(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[0]
OT[1x5 byo-yomi]
PW[AlphaGo]
PB[AlphaGo]
DT[2016-02-29]
C[Game 1 - Fighting
2016-2-29
White: AlphaGo
Black: AlphaGo
Komi: 7.5
Result: W+2.5
Commentary by Fan Hui
Go expert analysis by Gu Li and Zhou Ruiyang
Translated by Lucas Baker, Teddy Collins, and Thore Graepel
Before we begin, I would like to note that these games were played very quickly. AlphaGo's self-play games often take place under blitz time settings, with only 5 seconds per move. Obviously, this would be extremely fast for human players, and even AlphaGo understandably makes a few mistakes when playing at this speed.
This game was played on February 29, 2016, nine days prior to the Lee Sedol match.]
RE[W+2.50]
PL[B]
MULTIGOGM[1]
GN[AGvsAG-1]
;
B[dp]
;
W[dd]
;
B[qp]
C[Black 3:
As viewers of the match will remember, AlphaGo has a strong proclivity for the Chinese opening.]
;
W[pd]
;
B[qf]
C[Black 5:
In this game, Black approaches the corner before completing the Chinese opening.]
;
W[nc]
;
B[kq]
;
W[cn]
;
B[fp]
;
W[cq]
C[White 10:
White approaches from the left, then invades the corner at the 3-3 point. This opening appears frequently in professional games.]
;
B[cp]
;
W[bp]
;
B[dq]
LB[bo:A]
C[Black 13:
Black 13 struck Gu Li and Zhou Ruiyang as highly unusual. They wondered, "Will this be enough?" A professional would normally block at A. Nonetheless, as the match with Lee Sedol demonstrated, AlphaGo?�s strategic brilliance cannot easily be discounted. We know that it has an outstanding sense of the whole board, can control every aspect of the game, and takes great care to coordinate every stone with all the others. In light of this, from a global perspective, Black 13 was probably not a mistake.]
;
W[br]
;
B[cr]
;
W[bq]
;
B[fc]
C[Black 17:
After the corner is settled, Black plays not on the right side, but on the top.]
;
W[cf]
;
B[jd]
C[Black 19:
This extension appears frequently in AlphaGo?�s self-play games.]
;
W[op]
C[White 20:
White then approaches at the bottom, but doesn't Go theory advise to play on the wider side? Should White have played on the right? It is hard to say. The variations show two alternatives, but both turn out less than ideally for White. It seems that, if Black 19 is a product of AlphaGo's excellent positional judgment, then White's corner approach at 20 makes sense as a way to counter it.]
;
B[pn]
;
W[qq]
;
B[rq]
;
W[pq]
;
B[rr]
;
W[mq]
C[White 26:
The joseki through here is a common way to settle the local shape in the Chinese opening.]
;
B[qc]
C[Black 27:
Now Black is free to invade the upper right corner.]
;
W[qd]
;
B[rd]
;
W[pc]
LB[re:A]
C[White 30:
Although a few rare moves have appeared so far, all of them are understandable. After seeing this turn by White, however, our shock was palpable: "This move is so defensive!" "No human would play like this!"
White's playing style looks extremely cautious. Why not the hane? With the analysis shown in the variation, we began to understand White's reasoning, but Gu Li and Zhou Ruiyang emphasised that this would be very difficult prescription for professional players to swallow.]
;
B[rb]
;
W[qb]
;
B[rc]
;
W[hc]
;
B[lc]
LB[hc:B][pc:A]
C[Black 35:
In these few days of research, AlphaGo never ceased to amaze. The invasion at B is the proper follow-up to White A, but Black astonished us once again by responding indirectly with this knight's move. "There is nothing AlphaGo will not do! It fears nothing!" "No human would play this move either!". Of course, the more we saw of these audacious moves over time, the less they surprised us.
Why doesn't Black press down directly? See the variation.]
;
W[he]
;
B[of]
;
W[nf]
C[White 38:
An excellent attachment! Both Gu Li and Zhou Ruiyang applauded this move.]
;
B[ng]
LB[ne:A]
C[Black 39:
The commonsense response, but what about Black A instead - could Black have haned and cut immediately? See the variation.]
;
W[mf]
;
B[oe]
;
W[md]
LB[oe:A]
TR[md]
C[White 42:
In our research, Gu Li and Zhou Ruiyang concluded that Black A is problematic, since the exchange with the marked stone clearly strengthens White. Playing this move at the marked stone instead would have caused White more shape problems.]
;
B[jf]
;
W[mg]
;
B[nh]
;
W[mh]
;
B[ic]
;
W[hb]
;
B[ni]
;
W[mi]
;
B[nj]
C[Black 51:
All the moves up to here are mandatory. After Black extends, AlphaGo assesses Black's win rate at 50%: a completely balanced game.
AlphaGo plays a wide range of different openings, and many of them resemble this one to some degree. The more games it plays, the more chances AlphaGo will have to develop new variations of this opening family.]
;
W[ig]
;
B[jg]
;
W[jh]
;
B[ih]
;
W[ii]
;
B[hh]
;
W[hg]
;
B[kh]
;
W[ji]
;
B[ki]
;
W[gh]
;
B[lj]
TR[hh][ih]
C[Black 63:
After sacrificing the two marked stones in return for this kosumi, Black's win rate rises to 52%.]
;
W[ld]
;
B[kd]
;
W[lb]
;
B[kb]
;
W[mb]
;
B[nn]
C[Black 69:
The jump at 69 is a natural way of strengthening the right while increasing the pressure on the bottom. Black may have judged this result acceptable because it secures ample territory. However, it is also important to note that White's top side group is alive because of Black's aji.]
;
W[mo]
;
B[no]
;
W[np]
;
B[lp]
;
W[kr]
;
B[jr]
;
W[lr]
LB[mp:B][jq:A]
C[White 76:
The moves up to here are a middle game joseki for the Chinese opening, but it is important to note that Black usually exchanges A for B. Black omits this connection in the game, retaining the possibility of later cutting at B instead.]
;
B[cm]
C[Black 77:
The attachment at 77 is a common tactic for AlphaGo. The aim is simple: create a ladder breaker for the two stones in the centre.]
;
W[dm]
;
B[dl]
;
W[em]
C[White 80:
White has no choice but to reply this way.]
;
B[bn]
C[Black 81:
The double hane is a very aggressive play. As shown in the variation, connecting peacefully does not lead to a good result.]
;
W[cl]
;
B[bm]
;
W[mj]
;
B[mk]
;
W[dk]
LB[mj:A]
C[White 86:
The isolated push at A is curiously timed. What was the thinking behind this push? Was it only to limit the scope of possible variations? Or did White plan to cut, then reconsider and turn back? Unfortunately, we do not know.]
;
B[gk]
C[Black 87:
Black plays a double ladder breaker.]
;
W[fj]
C[White 88:
White hangs tough, but this move allows Black to run out on the left.]
;
B[el]
C[Black 89:
Here, Black's win rate reaches 60%! It is clear that Black holds the advantage in this fight.]
;
W[fl]
;
B[ek]
;
W[fk]
;
B[ej]
;
W[cj]
;
B[ei]
;
W[gm]
LB[gm:A]
C[White 96:
All the moves through 96 are forced. See the variation for what happens if White omits A.]
;
B[jj]
LB[hi:A]
C[Black 97:
Black is now quite confident about the outlook, as Black believes that White must capture at A. However, White comes up with a better plan.]
;
W[ij]
C[White 98:
White's turn at 98 is better for the fight in the centre, but does White need to respond at all? See the variation.]
;
B[ci]
;
W[bl]
LB[bo:A]
C[White 100:
This move loses a few points, but gains thickness over the block at A. White is concentrating energy in preparation for the cut.]
;
B[bj]
;
W[bo]
;
B[bk]
;
W[lk]
C[White 104:
Now White can finally slice through in the centre, unsheathing a sharp and deadly blade! But is this move really as severe as it looks? If you believe Black can no longer hold the position together, then you have underestimated AlphaGo!]
;
B[jk]
C[Black 105:
The only move for Black.]
;
W[ll]
;
B[ik]
;
W[im]
;
B[jm]
TR[hh][ih]
C[Black 109:
Black could also have saved the marked stones, but since the central dragon is not yet alive, running out to safety takes priority.]
;
W[jn]
;
B[km]
;
W[nk]
C[White 112:
White 112 is good move order, forcing Black to invest another move on the right side.]
;
B[ml]
;
W[lm]
;
B[kn]
;
W[ln]
LB[nl:B][in:A]
C[White 116:
Now Black faces a dilemma: protect the dragon with A, or reinforce the right side territory with B?]
;
B[nl]
LB[in:A]
C[Black 117:
Black chooses territory. See the variation for what happens if Black plays at A instead.]
;
W[ko]
;
B[in]
;
W[jl]
;
B[hm]
C[Black 121:
This counter-squeeze is the critical tesuji to save the dragon, by connecting back to the left side.]
;
W[kl]
;
B[hn]
LB[jp:A][mp:B]
C[Black 123:
Although at first glance White's capture of the three centre stones looks spectacular, it is important to remember that protecting with A is sente. If White does not respond, Black can cut at B to kill all the white stones above. From this perspective, Black's loss is limited.]
;
W[ok]
C[White 124:
Although Black survived in the centre, White has been plotting for some time, and extends at 124! Even after Black's reinforcement, can White still live?]
;
B[dn]
C[Black 125:
With this atari, Black resolutely abandons the right side for the left!]
;
W[gj]
C[White 126:
White fixes up the middle in sente.]
;
B[il]
;
W[ql]
C[White 128:
Finally, White ensures life on the right side with 128.]
;
B[qm]
;
W[rm]
;
B[rn]
;
W[om]
;
B[nm]
;
W[pg]
C[White 134:
Black can no longer hope to kill White.]
;
B[jp]
C[Black 135:
Black 135 makes a tiger's mouth to reinforce the bottom.]
;
W[io]
C[White 136:
This move, testing Black's response, is a very agile reply.]
;
B[jo]
LB[ir:A]
TR[io]
C[Black 137:
Because of the marked white stone, the aji of the clamp at A remains. Could Black have defended differently? See the variation.]
;
W[qg]
;
B[pi]
;
W[rf]
;
B[rh]
;
W[og]
C[White 142:
The great exchange continues!
Ignoring the cutting point, White turns to capture a portion of the right side, letting Black cut in the centre and swallow up the entire group.]
;
B[mp]
C[Black 143:
Which is bigger, the right side or the centre? I must admit we don't know. In an exchange this complex, one must consider many aspects of the position, all of which may vary in response to small changes. A single mistake can lead to total collapse.
While playing out these sequences, we often felt overwhelmed due to the number of factors and the complexity of the calculations. To think that AlphaGo considers all of this, using just five seconds per move, is truly frightening!]
;
W[po]
;
B[qo]
;
W[ir]
C[White 146:
White takes advantage of Black's shorage of liberties with the clamp.]
;
B[iq]
;
W[js]
;
B[jq]
;
W[lq]
;
B[kp]
LB[hq:A]
C[Black 151:
Up to here, Black has no choice but to retreat, and White still has the potential of the hane at A.]
;
W[an]
C[White 152:
White connects up the left side.]
;
B[co]
;
W[am]
;
B[cn]
;
W[ck]
LB[hq:A][dr:B]
C[White 156:
When I saw this move, I was shocked! If Black plays at B, it seems White can no longer live on the side, and besides that is short of liberties. How could White possibly escape? But AlphaGo's reading was extremely careful, and White indeed has a way to cheat death. See the variations.]
;
B[nr]
LB[cg:B][bh:A]
C[Black 157:
Although White has succeeded in connecting up the left, Black has sente moves at A and B, and will have no difficulty living. Interestingly, Black tenukis to make this placement on the bottom!]
;
W[dr]
C[White 158:
Still more puzzling is the reply: White plays at 158 and captures a stone! Can Black no longer live on the left? Is White dead at the bottom? What happened!?
White may have chosen this way because living on the bottom gives Black a valuable sente endgame sequence. See the variation.]
;
B[er]
;
W[cs]
C[White 160:
After White 160, although Black can capture White on the bottom, White would then reinforce and kill Black on the left.]
;
B[bf]
C[Black 161:
Weighing the pros and cons, Black decides to make life on the left side.]
;
W[bg]
;
B[ce]
;
W[be]
;
B[df]
;
W[af]
C[White 166:
Now the position is ko. Could White have killed unconditionally? See the variation.]
;
B[cg]
C[Black 167:
Black exhibits AlphaGo?�s most distinctive characteristic?�persistence!]
;
W[cd]
C[White 168:
What if White just connects when Black plays ko? See the variation.]
;
B[bf]
;
W[or]
;
B[nq]
;
W[cf]
;
B[fi]
;
W[gi]
;
B[bf]
;
W[hq]
LB[hr:A][hs:B]
C[White 176:
This ko threat is an extremely pleasant one for White. Black cannot atari at A, or White could simply create a second ko at B.]
;
B[hp]
;
W[cf]
;
B[rg]
;
W[pf]
;
B[bf]
;
W[lo]
;
B[ip]
;
W[cf]
;
B[fn]
;
W[fm]
;
B[bh]
C[Black 187:
Now the balance of ko threats favors White, so Black concedes the ko to make life.]
;
W[de]
;
B[ak]
;
W[al]
;
B[dg]
;
W[gp]
;
B[ho]
;
W[fq]
C[This tiger's mouth is the final tesuji of the game.]
;
B[hr]
C[Black has no option but to capture. See the variation for what happens if Black resists.]
;
W[oi]
C[The long battle has finally reached its conclusion.
We finally understood now that White never intended to make life at the bottom. Instead, White wedges at 196. Could it be that White always thought the right side was bigger? Regardless of how it arrived at this decision, White has chosen the path to victory.]
;
B[mr]
C[Black must secure the bottom.]
;
W[re]
C[Finally, with this move, the endgame begins.]
;
B[ol]
;
W[pk]
;
B[pl]
;
W[rk]
;
B[pb]
;
W[ob]
;
B[qa]
;
W[eq]
;
B[ep]
;
W[fr]
;
B[ff]
;
W[ee]
;
B[gq]
;
W[ie]
;
B[je]
;
W[ef]
;
B[eg]
;
W[fh]
;
B[ao]
;
W[ap]
;
B[dj]
;
W[ao]
;
B[eh]
;
W[fg]
;
B[gr]
;
W[es]
;
B[sm]
;
W[sk]
C[White 226:
Just when I thought everything was over, I saw this move. Why doesn't White connect? Is there really something here?]
;
B[if]
;
W[hf]
C[White 228:
Now the small endgame begins. The difference is narrow, but the result has been decided in White's favor.]
;
B[ib]
;
W[id]
;
B[ha]
;
W[ga]
;
B[ia]
;
W[gb]
;
B[hk]
;
W[en]
;
B[rl]
;
W[qk]
;
B[hj]
;
W[hi]
;
B[ka]
;
W[fo]
;
B[go]
;
W[eo]
;
B[le]
;
W[lf]
;
B[kf]
;
W[oa]
;
B[pa]
;
W[gl]
;
B[hl]
;
W[li]
;
B[kj]
;
W[lg]
;
B[kg]
;
W[sl]
;
B[rm]
;
W[sd]
;
B[sc]
;
W[se]
;
B[gn]
;
W[me]
;
B[la]
;
W[ma]
;
B[gs]
;
W[fs]
;
B[ah]
;
W[ag]
;
B[ke]
;
W[mc]
;
B[kc]
;
W[do]
;
B[gp]
;
W[lh]
C[White 274:
White wins by 2.5 points.
