Since the first recorded software failure in 1962, the prevailing approach to deploying software-intensive systems has been to hope for the best—only to be proven wrong repeatedly

If you can predict weather down to the hour, you can predict software defects