Life In 19x19
http://prod.lifein19x19.com/

Fast != slow
http://prod.lifein19x19.com/viewtopic.php?f=10&t=11467		 Page 1 of 2
Author:  John Fairbairn [ Fri Feb 13, 2015 11:17 am ]
Post subject:  Fast != slow
Some of us here have voiced concern over the increasing number of fast games in modern tournament go, a trend that has been paralleled by an apparent decline in Japanese go.

Whichever type of go your prefer, it seems you can't shrug it off as making no real difference.

Chess has faced the same pressures. They have responded by defining ranking lists for three types of chess: Classical, Rapid and Blitz. These are based on time allowances.

If you look at the ranking in Classical chess, which appears to have the highest status among both players and chess fans, and compare the ranking in Blitz, there are some startling differences.

The table below shows the players in the Classical top ten and their Classical/Blitz rankings with the Elo points difference between them.

    Carlsen.... 1/1 (86)
    Grischuk... 2/3 (42)
    Caruana.... 3/69 (130)
    Giri....... 4/13 (27)
    Topalov.... 5/>100 (? but big)
    Nakamura... 6/2 (91)
    So......... 7/34 (62)
    Kramnik.... 8/16 (20)
    Anand...... 9/14 (15)
    Vachier-L. 10/5 (51)


To add some context, a 100 points difference between opponents equates to the higher rated player having a 64% chance of winning. That seems to equate to the old go handicap of B-B-W, so 100 points elo = 1 dan at pro level. (200 points difference makes the winning probability 76%.)

On that basis, Caruana (who is third in all-time elo ratings list) shows much more than one dan difference, and Topalov (a recent world champion) must be at least two dans different. There are other big dan-level differences in the top ten, but of course all of this is magnified if you scan below the top ten.

The list also shows that while some are significantly better at classical chess, for others the forte is in blitz, so it is not the case that everyone's play deteriorates at the same rate when playing blitz.

My first impression is that this puts a big question mark over the elo ratings in go, and that they may need to be recast in the same way as in chess. I'm inclined to wonder, too, whether this lends support to the notion that older and/or Japanese players who are obliged to play too many fast games are misrepresented by the current ratings. I know we have some good statisticians in this forum (esp. Dr Straw), so I look forward to their views.
Author:  EdLee [ Fri Feb 13, 2015 11:40 am ]
Post subject: 
Quote:
...must be at least two [stones] different.

My first impression is that this puts a big question mark over the elo ratings in go, and that they may need to be recast in the same way as in chess.
Yes.

One data point: I drop anywhere from 2 to 4 stones in faster (than my normal) time settings.
Author:  Bill Spight [ Fri Feb 13, 2015 11:56 am ]
Post subject:  Re: Fast != slow
By comparison with go, contract bridge has almost a blitz pace. The normal speed of tournament bridge is like playing a game of go in 30 minutes. Still, there are fast, intuitive, bridge players and slow, deep thinking, bridge players. One interesting thing is that the fast players seem to maintain their strength well into old age. A prime example is Oswald Jacoby, who won his last US National Championship at age 80. B. Jay Becker won his last US National Championship at age 76.

Right now we are seeing oldsters losing to youngsters in (relatively) fast games. But 20 or 30 years from now, who knows? We may see our current young champions still in the running. :)
Author:  gowan [ Fri Feb 13, 2015 12:27 pm ]
Post subject:  Re: Fast != slow
I think it was Ishida Yoshio (of the joseki dictionary fame and honorary Honinbo) who, at one point in middle age, opined that he was not competitive in slow title go but he could hold his own in fast TV tournament games.
Author:  tiger314 [ Fri Feb 13, 2015 12:48 pm ]
Post subject:  Re: Fast != slow
The European ratings deal with this by requiring sufficient thinking time for a tournament to be included with full coefficient of 1 (75 minutes sudden death or 60 minutes plus 15 minutes for 45/60 moves in Japanese/Canadian byoyomi). There is also a minimal requirement (30 minutes or 25 minutes plus 5 minutes for 45/60 moves in Japanese/Canadian byoyomi) for a tournament to be included at all, and then, it is only included with a coefficient of 0.5 (or 0.75 for slightly slower games).
Since I have not heard of a single large blitz tournament organised in Europe (I don't know about the situation in the US and elsewhere), I think that a separate rating list for fast games would only have small local tournaments included. This would create enormous inconsistencies (a town/area could have everybody like 5 stones stronger when compared with equally rated players elsewhere). So the current European system, where blitz and fast games are never rated and other games count into one value, seems good to me. (Not commenting on situation in East Asia, Go has a totally different position there.)
Author:  jeromie [ Fri Feb 13, 2015 12:51 pm ]
Post subject:  Re: Fast != slow
I think it's reasonable to believe that time limits affect a player's strength, and that the effect is not constant from one player to the next. I think the biggest challenge in quantifying this difference in international go is the paucity of data points on the longer end of the spectrum. Are there any tournaments outside of Japan that have time limits comparable to the big Japanese title matches?

