more accurate results, or more plausible-looking random results?