Lol, we agree on one thing at least, weary indeed.
All I have to add, is you changed the stat by adding more variables, it loses it's descriptive power. The analogy was, you don't add variables (categorical or quantitative) to a stat which do not affect the description.
Here's another analogy. Suppose one of our NATO allies have troops in Afghanistan, but they are not there in a combat role. They send army engineers to build schools, bring water gathering technology, etc. Let us also say they rely on the security blanket provided by other nations. Now some agency releases a report. They find that coalition troops in active combat are responsible for 68% of violent deaths in the country (lets ignore the definition of violent deaths in this case, as it isn't important to this discussion.) Do you think it would be fair or accurate to list the hypothetical allies who do not contribute to the combat forces, and therefore by extension, not to the deaths described by that statistic?
That's really what I'm getting at. Whether or not you think the original statistic was useful or not, changing the stat by including an irrelevant variable is not even handed, and does not prove it's worth (or as you say, lack thereof.)