As I remember, a lot of people knew that the performance counters in 1.x didn't all work properly. As I see it, this is an issue that could slip by because performance counters aren't typically worked with in the main stream app development area.
Wally: I could see how one or two would have broke, but *most* of them is simply weird.
This explains a lot of some whacky behavior I saw on a project I worked on a couple of years ago; we never could track down some of the craziness we were seeing (mainly db connections/pooling). Good find.