Rank Factor classes categorize optimization hypotheses into seven core classes to balance exploratory edits using freshness weighting and candidate discovery via an automated Custodian model.