You can normalize against anything. Can we normalize Linux against the fact that millions of people potentially could have reviewed the source code, and prevented the bugs? You can bias things with proper justification any which way. At some point, we just have to accept an imperfect method :)