As with the first volume of Recent Top Pro Go Games, these very thorough game commentaries are somewhat similar to those published in my Deep Thought books. However, I give a lot of attention to the impact of the AI programs in these new commentaries. These programs did not exist before those earlier books were published. The top pros generally study the games of the AI programs extensively.
It is important to realize that the AI programs are still exploring the game. After millions of games they are still making changes in how they play. Some of the recent changes in how the AI programs play come up in the games in this book.
One of the striking things about the top AI programs that has not changed yet is that they all see White as having an initial advantage in the game with the seven and a half point komi. How Black can gain the advantage in the opening has not yet been discovered. White’s advantage is not huge: something like a two or three percent better chance of winning. Nevertheless, the message seems to be that Black cannot expect to gain an advantage in the first fifty or so moves unless White makes a mistake. This means that Black has to focus on finding a way to get ahead in the middle or end game. Even after millions of games, there are still possibilities in the middle and end game that the AI programs have not had a chance to thoroughly explore.
This means that Black must play more aggressively while White can play calmly and solidly.
Since AlphaGo retired, many AI programs have been developed. The Chinese AI program Fine Arts is the strongest at this point and functions as the coach of the Chinese national pro team. It is still able to push the top players to a handicap of two stones without komi. This is the AI program I will be referring to in the commentaries.
This game was played on July 31, 2020, in the second round of the international Chunlan Cup. The players are Ke Jie 9p of China, whom many believe is the best human player in the world. His opponent is Kang Dongyun 9p of Korea. Both of these players have won previous world championships. Ke Jie has won seven and is hoping for an eighth since that would tie the current Chinese record of eight world championships held by Gu Li 9p.
As with all international events during the pandemic, this game was played on the internet. We will see a number of the Fine Arts AI program’s moves in the play of both of these pros. Ke Jie is White and Kang Dongyun is Black. Since this event is sponsored by a Chinese company, it follows Chinese rules with a komi of seven and a half points and dame points count.
Moves 1–4
Kang Dongyun is a more territorial player so we are not surprised that he starts with a 3–4 point stone. The AI programs tend to play 4–4 points with the black stones, apparently seeing a 3–4 point stone as a little slow. Opposing black 3–4 point stones is a traditional territorial opening. It was very popular in the mid-twentieth century.
Move 5
In the past, Black would defend one of his corners with 5, but thanks to the AI programs, today we know that, to overcome the komi, Black needs to play a little faster, more aggressively. So Black approaches White’s corner first.
Move 6
Besides the game move, the AI programs play 6 at A or B in this situation. So no one plays the pincers anymore.
Moves 7–9
In the past this joseki was not considered good for White. However, since the AI programs play this way frequently as White and win the game, it is now popular. Nevertheless, a lot of human players still have doubts about this joseki being a good choice for White, but it is easy and a quick way for White to get sente. With 9 high it is easy for Black to make a further extension, but this also offers White a good attack at A. If 9 is low, it means Black is not planning to continue at the bottom. I would make the low play myself.
Move 10
Since the corners are bigger than the sides, White approaches with 10. So there is a big play waiting for White at A or for Black at B.
Move 11
Black does not pincer with 11. He chooses Honinbo Shusaku’s diagonal play instead. That this is a good move is confirmed by the AI programs. White has choices for a response. It is not necessary for White to respond locally immediately, unlike the earlier situation in the lower left, when White kicked with 6.
Move 12
White breaks Black’s upper left corner with 12. Breaking a corner with a 3–4 point stone is a high priority in the opening. Black has many choices now. Because of being lower, the value of the left side has decreased. So Black does not continue on the left side.
Move 13
Since White did not defend at the top right, Black 13 is a good follow-up to attack White. White still has the choice of going into the corner or coming out.
Move 14
The AI programs have made it clear that when Black plays elsewhere after White’s approach to the upper left corner, it is very important for White to play this 14 as soon as possible.
Moves 14–16
The AI programs see this development as a definite gain for White.
The normal response for Black is to jump to A, but makes this a low priority for Black. After Black A, White would take sente to play at the top.
Moves 17–18
Normally, this would be seen as good for White. Black is pushing from behind on the third line. White seems to be getting a much more valuable result. Black’s plan is to get sente here and go back to the top.
Move 19
So Black continues to push with 19. After another white response (at A), Black’s plan is to play at B. White would then turn at C and Black would settle his group with D. However, that would mean White was settling for only a small potential on the lower left side. This development would obviously be better for Black.
