(;CA[UTF-8]CA[shift_jis]RU[Chinese]MU[1]GM[1]AP[SmartGo:3.1.4]SZ[19]FF[4]GN[31]EV[AlphaGo selfplay]RO[31]PB[AlphaGo]PW[AlphaGo]KM[7.5]RE[W+R]MULTIGOGM[0];B[pd];W[dd];B[pq];W[dp];B[qk];W[nc];B[cq]LB[pf:A][oc:B]C[Black plays an early 3-3 invasion. It would have been feasible for Black to simply answer in the upper right at A which used to be the most common move. Otherwise, Black could have kicked at B, the more AI-like and modernly popular move. ];W[dq];B[cp];W[cn];B[co];W[do];B[bn];W[qc];B[qd];W[pc];B[od];W[nd](;B[ne]C[Black could have pushed through to cut with this move (see var).];W[me];B[mf];W[nf];B[oe];W[le];B[lf];W[ke];B[ng];W[kf]LB[oc:A][ob:B][nb:C]C[This sequence in the upper right has been played in a number of human pro games. Black still has the option of pushing through at A to cut at C later, after which White will have to choose between capturing on the left, or connecting on the right. ](;B[cm]C[Black 29 is always a big move, that punishes White for playing away mid-joseki. The upper side was also big, about the same value (see var). ];W[dn];B[cr];W[fc];B[hp]C[This is the distance AG likes to be from White's wall when threatening to attack. At first, my feeling was that it was too close, but it puts a lot of pressure on White by taking away eye space. ];W[pp]C[This attachment has become a common move to play against Black's Chinese opening pattern. Depending on how Black replies, White is willing to go to the right or the left. ](;B[oq]C[In this case the lower side was wide open and relatively important when compared to the right side, so Black pulled back. If Black had played a hane in the corner, White would have switched to the lower side (see var). ];W[qp];B[qq];W[rq](;B[op]C[At this point Black should have a plan to build on the lower side while attacking White on the left, so curling at 39 is better than covering in the corner (see var). ];W[pn]LB[oo:A]C[White still can make good shape with a play at A, so that is miai with a play on the side or corner to make a base. ];B[gm];W[el];B[fn];W[cl];B[dm];W[em];B[bm];W[en](;B[dg]C[Black 49 was a turning point in Black's game plan. From here on in the game, Black will focus on attacking White's group in the lower left, and reducing White's potential on the upper side. Otherwise, Black could have expanded the Black moyo on the lower side (see var). ];W[ce];B[gj];W[ei];B[gg]LB[fh:A][gd:B]C[Compared to the more direct play at A, Black 53 is a flexible move that is ready to sac one stone on the left and aims directly at the key point of B. ];W[pl]LB[pk:A][ql:B];B[gd]LB[oc:A][ob:B][nb:C][gd:D]C[As mentioned before, Black always has the option of pushing through at A to cut at C. Will White just capture one stone and sac the corner, or will White connect on the right? That depends, of course, and having a stone at D changes that by creating a group for Black's stone at C to potentially connect up to. Black D is not only a big move that reduces White's area on the top, but it is looking at this weakness in White's position. ];W[ph]LB[of:A]C[White threatens to escape at A. ];B[of];W[pk]LB[qj:A]C[The right side is no longer a Black territory. It is not yet a White territory either, as Black can pull the stone out by playing at A. ];B[gc]LB[oc:A][lc:B]C[After Black plays at 59, Black is still looking at the push through and cut with Black A, but also has the vital point at B to aim at. ];W[jq]LB[fd:A]C[This is a position where if I had White I would want to push at A, and I would also be thinking about how to settle my group on the left side. Instead, AG invades to create another goup! After this move, Black's final moyo has been invaded and Black's only territory to speak of is in the LL corner. While White is ahead in territory, we are uncertain about where exactly that territory will be at the end of the game. None of White's territories are finished, in that they all have weaknesses that Black can exploit. ];B[gq]LB[gq:A][er:B][hr:C]C[With Black A, Black is focusing on the coming attack against White on the left, and also strengthening Black's group with the potential to connect underneath at B. If Black had played at C instead, the focus would have been on an attack against White on the lower side. ];W[lq];B[ci](;W[gi]C[Not expected! The var shows a more straight-forward approach for White. ](;B[ej]C[Black counters strongly. The var shows White's plan if Black had covered on the right. ];W[gl](;B[hm]C[Black pulled back, so one migfht ask why not cover on the right (see