This should be the part where I rack my brains to comprehensively summarize the game. Yet all I can say is: so many battles! Countless sacrifices! Countless exchanges! And all with only five seconds per move! In a game as endlessly complex as this one, both sides surely made a few errors, but I believe they were not the least bit significant. After all, error is also an essential and fascinating part of the game.
I am grateful to Gu Li and Zhou Ruiyang for their help, as I would not have been able to discover and comprehend these many variations on my own. I could never have understood the complex interactions in this game without them. At every step, they diligently and conscientiously searched for the best moves.
After playing through this game, we felt incredible fatigue. Yet we sensed that if we had continued our study, we would have uncovered even more fascinating variations. I must bring my commentary to an end here - but perhaps your own investigation has just begun.]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[0]
OT[1x5 byo-yomi]
PW[AlphaGo]
PB[AlphaGo]
DT[2016-02-29]
C[Game 2 - Slaying the Dragon
2016-3-1
White: AlphaGo
Black: AlphaGo
Komi: 7.5
Result: W+R
Commentary by Fan Hui
Go expert analysis by Gu Li and Zhou Ruiyang
Translated by Lucas Baker, Teddy Collins, and Thore Graepel
Like the first game, this is a blitz with five seconds per move. Since the first 48 moves are identical, please refer to the commentary on the first game for a thorough analysis of the opening.]
RE[W+Resign]
GN[AGvsAG-2]
;
B[dp]
;
W[pd]
;
B[dc]
;
W[pp]
;
B[nc]
;
W[qf]
;
B[ci]
;
W[fq]
;
B[dn]
;
W[cq]
;
B[dq]
;
W[dr]
;
B[cp]
;
W[br]
;
B[bq]
;
W[cr]
;
B[qn]
;
W[nq]
;
B[pj]
;
W[de]
;
B[fd]
;
W[cc]
;
B[cb]
;
W[cd]
;
B[bb]
;
W[cg]
;
B[qc]
;
W[pc]
;
B[pb]
;
W[qd]
;
B[rb]
;
W[rc]
;
B[qb]
;
W[ql]
;
B[qh]
;
W[ol]
;
B[ne]
;
W[nf]
;
B[mf]
;
W[ng]
;
B[oe]
;
W[pg]
;
B[nj]
;
W[mg]
;
B[lf]
;
W[lg]
;
B[qk]
;
W[rl]
;
B[kg]
LB[kf:A]
C[Black 49:
Black 49 is the first departure from the previous game. Previously, Black extended at A, so perhaps AlphaGo was dissatisfied with the result in that game.]
;
W[kh]
;
B[jh]
C[Black 51:
Now, Black chooses the double hane. If White defends as shown in the variation, the game looks promising for Black.]
;
W[md]
LB[md:52][kg:49][jh:51][kh:50]
C[White 52:
However, White has other plans, disrupting Black's rhythm with the sharp placement at 52.]
;
B[ld]
;
W[mc]
LB[mc:B][ld:A]
C[White 54:
The exchange of A for B leaves Black hard-pressed to fix up all the weaknesses on the top side.]
;
B[oc]
;
W[li]
;
B[ki]
;
W[lh]
;
B[nl]
;
W[om]
;
B[ji]
C[Black 61:
Instead of shoring up the top side directly, Black decides to hang tough and connect on the outside, harassing White on the right while looking for an opportune moment to reinforce the top. ]
;
W[lk]
;
B[ok]
;
W[mk]
;
B[nk]
;
W[lm]
;
B[nm]
;
W[pn]
C[White 68:
Bold though it may be, Black's way of playing is extremely dangerous. At this point, there is still a gap through which Black can break into the open, but the game has already grown painfully difficult. See the variation for White's anticipated continuation.]
;
B[kl]
C[Black 69:
This peep is a probe to test White's intentions.]
;
W[kk]
C[White 70:
White's push on the outside leaves a minor weakness, but is the most uncompromising way of playing.]
;
B[kf]
C[Black 71:
In response, Black chooses to risk it all by connecting at 71! Now the top is much stronger, but can Black live with the dragon on the right?]
;
W[rj]
;
B[rk]
;
W[qi]
TR[qi][rj]
C[White 74:
With the marked stones, White seals Black's fate. AlphaGo is capable of killing!]
;
B[mn]
;
W[ko]
;
B[mp]
;
W[kq]
;
B[mq]
;
W[nr]
;
B[kr]
;
W[jr]
;
B[sl]
;
W[qo]
;
B[on]
;
W[pl]
;
B[oo]
TR[on][oo]
C[Black 87:
The marked stones create some eye shape, but Black is quickly running out of space to live. Carving out the second eye will be a very difficult task.]
;
W[po]
;
B[rm]
;
W[rn]
;
B[kn]
LB[lo:A]
C[Black 91:
At this point, Black briefly entertains the notion that everything will be fine. However, this is merely a fantasy, and White will soon put an end to Black's hopes.
Actually, Black could also have made a ko starting with A, but the result would have been a Pyrrhic victory.]
;
W[ln]
;
B[lo]
;
W[lp]
;
B[mo]
;
W[mm]
LB[km:A]
C[White 96:
Black had hoped for White A, as shown in the variation, but this merciless blow lays bare the harsh reality.
Although the following fight looks complicated, the fact is that Black has lost the chance to live. White?�s technique may appear suboptimal, but from AlphaGo's perspective, it is the simplest and clearest line.]
;
B[pr]
;
W[pq]
;
B[or]
;
W[oq]
;
B[qm]
;
W[pm]
;
B[sm]
;
W[rp]
;
B[sn]
;
W[ro]
;
B[qq]
;
W[rr]
;
B[mr]
;
W[ks]
;
B[ls]
;
W[lr]
;
B[lq]
;
W[kp]
;
B[kr]
;
W[js]
;
B[qr]
;
W[ns]
;
B[rq]
;
W[sq]
;
B[sr]
;
W[ss]
;
B[qs]
;
W[sp]
;
B[os]
;
W[sr]
;
B[ms]
;
W[iq]
;
B[pk]
;
W[qp]
;
B[np]
;
W[nn]
C[White 132:
The result here is technically ko, but Black has no reasonable means of starting the ko. See the variation for what happens if Black tries.]
;
B[rh]
;
W[pi]
;
B[ph]
;
W[oi]
;
B[oh]
;
W[ni]
;
B[mj]
;
W[lj]
;
B[sj]
;
W[pf]
;
B[ri]
;
W[qj]
C[White 144:
This is Black's last chance to start the ko, but it will still end in failure. ]
;
B[sf]
;
W[rs]
C[White 146:
White kills unconditionally. The remaining moves may be regarded as Black's dying breath.]
;
B[sc]
;
W[sg]
;
B[rf]
;
W[re]
;
B[rg]
;
W[se]
;
B[jn]
;
W[ik]
;
B[ip]
;
W[hq]
;
B[jo]
;
W[jq]
;
B[jk]
;
W[jl]
;
B[jj]
;
W[im]
;
B[gp]
;
W[km]
;
B[gq]
;
W[gr]
;
B[hp]
;
W[fr]
;
B[gm]
;
W[hj]
;
B[hl]
;
W[il]
C[
Black resigns at move 172.
AlphaGo's unparalleled positional judgment and outstanding ability to control the flow of the game are well known. However, some have wondered about whether it can handle life and death situations, even questioned whether such positions might drive AlphaGo berserk. This game should prove to everyone that, indeed, AlphaGo can kill.]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[0]
OT[1x60 byo-yomi]
PW[AlphaGo]
PB[AlphaGo]
DT[2016-02-29]
C[Game 3 - Freedom
2016-3-2
White: AlphaGo
Black: AlphaGo
Komi: 7.5
Result: W+R
Commentary by Fan Hui
Go expert analysis by Gu Li and Zhou Ruiyang
Translated by Lucas Baker, Teddy Collins, and Thore Graepel
Whereas the first two games allowed only five seconds per move, this game was played at a more classical pace of one to two minutes per move. The difference between these time controls is night and day, and the slower pace dramatically improves AlphaGo's calculations. This helps explain why, as you will soon see, White proves more than willing to spurn conventional wisdom in this game!]
RE[W+Resign]
GN[AGvsAG-3]
;
B[pd]
;
W[dp]
;
B[cd]
;
W[pp]
;
B[cn]
C[Black 5:
Before setting up the Chinese opening, Black approaches the corner. What is this exchange about? When I asked Gu Li and Zhou Ruiyang, they said that while it is unusual, it does not lose anything.]
;
W[fq]
;
B[ic]
C[Black 7:
Here, AlphaGo once again demonstrates its preference for the Chinese opening.]
;
W[qf]
;
B[nd]
;
W[qc]
;
B[qd]
;
W[rd]
;
B[pc]
C[Black 13:
Black's block here would strike professional players as soft, but it turns out to be an acceptable move. See the commentary for the first blitz game for a deeper investigation.]
;
W[rb]
;
B[qb]
;
W[rc]
;
B[nq]
;
W[qn]
;
B[jp]
TR[jp][nq]
C[Black 19:
The approach and extension, a common strategy in AlphaGo's games, are likewise analyzed in the commentary for the first game.]
;
W[dd]
C[White 20:
Now we come to White 20. No, that is not a misclick, AlphaGo really played there! Of course, this is not something AlphaGo dreamed up on the spot, but rather a strategy fashioned over time through many self-play games. So, what is the reasoning behind this move? See the variations for an investigation of some tempting but wrong ways for Black to answer it.
While looking through these alternatives with Gu Li and Zhou Ruiyang, we discovered many complex variations. Although we have not reached a definite conclusion about this attachment, we would not be surprised to see it appear in many games to come. AlphaGo has opened our minds and inspired us to reassess conventional Go wisdom regarding what is right and wrong. Sometimes we instinctively reject moves because they clash with our experience or training, or even because we fear the ridicule of others. Yet these concerns obstruct our progress. AlphaGo does not have this layer of human prejudice, and that is what makes it so free, bold and unrestrained! As Go players, shouldn't we also seek to embody such a spirit?]
;
B[ce]
C[Black 21:
Black ultimately chooses to extend on the outside.]
;
W[de]
;
B[dc]
C[Black 23:
When Black plays this hane, it feels as if the corner might transpose to the avalanche joseki. Does Black have any other options? See the variations.]
;
W[ec]
;
B[cc]
C[Black 25:
As the variation shows, Black would get a bad result from the small avalanche, so Black simply connects.]
;
W[df]
;
B[cg]
C[Black 27:
It looks like Black is poised to attack White...]
;
W[id]
C[White 28:
...but White 28 is a severe counter! Gu Li and Zhou Ruiyang praised this move highly. At first it looks like an overplay, but actually, Black has no good response. Thanks to its powerful calculations, AlphaGo can play these sorts of strong moves without reservation. This is perhaps its greatest strength.]
;
B[jd]
C[Black 29:
The best response. See the variations for Black's alternatives.]
;
W[hc]
;
B[jc]
;
W[ie]
C[White 32:
Here, Black and White have reached an agreement of sorts. Black hanes on the outside, and White hanes and extends for a very comfortable position.]
;
B[je]
;
W[lq]
LB[if:A]
C[White 34:
Just as we were thinking White would be satisfied with the extension at A, White pleasantly surprised us by invading directly at the bottom! Could it be that AlphaGo does not understand the importance of a hane at the head of two stones? Truly, AlphaGo never ceases to amaze. One would never find a tenuki like this in a game between professionals.
What will happen if Black hanes now? See the variation.]
;
B[lp]
C[Black 35:
Once again, both sides have reached a common understanding. Disregarding the normal imperative to hane at the head of two stones, Black attaches on top.]
;
W[kp]
;
B[kq]
;
W[ko]
;
B[lr]
;
W[jq]
;
B[mq]
;
W[kr]
;
B[lq]
LB[ip:A]
TR[kp]
C[Black 43:
Generally, White cannot wedge with the marked stone when the ladder at A is unfavorable, so I requested that Zhou Ruiyang and Gu Li explore White A a bit further. See the variations.]
;
W[iq]
;
B[pr]
;
W[co]
LB[dn:A][bo:B]
C[White 46
is a common attacking technique, but AlphaGo's attitude towards it is quite unusual. The typical reply would be to extend at A, but AlphaGo prefers to tenuki, either right away or after exchanging the hane at B and the atari in sente. It seems AlphaGo has no fear of allowing the tiger's mouth at A.]
;
B[fc]
C[Black 47:
Black now clamps on the upper side, prompting White to descend. In the end, it seems neither player was ever concerned with the hane at the head of two stones.]
;
W[eb]
;
B[fd]
;
W[ed]
;
B[fe]
;
W[dg]
;
B[ch]
LB[dh:A]
C[Black 53:
Black extends, probably anticipating White A, but White has a way to improve on that result.]
;
W[gb]
;
B[fb]
;
W[dh]
LB[fb:B][gb:A]
C[White 56:
With the well-timed exchange of A for B, White's position is clearly better than the variation.]
;
B[hd]
;
W[di]
;
B[ci]
;
W[dj]
;
B[cj]
;
W[cf]
;
B[bf]
;
W[be]
LB[bd:A]
C[White 64:
A well-timed cut - Black has no choice but to connect. The variation shows the consequences of Black's capture at A.]
;
B[bg]
;
W[hh]
TR[hd]
C[White 66:
White then leaps out into the centre with the magnificent move of 66! This one move embodies all of AlphaGo?�s formidable genius. It strengthens White's group on the left, reduces Black's top side, and aims at the aji around the marked stone. One can even hear the faint echoes of a White moyo forming in the centre. If there were an ear-reddening move for AlphaGo, this would be it!]
;
B[ql]
C[Black 67:
Black 67 is the only reply. Not only does Go theory say to play in the widest area, but this move begins to threaten White's safety in the corner.]
;
W[qq]
C[White 68:
This exchange shores up the corner, freeing White to make another big move.]
;
B[qr]
;
W[cl]
C[White 70:
White's moyo begins to crystallize.]
;
B[bo]
C[Black 71:
Black takes an extremely direct approach, starting with the hane and atari.]
;
W[bp]
;
B[do]
;
W[cp]
;
B[ck]
C[Black 75:
Black connects up the left side.]
;
W[dk]
;
B[bl]
;
W[cm]
;
B[bm]
C[Black 79:
Although the result is clear, this way of connecting is extremely crude, and most professional players would be loath to use it.]
;
W[dn]
;
B[bn]
;
W[em]
;
B[km]
C[Black 83:
Finally, Black plays a reducing move in the centre.
Up to here, this game conveys the feeling that White is playing with masterful lightness, while Black is being dragged around the board. Gu Li and Zhou Ruiyang felt this to such an extent that they declared the game "totally one-sided," almost as if White were playing by itself. Yet in AlphaGo's own calm-minded assessment, White has a win rate of just 51.5%, a lead by only the slimmest of margins. When I informed the two masters, they sighed deeply. In professional matches, they told me, one often encounters situations in which one feels sure of a tremendous advantage - but after a few soft moves, the outlook has reversed completely.
To analyse the position in detail, Black's territory is not small, and though White has great potential in the centre, it is still unknown how this will be converted to real profit. On top of that, White is not completely safe in the corner, which increases the pressure on the centre. Sometimes the difference between feeling good about a game and feeling great is just a single detail. In such cases, players may lean too much on feelings to guide their judgment. This reminds me of a comment Lee Sedol made after the AlphaGo match. He declared, "I will no longer rely on feeling - I will focus on precise calculation." This could be what he learned from his games with AlphaGo, and the secret behind his subsequent 9-game win streak.]
;
W[om]
C[White 84:
White starts a fight with the knight's move.]
;
B[qg]
C[Black 85:
Black 85 is a very interesting reply. AlphaGo often attaches when the local shape looks like this, and it seems that it has already developed the follow-up into a new joseki.]
