The Artifact Index is a live reading of cultural intensity. It does not score works, recommend works, or rank works. It describes the pressure they generate.
We treat criticism as signal archaeology. The data is the open record — Reddit threads at three in the morning, Letterboxd diaries kept for years, marginalia in academic essays, the half-life of a meme. None of that is truth. All of it is evidence of how an object is being lived with.
A high Obsession reading does not mean the work is good. A low Consensus reading does not mean the work is bad. A high Friction reading means the gap is open and probably will not close. That is information of a different kind than a percentage.
Art is not consumed evenly. Some works become shelters. Some become obsessions. Some become religions, wounds, identity systems, recurring dreams. Current review systems flatten all of that into percentages. We refuse the flattening, and accept the cost: the reading here is honest about being a reading.
Phase I, currently visible, presents twelve hand-curated artifacts with metric vectors set by the editors as a reference set. Phase II wires the vectors to live signal — Reddit and niche subreddits, Letterboxd diaries and lists, Rotten Tomatoes and Metacritic as consensus anchors, TMDb and the Criterion catalog for afterlife events. Language models are used only to cluster discourse and extract symbols, never to write criticism in the system's own voice.
The full method document will sit alongside the catalogue when Phase II ships. Until then: read the Lexicon. Each axis is a claim stated in language. Disagree with the language.