var). ];W[gk];B[fj];W[hj];B[hi];W[ij];B[il];W[hl];B[km];W[jn];B[im](;W[hh]TR[ei]C[Capturing in a ladder, White has saved the weak group. Meanwhile, Black got some profit on the left while reinforcing the Black group in the upper left, and is one step closer to attacking White on the lower side. If White had tried to save the marked stone beforetaking in a ladder, that would have been dangerous (see var). ];B[fi];W[ii];B[rr]TR[op][pq]C[Black makes a base for the marked group. ];W[rc]LB[oc:A][lc:B]C[White gets rid of the weakness at A while preparing to put some pressure on the outside Black group. Black can still aim at the weakness at B. ];B[qj]TR[of][ng]LB[qj:A]C[Black A is a big move that takes away White's potential side territory, and aims to link up to reinforce the marked Black group. ];W[pj];B[rh];W[fd]LB[rr:A][rc:B][qj:C][fd:D][dl:E]C[A~D were all urgent points, Now we can have an idea of where the territories will be. White's two main territories will be the UL and UR corner areas, and Black only big territory is the lower left corner. If Black gets to play at E, Black wull have a big left side. The game looks very close. ];B[hg]LB[ge:A]C[Black 87 indirectly protects against White A. ];W[hr]C[White settles the lower side group. ];B[gr];W[lo];B[gb];W[fb];B[cc]LB[cc:A]C[This invasion at A is always annoying, while White should have no trouble killing it. Black has ideas of squeezing from the outside and forcing White to use some extra stones to capture Black.](;W[df]C[White deals with the invasion while seeming to be ignoring it. The var shows the more mundane move for White, and how Black wants to use this invasion to squeeze White in the UL corner. ];B[mn];W[ql];B[cg];W[eg]LB[df:A][eg:B][de:C][fh:D]C[The point of White's plays at A and B is that now White has an eye at C, and is threatening to cut Black at D. ];B[fe];W[ee];B[ff];W[ef];B[fh];W[dl]C[White 104 was a big move that maintained White's lead in territory. ](;B[jh]TR[ne][mf][of][gg][ng][fh][hj][gk][el]C[Black 105 is a multi-purpose move that strengthens the two marked Black groups while starting to put some pressure on the marked White group. Black has some aji in the corner, but is not quite ready to make use of it (see var). ];W[eh];B[dj];W[kj];B[kg];W[lk]TR[ii][hj][pj][il][ql][gm][op][pq]LB[kj:A][lk:B][rp:C][jc:D]C[White's moves at A and B subtly change the status of a number of groups. None of the marked groups are in immediate danger, but White A and B reinforce White's groups and add value to endgame plays such as C. This was a very patient way for White to play when there was still a big move at D. ];B[lc]LB[lc:A][md:B][nb:C]C[Black A is now the biggest move. Aside from the obvious gain in territory on the upper side, Black has the direct cut at B or the attachment tesuji at C to play next. ];W[jc];B[jd](;W[id]C[White 114 is a good shape move. Pushing through on the fourth line loses the semeai (see var). ];B[kc](;W[mc]TR[ga]C[White backs off, keeping the semi-forcing move at the marked point. White could not have captured the Black stones in a semeai (see var). ];B[ml];W[mk];B[kn]LB[kn:A]C[Black's attack on the lower side starts with A. ];W[ko]TR[kn]C[The effect of the marked exchange should become relatively clear in the next few moves. ];B[iq];W[ir];B[jp];W[kp];B[jr];W[kq]LB[kn:A][ko:B][jo:C]C[Now it should be easy to see that Black has gained some profit by exhanging A for White B first. If Black had played A later, White would not answer at B, and would have the option to save a stone by playing at C. ];B[hs];W[jo];B[is];W[kr];B[hq];W[ho];B[go];W[nr]LB[lr:A][jp:B][mp:C]C[White has one eye at A, and potential eyes at B or C. ];B[on];W[be]C[White finally makes the upper left corner into a territory. ];B[oi](;W[ni]C[Black was still attempting to cut White off in the center, so White avoids that danger (see var). ];B[pm];W[qn];B[oj];W[oh];B[or];W[mr];B[pi];W[qi];B[ok];W[ol];B[qh];W[nl];B[je]LB[qh:A][je:B]C[Black has gained a lot of points, capturing at A to expand the right side, and detting the big point at B. As a human player, I have the feeling that White was slack and has lost some ground here. I guess that AG with the Google super-computer was able to calculate that this was good enough for White, but I doubt that any human player would have been able to do the