CU-LOG-SPEC-02 — Negative finite evidence against a 5-percent modulo-30 projection threshold

STATUS: UNREVIEWED RESEARCH ARCHIVE / NOT PEER REVIEWED.

Reproducible finite prime-race computations up to 90 million and a segmented-sieve check over four disjoint windows. The former exploratory 5-percent projection threshold fails in two of the four later windows. Dirichlet-character weighted covariance calculations concern a distinct logarithmic limiting model and must not be equated with the finite-window statistics. No prime number theorem, p-value, universal threshold, priority or mathematical novelty is claimed. UNREVIEWED RESEARCH ARCHIVE / NOT PEER REVIEWED. No priority or general theorem claimed.

Source repository: https://github.com/alexcour/couret-finite-verifications/pull/11 (frozen source commit: 94851055ba00464790d7664468be2abc71105e2b).

Licensing of author-owned materials:

• Executable software and scripts: MIT License
• Scientific text, notes and demonstrations: Creative Commons Attribution 4.0 International (CC BY 4.0)
• Original numerical outputs and tabular data: Creative Commons Zero v1.0 Universal (CC0 1.0)

Published to enable independent criticism, verification, and counterexamples. No novelty, priority, or general validity is claimed beyond the stated finite bounds.

Research artifacts and documentation were prepared with AI assistance. Internal machine or AI-assisted checks are not independent human mathematical review. This unreviewed archive does not establish originality or scientific priority.