;
W[pk]
;
B[rf]
;
W[pb]
;
B[ob]
;
W[qa]
;
B[qk]
;
W[pj]
;
B[qj]
;
W[pi]
;
B[rh]
C[Black 95:
White continues to build up the centre, and Black continues with the strategy of "profit, then erase." Up to here, Black has secured considerable territory across the board.]
;
W[lo]
TR[km]
C[White 96:
This turn is necessary to isolate the marked Black stone.]
;
B[jj]
C[Black 97
throws some support to the lonely centre stone.]
;
W[if]
;
B[hb]
LB[hb:B][if:A]
C[Black 99:
The exchange of A for B helps prepare White's coming attack. Here, the two sides diverge slightly in their thinking about the position. See the variation for Black's prediction.]
;
W[lj]
C[White 100:
White chooses to start the attack slowly, hitting the vital point of Black's shape.]
;
B[il]
C[Black 101:
When I asked Gu Li about these moves, he and Zhou agreed that 'these feel like good moves that will be hard to counter, but the calculations are fundamentally unclear.?? I believe AlphaGo calculated these moves deeply, but sadly we cannot see all of its internal variations.]
;
W[im]
;
B[hi]
;
W[jl]
;
B[ih]
;
W[hl]
C[White 106:
Through White 106, a trade has developed, and the outlook has reversed: Black's win rate now stands at 56%. In other words, Black believes that the fight in the middle has been a success. However, this judgment is predicated on Black's ability to further harass White in the centre.]
;
B[mo]
;
W[mn]
;
B[nn]
;
W[no]
;
B[oo]
LB[nm:A]
C[Black 111:
Black now seals in the White stones on the right. Note that White cannot atari at A - see the variation.]
;
W[mp]
;
B[np]
;
W[mo]
;
B[pm]
;
W[rr]
;
B[nm]
C[Black 117:
Black has successfully cut off the corner, but this has no impact on White's life and death status. Zhou Ruiyang thought Black played poorly here.]
;
W[mm]
;
B[rs]
;
W[rp]
;
B[sr]
;
W[rq]
;
B[nl]
;
W[ml]
;
B[pn]
;
W[pq]
LB[jr:A][or:B]
C[White 126
is a very strange move, and incurs a definite loss of territory. Before this move, connecting at A may have influenced the status of Black's group, but once White provokes Black B, the connection becomes completely gote. AlphaGo may like to play the clearest variations, but this move must be called a mistake.]
;
B[or]
;
W[qo]
;
B[gh]
;
W[lh]
;
B[rm]
;
W[rn]
;
B[nj]
C[Black 133:
At this point, Black's win rate stands at 53%.
The game has entered a stage of extreme suspense. When Black jumps to 133, Black believes that White will be overwhelmed trying to balance the middle and the right side. More precisely, Black thinks both sides are in danger of dying, and that is the reason behind this splitting move.]
;
W[hg]
;
B[gg]
;
W[gf]
C[White 136:
White also believed that the situation was difficult, but when White hanes at 136, the win rate begins to shift. It seems as though both sides failed to foresee this move.
But this move appears to simply give away points. Why make the exchange? See the variation for a deeper analysis.]
;
B[ff]
;
W[kg]
;
B[ig]
LB[jf:A]
TR[ff][gf]
C[Black 139:
Incredibly, Black chooses not to take the three stones! But why? We researched the variations behind the push and cut at A extremely thoroughly, and it seems there is no way for Black to come out ahead even with the three extra stones, as the marked exchange gives White a great deal of aji to work with. See the variations for all the intricate details.
However, if Black could not come out ahead with the three stones, Black is certainly behind without them, and this choice proves to be the point of no return.]
;
W[jf]
C[White 140:
Not only can White extract the three stones...]
;
B[hf]
;
W[oh]
C[White 142:
...but after Black captures, the kosumi at 142 connects up the whole centre. This is a huge loss for Black! Both sides understand the situation now, and White?�s win rate shoots up to 65%.]
;
B[kk]
;
W[ol]
;
B[on]
;
W[kl]
;
B[ap]
;
W[aq]
;
B[ao]
;
W[bq]
;
B[gk]
;
W[gl]
;
B[so]
C[Black 153
is an absolute loss, probably a reflection of Black's desperation.]
;
W[sn]
;
B[fk]
;
W[fl]
;
B[dl]
;
W[dm]
;
B[ek]
;
W[el]
;
B[nh]
;
W[mj]
LB[nh:A]
C[Black 163:
Black A is the last real try for a comeback, but White refutes it with this tesuji. See the variation.]
;
B[lg]
;
W[mg]
;
B[lf]
;
W[kh]
;
B[ng]
;
W[mh]
C[White 168:
At this point, White's win rate exceeds 80%. Black has no chance of victory.]
;
B[oa]
;
W[re]
;
B[pa]
;
W[sf]
;
B[rg]
;
W[qb]
;
B[sb]
;
W[sa]
;
B[kf]
;
W[mf]
;
B[me]
;
W[ne]
;
B[md]
;
W[jg]
;
B[jr]
;
W[ir]
;
B[ks]
;
W[ni]
;
B[is]
;
W[gr]
;
B[gs]
;
W[fs]
;
B[lk]
;
W[mk]
;
B[ll]
;
W[ij]
;
B[ji]
;
W[jk]
;
B[kj]
;
W[lm]
;
B[db]
;
W[da]
;
B[ca]
;
W[ea]
LB[dd:A][id:B]
TR[ff][gf]
C[Black resigns at move 202.
The most interesting aspects of this game are White's extremely creative attachment in the corner at A, and the stubborn attachment on the top side at B. Although we cannot say for sure that these are good moves, they are without a doubt inspiring and eye-opening. With moves like these to learn from, we can look forward to AlphaGo?�s innovations bringing greater diversity to Go.
I would like to thank Gu Li and Zhou Ruiyang once more for their vigorous, precise, and tireless analysis. The variations in the centre were incredibly complex, and we spent many hours in particular investigating White's critical exchange of the marked stones. I feel deep respect for the dedication these two world champions showed to the pursuit of perfection on the Go board.
Finally, I must say that in a game as rich as this one, there are surely many details we have not yet explored. If you have the time and inclination, please play out these fascinating games yourself, and you will be sure to reap the benefit of many new discoveries.]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[7200]
OT[3x60 byo-yomi]
PW[AlphaGo]
PB[Lee Sedol]
BR[9p]
DT[2016-03-09]
C[Game 1 - Dawn
2016-3-9
White: AlphaGo
Black: Lee Sedol
Komi: 7.5
Result: W+R
Commentary by Fan Hui 2p
Expert Go analysis by Gu Li 9p and Zhou Ruiyang 9p
Translated by Lucas Baker, Thomas Hubert, and Thore Graepel
I sometimes suffer from insomnia during important tournaments, but I slept well before the first game of this match. Perhaps it was because I was not the one playing. At breakfast, everyone on the team displayed some tension, but as for me, I was already free of doubts.
I entered the playing room at 12:35, twenty-five minutes before the match was due to start. Some members of the staff were testing the cameras one last time. A make-up artist called me to put powder on my face, to prevent the light from reflecting. Everyone looked very busy, but the real cause of the activity was just to occupy us while we waited.
At 12:40, Lee Sedol entered the room with his wife and daughter, and sat down in front of the Go board. His daughter looked at him inquisitively, as if this was the first time she had accompanied her father to an official match. Lee smiled back at her.
After the staff explained a few details about the live broadcast, Lee exited the room. The minute he left, without knowing why, I felt a strange sense of pressure descending. It was not the pressure of anticipating the outcome, but rather of uncertainty - of not knowing what would transpire when the match began.
The playing room was actually quite sizeable. Apart from the Go board and the referees' table, there was a space for the AlphaGo team and VIPs to experience the game up close, complete with a screen for the commentary. The cameras did not face this section, so the audience would only see the players and the board.
At 12:47, two Korean referees entered the room. One was a professional 8 dan, whose main role would be to announce the beginning of the game and the final result. A different professional would take on this role each day. The other was a woman, reportedly an amateur player, who would be in charge of reading out the seconds during byo-yomi. Like me, she would be here for every game. Myself, I had only one role: to count the final score if the game reached an end and neither player had resigned. Due to the language barrier, the other referees and I could only communicate by nodding our heads, but I could still see in their eyes their absolute confidence in Lee Sedol's victory.
Because this was the first game of the match, everyone looked a bit tense. No event of this scale had ever taken place in the history of Go. Although Google had no prior experience in organizing Go competitions, they had conscientiously seen to every detail: the camera angles, the timings, the order in which the players should enter, the setup of the postgame press conference, and everything else. Even we Go players can learn from the professionalism, enthusiasm, and sincerity they displayed in organizing this match.
As the countdown on the screen hit zero, Lee Sedol and Aja Huang entered the room and took their seats. Following the nigiri, Lee Sedol chose Black. Interestingly, even if the difference is minute, it is clear from AlphaGo's data that it prefers White.]
RE[W+Resign]
GN[Sedol-AG-1]
;
B[cp]
C[After thinking for just thirty seconds, Lee placed the first black stone. The historic battle had begun!]
;
W[pd]
;
B[dc]
;
W[pp]
;
B[nq]
C[Black 5:
Lee Sedol quickly played the two 3-4 points, followed by the corner approach.]
;
W[qn]
;
B[ch]
C[Black 7:
Next, after a minute of reflection, Lee played this unexpected extension! A new fuseki, perhaps? Or was it a strategic ploy? If Lee thought that AlphaGo's knowledge was based on pattern recognition from existing games, then playing a move no professional had played before might confuse it. In reality, Lee had underestimated AlphaGo.
Faced with a strange move like this, a human player would no doubt take some time to think, as the first impression would be that the opponent is laying a trap. Imagine two kung fu masters locked in battle. If one makes an unusual move, the other would surely react with caution to avoid being caught out. However, there is no concept of "fear" in AlphaGo's dictionary. It knows only one strategy: constantly adapting to the present situation.]
;
W[ep]
;
B[hp]
C[Black 9:
Against White's approach, Lee quickly pincered, as if he had prepared this response.]
;
W[cq]
;
B[bq]
C[Black 11:
When White attached in the corner, Lee instantly blocked. The classic thinking is that this way of playing is very heavy for White, not only strengthening Black's corner, but turning the pincer into the perfect attacking move. Could it be that AlphaGo did not know the joseki?]
;
W[dq]
C[White 12:
When the players in the commentary room saw White 10 and 12, everybody thought Lee had exposed a problem in AlphaGo's opening. After all, our first impulse when passing judgment is to look within the scope of our existing knowledge. However, this can sometimes distract us from the essence of the problem. Were the black stones on the left side truly working well together?]
;
B[bo]
;
W[en]
C[White 14:
When White jumped, the feeling in the room suddenly changed. The black stones on the left were too low, and Black's formation was nothing to be proud of after all!
In fact, at this moment, AlphaGo already felt that the game had begun to favor White. According to its analysis, the problem lay in the tiger's mouth at A, and Black should have chosen the knight's move instead.]
;
B[dm]
C[Black 15:
Although many professionals already thought the result was unfavorable for Black, Lee still looked confident, and played the large knight's move almost instantly.]
;
W[em]
;
B[dl]
;
W[kp]
C[White 18:
When White played here, Lee hesitated. For the first time since the game began, his hand left the bowl of stones. He pressed his left index finger to his upper lip, and started tapping the right one against his leg. He seemed to realize things weren't going as well as expected. Of course, when circumstances take a sudden turn for the worse, the pressure can be difficult to handle.]
;
B[hn]
C[Black 19:
Lee responded steadily with the jump. In AlphaGo's judgment, this was too slow and White had already taken the lead, with a win rate of 54%. See the variation for an alternative.]
;
W[el]
;
B[dk]
;
W[hl]
C[White 22:
After pushing once on the left, AlphaGo capped Black on the lower side.]
;
B[il]
C[Black 23:
Lee quickly responded with the attachment. Subconsciously, he may have believed that AlphaGo would be unwilling to initiate a fight. At the same time, this move was very much his style, brimming with fighting spirit.
After the game, many commentators questioned this move, but this was AlphaGo's suggestion as well. In this case, it seems great minds thought alike.]
;
W[io]
C[White 24:
Seeing the peep, Lee sighed, as if realizing he was in for a hard game.]
;
B[ho]
;
W[hm]
C[White 26:
Black had to connect, but White continued with this severe follow-up! This move challenged Lee to make the game's first crucial choice: War, or peace? It is impossible to calculate every variation fully, so how to choose? For most players, the decision is a combination of personal style and a feeling for shape. At the same time, it is important to consider the opponent's intentions. For instance, did White pause before playing? Did White look pleased or dissatisfied with the previous position? Normally, reading the opponent is one of Lee's strong points. Thanks to his exquisite perception, he can pick whichever way of playing gives him the greatest advantage. This time, however, he hesitated. Since the start of the game, AlphaGo had played at a constant rhythm, with no sign of doubt or confusion. It felt as if every move were inevitable.]
;
B[im]
C[Black 27:
This was the first time in the match that Lee found himself at a crossroads, and he had no intention of backing down before a machine. So, after three minutes of thought, Lee resolutely blocked, and White cut. At this point, AlphaGo's win rate rose to 56%. In AlphaGo's judgment, Black should have followed the variation and chosen the turn instead.]
;
W[in]
;
B[fo]
;
W[eo]
C[White 30:
Black peeped, and White connected. At this point, Lee Sedol had 1 hour and 43 minutes remaining, AlphaGo 1 hour and 44 minutes.]
;
B[jq]
C[Black 31:
Black continued to probe White's intentions. Lee was asking AlphaGo, ?�Dare you block???]
;
W[kq]
C[White 32:
Without hesitation, AlphaGo did. The curtain rose on the first major battle, and the following moves were a one-way street. However, AlphaGo only grew more and more confident, its win rate reaching 58%. According to AlphaGo, Black should have followed the variation starting with the turn.]
;
B[jp]
C[Black 33:
At this point, Lee had 1 hour and 40 minutes, AlphaGo 1 hour and 43 minutes.]
;
W[ko]
;
B[jo]
;
W[jn]
;
B[kn]
;
W[jm]
C[White 38:
With this turn, AlphaGo's win rate continued rising to 59%.
White is not particularly eager to capture the two stones.]
;
B[hk]
C[Black 39:
At this point, Lee had 1 hour and 34 minutes, AlphaGo 1 hour and 39 minutes.]
;
W[gk]
C[White 40:
AlphaGo thought this exchange was unnecessary.]
;
B[ln]
;
W[kl]
LB[ik:A]
C[White 42:
Many commentators asked why White chose this move instead of A. Actually, by this point, AlphaGo's win rate had already reached 64%, so AlphaGo felt confident that capturing the two stones was not needed.
After this kosumi, the value of the two stones decreased markedly.]
;
B[gr]
C[Black 43:
At this point, Lee was pausing to think after each move, no longer laying down stones one after another as he had in the opening.]
;
W[mp]
C[White 44:
This move was mandatory to protect White's three stones.]
;
B[ik]
;
W[gj]
;
B[ii]
;
W[kr]
C[White 48:
White descended here to trigger a series of forcing exchanges. Many commentators thought this unnecessary, but from AlphaGo's perspective, the win rate continued climbing to reach 66%. Did AlphaGo think Lee had made another mistake?]
;
B[er]
;
W[gh]
;
B[hg]
C[Black 51:
White jumped out, and Black kept up the pressure on White's dragon, attacking fiercely with the knight's move.]
;
W[fq]
;
B[fr]
;
W[ir]
;
B[hq]
LB[fq:A][hq:D][fr:B][ir:C]
C[Black 55:
The simplifying exchanges of A for B and C for D might seem early to a human player, but were very much in AlphaGo's style. It is difficult to evaluate the pros and cons of making these exchanges now.]
