Most Testers, QA engineers, project managers and software engineering theorics try to match the concept of quality metrics with something close to the real world, so we have bugs per line of code (even if not all lines of code are created equal, as if python were equal to assembler), bugs per functional point, total bugs per man-hour... add to this if is blocker/normal/cosmetic, and so on. QA Metrics are, until now, very evasive to become widely adopted.
Then my good friend Nahualito discovered the following document, deeply buried in a oscure web site:
Neat, clean, measurable. Add a 2x multiplier if the WTF happens on end user, 2x if a crasher.