I think that You should periodically increase rating of players who have played many games regularly ! I say this because it's normal that players improve their rating with time ! Also some players have reported lower rating then it really is ! For example Bella Nagy is european 5. dan in GO and here is only 2. dan and that is the strongest player on this site in GO! You should get some information about this somehow ! ;-)
I don't see the first part of the remark. Rating is a system that estimates the power (this also means that the rating here doesn't have to be the same as 'in real life') A playzer getting better, and with a rating that is not reflecting his skill will by playing and winning games adjust his rating himself. I mean that is what the rating system is about, isn't. You might put other info (like real ratings) in the info of the player.
I am not saing that rating here should be the same as in real life, but it should be as close to it as possible. It is the fact that if someone is playing a lot of games he is going to improve more than someone who wasn't playing for a long period of time. That all should be taken into consideration ( and in the formula )when calculating new rating based on the previous rating and on the result of the game.
As I understand it, rating is recalculated with each game finished, so consequently people who play a lot will change their rating faster then players playing little (except that if you only played few games, the difference will be greater), I still don't see your point, actually... except if you mean that their rating only changes when games are finished, and thus slower then in 'realtime' games, but that doesn't trouble me.
I think this rating problem is not really a problem, in fact, i'm here to play and enjoy playing (i guess we all do so). Being concerned about rating is not showing how good i am, is more like showing off!
I cannot judge anyone by rating, rather by game. Let's not do a "job" from this beauty!
I am not trying to make things with rating complicated ! I simply think that rating should be as real as posible. Rating should be earned with experience, not only with actual strength at the moment.
I thought that is the main (and only) purpose of the rating: show the actual strenght compared to other players IN the same cummunity (same rating system). The system evolves like it should. Okay, it may take some time in turn based games, but the number 5 of europe will reach the 5 dan (or higher, if the rest of the european hotshots don't play here).
Does anybody knows which is the formula for rating in Go ,
because the discussion fall in abstract :) .From my modest
experience on this site , when two players play a go game
here , the rating for the winner increase with one level
and the looser rating fall with one level.This sounds me like in Murphy`s law "The intelligence is constant on the
earth , and the number of people grows day by day " ;0)
[2002-07-07]
=RE: RATINGS=
Here's the formulas that's used on the site: http://www.playsite.com/t/games/board/hex/rating.html At least in Hex, but I think they're used in all the games (even in Go, but there it's rounded to Kyu's). ----- Hope this helps.
I am stil hoping that Rio should take into consideration some way to increase rating to players who play more, or to players who win a tournament, or to players who lose because thay also improved in the knowleadge of the game and thay would play better in the future because they have learned something and their rating is in the fact better. Community - I would say NO ! Rating should be as close to the rest of the world as posible. We shouldn't be so conservative in nature! ;0)
But a person that is 'European 5 Dan' is rated in a limited group. Ohterwise he would just be '5 Dan'.
Now he's Euro 5 Dan AND Little Golem 2 Dan. How he's rated in Asia, I don't know, Niether do I know his worldrating. Perhaps he's even lower over there (sorry for speaking in third person if You are reading this)!
I think that rating here cannot respond to rating in real life. On such a site like here the results of a player depend very strongly on average thinking time per move. Some players look at a positon and move very quickly, some players print their games and then analyze positions at home. It means that average player in real life, analyzing every position, can be here very strong and on the other hand very strong player in real life, moving quickly, can be here only average...
I think the Euro 5 dan would be Asian amateur 6-7 dan.
