The ratio versus difference optimization and its implications for optimality theory