If you want to go straight to the fact, just read the "Important part".
Silene and Pedros, you're definitely right. Some people above you don't understand what elo is.
I'll try to be clear, there might be some langage errors in such a long text I'm going to write - please ask me if they're something that isn't understandable. Since I don't write that in a straightforward way, just tell me if it seems I didn't finish a sentence or an idea.
ELO is a mathematical formula which takes the 2 previous ELOs, the result of the game (who won) and gives 2 new ELOs. In addition, for chess (I dunno how it's implemented here), there is already a coefficient which takes into account the "experience" (numbers of games already played). This coefficient (a round number) is higher (40) for new players (less than 30 games) : allows new players to get to their real place faster. This coefficient is also lower (10) than average (20) for people who already went above 2400 once (whether they're still above 2400 or not). It alows people proven solid players not to be taken down too easily. Since the field of players on BGA is not big enough to get a consequent number of players above 2400, I won't speak more about the coefficient for those real solid guys.
I don't know if BGA have different coefficient already but I guess they probably do.
Anyway, you guys want to make this coefficient
lower for beginners ... this is the contrary of what is done for chess. Some also want to make beginner start with lower elo ... Basically, you're seing it from a self-centered point of view. As Silene was depicting, all our ELOs are a kind of (swimming-)pool : when a new players gets in, it makes some waves, then after few seconds, everything's calm and even again. And if each new player is "smaller" (lower elo), it will just make the whole pool smaller too. The pool is fine. But you are a small water drop that takes this wave and gets stirred. In the process, you get "hit" by another water drop (the new player, with is still moving because it has an anaccurate elo), and you'll hit other water drops (other players) because your elo is now a bit inaccurate too. This is the wave spreading. But don't worry, the system is stable. Maybe after 10 new players, who appeared to truly be masters, dived in and after the wave got blurred ; you'll have 10 less elo points, but so will have the other players of similar level.
Mathematics are cool in a pool (sorry I put that in bold but it sounds good
).
Smaller coefficient would make the wave smaller ... but longer. This is worse for the new player, and overall for the pool.
Of course, as I just explained, the single water drops (you), gets "hurt" :/ But you shouldn't care about that, this is not what matters. Don't be offended, I'm not saying you guys are selfish : I think you're seeing it that way because of the restrictions. ELO is made for ranking. And ranking is made for competition. And competition is good for 2 things : playing with guys of your level, and distinguish the best guys (seen with labels GoodPlayer/Expert/Master and the prestige points related to them). Restriction takes place in the competition to use elo and allow you to play with players of the same level. And the current restriction system (1600/1800/2000) is too rigid. This rigidity makes you look at those artificial 1600-1800-2000 boundaries. The rigidity makes the wave caused by a new player looks uncomfortable and dangerous. It divides the pool in 4
Like 4 different pools of 4 different colors. It creates discrimination (I know I sound dramatic - just kidding). And the water drop is afraid it could be pushed in a pool in which it doesn't fit, founding itself with strange water drops, with a strange color. But those color are superficial, we're all brothers and sisters
(yeah, I went far). Let's break down walls which separate siblings ! Let's make a unique pool again, where we can play in the waves like children in the sea.
IMPORTANT PART
For me,
the center of the problem is indeed the "restriction", as the subject of this thread points out (creating new category). Yes, a guy with 1620 elo wants to play with a guy with 1585 elo but doesn't want to play with a guy who has 1795 or 1450. Not even talking of people with 1304 elo and 1596 who are supposed to fight in the same category.
So may
I suggest that player could set a custom elo-fork when restricting a game ? Of course, we'd keep the "bands" (as Pedros calls them) for Prestige points. Beyond that, what should be a fork ? I see several options :
- totally loose system where a player can input the values he wants (I'm 1589, I want players between 1557 and 1632).
- bounds of the fork can be ticked on a scale from 1000 to 2500 with 25- or 50-points steps, with possibility to let one of the bound open. Or instead, of ticking, could be sliders on a scale.
- make a whole new simply way to restrict the game to players whose elo is +/- 50,100 or 150 compared to the table's host's elo. Again, we would just have to tick "restrict to player who have +/- 50 elo points than me", "restrict to player who have +/- 100 elo points than me", "restrict to player who have +/- 150 elo points than me".
I like the last option best. I even think we could go for only 1 option "restrict to player who have +/- 100 elo points than me".
Since categories seems fine for me for some games (I think of Hanabi, which is collaborative and compares the players' elo with a fixed game's elo, depending on which dificulty we're playing), we could also keep current category restrictions. In the end,it would add just one button. Not much complexity here, and a fairer way to restrict.
END OF IMPORTANT PART
Of course, as it's been modified by the guys of BGA last year, and whatever option is chosen, one player couldn't set a fork in which he wouldn't be included. And, of course, to restrict is optional, as in the current system.
Since people want to create a new category to have more accurate restrcition, this new "restriction system" would allow us to avoid the creation of a new category for players below 1300 or 1350, which could make them feels bad and wouldn't be logical regarding the prestige-points giving labels - I mean, with an "average" label giving 0 prestige point.
The new restriction, though, doesn't bring any novelty about new players' plunge. Makes things look less dramatic but you'll still get a little bit "hurt". I'll just detail a bit more how I think about that. I wasn't totally honnest about that before
In chess, usually, a player gets his first elo at the end of his first tournament ... So it can be set to a number quite close to his real worth. Here, we can't make that. Our ELO is based on 1500. Bascially, players who are good are inclined to keep on playing and those who are less good are inclined to stop. That's why it can look like average elo is above 1500. But if you take all inactive players, 1500 is the average. This is how ELO works, changing the "1500" landmark would just move the problem a bit further. As said in other posts, setting to 1400 would make it better if the new player is a real newbie but way worse if the new player turns out to be experienced. So basically, we can't do much about that
Well ... I see a solution but it doesn't fit with the prestige points and a ultra-competitive spirit, and possible dishonnest guys. I'm fine with that since the main purpose for me is the game but I already hear a lot of protestations from players who are more attached to absolute ranking. The solution would simply be
to allow new players to set their initial elo by themselves. Of course, this wouldn't be perfect, be it would decrease the "wave", the volatil effects of playing against new players. To avoid people wondering "what the hell, I don't know anything about elo and they ask me to give a number", a new player, in his first game, could chose from
"never played that game before=1400elo", "already played some games=1500elo", "experienced and rather good=1600elo", "I'm so smart I'm beating all my friends irl=1700elo". But even like that, it could be seen as a nuisance for a new BGA-user or even regular BGA-players. Dunno ... maybe can someone think further about that
To finish, I'd just precise again that ELO is initially a system for 2-players game, and is quite fugace when it comes to rank people after a 4- or more players game (hello Saboteur, Perudo, Elfenland, ...). It's even more fugace when it comes to games who have a greater "random" factor than chess ... i.e. main games on this site. I don't know anything about other ranking systems and I'm not a big mathematician so I can't propose anything else (and honnestly I'm rather fine with the actual system
). Anyway I feel it would maybe make sense to make a lowering coefficient for games with higher number of player and/or for games with a higher luck factor. This coefficient would be related to the number of players. For luck factor, difficult to judge how high it is, but everythings that include dices or other random generators could get a lowering coefficient.
Please tell me someone read that ... took me some time to write XD