There is another problem with 'real life' rating... In Asian Go rating you con't drop back since a new level is earned and given to someone (like in Asian fight sports the colors, and dans of a band). Here it is 'chess'-like, which is different. I would add another comment in favor of the experience is gain, even by losing... In fact a player not playing reversi is on 1500, a player who plays and wins and loses some games, could well be lower 1470, let's say, while he should be considered the better player. Still the system seems to work good, to me
maybe adding experience
along with rating
that way the players can see if they are getting better in spite of losing alot of games
we coud track our rating and experience level through out the games and tournaments
and the games that aren't rated maybe atleast add exsperience
that way it would be pratice for the tournaments and the players could see which games they are better at
Why the champion ( winner of the championship ) shouldn't be avarded 1 dan or 16 rating points ? After all he showed that he is better than anyone else on this site ! ;0)
reg. tournament games
or even non-tournament games
gaining exsperiance would show players how well they are doing
i really don't see the harm in atleast suggesting it
I think that reasonable suggestion would be to increase category of players that play regularly at some rate. For example if someone has played 33 games he or she should get ( be awarded ) 9 rating points.
Why? I don't think there is a direct relationship between total # of games and skill! The only relevant relation with skill is the number of games won/lost combined with your opponents rating. And that is already in the current rating system.
I agree with klaashaas...
You can play 100 games with getting better skilled. While if your rating is not reflecting your skill, the games you play will adjust that by the system. The only thing which might help is to indicate somehow that a rating might be off in case somebody finished less then 10 games or so, with some sign. But if you played more then 10 games the rating should be a good estimate (I'm already better rated, than I should, in reversi, by playing a lot).
I have visit other sites to play Go that post a message saying that ratings on their site tend to be higher(lower) than face to face ratings. This allows the new player a better chance of guessing what rating to start at. After that initial rating the system should be allowed to control the ratings and we should all just have some fun.
I guess here you have to indicate your rating somehow, and then the system choses a initial rating... if that is done properly the system should take care of the rest... So I think it is just fine here. (Maybe for Reversi, one could also distinct initial rating for Beginners (1500) and experienced players (1700) in such a way)
You can't say 1500 is a beginner and 1700 is an experienced player. It depends on the the skill difference between all the players. For example: if 90% of the players is a beginner, it's very likely that a player with a rating of 1700 still is a beginner (with more luck than the others :-). I don't think you should try to translate ratings to something concrete like beginner/pro: that's impossible. You can only use it to compare with another littlegolem-rating in the same game.
that's true... it is only strange Smaughster is only 1500, I meant to say that people when starting here could indicate a bit their rating, although it's not really necessary. Rating always is only an estimate
Great idea, if he had a higher rating, I wouldn't lose so much points by playing the Smaughster. One way to solve that, is to give the rating a 'not established' status when 10 or less games are played. I know this is done elsewhere with succes). The ratingformula with 'not established' players is different than with established players. Example: if a newcomer (1500) beats a 1900 established player, the newcomer's rating will be between 1900 or 2000. The established players rating will not change significantly. So the first 10 games are used to find a propiate rating for a player.
Hmmm I also have experience with sites using a provisional rating system for beginning players and have seen that it can easily lead to inflated ratings. I'd prefer to slowly get there instead of the inflated rating problem.
Looks like there already is something like provisional rating: "xxx old rating: 1500 newrating: 1486 __________
klaashaas old rating: 1539 newrating: 1540"
People that have less than 10 games should have "provisional" rating because their rating is unknown and nobody cheks their rating. You shouldn't ask and thrust them when they say their rating. People who play regularly and have more than lets say 100 regular games should be threated diferently because if You don't award them for playing regularly at this site You encourage them to sign off and then sign on again and again, with diferent category (rating) each and every time! ;0)
The rating in first 10 or so games HAS to be provisional. If someone wins the championship or the Monthly Cup tournament he HAS to be awarded with some number of rating points because he has showed that he is THE BEST ( he is on the Main screen ). If not the rating is not corect, and there is no motive to win in a tournament! That's my oppinion.
????? Why should there be extra points given to the tournament winner? The player has proved best in that particular tournament, but that does not necessarily say he/she is best overall player. The rating system works fine as it is (except for some problem with players signing up with wrong rating). Dont mess with it!