same. ];W[jf];B[ie];W[if];B[he];W[rp];B[jk];W[lh];B[rd];W[mb];B[nh];W[ga];B[ki];W[li];B[jj];W[ji];B[kh];W[ig];B[kk];W[lj];B[lg];W[ih];B[ha];W[fa];B[mp];W[ip];B[ib];W[sr];B[bl];W[mm];B[ln];W[kl];B[jm];W[qr];B[dc];W[fp];B[fo];W[da];B[sc];W[sb];B[hn];W[io];B[sd];W[qb];B[bf];W[ck];B[db];W[ad];B[ab];W[ba];B[ca];W[cb];B[ea];W[eb];B[bj];W[rk];B[rj];W[dr];B[fm];W[fl];B[bb];W[ca];B[aa];W[ed];B[om];W[sj];B[ri];W[rm];B[ik];W[hk];B[ea];W[ca];B[nm];W[ll];B[cs];W[fq];B[gp];W[ka];B[kb];W[bk];B[ak];W[ja];B[la];W[cj];B[bh];W[lb];B[jb];W[ma];B[da];W[nj];B[si];W[sk](;B[mh]C[One final var to show that Black loses sente if it plays in the UL corner. ];W[bc];B[ns];W[ks];B[ms];W[ls];B[os];W[mq];B[fr];W[af];B[ag];W[ae];B[ac];W[cd];B[er];W[hf];B[fg];W[ec];B[bd];W[nk];B[qi];W[bc];B[oo];W[mo];B[bd];W[no];B[nn];W[bc];B[po];W[qo];B[bd];W[al];B[am];W[bc];B[qs];W[rs];B[bd];W[pg];B[rf];W[bc];B[pr];W[rr];B[bd];W[rg];B[qg];W[bc];B[np];W[cb];B[ia];W[la];B[oc];W[ob];B[js];W[ds];B[lm];W[cf];B[bg];W[mi];B[dh];W[es];B[gh];W[fs];B[hi];W[eq];B[eo];W[di];B[kd];W[ml];B[ps];W[gi]C[W+0.25 stones(W+0.5)])(;B[bc];W[mh];B[cf];W[cd];B[af];W[bd];B[ae];W[ec];B[dh];W[ba];B[fg];W[ac]LB[bc:A][mh:B]C[Black A was a vital point that gave Black a squeeze, but was not sente. White B was equally big. Black gets all of the squeeze moves in the upper left and will take the final big point at C, just to lose by half a point anyway. ]))(;W[pi];B[nk];W[nj];B[oj]TR[ph][kj][lk][ql]C[While this is probably not going to be a killing attack, Black has potential to gain some points by cutting White off and attacking the two marked White groups. ]))(;W[jb];B[je];W[jf];B[ie];W[kd];B[nb];W[oc];B[if];W[jg];B[kh];W[ih];B[ig];W[md];B[mb]LB[nb:A]C[Black A is again the vital point tesuji that helps Black get some extra liberties. Black wins the semeai to get a big territory. ]))(;W[kd];B[kc];W[jb];B[nb];W[oc];B[jf];W[je];B[id];W[ie];B[ic]TR[jb]LB[nb:A]C[Black captures the marked stones. Black's stone at A takes away White's chances to get an effective squeeze. ]))(;B[dc];W[eh];B[dj];W[bd];B[eb];W[ga];B[ea];W[fa];B[bc];W[hb]LB[ga:A][bb:B]C[If Black tries to make trouble in the corner, White will just play at A and let Black live. White has gained some territory on the upper side, and still has a ko in the corner with White B later. ]))(;W[bd];B[bf];W[cf];B[cg]LB[bf:A][bc:B]C[If White attacks the corner, Black will play from outside with A. Later Black can play at B to create a very distant step ko, so White is not in any great danger but Black will have opportunities to squeeze and get some more forcing moves from outside. ]))(;W[fi];B[ek];W[dl];B[dj];W[fm];B[hh]LB[fi:A][ek:B]C[If White tries to force with A before capturing in a ladder, Black can save the 3 black stones with a forcing move at B, and White loses sente. ]))(;B[hl];W[gk];B[fk];W[fl];B[hk];W[hj];B[fj];W[hm]LB[ij:A][dj:B]C[White has a forcing move at A that threatens to play next at B, so Black is falling apart in this var. ]))(;B[hi];W[gl];B[hl];W[gk];B[hk];W[fj];B[hj];W[cj];B[bj];W[di];B[ck];W[dj]LB[hi:A][gl:B][cj:C]C[If Black simply answers at A, White will now start the sequence from B, and gets sente to play the vital point at C. This is easy for White. ]))(;W[gl];B[hl];W[gk];B[hk];W[fj];B[dj];W[gi];B[hj];W[dk];B[hi];W[ej];B[cf];W[fo]LB[gl:A][dh:B][cc:C]C[The sequence from White A is relatively simple to understand, with the implied threat of White B to connect up to the corner group. While White easily lived, Black has reinforced its groups on all sides, and now can think of invading at C. ]))(;B[om];W[on];B[nn];W[nm];B[no];W[ql];B[pl];W[rl];B[nl];W[mm];B[ml]LB[om:A]C[Black A is the proper shape to attack White on the right. In this var, Black is focusing on the lower side area, while White will have opportunities to expand White's moyo on the upper side. The game is about even at this point. ]))(;B[rr];W[ro];B[sq];W[no]LB[rr:A]C[If Black takes White's base away with A, White can play lightly into the center with this sequence. ]))(;B[qp];W[oq];B[po];W[op];B[qq];W[kq];B[oo];W[mp]LB[qp:A][oq:B][hp:C]C[If Black A, then White moves to the lower side with B. In this variation Black stone at C has become weak, and Black has lost the opportunity to attack White's wall on the left. ]))(;B[fc];W[eb];B[fb];W[ec];B[fe]LB[fc:A]C[Invading at A is a big move that reduces White's potetial on the upper side.]))(;B[oc];W[ob];B[pb];W[qb];B[nb];W[pa];B[ne];W[me];B[nf];W[mf];B[mg];W[ld];B[nh]TR[od]LB[oc:A][pf:B]C[Pushing through at A first has always been a popular way for Black to play, but has become less common recently. The drawback for Black is that the marked stones have lost a valuable liberty when Black played at A, and White moves such as B will be more effective because of that. ]))