I'll never be a big supporter of stats by themselves. There's just so much more that comes with them that you have to look at like 3 stats together plus see the situation to really know whats up.
And the problem is people are often looking at them without considering those variables. so they make assumptions that could very well be wrong.
Plus you can pretty much make stats say wtv you want to say when you know enough.
A good example allllll the articles from Habs blogs about how Subban deserved the Norris, he was not playing the PK that season and was not playing top competition, how in hell can you justify winning the "best overall dman in the league" without playing PK and against top lines? They use stats based on /per 60 min and not breaking down PP, PK and 5 on 5 comparing with Suter and all the other guys and all kind of stats that if you know enough you can pin point whats wrong with it.
Subban was not playing the PK, was playing the full 2 minutes of PP, was not playing top lines and had most of his zone start in the offensive zone. And that season he didnt play a single second against West teams and playing in quite a crappy division.
lol so obviously he had better numbers black on white or was hanging with those guys cause his assignment was wayyyyyy more favorable to him.
The issue is never with the statistic itself, but the analysis and what you interpret it to be. Your "good example" isn't an issue with the stats, but with the analysis of them. I haven't read the piece but let's say you are right and then we have poor analysis and interpretation, of which there is plenty. That is not an indictment on the stat itself. The research and analysis that made shot based metrics popular is solid and there is a reason why people have gotten jobs in hockey using it as the primary category for analysis.