Ever wondered ....

snorri · **#41**

I think it is possible. If 19x19 is solved or, for example, computers become much stronger than pros, it could be that games will arise that are very strong but are quite different from modern pro games. On direction is incredible vulgarity: like in life&death, sometimes the answer is a move that normally would be bad shape but works in this specific case. So I can imagine that the strongest go would include many one-off moves that only work in that particular game because of some specific tactical situations. (It's like the joke: what's the difference between a Japanese joseki and a Korean jungsuk? Answer: In the Japanese joseki, each player gets an equally good result. In the Korean jungsuk, each player gets an equally bad result. )

Another possibility is mind-bogglingly boring, peaceful games---perhaps a single pushing battle dividing up the board into two parts with no contesting or invasion as it might be analyzed that it is futile.

paK0 · **#42**

Splatted wrote:

paK0 wrote:

Research is usually done because someone believes there is a reason for it to exist.

I'm not saying they know everything, but I firmly believe that if they would think than another opening move would be better then they would put the time in and research that. Obviously they can't consider every possible opening, but I guess a lot of the opening moves that have potential have been explored to some extend, and the ones that we don't see have been dismissed for one reason or another(so if we see a new opening move its probably more likely that its a move that has been considered before and has been given a new twist, rather than a completely new invented move).

Unfortunately I can't remember the source, but I once watched an interview with a pro who talked about the aims of study groups. He explained that finding a better move was just a one off advantage because once it's been used in a public match everyone else will have studied it too, and so the goal of these groups is not so much to find the best moves, but to find new moves that will put them in a situation that their opponents haven't studied and they have. A new type of opening might be one way of doing this, but I imagine it would also take a lot more work than slightly deviating from a standard sequence and the advantage would be just as short lived.

Its not really about surprise value, but a move that gives an objective advantage over what we have now. Right now the pros play 4-4 because they think its good, noone is like:"Yeah, I'll open 4-4, my opponent will never see that coming".

snorri wrote:

I think it is possible. If 19x19 is solved or, for example, computers become much stronger than pros, it could be that games will arise that are very strong but are quite different from modern pro games. On direction is incredible vulgarity: like in life&death, sometimes the answer is a move that normally would be bad shape but works in this specific case. So I can imagine that the strongest go would include many one-off moves that only work in that particular game because of some specific tactical situations. (It's like the joke: what's the difference between a Japanese joseki and a Korean jungsuk? Answer: In the Japanese joseki, each player gets an equally good result. In the Korean jungsuk, each player gets an equally bad result. )

Another possibility is mind-bogglingly boring, peaceful games---perhaps a single pushing battle dividing up the board into two parts with no contesting or invasion as it might be analyzed that it is futile.

I think the first version is quite likely, but it may take some time until we are there. One of the top chess players once said he hates to play against computers, its like playing an idiot that always wins^^.

Wasiqi · **#43**

I think an interesting question to ask is how is Chess different from Go. I'm not really knowledgable about such things but how do computers evaluate a situation like ko? How do computers perform positional judgement?

I think Chess has some clear fundamental advantages in regards to computing in comparison to Go. Just the search space alone is much much smaller. I'm reminded of this article

http://en.chessbase.com/home/TabId/211/PostId/4008047

If you look at the hardware requirements needed to solve that opening, it seems like quite a bit.

I imagine it would be possible to solve an opening like san ren sei but at what computational cost?

While this isn't to say that some relatively strong Go computer programs don't exist or can't exist, what would be the source of their strength? Tactical calculation?

I wonder how computer Go treats the opening? Does it just copy known openings?

judicata · **#44**

Someone should condense this into a few questions and ask Takemiya what he thinks at the upcoming US Go Congress.

wineandgolover · **#45**

judicata wrote:

Someone should condense this into a few questions and ask Takemiya what he thinks at the upcoming US Go Congress.

And he will say, "Play the move you want to play," like he always does*. He likes people to play moves that feel right to them and make them happy. That's his schtick.

I wonder if software "wants" to play a move?

*always = the last two Congresses he attended.

SmoothOper · **#46**

Theoretically if everyone opened at different openings 8-8,7-8,8-9,9-5 there would be no standard openings.

