It would be nice to use these results see some 'good' implementation, rather than just highlighting the worst offenders.