;
W[gg]
;
B[hf]
C[Black 57:
Lee played quickly during this fight. When Black extended, Lee had 1 hour and 30 minutes, AlphaGo 1 hour and 29 minutes.]
;
W[ih]
C[White 58:
With this attachment, however, Lee became contemplative once more.
In normal matches, Lee often seeks opportunities to play startling moves that throw his opponents off their rhythm. This time, the roles had been reversed. It was AlphaGo that played these moves, rattling the opponent while maintaining equanimity itself.]
;
B[hh]
;
W[hi]
;
B[jj]
C[Black 61:
After 4 minutes of reflection, Lee chose the tiger's mouth, a bizarre shape in this position!
Compared to the rest of the match, Lee played quite quickly in this game. Subconsciously, it seemed, he believed that entering byo-yomi against AlphaGo would count as a failure in itself. Of course, playing so fast might also induce careless mistakes. AlphaGo thought this was one such move. See the variation.
At this point, Lee Sedol had 1 hour and 27 minutes left, AlphaGo 1 hour and 26 minutes.]
;
W[jh]
C[White 62:
When AlphaGo extended here, David Silver, technical lead of the AlphaGo team, came to the players' room and sat down to observe. I asked about the outlook, and he replied only with a smile. This confirmed my analysis - White was ahead!]
;
B[lj]
;
W[lk]
;
B[mj]
LB[lj:A][lk:B]
C[Black 65:
After the jump at A, it is painful to endure White's bump at B, but AlphaGo disapproved of Black's extension here. See the variation.]
;
W[mn]
C[White 66:
AlphaGo thought the nose attachment was the only reply.]
;
B[kk]
C[Black 67:
At this point, Lee had 1 hour and 24 minutes left, AlphaGo 1 hour and 22 minutes.]
;
W[jl]
;
B[kj]
C[Black 69:
This move was ugly but necessary, giving White sente to return to the left side.]
;
W[gf]
;
B[he]
;
W[ge]
;
B[hd]
C[Black 73:
Lee had played the preceding moves almost instantly, so after this move, Black had 1 hour and 23 minutes to White's 1 hour and 18 minutes. Lee glanced at the clock and shook his head.]
;
W[gd]
;
B[hc]
C[Black 75:
AlphaGo thought this move should have been the hane.]
;
W[gc]
C[White 76:
With this move, AlphaGo?�s win rate rose a little further to over 70%. Although this value theoretically represents the chance of victory, in practice it is an extremely conservative indicator of AlphaGo's true estimate. A value of 70% means that AlphaGo already believes the outcome of the game is nearly decided.]
;
B[kg]
C[Black 77:
When Lee played 77, he shook his head unhappily. Sure enough, Black's outlook was pessimistic. For the first time in the match, Lee left the room for a smoking break. (This also gave the referees the chance to use the restroom.) According to the competition rules, there would be no stopping the clock during the match, so when White played the next move, Lee's clock continued to run.]
;
W[ce]
C[White 78:
With one player absent, the tension lessened dramatically. It was as if the room were a bow, and the hunter had relaxed the bowstring. It was then that we realized how strongly affected we were by the pressure of the game.
At this point, both AlphaGo and Lee Sedol had 1 hour and 15 minutes remaining.]
;
B[nc]
C[
Black 79:
Apparently, AlphaGo thought that this approach was not best, preferring the variation instead.]
;
W[mm]
C[White 80:
When Lee resumed his seat, he quickly played the corner approach. Seeing AlphaGo's response at 80, I wrote down in my notebook: "Statement of victory!"
During the match, all the commentators thought this move was extremely soft, and answering in the upper right corner was far bigger. However, I believe I know AlphaGo as well as anybody, and in my view a move like this is really a demonstration of its matchless confidence. AlphaGo does not pursue the biggest victory, only the most certain one. Once it has a lead of two or three points, AlphaGo is completely capable of holding it until the end.
Looking at AlphaGo's 74% win rate confirmed my impressions.]
;
B[qf]
C[Black 81:
After White protected, Black had no choice but to seek compensation from the double approach.]
;
W[pf]
;
B[pg]
;
W[nd]
;
B[md]
;
W[mc]
C[White 86:
A beautiful move! Sacrificing a stone where the opponent is strong is a common strategy, but it is nonetheless refreshing to see White brazenly cutting through in the face of Black's wall.
At this moment, I looked over at the 8-dan professional who was that day's referee. He shook his hand under the table, indicating that he thought the game was already over for Lee.
One might have the impression from this commentary that Lee had not played well so far, or that he had not deployed his full power. Yet, I believe that's not the case. Rather, it is AlphaGo's strength that makes Go look easy.
At this point, AlphaGo and Lee Sedol had 1 hour and 8 minutes each on the clock.]
;
B[nb]
;
W[qe]
C[White 88:
The cut was breathtaking, but the follow-up was baffling. With this tiger's mouth, AlphaGo simply let Black swallow up the center! Was it already so confident in White's territorial lead? One way or the other, I thought White's move was far too slow. AlphaGo's judgment is not based on the size of its lead, but on the ease of controlling the outcome, so from that perspective there may be nothing wrong with this move. However, Lee seemed happy, as the game had certainly taken a turn for the better.]
;
B[of]
;
W[pe]
;
B[od]
;
W[qg]
;
B[ph]
;
W[hb]
;
B[ib]
;
W[br]
;
B[jr]
;
W[js]
C[White 98:
At this point, Lee Sedol had 1 hour and 4 minutes left, AlphaGo 59 minutes.]
;
B[gn]
;
W[gl]
;
B[fp]
;
W[cj]
C[White 102:
The last few moves had been a string of minor exchanges. To me, they felt like shots of anaesthetic in preparation for surgery. However, anaesthetic or no, Black must have felt the sharp pain of the incision at White 102.
Perhaps the answer to the riddle is right before our eyes, and the point of the tiger's mouth at 88 was to win back sente for the invasion on the left side?
A smile crossed Lee's face when he saw White's invasion, and he sank into deep calculation. This was his longest reflection of the game. At last, the deciding moment was at hand! The time had come to fight as if his life depended on it - but how does one fight an opponent that knows no fear?
At this point, Lee Sedol had 1 hour left, AlphaGo 56 minutes.]
;
B[dj]
C[Black 103:
After seven minutes of contemplation, Lee finally blocked from above.]
;
W[ci]
C[White 104:
AlphaGo answered here immediately. Although it normally plays at a constant pace, it seems even AlphaGo sometimes has a sense of "the only move."]
;
B[dh]
;
W[di]
;
B[ei]
;
W[bl]
C[White 108:
When white played the knight's move, Lee Sedol sighed again. The game held so many possibilities, but with no way to read the opponent's intentions, it became more difficult to accept or reject any one of them. Because AlphaGo lacked any sense of mental or emotional pressure, the psychological burden grew twice as heavy on Lee's shoulders!
At this point, Lee Sedol had 47 minutes left, AlphaGo 53 minutes.]
;
B[bh]
C[Black 109:
In the game, Lee Sedol chose to descend. AlphaGo, however, thought Black's only chance was to stake the game on the desperate fight.]
;
W[cn]
;
B[bk]
C[Black 111:
Attaching this way is clearly problematic, and Black should have attached on the other side as shown in the variation. In the game, White was able to capture the three stones in the lower left in sente, an enormous gain in both security and profit.
At this point, Lee Sedol had 44 minutes left, AlphaGo 51 minutes.]
;
W[bn]
;
B[dn]
;
W[co]
;
B[cl]
;
W[qq]
C[White 116:
When AlphaGo took the 3-3 point, its win rate rose to 82%. A value this high means AlphaGo already considers the game won, even if the absolute difference is small.]
;
B[rf]
;
W[rg]
;
B[rh]
;
W[re]
;
B[bm]
;
W[gb]
;
B[de]
;
W[jb]
;
B[ic]
;
W[df]
;
B[ee]
;
W[cd]
;
B[ef]
;
W[dd]
;
B[ed]
;
W[cc]
;
B[eb]
;
W[db]
;
B[ec]
;
W[dp]
;
B[cb]
;
W[bb]
;
B[da]
;
W[ba]
C[White 140:
Although many commentators found better ways for Black to handle the upper left corner, Black had no chance of victory either way. Against a human player, one can still try to provoke a blunder, but against the untiring AlphaGo such hope is futile.
When white lived in the corner, AlphaGo's win rate was 90%.
The tension faded from Lee's expression as he faced the inevitable.]
;
B[bf]
;
W[be]
;
B[cf]
;
W[ae]
;
B[qh]
;
W[sf]
;
B[ia]
;
W[eg]
;
B[eh]
;
W[qk]
;
B[ll]
;
W[lm]
;
B[pb]
;
W[pj]
;
B[rc]
;
W[qc]
;
B[qb]
;
W[rd]
;
B[pc]
;
W[rb]
;
B[qd]
;
W[sd]
;
B[oi]
;
W[nk]
C[White 164:
As the endgame continued, it became clear that although the score remained close, Black's defeat was unavoidable.]
;
B[rj]
;
W[rk]
;
B[dr]
;
W[cr]
;
B[an]
;
W[bp]
;
B[sk]
;
W[sl]
;
B[sj]
;
W[rl]
;
B[mk]
;
W[nj]
;
B[ni]
;
W[eq]
;
B[gq]
;
W[fg]
;
B[dg]
C[Black 181:
When Lee played 181, he muttered a few words. Sadly, since they were in Korean, I cannot tell you what they were, but they must have shown the depth of his dissatisfaction.]
;
W[qj]
;
B[sh]
;
W[sc]
;
B[ml]
;
W[nl]
C[White 186:
Lee picked up two black stones and lightly tapped them against the bowl. The sound echoed crisply through the room, and it felt as if time had frozen. I knew Lee was about to resign. Afterwards, many people said that I must have been happy in that moment. Now the world would know AlphaGo's true strength, and I would not be the only one it had defeated. The truth is, in the moment when he tapped those two stones, my mind was blank. Although I had known this day would come sooner or later, it had always felt so remote before, suddenly, it arrived. I felt as if I had been cast into outer space, adrift on waves of nothingness.
Lee Sedol resigned, and I returned to the present. The Korean referee's announcement of the result marked a historic shift. From that moment on, humanity was no longer alone in exploring the mysteries of Go. We had gained a new partner in our quest: AlphaGo!]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[7200]
OT[3x60 byo-yomi]
PW[Lee Sedol]
PB[AlphaGo]
WR[9p]
DT[2016-03-10]
C[Game 2 - Invention
2016-3-10
White: Lee Sedol
Black: AlphaGo
Komi: 7.5
Result: B+R
Commentary by Fan Hui 2p
Expert Go analysis by Gu Li 9p and Zhou Ruiyang 9p
Translated by Lucas Baker, Thomas Hubert, and Thore Graepel
AlphaGo's victory in the first game stunned the world. Many Go players, however, found the result very difficult to accept. Not only had Lee's play in the first game fallen short of his usual standards, but AlphaGo had not even needed to play any spectacular moves to win. Perhaps the first game was a fluke? Though they proclaimed it less stridently than before, the overwhelming majority of commentators were still betting on Lee to claim victory.
Reporters arrived in much greater numbers that morning, and with the increased attention from the media, the pressure on Lee rose. After all, the match had begun with everyone expecting Lee to win either 5-0 or 4-1.
I entered the playing room fifteen minutes before the game to find Demis Hassabis already present, looking much more relaxed than the day before.
Four minutes before the starting time, Lee came in with his daughter. Perhaps he felt that she would bring him luck? As a father myself, I know that feeling well.
By convention, the media is allowed a few minutes to take pictures at the start of a major game. The room was much fuller this time, another reflection of the increased focus on the match.
Today, AlphaGo would take Black, and everyone was eager to see what opening it would choose. Whatever it played would represent what AlphaGo believed to be best for Black. Perhaps that opening would become the new fashion in the Go world?]
RE[B+Resign]
GN[Sedol-AG-2]
;
B[pd]
C[Black 1:
Unlike the first game, AlphaGo played its first move after less than thirty seconds. Rarely does it decide so quickly.]
;
W[dp]
C[White 2:
Lee played White 2 in the lower left corner, on the diagonally opposing star point, indicating a strategic departure from the first game. Since the fighting had turned against him before, perhaps he wished to play a quieter and more solid opening this time. In that case, White 2 may have aimed to prevent AlphaGo from taking that point for Black, since this kind of cross-opening can easily turn violent.]
;
B[cd]
C[Black 3:
When Black played the 3-4 point, I felt that AlphaGo would opt for the Chinese opening, one of its favorites.]
;
W[qp]
C[White 4:
Compared to the first game, Lee played much more slowly, pondering over each move. Many players had criticized him for playing too quickly in the first game, and the shift may have been partly in response to that. Sometimes changing pace is also a useful meta-tactic to disrupt the opponent's mindset, but of course this would have no effect on AlphaGo.]
;
B[op]
C[Black 5:
White took the remaining 3-4 point, and AlphaGo approached the lower right corner with 5.]
;
W[oq]
C[White 6:
I had heard that Lee was very fond of coffee, and always asked the staff to bring more as soon as his cup was empty. During this game, and the rest of the match, he fully lived up to his reputation.]
;
B[nq]
;
W[pq]
;
B[cn]
;
W[fq]
;
B[mp]
;
W[qn]
C[White 12:
Through here, everything proceeded normally.]
;
B[ic]
LB[qn:A]
C[Black 13:
Black 13 came as an enormous surprise. Even here, Black can tenuki? It seems anything is possible for AlphaGo!
Just as interesting, AlphaGo did not even think White should have finished the joseki at A! See the variation.
Gu Li and Zhou Ruiyang were just as shocked as I was by this revelation, and we began to wonder: could there be a problem with even such an established, fundamental joseki as this?
Here, AlphaGo?�s win rate was 49.7%, so it thought the game more or less in equilibrium.]
;
W[dj]
C[White 14:
After seeing Black 13, Lee sank into deep contemplation. Compared to his confident appearance in game 1, he now seemed much more prudent. He even declined the opportunity to attack the three stones in the lower right, opting instead to bide his time and split the left side. However, AlphaGo thinks Black should have played at the bottom as shown in the variation.]
;
B[po]
C[Black 15:
The peep at Black 15 stunned the Go world for the second time in two moves! Any Go teacher would say that such a crude move cannot possibly be good, and would criticize any student who played it. Then again, AlphaGo never worried about other people's opinions.
As we walk the path of improvement, we must study and experience all aspects of the game: joseki, fuseki, shape, and direction, just to name a few. After we absorb this knowledge, we learn over time to apply it flexibly. But to reach the level of grandmasters, even this is not enough! As we gain experience, our knowledge fetters our creativity. To truly throw off these shackles and liberate ourselves from what we have learned, we must discard labels of "right" and "wrong." In their place, we must consider the essence of Go: the role of each stone, and the relationships between them. Only in this way can we reach the level where invention prevails over tradition. AlphaGo began from the same fundamentals as humans, but the rigid attachment to knowledge is simply not in its nature. Thus, it is only natural that AlphaGo possesses a talent for creativity.
In the history of Go, I can think of two periods when a decisive shift occurred in Go theory. The first was the age of Go Seigen. The most famous example of his innovation is perhaps the game in which he opened on the star point, the tengen, and the 3-3 point, in a time when no other professional player would think of starting anywhere besides the 3-4 point. But beyond that, he showed us that Go, even at the highest level, can be played in any style one likes. Out of that age of 3-4 points, a hundred other styles blossomed and grew, Black 15:
including the Chinese opening we know
today. However, many joseki and fuseki still remained purely theoretical.
