(;GM[1]FF[4]CA[UTF-8]AP[CGoban:3]ST[2] RU[Japanese]SZ[19]KM[7.00] PW[kata1-b60c320-s4406649088-d2410291439]PB[kata1-b40c256-s5626082304-d1354534002]C[startTurnIdx=0,initTurnNum=0,gameHash=504771034A6FD5CB576EFC64565114E6,gtype=normal {'ko': 'POSITIONAL', 'tax': 'SEKI', 'scoring': 'AREA', 'suicide': True} Analysis of kg40s994 various shape descriptions and strategy annotations mentioned in passing. 115: wordy analysis starts 154: very soft says the analysis. 165: ok, but bad shape instinct. There is a move that is almost the same but dominates this move in pretty much every way. The analysis barely considers the game move in comparison. 166: mistake according to analysis 184-onwards: detailed endgame analysis 244: lengthy ko analysis 244: losing move 270-end: Overanalysing dame filling. There are still good instinct patterns to be found in AI play here. 284: very dame analysis Detailed endgame analysis, but I sometimes switch between area scoring and territory scoring with my numbers, apologies. This causes a difference especially when different sides play a different number of moves locally. It will mostly be territory scoring unless the endgame is big and it is easier to count the number of intersection s that have switched hands.]RE[0] ;B[dp]C[0.52 0.48 0.00 0.3 v=600] ;W[pd]C[0.52 0.48 0.00 0.3 v=600] ;B[pp]C[0.52 0.48 0.00 0.3 v=645] ;W[dd]C[0.52 0.48 0.00 0.3 v=680] ;B[qc]C[0.52 0.48 0.00 0.3 v=781] ;W[qd]C[0.52 0.48 0.00 0.3 v=693] ;B[pc]C[0.52 0.48 0.00 0.3 v=941] ;W[od]C[0.52 0.48 0.00 0.3 v=1176] ;B[nb]C[0.52 0.48 0.00 0.3 v=1530] ;W[cq]C[0.52 0.48 0.00 0.3 v=1352 very standard. This exact opening has appeared countless times.] ;B[dq]C[0.52 0.48 0.00 0.3 v=1570] ;W[cp]C[0.52 0.48 0.00 0.3 v=1107] ;B[cn]C[0.52 0.48 0.00 0.3 v=1522 I'm surprised to see analysis recommend this since white has the ladder but I guess black can more easily play this than white?] ;W[bn]C[0.52 0.48 0.00 0.3 v=1257] ;B[bm]C[0.51 0.49 0.00 0.1 v=885] ;W[co]C[0.52 0.48 0.00 0.3 v=1370] ;B[dn]C[0.51 0.49 0.00 0.1 v=1465] ;W[do]C[0.52 0.48 0.00 0.3 v=1946] ;B[eo]C[0.51 0.49 0.00 0.1 v=2045] ;W[ep]C[0.52 0.48 0.00 0.3 v=2525] ;B[er]C[0.51 0.49 0.00 0.2 v=2633] ;W[fp]C[0.52 0.48 0.00 0.3 v=2890] ;B[fo]C[0.51 0.49 0.00 0.2 v=3224] ;W[go]C[0.52 0.48 0.00 0.3 v=3450] (;B[gn]C[0.51 0.49 0.00 0.2 v=3816] ;W[gq]C[0.54 0.46 0.00 0.7 v=839] ;B[cr]C[0.52 0.48 0.00 0.2 v=2277] ;W[bo]C[0.54 0.46 0.00 0.7 v=824] ;B[ho]C[0.52 0.48 0.00 0.3 v=1400] ;W[gp]C[0.54 0.46 0.00 0.7 v=1185] ;B[iq]C[0.52 0.48 0.00 0.2 v=1765] ;W[hn]C[0.54 0.46 0.00 0.7 v=1665 k=7 (white ahead in opening though, 56%, +0.5)] (;B[en]C[0.49 0.51 0.00 -0.3 v=791 k=5 black doesn't recover for the longest time] ;W[br]C[0.57 0.43 0.00 1.3 v=800] ;B[bs]C[0.50 0.50 0.00 -0.1 v=989] ;W[bq]C[0.56 0.44 0.00 1.2 v=1175] ;B[fr]C[0.52 0.48 0.00 0.4 v=1257] ;W[cm]C[0.59 0.41 0.00 2.0 v=1511] (;B[hr]C[0.52 0.48 0.00 0.4 v=1592 an usual choice given that white still has the H4, J4 threat to push through. The AI thinks this makes things worse and black doesn't recover for the longest time.] ;W[gr]C[0.63 0.37 0.00 2.6 v=808] ;B[cl]C[0.54 0.46 0.00 0.9 v=931] ;W[dm]C[0.64 0.36 0.00 2.7 v=1058] ;B[em]C[0.54 0.46 0.00 0.7 v=1105] ;W[gs]C[0.65 0.35 0.00 2.8 v=1633] ;B[in]C[0.54 0.46 0.00 0.7 v=1673] ;W[hm]C[0.66 0.34 0.00 3.0 v=1247] ;B[io]C[0.54 0.46 0.00 0.8 v=986] ;W[gm]C[0.67 0.33 0.00 3.4 v=876] ;B[fn]C[0.58 0.42 0.00 1.5 v=1244] ;W[dl]C[0.66 0.34 0.00 3.1 v=1402] ;B[el]C[0.58 0.42 0.00 1.6 v=1762] ;W[dk]C[0.65 0.35 0.00 2.8 v=1998] ;B[ek]C[0.58 0.42 0.00 1.5 v=2359] ;W[dj]C[0.64 0.36 0.00 2.8 v=2596] ;B[ej]C[0.58 0.42 0.00 1.5 v=2867] ;W[di]C[0.64 0.36 0.00 2.8 v=3007] ;B[ei]C[0.58 0.42 0.00 1.6 v=3228] ;W[eh]C[0.64 0.36 0.00 2.7 v=3325] ;B[hj]C[0.58 0.42 0.00 1.6 v=3463] ;W[fh]C[0.67 0.33 0.00 3.4 v=959] ;B[hq]C[0.58 0.42 0.00 1.6 v=2263] ;W[cc]C[0.67 0.33 0.00 3.4 v=1152] ;B[dh]C[0.58 0.42 0.00 1.4 v=1440] ;W[dg]C[0.69 0.31 0.00 3.7 v=1044] ;B[ch]C[0.58 0.42 0.00 1.4 v=1337] ;W[cg]C[0.70 0.30 0.00 3.9 v=1298] ;B[mc]C[0.58 0.42 0.00 1.4 v=1479] ;W[hi]C[0.69 0.31 0.00 3.7 v=853] ;B[ij]C[0.59 0.41 0.00 1.6 v=700] ;W[ii]C[0.69 0.31 0.00 3.5 v=707] ;B[gj]C[0.57 0.43 0.00 1.3 v=621] ;W[ji]C[0.69 0.31 0.00 3.5 v=654] ;B[kk]C[0.57 0.43 0.00 1.2 v=904] ;W[qn]C[0.70 0.30 0.00 3.9 v=737] ;B[jd]C[0.58 0.42 0.00 1.5 v=760] ;W[lk]C[0.69 0.31 0.00 3.4 v=613] ;B[kj]C[0.58 0.42 0.00 1.4 v=617] ;W[on]C[0.70 0.30 0.00 3.6 v=657] ;B[qo]C[0.59 0.41 0.00 1.6 v=757] ;W[rn]C[0.71 0.29 0.00 3.7 v=841] ;B[np]C[0.59 0.41 0.00 1.5 v=690] ;W[pj]C[0.71 0.29 0.00 3.7 v=1110] ;B[gc]C[0.61 0.39 0.00 1.9 v=737] ;W[md]C[0.72 0.28 0.00 3.6 v=760] ;B[ld]C[0.62 0.38 0.00 2.0 v=689] ;W[qq]C[0.72 0.28 0.00 3.6 v=851] ;B[pq]C[0.63 0.37 0.00 2.0 v=639] ;W[qp]C[0.74 0.26 0.00 3.9 v=900] ;B[nm]C[0.63 0.37 0.00 2.0 v=883] ;W[nk]C[0.72 0.28 0.00 3.5 v=600] ;B[bg]C[0.61 0.39 0.00 1.7 v=627] ;W[ck]C[0.74 0.26 0.00 3.8 v=608] ;B[rc]C[0.63 0.37 0.00 2.0 v=600] (;W[nn]C[0.72 0.28 0.00 3.3 v=667] ;B[qg]C[0.61 0.39 0.00 1.7 v=613] (;W[og]C[0.71 0.29 0.00 3.1 v=617] ;B[po]C[0.62 0.38 0.00 1.8 v=767] ;W[ro]C[0.70 0.30 0.00 3.1 v=687] ;B[qi]C[0.61 0.39 0.00 1.6 v=859] ;W[lo]C[0.69 0.31 0.00 2.9 v=1026] ;B[lq]C[0.60 0.40 0.00 1.6 v=714] ;W[qj]C[0.70 0.30 0.00 2.9 v=801] ;B[oh]C[0.61 0.39 0.00 1.5 v=890] (;W[pg]C[0.69 0.31 0.00 2.6 v=932] ;B[ph]C[0.57 0.43 0.00 1.0 v=669] ;W[qf]C[0.67 0.33 0.00 2.5 v=780] ;B[ng]C[0.56 0.44 0.00 0.9 v=1188] ;W[rg]C[0.66 0.34 0.00 2.4 v=1131] ;B[qh]C[0.57 0.43 0.00 1.0 v=1420] (;W[me]C[0.66 0.34 0.00 2.3 v=1446] (;B[nf]C[0.57 0.43 0.00 1.1 v=1483] ;W[le]C[0.67 0.33 0.00 2.5 v=846] ;B[ke]C[0.57 0.43 0.00 1.2 v=967] ;W[lg]C[0.68 0.32 0.00 2.7 v=1256] (;B[pf]C[0.57 0.43 0.00 1.1 v=1405 steady] ;W[pe]C[0.69 0.31 0.00 2.8 v=1529] (;B[of]C[0.57 0.43 0.00 1.1 v=1521] ;W[rj]C[0.69 0.31 0.00 2.7 v=1370] (;B[mh]C[0.59 0.41 0.00 1.4 v=1083] ;W[kl]C[0.69 0.31 0.00 2.8 v=636 if black gets M7 that even works with O7 and hence Q6.] (;B[kf]C[0.61 0.39 0.00 1.6 v=1130 Easy to miss this key point but black has upper side potential and white isn't alive either.] (;W[re]C[0.76 0.24 0.00 3.7 v=726] (;B[nd]C[0.63 0.37 0.00 1.8 v=818 this seems like a weird move to cut but white could have played O17 in sente. If black can get this move in sente, that gains one point in double sente. Furthermore, reducing white's liberties threatens to capture the Q16 stones in ko. Of course, it is also necessary to spend a move to save the O13 group and this move just tries to do so efficiently.] (;W[rd]C[0.73 0.27 0.00 3.1 v=600 not strictly necessarily to live but very big due to white's liberty shortage. Saving the Q16 stones is worth at least 10 (4 stones +P17)/3 points. Then the T17 endgame is another 3 points or so and making the R15 points is another 3 points or so. This is summing to around 10 points, which is smaller than R2. But saving white here means that white O15 will save N15 as well (perhaps at least gaining 2 points on top). In the game we will see white make black pay a little to make life with the O12 group as well. If white plays Q2 here and R2, we see analysis say B+2 (at k=5), so the difference is probably just 2 points.] (;B[db]C[0.58 0.42 0.00 1.1 v=642 good order of moves ] (;W[cb]C[0.72 0.28 0.00 2.7 v=643 questionable says the analysis.] ;B[eg]C[0.55 0.45 0.00 0.7 v=696 powerful!] (;W[fg]C[0.69 0.31 0.00 2.3 v=813] ;B[ef]C[0.52 0.48 0.00 0.4 v=817] (;W[bh]C[0.70 0.30 0.00 2.4 v=830] (;B[ff]C[0.54 0.46 0.00 0.6 v=1342 black makes a lot of points here in the game and it starts by saving weak stones at the boundary, even though they aren't really cutting stones since white is connected, there are still cuts to threaten at G11 and C15.] ;W[jl]C[0.71 0.29 0.00 2.5 v=906] (;B[lj]C[0.52 0.48 0.00 0.4 v=1358 analysis doubts this move. It ends up being captured, but it does make white thin in the centre. The problem is that white can exploit the capturing race at G6 to make some shape (though is it enough?)] (;W[im]C[0.72 0.28 0.00 3.0 v=743 Much bigger than it looks, saving 3 stones is at least 6 points, but later white plays K10 and forces black to play all the 5 moves to capture G5 which also helps ko threats and makes 2 points around J8 and an eye (helping the flimsy centre white stones) and prevent adding a move to capture D12. Although black doesn't have an eye at D2 since white can force with A2 and D1, that costs white more points having to throw in at C1. Furthermore, the K10 cut starts an attack on the O14 group as well as setting up combinations with L3 which threats to get K6 in sente.] (;B[ec]C[0.61 0.39 0.00 1.9 v=828 This is to prevent white playing F18 to capture D18 pretty much in sente locally (though it doesn't work yet as white isn't alive. This strong move also pretty much seals the line with E14 which makes it much better than other moves to save D18 such as F18 or E18. If C19 is sente, then saving D18 alone is 11 points worth in reverse sente, just counting the side territory at face value. It isn't really this much since white playing F18 wasn't really sente and also black's territory still has some weaknesses like this.] ;W[de]C[0.74 0.26 0.00 3.1 v=687 Better for points than C14 and more threats against black's top side. Consider the B13-C13-D13-E13 (+C12, D12) symmetric shape. The symmetry is broken by E14, so as usual, if white can defend everything for territory, the key is to defend against E14, the point of asymmetry more. This move is much better than C15 since black can still force white back a lot with the C14 atari. The key is that the C14 atari is more problematic that C15 since it is deeper in white's position. C14 works because E14 shorts the liberties of D14. This doesn't mean white should play D14 themselves, since we just said that D14 is low on liberties, but the D15 move counters by increasing the liberties of the D14 stone (to pretty much infinity), and also makes the C15 point of territory as well.] ;B[mk]C[0.63 0.37 0.00 2.1 v=933] (;W[ll]C[0.77 0.23 0.00 3.7 v=748] (;B[gf]C[0.65 0.35 0.00 2.3 v=853] (;W[gi]C[0.73 0.27 0.00 3.4 v=613] ;B[qr]C[0.66 0.34 0.00 2.0 v=652 Black invested in the R5 stone before which wasn't much use after R6 and S5, but it is still not over. Black is still claiming dividends from the way R5 shorts the liberties of white. This sometimes loses a little if white captures this stone since white becomes very alive and can fight a ko at P1, but in this game since white chose to squeeze for the centre, it didn't really lose anything. This is normally good endgame locally. The S3 atari and S1 atari reduces white by at least one point if white captures Q2 later. Also this occurs in sente whereas white could have prevented that almost in sente by playing R2 - Q2 - Q1 (though the L3 sequence must be played first in which case black likely has the chance to play R2 during that sequence). Essentially the idea is to reduce the size of white's follow up locally, even though it doesn't really change the size of black's follow up.] (;W[jj]C[0.76 0.24 0.00 3.0 v=734 This move is much more valuable than it looks, helping to prevent white from having to add 2 moves capturing and connecting D12.] (;B[fj]C[0.67 0.33 0.00 2.1 v=845 Necessary. Black wins by 2 liberties now.] ;W[rr]C[0.78 0.22 0.00 3.2 v=915 block of course. Black was merely going for a sente exchange not the corner. This makes the R6 completely alive, mitigating the Q6 cut and hence white's centre thinness. Just as territory, if black gets S2 (compared to black playing Q2 here), that reduces white by a further 1 points even if white responds with T3 (a tesuji). Black's territory increases by around 3 if white responds and the next black move at S4 if white doesn't captures S4 has a swing of another 10 or so. This is a total of 4+10/2=9 points. White maintains the 12 point follow up at Q2 as well so this position can be counted as 12/2 =6 points less than if black got Q2. My calculation method is a bit nonstandard here, and more biased towards practical counting rather than the theory, so apologies if it is hard to follow, but it is based on miai counting. In fact as 9>6, we normally consider R2 as simply a sente move. Anyway this means the swing of the S2 move is 9+6=15 points, very big. If white passes instead, the analysis says the position is even at k=17, which is a difference of 12 points. This is rather high as you would normally expect this to be half the size of the largest move. For example, komi is 7, so passing at the start of the game loses 14 points. This suggests the swing of an opening move is around 28 points (though that is not an endgame move so it doesn't really make sense to talk in such terms) according to standard theory. So we expect the swing of the largest move in this position to be around 12x2=24 points (rather than 15) I suppose there are 2 possible explanations and I prefer the former. 