Bug life times

by Karel Thönissen

Earlier today I wrote that we do not have a quality problem. The interesting question is: how do I know? Not finding many bugs is an indication for good quality code, but also for bad quality assurance practices. So where do we stand?

I am not suggesting that we are perfect programmers, far from that. Our first cut is not better than that of any other programmer. So when we first run our software, we find many bugs. However, the telling bit is that after the first run the number of bugs found almost instantly drops to zero: we hardly ever find a bug in code that is more than a few days old. This is in code that is dog fed, reviewed, unit tested, reused, etc. We are not talking about idle code here.

Unfortunately, I do not have the hide-out time statistics of the bugs we find, our systems do not allow us to obtain that information easily (nor is their an urgent need for that). However, I must think very hard to come up with a bug that was found in our code after a module was deemed ready for release.