The second such period was the age of Lee Changho and the Korean style. This movement was by no means the work of one person, but it was Lee Changho's rise that inspired Korean players to doubt established theory, and experiment with ever-bolder departures from the prevailing style. At first, many of these moves were called "crude" or simply "bad," but the results spoke for themselves. As research progressed, many older joseki were disproved and discarded. We realized that many of these joseki had been established not by study, but by old grandmasters who had claimed that one way or another was the right one. The more we explored the new variations, the more fatal weaknesses were exposed in the old ones. These discoveries liberated us, stimulating the development of more and more new styles and variations. In the final analysis, this period marked a great advancement in our understanding of Go, simply because we had the courage to try new things.
However, the emergence of the Internet has stimulated ever-faster development and progress. As the strength gap between players shrinks, so too does the apparent window for innovation. But Go is a rich, abundant world - can we truly say that we have reached its limits? Perhaps AlphaGo can usher in a third wave of exploration and invention. When we are brave enough to defy theory, only then can we Go players call ourselves "free." Even from the perspective of amateurs and fans, wouldn't this spirit make Go that much more fascinating to witness?
In fact, it is gratifying to note that since AlphaGo played 15, more and more professionals no longer find it so crude. Quite often, professionals find themselves wanting to exchange this move after the opportunity has disappeared, so many players have started trying to work it in earlier. It seems we have already gained some courage!]
;
W[qo]
;
B[cp]
;
W[cq]
;
B[bq]
C[Black 19:
In the lower left, Black played more normally, choosing the standard moves to settle the corner.]
;
W[co]
;
B[bp]
C[Black 21:
When Black connected with 21, Lee nodded. He seemed to be reaffirming his decision to play on the left side.]
;
W[bo]
C[White 22:
White chose to block, but AlphaGo prefers the connection shown in the variation.]
;
B[do]
;
W[bn]
;
B[dq]
;
W[ep]
;
B[dr]
;
W[cm]
C[White 28:
Up to here, the moves in the lower left form a classic joseki.]
;
B[jp]
LB[iq:A]
TR[dr]
C[Black 29:
Taking the initiative on the bottom, AlphaGo once again smashed convention with the beautiful two-space jump at 29! According to Nie Weiping, when he saw this move, he exclaimed, "Hats off to AlphaGo!" The rationale behind this move is not difficult to understand: since Black already has the marked stone in the corner, the lower side has lost much of its value, and the normal extension at A would be less efficient in view of White's thickness on the left. Furthermore, although Black 29 takes less territory than A, it gives Black more potential to expand in the center while restricting White's influence there. The real beauty of this move lies not just in the move itself, but in the way it demonstrates AlphaGo's ability to adapt joseki according to context. It is this sense of flexibility that deserves recognition.
At this point, Lee Sedol had 1 hour and 40 minutes left, AlphaGo 1 hour and 43 minutes.]
;
W[cg]
;
B[ed]
LB[ed:B][cg:A]
C[Black 31:
The exchange of A for B looks normal, but AlphaGo's win rate clearly increased, so it must think this benefits Black! See the variation.]
;
W[qf]
C[White 32:
After Black enclosed the corner, Lee Sedol briefly hesitated over the direction of play in the upper right. He eventually chose to approach from below.]
;
B[qe]
C[Black 33:
The kick at 33 was a good choice for Black. White was already solid on the right, so the loss from strengthening White further was minimal, and 33 prevented White from invading the corner. See the variation.]
;
W[pf]
;
B[nd]
;
W[pi]
C[White 36:
Up to here, both sides played normally.]
;
B[oj]
LB[qn:A]
C[Black 37:
Black 37 is one of the two moves from this match sure to go down in Go history. This move proved so stunning that, when it appeared on the screen, many players thought the stone had been put down in the wrong place.
Because my main duty was to count the score at the end of the game, I had nothing in particular to do in the meantime. However, for such a historic event, it would be a terrible shame if there were no one to record the scene in the playing room, so I took this role upon myself. Though I had no prior experience, my aim was simple: to capture the atmosphere, the players' expressions, even their gestures, and in this way to communicate the experience of the match.
On seeing Black 37, I wrote down the following: "Here?! This goes beyond my understanding. Globally, there's nothing wrong with it, it's going in the right direction...and AlphaGo always pays special attention to coordinating the stones. It seems anything is possible in Go! Everyone will be talking about this move! A human
would never dare play it, it's too difficult to estimate. But AlphaGo can. Perhaps this move is a sign of its confidence."
This move made a deep impression on me during the game. I experienced first confusion, then shock, and finally delight. It reminded me of an old Chinese saying: "A beginner plays the corners, an average player the sides; but a master controls the center." These days, due to the convergence of strengths and the pressure of competition, something close to the opposite is true, with most players focusing on the corners and sides. In contrast, AlphaGo's talent for central control is second to none. Perhaps, through AlphaGo, we too can become the "masters" of which the proverb speaks.
Returning to the game, we may say that Black 37 casts an invisible net across the board. Together with the lower side, Black's shoulder hit creates potential all across the center. Although it helps White make territory on the right, the presence of White A means that a Black invasion there would not have been valuable anyway. Of course, Black should be reluctant to give away fourth-line territory too easily, but one must give to get.
After the match, when I examined the data back at DeepMind, I saw that AlphaGo had not even been thinking about 37 only a few moves before. It had been expecting diagram 6, and its data indicated that a human player would hardly consider the shoulder hit a possibility. It was only when White played 36 that AlphaGo discovered 37, and boldly decided that this move would work even better.
The pace of the game was much slower than the previous day, so Lee had already gone out to smoke before 37. The minute he caught sight of AlphaGo's reply, he stared blankly at the board. Then he smiled, sat down, and started thinking. The longer he thought, the more serious his expression became, while the clock ran down little by little.]
;
W[oi]
C[White 38:
After twelve minutes of thought, Lee finally pushed up at 38. Perhaps he was feeling the pressure, as the direction of this move is clearly problematic: White is helping Black build up the bottom. See the variation.]
;
B[nj]
;
W[mh]
C[White 40:
Even now, AlphaGo thought the knight's move at 40 was inappropriate, and recommended pushing as shown in the variation.
It seems that Black 37 not only helped AlphaGo on the board, but also threw Lee off balance psychologically.
At this point, AlphaGo's win rate reached 55%.]
;
B[gp]
C[Black 41:
Black 41 further restricted White's potential on the left, while enlarging Black's in the center.]
;
W[gq]
C[White 42:
At this point, Lee had 1 hour and 16 minutes, AlphaGo 1 hour and 34 minutes.]
;
B[dn]
C[Black 43:
This connection was a definite overplay for AlphaGo.]
;
W[dm]
C[White 44:
After White pushed here, even AlphaGo's own win rate suffered a rare decline. Lee Sedol stared at the board with a scowl, evidently confused by AlphaGo's strategic slip. Drawing on the lessons of the first game, he remained cautious, taking time to verify each move.]
;
B[fo]
;
W[hp]
;
B[ho]
;
W[eo]
;
B[en]
C[Black 49:
When Black blocked at 49, AlphaGo's win rate fell to 50%. This may have been Lee's only chance to wrest control of the game. See the variation for one way he might have done so.]
;
W[fn]
C[White 50:
Regrettably, Lee missed the crucial opportunity, and chose this atari instead.]
;
B[em]
C[Black 51:
When Black extended, AlphaGo's win rate shot up to 59%! ]
;
W[el]
;
B[fm]
;
W[gn]
;
B[fl]
;
W[go]
;
B[ek]
;
W[dk]
;
B[dl]
;
W[cl]
;
B[eh]
;
W[di]
LB[eh:A][di:B]
C[White 62:
Lee still looked unconcerned. Through 62, Black had gained nothing locally, while White had captured two stones at the bottom, and furthermore the center group was not yet safe. On the face of it, White should at least not have been worse off. Globally speaking, however, Black had reduced White's left side in sente, made the profitable exchange of A for B, and come away with sente to take the last big point at the top. Overall, it would be hard not to call this a strategic success for Black.
Many of the professionals in the commentary room did not understand Black's way of playing, but after game 1, they were hesitant to pass judgment.]
;
B[pj]
;
W[qi]
C[White 64:
Lee spent a few minutes in thought, and patiently descended with 64. AlphaGo felt this move was too soft, as did Gu Li and Zhou Ruiyang, and would have chosen to come back to the center as shown in the variation.]
;
B[rf]
;
W[rg]
C[White 66:
At this point, Lee Sedol had 59 minutes, AlphaGo 1 hour and 21 minutes.]
;
B[kd]
C[Black 67:
When Black defended the upper side, AlphaGo pulled clearly into the lead.]
;
W[hn]
C[White 68:
Lee followed with several more slow moves, perhaps still misled by White's apparent gains on the left. This was one such move - see the variation.]
;
B[om]
;
W[re]
C[White 70:
Taking a stone on the second line was even slower. This choice left a deep impression on Gu Li and Zhou Ruiyang, who felt that Lee must have thought White ahead to play like this. The variation shows a better option for White.]
;
B[rd]
;
W[sf]
;
B[fi]
C[Black 73:
For AlphaGo, the outlook was continually improving. When Black played 73, its win rate reached 64%.
For a human player, Go requires the capacity to feel. The better the player, the more effectively their feelings steer them towards good moves and away from bad ones. However, if we rely too much on feeling, it begins to obscure rational calculation and judgment. After the AlphaGo match, Lee Sedol won nine consecutive games. When asked for his secret, he replied, "Do not rely on instinct. Calculate with the utmost precision." That may be the lesson he drew from this game.]
;
W[gk]
LB[fi:A]
C[White 74:
After Black protected at A, White's attack at 74 became much less severe.]
;
B[hm]
C[Black 75:
On seeing 75, Lee finally seemed to wake up.]
;
W[in]
C[White 76:
He sighed, and made the necessary extension.]
;
B[hl]
C[Black 77:
By now Black's advantage was obvious.]
;
W[ko]
;
B[kp]
;
W[gc]
C[White 80:
Seeking a comeback, Lee launched the do-or-die invasion at White 80! However, AlphaGo thinks White should remain patient, as shown in the variation.
At this point, Lee Sedol had 50 minutes, AlphaGo 1 hour and 12 minutes.]
;
B[df]
C[Black 81:
Just as everyone was debating how Black should answer locally, AlphaGo played the shoulder hit at 81, reinforcing the center while subtly pressuring White's invading stone. It felt like a breath of fresh air - truly, an amazing example of whole-board thinking!
Not only did this move connect up the center, it completed a huge wall. Although Black had done nothing to attack White directly, the invisible pressure was suffocating! The saying normally goes "attack is the best form of defence," but in this case, defence was the best form of attack.]
;
W[id]
C[White 82:
Facing 81, Lee sobered up completely. He sighed, his face reddening.
Because Black's center potential is so great, White can no longer be satisfied with living locally. The attachment aimed to escape.]
;
B[jc]
C[Black 83:
Drawing back was a calm response.
At this point, Lee had 41 minutes remaining, AlphaGo 1 hour and 10 minutes.]
;
W[ge]
;
B[dg]
C[Black 85:
Black 85 was solid defense.]
;
W[cf]
;
B[ch]
;
W[bh]
;
B[dh]
;
W[bi]
C[White 90:
Up to here, Black had not only compressed White's territory on the left, but also reinforced the connection between the center and the corner.]
;
B[hd]
C[Black 91:
Following this hane, it became difficult for White to save the initial invasion stone.]
;
W[he]
;
B[gd]
;
W[fd]
;
B[hc]
;
W[fe]
;
B[ec]
;
W[gh]
C[White 98:
White had no choice but to escape, since connecting would have meant the death of the whole group. See the variation.]
;
B[fc]
C[Black 99:
At this point, AlphaGo's win rate reached 73%.]
;
W[gi]
C[White 100:
After the cut, Lee had no real chance of winning.
We have all had the experience of playing a game where we are behind. Against a person, the best hope is to stay calm and patiently await the opponent?�s mistakes. After all, to err is human. Against AlphaGo, of course, this strategy is futile. It is calmer than you, more patient than you, and, by the time you realize you are in trouble, the game is already decided.
Lee had just experienced this painful process, the desperation of knowing one's fate while being helpless to change it. But, as a professional Go player, he could only steel himself and press on.]
;
B[ii]
;
W[hk]
;
B[ik]
;
W[il]
;
B[im]
;
W[ij]
;
B[jl]
;
W[jj]
;
B[if]
C[Black 109:
Here, AlphaGo's win rate reached 78%.]
;
W[km]
;
B[kl]
;
W[lj]
;
B[lk]
;
W[lo]
;
B[li]
C[Black 115:
At this point, Lee had 10 minutes, AlphaGo 48 minutes.]
;
W[kj]
;
B[ci]
;
W[cj]
;
B[mj]
;
W[nr]
;
B[mr]
;
W[lq]
;
B[lp]
;
W[mq]
;
B[np]
;
W[lr]
;
B[lm]
;
W[kh]
;
B[hg]
;
W[qc]
;
B[qd]
;
W[rc]
;
B[pc]
;
W[sd]
;
B[gg]
;
W[ce]
;
B[bd]
;
W[qb]
;
B[hi]
;
W[jg]
;
B[hj]
C[Black 141:
In the end, AlphaGo never went for the kill in the center, instead nibbling away at White's stones bit by bit. Although it looked as if White had destroyed much of Black's territory at the top and bottom, the actual score tilted ever more strongly in Black's favor.]
;
W[ob]
;
B[pb]
;
W[pa]
;
B[nb]
;
W[de]
;
B[ee]
;
W[gj]
;
B[hh]
;
W[ej]
;
B[nf]
;
W[mf]
;
B[me]
;
W[rk]
;
B[fh]
;
W[el]
;
B[nh]
;
W[ng]
;
B[lg]
C[Black 159:
Even though the result was already decided, the beautiful tesuji at Black 159 is worth remembering! With this move, White's center collapsed.]
;
W[lh]
;
B[mg]
;
W[og]
;
B[kg]
;
W[ni]
;
B[jh]
;
W[na]
;
B[ki]
;
W[mi]
;
B[ji]
;
W[nc]
;
B[mb]
;
W[od]
;
B[mc]
;
W[oc]
;
B[kr]
;
W[ms]
;
B[io]
;
W[ip]
;
B[jo]
;
W[jn]
;
B[ir]
;
W[hr]
;
B[ql]
;
W[rl]
;
B[qm]
;
W[rm]
;
B[ao]
;
W[bm]
;
B[ln]
;
W[kn]
;
B[mo]
;
W[be]
;
B[ae]
;
W[af]
;
B[ad]
;
W[ma]
;
B[la]
;
W[oa]
;
B[dd]
;
W[bg]
;
B[lb]
;
W[pn]
;
B[on]
;
W[er]
;
B[cr]
;
W[fp]
;
B[iq]
;
W[hq]
;
B[qj]
;
W[rj]
;
B[ks]
C[Black 211:
As the game neared its end, Lee continually recounted the score, looking for some sort of miracle. But he searched in vain: as the world had already realized, AlphaGo had won.
After playing on doggedly in byo-yomi, Lee Sedol resigned at move 211. In that moment, I heard the reluctance in his voice, and saw the regret etched upon his face.
During the press conference, Lee's voice was choked with emotion, and he repeated many times how impressed he was with AlphaGo's near-perfect play. The way he spoke, it seemed as if he bore the weight of the entire Go world on his shoulders.
Although AlphaGo's play in the first two games had verged on perfection, the match was not yet over. Behind Lee's trembling voice there stood a firm conviction: "I may have lost the second game, but there is still a third game left to play."
At the end of this game, the entire Go world was left astonished and bewildered. Perhaps the time had come to rethink our understanding of AlphaGo, and of ourselves.
