Wishy wrote:
And we're back to grinding a few maps, plus if for some reason you don't happen to like most of the maps you're likely not gonna play for a whole cycle. Just get a huge pool and make people download lots of packs.
I think the realistic problem with this might be that, as someone pointed out in this thread earlier(too lazy to quote, sorry), players won't bother to practice all maps in the huge mappool before queuing, which makes it an imperfect elo system. Because the elo system was originally designed for chess where there is only one kind of board(8x8) and one set of pieces(king, queen, 2 rooks, etc.), it fails to work 100% in other systems with different backgrounds.
Take LoL for example, and put aside the fact that LoL is a 5v5 game, not 1v1, for now. There are slightly more than 100 champs in LoL, and I'm pretty sure nobody plays all champions in his highest skill level; most of the time, a player is only good at several champs, not so good on others, and honestly have no idea about some of the champs. If the opposing team decides to play with a champ that the player has no idea about, then his game has already been affected; ideally, this shouldn't happen in an elo system. One may argue that this is the same with chess players coming up with 'cheese' strategies to win games. It is not. Cheese strategies in LoL would be CLG NA's 3TP 1Promote or AZB's lane swaps (which actually now is a common meta in the pro scene). The set of pieces in chess remains the same always, and this is a premise for the elo system to work.
Back to osu!. 100+ champs in LoL already make the elo system imperfect, which means that we can expect a mappool with 100 maps to function imperfectly. Nobody is going to practice all 100 maps, they will most likely just practice maps they are going to pick and hope that they win all their picks and try to take at least one game from the opponent's pick. Quickly, this could come down to a game of "who has practiced the tiebreaker map more", and while this system may be an accurate portrayal of who is the best player on the tiebreaker map, it is certainly not an accurate portrayal of who is better at osu!.
Reducing the size of the mappool reduces this inaccuracy. As the size of the mappool is reduced, players are more and more able to practice all maps, which makes for an actual skill matchup. Of course, if a player does not like the mappool, he may choose to skip the entire mappool and wait for the next one. If we are to implement the elo system in osu!, this cannot be fixed in any way, because osu! is not chess.
Think of it as this way: every week in LoL, rotation champions are announced, which players can play for free even if they do not own the champions. If LoL worked so that players can only play rotation champs in ranked games, would it be a better representation of the elo system than it is currently now? I think the answer is yes. One may point out that the games will be boring as shit, and this is true for LoL.
Not for osu!. There aren't strategies for single maps in osu!, you just hope you don't make a mistake. There is no element to make the games boring in the first place.
After writing this, I'm now convinced that the elo system is probably a bad way to represent any skill level in osu!. I just think that having a small, changing mappool for periods of time is a more accurate implementation of the elo system into osu! than having a huge, fixed mappool.