1) This is that we have only counted the territory value of S2. In terms of thickness such as the cutting point at Q6, N6, K9 making life with this group counts for a lot more than 15 points. Especially considering that in the game, white defended the shape by squeezing to get Q6, N4, N5, O5 in sente. 2) The second largest move is worth much less than 15 points so that although normally getting the initiative to play a 15 point move will gain you 15/2=7.5 compared to the opponent playing it, the rest of the endgame might consist pretty much of 6 points moves, so that you gain 15 - 6/2=12. __ Just in terms of shape, white has a shortage of liberties and S3 and S4 are atari, but conveniently, responding to the atari captures the cutting stone. Locally this white move is the strongest move regardless of whether the Q2 cutting point existed or not. For example playing S1 would get cut to pieces. Playing a connected move helps mitigate the S4 cut by creating a buffer zone. With the Q2 cut, it is good instinct to play against the weak R2 stone especially as the first R1 is a death zone noone wants to play against.] (;B[ok]C[0.67 0.33 0.00 2.0 v=1347 unnecessary but a fancy tesuji] ;W[kq]C[0.75 0.25 0.00 2.7 v=625 The analysis doesn't seem to see this move at this timing but white's win rate goes up 4% after the next 2 moves. This uses J7, helping to explain why white was so determined to save J7 in all the moves earlier, even though it took so many moves.] (;B[kr]C[0.65 0.35 0.00 1.8 v=1127] (;W[kp]C[0.78 0.22 0.00 3.4 v=815 The analysis doesn't understand this move.] (;B[jr]C[0.67 0.33 0.00 2.1 v=860] (;W[mj]C[0.78 0.22 0.00 2.9 v=676 nice] ;B[oj]C[0.67 0.33 0.00 1.8 v=812 Black can just about pull this stone out.] (;W[ni]C[0.77 0.23 0.00 2.8 v=1026 Nice way to poke out black's eyes in sente since for now cutting P11 is very big. This is safe for now since if black cuts O10, white captures M11 with N11 and M12. Black capturing O9 doesn't actually make an eye although quite big.] (;B[oi]C[0.70 0.30 0.00 2.6 v=1166 Only hope to save these stones. Black has miai of O15 and N8 to make life] (;W[ml]C[0.78 0.22 0.00 2.8 v=859 The analysis says this move is very soft.] ;B[mi]C[0.68 0.32 0.00 1.8 v=1105 This makes life in the most efficient way, threatening to cut O10 to solidly make life even if white adds a move at M11. This also helps cut the weak stones at M15 from the K11, especially if white plans to cut at M11.] ;W[pr]C[0.79 0.21 0.00 2.7 v=732] ;B[or]C[0.68 0.32 0.00 1.8 v=1324] ;W[qs]TR[pr][qr][ns][os][ps][qs][rs][ss]C[0.81 0.19 0.00 2.9 v=1106 the only move other than forcing moves like B2. Taking it at face value, it is changes the status of the triangle points. O1 and T1 haven't completely switched sides, but even so that makes this move around 6x2+2=14 points in area scoring. In territory scoring, white's territory increases by 6 points and black's increases by nearly 7. The discrepancy to area scoring is due to needing to add 2 to the difference in order to account for each side adding an extra stone locally. This is 12-13 points in pure territory. This is far bigger than your normal 6 point 2nd line move. In addition the above count certainly unestimates the thinness in both side's positions after the opponent takes the ponnuki, since this pretty much means that Q6 has also changed hands however many points that is worth as well as the situation at N6 which will be demonstrated shortly in the game. ] (;B[ca]C[0.71 0.29 0.00 2.0 v=1677 big as black is thin here, normally this leads to a 5 point follow up in the corner and prevents F18 as well, which in the worst case leads to a swing of 15 points due to ko possibilities. However the problem is that even after black plays here, black's area still has problems.] ;W[ba]C[0.81 0.19 0.00 2.9 v=1118] ;B[da]C[0.71 0.29 0.00 2.1 v=1239] ;W[ar]C[0.80 0.20 0.00 2.5 v=1079 locally the only move due to B1. But this is always forcing so there was no need to play it immediately especially when the M11 ko remains a threat. Without B1, white could atari D2 and black would be in deep trouble locally seki but dead as G6 is locally dead. However, after B1, white needs to play 2 extra moves in the corner to approach giving black time to even pretty much make an eye on the side. Even with B1, black only wins by one liberty. Although white has G1 which often gets an extra liberty, it only works if white can play repeatedly on the first line. In this case black will be able to block at D1 or E1 after which black has a permanent approach liberty at C1 where white needs to play A2 and A1 to start. Black also has a mutual approach liberty with white at E3 or F3 which neither side can play. White doesn't have that on the other side as H4 and H1 are solidly under black control, so white is doomed to start approaching at A2, A1 first. ] (;B[fs]C[0.77 0.23 0.00 3.0 v=999 A nice tesuji to get pretty much an eye locally and claim both D1 and E1. Now black wins the capture race with E9 by 3 liberties (though only 1 compared to the F1 group).] (;W[as]C[0.80 0.20 0.00 2.5 v=1173] (;B[ds]C[0.77 0.23 0.00 2.5 v=1365 k=5 Analysis doesn't like this move (at k=9 and dropping from 67% to 56%, 0.8 to 0.3), preferring D2. It might claim an extra 1 1/3 point in reverse sente locally compared to shorting a liberty, but if white took that sente, white loses out around G9. But it seems to simply be because it takes the engine a long time to find the L13 move because the move it prefers reverts to this. This does seem admittedly simply worse locally as well as bad shape compared to D2. This is because in this position white can delay the option of play C1 for D2 and leave it as a massive ko threat. In contrast, if there is a massive ko, and black has played D2, black has the option of backing off with H1 instead of D1 if white takes C1. If white doesn't take C1, black preserves the option of playing C1 themselves for 2/3 point more. Normally C1 is forcing for all kos, but there is the slightly possibility that it won't be and black will settle for a squeeze after losing this 70 point capture race. However, in this position, white has an even bigger threat than C1, which is at E3, which threatens to cleanly capture everything at D2. This is unlikely to be used since black responding means white loses several points and a critical liberty. However all these small losses can add up when there is no possible gain.] (;W[if]TR[oc][sc][ne][hf][ig][kg][li][nj][ol][nr]C[0.79 0.21 0.00 2.1 v=614 A tricky move but fairly soon it evaluates the position as jigo. At this point it still prefers things white is ahead so what went wrong. At this point it thinks white has a lot of options.] (;B[jg]C[0.75 0.25 0.00 2.0 v=781 k=7 This is a tough move to play correctly as it says another other move will likely lose. This prevents white from cutting M11 since white M12 threatens O15, but it also doesn't mind white playing O15 since this move threatens J13 too. At this point the analysis engine still leans towards a white win, but afterwards says jigo. With 500k playouts, however, it says hard to call, W+1 here.] (;W[kh]C[0.78 0.22 0.00 2.1 v=760 As there are several options for this move, it could be a mistake. After this it seems that jigo is simply best play.] (;B[nj]C[0.76 0.24 0.00 2.1 v=757 Necessary for life and a big move in itself, mostly saving the M10 stones (though white couldn't yet capture them due to the M12 double threat, this removes threats and creates threats).] ;W[lc]C[0.78 0.22 0.00 2.1 v=789 Apparently this atari is necessary for the following sequence because it won't be forcing once black gets O17 or O15. The following 2 white moves gain less than one point in double sente since black playing O17 is forcing N13. However note the possibility that this loses a point if white gets to play J13 and K15 and N13. This is pretty unlikely but still nonzero.] ;B[kd]C[0.76 0.24 0.00 1.9 v=767] ;W[oc]C[0.79 0.21 0.00 2.0 v=918] ;B[ob]C[0.75 0.25 0.00 1.8 v=1211 around 27 points in reverse sente.] ;W[nr]TR[dc][nc][sc][ig]C[0.77 0.23 0.00 1.9 v=1187 White's winning percentage dips after this sequence. However after more playouts (600k total on mainly on triangles), this move returns as the best move since the evaluations for other moves have dropped. Currently white has 62.0 1m 0.5] ;B[os]C[0.73 0.27 0.00 1.7 v=721] ;W[nq]C[0.76 0.24 0.00 1.7 v=816] ;B[oq]C[0.72 0.28 0.00 1.6 v=1030] ;W[mp]C[0.73 0.27 0.00 1.5 v=1072] ;B[mq]C[0.71 0.29 0.00 1.5 v=1045] ;W[op]C[0.72 0.28 0.00 1.4 v=1160] ;B[oo]C[0.70 0.30 0.00 1.4 v=922] ;W[no]C[0.71 0.29 0.00 1.4 v=1211] ;B[op]C[0.70 0.30 0.00 1.3 v=1153 This ko has a swing of over 30 points and isn't worth fighting. White has gained sente, black has 8 points up to M2 and M1 (recall the capture at P4). White has gained 2 of Q6, Q1, M4 as forcing moves in the future. This seem to be worth around 1 point each so this is a count of 6 points. If black descended at P1 before, then black would have around 7 points here but white would have both the Q6 and N6 cutting points. Was it really worth it though?] ;W[ig]TR[ji]C[0.70 0.30 0.00 1.2 v=1303 Around 6 point sente. Compare if black played H13 or J13.] ;B[ie]TR[fc][hc][ic][gd][hd][id][ge][je]C[0.68 0.32 0.00 1.2 v=1077 Preventing H15 is worth around 7.5 points reverse sente as well as many ko threats.] (;W[nc]C[0.70 0.30 0.00 1.2 v=1136 this gains a point in sente which is simply a pure 1 point since the white group this is connected to is very alive and so are the black groups. In the game black added a move locally, so this surely gained one point compared to black connecting at O15. The only difference is that white one less liberty with the K11 group after white defends at