Chapter
TRACE: Contrast-Guided Discovery — Overview
An accessible guide to a proposed framework for using conflicting accounts to build measurable explanations, lock predictions, and seek observations that could refute them.
An unvalidated research proposal for turning conflicting accounts into testable explanations, committed predictions, and discriminating observations.
Research proposal · Version 0.1 · 15 November 2026. TRACE asks whether explicit source-conflict localization improves predictive discovery beyond established contrastive and abductive methods. Its four stages align questions, account for source dependence, generate rival explanations and investigate committed predictions. Shared records connect every stage to the proposed evaluation. No TRACE implementation or experimental results are reported.
Start with the illustrated overview, then read the full research paper and proposed evaluation specification.
TRACE builds on earlier CNS proposals while shifting evaluation toward prospective predictions and discriminating evidence. The CNS 8.0 framework and CNS 7.1 / GCTS material remain available as prior work.
Chapter
An accessible guide to a proposed framework for using conflicting accounts to build measurable explanations, lock predictions, and seek observations that could refute them.
Chapter
A proposed, unregistered protocol to test TRACE against resource-matched alternatives on contrastive correctness, independently predictive synthesis, and evidence-acquisition decisions.