Move 20
So White sees that an extension at 20 would be much better for him. Note that if Black now pushes at A and cuts at C after White B, a ladder is involved. This ladder would be good for White, which is why White can play as in the game. If White had played 18 () at 20, there would not have been a ladder if Black pushed and cut.
Move 21
Normally, a turn at A would have been good for Black, but not with in place. Also, normally, Black would play at B with 21, and White would take sente to play elsewhere. But even after Black B, this corner would still have a lot of aji after White C, Black D, White E. So Black B would not secure this corner for Black. So Black attaches with 21, hoping to reduce White’s potential to the left while expanding Black’s potential in the upper right center.
continued (21)
Black wanted White to hane at A so he could pull back at B, leaving a cutting point at C that White would have to defend.
Moves 22–24
White does not follow Black’s plan. Instead, White aims to connect to his stone in the upper right corner. Normally, this would not be seen as good play by White since it invites Black to hane at the head of two stones at A. That would put a lot of pressure on White since White’s liberties would be very tight. However, White sees the value of connecting to as bigger. Ke Jie’s play is very flexible, which is part of why he is considered the most clever human player in the world.
A Normal Continuation
A normal continuation for 25 would be to double hane at 1 with this result. Most players would say Black should play this way.
Moves 25–26
Black doesn’t want to give White a prisoner as in the normal continuation and thought he could do better than normal here by starting with this 25. However, White doesn’t respond at A as in the normal sequence shown in the previous diagram. Instead he cuts at 26.
In the full book, this game is analyzed to the end.
PK pY OPS/img/PK pYvǠ OPS/img/sg0160_ss_rtp2_cover.jpg JFIF tExif MM * > F( i N 8Photoshop 3.0 8BIM 8BIM% ُ B~
0ICC_PROFILE
appl mntrRGB XYZ acspAPPL APPL -appl desc P bdscm cprt #wtpt rXYZ gXYZ bXYZ rTRC $ aarg 0 vcgt P 0ndin >mmod (vcgp 8bTRC $ gTRC $ aabg 0 aagg 0 desc Display mluc ' hrHR koKR nbNO id huHU csCZ slSI daDK nlNL fiFI itIT esES roRO frCA ar ukUA heIL zhTW viVN skSK zhCN ruRU enGB frFR ms hiIN thTH caES enAU esXL deDE enUS ptBR plPL elGR svSE trTR ptPT jaJP L G U l t r a F i n etext Copyright Apple Inc., 2024 XYZ XYZ =XYZ J `
XYZ (] ȼcurv
# ( - 2 6 ; @ E J O T Y ^ c h m r w |
%+28>ELRY`gnu|&/8AKT]gqz !-8COZfr~ -;HUcq~
+:IXgw'7HYj{+=Oat2FZn % : O d y
'
=
T
j
"9Qi*C\u
&
@
Z
t
.Id %A^z &Ca~1Om&Ed#Cc'Ij4Vx&IlAe@e Ek*Qw;c*R{Gp@j>i A l !!H!u!!!"'"U"""#
#8#f###$$M$|$$% %8%h%%%&'&W&&&''I'z''(
(?(q(())8)k))**5*h**++6+i++,,9,n,,--A-v--..L.../$/Z///050l0011J1112*2c223
3F3334+4e4455M555676r667$7`7788P8899B999:6:t::;-;k;;<' >`>>?!?a??@#@d@@A)AjAAB0BrBBC:C}CDDGDDEEUEEF"FgFFG5G{GHHKHHIIcIIJ7J}JKKSKKL*LrLMMJMMN%NnNO OIOOP'PqPQQPQQR1R|RSS_SSTBTTU(UuUVV\VVWDWWX/X}XYYiYZZVZZ[E[[\5\\]']x]^^l^__a_``W``aOaabIbbcCccd@dde=eef=ffg=ggh?hhiCiijHjjkOkklWlmm`mnnknooxop+ppq:qqrKrss]sttptu(uuv>vvwVwxxnxy*yyzFz{{c{|!||}A}~~b~#G
k͂0WGrׇ;iΉ3dʋ0cʍ1fΏ6n֑?zM _ɖ4
uL$h՛BdҞ@iءG&vVǥ8nRĩ7u\ЭD- u`ֲK³8%yhYѹJº;.!
zpg_XQKFAǿ=ȼ:ɹ8ʷ6˶5̵5͵6ζ7ϸ9к<Ѿ?DINU\dlvۀ܊ݖޢ)߯6DScs
2F[p(@Xr4Pm8Ww)Kmpara ff
Y
[vcgt ndin 6 R C &