> pretty big change performance-wise for the better or worse? what is a > potential negative outcome of dirty reads? dirty data...dirty results -- ray