The time clause in the contract is to specify the maximum wall clock time that the end user is willing to wait for the completed execution of the application. The decision to use wall clock time instead of CPU time was not arbitrary. Under our new paradigm, the end user is insulated from hardware and system details. Therefore, it makes no sense to have the end user think about such things as time sharing, job swapping, and all the other events that cause CPU and wall clock times to differ. Rather, we view the end user as having little if any prerequisite computer science background; all that the end user cares about is reliable obtainment of accurate answers in a reasonable time and cost. Since human interaction can have a major impact on wall clock time especially when input is entered from the keyboard), we give the end user the option of excluding that interaction time in the wall clock time. We also give the option of specifying time as an average over many runs, a hard maximum, or a degree of confidence that it will finish within that time.