Intel Report 005 describes synthetic consensus loops: repeated frames, ranking pressure, context collapse, and silence penalties. What signals help distinguish real agreement from algorithmic repetition?
