Biggest problem I have: Years of finding surprising bottlenecks and bugs where you least expect it has lead me to question every line of reasoning\anecdote people say unless they are able to back it up with something sufficient.

Maybe it's the correct thing to do, but it also annoys the shit out of people when I ask them if its possible that their single data point could have been an outlier or just a complete misinterpretation.