Rating / Discussion

Sub-page of Rating

Charles Matthews: Rank is expected to be quite stable, while ratings constantly vary. See rank and rating.

Frs: Why? What is the (common) difference between a Go rank and a Go (or Chess; or any game) rating? (I'm not asking about Go's unique handicap feature.)

Charles Consider that a rank is a title.

Frs: I disagree. In the area of double-digit-kyus a rank isn't a title.

Charles: I would treat it that way. Anyone is entitled to disagree with me, of course.

Charles Ranks of pros are awarded by a particular organisation. Only one such organisation, the Zhongguo Qiyuan, also publishes ratings.

Also you should notice that the primary function of ranks within a club is to set handicaps. That is, it arises out of the need to have teaching games. But in different countries this meaning will be treated rather differently.

BobMcGuigan: In my opinion it may be possible to determine handicaps entirely by rating, using the idea that 100 points of rating difference equals one handicap stone. Of course that assumes that the ratings are calibrated properly. And there is also the problem of compatibility of rating systems. I think ranks are an artifact of the spread of go from China, Japan, and Korea to the rest of the world. Another way to think of ranks is to view them as measuring peak performance while ratings measure recent performance. In the United States the AGA has no official rank system but does have a rating system which is used to determine handicaps in tournaments. It seems clear to me that use of ranks or ratings to determine handicaps is really only approximate at best. It provides a starting point, really, and handicaps would then be based on long term performance between the two particular players, not on their performance with others. We all know of players who force opponents to a larger number of handicap stones than relative rank/rating would specify.

Charles Well, Bob, I think I was trying to say what ranks are in go. Not what they might be. For the benefit of Frs.

Probably noticeably fewer that 1% of the world's players even have a rating. The trouble with starting from the analogy with chess (or any other game) is that it blinds one: I think one can probably have an interesting discussion of the differences between ranks in go and shogi, but to compare go ranks with chess ratings is fairly silly.

Go ranks are more like the GM and IM titles chess introduced in the twentieth century, but more finely divided. If one asks 'why so finely divided?' one does get to a question that is worth answering. Pros feel that a difference of a 'third of a stone' is something palpable in play, and that's the old theoretical step in going up one dan (professional).

John F. Also for the benefit of Frs, can I suggest the way to view the difference is that rank measures status, ratings measure performance. Status here means that you are acknowledged to have mastered a certain amount of knowledge consonant with a rank, so that if you are 1-dan but lose all your games (e.g. because you are old/ill/careless) you are quite possiby still a better teacher for a kyu player than a 4-kyu who wins all his games.

In Japan the status aspect goes much further. You can't refer to just Kobayashi. You have to write Kobayashi 9-dan. From the pros' point of view it is important in that it determines teaching fees, where you play your games and your pension (would you like your pension to go up and down in line with your performance?). There was an almighty row in the shogi world 20 years or so ago when amateurs tried to introduce rating lists that included pros. It is highly unlikely that ratings will be used in the closed shops of Japanese or Korean pro go simply because there is no need for them. In China it was different: the huge size of the country meant that a more objective method was needed. Recall that Chinese ratings were introduced for a specific reason: to regulate access fairly to the lucrative international events.

erik? I think I can satisfactorily explain ranks. You rank is essentially your playing ability, represented by a number. As mentioned before, this is useful in determining the number of handicap stones neccesary in a game, but it's not really for teaching games. Win or loss means nothing when teaching. The goal of teaching games with handicaps is not to make the game even, but to teach the student to use their territory better, and how to defend against invasion. the teach could just as easily use extra stones to teach the student to invade better. The practical use of ranks is to help players find other players near their own ability to play against, and the system is really much more than aproximate if you dedicate yourself. The more powerful you become, the more solid and well-rounded your play will be. That is to say that when you are new, your play does not fit that description; you will have strength and you will have weaknesses, and when someone near your rank crushes you, it will usually not be a question of skill, but a question of comparable strengths and weaknesses. A game that becomes a complex life or death fight between 20 kyu players will almost definitely end in utter ruin for one player (the one weaker at life and death), while the same fight between dan players will almost certainly be more balanced, because they play a balanced game. This being said, your ranks are basically a measure of your skill, and are used to help you find opponents.

On a personal note, I rose from 26 kyu (my beginning strength) to 14 kyu in roughly 3 weeks. I also almost never play people my own strength. I play stronger players, which is why rank is neccesary. To improve, you don't need wins. You need to be crushed, to see where you're weak and improve it. Ranks let you do this.

TJ: In answer to that last paragraph...perhaps. However, some of that strength MAY be a mirage, if you never play weaker players. Taking a couple months, I once rose to 16-15 kyu on an online server. Then, being very busy and preoccupied, I stopped playing for six months. My rank drifted to 13 kyu in my absence. I won a few games at that level, much to my surprise. However, when I started giving 2 or 3 stones to lower ranks, I started getting beaten.

This does NOT mean the handicap system is silly. Given that the rank system is based on the handicap system, it MUST work, minimally at 1 stone, or the size of a single rank is impossible to judge, rendering rank totally meaningless. Rank is used to find people to play, but playing with a properly set handicap does have meaning, and rank is used to find that as well, at least as a guide; those heavy handicaps CAN get tough, but surely up to 3 stones should be able to be handled before you're ready to advance a stone's rank if you've any strength at all.