In my opinion the standard openings being prevalent at top levels is due to the players learning to play at such a young age, and that the players are so heavily invested in those openings that they couldn't play a different opening if they wanted, it kind of reminds me of a game I saw where both players played non-standard corner openings but then at move forty, you couldn't tell that they hadn't just played hoshi points. I mean 12 years of research on a hoshi opening isn't going to happen overnight. Of course these young players are so strong, it takes a tectonic shift to dislodge them, and persuade them to think differently. Computer based research may lead to a situation, where a go program is able to prove that alternate openings are indeed better, or allow adept younger players to research new lines of play faster and independently of study groups, facilitating less hegemonous lines of play.

oren · **#47**

Or simply they would come back and find starting closer to the corners worked better to get their win rate higher.

Yamashita Keigo and many other players experimented for a time with tengen. The win rate was pretty much 50% but they were players whose career win rate was generally higher than that.

SmoothOper · **#48**

I forgot to mention, don't forget the handicap system encouraging hoshi play, as a ranking system discouraging people from exploring alternate plays while developing skills.

judicata · **#49**

SmoothOper wrote:

I forgot to mention, don't forget the handicap system encouraging hoshi play, as a ranking system discouraging people from exploring alternate plays while developing skills.

Since I more often lost as black (and still do) in handicap games, for a long while, handicap games encouraged me not to play 4-4 openings.

moyoaji · **#50**

SmoothOper wrote:

I forgot to mention, don't forget the handicap system encouraging hoshi play, as a ranking system discouraging people from exploring alternate plays while developing skills.

Well, if you go back 100 years the star point was not seen as a viable opening. The star point stone given as a handicap meant that you still didn't get any territory at the start of the game, only influence. Influence was not as highly valued before the shin-fuseki era. If you look at Shusaku's games the only opening moves were the 3-4 and 5-3. You'll also see that almost every approach to the komoku is low - a more territorial move. Even in handicap games white would play a 3-4 point away from the hoshi stone, which, according to Kajiwara Takeo in the Direction of Play, is something that shouldn't be done because it gives black an ideal approach situation since 4-4 stones are biased toward the sides.

The star point stone is a good example of how conventional thinking has given way to modern playing style. A move that was once seen as bad is now seen as the most common opening move in the game.

snorri · **#51**

SmoothOper wrote:

Theoretically if everyone opened at different openings 8-8,7-8,8-9,9-5 there would be no standard openings.

I've considered the following idea. Bring two 8-sided dice to a tournament. Whenever I'm black, roll them before my 1st move. Add two points to each number, and play that move. (So nothing starting on the 2nd line or below, which I guess we assume is inferior, despite the habits of certain Polish 4-dans.) I'm concerned about insulting people, though. I guess I can lessen the effect by not rolling the dice in front of them.

BTW, I don't think this makes sense as white. Once black has made a move, white has to immediately start thinking about what that move means in relation to future plays.

judicata · **#52**

snorri wrote:

I've considered the following idea. Bring two 8-sided dice to a tournament. Whenever I'm black, roll them before my 1st move. Add two points to each number, and play that move. (So nothing starting on the 2nd line or below, which I guess we assume is inferior, despite the habits of certain Polish 4-dans.) I'm concerned about insulting people, though. I guess I can lessen the effect by not rolling the dice in front of them.

Occasionally, I would use a somewhat similar method on whether to check or raise some hands in poker (to add some unpredictability to my strategy). If a clock with seconds or a second hand was nearby, I'd check if it was under 30 seconds and raise if over 30. This would avoid the blatant rolling of dice in front of your opponent (although the irony/humor of seeing someone rolling dice while playing go favors the dice method). But no simple/clean way to relate the seconds to which points you use immediately comes to mind. Sure, you could assign different coordinates to different ranges of seconds, but that's a little forced.

ez4u · **#53**

Wasiqi wrote:

...
I'm reminded of this article

http://en.chessbase.com/home/TabId/211/PostId/4008047

...

Amazing article! Thanks for posting this link.

The posted article - 'Rajlich: Busting the King's Gambit, this time for sure' wrote:

How much processing time went into your project?

