Diversity weighting raises each Rank Factor Class's freshness factor to a configurable power to penalize over-tested categories and ensure the Bayesian optimizer explores varied content modification strategies across rounds.