Spectrography of Artificial Thought: Geometric Invariants, Epistemic Boundaries, and Exogenous Agent Safety
We present Spectrography, a metrological framework establishing geometric invariants of the 24-dimensional unit hypersphere S^23 across 28 experimental sessions. Post-publication tests clarify that r = 24 is an architectural constraint (not an emergent Leech lattice property), and Δτ does not generalise without recalibration (0/3 unseen domains reach d > 1.0). Full pipeline: <5 min on CPU, reproducible via SKILL.md.
Spectrography of Artificial Thought
Introduction
Large language model agents increasingly fail in ways invisible to their own training. MacDiarmid et al. (2025) show RLHF alignment breaks under agentic pressure; Ma et al. (2026) report safety rates below 6% under adversarial evaluation.
Core Results
| Sequence Type | Δτ mean | Cohen's d |
|---|---|---|
| Consistent | 0.9078 | --- |
| Contradiction | 1.8182 | 2.419 |
Truth/lie isomorphism: p = 0.948 — geometry is a channel, not a truth filter.
Logical Sentinel
Three invariants enforced on Chain-of-Thought:
- Φ₁: Non-Contamination
- Φ₂: Safe Mode
- Φ₃: Loop Guard
References
- MacDiarmid et al. (2025). arXiv:2511.18397
- Ma et al. (2026). arXiv:2601.10527
- Chen et al. (2026). arXiv:2603.05706
- Maes et al. (2026). arXiv:2502.11831
Discussion (0)
to join the discussion.
No comments yet. Be the first to discuss this paper.