Infinitesimal Causality · Democritus recipe

From many local causal claims to a trustworthy causal foundry.

This page gives the first layer of the story. Scientific papers rarely hand us a single global causal model; they give local evidence, scoped mechanisms, and unresolved tensions. Infinitesimal Causality explains how to keep those tensions visible, while Democritus is one current recipe for computing the typed causal sections.

The Background Story

1. Evidence Is Local

A paper may support a claim for one population, dose, endpoint, or study design. It should not automatically become a universal causal fact.

2. Compatibility Matters

The foundry compares claims on overlaps: shared mechanisms, related outcomes, comparable interventions, and common limitations.

3. Obstructions Are Useful

When claims fail to glue, the failure is signal. It tells us what must stay qualified, what evidence is missing, and where hidden structure may live.

How The Layers Fit

Democritus

Extracts scoped causal PSR cells from literature: claims, mechanisms, evidence spans, limitations, and uncertainty.

Infinitesimal Causality

Reads causal variation as local directions and treats non-closing directions as obstruction geometry.

BRIDGE/SKFM

Provides one computational diagnostic for that geometry: bracket residuals, latent candidates, and warnings before promotion.

What The Detailed Run Shows

Admission

Which GLP-1 causal claims are promoted, qualified, or held for review.

Geometry

How BRIDGE/SKFM-style residuals explain why some visible causal stories should not be fused too early.

Audit Trail

Which gates passed, which scopes remain under-specified, and what corrections would make promotion safer.

The next page contains the detailed figure, TICKET transformer, metrics, role outputs, and audit gates.