Artificial intelligence chess

liu, tempo Date: 2021-08-17 09:39:07
Views:48 Reply:0

Go fans often use the term “draw” to make fun of Chinese chess. It is generally believed that the reason why there are so many draws in Chess is that there are not as many changes in chess as in Go, and many moves have been thoroughly studied. Others believe that in chess, the first player is almost certain to win because the red player takes advantage of the first player, and the analysis by chess software shows that the first player is almost certain to win, while in Go, there is no first player to win. Therefore, it is concluded that the lack of changes in chess leads to the first player winning.


In fact, this is a very wrong understanding. No matter in chess or go, the probability of winning is very high when the first move is made. Since there was no rule of keeping eyes on go in ancient times, the situation of drawing chess often occurred in ancient times. The field of Go is not like the outside world now: “there will not be a draw”! There is no such thing.


If you look at go alone (regardless of the score), black’s win rate is close to 80%(katago’s latest weight score). Such a win rate is almost a must for black in a game between AI, which is no different from chess’s win rate! In the case of a go draw, there is also a high chance of a draw, according to the training data given by Leela Zero Sai.


It can be said that the computer AI chess almost exhausted the changes in chess at the same time, also almost exhausted the changes in go! Many go layouts, such as “Chinese flow” layout, “Three consecutive stars” layout, “Xiucuryu” layout and “universe flow” layout, have been eliminated by computers. At present, in the world of artificial intelligence, there are only stars and two consecutive stars left. Isn’t this just the result of numerous changes?


If this is not an end, how can it be?


The reason why there is no draw in the go game (except for three rounds) is the result of external intervention. To put it simply, it is not the essence of Go. After all, the pure rules of Go have no such thing as stickers.


Artificial intelligence chess


The creation of the go-playing rule is a new system implemented since The professionalization of Go in Japan in the last century. The purpose of the rule is to reduce the first-move advantage of black with the number of eyes, and to achieve the balance of power between black and white. At that time, the Japanese and South Korean rules were black and white five and a half, and later changed to six and a half. Since the weiqi system has not formed a completely unified standard, so, the Chinese rule is black stick 2 and three quarters, and then changed to 3 and three quarters.


Black stick 7 and a half, white win rate is close to 60%, indicating that there is a problem with this stick itself! And some people still want to ask the computer to play the optimal solution under the problematic rules, ask: you have a problem with the rules, how can you make the computer play the optimal solution of go like chess? It’s not a computer problem, it’s a human rule problem.


Therefore, it is not easy to judge whether go has more changes or chess has more changes.


Mr. Nie Weiping, a national weiqi player, said that the more chess is played, the easier it is, because the more chess is played, the fewer pieces there are. In fact, this view is extremely one-sided, chess endgame, although the pieces are less, but the moves have become more, more moves, all kinds of traps also followed! Therefore, it is not unreasonable for many people to think that “the best part of chess is the end game”.


Chess fans always counter go fans by saying that it is easy to go to the end because as the number of pieces increases, there are fewer Spaces on the board, so there are fewer changes to choose from. In fact, this view is extremely one-sided, just as Mr. Nie Weiping’s view of chess is one-sided.


In Weiqi, if both sides enter the official stage when all sides are completely set, then you can say: The more you play weiqi, the easier it will be. If the edges and corners are not finalized, the two sides into the official contention, this situation is extremely complex, even more complex than the mid-game battle. In most cases of Go, when black and white are evenly matched in the middle set, there is no time to deal with the corners.


Therefore, whether the situation of Chess and Go is more complex, or even simpler, can not be generalized, everything depends on the specific situation.

Leave a comment

You must Register or Login to post a comment.