Where Is Go Going

NOTE: almost all of this page is from 2002-2003, except for the very end.

AvatarDJFlux

Dear all, I'd just like to share some thoughts and feelings that I've been munching in my head lately. I'd be very glad to hear your opinions and criticisms, of course.

I started to play go in the early eighties, giving up chess in the meanwhile: it looked to me as if the latter was becoming too cramped, almost claustrophobic.

Two aspects of go I loved most (and still do): on the one hand, its subtlety, deepness and stress on flexibility and on keeping open as much and as long as possible the largest number of possibilities.
It really fascinates me how professionals use probes in order to force your opponent to choose a strategy, making him consequently lose some degrees of freedom, so to speak.

In the eighties, if I remember correctly, very few professionals played long, complicated and drawn out joseki: Cho Chikun once said that he didn't like them because they simplified the game too much, taking away subtlety. Kobayashi was criticised very often by his fellow professionals (especially Takemiya...) for settling the shape as much as he could.

On the other hand, I adore the great role played by intuition and "feeling" (something that I couldn't find in chess). I'm not saying that brute force reading and calculation are not important, especially so when you go through chuban to yose, but in fuseki and early chuban intuition has wide scope for action: no fuseki is like another one (not so in chess...). Not for nothing are my heroes Shuko, Takemiya and Otake!

It is true that intuition and feeling can be seen as skills that one acquires with the experience, playing and studying literally thousands of games.

(Mind you, I am a post-hippie, my generation is the one that looked toward India and the "Orient" at large to find something that we couldn't find in our western, rational mind. So it is quite obvious that I would cherish these aspects of the game of go, maybe missing or misunderstanding some others.)

Then I stopped playing for ten years, and when I resumed a couple of years ago, what kind of situation did I find (and please correct me if I'm wrong)?

In the East, the arrival of strong, very strong Korean and Chinese players: this is really great, as otherwise the game could have ended up as something for a few pros and for a bunch of aged people in Japan... But this meant also, as I perceive it, the affirmation of a very aggressive style right from the beginning, and the appearance of almost established fuseki sequences where long, complicated, drawn-out joseki are instrumentally played in order to settle the shape quickly and go directly to a sort of oyose stage, where it's easier to calculate the value of moves.
Is go evolving towards chess??!? ;-) What happened to subtlety and flexibility?

In the West, there are a number of attempts to formalise the game using game theory and other mathematical and logic tools. One of the reasons for this is of course the possibility to arrive at writing a software that would be able to play at a reasonable level, but we have seen few results so far. Luckily, I say.
I think the real reason is the very western passion for rational and formal thinking, for putting everything in the right box identified with its proper label.
It almost looks as if people are trying to find the perfect move in the fuseki by applying some suitable theorem instead of relying on experience and "feeling", as described above.
Is the western, rational mind taking over??!? ;-) What happened to intuition?
Are there any pros who use such approach??!?
Another aspect that seems weird to me is the hugely large discussion going on on how to write a perfectly coherent and logic set of rules and definitions, able to take into account even the remotest and weirdest of the possibilities. (Ah, Gödel, where art thou... ;-)

Couldn't all the energies spent in these intellectual quests be spent instead on playing and spreading the game? It seems to me that all that mathematical, logical and theoretical arguing would scare any beginner (not to mention myself), making him/her believe that without a PhD in mathematics you cannot play go: not a good advertisment indeed!

I'm not saying that those approaches to the game have no validity (and after all I'm still 3k and shouldn't talk much...).
Koreans and Chinese players seem to win more international tournaments than the Japanese do.
Intellectual formalisation is surely fun for those who practise it.
But I have the sensation that such approaches somehow diminish and simplify the game, an attitude that is only natural when one is afraid of empty spaces (agoraphobia) and wants to grasp a sure way around...
I, of course, have no answers, and perhaps I just like to grope in wide darkness...

Now I await your criticisms and opinions.
But I promise: while still liking better to develop the left side of my brain, I will practise reading and counting as well, otherwise I'm afraid I'll never reach shodan...

HolIgor: I am not strong enough to judge the trends in modern professional go, but I know that there is one big difference between go and chess. Chess suffocates in draws and go does not. I remember that when I read the rules for the first time I was amazed by the absence of draws. It seemed to me a very unjust thing. Imagine, you work hard, you play well, not worse than your opponent and in the end everything is decided by the last ko.

Today I lost a game by a half point again. I won a lot of games by a half point. It is a lottery.

I imagine that playing against Yi Ch'ang-ho many players would agree to a lottery. Yi Chang-ho would not, of course. This is a matter of strength and weakness.

Freedom is a strange thing. According to Marx it is recognized necessity. At some point I understood that I can't win making any moves. I have to make correct moves. This point in the games moves closer and closer to the opening, to fuseki. But it is still very far. At move 50 or so. Recently I discovered importance of turning first. Playing with it now. I admire the way dan players discuss the moves in the fuseki. I have no clue. I call it divination. How much will the opponent be able to get by attacking my weak group? Recently I discoved that a line of ikken-tobi has a good fighting and eye forming potential. But to what extent?

