Can we just allow the kingdom selection methods to be very external and sandboxy and ask for several leaderboards? Default leaderboard is when kingdom selection is the way it is now. Then you have a koala leaderboard, a giraffe leaderboard, and a zebra leaderboard, and all of those only consist of games where the kingdom was handpicked, but that can have a significance people decide upon later. koala might mean that it's fully random but people are gentlemanning out of IGG and stuff, giraffe might mean an isotropish veto mode was used in an external app, zebra might just mean it was a game for dominion league and hey lets see if the ratings come out different when people are playing mega serious. Dingo might mean LastFootnote style kingdoms and crocodile might mean drafted kingdoms. Then you don't have to ask MF to implement any of the kingdom selection methods or the system of verifying a certain kingdom selection method was used, all they have to do is instantiate several other leaderboards in addition to the default one, which shouldn't be that much harder than instantiating the default one.
The game would only count for a certain auxillary leaderboard if both players agree that the game meets the requirements of that leaderboard. So when some guy makes custom trap boards and asks people to play with him all game, that never makes it onto Dingo board because other players will either be like "wtf does Dingo leaderboard mean" or "I read on f.ds Dingo board is for such and such kingdom selection method, I don't trust you that that was how this kingdom was selected."
Well, anyway, it's just an idea I had. And I think stuff that requires minimal amounts of work from MF is a good thing.