Methodology
Chorus currently tracks 1,058 active albums, 121 active sources, and 6,532 enabled reviews. Every public answer should be understandable without reading this page, but this is the source of truth for how those answers are built.
Chorus indexes professional review publications and stores them as source entities. Sources can be disabled, corrected, or reprocessed without deleting their history. Album pages link back to the source page and, when available, the credited critic page.
Chorus uses a single 0-100 score scale across album pages, metadata, APIs, and schema. Source-native scores are normalized into that scale before consensus is calculated. Chorus never mixes /10 and /100 displays on public critic-consensus surfaces.
If a review has no source score, Chorus can still use its text for track mentions, themes, and summary support. Unscored reviews do not create a fake source score. They enrich the answer only where the underlying review text supports it.
Confidence is displayed conservatively. It is an internal confidence-derived signal, shown as a percentage only when present, and should be read as “how settled this critic read looks” rather than as a promise of objective truth.
Awaiting consensus. No stable public answer should sound settled.
Early read. Useful signal, but not a durable consensus.
Consensus forming. Direction is visible, but the page should stay cautious.
Established consensus. This is where Chorus should read most confidently.
Consensus data is refreshed when new reviews are ingested or when an album is reprocessed. Public pages should expose the last consensus update time so readers can tell whether a page is fresh or stale.
Source URLs, scores, and metadata can be corrected. When a source is wrong or unreliable, Chorus disables or fixes it rather than leaving a bad record live. See the corrections page for the reporting path.