Manual Tests Are Unreliable

When manual testing is subjected to schedule pressure, the quality of the testing goes down. People begin to cut corners, omit tests, and miss problems. This is the kind of dysfunctional behavior for which traditional software development is famous. All those manual test cases look great on paper, but when crunch time hits, all the paper goes out the window. The warm, comfy feeling the manual tests gave us by promising to keep defects from getting through to the customer is replaced by the burning flames of perdition (which can be even worse than a bad hair day).

It's better to have no test at all than to rely on one that might happen or might not, depending on how the schedule goes and how attentive we are at four o'clock in the morning. Instead of counting on this dubious mechanism to eliminate defects from the system, we must omit the defects in the first place or find them immediately. Which is, of course, exactly what the XP unit testing practice is all about when it calls for the unit tests to be written before the code (omit defects in the first place) and for a 100% pass rate before code is integrated into the system (find defects immediately).