Sedol resigned.]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[7200]
OT[3x60 byo-yomi]
PW[AlphaGo]
PB[Lee Sedol]
BR[9p]
DT[2016-03-12]
C[Game 3 - Revolution
2016-3-12
White: AlphaGo
Black: Lee Sedol
Komi: 7.5
Result: W+R
Commentary by Fan Hui 2p
Expert Go analysis by Gu Li 9p and Zhou Ruiyang 9p
Translated by Lucas Baker, Thomas Hubert, and Thore Graepel
Although the match had begun just two days ago, it had already delivered the greatest shock in Go history. According to the rules, every two games would be followed by one day of rest, so it was vital for Lee to use this day to regain his composure. The next game might well decide the match. The rest day also gave the AlphaGo team a temporary respite from the pressure. For those of us visiting Korea for the first time, it also presented an opportunity to explore the city of Seoul.
At 10AM on March 11th, the AlphaGo team set off to visit the city center. It was then I realized that, apart from eating, I hadn't left the hotel once since I had arrived. Although I tried my hardest to relax and enjoy the surroundings, before I knew it, my thoughts always returned to the match. The past two days had been simply too astonishing.
One fragment stands out from the day of sightseeing. As we passed by a small shop selling dragon's beard candies, a popular sweet in China and Korea, we peered inside to see how they were created. The friendly shopkeeper came out and told us about their history, as well as how to make them. As we chatted, he asked where we were from, and one member of the team replied "England." He hesitated a bit and asked, "AlphaGo?" The moment we confirmed his suspicions, he grew excited and began repeating, "Incredible! AlphaGo is truly incredible." Even after only two days, AlphaGo had become a household name throughout Korea.
Despite the rest day, it seems our minds never left the match, with the historic third game imminently approaching.
When we returned that evening, we looked at the online commentaries, and found that almost no one still believed that Lee would win. Nonetheless, there was no lack of moral support and strategic advice. The more I read, the more I felt that Lee was bearing the weight of the world on his shoulders. Ever since I was young, my teacher had taught me that Go demanded inner peace. Without a calm heart, one could not play at the highest level. But in an atmosphere like this, how could Lee possibly achieve such equilibrium?
At 1pm on March 12th, the third game began. Lee took Black.]
RE[W+Resign]
GN[Sedol-AG-3]
;
B[dp]
;
W[dd]
;
B[qp]
C[Black 3:
As Lee put down move 3, I was surprised to see his hand shaking slightly. I had heard that Lee sometimes suffered from insomnia before important games. Perhaps yesterday was one such sleepless night.]
;
W[pd]
;
B[qf]
;
W[nc]
;
B[kp]
C[Black 7:
Black 7 set up the high Chinese fuseki, a fighting stance that seeks to take the initiative. It seemed that Lee was done with waiting patiently.]
;
W[cn]
;
B[fq]
C[Black 9:
Facing the corner approach, Lee hesitated for three minutes, eventually choosing the knight's move at 9. He must have worried about White invading the corner directly at the 3-3 point if he had jumped on the fourth line. Nonetheless, AlphaGo thought this was the best plan for Black. See the variation.]
;
W[cg]
C[White 10:
White protected the upper left corner with the large knight's move at 10, a relatively common strategy. AlphaGo?�s win rate at this point was 51.5%.]
;
B[pj]
;
W[op]
C[White 12:
Black's extension was necessary, and White 12 also seems like the only move.]
;
B[pq]
C[Black 13:
Black's kosumi with 13 is an uncommon way of playing - Lee must have prepared it to work with his high Chinese opening. Later, I found that AlphaGo would also have played this move.]
;
W[om]
C[White 14:
When White played the two-space jump at 14, it felt as if AlphaGo were looking down from the heavens, comprehending the entire board at a glance.]
;
B[on]
C[Black 15:
I do not know whether anyone had played the jump at 14 before, but it seemed to infuriate Lee, who slapped down the astonishing attachment at Black 15!
Playing with AlphaGo can feel distressingly like euthanasia: by the time we feel what is going on, we are already dead. Perhaps this is why Lee seized the first opportunity to attack, attempting to rip open AlphaGo's weak points before its global advantage grew overwhelming. Nonetheless, from the perspective of Go fundamentals, Black 15 is simply too extreme.
Many people have asked me about Lee Sedol's style. Most would call him a fighter, but I feel this is too broad a label. Gu Li is also a fighter, but he is the tough, courageous, and direct kind, whereas Lee resembles a wolf who ambushes his prey. He conceals himself, awaiting the best opportunity. His slow preparation lulls you into unawareness. Patience and keen senses are his weapons. Finally, the moment he detects his chance, he strikes, swiftly and fatally.
But in this game, Lee betrayed his style and showed his fangs too early. Perhaps he succumbed to the pressure, to the desire for a clear victory to prove himself.
AlphaGo also thought Black 15 was overplay. See the variation.]
;
W[po]
C[White 16:
After White hit the elephant's eye with 16, Black's shape was hard to bear. The next move would be mandatory for a human player, but AlphaGo believed there was still time to change course. See the variation.]
;
B[pn]
;
W[qo]
C[White 18:
Playing against AlphaGo can induce a certain sense of despair. It is like a black hole, sucking you in little by little. No matter how you try to break away, you will slowly discover that your efforts count for nothing.
After White 18, it was Black's shape that was torn apart. Lee's face began to redden.]
;
B[oq]
;
W[np]
;
B[nn]
C[Black 21:
Many commentaries disagreed with Black 21, even calling it the losing move, but AlphaGo saw no problem with this way of playing.
At this point, Lee had 1 hour and 41 minutes left, AlphaGo 1 hour and 48 minutes.]
;
W[nq]
C[White 22:
Lee sighed when he saw this move, although he must have known it was coming.]
;
B[rp]
;
W[lo]
;
B[ko]
;
W[lp]
C[White 26:
Lee hesitated over the choice presented by this move: enclose White and let the group live, or give no quarter and continue pursuing the attack?]
;
B[kq]
;
W[lm]
C[White 28:
Black continued attacking, but when White came out at 28, Lee sighed even deeper. Did he regret the initial fight, or the decision to continue? Either way, in a game against AlphaGo, the more one plays, the more doubts arise, and these doubts are what will ultimately prove fatal.
In any case, AlphaGo thought Black should have extended as shown in the variation.
When White jumped here, AlphaGo?�s win rate stood at 59%.]
;
B[kn]
;
W[mm]
C[White 30:
When White played 30, Lee Sedol rested his hand on the bowl and started tapping his finger. This decision would take courage.]
;
B[jl]
C[Black 31:
Courage he may have had, but Black 31 was a definite overplay! Lee must have known this, but since he could not best AlphaGo in a contest of positional judgment, perhaps he hoped to break through its defenses in local fighting.
Of course, AlphaGo did not agree with this move. See the variation.]
;
W[qk]
C[White 32:
After the baptism by fire of the last two games, Go players were prepared to appreciate AlphaGo's ability. When move 32 appeared onscreen, everyone agreed it was a beautiful, perfect move! Although White's local shape was strange, the situation already looked out of Black's control. See the variation for some alternative lines.
Lee glanced at the clock. Then he began to rock back and forth, nervously, like a boat drifting in windy seas. At this point, AlphaGo's win rate had reached 62%.]
;
B[pm]
C[Black 33:
In the actual game, Lee had no choice but to turn at 33, after which he left the room for a smoke. He needed to regain calm, but it was a bit too late.
In the sequence that followed, AlphaGo's fighting strategy was watertight. If Go were boxing, then Lee came in with fists swinging, but he may as well have been punching cotton. The way his strength rebounded back on him must have been difficult to bear.]
;
W[pk]
;
B[ok]
;
W[rm]
;
B[rl]
;
W[ro]
;
B[rk]
;
W[qj]
;
B[pl]
;
W[pi]
;
B[oj]
;
W[qi]
;
B[ql]
;
W[mk]
;
B[nl]
;
W[pg]
C[White 48:
As Lee pressed beyond the point of no return, AlphaGo's win rate continued to rise. By the time White jumped at 48, it had already topped 72%.
At this point, Lee had 1 hour left, AlphaGo 1 hour and 34 minutes.]
;
B[pf]
;
W[nm]
;
B[ol]
;
W[ng]
;
B[oi]
;
W[oh]
C[White 54:
During this game, Lee was clearly in a different state of mind than he had been before. He murmured and sighed incessantly, shook his head back and forth, even occasionally ventured a helpless smile.
The battle grew more difficult for Black, and when White played 54, the win rate stood at 74%. To reach such a high number so quickly meant that the game was essentially decided, but Lee continued to persevere.]
;
B[mi]
;
W[kk]
;
B[qg]
;
W[ne]
;
B[of]
;
W[nh]
;
B[ri]
LB[ph:B][qh:A]
C[Black 61:
Black 61 was an interesting move. Normally, Black would first exchange A for B to secure an unconditional connection with 61. In the game, there remained a dangerous possibility of ko.
Since the beginning of the match, there had been a rumor that AlphaGo could not handle ko correctly. Especially since it had avoided a large ko near the end of game 2, it seemed that AlphaGo sought to prevent ko. Perhaps Lee wanted to verify the rumor.]
;
W[rj]
;
B[sj]
;
W[lg]
C[White 64:
When White jumped with 64, AlphaGo's win rate reached 77%.
See the variation for an idea of what might happen when White triggers the ko.]
;
B[lj]
;
W[lk]
;
B[mj]
;
W[jk]
;
B[kh]
C[Black 69:
One look at Lee's face was enough to see how the game was going. Perhaps the last hope for Black was that White's dragon at the bottom remained a little insecure.
Here, Lee had 39 minutes, AlphaGo 1 hour and 21 minutes.]
;
W[jf]
;
B[qd]
;
W[qc]
;
B[pc]
;
W[pb]
;
B[rc]
;
W[oc]
;
B[dg]
C[Black 77:
With the attachment at 77, Lee admitted that he could not win by normal means. Instead, he resorted to his famous "zombie" style! As the name implies, this style is one in which a player is already "dead," or at least doomed, but thrashes about frighteningly in an attempt to catch the opponent off guard. This strategy has won back many a lost game, but against AlphaGo, it seemed more tragic than anything else.]
;
W[dh]
;
B[cf]
;
W[eg]
;
B[df]
;
W[ch]
;
B[eh]
;
W[ef]
C[White 84:
When White pressed at 84, AlphaGo's win rate was 84%.]
;
B[bf]
;
W[ee]
;
B[ei]
;
W[ck]
;
B[hh]
;
W[lq]
;
B[cc]
;
W[cd]
;
B[bd]
;
W[dc]
;
B[bc]
;
W[cb]
;
B[bb]
;
W[hf]
C[White 98:
The battle in the upper left drew to a close with 98. Black had lived in the corner, but White had gained safety for the group on the lower side, and secured an enormous swath of sixth-line territory on the top! Black's position was already hopeless, as confirmed by AlphaGo's win rate of 87%.]
;
B[co]
;
W[bo]
;
B[bp]
;
W[do]
;
B[cp]
;
W[rr]
;
B[pp]
;
W[oo]
;
B[qr]
;
W[rq]
;
B[sp]
;
W[or]
;
B[rs]
;
W[mr]
C[White 112:
Sitting beside the board, I was expecting Lee to resign any minute, a feeling that only intensified when White played 112. White was unconditionally alive, Black was short on points, and there seemed to be nowhere left to play.]
;
B[ci]
;
W[di]
;
B[he]
;
W[gf]
;
B[dj]
;
W[cj]
;
B[bh]
;
W[bi]
;
B[bg]
;
W[ci]
;
B[qb]
;
W[dn]
;
B[kc]
C[Black 125:
In reality, Lee was far more tenacious than I had thought. When he invaded with Black 125, I gasped in surprise! Could he possibly live there!?]
;
W[ic]
;
B[hc]
;
W[jc]
;
B[je]
;
W[if]
;
B[lb]
;
W[kd]
;
B[ld]
;
W[ke]
;
B[jb]
;
W[ib]
;
B[le]
;
W[mc]
C[White 138:
For a human go professional, it would be unbearable to let Black live, but AlphaGo has no such restrictions. It only picks the move with the highest win rate. So, when White played 138, Black earned an unexpected chance to live in ko. See the variation.
Lee did not choose this ko, however, because Black has no way to win it.]
;
B[lc]
;
W[kf]
;
B[nf]
;
W[mf]
;
B[me]
;
W[mb]
;
B[ma]
;
W[na]
;
B[ka]
C[Black 147:
Turning down one ko, he tenaciously cut off the White group on the right side to set up an even bigger one. With this move, Lee secured an approach-move ko for life.]
;
W[kr]
C[White 148:
However, White did not even bother to answer! Instead, White took the big point at 148. By this point, the win rate was already over 98%, a declaration of total victory.]
;
B[ia]
;
W[hb]
;
B[pc]
;
W[lf]
;
B[pa]
;
W[qc]
;
B[nd]
;
W[oe]
;
B[pe]
;
W[od]
;
B[ob]
;
W[la]
;
B[cl]
;
W[dl]
;
B[ma]
;
W[pc]
;
B[rb]
;
W[la]
;
B[dk]
;
W[el]
;
B[ma]
;
W[rd]
C[White 170:
The final ko fight of the game felt as solemn as a dirge. Lee's last hope was gone: AlphaGo had shattered the notion that it had any problem with ko.]
;
B[la]
;
W[qe]
;
B[re]
;
W[rf]
;
B[qd]
;
W[qa]
C[At 176, Lee Sedol resigned.
Although, with this game, AlphaGo had already claimed victory in the match, the rules stipulated that all five games would take place regardless of the results.
During the postgame press conference, Lee stated his new objective was to win at least one game. Even this seemed very unlikely.
At the same time, the idea that AlphaGo's victory would somehow ruin Go began to disappear. Confronted by defeat, people suddenly seemed to understand that humanity had not lost to a machine, but rather, had only surpassed itself! Furthermore, interest in Go had risen to unprecedented levels in both Asia and the West, creating a wonderful opportunity for further development.
I, too, believe we must not regard AlphaGo as a rival, against whom we can win or lose. Instead, we should see it as a partner, helping us more deeply explore the marvels of the game.
Perhaps, now that Lee could finally lay down the burden of winning the match, the fourth game would prove to be an especially brilliant one.]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[7200]
OT[3x60 byo-yomi]
PW[Lee Sedol]
PB[AlphaGo]
WR[9p]
DT[2016-03-13]
C[Game 4 - Endurance
2016-3-13
White: Lee Sedol
Black: AlphaGo
Komi: 7.5
Result: W+R
Commentary by Fan Hui 2p
Expert Go analysis by Gu Li 9p and Zhou Ruiyang 9p
Translated by Lucas Baker, Thomas Hubert, and Thore Graepel
When I arrived in the playing room on the morning of the fourth game, everyone appeared more relaxed than before. Whatever happened today, the match had already been decided. Nonetheless, there were still two games to play, and a true professional like Lee must give his all each time he sits down at the board.
When Lee arrived in the playing room, he looked serene, the burden of expectation lifted. Perhaps he would finally recover his composure, forget his surroundings, and simply play his best Go. One thing was certain: it was not in Lee's nature to surrender.
There were many fewer reporters in the press room this time. It seemed the media thought the interesting part was finished, and the match was headed for a final score of 5-0. But in a game of Go, although the information is open for all to see, often the results defy our expectations. Until the final stone is played, anything is possible.]
RE[W+Resign]
GN[Sedol-AG-4]
;
B[pd]
;
W[dp]
;
B[cd]
;
W[qp]
;
B[op]
;
W[oq]
;
B[nq]
;
W[pq]
;
B[cn]
;
W[fq]
;
B[mp]
C[Black 11:
For the fourth game, Lee took White. Up to move 11, the opening was the same as the second game. AlphaGo is an extremely consistent player, and once it thinks a move is good, its opinion will not change.]
