Defect Prevention Experiment
Experiment: defect prevention
Problem: When we test our code, it is always a small subset of all of the possible paths. Therefore for every bug test finds, there is a corresponding number of bugs we never find! If we test 10% of our paths and find 10 defects, we can assume that the other 90% of the paths we didn't test contains 90 defects! This means they could be found in production!
Hypothesis: If we study the patterns in our defects and come up with new techniques to prevent or eliminate them, we not only eliminate 10 bugs, we eliminate the 90 as well! Test cannot be about find and fix and must become find and prevent!
Time Frame: 12 sprints, pivot or persevere in 2 sprints
Typical Outcomes: decrease number of defects found in production, number of defects found during test of a specific defect category
Skill Growth Criteria: You have harvested a list of real defects, listed them, studied them, found categories, created a pattern to eliminate the worst category, and saw that category drop or disappear during subsequent test cycles.
Shirt Today: seibels cove