Testing is a search process and a test suite is complete when the search has examined all the comers of the program. Standard models of test suite sizes are gross over-estimatesince they are unaware of the nature of that search space. For example, only a small part of the possible search space is ever exercised in practice. Further, a repeated result is that a few random searches often yields as much information as more thorough search strategies. Hence, only a few tests are needed to sample the range of behaviours of a program.