;
W[po]
LB[ed:B][qn:A]
C[White 12:
During the commentary for game 2, I mentioned that AlphaGo prefers to tenuki with 12 and approach at B. Previously, Lee chose to finish the joseki normally at A. White 12, however, is an interesting alternative! Perhaps it was a test: would AlphaGo still tenuki as in game 2?]
;
B[iq]
C[Black 13:
This time, AlphaGo chose the ordinary extension. After this move, AlphaGo's win rate was 50.5%. Both players had 1 hour and 52 minutes apiece.]
;
W[ec]
C[White 14:
White played the usual corner approach.]
;
B[hd]
;
W[cg]
C[White 16:
Against Black's pincer, Lee counter-pincered to emphasize the left side. However, AlphaGo preferred the press as shown in the variation.]
;
B[ed]
;
W[cj]
;
B[dc]
;
W[bp]
;
B[nc]
C[Black 21:
At this point, Lee had 1 hour and 40 minutes left, AlphaGo 1 hour and 47 minutes.]
;
W[qi]
C[White 22:
The opening had been a balanced one, and AlphaGo's win rate stood at 53%.]
;
B[ep]
C[Black 23:
The better people got to know AlphaGo, the more started calling it by friendly nicknames, such as "Master A" in China and "Master Al" in Korea. Although such names mean little on their own, they represented a growing acceptance of AlphaGo as a partner and teacher, together with whom we could advance our understanding.
During this competition, "Master A" never ceased to surprise, playing at least one extraordinary new move in every game. This move was the first such display in game 4.
Theoretically, Black 23 is quite vulgar, as it induces White to strengthen the corner. However, since White's lower side is already solid, if this move helps Black on the outside, is it really as crude as it looks? See the variation.
One of the major ideas in Chinese philosophy, from kung fu to Taoism, is the notion that "formlessness defeats form." This is not to suggest that one should do nothing, but rather that one should be ready to make use of any resource at any moment. In other words, complete flexibility is the surest path to success. When one is not committed to any style, there are no weak points at which to aim. Naturally, reaching this state demands a strong foundation and a formidable repertoire. AlphaGo seemed to have attained this "formless" style already: simple, easy to understand, and totally unexploitable.]
;
W[eo]
;
B[dk]
C[Black 25:
Lee chose the outside hane, and just as I was reading out the cut, AlphaGo changed course with the shoulder hit at 25. Lee cracked a smile, as if he were looking at a naughty child. AlphaGo was being mischievous indeed!]
;
W[fp]
C[White 26:
Coolly, Lee protected the corner with 26. Most players found this move too slow, and even AlphaGo thought White should respond with the crawl as shown in the variation. Nonetheless, I saw this move as a sign that Lee had finally found the confidence to play his own game, regardless of anyone's approval. This was the Lee Sedol I knew: the wolf that, starving in the winter winds, still waited for his prey to come closer, biding his time for the moment his intuition knew would come.]
;
B[ck]
C[Black 27:
When Black blocked here, AlphaGo's win rate was 55%.]
;
W[dj]
;
B[ej]
;
W[ei]
;
B[fi]
C[Black 31:
Perhaps because it was playing Black, AlphaGo played very aggressively during this game. With the double hane at 29 and 31, it seemed as if Black was attempting to completely overwhelm White.]
;
W[eh]
;
B[fh]
;
W[bj]
C[White 34:
Even more shocking is that Lee submitted to it! Meek as a lamb, he let AlphaGo blockade the center and seal in White's group on the left side.
Sometimes to endure is difficult, sometimes it is irrational, and sometimes it is futile. Whatever it meant, Lee endured, and perhaps this was way of showing us his unwavering faith and unwillingness to surrender.
However, White did not need to suffer quite so badly. See the variation.]
;
B[fk]
;
W[fg]
;
B[gg]
;
W[ff]
;
B[gf]
C[Black 39:
At this point, AlphaGo's win rate reached 60%.]
;
W[mc]
C[White 40:
After Black had constructed such a thick wall, it became urgent for White to invade the top. The attachment at 40 was a classic technique to do just that.]
;
B[md]
C[Black 41:
At this point, Lee Sedol had 1 hour and 15 minutes left, AlphaGo 1 hour and 35 minutes.]
;
W[lc]
;
B[nb]
;
W[id]
C[White 44:
AlphaGo thought this move was problematic. In view of Black's thickness, White's most attractive option would have been to live on the spot.]
;
B[hc]
;
W[jg]
;
B[pj]
C[Black 47:
Compared to the previous games, Lee appeared much more relaxed and focused. Gone were the sighs, and the shaking of the head. Instead, he wore a look of intense concentration, as if waiting for something to arrive.
White gracefully leapt out into the center, but Black initiated a magnificent leaning attack!]
;
W[pi]
;
B[oj]
;
W[oi]
;
B[ni]
C[Black 51:
The greater the pressure became on White's top group, the more time Lee took to ponder each move. When Black played 51, Lee hesitated even longer. If Black dared to play so boldly, choosing the hane with only two stones to White's three, I could not imagine any move for White except the cut! Even AlphaGo thought cutting was the only move.]
;
W[nh]
C[White 52:
But Lee continued to endure! Against Black's hane, he haned in return.]
;
B[mh]
;
W[ng]
C[White 54:
Against the double hane, he extended!
"I can feel Lee's conviction," I wrote in my notebook, "waiting for the critical moment. But will that moment ever come?" As the game progressed, it seemed to everyone that Lee was once again on the verge of defeat.
At this point, Lee's clock had 51 minutes, AlphaGo's 1 hour and 28 minutes.]
;
B[mg]
;
W[mi]
;
B[nj]
;
W[mf]
;
B[li]
C[Black 59:
When Black ataried at 59, White had the option of linking up the group on the top. See the variation.]
;
W[ne]
;
B[nd]
;
W[mj]
C[White 62:
Instead, however, Lee chose to pull out his center stone. With White's group now isolated and in grave peril, Lee's heart must have been overwhelmed with emotions. But perhaps he also sensed the long-awaited moment at hand.]
;
B[lf]
C[Black 63:
Lee now had 42 minutes, AlphaGo 1 hour and 22 minutes.]
;
W[mk]
;
B[me]
;
W[nf]
;
B[lh]
;
W[qj]
C[White 68:
At this point, people suddenly began to congregate near the playing room. The rumor was spreading that the game was about to end, with AlphaGo victorious as expected. But Lee looked cool-headed as ever as he played the turn at 68. Was he really not afraid of dying?]
;
B[kk]
C[Black 69:
AlphaGo continued striding forward with the knight's move, aiming to swallow the white group whole. Lee pondered deeply over his next move. If White was to have any chance, he would have to seize it now.
At this point, Lee had 34 minutes, AlphaGo 1 hour and 19 minutes.]
;
W[ik]
;
B[ji]
C[Black 71:
When Black closed off the center with 71, Lee had 27 minutes, AlphaGo 1 hour and 17 minutes.]
;
W[gh]
;
B[hj]
C[Black 73:
After ten minutes of thought, Lee played the cut. Right on schedule, one minute later, AlphaGo enclosed White's cutting stone. Lee sighed, and kept thinking.]
;
W[ge]
;
B[he]
;
W[fd]
;
B[fc]
C[Black 77:
At move 77, Lee had only 11 minutes left. Moreover, AlphaGo?�s win rate had climbed over 70%. It seemed the game was over.
In fact, Lee had just completed the last of the preparations for his final charge!]
;
W[ki]
C[White 78:
At last, Lee Sedol launched his attack. Like an earthquake, the wedge at 78 tore apart the cracks in Black's fortress! None of us had anticipated this. When Gu Li saw White 78 from his broadcasting studio in China, he shouted: "The divine move!" All of Lee's painstaking preparations were finally about to bear fruit.
Actually, Lee spent very little time on this move itself. Later, during the press conference, he told the assembled reporters that he had not spent much time calculating. He had simply played what felt right.
Without a doubt, this move is a spectacular flash of insight - but does it really work? See the variation for an explanation of White's plan and Black's best response.
Regardless, this move cast AlphaGo into complete confusion.]
;
B[jj]
;
W[lj]
C[White 80:
Move 78 might not really work, but AlphaGo was at a complete loss to deal with it. When Black pulled back, White blocked at 80, and Black could no longer kill the white stones unconditionally.]
;
B[kh]
;
W[jh]
C[White 82:
Everybody still thought AlphaGo was aiming for the ko shown in the variation, but this was not the case.]
;
B[ml]
;
W[nk]
;
B[ol]
C[Black 85:
Inexplicably, AlphaGo began trying to extract the dead stones on the right side!]
;
W[ok]
;
B[pk]
;
W[pl]
;
B[qk]
;
W[nl]
;
B[kj]
;
W[ii]
C[White 92:
When White haned here, it was already difficult for Black to contain the white stones.
At this point, AlphaGo?�s win rate was in free fall. When White played 92, it dropped all the way to 55%, a full 15 points lower than before! What was going on?!]
;
B[rk]
;
W[om]
LB[rk:A][om:B]
C[White 94:
AlphaGo seemed to have gone crazy, and began thrashing around wildly. The exchange of A for B reinforced White's center with no compensation.]
;
B[pg]
;
W[ql]
;
B[cp]
LB[ki:A]
C[Black 97:
This wedge was completely beyond understanding! Aja Huang, who had been the picture of calmness since the beginning of the match, now looked at me as if to ask, "What's happening!?" I answered with a look that said, "I don't know."
Even now, we still do not know why AlphaGo lost its mind, playing senseless blunders one after another. Only one thing is certain: the original cause was the wedge at A, Lee Sedol's mystical "divine move." White 78 was incontrovertible proof of his determination to fight on, and his perseverance was rewarded with victory.]
;
W[co]
;
B[oe]
;
W[rl]
;
B[sk]
;
W[rj]
;
B[hg]
C[Black 103:
Little by little, AlphaGo recovered its sanity, but too late to save the game. When Black played 103, AlphaGo?�s win rate had fallen to 30%, the first time such low numbers had appeared since the match began.
At this point, Lee entered byo-yomi, but for the first time, he was seeing the light of victory.
The press room had been half-empty for some time, but now people began to pour in. Many reporters had already left, but when they heard that Lee Sedol might win, they turned around and came rushing back!]
;
W[ij]
;
B[km]
;
W[gi]
;
B[fj]
;
W[jl]
;
B[kl]
;
W[gl]
;
B[fl]
;
W[gm]
;
B[ch]
;
W[ee]
;
B[eb]
;
W[bg]
;
B[dg]
;
W[eg]
;
B[en]
;
W[fo]
;
B[df]
;
W[dh]
;
B[im]
;
W[hk]
;
B[bn]
;
W[if]
;
B[gd]
;
W[fe]
;
B[hf]
;
W[ih]
;
B[bh]
;
W[ci]
;
B[ho]
;
W[go]
;
B[or]
;
W[rg]
C[Everyone was anxiously awaiting the news of Lee's victory. As the professional commentators grew more and more certain of the conclusion, the game continued to progress onscreen. Playing in byo-yomi, Lee continued to answer AlphaGo's every move with utmost caution.]
;
B[dn]
;
W[cq]
;
B[pr]
;
W[qr]
;
B[rf]
;
W[qg]
;
B[qf]
;
W[jc]
;
B[gr]
;
W[sf]
;
B[se]
;
W[sg]
;
B[rd]
;
W[bl]
;
B[bk]
;
W[ak]
;
B[cl]
;
W[hn]
;
B[in]
;
W[hp]
;
B[fr]
;
W[er]
;
B[es]
;
W[ds]
;
B[ah]
;
W[ai]
;
B[kd]
;
W[ie]
;
B[kc]
;
W[kb]
;
B[gk]
;
W[ib]
;
B[qh]
;
W[rh]
;
B[qs]
;
W[rs]
;
B[oh]
;
W[sl]
;
B[of]
;
W[sj]
;
B[ni]
;
W[nj]
;
B[oo]
;
W[jp]
C[At move 180, AlphaGo finally resigned.
Lee Sedol entered the press room to thunderous applause. At long last, he could smile. It was a simple, modest smile, nothing world-changing - but then, aren't little moments like this the real source of joy in life?
AlphaGo resigned.]
)
(
;
GM[1]
FF[4]
CA[UTF-8]
AP[Jago:Version 4.25]
ST[2]
RU[Chinese]
SZ[19]
KM[7.5]
TM[7200]
OT[3x60 byo-yomi]
PW[AlphaGo]
PB[Lee Sedol]
BR[9p]
DT[2016-03-15]
C[Game 5 - Renewal
2016-3-15
White: AlphaGo
Black: Lee Sedol
Komi: 7.5
Result: W+R
Commentary by Fan Hui 2p
Expert Go analysis by Gu Li 9p and Zhou Ruiyang 9p
Translated by Lucas Baker, Thomas Hubert, and Thore Graepel
Lee Sedol's triumph in the fourth game attracted a surge of interest in the match throughout the world. Even the Western media, including CNN and the BBC, came to conduct reports and interviews. For Go, this level of attention was utterly unprecedented. More and more people started trying to learn and understand the game. To give just one example, Go Game Guru, the largest Go website in English, reported that its daily visitors had jumped tenfold.
In Korea, the heart of the craze, the media could talk of little else. For the next two days, the top three items at every news station invariably had something to do with the match. After winning the fourth game, Lee Sedol had risen to the status of a national hero. His brave request during the press conference after the fourth game was particularly inspiring. Normally, the two players would have chosen colors in the fifth game by nigiri, but Lee asked to play Black. The reason was simple: although AlphaGo prefers White, Lee had won as White already, and wanted to prove he was equally capable of winning with Black. His courageous spirit deeply moved the team, so for the fifth game, there was no need to choose colors. Lee would take Black.
After the fourth game, people had finally seen that AlphaGo was not invincible. Though the outcome of the match had already been decided, the result of the fifth game now seemed more important than the winner of the contest. If Lee prevailed again, it would perhaps mean that he had found the key to defeating AlphaGo. If so, the match score would no longer reflect the real balance between the two players.
During the rest day, reporters from around the world poured into Seoul, and any news stations that had lessened their coverage regretted their lack of foresight. On the morning of the final game, the press room was packed beyond capacity.
Since the match was taking place in Korea, Go fans in Europe had no choice but to wake up long before dawn, to catch the start of the game at 5 or even 4 AM. It felt like the World Cup!]
RE[W+Resign]
GN[Sedol-AG-5]
;
B[cp]
C[Black 1:
Lee entered the playing room looking relaxed and confident. He carried himself as if destined to win.]
;
W[pp]
;
B[dc]
C[Black 3:
Drawing on the experience of game 4, Lee started with a territorial fuseki, taking double 3-4 points.]
;
W[pd]
;
B[eq]
;
W[de]
;
B[ce]
;
W[cf]
;
B[cd]
;
W[dg]
;
B[fc]
;
W[co]
;
B[do]
;
W[cn]
;
B[bp]
C[Black 15:
Lee played the opening up to 15 unusually fast, as if he had prepared it in advance and anticipated AlphaGo's responses.]
;
W[dn]
C[White 16:
When White turned with 16, Lee stopped to think. Had this turn ever been played before? See the variation.
At this point, Lee's clock showed 1 hour and 55 minutes, AlphaGo's 1 hour and 51 minutes.]
;
B[ch]
C[Black 17:
Lee had probably researched this shape already, and after a few minutes of thought, he decided to attack White's top group. However, AlphaGo thought Black should simply extend, as shown in the variation.]