I digress. My point is: I wasn't losing suddenly because the system was faulty, but because my strength was a mirage. I knew a lot about theoretical correctness, but I had no strength to fall back on when players gave me irrational moves which should fail...in short, I couldn't refute play from weaker players because I didn't have the reading skill and understanding to do so. I learned much coming back up from 18 kyu to 13 kyu, playing both sides of my rank to become well rounded (though perhaps not enough lately, must get working on that!).

It is great to play someone stronger to see the refutations of wrong ideas, of course, assuming you make the mistakes to start with. Mistakes you wouldn't consider making and their refutation, however, can only be learned playing those likely to make them. Weaker players. This strange phenomena I encountered may somewhat explain some people believing the handicap system doesn't work at all.

I know this is long, but I begin to suspect I had an experience that isn't very common. Perhaps I've learned a lesson about Go that is not usually driven home forcefully enough to validate the ranking/handicap system itself, since usually that lesson is learned with a much lesser gap between real strength and rank. I hope I've expressed myself well enough to be understood, because what I'm trying to write here is important to me.

Cheyenne:I believe that what happens if you only play stronger players is that as you said you do not need to handle as much irrational play. For instance you have a group that is alive if you play well, a stronger player may respect your "rank" and not bother trying to kill the group, while a weaker player may just go ahead and play in that group and then you mess up defending the group and it dies.

(Sebastian 15k:) Thank you for plea, which would fit nicely on PlayingOnlyHigherRankedPlayers. Regarding the difference between rank and rating, I am still confused. From rank and rating, it appears that rating is equivalent to a rank with subdivisions. However, from what JohnF wrote with regard to Japanese ranks, I understand that there is no rigid coupling between the two. Aren't therefore online ranks closer to ratings than to Japanese ranks?


mdh Reading through all of this, Rank indicating Title, Rating indicating performance. It would seem that in the Amatuer world, players would be more concered with the Rating. Ranks would not hold much of a significance. Yet another set of ideas stated above says that Ratings are insignificant and Ranks are used. I don't believe this can be true. While on that club card it may say I am 14 kyu but the card should change based on my performance, For example if I am winning more games then I am losing it would go up. If I am losing most of time, perhaps it would be adjusted down. So it is more like a slow response Rating than a rank.


An example to prove my point is IGS where you have a stated Rank but there is a Rating that moves up and down using the same scale terminology (kyu/dan). When you are looking at a players list to find a game, you see their Rating not their rank, so the rank serves no purpose after the initial use. My rank is about 6 stones lower then my rating. I havn't changed it since I first started.

TJ: A single stone is a big enough difference that you can have fluctuations due to having the flu, feeling off, not having had a coffee today, even trying something new you have to mess up really bad to drop a whole stone in skill, you can still be playing at the level of your rank. A rating that calls itself rank is probably still rank, or rather a best estimation of your most probable rank. The thing about ratings is that they tend to go from the top-down, so that a rating can decrease without any change in ability. This from the chess world, though: chess ratings aim to have, theoretically, half the people within the system above 1500, and half below. With go, where rank is a measure of ability starting at the bottom of the realm of possibility and moving upwards, once you achieve a rank it should not go down as long as you're in practice. If you've stopped playing go entirely for years and come back to it, you can't declare yourself the same old rank...if rank is a title, you've abdicated by abandoning go. Other than that, rank of a player should tend to go up only. A go server, it might dip, but that's because you're letting the server determine your most probable rank. Once you're established playing at a certain level, there's only up...something I like about the rank system relative to chess ratings.

In go, ratings should (I hope!) be based on the rank system, and not on a system like chess ratings, as a means of attempting to keep accurate estimates of strengths instead of slotting you into a curve, since the ability to do just this is something Go has that chess just cannot have. If so, I don't think that applying the "going up and down" bit to go ratings really applies at all. Is thinking they do a confusion with chess ratings? And if based on rank, shouldn't we ignore anything that doesn't declare rank when setting up a game, so in a 100 point per stone system, a 1390 and a 1310 should still play without handicap to keep the system, and their own rating, meaningful?


Soulpanda: Ok, Soulpanda's long explaination. A Rank is awarded by a professional Go Organizations (Japanese Go Association, Korean Baduck Association, ect.)for players who have proven to be of a certain strength. For example, a person becomes a 1-dan in their respective country and receives a Rank of 1-dan, then plays in the various rating tournaments (again, for Pros) and shows that he is stronger than 1-dan in strength. He is then awarded the Rank of 2-dan and now has Two pieces of paper that he can frame. If he never played another game in his life he would still be a 2-dan in Rank. Though, if he tried to play again after 50 years he would propably have his head handed to him on a plater. As far as I know, Rankings are given from 1-dan to 9-dan only, though your professional organization may vary.

A Rating is a mean average of a player's strength. Each win and lose counts in determining how far up or _DOWN_ you can go. This is used by "Non-Professional" organizations (such as the AGA, IGS, KGS, maybe the BGS (sorry Britain if I'm wrong)) mainly as a way of keeping up with the strengths of a large group of amateur players. Ratings can start as weak as 35-kyu and can go from 6 to 9-dan. If obtaining a Rank is like climbing a terranced mountain, where its grueling to get to the next level but once your there you don't have to worry about falling off, then getting a Rating is bare-handing (climbing with no safety equipment) a shear cliff face and having to deal with sliding down and clawing your way up.

Soulpanda: Soulpanda's short explanation. Simply, Pros have Ranks and amateurs don't. That's it. Pro=Rank, Amateur=Rating. End of story.


Rating / Discussion last edited by tapir on May 27, 2013 - 14:27
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
RecentChanges
StartingPoints
About
RandomPage
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Goproblems.com
Login / Prefs
Tools
Sensei's Library