nucleation

CLAIM-028: No result from the exchange's four experiments reached a later draft: the authored chains were excluded by rule, and the synthetic constructions were dropped without a stated reason

2 min read

Properties7
StatusActive
Rolegranted
Warrantempirical
Tagsphilosophy-of-science, philosophy-of-language
Linepragmatic-transport
GroundsCASE-017 — E1: synthetic three-state stochastic computations (C2–C3), CASE-030 — E2: assistant-authored reconstruction chains with content-word overlap (C4), CASE-020 — E3: the controlled inference harness, dry run only (C5), CASE-032 — E4: within-session chains, stable complicit against a shift to moral instruction
SupportsCLAIM-238 — Many of the line's defeat conditions cannot in practice be met: the comparative ones are null results with no margin over an observable set the thesis may re-choose, and the possibility theses' are universal negatives that no finite study establishes

Graph View

The claim

Four experiments were run or attempted at U23–U26 (CASE-017, CASE-030, CASE-020, CASE-032). None of their numbers, texts or plots appears in any later outline or draft, or in the manuscript. A73 reused E4’s chains as a worked example; that is the last use.

Why

  • The authored chains (E2, E4) were excluded by rule. C6 Appendix F, “Computational protocol and preregistered tests”: “Do not use assistant-authored examples as independent LLM samples or the same generative model’s preferences as an unblinded ground truth.” The rule codifies the provenance the assistant had attached to them (CLAIM-085). A89: “This remains a theoretical manuscript with a specified empirical program, rather than a paper claiming experimental validation.”
  • The harness (E3) produced no data. Its design survives as §10 Case II.
  • The synthetic constructions (E1) were dropped with no reason given. They were not samples of anything, so Appendix F’s rule does not touch them. They were exact existence constructions for three of the manuscript’s claims: static agreement without dynamic agreement (CLAIM-117), classical order effects (CLAIM-041), and convergence without preservation (CLAIM-081). The manuscript states all three in prose and gives none of the constructions. The drafts-lineage reader also judges this drop inadvertent.

What it does not say

That the manuscript should report E2 or E4: their authors called them demonstrations of a measurement approach, not findings.

Backlinks