DL, I like your analisys and, as usual, you wonder right questions. I don't have an answer for all. So, what I want to say is that I believe you missed to wonder one important questions before typing your thoughts about stats.
Do we have the right stats?
You're right if you say that the reason why Egypt in WWIV hasn't won a game yet could be simply that games played are 27, while countries are 35. And that statistic is a science that needs large samples, so Egypt not winning a game out of 70 or 105 could still mean nothing. Evidence of this is Migraine, symmetrical map, that due to his small sample (7 games) has weird stats. The only conclusion you could get from that stat is that Sigma has been played by a noob most of the times (or played by a player weaker than Theta, because, in Diplomacy, the issue is not only how many good players there are into a certain game, but also WHERE they are).
Nonetheless, Egypt being ELIMINATED in 23 games out of 27 MUST mean something different than just "Egypt has been unlucky having neighbours stronger than him most of the times". Same way Cuba and South-Africa never won, but they were eliminated only 12 times being able to take part in a draw 9 times each, a third of the sample. While the only 3 times Egypt had part in a Draw look quit games, being them 25-ways-draws or so. It must mean something.
I believe that the issue with the current stats, aside sample's size that is an issue with any stat, is that they're unable to tell us what we want to know. They're too Diplomacy-minded and they divide results in W/D/S/E that is not an accurate way to measure strength of countries (and therefore inbalance of the map). Even less accurate is that ugly rating system (Performance = (15 x Solos + 5 x Draws + 1 x Survivals) / Games). If you're trying to understand the potential of a Country in order to evaluate a possible inbalance of a map, you must consider that there is draw and draw, there is survival and survival.
A survive with 30 SC looks rather different than a survive with 1. And we all agree that 2 PLAYERS in a x-way draw are equal, but their COUNTRIES' POTENTIALS are not, if one ended up with 30 SC and another with 1. So a Country Rating System that makes equal all draws and all survival is wrong.
Knowing how many SCs a Country usually ends a game describes much better its potential and therefore possible map's unbalancies.
Supply Centers.
That's what everybody fight for, each turn. A stat about each country's average ending SCs would start being meaningful after not many games and therefore it could be used for a balancing system. A sample of only 15-20 games could already give precious informations, any the size of the map be.
Think of a WWIV pick-your-country game. Egypt wouldn't be the last country picked up, because you'd know that, although risky, Egypt would be much rewarding in case of a good game. Good players would risk & choose it!
Stat would be better each time a game ends up, without ever getting perfect because statistics isn't perfect. So a system keeping into count these stats would always have some mistake, but a system pretending that all the variants are perfectly balanced would be a MUCH BIGGER MISTAKE.
Of course, such system should be sophisticated enough to handle some "corner case". IE, Countries of symmetrical maps like Migraine or Rats should always be considered equal, any their stats be. The same should go for maps with less than x games played, because their stat's sample would be still too small.
- - -
The perfect system does not exist. That said, I believe that any new rating system running alongside the current vDip-point system (perhaps with some fine tuning to it) must be welcome. Every Rating System would have its flaws, but together they would give a better idea of the player in question than just one system.
________
Update (because I started typing this post yesterday, right after DL's one, and I checked new posts just now):
- maybe we could try to merge all those systems into one, likewise AIRBORNE suggested. It's a good idea, but I'm scared thinking about how to do it and about related discussions! :)
- Obviousely I agree with LEIF_SYVERSON. He explained some points I had in mind much more properly than me.