We often get to see how top Japanese players perform in international matches with shorter time limits; it would certainly be interesting to observe the reverse. I suspect there would be a few Chinese and Korean players who would do very well under the longer time setting, though perhaps not the same set that dominate shorter tournaments.

John, I am curious: where would you draw the lines between classical, rapid, and blitz play in professional go?
Author:  palapiku [ Fri Feb 13, 2015 1:05 pm ]
Post subject:  Re: Fast != slow
It is pretty common for people to create separate KGS accounts for slow and blitz games because the ranks can be quite different. The chess world definitely has the right idea here.
Author:  Boidhre [ Fri Feb 13, 2015 1:29 pm ]
Post subject:  Re: Fast != slow
It's not strange to see strong amateur chess players playing and beating Master and IM level players in online chess when the time limit is absurdly short like in bullet chess where these amateurs comment that this person would crush them in a slow game. When each side has 1 minute and no overtime it does change the playing field somewhat.
Author:  Bantari [ Fri Feb 13, 2015 2:25 pm ]
Post subject:  Re:
EdLee wrote:
Quote:
...must be at least two [stones] different.

My first impression is that this puts a big question mark over the elo ratings in go, and that they may need to be recast in the same way as in chess.
Yes.

One data point: I drop anywhere from 2 to 4 stones in faster (than my normal) time settings.


Out of curiosity, how do you determine that?

Here is what I struggle with, and have so each time I hear such statement.
When you play faster, or slower, you have a few metrics I can think of:
(1) general win/loss record/ratio
(2) win/loss record/ratio against known players (the ones you also play at other time settings)
(3) the quality of your moves and the magnitude of your mistakes

To me, only case (3) would warrant any quantitative statements like that.
But this would take a lot of serious analysis of both your games and the games of your opponents, at all kinds of time settings, and still - probably be impossible (or at least - very hard) to actually qualify to any degree. And even then, it would probably change between levels.

In cases (1) and (2) one cannot really say, since everybody is playing on different settings, so the whole field changes and the frame of reference is completely different. You can never tell if it is they who play better/worse, or you, or its somewhere in the middle. Or they play better while you play worse, or the other way around, or whatever. It might even change for different phases of the game - you might play fast fuseki better than him, but he plays fast yose better thank you, while in slow games it is the other way around. What does it say about your respective levels?

Confusing...
Author:  EdLee [ Fri Feb 13, 2015 2:30 pm ]
Post subject: 
Bantari wrote:
how do you determine that?

(3) the quality of your moves and the magnitude of your mistakes
Hi Bantari, mostly it's my feeling. And yes, (3) is a big factor.

But mostly, I am just slow, in reading. So it's very obvious, to me. :)
Bantari wrote:
probably be impossible (or at least - very hard) to actually qualify to any degree.
On the contrary, it's very easy (for me).

It's like this:
monkey memory test.

The difference is like that -- for me, blitz is like the time given to the monkey,
whereas my normal time is like 5 minutes displaying the numbers -- there's absolutely no comparison.

BTW, is anyone here on the forum able to perform as well as the monkey ?
Author:  RobertJasiek [ Fri Feb 13, 2015 10:33 pm ]
Post subject:  Re: Fast != slow
OC, playing strengths or ratings greatly depend on time settings and lots of other factors. I have said so for decades and suggested multi-dimensional values.
Author:  Javaness2 [ Sat Feb 14, 2015 12:15 am ]
Post subject:  Re: Fast != slow
It is strange that no Go server has ever supported blitz ratings, but there you go, they never have. There doesn't seem to be much demand for them to appear. I rather doubt that any amateur organisation has the capacity to construct a second rating list.
Author:  RBerenguel [ Sat Feb 14, 2015 1:22 am ]
Post subject:  Re: Fast != slow
John Fairbairn wrote:
Some of us here have voiced concern over the increasing number of fast games in modern tournament go, a trend that has been paralleled by an apparent decline in Japanese go.

Whichever type of go your prefer, it seems you can't shrug it off as making no real difference.

Chess has faced the same pressures. They have responded by defining ranking lists for three types of chess: Classical, Rapid and Blitz. These are based on time allowances.

