Well kaner, isn't it just a matter of limits?
Take a popular variant like Colonial for example. According to the current stats, China has been defeated in 73 games out of 116. We can say that your expectation, when you start a game playing that country, is to be eliminated 2 times out of 3. So any China's result different from defeat looks heroic, but the system doesn't rate it differently from the same result of Japan, whose islands in its corner look almost inaccessible. Japan has only 19 defeats and its solo+draw ratio is double than China's. So I wonder why on earth 2 equally rated players have the same Re Ø when one is China and the other is Japan.
I'm just saying that, instead of setting a limit and keep unrated the variants beyond such an arbitrary limit, it would be better to find an adjustment for any inbalance.
Which won't be easy to make and it will never be perfect, but it will be much less inaccurate than just pretending that all the games you rate are as balanced as a Migraine or a Pure.
- - -
On an overall view, Oli, your score will _not_ "even out after a few games".
I'm the #10 just because I drew Britain in a Colonial, Almoravids in a FirstCrusade, Spain in an Abstraction etc., and it's going on since months. Maybe since a year or more.
I could be even better placed if only I used to join games without anti-NMR options enabled and take advantage from NMRs, since I would be much more reliable than my opponents. Playing low-quality games is very rewarding in the long term and I would do it, if I cared about my ranking more than about the quality of my games.
So, IMHO, this supposed "auto-fix" feature of the system can't be considered a panacea for any flaws and it shouldn't prevent you from implement further enhancements on this new system.
Especially if you plan to make of this the only points system in the site and you plan to build features on it, like the one proposed by DL, which would be fantastic if only this system was accurate enough already.