Lab Experiment
Work In ProgressInformation Inspection
This experiment turns internet articles and web pages into structured diagnostic signals and a Reliability score, using weighted inputs and calibrated banding (High through Poor).
Each article is scored across dimensions like content quality, evidence density, attribution, headline fit, and specificity, with sensationalism and clickbait applied as penalties. These signals are combined into a single Reliability score using a consistent weighting model designed to reflect how well the article is supported and presented.
Signals are scored on a 1–10 scale, while penalties are normalized. Political lean and framing are shown separately and do not affect reliability. This is a diagnostic workspace, not a summary.
Use filters to slice the feed by reliability, bias, source, and tags, and compare how articles perform across different dimensions.
Recent scans
Reliability band
Limit the feed and metrics to a score tier.
Political bias (coarse)
Collapses the five stored labels into left, center, or right.
Source Domain
Narrow the feed to one publisher. Open the menu and type to filter domains; matches any substring in the name.