Claesar is absolutely correct. WebDiplomacy is not short of good and wide ranging ideas for improvement. What it *is* short of is people willing to dedicate the time to improving the site.
On the subject of GR and takeovers, it is clear that treating them the same as a game that you played from start to finish is not the best choice.
Part of the reason that takeovers are not considered at GR time is historical - for a long time, takeovers (and CDs) were not recorded in the database. It's not possible to reconstruct them accurately from the move log, either (for example, all holds is a legitimate and common choice for small powers on the stalemate line, waiting out for the construction or failure of the rest of the line).
We do have more recent takeovers - but if we only remove recent takeovers, then players who have been on the site for a long time will have an advantage.
We *could* do a ratings wipe, and start again. But I can't imagine that being very popular.
GR games are currently "worth" a predictable sum. If this amount changes because of takeovers (or not), it could have unpredictable results on the rest of the rating system.
But having said that, GR is not actually a very effective rating system (you can measure this by comparing the error in the predicted outcomes each game). I've written extensively about this before, but the major problems are that:
1) All losses are equal, regardless of who you are playing against
2) The amount a game is worth scales with the rating of the players you're playing with (including your own). This creates a bit of a rich-get-richer economy
A proper analysis of the problems isn't something I want to repeat here (it's somewhere in the archives for those dedicated souls who want to look for it), but I generally feel that if we were to modify the system, we'd be better off replacing the whole thing with something more principled.
Personally, what I'd like to see is a separate ladder for takeovers. Perhaps this is scored by the difference between what a position was worth (under the particular game's scoring system) when you took it over, and what it was worth when you finished the game.
This would especially incentivise takeovers of small positions - since they have the smallest possible loss, and the largest potential for gain. And I think that's a good thing for the site.