Persitent Patient Parnoia

The recent trading software glitches has created a buzz about testing computer systems ( e.g. Knight loses $440 million).  I was fortunate to be trained at Accenture in proper system testing.  It was always a battle of cost benefit trying to convince clients on the proper way to assure a smooth conversion of a new computer system.  The potential business disruption, impact on employee confidence and acceptance, and even revenue loss factored into the risk of less testing.  There were unit tests, string tests, integration tests, user tests, volume tests, stress tests, and complete system tests to perform.

Obviously testing every logic path in a system is impossible (and not practical).  But only testing the most probable paths is equally not practical.  Many clients just wanted to use historical data to test the new system.  Typically this was the worst test methodology as it tested millions of transactions within one logic path instead of fabricating unique test data that would test thousands of logic paths once.

Susan has always commented on my intense skeptism and lack of confidence in so many things.  I think this comes from a career of testing and discovering system bugs. 

It takes patience, persistence, and partial paronoia.