Approximately – well actually quite precisely – 10,750,000 hours of single-core CPU time. The King's Gambit run took a little over four months in total, elapsed time, to calculate.

That was on about 3000 cores, right? What was the system doing?
...

Will this take all of the creativity and flexibility out of opening play?

Quite the opposite. I think players will see that many surprising moves are playable, giving them more options to choose from. For example, take the game Topalov-Kramnik from Wijk aan Zee 2008. The genius of Topalov's preparation wasn't that 12.Nxf7 is some sort of great move. It was simply realizing that 12.Nxf7 is playable and doesn't lose by force. Once you establish that, it's not that hard to work out a way to put a ton of pressure on your opponent.
...

Okay, cough it up: what next?

I say this with some apprehension, but we've been looking at the 6.Bg5 Najdorf (10.e5! looks promising as a forced win for White). But if we decide to do the full classification we are going to need much faster hardware. Maybe a factor of a thousand over what we used for the King’s Gambit. There are companies that periodically test that kind of hardware for new server parks, and run very CPU-intensive applications on them to locate faulty processors. We have made contact with the technical director of the Google server farms, who was definitely interested. So maybe in a year or two we will have solved the Najdorf.
...

[Emphasis added]

Boidhre · **#54**

What's really scary there is that they didn't solve that opening position in 4 months and over 3000 cores, they merely tested a small fraction of the variations and assumed a win once either side had better than a 5 pawn advantage. This is a fairly reasonable assumption for top level players, yet it's a very long way from totally solving it. When you consider the extreme difficulty of accurately evaluating a go position in the midgame for a computer compared to how relatively easy it is for a chess engine...

SmoothOper · **#55**

I wonder if there is an analog to the term gambit. Cho Chikun has a book in the mycom series, that google translate translates the title as gambits, I am tempted to get it, but I am afraid it won't be like the other problem books in the series. However, I gather risky plays are mostly avoided.

Here is the translation of the reviews I am talking about.

http://translate.google.com/translate?s ... nts%3D1%0A

It looks like there are more reviews saying that it is a problem book, it might be worth a try.

Boidhre · **#56**

Well a gambit in chess is an early sacrifice of material for positional or future gain. An analogy in go would be sacrificing a group of stones during the opening in exchange for "gaining a tempo or two" to develop faster globally or some other benefit. Crucially though, it can't be a forced capture for the opponent, the opponent would have to have the option of taking the stones or not. At its simplest it's just an offer of a trade of some stones and the attached territory for potential points very early in the game from my understanding.

SmoothOper · **#57**

Boidhre wrote:

Well a gambit in chess is an early sacrifice of material for positional or future gain. An analogy in go would be sacrificing a group of stones during the opening in exchange for "gaining a tempo or two" to develop faster globally or some other benefit. Crucially though, it can't be a forced capture for the opponent, the opponent would have to have the option of taking the stones or not. At its simplest it's just an offer of a trade of some stones and the attached territory for potential points very early in the game from my understanding.

I agree, that the analogy is odd, in the chess sense, gambits simplify the game with fewer pieces, whereas trading stones and leaving weak groups usually leaves aji, that can become very complicated. Maybe a better analogy would be making overly thick/heavy groups, and capturing groups too early in gote.

oren · **#58**

They translates 'fuseki' to 'gambit'.

The title is just breakthrough to shodan, fuseki at a glance.

Bill Spight · **#59**

On various plays in the opening:

Tengen.

1670.

I believe that when go had setup stones on diagonal 4-4 points, tengen was popular in medieval times. 14th century?

3-3 vs. 4-4.

1838.

yoyoma · **#60**

Wasiqi wrote:

...
I'm reminded of this article

http://en.chessbase.com/home/TabId/211/PostId/4008047

...

This was an April fool's joke. I added emphasis below:

From the article wrote:

On March 31 the author of the Rybka program, Vasik Rajlich, and his family moved from Warsaw, Poland to a new appartment in Budapest, Hungary. The next day, in spite of the bustle of moving boxes and setting up phone and Internet connections Vas, kindly agreed to the following interview, which had been planned some months ago.

Ever wondered ....

Who is online