Or maybe know other things that you don't know...
There are *tons* of things which I don't know, but it would have to be pretty miraculous secret knowledge in order to offset what we do know.
Your idea of categorising the different points is interesting but relying on a mathematical model is dangerous as the info is usually inaccurate.
Well, we can only go by what the records say. Unless there is a good reason to doubt certain stats, then we should tentatively accept them.
Some of the records are pretty good. For example, let's take all of the candidates for top pilot that anyone can think of, and rank them according to the following statistics:
1. Total number of kills
2. Total number of combat missions (surviving more missions = indication of being a good pilot)
3. Ratio of kills per mission.
For each of these categories, we have a different ranking. Everyone knows the ranking for category #1. #2 isn't *that* hard to find out. Once you have #1 and #2, you can derive #3.
If we have these three different rankings, then we can do what I suggested in part one of this poll: Create a weighted metric on these three rankings, and then iterate over all possible weightings of these characteristics for some small step value. So for the weighting 100%,0%,0% the ranking comes out to exactly the same as the normal kill ranking. Then we calculate the rankings when we use a weighting of 99.9%,0.01%,0%, and so on, keeping a running total of pilot rankings. We iterate over the entire area, and in the end we come up with a pilot ranking which is about as robust and comprehensive as you want *for those three characteristics*.
We can add as many categories as we want, provided that we have reasonably accurate rankings for that category.
We could also incorporate some way of scaling the values according to quality of enemy faced (eg. shooting down a Japanese fighter in 1945 isn't worth as much as shooting down a German ME-109 in 1940.
If you guys are interested in helping, then I'd be up for writing the program which computes this. I'd need help gathering all of the statistics, though. There are enough people here so that this is doable, though. Who knows, maybe we can even publish a paper in some history journal or conference. It would be interesting for ww2aircraft.net to have a publication!