An independent research initiative

Follow the path.
Question the claim.

The world is connected. Explanations should be inspectable. Causals Lab is building a versioned map of entities, events, relationships, and evidence—and studying how language models can reason with it.

Evidence before certainty✳Mechanisms and observations kept distinct✳Open to correction

Why this exists

A graph should explain its answers.

A line between two things can mean many things: ownership, a supply relationship, an observed association, or a causal claim. Treating them all as the same would produce a compelling picture—and a misleading answer.

Our approach keeps the type, direction, provenance, time context, and uncertainty of each relationship visible. The graph is a place to investigate a hypothesis, not a machine that turns every connection into proof.

Explore the ontology

How reasoning earns trust

Evidence has layers.

A plausible path and a past pattern are different kinds of evidence. We want people to see both—and the misses.

01 / MECHANISM

Why might it propagate?

Trace a route through events, physical flows, capital relationships, and affected entities. Inspect the meaning of each step.

Structural path
02 / EXPERIENCE

Has it happened before?

Compare similar historical cases and their observed outcomes. A useful record includes failures, not only successful predictions.

Historical validation
03 / JUDGMENT

What remains uncertain?

Surface alternative explanations, disputed edges, missing observations, and limits of the source material.

Human review

Four connected efforts

One lab. Many ways in.

A living project

Bring a question.
Bring a source.
Make the map better.

Spot an error, add context, or challenge a relationship. You don't need to write code to contribute.

How to contribute