[pron]

There's a fundamental problem with this kind of debates. Because we don't have a lot of empirical data and because human interaction is complex and often unintuitive, we don't have a good theoretical model of how the development and maintenance process actually works, the arguments tend to be based, in the best case, on unrepresentative and biased samples, and in the worst case, on personal aesthetic preference. And when we look at empirical data in the field, obviously whichever the dominant paradigm is, it will show the actual effect of programmers and shops at wildly different skill levels using it, while the non-dominant paradigm won't be able to show success at doing things better because, well, it doesn't have enough people at different skill levels using it.