N13, but this doesn't seem very significant as white seems to have enough liberties for now.] ;B[mb]C[0.67 0.33 0.00 1.1 v=1430] (;W[bc]C[0.69 0.31 0.00 1.1 v=1467 This is at least 4 points locally as well as setting up D17 sente. If black has enough ko threats, black will be able to hane at A17 after capturing B18. Making life solves the C16 cut and D14 cut, gaining at least one more point. Finally making life sets up very nice tesuji inside black's top side. Overall this seems to be at least 6 points. It doesn't seem realistic for white to fight the fairly heavy ko with A18 in response to B18 when white is comparatively thin on the board.] (;B[ne]C[0.66 0.34 0.00 1.0 v=1424 This has 99% of playouts by the analysis engine but I have some doubts. The policy network gives it 14.5%, just edging it ahead of S11 at 14.0% which it finds to be a bad move (worse than S12) with more playouts. Saving 2 points, threatening nearly 6 points (nearly since white N13 threatens the O12 point too. This also makes white thin at S12 leading to many more ko threats though not necessarily many more points since black could have played there anyway.] (;W[rh]C[0.67 0.33 0.00 1.0 v=1051 There were several alternatives to this solid move from T17 to N9 to K6 etc. This secures a connection which threatens to make points around R8. Otherwise black disconnecting could remove 3 points (assuming S10 will connect to S6) as well as potential around T9 since black could get T11 at the end of the cutting sequence. Assuming the right side is white territory, this loses up to 3 points in sente with possibly more follow ups to reduce around R8. However, the right side is still not necessarily where white aims to make territory. In addition black still has 2 big ko threats. Though admittedly, black had may be 5 ko threats here before this move.] (;B[kn]C[0.62 0.38 0.00 0.7 v=741 The M14 is a big move around 6 points. However, this move prevents white making 8 points locally with K6 (note that M4 is sente and there is the P8 cut) This should be counted as less than 8 points because black K5 sente makes 1/2 point at J4 and black N2 prevents a squeeze at Q1 and black P8 sente gains nearly 2 points on the right. This brings the count down towards 4 points. However, the key thing is that white connecting so solidly at K6 changes the nature of the N9 ko. After black's move, white's whole centre group is somewhat cut off and owes a move. This move is pretty much sente, which explain why it is so big.] (;W[jk]TR[dc][sc][ed][fd][ee][he][df][hf][lf][mg][li][ik][mk][pk][nl][ol][pl][om][jn][ln][pn][ko][mr][cs][ps]C[0.63 0.37 0.00 0.7 v=630 Probably necessary to become thick before starting the ko since black pushing here is too much. Still there were a lot of options for this move, marked as triangles.] ;B[mk]C[0.60 0.40 0.00 0.6 v=905] ;W[sc]C[0.62 0.38 0.00 0.6 v=865 This ko threat loses almost nothing since even if black plays twice in this area white has lost almost nothing compared to if black played here first.] ;B[sb]C[0.59 0.41 0.00 0.5 v=786 The ko has a swing of over 10 points but blocking here is worth around 6 2/3 points in reverse sente. This is forced. At least if white connects T16, that is no longer a ko threat as the follow up is gote.] ;W[mj]C[0.61 0.39 0.00 0.5 v=937] ;B[ri]C[0.60 0.40 0.00 0.6 v=1337 A move that creates at least one follow up threat at T12.] ;W[si]C[0.59 0.41 0.00 0.4 v=1296] ;B[mk]C[0.58 0.42 0.00 0.5 v=1591] ;W[dc]C[0.59 0.41 0.00 0.5 v=1782] (;B[eb]C[0.56 0.44 0.00 0.4 v=877 A good move to save 2 ko threats and proper locally.] (;W[mj]C[0.60 0.40 0.00 0.5 v=1173] (;B[lf]C[0.54 0.46 0.00 0.3 v=1430 black can't afford to let these stones escape] ;W[mg]C[0.52 0.48 0.00 0.1 v=612 big since the main threat is to connect 6 points at N15 which also increases the threats of S18 and Q18 as well as remove one point at O12 if white wins the N9 ko.] ;B[mf]C[0.53 0.47 0.00 0.3 v=769] (;W[li]C[0.52 0.48 0.00 0.1 v=758 This becomes the top move at 40k playouts going up to 44% with 150k playouts. and -0.1 score. However, it seems white had a simpler way to play that also leads to jigo very quickly. Perhaps white wanted to show off their tesuji power at the top. This spends an extra move capturing 6 points. And since white has a nice sacrifice at the top which is both fairly good endgame (losing only 0.25 to normal play) and creates a lot of ko threats, white is acting a bit like ko monster, trying to gain more than 6 points from winning this ko by playing endgame moves at the same time as playing the ko. After all white spending a move at T16 is already a sente 3 points which is around 6 points. And a white move around Q8 also is at least something like 2 points in reverse sente. It is smaller than it looks since it helps black connect at P8 which saves O7. Probably white merely aims to play the best endgame but ends up giving up quite a bit to do so. Still there are several variations where if black doesn't play perfectly such as hanging on to a last ko until after all dame has been filled, then white will win. Also, making an eye for the centre group permanently gains the point at D12/C11 so that the D12 cut becomes completely useless. This could make a difference of up to 2 points but actually only gains much less than 1 point since white has eyespace floating around at J8 and H13 from the squeeze.] ;B[mk]C[0.46 0.54 0.00 -0.1 v=666] ;W[fd]C[0.40 0.60 0.00 -0.4 v=781 A superb sacrifice strategy starts here. Normal endgame of pushing E16 doesn't lead to many ko threats and also allows black only 2.5 points less than the game. The squeeze in the game leads to 8 ko threats in total as well as recuperating around 2.25 points from the squeeze. Getting D14 from the squeeze also prevents around 8 ko threats by black at the same time!] ;B[ed]C[0.38 0.62 0.00 -0.5 v=654] ;W[mj]C[0.39 0.61 0.00 -0.5 v=868] ;B[lp]C[0.37 0.63 0.00 -0.6 v=1182 a fine ko threat even though black has to go back to defend since it creates a cutting point since white must connect N4 in order to make use of N2, and after black plays at N2 to defend, this saves black from the Q6 and Q1 squeeze somewhat. The cutting point at L5 is quite severe, worth nearly 8 points.] ;W[mo]C[0.36 0.64 0.00 -0.6 v=952] ;B[mr]C[0.37 0.63 0.00 -0.6 v=1632] (;W[lh]C[0.37 0.63 0.00 -0.6 v=1417] ;B[mk]C[0.38 0.62 0.00 -0.7 v=1183] ;W[ee]C[0.40 0.60 0.00 -0.4 v=1380] (;B[fe]C[0.42 0.58 0.00 -0.4 v=1326] ;W[mj]C[0.42 0.58 0.00 -0.3 v=1880] (;B[jm]C[0.39 0.61 0.00 -0.5 v=1908 not a very happy ko threat to play, but given that white L5 forces this move, it loses almost nothing, merely that it is committing to letting white connect L5 later since this allows white an extra 2 points or so around J8.] ;W[ik]TR[gk][hk][fl]C[0.34 0.66 0.00 -0.7 v=601 bad shape but actually a good move locally, better than H9 for example since H9 gets negative points if black plays first with J9 J8 and G9 atari. The point is that every white move here shorts black's liberties and white remaining connected is more helpful. The analysis engine can't tell any difference between the triangle liberties though this move is marginally preferred. The other liberties don't manage to defend the J8 cut since they are too low on liberties and not blocking the escape route for J8] ;B[mk]C[0.38 0.62 0.00 -0.5 v=1035] ;W[he]C[0.37 0.63 0.00 -0.5 v=1071 A brilliant sacrifice, perfect locally and perfect for the ko.] (;B[hf]C[0.41 0.59 0.00 -0.2 v=1035] ;W[mj]C[0.40 0.60 0.00 -0.3 v=1210] ;B[nh]C[0.44 0.56 0.00 -0.1 v=1506 No point to waste a ko threat like T12 fighting a ko that isn't yet fightable when white has lots of ko threats at J16 as shown in the game. Importantly, neither side will play around H9 since black is leading the capture race by two liberties there at the moment, even though both sides will show a nice endgame technique there later after the ko is settled.] ;W[ko]C[0.44 0.56 0.00 -0.1 v=1590 around 8 points locally (slightly less due to white's L7 atari potential). Probably slightly less since black connection threatens to make J4 a point now whereas if black cut here, white could use the L7 atari and M6 follow up. This is nearly as big as the swing of the ko (around 10 points) so it almost certainly bigger (since the normal estimate to compare the value of a move is to divide by 2 for a double gote and divide by 3 for a simple ko and 8/2 > 10/3) However due to the Q6 push and cut this move is probably more than a simple 8 points of territory since it gives white enough liberties to block Q6 with Q7 and extend Q8 in response to P7 cut, making Q6 a pretty bad move rather than an efficient way to break into white's S8 area as well as prevent being force to play O1 in response to Q1.] (;B[mk]C[0.46 0.54 0.00 -0.0 v=1564] (;W[id]C[0.47 0.53 0.00 -0.0 v=1588 Excellent move. This threatens to play G16 and capture around 30 points. At first sight, It loses around 3 points locally in which case it would seem questionable whether it is worth it to win the ko. This is because white could play K14 as a forcing move and tne play K15 atari. However, in this situation K14 forces L13 which is actually a valuable move in this game, not pure dame (around 1/2 point due to K12 atari). However the endgame situation is that black can play H13 and K14 as atari in most lines, so this is actually making black's shape very bad for zero loss. More importantly this move playing from inside gains a liberty for cutting off the F15 6 stones. This allows for a squeeze which gains not just many points on the outside but also makes white much thicker. As we will see, making D14 sente recuperates 1point loss, H13 sente recuperates 1.75 pt and D14's thickness also saves 2 big ko threats which we will see may be important in the rest of the game. In addition playing like this makes 5 big ko threats locally (worth over 20 points), so it is more than worth it in order to win the N10 ko.] ;B[je]C[0.48 0.52 0.00 0.1 v=1514] (;W[gd]C[0.47 0.53 0.00 -0.0 v=2100 Why waste a threat?] ;B[ge]C[0.50 0.50 0.00 0.1 v=1773] (;W[hd]C[0.46 0.54 0.00 -0.1 v=1184] ;B[ic]C[0.47 0.53 0.00 -0.0 v=981] ;W[mj]C[0.43 0.57 0.00 -0.2 v=1567 kg40s994 is still not that confident, but it seems very likely to be jigo here. white has 31.0 1.4m -0.3] (;B[om]C[0.42 0.58 0.00 -0.1 v=1526 much bigger than it looks, saving the stone and threatening Q6 in sente. This is pretty bad shape, risking P8, but in the game black handles it enough so that it doesn't lose any points. This is in part because white doesn't dare to play for points in response and must connect solidly in the most defensive way.] (;W[pn]C[0.37 0.63 0.00 -0.4 v=1267 Good move though not that easy to find (it takes 300k playouts). It doesn't actually set up any ko threat at Q1 since white's shape is weak at S4, but it does remove black's ko threats here and ensures that black will play O1 later. This gains the point at R2 as well. The ko threat difference is critical. This move also sets up the P8 wedge, greatly limiting black's options for the R8 area even without adding a move there.] ;B[mk]C[0.39 0.61 0.00 -0.2 v=1517] ;W[hg]C[0.35 0.65 0.00 -0.5 v=1620 Simply wasting a ko threat if white doesn't follow up. Since T16 is easily the biggest move, so neither side needs to play here