It is currently Thu Mar 28, 2024 4:05 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 38 posts ]  Go to page Previous  1, 2
Author Message
Offline
 Post subject: Re: Statistically derived komidashi?
Post #21 Posted: Wed Feb 26, 2014 2:53 pm 
Lives in sente
User avatar

Posts: 1103
Location: Netherlands
Liked others: 408
Was liked: 422
Rank: EGF 4d
GD Posts: 952
hyperpape wrote:
A good question, and I think you can make an argument that draws should be possible in some ideal sense. But if komi suffices to reduce the advantage to less than a 52% win percent (for either White or Black), that's excellent balance. I don't find it worth tolerating a 5% chance of draws in competition to remove that advantage. Perhaps if it were the old days in which a series could continue as long as it needed, that would be ok, but we already have scheduling conflicts.

Chess gets by with far more draws than I think we'd ever have, I feel like competitions could probably handle it.

Quote:
(Btw: even allowing draws would not necessarily make the result 50-50. We could be talking about moving the needle by less than 2% advantage to a point that is still slightly unbalanced).


Certainly, If the exact 7 pt komi doesn't make the results closer to 50-50, then there's no point. It's just interesting that we have limited our calibration to only adjusting komi by a full point, rather than a half point.

_________________
Tactics yes, Tact no...

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #22 Posted: Wed Feb 26, 2014 4:01 pm 
Oza
User avatar

Posts: 2777
Location: Seattle, WA
Liked others: 251
Was liked: 549
KGS: oren
Tygem: oren740, orenl
IGS: oren
Wbaduk: oren
shapenaji wrote:
Certainly, If the exact 7 pt komi doesn't make the results closer to 50-50, then there's no point. It's just interesting that we have limited our calibration to only adjusting komi by a full point, rather than a half point.


I don't think we'll see that change in professional play. Sponsors and tournament directors want a winner and loser. Having to do repeat games can be a bit annoying.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #23 Posted: Wed Feb 26, 2014 4:09 pm 
Oza
User avatar

Posts: 2401
Location: Tokyo, Japan
Liked others: 2338
Was liked: 1332
Rank: Jp 6 dan
KGS: ez4u
HermanHiddema wrote:
ez4u wrote:
I am not sure why you think that 8p both sides is particularly relevant in the present pro environment (consider who gets excluded from the results). Anyway my copy of GoGoD (summer 2013) has 9,340 games with komi of 3.75 (i.e. Chinese rules with 7.5). White wins 52.3%. The not quite 19,000 games with komi of 6.5 have Black winning 50.2%


Is the weaker player (lower pro rank) more likely to get black in that sample? I wanted to avoid that option.

There are no events like that AFAIK. There are a few special events like the new-shodan games each year in Japan, but they do not use the standard komi.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #24 Posted: Wed Feb 26, 2014 7:08 pm 
Tengen

Posts: 4380
Location: North Carolina
Liked others: 499
Was liked: 733
Rank: AGA 3k
GD Posts: 65
OGS: Hyperpape 4k
shapenaji wrote:
Chess gets by with far more draws than I think we'd ever have, I feel like competitions could probably handle it.
Perhaps you just mean to say that the low proportion of draws in Go would be ok, but comparisons to Chess don't make the point. There is a large and vocal contingent of people complaining about the frequency of draws in high level play, and there have been proposals to change the rules to reduce the number of draws.

_________________
Occupy Babel!

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #25 Posted: Wed Feb 26, 2014 11:09 pm 
Lives in gote

Posts: 370
Liked others: 91
Was liked: 254
Rank: Weak
Assume that
  • the players do not resign but always go to counting
  • non-counting draws (due to triple-ko, etc.) do not occur
  • black wants to maximizes black territory minus white territory
  • white wants to maximizes white territory minus black territory
Zermelo/Von Neumann/Morgenstern wrote:
Under perfect play, some player (presumably black) always wins by K points with no komi.
Suppose that, as the skill of the players approaches perfection, the probability of black winning by K points approaches 1. Then
  • Under a komi of K+0.5, the probability of white winning by 0.5 approaches 1 as the players approach perfection.
  • Under a komi of K-0.5, the probability of black winning by 0.5 approaches 1 as the players approach perfection.

This seems like an irrelevant mathematical oddity because perfect play is so far off. However, suppose we assume symmetric probabilities (without komi) for Black winning by (K+n) and Black winning by K-n.

