I think a good software analogy for this would having a piece of multi-threaded code that accesses a shared resource using locks. There will probably be some w

I think a good software analogy for this would having a piece of multi-threaded code that accesses a shared resource using locks. There will probably be some waiting involved (analogous to defects). The engineer determine wha rate of waiting/working is acceptable, and if the software actually does perform like predicted. If it does not, as in the wait(defect) rate is higher, it might indicate a genuine bug in the system (like locks not being released in a timely manner), or a failure of planning.