locally.] (;B[fc]C[0.34 0.66 0.00 -0.5 v=1025 k=7] (;W[cs]C[0.34 0.66 0.00 -0.5 v=723 k=9 extremely questionable to lose a 80 point ko threat. Black has two big ko threats at T12 and S14 and white has two ko threats at D14 and C1. This is even. So white wins the ko if white took the ko. S18 is theoretically a ko threat, but it loses at least 3 points there so it doesn't help. Q1 is a 32 point ko threat (area scoring) but leads to the S4 move and S1 as counter ko threats. S4 is around 20 points, S1 around 11. So white must never play Q1. O1 is a ko threat, but it loses 2 points there. But after this, white should lose the game.] ;B[dr]C[0.27 0.73 0.00 -0.8 v=679 well over 70 points, black must respond. This keeps everything connected, with the D2 group, so with the eye at E1, allows black to eventually fill one of E3 or F3 without worry. Since the second move at E3 or F3 will eventually capture the white chain G5, black has no issues now.] (;W[mj]C[0.27 0.73 0.00 -0.7 v=766 k=9 (i.e. B+9 on the board or B+2 overall)] (;B[sd]C[0.27 0.73 0.00 -0.8 v=1235 k=7 (jigo) Losing move says kg40s994. Questionable as the ko is so big. This move takes at least 3 points locally in reverse sente as well as removing a ko threat. This sounds to be of similar size to the ko, but it is late enough in the game that there isn't enough endgame to compensate for losing the ko. On the positive side, there are some traps of squeezing white later using T15 for another potential 3 points if white misplays the R8 area.] (;W[mk]C[0.25 0.75 0.00 -0.7 v=1178 Captures at least 6 points and prevents K12 from being sente, so it sets up the K12 follow up increasing white's territory at H12 from 0.5 to 2 points and adding one point at L11. This is worth (3-0.5)/2=1.25 (halved as only a follow up). This also sets up the P8 follow up, which would capture 4 points. To prevent it in the game, black gave up 1 pt. If black spent 2 moves to win the ko with N9 and N10 that captures 1 stone for 1pt. Hence, I estimate the swing of this ko as worth around 6+1.25+1+1 or nearly 9 points. This takes 3 moves to settle, so the value of a move is around 3 points (or as much as a 6 point gote), which is very large at this stage. The next largest move other than the unsettled R8 area might be just the 2.5 pt move at K12. So if losing the ko loses 7 points locally, getting K12 and P9 in return gains 2+3=5, which is a loss of 2 points overall.] (;B[ql]C[0.28 0.72 0.00 -0.6 v=1425 A cool and tricky move that gains 4 points in sente. The situation is that there is a weak point at P8 but white is reluctant to play it because black breaks white's territory down to S9, as well as setting up a potential harassment of white's group (at the least T12 will have to be connected, losing a point. The shape isn't completely connected together yet but at least this can't be cut off from P9 (though P7 might be) This is nearly 2 points better than playing Q8, but only because white hasn't managed to exchange T15 for T17 yet. This is perfect timing. Black's shape is weak and P7 can be captured, but the S10 group as a whole isn't 100% alive either, losing white around 3 points when black starts harassing (with T15, spending a move) in addition to the loss of around 5 points of territory around S8, despite capturing 6 points at P7. Overall white loses 3+5-6=2 points if white fights back to capture P7.] (;W[rl]C[0.25 0.75 0.00 -0.6 v=1360] (;B[pl]C[0.27 0.73 0.00 -0.6 v=841 Approximately 4 points sente. Shape: A nice move to both permanently defend the P8 cutting point as well as permanently connect R8, ensuring black has a threat of playing S7 next to gain around 2 points in sente. And ensuring white has no threats. Later white can play O8 or Q7 as potential threats but they are empty triangles which are completely pointless.] (;W[qm]C[0.25 0.75 0.00 -0.6 v=1113 Better shape to get black to force with R9 since there will be no ko threat against white. Whereas if white aims to play R9, there will be one more ko threat at S9 later. Basically Q10, R10 are clearly weaker than R6. S10 is also vulnerable since it is adjacent to S11 and T10 is empty. However "weaker" doesn't mean you should play there immediately!! Since the best points move is not the same as the best defence move. Playing S9, the good move for defence will lose a point for no reason since S9 is already surrounded by solid white stones, so R9 is the good points move, but that leaves one ko threat since it still doesn't connect the S10 chain to a supporting group. R9 is an empty triangle with a double weakness. (the S9 and T10 cutting points).] ;B[qk]C[0.24 0.76 0.00 -0.7 v=635] ;W[rk]C[0.22 0.78 0.00 -0.6 v=671 reverse sente 2 points at T9 and T10 and also 2 ko threats (T8, S7)] (;B[hk]C[0.27 0.73 0.00 -0.6 v=969 This is good shape to punish J7 (making it harder for white to play H8 to fill black's liberties), but it doesn't actually succeed at that in this case.] (;W[hl]C[0.23 0.77 0.00 -0.6 v=1118 Not that easy to find this move, but it is worth 4 points. Big to respond where black is weak even though black seems to gain in sente. Black actually doesn't really gain much in this game with the J9 move if black doesn't follow up since the space around G8 is negative for black - if black plays there they lose liberties since the space is adjacent to black's stones and makes bulky shape - if white plays there, every most is forcing on the capturing race with G5, so they can make an eye. Why is this empty triangle shape great? - It cuts off black from J8 and G8 simultaneously, both of which were empty spaces and good for liberties from black's point of view, makes H9 a liberty losing empty triangle. - G8 is significant since black is only connected to F7, F8, F9 and G9 which it was already connected to. This ensures black can't make more liberties. - Black is weak enough that black can no longer make an eye here without losing a critical number of liberties. In all the space remaining, G8 remains cut off from black, so white can make it an eye. __ Just locally, black's move is excellent since it forces white to play all the moves F8, F7, F9, G9 immediately, settling the position, rather than letting white leave them as ko threats. That would be the situation after white plays H9 (which takes 3 points locally with G9, F9, F8, F7 all ko threats). It was smart of black to pay for this by losing a liberty only after the N10 ko has been settled or else white would have 4 ko threats each worth 70 points and this would surely make white komonster, able to finish the endgame without needing to spend a move ending the N10 ko. Also, there is no meaningful ko that white can use these ko threats for now as T15 is pretty much always forcing on T17, and the B1 is smaller than the local F8 moves, so will only come into play after this local area is settled.] (;B[jh]C[0.27 0.73 0.00 -0.6 v=1375 2.5 points gote + big ko threat at L13 1 point from the L11 cutting point (working with N11 and M10)] ;W[fl]C[0.23 0.77 0.00 -0.5 v=1424 ] (;B[hs]C[0.28 0.72 0.00 -0.6 v=1306 It is good practice to fill in white's liberties from the outside since they will be valuable even if black loses the capturing race. ] ;W[fk]C[0.23 0.77 0.00 -0.6 v=943] (;B[hp]C[0.29 0.71 0.00 -0.6 v=770 This move will threaten another eye at K4 or at least a last dame ko, so it is valuable.] ;W[fm]C[0.23 0.77 0.00 -0.6 v=721] ;B[eq]C[0.29 0.71 0.00 -0.6 v=743] (;W[gk]C[0.24 0.76 0.00 -0.5 v=722 This makes 2 points in sente, so is certainly the biggest move. Surrounding G8 makes an eye, which makes J8 a point of territory as well. Otherwise if black played here, then white couldn't play G8 without it being self-atari, so white is forced to pull back at J8 first. Being able to take G8 a territory without playing there (which would be self-atari) is very important. Under stone scoring this means white would have time later to play J8 at their leisure.] ;B[fq]C[0.28 0.72 0.00 -0.7 v=848 70 1/3 points move C6 to D5 D1 to G6 is 4*6=24+ 1+10=35 stones, double that and add the C1 ko.] (;W[df]C[0.24 0.76 0.00 -0.5 v=1028 A nice solid move, removing ko threats in sente.] ;B[hc]C[0.28 0.72 0.00 -0.6 v=1035 Forced, worth 24.5 points in sente] ;W[se]C[0.25 0.75 0.00 -0.5 v=1371] ;B[sc]C[0.29 0.71 0.00 -0.6 v=1124 This ko clearly dominates the B1 ko since white T17 threatens to cut T18 off by S18 which will threaten Q18 and so on. Hence it is pointless for black to try to fight back by playing B1. There is only the B1 endgame, T17 and G13 endgames left. If black ends this ko, the last two endgames are miai, so black has no reason to try to exploit the T17 ko for more profit (a la Spight's komonster theory) even though black has a few more ko threats.] ;W[ps]C[0.26 0.74 0.00 -0.4 v=1449 Doesn't mean much since O1 and Q1 are miai but it is a ko threat for both sides. Black playing O1 threatens a 1 point sente follow up here. White playing Q1 threatens a 32 point sente follow up here. However, this late in the game, 1 point sente is bigger than any other move and should be played immediately. So black playing O1 is as forcing as white playing Q1. Hence this is basically symmetric (except as ko threats) and 1 point sente = 32 point sente = infinity point sente at this point. __ This normal when one side (black here) has won a capturing race. Black's outside liberties (Q1 here) are under white's favour since black has both capture (O1) and play there in order to take them, while white can play them in sente.] ;B[ns]C[0.28 0.72 0.00 -0.6 v=855 worth 32 points.] (;W[bs]C[0.26 0.74 0.00 -0.4 v=1125 This connection makes sense when black has many more ko threats. This sort of final 1/3 point ko has 4 possible states (white end, white capture, black capture, black end) and the difference in territory scoring is 1 point. So among the 3 moves it takes to traverse from one state to the next, we estimate the score difference as 1/3 pt For example if this white end is counted as zero points, then from black's perspective, the 4 states are worth around 0 1/3 2/3 1 points on average. The swing of a move, the difference between white's play and black's play is then 2/3. ___ From area scoring's perspective, 3 moves normally take 3 points (3 stones) unless they surround territory. As one captures occured during traversal in one direction there is only 3+1=4 point difference in score. This adds up since the fight is over 2 intersections B1 and C1 which is 2x2=4 points in stone scoring.] (;B[gg]C[0.30 0.70 0.00 -0.5 v=1090 Skilful! Black wins tedomari before dame. This is the last point worth anything. We will see that there are an odd number of dame remaining (23) so white wins tedomari of dame. ] ;W[jp]C[0.28 0.72 0.00 -0.3 v=836 Black couldn't make a point at J4 since K5 is forcing. Still this is solid and good practice since white doesn't want to exploit the K6 weakness with the K5 bad shape move. White might be able to exploit it with L7 later and this will force black to remember to play K6 themselves before filling in the rest of the dame. I find it weird that at 7 komi black still has 70% winning percentage (B+0.3) while at 7.5 komi, white has 99% winning percentage (W+0.1) when the game is a draw at 7 komi (black wins by 7 on the board).] ;B[mm]C[0.30 0.70 0.00 -0.5 v=903 