Methodology
Observational analysis only. Not a fact-check.
Baseline is a measurement surface for public speech. A statement is captured from a verified public source, processed independently by three AI systems, and displayed side-by-side with source context. A separate consensus layer is computed after independent outputs are produced and displayed.
Diagram summary: statement input is captured with source and context, normalized, processed independently by three systems, displayed side-by-side, then a separate consensus layer is computed and stored as an append-only artifact.
The system captures the statement text plus metadata (source link, timestamps, figure identity, and context pointers). The input text passed to analysis systems is normalized to a single canonical string, so each system receives identical input.
Three AI systems process the statement independently. Outputs are never combined before display. This preserves separation between lenses and makes variance observable. No manual rewriting is applied to model outputs.
The app displays system outputs side-by-side, with sources and context visible. Context is presented as supporting metadata and links, not as editorial judgment.
After independent outputs exist, a consensus layer is computed as a separate surface. Consensus summarizes shared patterns and highlights variance, without overriding individual outputs.
Baseline includes a framing measurement surface across five axes: Adversarial/Oppositional, Problem Identification, Commitment/Forward-Looking, Justification/Reactive, and Imperative/Directive. These axes describe rhetorical structure, not moral character.
Baseline artifacts are designed to be immutable once written: the system stores inputs, outputs, and consensus as append-only records. This supports reproducibility and auditability over time.
Sources are presented as URLs to public records or verified platforms. The system does not replace source reading; it preserves the path back to origin for verification by the user.
This methodology page is a high-level description of the measurement surfaces. Specific feature limits and subscription details are listed on the Pricing page.