Hi Keith,

Making sure you are running the right tests and actually running those tests are often done by different people. It all comes back to outcome-oriented thinking (as someone at Microsoft just emailed me to remind me — having drawn parallels for himself between TDD and GTD). The tests should represent an accurate measure of the successful outcome for that particular feature or component of the system. Then running the test could involve a real person going through steps on a checklist, or an automated unit test (what TDD is all about) that can be run by software on software to verify the expected output for a given input is achieved (i.e. it worked). Automated versus real-person depends on the circumstances and what’s being tested. But devising the tests always comes back to defining success for that component — which is ideally always something you do the moment you define the project associated with it. Hope that helps.

Cheers, Robert