Of Crazy Numbers and Realease Criteria

he rolled his eyes and asked me, " Why 36? What is the significance of 36? " I explained that when I realized we could not have a successful beta with the current software stability, I wanted the team to realize just how many defects we were dealing with. A real number of defects would do this because this team didn't think in terms of other defects, just the one they were working on. The team had an interesting view of defects. At any given time, the developers operated on the assumption that the problem they were working on was the last one in the product. In hindsight, this seems a bit silly. But imagine you're in this situation. For the past five years, you've been stuck between two states: thinking that you're close to shipping and realizing that what you've just completed is not successful. Your senior management wants a release last quarter. Your customers want you to get to beta. Not all the developers understand the entire product and you're all exhausted. The testers don't completely understand how to test the product and they're tired, too. What would you do? You very well might cling to the belief that each defect is an exception to the rule that your product is ready to go. You would keep your head down, work like crazy, and hope that this defect is the last one. I wanted to burst this misconception. Rather than hope we were on the last problem, I wanted to create a buffer of problems and give the project permission to ship with some known and acknowledged problems. If we could agree on nonzero defect criteria, the team could get out of the rut they were in. They could stop thinking that they had created perfection and stop being disappointed when they did not achieve it. We would then be able to move the project to beta and ultimately to release. In truth, my original suggestion was that all known bugs be entered into the bug tracking system and that there be fewer than 40 open high-priority bugs. When I made this suggestion, I was called into the division manager's office. We had J ames Bach and I have had an ongoing, multiyear discussion about my definition of release criteria. He thinks I sometimes come up with crazy measurements based on no apparent foundation. He's right. I do. …