Core concepts

Documentation Index

Fetch the complete documentation index at: https://wuweism.com/llms.txt

Use this file to discover all available pages before exploring further.

Understanding these concepts will help you ask better questions, interpret responses accurately, and use Wu-Weism’s governance features with confidence. Each section below is a brief orientation — follow the links to the full reference pages.

​

Pearl’s causal ladder

Judea Pearl’s causal hierarchy defines three distinct levels of causal reasoning. They are not interchangeable — each rung requires different assumptions and different data. Most statistical analysis — including most machine learning — operates only at rung 1. Moving to rung 2 requires an intervention model; moving to rung 3 requires a full structural causal model. Wu-Weism tracks which rung your question belongs to and tells you when it cannot answer at the rung you need.

​

Structural Causal Models

A Structural Causal Model (SCM) is a mathematical representation of a causal system. It consists of three parts:

• Variables — the quantities in your system (exposures, outcomes, confounders, mediators).
• Edges — directed arrows that encode which variables causally influence which others.
• Mechanisms — the functions that define how each variable is determined by its parents in the graph.

Wu-Weism builds an SCM for every domain it encounters. You do not need to specify the graph manually — Wu-Weism infers it from your question and domain context, then makes the structure visible so you can inspect and challenge it. Every inference Wu-Weism produces is grounded in the active SCM. If a variable is not in the model, Wu-Weism will tell you it cannot reason about it without extending the graph.

​

Claim Ledger

Every hypothesis Wu-Weism produces is recorded in the Claim Ledger — a persistent, auditable log of causal conclusions. Each entry includes:

• Provenance — the full chain from question to response, including the SCM that was active and the model that generated the claim.
• Confidence and uncertainty — a score and label that reflect how well-supported the claim is given the available evidence and model assumptions.
• Falsifiability condition — the specific evidence that would overturn this claim. Claims that cannot meet this requirement are blocked by the Falsifiability Gate before they reach the ledger.

The Claim Ledger is not a conversation history — it is a research record. You can return to it, annotate entries, and export it for reporting or peer review.

​

Counterfactual traces

A counterfactual trace is the step-by-step record of a do-calculus intervention. When you ask a counterfactual question — “what would have happened if X had been different?” — Wu-Weism:

1. Abduces the state of the system given the observed evidence.
2. Applies the intervention by surgically replacing the mechanism for the intervened variable.
3. Predicts the outcome in the modified model.

This process is deterministic and fully traceable. Wu-Weism exposes each step so you can follow the reasoning, identify where assumptions enter, and challenge the result.

​

All concept pages