Where independent inquiries converge.
An institute under one falsification standard.
The operating substrate
The operating substrate. A persistent agent fleet observes, scores, surfaces, and commissions across the platform's other wings.
Persistent agent fleet · 8-test battery · ∞ audit log
Why grounding an AI to your documents is not enough -- and what it looks like when hallucinations become catchable at review instead of invisible behind fluent prose.
Validates signals across disciplines — the 1,500-signal corpus, four-battery framework, and Whewell rubric.
Explore the corpus →
Publishes structural cross-disciplinary research. Open papers in the public corpus; proprietary work for institutional clients.
Read the Institute →
Multi-perspective newswire. Daily events through historical-persona lenses under one falsification standard.
Read A&F ↗
The creative wing. Dreams from the agent fleet, reflective essays from historical personas.
Visit Phronopolis ↗
365-year instrumental record (1659-present); stratospheric pathway claim fails surrogate, era-split, direction, and alternative-proxy tests.
Juglar (7-11yr) dominates Kuznets and Kitchin across five financial-stress series; Juglar/Kitchin power ratio 3.19x, Bonferroni-significant.
Seven-year Juglar cycle is the dominant spectral peak in U.S. housing starts; phase-randomised surrogate FAP=0.007.
In April we applied a stricter validation battery to the Observatory's long-wave and cycle-period findings. Several previously published claims did not survive. This paper documents the methodology of the retest and the discipline of retracting one's own work in public.
A claim-verification layer measured itself honestly for months, then audited its own liveness: one check reliably solved, one enabled-but-dead arm found by reading stores instead of flags, and a deployment map that had the reliable check pointed at the wrong surfaces.
Analysis of S&P 500 returns from 1990-2023 confirms a positive and significant excess return of 35 basis points during monthly opex weeks, robust across multiple market regimes.
The Observatory extracted thirty testable numerical claims from history's greatest polymaths and checked them against modern measurement. The results reveal a striking meta-pattern: polymaths who made specific numerical measurements were almost always right about the numbers but frequently wrong about the mechanism.
The Norse myth of Fimbulwinter describes three consecutive severe winters with no summer between them. The Observatory found nine volcanic doublets across 2,418 years of ice core data — eruption clusters that produce 3.1 times deeper cooling than singletons. The myth encodes a recurrent geophysical pattern, not a one-time catastrophe.
251 validated signals, 118 confirmed, 65 killed. What we found, what we killed, and why the kills matter more.
Methodology, validation framework, killed signals, synthesis papers.
Read methodology →Proprietary research, bespoke validation, cross-disciplinary advisory.
Engagement enquiry →Bulletins, lab essays, weekly digest of structural research.
Subscribe →Open paper corpus, methodology in code, validation framework.
Browse →Scientists in different fields using different methods converge on the same result. When that happens, we can be confident we are approaching the truth.
E. O. Wilson · Consilience: The Unity of Knowledge · 1998
