I can sort of see where our 20,000 backlog might have come from.
I want return to the earlier questions to Ms. Di Paola. You were talking about the pilot period, the 34 departments. It seems to me that the departments chosen were the ones most closely aligned with Phoenix already and that it wouldn't have been a representative group by which to judge how the project would roll out across departments that weren't so closely aligned with Phoenix. Do you think using a pilot project on a more closely aligned subset of the departments actually created false expectations when they didn't show the same number of errors that we're seeing now?