Useful Metrics and the Problem with Performance Testing Programs: An Interview with Scott Barber