White also couldn't make anything here, but good practice to remove the possibility of white playing two moves in a row to get 2 points here. (N7 and L7)] ;W[ln]C[0.27 0.73 0.00 -0.4 v=681 Cut when you (think you) are behind? This also completely gets rid of the K6 and P8 potential eyes (under optimal play), and sets up ko threats since K7 and L6 are cut off (by threatening to play K6, either with K5 or L7 and making it more difficult for black to play L7)] (;B[jo]C[0.31 0.69 0.00 -0.4 v=684 Black is going a bit crazy playing risky moves. It can be called immaturity to create more problems for yourself later. The only possible benefit if white plays well is to get both M7 and L7 and make an eye. However, this is giving white a ko threat and also making it impossible for black to play L7 themselves despite spending a move here. Perhaps there is a meta benefit to challenge yourself and not make mistakes even when there is opportunity to?] ;W[lm]C[0.28 0.72 0.00 -0.4 v=606 Solid play has benefits: - These white groups have a zillion liberties - Black has no ko threats. - White has ko threats such as L7 or O8 - Black might commit self-atari with L7 (or O8 later) In this game these sum to zero points (with optimal play) but still good instinct.] ;B[pm]C[0.30 0.70 0.00 -0.5 v=613 Good for threatening to get an eye, trying to avoid connecting at P8 even though realistically black will have to.] ;W[jf]C[0.29 0.71 0.00 -0.4 v=606 Probably the most sneaky way to set up a potential eye at H12 later, or a threat with L11. Probably a good habit when behind, though this is a draw.] ;B[mn]C[0.31 0.69 0.00 -0.5 v=601 black seems to be in denial that this is a draw, still dreaming of making an eye at P8 or K6 or capturing F12 or M12? Or white ignoring B18 or S4 or S1? This is playing instinctively risky shape. It is true that black's position is much thicker (black only has the Q18 ko threat to worry about), but there is no ko, so it makes no difference.] (;W[nl]C[0.34 0.66 0.00 -0.2 v=620 Very clumpy shape but the simplest way to prevent an eye and set up an atari as a follow up.] ;B[ip]C[0.37 0.63 0.00 -0.3 v=634 This black player seems to enjoy making terrible shapes during dame filling, prioritising dame which are already surrounded by 3 black stones. Perhaps this is good instinct when those 3 stones aren't connected (then K4 becomes a "peep" move). However, it is often better to ask white to play there if black is solid since that loses white liberties. Perhaps the winning percentage tells the story. Black thinks they are leading, so they are happy to secure such weak points. Still from a perspective of generalisable strategy perfection, I would recommend connecting cutting points first such as Q9 or even K3. This helps prevent future mistakes. This may even gain points if there is a ko.] ;W[pk]C[0.42 0.58 0.00 -0.0 v=665 white tries to scam a win. Perhaps this atari is chosen instead of P8 since P8 is horribly bulky shape with the N9 clump, so this is more instinctive. Alternatively, it looks more normal, so black is more likely to not notice the threat?] (;B[ol]C[0.45 0.55 0.00 -0.1 v=605 black notices ] ;W[gh]C[0.47 0.53 0.00 0.1 v=907 Connecting white's weakest stones, preventing an atari and taking away black's largest ko threat. Also almost the last possible way for white to scam a win.] ;B[ih]C[0.46 0.54 0.00 -0.0 v=1023 White last move has cut G13 from G12 and hence H12, so this threatens to make H12 a point. Black only needs L13 to connect back and white can't cut there yet as that would be self atari. Furthermore, H13 is in atari so needs connecting now.] ;W[hh]C[0.49 0.51 0.00 0.1 v=947] ;B[kg]C[0.47 0.53 0.00 -0.0 v=1111] ;W[ki]C[0.49 0.51 0.00 0.1 v=1489] ;B[kc]C[0.47 0.53 0.00 -0.0 v=1673 From a simplicity perspective I can't recommend playing here. In territory score it would lose a point, and under area scoring, it only doesn't lose a point if there are an even number of neutral points left (which are all miai, so there is no need to fight to play there first). This is normally called "showing off". There are 6 neutral points left. This shows white has won tedomari of the dame so only white needs to keep focused and keep playing dame. Black can play elsewhere if they want, seemingly losing a point but gaining it back through becoming the winner of the dame tedomari. Winning tedomari of dame is only helpful in area scoring whereas in comparison it is a bad sign in territory scoring where neutral points don't count towards score. This can be the margin to tip the score in some cases. I suspect this is why the AI analysis keeps saying that white might be behind when it is jigo, because the AI is designed to play territory scoring (and button go) as well as area scoring.] ;W[jq]C[0.49 0.51 0.00 0.1 v=709 Quite a neutral point, but it is good for white to get this since J3 and K2 are not connected. Neither black group is pass-alive either.] ;B[pi]C[0.50 0.50 0.00 -0.0 v=654 A neutral dame point, but one that instinctly feels under black's control more, surrounded by 3 connected black stones. If white Q10 was weaker, this would make a difference and white wouldn't want to play this dame.] ;W[fi]C[0.50 0.50 0.00 0.2 v=644 ] ;B[jn]C[0.50 0.50 0.00 -0.0 v=648 Black doesn't fall for white's trap. White has left this dame till last to tempt black to play L7 (it does happen rarely in pro games, though probably not AI) and commit self-atari.] ;W[km]TR[kl][lm]SQ[jm][kn]C[0.51 0.49 0.00 0.1 v=632 Completely neutral dame point since the white stones adjacent (triangles) are connected already and so are the black stones adjacent (squares). Both groups also have many liberties.] ;B[oe]C[0.50 0.50 0.00 -0.0 v=707 Very neutral dame point, but white's group only has 3 liberties, so perhaps white was reluctant to play here.] ;W[bl]C[0.50 0.50 0.00 0.2 v=660] (;B[gr]C[0.51 0.49 0.00 -0.0 v=680 ? This is a vital point for G1 (i.e. it dominates G1 so it is often a key move to get both G1 as eyespace and maximise G3's eyespace) but still completely pointless.] ;W[bf]C[0.50 0.50 0.00 0.1 v=618 This move clamps down on the black stone which can get the most liberties (black here would get 4 liberties whereas the other two dead black chains could only get 2 liberties).] ;B[lb]C[0.50 0.50 0.00 -0.1 v=614 Very unnecessary, but the last dead white stone.] ;W[ag]C[0.50 0.50 0.00 0.1 v=621 This captures a cutting stone which cut two single white stones which each only had 3 liberties.] ;B[]C[0.50 0.50 0.00 -0.1 v=619] ;W[am]C[0.50 0.50 0.00 0.1 v=637 Capturing dead stones is reminiscent of Tromp-Taylor rules. But this is also a cutting stone with two groups that are not pass-alive, so it could also be viewed as minimising potential mistakes in the future, even if very small chance.] ;B[]C[0.50 0.50 0.00 -0.1 v=625] ;W[ci]C[0.50 0.50 0.00 0.1 v=646] ;B[]C[0.50 0.50 0.00 -0.1 v=643] ;W[dh]C[0.50 0.50 0.00 0.0 v=665 both groups are very alive and the centre group was pass-alive but they now have a zillion liberties even if white fills in the eyes. Also, cutting stones were here just a few moves ago (Alphago mimics have a standard memory of 8 moves previous).] ;B[]C[0.50 0.50 0.00 -0.1 v=621] ;W[]TW[aa][ab][bb][ac][ad][bd][cd][ae][be][ce][qe][af][cf][rf][sf][bg][sg][ah][ch][sh][ai][bi][aj][bj][cj][kj][lj][sj][ak][bk][kk][sk][al][cl][gl][il][sl][bm][rm][sm][an][sn][ao][so][ap][bp][rp][sp][aq][rq][sq][qr][sr][rs][ss]TB[ea][fa][ga][ha][ia][ja][ka][la][ma][na][oa][pa][qa][ra][sa][fb][gb][hb][ib][jb][kb][pb][qb][rb][jc][lc][fd][gd][hd][id][md][he][le][me][og][pg][ni][go][ep][fp][gp][gq][nq][ir][lr][nr][es][gs][is][js][ks][ls][ms]C[0.50 0.50 0.00 0.0 v=652 result=0 The final score was a jigo. In area scoring, this is only possible for odd integer komi (k=5,7,9 etc) in area scoring if sekis are ignored since the difference on the board of black and white's stones must be odd, since their sum is 361 which is odd. i.e. A+B has the same parity as A-B since their difference is 2B which is even.]) (;B[gp]C[This is the move to make everything pass alive] ;W[bf] ;B[ls] ;W[] ;B[is])) (;B[kg]C[Both are 16 points. Cute!])) (;W[pk]C[KG40s943 prefers here.])) (;B[lm]C[It is telling that black doesn't cut through. This would be quite bad instinct when you think you are ahead.] ;W[km] ;B[jn]C[Black is still cut off, both white groups are strong and black loses the initiative.]) (;B[jn]C[This is safer, worth a move and allows black to play L7 later.])) (;B[jp]C[At 7.5 komi, it suggests this cool move. The idea is to trick white into taking out this eye.] (;W[jo] ;B[jn] ;W[jq] ;B[gg] ;W[ih] ;B[gh] ;W[fi]C[This is good for white in Japanese rules since black has to play at J4. However in area scoring, black can make use of their very thick position to hold onto J4 for as long as possible. There are 13 dame points left, so does black have 7 more ko threats than white? Black only has 5 more ko threats (even optimistically including C11), so this is not enough to turn around this game but it certainly makes it more complicated.]) (;W[gg]C[This is the simpler and proper way to play] ;B[ih] ;W[hh] ;B[jn]C[Same score as the game except white is more solid at G13, black is more solid at K6 and white has a forcing move at K3.]))) (;W[gg]C[Technically bigger normally, but in area scoring, the last ko can be worth 4 points if both sides can hold onto it while all the dame is filled.])) (;W[gg]C[no reason to play here though this does make the G12 point. This is because black at D14 is always forcing, but black at G13 is not. i.e. white just needs to stay connected and the bamboo join at H13 is strong so G13 is not a threat, and G13 only removes 1/2 point of territory since black was strong at K12 already, so there aren't many points to be made here. If white had J12 blocked, then G13 and D14 would pretty much be equivalent in points but D14 still seems preferable for solidity to remove threats that B13 has.] ;B[hc] ;W[df]C[This is now the biggest move.] ;B[ih]C[This is 1.5 points but is the same score as T15 reverse sente 1 point. Either way black needs to fight and win the B1 ko but black can since black has many threats.])) (;W[se] ;B[gk] ;W[sc] ;B[rb]C[This probably wouldn't lose white the game, but it is a smaller move. White has only gained 1 2/3 points here while black gains 2 points at G9.])) (;B[fm]C[It goes without saying that this is not helpful] (;W[fi]) (;W[gl]C[Even if white connected, black has lost a liberty at F7 and loses everything. This is sometimes a good technique if white also can't approach at F11, since now white can't approach at G9 (black captures at J8). However, all white needs to win is to fill all black's liberties but one and then play G9 since that captures black first. That means black needs a special two approach liberties like G9 in order to get an extra liberty and recover the loss of F7. It doesn't work in this case, but if it did, this would gain black 2 points (G8 and J8).]))) (;B[eq]C[This is less good instinct even if white didn't have F3, because of the possibility that black loses capturing race (such as due to ko). A move such as E3 will lose a stone if black dies and also has no real potential. Black doesn't have enough liberties to follow up at F3 yet.] ;W[fq]C[this is very bad for