Then all we need for White to win 50+x% of the time (with komi=K+0.5) is for the probability of Black winning by K (without komi) to approach 2x%, which is a far cry from 100%. Pros can easily keep any game from getting beyond a 10 point difference with the current komi if they don't care about winning. That actually seems like a conservative estimate if we believe that any top 20 pro could win 90%+ of the time if he was taking black with no komi.

What is the conclusion here? A stable non-integer komi that produces 50-50 outcomes is not a realistic long-term goal. It may be achievable in the short term though.

I could see komi cycling between two values as players research black's fuseki when komi=K+0.5 and research white's fuseki when komi=K-0.5. In the long-run we will either have to keep changing komi every once in a while or switch to a system that allows ties.


This post by lemmata was liked by: shapenaji
Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #26 Posted: Wed Feb 26, 2014 11:32 pm 
Lives in gote

Posts: 370
Liked others: 91
Was liked: 254
Rank: Weak
Addendum: The SG Pair Baduk tournament uses a system called "dum betting" (komi betting). Each team secretly writes down a komi value on a piece of paper. The team that wrote down the higher komi value takes black and offers that komi. In the case of a tie, they use nigiri to see who goes first. Such a system, if adopted for one-on-one pro games, may obviate the need to adjust komi values via arduous bureaucratic processes long after the winning percentages get out of hand.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #27 Posted: Thu Feb 27, 2014 3:09 am 
Lives in sente
User avatar

Posts: 1103
Location: Netherlands
Liked others: 408
Was liked: 422
Rank: EGF 4d
GD Posts: 952
hyperpape wrote:
shapenaji wrote:
Chess gets by with far more draws than I think we'd ever have, I feel like competitions could probably handle it.
Perhaps you just mean to say that the low proportion of draws in Go would be ok, but comparisons to Chess don't make the point. There is a large and vocal contingent of people complaining about the frequency of draws in high level play, and there have been proposals to change the rules to reduce the number of draws.


Well, they certainly argue about the frequency of draws, but not their existence. I think most chess players are fine with the possibility of draws, they just don't like that top players have methods at their disposal to ensure draw-like games. I don't think go lends itself to a drawing strategy in the same way.

_________________
Tactics yes, Tact no...

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #28 Posted: Thu Feb 27, 2014 8:25 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
shapenaji wrote:
hyperpape wrote:
shapenaji wrote:
Chess gets by with far more draws than I think we'd ever have, I feel like competitions could probably handle it.
Perhaps you just mean to say that the low proportion of draws in Go would be ok, but comparisons to Chess don't make the point. There is a large and vocal contingent of people complaining about the frequency of draws in high level play, and there have been proposals to change the rules to reduce the number of draws.


Well, they certainly argue about the frequency of draws, but not their existence. I think most chess players are fine with the possibility of draws, they just don't like that top players have methods at their disposal to ensure draw-like games. I don't think go lends itself to a drawing strategy in the same way.


Wasn't there a flap several years ago about an arranged draw (or no result) in a go tournament?

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #29 Posted: Thu Feb 27, 2014 8:43 am 
Lives in sente
User avatar

Posts: 1103
Location: Netherlands
Liked others: 408
Was liked: 422
Rank: EGF 4d
GD Posts: 952
Bill Spight wrote:

Wasn't there a flap several years ago about an arranged draw (or no result) in a go tournament?


Have a handy link?

I think that's the big difference though, in chess it doesn't take collusion to force a draw with high probability (as white).

_________________
Tactics yes, Tact no...

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #30 Posted: Thu Feb 27, 2014 9:20 am 
Lives in sente

Posts: 946
Liked others: 1
Was liked: 41
Rank: IGS 5kyu
KGS: KoDream
IGS: SmoothOper
shapenaji wrote:
hyperpape wrote:
shapenaji wrote:
Chess gets by with far more draws than I think we'd ever have, I feel like competitions could probably handle it.
Perhaps you just mean to say that the low proportion of draws in Go would be ok, but comparisons to Chess don't make the point. There is a large and vocal contingent of people complaining about the frequency of draws in high level play, and there have been proposals to change the rules to reduce the number of draws.


Well, they certainly argue about the frequency of draws, but not their existence. I think most chess players are fine with the possibility of draws, they just don't like that top players have methods at their disposal to ensure draw-like games. I don't think go lends itself to a drawing strategy in the same way.


I think the rules in chess encourage draws, by allowing repeated states, though it is difficult to keep track of the repeated states in casual settings since chess has so many configurations with the same set of pieces, however if they had a super ko type rule, they wouldn't have this problem.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #31 Posted: Thu Feb 27, 2014 9:53 am 
Tengen

Posts: 4380
Location: North Carolina
Liked others: 499
Was liked: 733
Rank: AGA 3k
GD Posts: 65
OGS: Hyperpape 4k
In theory, superior would avert draws (though perhaps only if you removed the stalemate rule), but in practice it might lead to unbearably long almost-cycles that would be beyond effective grasp by players. You'd have to analyze it to see if it would work properly.