I've heard that pros study fuseki to yose now. But at the same time somebody in the rgg tried to find in a database the longest fuseki sequence common for more than 3 games. I don't remember the size of the database, but it was not small. The longest line was 6 moves. It is really unbelievable.

Arno: 6 moves? Maybe they were not counting for "normalizing" the board, i.e. the first black move on 4-4 would be a different game from the first black move on 16-16. If you normalize the board (mirror and/or turn around) you get something like my fuseki_db at http://xmp.net/arno/fuseki.html -- a quick look showed a sequence up to move 20 with three games being the same (Black's low chinese fuseki against White's ni-ren-sei). Base: roughly 2000 games.

DaveSigaty: Alternatively from the Go Games on Disk CD (January 2002 update) the following is the widest path (following the most frequent first move and the most frequent reply to that move...). The population analyzed for the game counts shown consists of the 1470 games from 2000 and 2001 out of which White won 667, Black won 800, and 3 were drawn or void. The analysis ignores mirrored moves since my machine is too slow on the searches :-)

"Widest Path" Fuseki 1-10  

B 1 - 1056 games: W won 470 and B won 586

W 2 - 564 games: W won 249 and B won 315

B 3 - 257 games: W won 106 and B won 151

W 4 - 124 games: W won 51 and B won 73

B 5 - 98 games: W won 40 and B won 58

W 6 - 53 games: W won 25 and B won 28 (notice how White suddenly narrows the gap in results!)

B 7 - 46 games: W won 21 and B won 25

W 8 - 46 games: W won 21 and B won 25

B 9 - 46 games: W won 21 and B won 25

W 10 - 26 games: W won 14 and B won 12

"Widest Path" Fuseki 11-15  

B 11 - 26 games: W won 14 and B won 12

W 12 - 26 games: W won 14 and B won 12

B 13 - 13 games: W won 7 and B won 6

W 14 - 13 games: W won 7 and B won 6

B 15 - 8 games: W won 4 and B won 4

There are a couple of interesting points here (based in part on some further analysis of the data on the CD). The reference point is White 10 where the winning percentage for White exceeds that for Black:

• The sequence with White 10 first appeared in 1997 (checking back only to 1990, it may have appeared even earlier) in the 2nd Samsung Cup final between Yi Ch'ang-ho (B) and Kobayashi Satoru.
• Despite the fact that White 10 seems to have been invented by a Japanese pro, only 1 game featuring this move was from a Japanese domestic tournament (Kuwabara Yoko vs. Kobayashi Izumi in the Women's Honinbo in 1999). All others are from Chinese, Korean, or international tournaments. Is this an indication of the greater extent of fuseki research going on in China and Korea compared to Japan?
• White 10 appeared as follows by year:
  • 1997 - 1 game
  • 1998 - 1 game
  • 1999 - 4 games
  • 2000 - 15 games
  • 2001 - 11 games (last appeared in July). Does this indicate that the pros have reached their conclusions concerning this line and moved on to other ideas during the last 6 months? This may be so. The move that may have killed this line is Black a instead of 13. This was first played in February 2000 in a game between Cho Hun-hyeon (B) and Yu Bin in the LG Cup. That game was won by White. The move was not used again until May 2001 when Yu Ch'ang-hyeok played it in the Chunlan Cup against Wang Lei Sr.. In June Yu used the move again and won again. Finally in July Ch'oe Myeong-hun won as Black with the move and suddenly the pros were no longer interested in this line.

--Stefan: Fascinating! I knew that pros (especially in Korea) spend a very, very large amount of time on research, and I knew that there were certain preferred fuseki at any given point in time. But I didn't know the phases in fuseki fashion were as fast or frequent as this evidence seems to indicate.

HolIgor: It seems that the process that AvatarDJ mentions is really under way. It is a natural thing though. White did well with the position after move 10 so, the variation was popular. Then Black comes with a new move and starts to win more often than lose. Naturally nobody wants to continue it for white till somebody comes with a new idea for White countering Black's move at a. It is natural that the professionals study what other people play and remember the result. This is like the situation with chess, but it seems to me that this is not a new thing under the moon. In the era when everybody played Shusaku fuseki the move sequences were long too. They were much more difficult to collect and analyze though. With computer databases every pro can actually see statistics of any variation to the moment where the game really starts.

In go the choice of the moves in fuseki is wider than is chess, of course, and the variations should be shorter, but still they would be there. The actual game would happen in the chuban and yose, while there would be a quite decent amount of openings to use.

I would use this approach as I did not develop that fuseki intuition and play first 20-30 moves just scattering stones on the board trying to avoid situations that I know would be unpleasant for me.

