Not sure what the selection bias for this report is, perhaps that we care about code and believe in the value of static code analysis. Some interesting results in there either way.