_________________
Occupy Babel!

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #32 Posted: Thu Feb 27, 2014 9:59 am 
Tengen

Posts: 4380
Location: North Carolina
Liked others: 499
Was liked: 733
Rank: AGA 3k
GD Posts: 65
OGS: Hyperpape 4k
Lemmata, why does your claim not work for the short term?

_________________
Occupy Babel!

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #33 Posted: Thu Feb 27, 2014 10:02 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
shapenaji wrote:
Bill Spight wrote:

Wasn't there a flap several years ago about an arranged draw (or no result) in a go tournament?


Have a handy link?

I think that's the big difference though, in chess it doesn't take collusion to force a draw with high probability (as white).


IIRC, it was a European tournament and the two players in question were Russian.

As for chess, way back when, Bobby Fischer accused Russian grandmasters of arranging draws in order to save their energies to play against him.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #34 Posted: Thu Feb 27, 2014 10:45 am 
Lives in sente

Posts: 946
Liked others: 1
Was liked: 41
Rank: IGS 5kyu
KGS: KoDream
IGS: SmoothOper
hyperpape wrote:
In theory, superior would avert draws (though perhaps only if you removed the stalemate rule), but in practice it might lead to unbearably long almost-cycles that would be beyond effective grasp by players. You'd have to analyze it to see if it would work properly.


Wouldn't it work just like a ko threat? I think the only time you would have issues would be if you had multiple semi-independent almost-cycles(almost-cycles within almost-cycles but not quite cycles), just like jigo. I think the reason it would work, is that they have to keep track of the cycles anyway to prove that it is a draw.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #35 Posted: Thu Feb 27, 2014 3:29 pm 
Lives in gote

Posts: 370
Liked others: 91
Was liked: 254
Rank: Weak
hyperpape wrote:
Lemmata, why does your claim not work for the short term?
Because I am lazy and assert the conclusions I want. :lol: I was thinking that there might be asymmetric short-term jumps in knowledge about how to play as one color. It's certainly not a watertight argument though.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #36 Posted: Fri Feb 28, 2014 12:26 am 
Lives in gote

Posts: 309
Liked others: 3
Was liked: 41
Rank: 5 dan
ez4u wrote:
HermanHiddema wrote:
Matti wrote:
I think we also have enough data for 7½ komi. Does some one have them at hand and be able to tell whether black still has an advantage? If yes, then we would need, say 1000 games with 8½ or preferably 9½ to know better.


At http://senseis.xmp.net/?Komi%2FStatistics there is a section I posted about 7.5 or 8.0 (Ing) komi, which has white winning 50.5% of games in a sample of 410 games from gobase.org where both players are at least 8d pro.

I am not sure why you think that 8p both sides is particularly relevant in the present pro environment (consider who gets excluded from the results). Anyway my copy of GoGoD (summer 2013) has 9,340 games with komi of 3.75 (i.e. Chinese rules with 7.5). White wins 52.3%. The not quite 19,000 games with komi of 6.5 have Black winning 50.2%

Judging from the numers only, it seems that with 7.5 komi white has a slight advantage. However, I would like to eliminate the possibility that white's winning percentage has disappeared since the 7.5 komi was introduced.

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #37 Posted: Fri Feb 28, 2014 2:41 am 
Oza
User avatar

Posts: 2401
Location: Tokyo, Japan
Liked others: 2338
Was liked: 1332
Rank: Jp 6 dan
KGS: ez4u
Through 2007 we have 3406 games with White winning 52.8%
From 2008 through 2010 we have 3129 games with White winning 52.1%
From 2009 through the first half of 2013 we have 2805 games with White winning 52.1%

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

Top
 Profile  
 
Offline
 Post subject: Re: Statistically derived komidashi?
Post #38 Posted: Fri Feb 28, 2014 3:24 am 
Lives in gote

Posts: 309
Liked others: 3
Was liked: 41
Rank: 5 dan
ez4u wrote:
Through 2007 we have 3406 games with White winning 52.8%
From 2008 through 2010 we have 3129 games with White winning 52.1%
From 2009 through the first half of 2013 we have 2805 games with White winning 52.1%

Ok. The winning percentage seems quite stable.

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 38 posts ]  Go to page Previous  1, 2

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: Bing [Bot] and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group