Dieter: I'd like to react to one statement in AvatarDJFlux's highly interesting post. Couldn't all the energies spent in these intellectual quests be spent instead on playing and spreading the game?

As with all cultural and intellectual features, the quest for perfect play or formalization does not necessarily harm spreading and enjoying the game. I've been teaching math for a week now to 13-year olds. The methods used in the textbook are highly intuitive at the expense of logical reasoning and proof. The kids adore it. Meanwhile, others continue to formalize math at a very high level, or explore the extremities of the well known paths.

Learning the basics and exploring the limits are the opposite ends of the intellectual process. There is little reason to mix them up - although refreshing the basics does good to every explorer.

Charles Matthews A couple of comments. Firstly John Fairbairn of GoGod does collect nadare games as a priority. Secondly I find it helpful to attribute purely aggressive play to the Chinese influence in contemporary go. The Korean approach is more like an intermediate between that and the Japanese, artistic tradition in the game. It also does represent a convergence with the way chess has been studied. In some sense this seems to be a consequence of the general adoption of 4-4 points - and the difficulty of coming to any conclusions.

BobMcGuigan: What about Go Seigen's criticism of modern (Japanese?) go and his exposition of his ideas in his "21st Century" series? I don't know all the details but I remember he condemned the study of joseki saying that it is regrettable that the idea of "joseki" is all that has come out of go study in Japan. It seems to me that there is a lot more experimentation, and a lot more new patterns being played than 20 years ago. On the other hand the contrast between territory-oriented styles and influence (moyo) or thickness-oriented styles persists.

Charles When I have gone through recent yearbooks (Igo Nenkan) from Japan actually looking for innovations, that isn't the impression I have got. Joseki innovation is at a low ebb, it seems, compared to what one could read in the annual books by Abe Yoshiteru in past years. To some extent that's because modern fuseki patterns bypass older ideas on corner joseki, placing more emphasis on the sides. But I believe that if you did a comparison between a Japanese yearbook and the corresponding Chinese or (particularly) Korean books, you'd end up thinking that the Japanese are more conservative. That can certainly be quantified by breaking down the fuseki patterns (I do this for my own purposes, into about 100 types).

Bob: I was actually thinking of the Chinese and Koreans when I spoke of innovations. I haven't looked at Abe's new move and new pattern books for a while but I have the impression that most of the innovations occur in games between lower ranked players rather than the title holders. By the time the top players use an idea it probably doesn't qualify as new any more, but I think Cho Chikun has tried new things in title match games occasionally. Go Seigen's ideas reject the large scale joseki trends and emphasize overall fuseki ideas. Some of these have been tried in title matches, for example white plays the high two space approach to black's 3-4 corner stone, black answers at 5-3 and white tenukis. This does emphasize the side more than the corner, as Charles said.

Charles Yes, following this up on a database, a new pattern emerging since 2000.

A Kobayashi variant  

This is being tried, with the Kobayashi fuseki when is played in particular. So White avoids at a, a well known pattern - instead trying a lighter and more pacy play.

I get the feeling that this is typical enough, incremental change within a very familiar setting. The move at is imaginative, somewhat in the shinfuseki style, but not a radical departure.

BobMcGuigan: is an interesting move. After it is played a white move at a, resuming the joseki for the , exchange, becomes very good for White since she is now one move ahead in the sequence. So Black would want to play at a very soon, if not right away. Then White might abandon temporarily, treating it lightly.

ThorAvaTahr (2008): I have the Smartgo 2.82 software, which incorporates a large database of 43 100 pro games (going back to very old game records of shushaku etc...). I added to this about 1100 pro games played in the years of 2007-2008 and some 5600 undoubled pro games i found in a combination of other database sources. In this database I walked through the widest path (symmetries accounted for). The widest path is 41 moves long, before it is less than 3 games wide: Widest Path/SmartGo

RandomThoughts (2015): I think this critizism is missing an integral point about the relationship between knowledge and intuition. It is my understanding that intuition / feeling is really guesswork, which is applied in the absence of concrete knowledge.

In comparison, there is no intuition involved in Tic-Tac-Toe because that game has already been solved, we have exhaustive trees of all possible moves and the results they produce. One doesn't have to guess what a good move would be in any Tic-Tac-Toe game because one can look it up (if one hasn't already memorized it).

The difference between Go and Tic-Tac-Toe is that Go has not been solved, not by a long shot, and the body of knowledge produced so far widely extends beyond what most of us "normal mortals" can ever memorize - which is where the relevance of Intuition lies:

While playing games, we will inevitably run into situations where we don't know what the best possible moves are - so we guess. We rely on our experience, our instinct, our feel and put the stone down and hope for the best.

And no research done by any Dan players anywhere will ever change that - not until they solve the game and devise the "hand of god" algorythm that provides the perfect move in any possible situation.