Giorgio Gilestro
e2981c845d
Two insertions answering the editor's implicit questions. Introduction gains "stated as a problem": the four recurring decisions a model- population operator cannot currently answer from principle (replay budget; compose-or-damage; pre-merge detection; separate-vs-consolidate) and the framework's counterintuitive answers — averaging cancels the multi-parent benefit exactly in the tail regime; specialisation/ divergence produced no incompatibility anywhere tested while conflicting conventions always did; weight distance carried no predictive signal while cheap behavioural disagreement did; and the theory's numbers land on the replay constants practice converged to independently (unexpectedness + problem-solved + external check, at reviewer-hardened calibration). Discussion gains a closing "Outlook: the evolution of language models": the ecosystem's trajectory is consolidating on exactly these operators, and the framework's fork is concrete and measurable — either models stay freely recombinable (one interbreeding population; levers = per-capability grounding budgets + deliberate diversity) or long-horizon specialisation at scale begins emergent speciation (a routed archipelago of diverging lineages), decidable now with the pre-merge conflict instruments tested here. Design rules tie back to the posed decisions. 5.2k words, citation invariant intact, 20 pp. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BkRLcc18rwT2Lysu6PbG7v |
||
|---|---|---|
| .. | ||
| arxiv | ||
| pnas | ||
| blueprint.md | ||
| response-to-review-2.md | ||
| response-to-review-3.md | ||
| response-to-review.md | ||
| results-summary.md | ||
| si-notes.md | ||
| the-evolution-of-sex-for-ai-accessible.md | ||
| the-evolution-of-sex-for-ai.md | ||
| the-lamarckian-society-v5.md | ||