;
W[eo]
;
B[dh]
C[Black 19:
When White ataried on the lower side, Lee mumbled a few words and pushed up on the left.]
;
W[eh]
C[White 20:
The hane was White's only reply.]
;
B[eg]
C[Black 21:
Looking carefully across the board, Lee sighed and cut at 21. Afterwards, many professionals asked what happens if Black had haned on the outside instead. In fact, AlphaGo would have preferred the hane. See the variation.]
;
W[ei]
;
B[fg]
C[Black 23:
White discards the three stones. The impression is that Black has claimed considerable territory, but has he really? AlphaGo thought Black 23 was not the biggest move. See the variation.]
;
W[cj]
;
B[bf]
CR[de][cf][dg][eh]
LB[fc:A][eg:C][ei:B]
SQ[cd][bf][fg][dh]
C[Black 25:
Among professionals, there is a consensus that the most difficult aspect of Go is judgment. This means not only evaluating the overall shape of the game but, most importantly, judging the pros and cons of local variations. The hardest aspect is finding a reference point to which each variation can be compared. Of course, the higher a player's level, the deeper their understanding of local shape. Professional players leverage this understanding through tewari analysis, either pairing off moves or permuting their order to create a clearer reference point. During a dinner I attended with Nie Weiping, he raised the subject of this opening.
Nie Weiping evaluates the position as follows. Imagine we remove four white stones, marked by the circles, and four corresponding black stones, marked by the squares. The overall balance remains about the same, but it immediately becomes clear that Black's shape is no good, as the extension at A is too narrow and the exchange of B for C makes Black overconcentrated.
Of course, Black benefits from the exchange of the circled stones for the squared ones, so the position is not exactly equivalent. Regardless, Black has lost more from inefficiency than he has gained from the exchanges, so the real game still favors White.]
;
W[mc]
C[White 26:
White 26 protected the corner while diminishing Black's potential on the top. At this point, AlphaGo's win rate was 56%.]
;
B[qn]
;
W[nq]
;
B[rp]
C[Black 29:
After Lee played the knight's move at 29, he removed his watch, and I noticed his hands were shaking tremendously. Was it stress, or excitement? I could not tell.
At this point, Lee Sedol had 1 hour and 38 minutes, AlphaGo 1 hour and 42 minutes.]
;
W[ql]
LB[fq:A]
C[White 30:
White pincered here, and this is one of AlphaGo's favorite joseki. Later, Chang Hao mentioned that Go Seigen felt that this joseki, which encloses Black in the corner, clearly favors White. The great master was one of the few professionals who placed great emphasis on the centre. Regrettably, he passed away in 2014 at the age of one hundred. It would be fascinating to learn his views on AlphaGo if he were alive to see it today.
After the match, during a Go program with Meng Tailing, I raised the question of the proper timing for White's attachment at A. According to him, professional research has shown that although A is tempting, the result may turn out badly if one plays it too early. See the variation.]
;
B[pl]
C[Black 31:
The attachment at Black 31 emerged from Lee's desire to avoid being sealed in, especially considering AlphaGo's outstanding strength in the centre. Nonetheless, AlphaGo thought Black should have taken the corner. See the variation.]
;
W[pk]
;
B[pm]
;
W[qq]
;
B[qk]
;
W[qj]
;
B[rk]
;
W[pj]
;
B[rl]
C[Black 39:
The right side was settled for the time being, but the result was difficult to evaluate. Lee sighed again as he looked at the board, perhaps dissatisfied with the direction the game had taken.]
;
W[gh]
LB[pd:A][cj:C][pj:B][nq:D]
TR[kd][je][if][hg]
C[White 40:
His expression only grew more serious when he saw White 40. This jump was an outstanding move, shining light across the whole board! Once again, AlphaGo's move had surpassed our imagination, but having seen it, we all agreed it was the right one. What makes this move so beautiful? Let's try to explain its impact more clearly.
The gap between A and B on the right is fairly wide, so White must stay on guard against the possibility of invasion. On the left side, the group at C is not thin, but its potential for expansion is limited. The group at D is also difficult to develop because of Black's strength on the right. Although these groups were not very strongly related at first, after White 40, the left and right groups become loosely connected. Not only does this jump restrict Black's potential on the top, it also increases White's potential on the left. Finally, it begins to form an invisible web around the centre. If White can seal Black in on the right side, the entire centre will fall under White's influence.
One move that links all the groups together: now that's the essence of whole-board thinking!]
;
B[gg]
C[Black 41:
Lee cautiously responded by extending at 41.]
;
W[hg]
;
B[fh]
C[Black 43:
When White haned, Lee shook his head and grudgingly pushed in at 43. However, AlphaGo disliked Black 43, and instead suggested that Black hane directly as shown in the variation.]
;
W[fi]
;
B[hf]
;
W[hh]
;
B[if]
;
W[gf]
LB[if:B][fh:A]
C[White 48:
Although Black A had ensured that the extension at B was possible, it also lost a liberty, making the cut of 48 more severe.
At this point, Lee had 1 hour and 22 minutes, AlphaGo 1 hour and 34 minutes.]
;
B[ge]
;
W[bg]
C[White 50:
Black's unyielding move at 49 aimed to swallow up all of the white stones, but White's hane at 50 left a great deal of bad aji for Black due to the shortage of liberties. Lee's expression at this moment was grim, and he sighed heavily. It seemed as if Black would have to compromise. See the variation.]
;
B[bh]
;
W[be]
;
B[af]
C[Black 53:
The fight grew more complicated, but Lee appeared to be at a disadvantage. If White managed to get out, the game would be decided on the spot. In fact, things seemed to be heading in that direction.]
;
W[df]
C[White 54 is excellent move order.]
;
B[ee]
;
W[ed]
;
B[fe]
LB[ed:A][fe:B]
C[Black 57:
After the exchange of A for B, many of the live commentaries had already discovered a problem for Black. See the variation.]
;
W[ag]
C[White 58:
Shockingly, at this moment AlphaGo chose to atari at 58, and White's group simply died! Had AlphaGo made another critical blunder?
Studying the data, I found that AlphaGo was not confused the way it had been in game 4. In fact, the win rate had barely changed at all. Of course, from the human perspective of seeking the best moves, AlphaGo's choice was a foolish one. From a global perspective, however, White is by no means behind, and as we will soon see, the dead stones are far from meaningless.
Regardless, Lee had just gone from hell to paradise. He looked as if he sensed another sudden reversal at hand. Perhaps this was the cause of his tremendous blunder just a few moves later.]
;
B[dd]
;
W[fq]
C[White 60:
White 60 was extremely large. We have already analyzed this move, and earlier it would have failed for White. However, due to the exchanges on the left side, Black's refutation no longer works! This may have been the compensation White was looking for by sacrificing the stones on the left.]
;
B[fr]
;
W[er]
C[White 62:
See the variation for what happens if Black tries to avoid being sealed in.]
;
B[dr]
;
W[ep]
;
B[dq]
;
W[gr]
;
B[es]
;
W[gq]
LB[bi:A][di:B]
C[White 68:
Up to here, White had reaped a large profit on the bottom. Black was completely sealed into the corner, while White had built a powerful formation on the outside. Considering that White's moves at A and B had also become sente, the thickness of the upper wall was almost perfect. AlphaGo may have missed the chance to wrap up the game, but on the whole White was not doing badly at all. At this point, AlphaGo's win rate had already climbed above 70%.
At the same time, Lee Sedol was equally pleased, still basking in the warm afterglow of Black's profits on the left.]
;
B[mp]
C[Black 69:
White's aim to build up the bottom was clear, so Lee played the shoulder hit to reduce White's potential. Some commentators felt that this move went too deep, but AlphaGo would have made the same choice in Black's position.]
;
W[mn]
C[White 70:
The cap was the only move to start the attack.]
;
B[mq]
;
W[np]
;
B[mo]
;
W[nn]
;
B[ln]
;
W[lm]
;
B[kn]
;
W[km]
C[White 78:
Up to 78, the battle intensified, seemingly heading towards a climactic life-and-death battle.]
;
B[kp]
LB[km:B][lm:A]
C[Black 79:
Unexpectedly, however, Lee hesitated. Instead of extending, he turned back towards safety, making an eye with 79. Should Black have risked the life of the group to extend out into the center? See the variation.
If Lee were to play ten games from this exact position, he would have played the extension in perhaps nine of them. After the game, he must have asked himself, when it would have been so much his style to seek life in the centre, how could he possibly have chosen to back down? There are no doubt many causes, but the sense of superiority born of Black's unexpected gains on the left side must be the main one.
Besides this, White's moves at A and B would have taken immense courage for a human to play. White would be in deep trouble if Black lived inside White's moyo, so a human player would experience great doubt and hesitation before attacking this way. In contrast, AlphaGo's constant, unfeeling rhythm can easily unnerve its opponents. This may well be the most difficult aspect of playing against AlphaGo.
Black 79 was perhaps the move that Lee regretted most from the entire match. If he had extended instead, would it have changed the outcome? Sadly, we will never know. The French have a proverb: "With enough 'ifs', one could fit Paris in a bottle."]
;
W[jn]
C[White 80:
When White haned at 80, AlphaGo?�s win rate climbed to 76%. Now the centre and left side would become White territory, and Black would have to make a painstaking life at the bottom.]
;
B[jr]
;
W[jp]
;
B[jq]
;
W[jo]
;
B[ml]
C[Black 85:
Lee Sedol suddenly realized his plight. With this jump, he attempted to reduce White's influence, but only achieved the opposite. ]
;
W[nr]
;
B[mm]
;
W[no]
;
B[lq]
;
W[lk]
LB[no:B][nr:A]
C[White 90:
Not only had White gained points while making life with A and B, but after White jumped here, it was plain to see that Black's two stones were facing the wrong direction. Truly, blessings do not come in pairs, and misfortunes never come singly.
At this moment, AlphaGo?�s win rate was 84%. Lee had 46 minutes left, AlphaGo 1 hour and 9 minutes.]
;
B[jh]
;
W[lr]
C[White 92:
AlphaGo never gets too attached to its own territory, and when Black played the knight's move to threaten the centre, White showed no inclination to protect it.]
;
B[mr]
;
W[ms]
;
B[kr]
;
W[ls]
;
B[ks]
;
W[ns]
;
B[is]
;
W[kc]
C[White 100:
After some exchanges at the bottom, White chose the one-space jump at 100. This solid way of playing could only mean AlphaGo felt very optimistic about the outcome.]
;
B[kk]
C[Black 101:
AlphaGo thought this and the following moves were no good - see the variation. In any case, Black faced a very difficult game no matter what.]
;
W[ll]
;
B[lj]
;
W[mk]
;
B[mj]
;
W[qf]
C[White 106:
When White protected the corner with 106, AlphaGo's win rate stood at 86%. Lee had 29 minutes left, AlphaGo 57 minutes.]
;
B[pc]
;
W[qc]
;
B[nc]
;
W[oc]
;
B[nd]
;
W[md]
;
B[pe]
;
W[qe]
;
B[od]
;
W[pb]
;
B[ob]
;
W[pc]
;
B[mb]
;
W[ne]
;
B[oe]
;
W[nf]
;
B[lb]
;
W[kf]
;
B[kb]
;
W[jc]
;
B[ib]
C[Black 127:
Lee tried everything to get back in the game, but to no avail. Although Black had lived and connected out following the invasion, White had gained strength on the outside in compensation.]
;
W[jb]
;
B[ja]
;
W[ic]
;
B[hb]
;
W[hc]
;
B[gc]
;
W[gb]
;
B[fb]
;
W[ki]
C[White 136:
With White 136, AlphaGo struck Black's vital point, slowly building further potential in the centre.]
;
B[nk]
;
W[nl]
;
B[ji]
;
W[kj]
;
B[jj]
;
W[kh]
;
B[jg]
C[Black 143:
Many commentators felt that Black 143 was too passive, and that Lee should have wedged instead. AlphaGo agreed, but White was winning either way.]
;
W[kg]
C[White 144:
With this move, AlphaGo's win rate reached 92%.]
;
B[ol]
;
W[nm]
;
B[gj]
;
W[hj]
;
B[hk]
;
W[gk]
C[White 150:
As the endgame began, Lee looked fiercely for an opportunity, exhausting every possible resource. But no matter what he did, White remained a little bit ahead.
At move 150, Lee entered byo-yomi.]
;
B[hi]
;
W[fj]
;
B[ij]
;
W[ok]
;
B[ri]
;
W[rj]
;
B[sj]
;
W[ka]
;
B[la]
;
W[ae]
;
B[bf]
;
W[bi]
;
B[ah]
;
W[sk]
;
B[sl]
;
W[qh]
;
B[rh]
;
W[rg]
;
B[pg]
;
W[pf]
;
B[qg]
;
W[ph]
;
B[og]
;
W[of]
;
B[nh]
;
W[nj]
;
B[rf]
;
W[si]
;
B[sh]
;
W[nb]
C[White 180:
White 180 concluded the last major exchange, and Black was still a few points short. Lee must have recognized now that there was no way to win.]
;
B[gl]
;
W[fk]
;
B[hm]
;
W[ia]
;
B[ha]
;
W[rq]
;
B[re]
;
W[rd]
;
B[em]
;
W[gn]
;
B[fl]
;
W[dl]
;
B[qo]
;
W[gi]
;
B[am]
;
W[bm]
;
B[mh]
;
W[bo]
;
B[ao]
;
W[an]
;
B[bn]
;
W[al]
;
B[hp]
;
W[go]
;
B[sq]
;
W[sr]
;
B[sp]
;
W[qp]
;
B[rn]
;
W[iq]
;
B[ir]
;
W[el]
;
B[fn]
;
W[hq]
;
B[rr]
;
W[qr]
;
B[ss]
;
W[mi]
;
B[lh]
;
W[li]
;
B[jl]
;
W[im]
;
B[hn]
;
W[il]
;
B[hl]
;
W[ho]
;
B[in]
;
W[jm]
;
B[se]
;
W[hj]
;
B[ik]
;
W[jf]
;
B[ig]
;
W[af]
;
B[ef]
;
W[bd]
;
B[bc]
;
W[di]
;
B[cg]
;
W[an]
;
B[ap]
;
W[gs]
;
B[fs]
;
W[ko]
;
B[lo]
;
W[lc]
;
B[ga]
;
W[ie]
;
B[hd]
;
W[id]
;
B[he]
;
W[na]
;
B[ia]
;
W[oh]
;
B[ng]
;
W[lg]
;
B[sg]
;
W[ii]
;
B[fo]
;
W[en]
;
B[po]
;
W[sd]
;
B[oo]
;
W[op]
;
B[io]
;
W[ip]
;
B[dp]
;
W[dm]
;
B[fm]
;
W[ai]
;
B[bf]
;
W[ni]
;
B[ih]
;
W[mg]
;
B[rg]
;
W[hi]
;
B[qi]
;
W[pi]
;
B[ci]
;
W[bj]
C[White 280:
The endgame stretched out endlessly, making this by far the longest game of the competition. Even AlphaGo entered byo-yomi in the late endgame. However, the result had long since been decided.
Just as I began to believe we would finally have a game come down to counting, Lee resigned at move 280. In that moment, he looked utterly helpless.
Lee Sedol resigned.
In the end, the result of the match was a 4-1 victory for AlphaGo.
It was an unforgettable story, this week of intense competition between man and machine. Even now, when I recall the scene, I am just overwhelmed by emotion as I was then. I believe the most important contribution of this match is not the result, but rather the opportunity it has given us to rethink our understanding of Go, and of ourselves.
I would like to finish this commentary with a quote from Eric Schmidt, the executive chairman of Alphabet, spoken at the press conference in Seoul before the match:
The winner here, no matter what happens, is humanity."]
)