We've had this discussion before, and you're right. However, the caveat that one of our stat gurus explained a few years ago is that K/AB is more appropriate than K/PA. I can't split that hair and explain why. But either K/PA or K/AB would be better than K/9.

Essentially, a poor pitcher can have the same K/9 as a good one. The lousy one achieves his K/9 by facing more hitters, and the good one is penalized for not giving up hits and walks.

I'd submit that we use K/9 because it's conveniently offered by websites and because it paints an easy mental picture. Six Ks in nine innings is easy to picture. Numbers like .166 aren't. What would be great is an adjusted K/9, which would actually be K/AB multiplied by the average number of ABs in a game. That would make it look like K/9 and eliminate the learning curve that would come with understanding the number.