If you look at the ranking in Classical chess, which appears to have the highest status among both players and chess fans, and compare the ranking in Blitz, there are some startling differences.

The table below shows the players in the Classical top ten and their Classical/Blitz rankings with the Elo points difference between them.

    Carlsen.... 1/1 (86)
    Grischuk... 2/3 (42)
    Caruana.... 3/69 (130)
    Giri....... 4/13 (27)
    Topalov.... 5/>100 (? but big)
    Nakamura... 6/2 (91)
    So......... 7/34 (62)
    Kramnik.... 8/16 (20)
    Anand...... 9/14 (15)
    Vachier-L. 10/5 (51)


To add some context, a 100 points difference between opponents equates to the higher rated player having a 64% chance of winning. That seems to equate to the old go handicap of B-B-W, so 100 points elo = 1 dan at pro level. (200 points difference makes the winning probability 76%.)

On that basis, Caruana (who is third in all-time elo ratings list) shows much more than one dan difference, and Topalov (a recent world champion) must be at least two dans different. There are other big dan-level differences in the top ten, but of course all of this is magnified if you scan below the top ten.

The list also shows that while some are significantly better at classical chess, for others the forte is in blitz, so it is not the case that everyone's play deteriorates at the same rate when playing blitz.

My first impression is that this puts a big question mark over the elo ratings in go, and that they may need to be recast in the same way as in chess. I'm inclined to wonder, too, whether this lends support to the notion that older and/or Japanese players who are obliged to play too many fast games are misrepresented by the current ratings. I know we have some good statisticians in this forum (esp. Dr Straw), so I look forward to their views.


John, it is worth noting several factors in Caruana. First, I'm pretty sure he does not join as many blitz tournaments as the others (from my following of chess events, I rarely see Caruana & blitz in the same way I see Nakamura, Giri or Carlsen.) This also seems to hint that either he doesn't care or he doesn't like blitz. The former seems weird, since he's a very focused player and I'd guess he would like to improve this area. So, probably, he just doesn't enjoy blitz and just endures it when needed.
Author:  RBerenguel [ Sat Feb 14, 2015 1:23 am ]
Post subject:  Re: Fast != slow
Javaness2 wrote:
It is strange that no Go server has ever supported blitz ratings, but there you go, they never have. There doesn't seem to be much demand for them to appear. I rather doubt that any amateur organisation has the capacity to construct a second rating list.


Doesn't OGS now do it? I can't check quickly since I'm on iPad and OGS "hover to find out what this is" doesn't work that well here.
Author:  Tryss [ Sat Feb 14, 2015 10:51 am ]
Post subject:  Re: Fast != slow
As a sidenote, I'm glad to see Vachier doing so well. He was in the same class as me at university and was a very nice guy.
Author:  uPWarrior [ Sat Feb 14, 2015 11:44 am ]
Post subject:  Re: Fast != slow
RBerenguel wrote:
Javaness2 wrote:
It is strange that no Go server has ever supported blitz ratings, but there you go, they never have. There doesn't seem to be much demand for them to appear. I rather doubt that any amateur organisation has the capacity to construct a second rating list.


Doesn't OGS now do it? I can't check quickly since I'm on iPad and OGS "hover to find out what this is" doesn't work that well here.


They do, there are blitz, live and correspondence ratings now (there is also an "overall" rating, but I don't know how it's calculated. Maybe ignoring the timesettings?)
Author:  snorri [ Sat Feb 14, 2015 2:48 pm ]
Post subject:  Re: Fast != slow
An amateur organization could decide to implement blitz ratings, but the discussion over where to draw the line for how fast blitz is would likely end in fisticuffs.
Author:  Uberdude [ Sat Feb 14, 2015 9:49 pm ]
Post subject:  Re: Fast != slow
EGF ratings already have a tournament class weighting factor so that faster games change your rating less.
Author:  palapiku [ Sun Feb 15, 2015 2:31 am ]
Post subject:  Re: Fast != slow
I believe the reason it's common to have a single rank, instead of different ranks for different classes, lies in the origin of the dan/kyu system. Your rank is not just a number, it's almost like a title. Nowadays it is usually calculated in the same way as elo-style ratings, but it can also be given after an examination, or simply announced by your teacher, or given as an honor. Sometimes for life. It's not really meant to be an exact estimate of your current ability the way chess ratings are. This is still generally the case with pro dan ranks.
Author:  EdLee [ Sun Feb 15, 2015 2:49 am ]
Post subject: 
Uberdude wrote:
EGF ratings already have a tournament class weighting factor so that faster games change your rating less.
Nice.
Page 1 of 2 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/