Since I don't come from a CS background (IS major at a business school), I never really knew much about strict testing methodologies except "make sure your shit ain't broke", and thus did the best I could early on by just testing the most probably use cases for my code. It's worked well enough for me that I still test code this way. This article makes me feel a little bit better about what I used to think was my own noobie way of testing :)

This is meandering rambling from someone who doesn't know how to write unit tests. He complains about bad unit tests ("many tests tend to...") as an argument against unit testing. Yes, there are bad tests out there. But there is a way to write good valuable unit tests that protect you from future regression (thus future pain and needless effort).

Get better at writing unit tests, don't use this article as an excuse to avoid a highly valuable practice.