Here are my responses to some of the thoughts in the thread:
The calculation of ratings' changes after the game end are based on a player gain victory vs every player. [5th player loses a lot of ELO, etc...]
Seems fair to me... In a 5-player game it's much easier to avoid 5th place than avoiding 2nd in a 2-player. So getting 5th should give a bigger penalty.
ELO prevents people from playing, especially in luck-based games.
Playing ELO-off seems a good solution. This works great in 2-player games, especially if you just want to learn from the master and don't care too much about ELO. I can see that it might be hard to start a game with more players and ELO off.
ELO prevents high-ranking players from joining tournaments. They are afraid of playing with people with far less rating. ELO doesn't have the maximum of rating's change in one game.
The solution I've seen is that players start ELO-off tournaments, which I've been happy with. I think ELO-off should be the standard case for tournaments.
ELO promotes opponent selection. Players try to play with convenient opponents to increase their rating.
Literally all rating systems seem to have this problem.
ELO promotes table restrictions. You can often see players with rating about 1610 that restrict table for 1600+.
You could start your own game without restrictions then? Some games I wouldn't play without restrictions like that. In some games, I want to count on other players denying opponents and playing with more strategy, and restrictions are necessary for that.
ELO doesn't show how good player is. In some games 1850 is very high level. It depends on number of active players and history of playing the game.
ELO does vary by game. Games with more luck have a lower ELO spread. I don't think number of active players has much effect, actually (for example, check out ratings for Dark Agent which has few players - around 4000 plays total - but the highest ELO is 2400).
If you're worried about titles being impossible in some games (for example, there are no Master players in Senet), that could be a concern. But varying titles earned based on which game is played and its ELO-spread just seems needlessly confusing to me.
1. The rating of a player with big number of games should change very slow. If you played 200 games every new game give you only a little expirience. The rating should reflect your skill.
I would not like this. Oftentimes my first games are learning games and I lose ELO. I like the ability to climb the ranks quickly after learning a game (even if it means I can also fall quickly). The ability to rise/fall quickly with ELO is exciting to me.
2. The changes of the ratings in result of a game should take into account the randomnicity of the game. A victory in a skill-based game should increase your rating more than a victory in a luck-based game.
This already seems taken into account to some extent since in skill-based games you can achieve higher ELO's than in luck-based games.
3. The rating of a player who play too little should decrease over time.
I have played on sites that use this type of system and I hate it. HATE IT.
It makes me feel bad for not playing on a website for a while. It makes me less likely to return to that website.
If its implemented for specific games, it would make me feel bad for not playing a specific game. I have played almost all of the 90ish games on BGA and imagining all of those ELO ratings dropping if I didn't keep playing all the games. Ecchh....
4. (Not necessarily, but can make rating system more stable.) New players should start with no rating. They'll receive a rating after 10 games or after playing vs 20 players total in games with rating. Until they get the rating, their results won't affect the ratings of other players.
This would probably make people less likely to play with new players.