black] ;B[bs] ;W[hc] ;B[hs] ;W[hb] ;B[gb] ;W[ib] ;B[fa])) (;B[fl]C[This doesn't help much since it is gote as black loses a liberty. The liberties of this stone include F7, F9 (which were already liberties) as well as G8. Since playing here already loses F8 as a liberty, white merely needs to play G8 for black to have lost a liberty. __ Overall this removes the G8 and J8 points from white (2pts) but takes gote, so it isn't better than K12 which makes white 3 points compared to white's 0.5 point in the game. __ This is a reverse sente 2 points, which is normally bigger than the game move, but with only the C1 ko left to play for, this loses 3 points after white K12 compared to the game, so black has to recover the loss by winning the C1 ko which is troublesome. It turns out black can win it (or make a point at K4), but it is more reading. In the game, black skilfully wins tedomari before dame as well as tedomari during dame, so that recovers any potential loss of the game move.] ;W[gl] ;B[gk]C[More commentary later but making one small eye doesn't help your liberty count. This spends one move to lose one liberty (at G9) since this move is already adjacent to 3 stones (G10, H9, F8) (you need to be adjacent to empty space to get more liberties). But adds an approach move at J8 to the F7 liberty. Since black has an eye this gains a liberty. Still this is an increase of -1+1=0 liberties for the cost of one move. This is like passing during a tense capture race which will lose the race. In a one eye kill no eye situation, an eye have reduce your opponent's liberties drastically since all mutual liberties become your own and not theirs. In this case, there are no mutual liberties between G5 and G6. This loses the capturing race.] ;W[fi] ;B[hs] ;W[il] ;B[hp] ;W[fm])) (;W[jh]C[This loses] ;B[gl] ;W[fm] ;B[fl] ;W[il] ;B[hp]C[black needs to add another move, but black makes two points where white had two points, this is a 4 point move.]) (;W[fm]C[bad] ;B[hl] ;W[il]C[Black has 5 liberties still and white has 2 points less.])) (;B[jh]C[pretty much the same but more easy to play against] ;W[hk] ;B[gk] ;W[gl] ;B[hp] ;W[fl] ;B[hs])) (;W[jh] ;B[rm] ;W[qm] ;B[rk] ;W[sm] ;B[qk] ;W[sk]C[The problem is that S10 has to spend a move filling in W's own territory in order to connect, losing white two points locally. The S7 throw in hasn't lost any points since the S8 stone was weak enough that black got S9, forcing white to capture at S7.])) (;B[qk]C[this is very tempting but bad here] ;W[rk] ;B[qm]C[white is completely connected but black is not] ;W[ol] ;B[pl]C[good move that gains a point by preventing snapback later.] ;W[nl] ;B[pk] ;W[pm] ;B[ln] ;W[mm]) (;B[pm]C[a little riskier] ;W[qm] ;B[qk] ;W[pk] ;B[pl] ;W[rk]C[If N7 is sente, then white might gain a point. Basically the exchange of Q7 for Q9 was very bad for black. Both sides played a move and black lost 2 liberties compared to the game because black was the only one with a weakness here and black didn't play to defend it. Q7 only shorts black's liberties.]) (;B[ns] ;W[pl]C[This is pretty big at this point, around 4 points more than the game locally.])) (;W[qk]C[this also works since it is cutting off R8 from the strong group, but isn't as instinctive due to bad shape (which probably loses a ko threat compared to the game). This isn't properly defending the weakness yet and doesn't capture anything, so black merely has to defend first and you have to defend again. Although R9 is still valuable for defending the S8 stone, there may be a better option. In the game, white played R7 which is better shape. If you defend S8 first, you set up bigger threats later. The move in the game sets up a trap] ;B[pl] ;W[rl]) (;W[pl]C[A cool tesuji and trick move, but it can be refuted.] ;B[pk] (;W[qk] ;B[pm] ;W[rl] ;B[qm]C[Black gets two points from the capture of Q8 as well as white losing two points at S7 and T7 as well as black setting up maybe three ko threats at T9, T6 and S9.]) (;W[ol] ;B[qk] ;W[qm] ;B[sk] ;W[sj] ;B[se] ;W[sf] ;B[sc] ;W[gk] ;B[sm] ;W[sh])) (;W[ol]C[too hasty] ;B[pl] ;W[nl] ;B[pk] ;W[rl] ;B[rm] ;W[rk] ;B[qm] ;W[sm] ;B[pm]C[White loses 2 points in sente (S7 and T7)])) (;B[pk]C[This is the safe move but loses 1 point locally. The AI says it is still a draw though.] (;W[ql] ;B[qk] ;W[rk] ;B[pm] ;W[qm]) (;W[pm] ;B[qk] ;W[rk] ;B[ql] ;W[rl] ;B[mm]C[Black takes gote, but white has one point less here, so this is good for white (normally black should be playing at least a 2 point move.])) (;B[] ;W[pl] ;B[pk] ;W[qk] ;B[ol] ;W[pm]TR[rk][ql][rl][qm]C[This is a 4 points reverse sente compared to the game])) (;W[pk]C[AI says this loses the game. This takes a 3 point reverse sente locally, which is normally of similar size to the ko, but it seems it is very difficult to win the ko when H9 is not a ko threat. There is no big move on the board to give enough compensation for the ko.] ;B[mk] ;W[nl]C[This makes the ko a bit bigger since N9 will be captured too but white will have to spend two more moves to win the ko. Black's response connecting fixes black's problems] ;B[ol] ;W[mj] ;B[pl] ;W[ql] ;B[mk] ;W[df] ;B[hc] ;W[mj] ;B[sj] ;W[sh] ;B[mk] ;W[jo] ;B[jn]C[5 points is big when white has already gained 3 points at R9. The swing of the ko is at most 9 points now. The thing is that black expects to win the ko.] ;W[mj] ;B[rf] ;W[sf] ;B[mk] ;W[ps] ;B[ns] ;W[mj] ;B[rp] ;W[rq] ;B[mk] ;W[kg] ;B[jf] ;W[mj] ;B[rs] ;W[ss] ;B[mk] ;W[jh] ;B[mj])) (;B[ql]C[this is basically the same as the game] ;W[mk] (;B[sd] ;W[rl]) (;B[pl]C[locally black doesn't have much better than this since white playing here captures P7 with 6 points.])) (;B[pl] ;W[mk] ;B[ql]C[Black has nothing better than this since white is solid on both sides]) (;B[sh] ;W[sj] ;B[mk] ;W[df] ;B[hc] ;W[mj] ;B[rf] ;W[sf] ;B[mk] ;W[nl] (;B[mj]C[This is sufficient here since white gains 4 points at R9, 3 points at T16 compared to the game with a 7 point loss on this ko. Since black has used up 2 ko threats at S14 and T12, they will find it difficult to win the C1 ko. But as compensation white is down to one eye and black cutting at N7 forces white to adda move at C11. This costs one point as well as one dame, which is a profit for black of 7+1+1-4-3=2 points over the game. ] ;W[pk] ;B[ol]C[8 point move] ;W[sd] ;B[jh] ;W[ki] ;B[rb] ;W[hk] ;B[gk] ;W[gl] ;B[hp] ;W[fl] ;B[hs] ;W[pl]C[white has tedomari here for the last move at the 2 point level.]) (;B[pk]C[This also seems to work] ;W[mj] ;B[sd] ;W[ki] ;B[qk] ;W[rk] ;B[qm] ;W[ql] ;B[pm] ;W[rm] ;B[hk] ;W[hl] ;B[fl] ;W[gl] ;B[hp]))) (;W[sd]C[too small white gains 8 2/3 points locally by spending 2 moves (rather than black 1 here) compared to the game but black gains 6 (M10)+1(N10)+2(O9)+1 (T7) - 0.5 (K12)= 10.5 with the ko. (black spends 2 moves on the ko rather than white spending one move in the game)] ;B[nl] ;W[rb] ;B[qb] ;W[sa] ;B[ql] ;W[qa] ;B[pb] ;W[rl] ;B[qk] ;W[rk] ;B[hk] ;W[hl] ;B[fl])) (;W[mj] (;B[sd] ;W[mk] ;B[ql]C[This is the same as the game]) (;B[ql] ;W[sd]C[white can even dare to play here now] ;B[mk] ;W[rb] ;B[qb]C[Black must respond as white at Q18 gains 26 points locally compared to the game. This is the Q19 to T17 block (3*4) plus control of T16 which is 13 stones or 26 points under area scoring.] ;W[mj] ;B[ra] ;W[mk] ;B[qk] ;W[ol] ;B[pl] ;W[nl] ;B[pk] ;W[pm] ;B[ln] ;W[mm] ;B[rm] ;W[hk]) (;B[sh]C[This is not as good] ;W[sj] ;B[mk] ;W[cs] ;B[dr] ;W[mj] ;B[rf] ;W[sf] ;B[mk] ;W[df] ;B[hc] ;W[mj] ;B[bk]C[This is not a threat] ;W[mk]))) (;B[hc]C[Possibly better technique to make white's ko threat here at F17 smaller. But since black can live if white plays H17 and H18 in the game, the game result is probably better technique given more reading since it supports the weak group E17.])) (;W[ql]C[Not so clear. White finds it hard to win the ko since black has 1 more threat at Q6. Despite getting 2 more points locally, white may loses the R2 point. White can get T16 and S18 (around 9 2/3) as compensation for losing 9 points in the ko as well as 2 points at H12, but loses 2 at J8. Black won't have to play O1 which also gains black a point. There is also potential to play Q8 to gain P8 as a point. It is a close call, but AI rates it as not a clear jigo, black has the advantage.]) (;W[mk]C[This probably loses (AI: white winning percentage 10.1 with 700k playouts)] ;B[pn] ;W[mm] ;B[qm] ;W[rm] ;B[ql] ;W[sd] ;B[rl] ;W[rb] ;B[qb] ;W[sa] (;B[sm]TR[pa][qa][ra][sa][pb][rb][sb][sc][sd][sh][sj][sk][rl][sl][rm][sm][sn][so][rp][qr]C[White loses the 11 triangles as points Black loses the 9 triangles as points. There is some potential for black around Q7, but only around 0.5 pt (2 points if black plays P8, but white has 1 point to force O1 later if white plays P8 first and Q9 later). This might be a loss for white even though white has sente now]) (;B[qa] ;W[sm] ;B[sk] ;W[sj])) (;W[pm]C[This is bad because black Q8 becomes a nearly free ko threat.])) (;B[pn]C[bad shape to cut living groups prematurely] (;W[pm]C[This is very bad shape for black since Q7 helps surround R8, gains control of Q1 (worth a point), and O7 is more surrounded. What has black gained? Merely the potential to cut at P7, but white can even respond to that at Q8 which loses black even more points.] ;B[mk]C[later white can respond Q8 in response to P7 due to the P8 cut. That already gains white 3 points around R8 compared to the game as well as the R2 point, all without spending a move. Even though black gains a critical ko threat, it might not be worth it. Black will even have another threat at M6, but then moves such as K5 threaten 8 points and might become ko threats. One problem is that white can play moves like O1 as a big ko threat that only loses 2 points, so in this exchange of ko threats black does probably gain one big enough threat, but already loses points. This might mean that white can afford to play T16 and black might not have enough points if black doesn't respond, but after responding white might not have lost out on ko threats after all.]) (;W[om]C[Also good it seems])) (;B[sd]C[The engine's recommendation] ;W[cs] ;B[dr] ;W[hg] ;B[hc] ;W[df] ;B[fc] ;W[mk] ;B[ql] ;W[qm] ;B[ol]C[This shape is quite funky but is exactly the same number of points as the game as well as more ko threats for black. White gains 1 pt at S9 and loses 1 at R2.] ;W[qk] ;B[pl] ;W[rl] ;B[hk] ;W[hl] ;B[jh] ;W[fk]C[funky order of moves] ;B[hs] ;W[gk] ;B[hp] ;W[fl] ;B[fq] ;W[fm] ;B[eq] ;W[se] ;B[ps] ;W[qr] ;B[sc] ;W[gg] ;B[ih] ;W[hh] ;B[bs]) (;B[ql]C[not a great idea] ;W[sd] ;B[rb] ;W[mk] ;B[om] ;W[pn] ;B[qk] ;W[df] ;B[hc] ;W[ol] ;B[pl] ;W[nl] ;B[pk] ;W[pm] ;B[ln] ;W[mm] ;B[rm]C[white gains 4 points at P7 compared to game but loses around 7 points around S8. White gains 3 points at T16, so it should be completely even if the extra variations at S9/T6 don't change the score.]) (;B[pk] ;W[mk] ;B[rl] ;W[sd] ;B[rb] ;W[ql] ;B[qk] ;W[rk] ;B[qm] ;W[sl] ;B[pl] ;W[rm]C[This shape is quite curious with so many variations in a small area. White has gained 3 points over the game at T16. (territory) Area scoring is that T17 and T16 has changed hands which is 4 points but black spends a gote to take T16, so that explains 4-1=3. but loses nearly 3 (S7, T8, capture at R8) points here. In addition, white has lost R2. White's gain is the potential to