It is currently Thu Mar 28, 2024 2:09 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 4 posts ] 
Author Message
Offline
 Post subject: Forward an interesting article
Post #1 Posted: Sun Oct 10, 2021 5:24 am 
Beginner

Posts: 2
Liked others: 0
Was liked: 1
Rank: kgs 6
GD Posts: 9
KGS: 6 dan
Tygem: 20211010
IGS: b
Wbaduk: c
DGS: s
OGS: s
Kaya handle: h
Universal go server handle: go999999
Ranking of Japan and China replacing modern chess power。
http://home.yikeweiqi.com/#/gonews/detail/49602?type=1


Attachments:
73f36f820d701659fc1ac322b7502293.png
73f36f820d701659fc1ac322b7502293.png [ 553.82 KiB | Viewed 2297 times ]

This post by go999999 was liked by: dhu163
Top
 Profile  
 
Offline
 Post subject: Re: Forward an interesting article
Post #2 Posted: Sat Oct 16, 2021 2:41 am 
Lives in sente

Posts: 905
Liked others: 22
Was liked: 168
Rank: panda 5 dan
IGS: kvasir
It would be nice when linking articles to provide a summary. Especially when the article is not in English (because most readers here won't be able to read it) but even then it is still useful because many people may not actually read linked articles.

Top
 Profile  
 
Offline
 Post subject: Re: Forward an interesting article
Post #3 Posted: Sat Oct 16, 2021 3:40 pm 
Oza

Posts: 3647
Liked others: 20
Was liked: 4626
Quote:
t would be nice when linking articles to provide a summary.


I haven't looked at the source yet, but the histogram shows a comparison, based on katago set at the appropriate komi for each period, for lots of Chinese ancient and modern masters, in chronological order starting with Guo Bailing and ending with Ke Jie. (In one case the entry is for the famous "18 Qing masters" collectively. The profile for each player is based on 10 to 15 games at his peak period (which is presumably a matter of guesswork for the old players, as their games are undated). Without seeing the original paper, I can't really say what the histos series really represent. The first two bars are minimum and maximum readings of something. The third bar (the maroon bar) represents an average of some sort, and I infer it's the one to take most notice. The last bar is some measure of complexity.

There have been several similar exercises in Japan and China, most not scientifically rigorous, but they all seem to show a similar pattern of the old masters faring very well in comparison with modern masters. I have mot seen any deep discussion of this phenomenon, though I imagine the longer time limits characteristic of older play might be factor. But in general I don't feel too surprised, despite the fact that many modern people automatically assume modern players must be stronger. Why? Do we really regard Bernstein as better than Mozart and West Side Story as musically outshining Cosi Fan Tutte? Or Gershwin better than Beethoven?

There seems to be a better argument to be made for the skill of players or musicians to reflect the times they live in. Times of great patronage or popularity produce greater geniuses. The graph above is a good example. After Fan Xiping and Shi Xiangxia, there is a sustained dip that has long been noticed, well before AI (in fact the whole graph seems to show that AI opinion matches the pre-AI views of professional commentators). This dip, when patronage and book publication dropped noticeably, was also the period of the 18 masters (guoshous to a man but none considered truly outstanding).

On the basis of some other recent Chinese research it looks like there may be a spike on the way. In China and Korea young players have hitherto made more use of Leela than Japanese young players and have shown an increase in rating as measured by matching AI plays (though that seems to have changed since in Japan, and pr4esumably katago has displaced Leela). In other words, modern players now have more resources, as well as more tournaments, and so will eventually show up as more skilful for those reasons and not because they have bigger brain pans (or, just as plausibly, will simply peak younger?)

For the curious, Huang Longshi is second from the left. The peak player near the middle is Zhou Xiaosong (the Chinese Shuwa), who, as it happens, is the one of the main players in the next Museum of Go Theory sitting on my desk waiting to be proof-read. The other main player is Chen Zixian, and there has been a long-running argument about who was stronger, Zhou or Chen. The graph appears to show it was Zhou, but the reality was that Chen died young, and Zhou published his own selection of his own games. Just as average willy size has (if you believe Youtube) been found by recent rigorous medical measurements to be about 2 inches less than when sizes are self reported, we may have to regard Zhou as a self-reporter.


This post by John Fairbairn was liked by: dhu163
Top
 Profile  
 
Offline
 Post subject: Re: Forward an interesting article
Post #4 Posted: Sun Oct 17, 2021 1:29 am 
Lives in sente

Posts: 905
Liked others: 22
Was liked: 168
Rank: panda 5 dan
IGS: kvasir
Thank you for the summary of the graph.

I only got as far as reading off some of the names in the graph (from left to right) Ke Jie, Gu Li, Chang Hao, Ma Xiaochuan and Nie Weiping. I was confused that I didn't recognize any Japanese names because something was said about comparing Chinese and Japanese players. I didn't find Go Seigen in the graph.

I feel all the exercises in using computers to evaluate historic Go players, that I have seen, fail because the metrics used don't really mean much and are not directly comparable. The analysis is based on too few games and the games have very different character.

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 4 posts ] 

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users 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