It's because software is incredibly amorphous. It's incredibly hard to compile meaningful metrics about it, partly because every project is different and used differently. We all know this in the back of our heads, and how incredibly daunting it is to squeeze out quantifiable data from it, so we instead don't tackle it (which is normal behavior for humans) and instead go with stuff that sounds and feels good. Pairing increases your code output Well, two it better than one, so it must be better! TDD prevents bugs? Well, writing the tests first sounds good to me, so it must be better!
Yeah, it's a dangerous mindset, and I don't really know how to get out of it, especially since we're all hypocrites to some degree.
Yeah, it's a dangerous mindset, and I don't really know how to get out of it, especially since we're all hypocrites to some degree.