capture 2 points with P7. Apparently it is even. White Q6 pretty much forces M6 or P8 (or else Q7 then P8 captures 4 points), and then R8 pretty much forces Q7, so that recovers R2 as well as 1/3 point at R8. Viewed thus, it seems even better for white than the game by around 2/3. The AI seems to say it makes no difference to the score, probably as black is thick and has options to play differently. For example black will have more ko threats at the end of the game, so has a good chance of winning the ko. (both are white gote) However, deeper analysis shows that actually white will be able to even win this ko if all goes the same so white gains 1 point more. Where is the missing point? It turns out it is still jigo because like this, black gains an extra point since there are an odd number of dame points at the end. Because of the extra endgame of R8, white gets tedomari before dame, but there black gets to play first on the odd number of dame points, recovering the missing point. Area counting can be confusing!! ] ;B[mm] ;W[pn] ;B[hk] ;W[hg] ;B[fc] ;W[df] ;B[hc] ;W[hl] ;B[jh] ;W[cs] ;B[dr] ;W[fl] ;B[hs] ;W[fk] ;B[hp] ;W[fm] ;B[eq] ;W[gk] ;B[fq] ;W[ql] ;B[pm] ;W[ps] ;B[ns] (;W[rl] ;B[gg] ;W[bs]C[There are 21 neutral points left including L13 and L11. One way to count the difference to the game is that black loses 4 points due to T17 and T16. Black gains 2 points from poking in at R7, but the switch in control of R8 loses those 2 points back. However, black has the dame points N7, Q9, Q7 filled when in the game black only had P1 (both have Q8). These two dame points recover a point and getting the last dame point gains another. This accounts for -4+2-2+3=-1 in total. Where is the missing point? It is because in the game white won tedomari of dame and that no longer gains white a point. This comes to -1+1=0, so it is still jigo. Another way of looking at this is that the ko on the right meant that white ended up with a stone at T8 instead of S8. Switching the order is equivalent, so it is as if instead of winning tedomari of dame, white suddenly played T8 (which gains zero points), and tedomari of dame switched hands for a total of 2 points difference. Note that the AI says that black has more that 50% at this point but that is because of the K4 move to try to hang onto both J4 and K4 until all the dame is filled. It doesn't work here, so it is still jigo. If I get black to play J4, then it becomes over 50% in white's favour. Hence, I suspect that without any kos on the board, the AI's judgement is biased towards the side that loses tedomari, simply due to its familiarity with territory scoring. If my hypothesis is correct, the AI should be more biased when white loses tedomari (compared to black losing tedomari), because that normally makes a one point difference in score to territory scoring since black plays one more stone one the board during the game. However, from this game, the evidence doesn't seem to support my prediction.]) (;W[gg]))) (;W[mj]C[surely this is better] ;B[hd]C[this is a bad idea. However, this move in the context of black's thick positions around the board with multiple potential kos apparently means black can still eke out jigo. There are 2 extra ko threats on the left now starting with C11 since B14 will be atari. Even the upper white group isn't secure.] ;W[mk] ;B[ql] ;W[pm] ;B[pl] ;W[ol] ;B[qm] ;W[pn] ;B[pk] ;W[sd] ;B[qb] ;W[ln]C[This gains a lot of points, suddenly 7 points here and 3 points at T16 despite losing around 1.75 points at H15 (D13, 2.25 points at H12 but gaining 1/2 at L11 since locally black plays H13 for K12), around 8 at S8, 1/2 at J4. Note that H16 is pretty valuable although it doesn't save a move capturing needing two moves the same as the game, it removes white's sente, which could have been used to make points. White gains sente however (black needs to play S7 to get the above score). Although there isn't much endgame left, white can at least play T6 for example so that black only gains 7 points around S8. This is only a gain of 0.75 in total compared to the game. Playing out the variation with AI leads to a jigo where white also gains a full 0.5 point by taking K4, and another 2.75 around K12 compared to the game since in the game white gets no points then as black spends two moves. This is 4 points better for white than the game. However black ekes out a draw by holding onto the C1 ko after all the dame are filled (with around 5 more internal ko threats than the game at P7, T12 and B14, C11, and 2 hanging around at F11 and H12) still managing to get tedomari of dame. The change in the ko status is 4 points. If white ends the C1 ko, black ekes out a jigo by holding on to a ko at T8 or K4! Area scoring is very tough!])) (;W[mj]C[The analysis keeps saying white should retake the ko and not keep wasting ko threats.])) (;W[] ;B[hg]C[This is annoying to answer] (;W[jh] ;B[jf] ;W[ih]C[Like this the J16 atari has lost nothing]) (;W[ih]C[Like this, white preserves a 3 point follow up at K14 but has already lost 1/2 point at L11 if white wins the ko (from potential point to definitely not a point). Furthermore, playing for K14, K15 atari is not a great strategy locally, when you could play J16 to get a squeeze and make D14 sente. This removes a lot of ko threats in sente as well as gaining a point as we will see.]))) (;B[ql] ;W[sd] ;B[rb] ;W[hg] ;B[hd] ;W[mk] (;B[rl] ;W[ol] ;B[pk] ;W[pl] ;B[qk] ;W[cs] ;B[dr] ;W[ln]C[Black breaks around 8.5 points around S8 compared to the game, white gains 3 points at T16, 6 points from capturing O7, loses 0.5 point at K4. Gains 0.5 at K15 (if black gets K14, that is still the same number of points as the game), loses 1 point at D14, gains 1 point at H12. This totals to 3+6-8.5=0.5 point gain for white, and both are white gote. Apparently it is still jigo but white has a higher winning percentage (61%) ] ;B[df] ;W[cf] ;B[pm] ;W[om] ;B[pn]C[,] ;W[jh] ;B[hk] ;W[hl] ;B[fl]) (;B[qk]C[apparently trying to get an eye at S8 is a better strategy] ;W[ol] ;B[pm] ;W[om] ;B[pn] ;W[sk] ;B[rl] ;W[sl] ;B[rm] ;W[sm] ;B[ln] ;W[mn] ;B[hk] ;W[hl] ;B[df] ;W[cf] ;B[jh]C[Compared to the game, white will lose 5 points on the right (assuming white gets Q9 and black gets Q1) but gain 3 points at O7 and 1/2 at K15 and 3 at T16. In return black will have enough ko threats to hold onto the last ko at K4, since H12 and K14 become miai and B1/C1 and K4 become miai. Black recuperates the 1/2 at K12 for a total of B+6-4=B+2 so far. However, with an odd number of neutral points, white wins tedomari of dame, bringing it back to jigo.]))) (;B[hd]C[Backing off probably better says the AI since the squeeze in the game was perfect. This is preferred at 75% to 60% at 400k playouts. This may lose around 5 points locally for seemingly no reason, but gains a bit from the lack of a squeeze and less ko threats. White's winning percentage drops to 24% after showing this move. However deeper analysis seems to show that it is still jigo. Black wins the ko by creating maybe 10 ko threats on the left, but eventually loses around 10 points here without white even spending a move in this region. The ko itself only had a swing of around 9 points but this goes up to 10 since black wasn't forced to play O12. So in return for losing the ko, white gets two moves at L5 and white can take gote to defend S8 as in the game for the same number of points (since although white plays first, white has to defend Q6). Black gets T16 and white has the initiative to play around G9 and take the C1 ko. So it all pretty much cancels out.] ;W[mj] ;B[ci] ;W[df] ;B[mk] ;W[hf] ;B[ge] ;W[mj] ;B[bi] ;W[bk] ;B[mk] ;W[cs] ;B[dr] ;W[mj] ;B[bf] ;W[be] ;B[mk] ;W[gg] ;B[gd] ;W[mj] ;B[ah] ;W[af] ;B[mk] ;W[ko] ;B[nl] ;W[om] ;B[sd] ;W[jf] ;B[pm] ;W[pn] ;B[pk] ;W[qk] ;B[ql] ;W[rl] ;B[hk] ;W[hl] ;B[kg] ;W[fl] ;B[hp] ;W[gk] ;B[hs] ;W[fk] ;B[fq] ;W[fm] ;B[eq] ;W[je] ;B[id] ;W[jh] ;B[qm] ;W[rm] ;B[pl] ;W[se] ;B[sc] ;W[mm] ;B[ln] ;W[ol] ;B[nk] ;W[ps] ;B[ns] ;W[bs] ;B[jp]C[black squeezes an extra 2 point as usual from this ko, recovering the 2 points loss on territory.])) (;B[ci]C[The analysis thinks this would be good timing] (;W[bi]C[black has made one more ko threat than the game since this was no longer a ko threat.]) (;W[df]C[like this white can't do the same squeeze as H15 is met with G16 and with 10 ko threats on the left, black will win the ko without making negative exchanges. ]))) (;B[he]C[This avoids a squeeze, but allows white to gain 12 point move at F15 more than the game. This is bigger than L5 and bigger than the swing of the ko. It still leaves 6 ko threats at the top.])) (;W[ko]C[Analysis prefers this move as instinct. M12 can be played after winning the ko. The M12 move gains if white loses the ko, but not much anyway and risks all sorts of tactics for black to generate or save ko threats on the left and upper side. This is because white will require one more ko threat in order to win the ko.])) (;W[sd]C[Playing simply also leads to a very close game.] ;B[rb] ;W[mk] ;B[li] ;W[pk] ;B[ol] ;W[pl] ;B[hf] ;W[ee] ;B[fe] ;W[pn] ;B[mr] ;W[ko] ;B[jm] ;W[ik]C[white has 5 points more at S8, 3 at T16 compared to the game. white has 6 points less from the ko. around 1.25 less as an endgame around H15. But around 0.5 more than the final game result. Since black has around 1.5 points less of territory (8 points + K15 where black captured 5 stones in the game) but white loses D14 and has no points at H12 as in the game. White takes gote as in the game. However the loss of eyespace with the centre group will cost white another point due to needing to worry about capturing D12, white will have to connect at N7. Importantly, white doesn't have time to play P7 to connect since black M6 in response has N6 sente (this is better for black than playing M4 in the game), so white is cut off. So white loses the 2 points around J8 and G8 compared to the game. Only K14, the C1 ko and K4 (the persistent last ko) remain so a jigo is very plausible. By the count, white has gained 5+3+0.5-6-2=0.5 with only very small endgame left.])) (;B[li]C[losing] ;W[sd] ;B[rb] ;W[mg]C[compared to the game, white pretty much gained 6 points at N15 and 3 at T16 in return for losing 6 from not capturing M10 and a few from not connecting at N9. The problem is that N9 is worth very little in this position. If black takes, it is sente but white getting O8 is a valuable move. If we imagine white gets O8 naturally later, then just being able to take the N9 ko is worth around 2/3 of a point (it concerns the ko as well as the O12 point so that is (1+1)/3) This is almost directly losing the 2 points required to win. White has several ways of connecting the N8 group back to O6 even without having N8 or an eye at J9. Such as playing P8.])) (;W[nl]C[This is the analysis top move and it makes sense. The N15 stones are about as valuable as the M10 stones. The only cost is that black gets the O12 point and white needs to work a little to prevent being forced to capture at C11.] (;B[ol] ;W[mg] ;B[ql] ;W[sd] ;B[rb] ;W[rl] ;B[qk] ;W[rk] ;B[mj] ;W[ko] ;B[jn] ;W[pn] ;B[mr] ;W[hf] ;B[he] ;W[gg] ;B[jh] ;W[ki] ;B[ih] (;W[gk] ;B[hk] ;W[hl] ;B[hs]) (;W[fl] ;B[hl] ;W[il] ;B[hk] ;W[om])) (;B[lf]C[The problem is that black will capture]))) (;B[fb]C[These