
 
 ## 1. The averaging proposition (your §2) — you are right, and the text now proves what it claims

paper/the-evolution-of-sex-for-ai.md

@@ -365,7 +365,7 @@ prematurely converges onto one basin and finds a *worse* champion, while pure mo
 Throughout, wide mating lifts the *typical* model but monotonically **destroys diversity** — so on rugged
 problems, where the best model needs preserved diversity to be found, structured (partly monogamous)
 merging wins. The design rule extends the one above: *merge widely when skills are additive; keep
-structured sub-populations — island-style merging — when skills are rugged.* (Figure: `results/E14/E14.png`.)
+structured sub-populations — island-style merging — when skills are rugged.* (Figure: `results/figS13_mating_breadth/E14.png`.)
 
 *AI can do sex better than biology can.* Biology is stuck with two parents, mating roughly at random,
 and cannot inspect an offspring before it is born. An AI has none of those limits. It can recombine
@@ -423,7 +423,7 @@ increasing specialisation eventually breaks merging and that one should then rou
 trained models is a coordinate artefact removable by aligning neurons (Git Re-Basin — Ainsworth et al.,
 2022). What the frame adds is the *theory* of the phenomenon they observe: its functional form, its
 super-linear (snowball) onset, and its dependence on epistasis — merge failure as a Dobzhansky–Muller
-event. (Figure: `results/E12/E12.png`.)
+event. (Figure: `results/fig5_speciation_bdm/E12.png`.)
 
 **The real-weight confirmation.** The obvious objection to the analytic model is that its
 "incompatibility" is a re-labelled loss barrier, and loss barriers between independently trained