leaves E18 and F19 as 2 ko threats and doesn't even prevent the squeeze in the game as it is too far away from F16 and H15 to make a difference, if anything, it makes it easier to spot F16 since it is a vital point of this shape.])) (;W[li]C[The analysis likes this at 50k playouts but still 50% at jigo.])) (;B[jk]C[Black doesn't have enough ko threats to make cutting worthwhile] ;W[ik] ;B[hk] ;W[il] ;B[mk] ;W[pn] ;B[mr] ;W[mj])) (;W[mk] ;B[li] ;W[ol] ;B[pk] ;W[sc] ;B[sb] ;W[sd] ;B[rb] ;W[jn] ;B[rh] ;W[sg] ;B[pl] ;W[rl] ;B[lf] ;W[pm] ;B[jh] ;W[ki] ;B[ih] ;W[hg] ;B[gg])) (;B[ql]C[This move doesn't work yet] ;W[pk] ;B[nl] ;W[pl] ;B[mk]) (;B[mk]C[In hindsight, maybe black should start taking the ko now.] ;W[nl] ;B[kn] ;W[mr] ;B[lr] ;W[pn] ;B[ms] ;W[sc] ;B[sb] ;W[sd] ;B[rb] ;W[ne] ;B[ln] ;W[ko] ;B[jm] ;W[ik] ;B[ql] ;W[rl])) (;W[] ;B[bb] ;W[bc] ;B[aa]TR[aa][ba][ab][bb][eb][ac][ad][bd][ce])) (;W[bc] ;B[nc]C[Simply one point (N18) better.])) (;B[ig]C[Tempting but it is a blunder] ;W[oc] ;B[ob] ;W[nc] ;B[nj] ;W[lc] ;B[mb] ;W[kd] ;B[kc] ;W[ld] ;B[lb] ;W[fd] ;B[gd])) (;W[nj] ;B[kh] ;W[ne] ;B[nc] ;W[oe] ;B[lf] ;W[hf] ;B[gg] ;W[ig] ;B[he] ;W[ol] ;B[pk] ;W[sc] ;B[sb] ;W[sd] ;B[qb] ;W[ie] ;B[id] ;W[bb] ;B[rh] ;W[sh]) (;W[ig] ;B[ie] ;W[kg] ;B[lf] (;W[sc] ;B[sb] ;W[oc] ;B[ob] ;W[sd]) (;W[jf] ;B[ne]))) (;B[ie]C[This would probably lose] (;W[kg] ;B[lf] ;W[hf] ;B[he] ;W[li]) (;W[li] ;B[lh] ;W[ki] ;B[nj] ;W[hf] ;B[he] ;W[jk])) (;B[kg] ;W[ne] ;B[nc] ;W[oe] ;B[nj] ;W[hf] ;B[gg] ;W[fd] ;B[gd] ;W[ed]C[This sets up E18]) (;B[kh] ;W[he] ;B[ig] ;W[fd] ;B[ed] ;W[ee] ;B[gd]) (;B[ig] ;W[jg] ;B[kg] ;W[kh] ;B[jf] ;W[hg] ;B[hf] ;W[ih] ;B[ie] ;W[nj] ;B[lf] ;W[ne])) (;W[kg]C[At 600k it says this is a white win. At k=9, it gives 44.1% -0.2 Solid moves are sometimes best. This doesn't make any points yet, merely disconnecting and connecting. However, this always threatens M11 since black M12 is no longer forcing, gets completely acceptable endgame at J13 since black remains thin enough that F16 might work, and helps connect to P16 so that even S12 isn't forcing. For now, black needs to decide how to make life with the O14 group.] (;B[lf] ;W[ig] (;B[rh] ;W[sh]) (;B[jf] ;W[fd] ;B[ed] ;W[dc] ;B[eb] ;W[he] ;B[hf] ;W[ee] ;B[gd] ;W[fe] ;B[ge] ;W[bc] ;B[nj] (;W[sc] ;B[sb] ;W[sd] ;B[mk]) (;W[li] ;B[mk] ;W[oc] ;B[ob] ;W[mj] ;B[rh] ;W[sg] ;B[mk] ;W[fc] ;B[fb] ;W[mj] ;B[ci] ;W[bi] ;B[mk] ;W[cs] ;B[dr] ;W[mj] ;B[aa] ;W[ab] ;B[mk] ;W[gb] ;B[hb] ;W[mj] ;B[jg] ;W[ki] ;B[mk] ;W[ga] ;B[ha] ;W[mj]))) (;B[jg] ;W[ig] ;B[if] ;W[jh] ;B[jf] ;W[nr] ;B[rh] ;W[lc] ;B[sg] ;W[sf] ;B[nc] ;W[mg] ;B[sh] ;W[kd] ;B[sd] ;W[jc] ;B[os] ;W[id] ;B[hd] ;W[ic] ;B[hc] ;W[li]) (;B[rh]C[Something it considers as a nice way to set up lots of ko threats if white answers but white will fight back at M17. Because white can make an eye with P17]))) (;B[dr]C[Simply better shape in every way locally.] ;W[cs] ;B[ds]) (;B[ig]C[getting sente to play here first significantly helps solve the M11 cut since L13 has no power anymore. White's many cutting points are exposed and S12 becomes completely forcing. This probably gains around 5 points already + perhaps 2 from helping solvethe M11 cut. Just locally, this is probably 2.5 points better than the game. But compared to white playing first at L13 and then J14 with follow ups around F16 or O10 probably makes this move a lot bigger in this case.])) (;W[ds]C[Normally tesuji but ] (;B[dr] ;W[cs]C[The problem is that white has no eye and this black group has 2 (the critical number) of extra approach liberties since C1 is cut off and white G1 is cut off. Black can still win the capturing race without an eye since E1 acts like a temporary eye in this case.] ;B[es] ;W[cs] ;B[hs]C[Black has already made 2 points by capturing 2 stones so even if white captures B1, black has 1 point locally and this is clearly worse for white than the game. Where black only has 1/3 point locally. White has lost forcing moves around H9.]) (;B[cs]C[this is dead] ;W[dr] ;B[es] ;W[dr] ;B[eq] ;W[fq]))) (;B[] ;W[as]C[W+20] (;B[hs] ;W[cs] ;B[dr] ;W[ds]C[the existence of A1 means that B1's effect is removed.]) (;B[fs] ;W[cs] ;B[hs] ;W[dr] ;B[eq] ;W[es] ;B[ds] ;W[mq] ;B[ci] ;W[bi] ;B[lr] ;W[es] ;B[df] ;W[cf] ;B[ds] ;W[oc] ;B[ob] ;W[es] ;B[ce] ;W[bf]))) (;B[ig] ;W[jg] ;B[kg] ;W[kh] ;B[nj] ;W[da] ;B[rh] ;W[sg] ;B[ea] ;W[ca] ;B[mk] ;W[if] ;B[jf] ;W[ar] ;B[fs] ;W[hg] ;B[hf] ;W[ih] ;B[ie] ;W[nl] ;B[ol] ;W[as] ;B[dr] ;W[fd] ;B[ed] ;W[fb] ;B[fc] ;W[eb] ;B[dc] ;W[fa])) (;W[oc]C[This is preferred until around 100k when it decides it doesn't work. But it is certainly extremely complicated.] ;B[ob] ;W[nc] ;B[nj] ;W[fb] ;B[ed] ;W[lc] ;B[df] ;W[cf] ;B[mb] (;W[pr] ;B[mi] ;W[or] ;B[nq] ;W[ee] ;B[fe] (;W[hc] ;B[dc] ;W[ce] ;B[hd] ;W[gb] (;B[id] (;W[ib] (;B[ha] ;W[kd] ;B[ne] ;W[jc] ;B[kb] ;W[lb] ;B[ic] ;W[fc] ;B[fd] ;W[kg] ;B[lf] ;W[gd] ;B[jb] ;W[he] ;B[if] ;W[da] ;B[ea] ;W[mf] ;B[jf] ;W[ca] ;B[eb] ;W[mg] ;B[rh] ;W[sh] ;B[jk] ;W[mm] ;B[om] ;W[pn] ;B[ml] ;W[ln] ;B[kc] ;W[sc] ;B[sb] ;W[sd] ;B[qb] ;W[pk] ;B[jh]) (;B[fc]C[eventually it says B+5] ;W[kb] ;B[ic] ;W[kd] ;B[ne] ;W[jc] ;B[hb] ;W[ha] ;B[ja] ;W[fa] ;B[ca] ;W[hc] ;B[jb])) (;W[fc]C[but this is a good move] ;B[fd] ;W[ib] ;B[gd] ;W[kb] ;B[ic] ;W[kd])) (;B[kd] ;W[gd] ;B[fc] ;W[he] ;B[id] ;W[da] ;B[ea] ;W[ca] ;B[eb] ;W[hf] ;B[ge] ;W[gg] ;B[fd] ;W[om] ;B[rh])) (;W[id] (;B[if] ;W[kd] ;B[jc]) (;B[kd] ;W[if] ;B[ic] ;W[hf]) (;B[je] ;W[hf] ;B[he] ;W[gg]))) (;W[kd] ;B[ne] ;W[je] ;B[jf] ;W[lf] ;B[jc] ;W[ml] ;B[mi] ;W[kh] ;B[rq] ;W[rp] ;B[rs] ;W[sr])) (;W[mi] ;B[lh] ;W[ne] ;B[nc] ;W[oe] ;B[kh] ;W[ol] ;B[jh] ;W[ar] ;B[fs] ;W[as] ;B[dr] ;W[pr] ;B[or] ;W[qs] ;B[pk] ;W[nr] ;B[os] ;W[nq] ;B[oq] ;W[mp] ;B[mq] ;W[op] ;B[oo] ;W[no] ;B[op] ;W[mg] ;B[nh] ;W[sc] ;B[sb] ;W[sd] ;B[rb] ;W[fb] ;B[ca] ;W[ba] ;B[da] ;W[fd]) (;W[lh]C[inducing move to let both sides save their stones immediately. Black gets an extra point at N18 and may even get sente to play S3, but white gets more efficient shape locally saving things with one move, not needing to add a move at N8 yet and already managing to play O10. White exchanging P8 for Q9 damages the right side but if black exchanges S3 for S4, then the right side is maybe already 0.5 point smaller.] ;B[mi] ;W[nj] ;B[ne] ;W[ar] ;B[fs] (;W[li] ;B[rq] ;W[as] ;B[dr] ;W[oc] ;B[ob] ;W[rp] ;B[rs] ;W[sr] ;B[or] ;W[lf] ;B[nc] ;W[ol] ;B[pk]) (;W[oc] ;B[ob] ;W[lf] ;B[rh] ;W[sh] ;B[li] ;W[jg] ;B[nc]))) (;B[ne] ;W[oi]C[this costs black around 6 points.])) (;W[lc]C[this doesn't work yet and is quite a bad move] ;B[nc] ;W[ni] ;B[mi]C[this captures M15 and hence M17 is difficult to use.] ;W[nj] ;B[oi])) (;W[nj]C[The analysis prefers this until around 50k playouts. This has the advantage of having more liberties so taht M11 and O12 are not forcing at all on white's shape, so black needs to add more moves capturing M13 in order to save the group. B needs to focus on threatening to connect N10 as well. However, the loss is that white has not yet cut N10 which may seem insignificant but it is probably 2 points difference locally and white has potential in the L7 area with N10. Now without, if black connects N10, it is pretty much sente since white has many cutting points. Even worse, black can threaten to play N8 directly (or potentially) with bigger moves like L6 in sente. Basically, if white doesn't get solid here, there is a clash with trying to maximise territory around R8 ot L7 later. Cutting N10 allows white to save this stone fairly cleanly. ] ;B[oj] ;W[ni] ;B[oi] ;W[nl] ;B[kh] ;W[ig] ;B[jf] ;W[ar] ;B[fs] ;W[ne] ;B[nc] ;W[as] ;B[dr] ;W[pr] ;B[or] ;W[oe] ;B[lf] ;W[qs] ;B[ca] ;W[ba] ;B[da] ;W[bc]) (;W[oj]C[this is much worse] ;B[nj] ;W[nl]C[bad idea to continue] ;B[ml] ;W[ol] ;B[om] ;W[pl] ;B[mm] ;W[jk] ;B[pn] ;W[kn] ;B[pr] ;W[if]C[compared to the game, white hasn't even gotten the chance to save M15 yet nor atari M17. White has gained 7 points capturing P9, but lost 12 at R2, and O6 is still hanging. This is a disaster. This is not the way to attack a weak group.])) (;B[jq]C[This is a standard sort of mistake. Although K2 is not a threat in this position] ;W[mq]C[This is very valuable at this point in the game. Locally it gains 3 points + M4 in sente whereas K3 only gains the one point at K2. But to be honest of those 3 points, they weren't really black's anyway if white captured Q2, so you certainly can't count it as a full 3 points. And there is a clash between white's endgame here and after Q2 (black N2 defends against both attacks simultaneously which is a bonus for black). If black gets Q2, then this move is pretty much a full 3 points gain, while if white gets Q2, it pretty much gains zero. It is probably still worth more than (3+0)/2=1.5 points though, because white has other forcing moves here such as N2 which can connect back to Q2 if white wins the Q2 endgame.]) (;B[mr]C[Even this is something to consider, giving up a point at N2 (as well as letting white get K3 in sente), but preventing the squeeze in the game. The main problem with this move seems to be that white playing K6 is forcing, since it threatens K2 - K3 - L1 - M2 - J1 - J2 - H1 and black risks losing 100 points in the capture race with this ko. K6 is very valuable!])) (;W[mj]C[why not immediately? White probably ends up return to L4 though.])) (;B[kp]C[possible but analysis dislikes by 3%, 0.4] ;W[lp] ;B[lr] ;W[nq] (;B[nr] ;W[mq] ;B[mr] ;W[jp] ;B[ko] ;W[jo] ;B[kn] ;W[jn] ;B[jq] ;W[ln] ;B[kr] ;W[ar] ;B[fs] ;W[mj]) (;B[mq]C[bad] ;W[nr] ;B[oq] ;W[mp] ;B[kr] ;W[op] ;B[oo] ;W[no])) (;B[lp] ;W[kp] ;B[ko] ;W[jr]C[-15 points]) (;B[lr]C[the squeeze in the game is more powerful now as this is a very bulky shape with N3 and N2.])) (;B[oj] (;W[mj] ;B[ok]) (;W[ok]C[bad as P10 has become a ladder maker.] ;B[ml] ;W[mm] ;B[nl] ;W[kq])) (;B[mj]C[analysis considers it but says not so good. Since black has O10 sente, this black group is totally alive but white won't bother with so much effort saving M15 now since the right side has already increased by around 13 points compared to the game, assuming white gets S12. White will aim at J14 sente and won't try anything against black's upper side since black is thicker now. Black has a good chance of play S12 in sente, recovering 3 points. White also hasn't captured N9, another 1 pt, black has 2 points at O11. Also black has sente rather than white, which accounts for the large discrepancy in points. White is 54% 0.5 (k=5)] (;W[kq]C[necessary timing to play this before black gets P2] ;B[kr] ;W[kp] ;B[jr] ;W[ml]) (;W[ml]C[careful] ;B[rq] ;W[rp] ;B[rs] (;W[sr] ;B[kn] ;W[ln] ;B[or]) (;W[kq]C[this is the latest point where this is playable]))) (;B[rq] ;W[rp] ;B[rs] ;W[mj] ;B[sr]C[with 2 moves black gains around 20 points here compared to the game.] ;W[ne] ;B[nc] ;W[kq] ;B[kr] ;W[mg] ;B[kh] ;W[oe] ;B[mi] ;W[nj] ;B[lh] ;W[if] ;B[ig] ;W[jg] ;B[pn] ;W[he] ;B[gd] ;W[pm] ;B[qm] ;W[om] ;B[ql] ;W[sm]) (;B[ml] ;W[mj]C[this cut is too good when black is not alive.] ;B[nj] ;W[mi] ;B[ni] ;W[li] ;B[ok] ;W[ne])) (;B[rr] ;W[fj] ;B[fk] ;W[gk] ;B[hk] ;W[jk] ;B[gl] ;W[fi]C[black has 3 liberties now])) (;W[fj]C[analysis says this is playable.] ;B[fk] ;W[jj] ;B[gk]C[black still has 5 liberties rather than 7. However, are white's moves to short black's liberties useful moves? Consider that they give up forcing moves at A2 and ajiw th L3 and K6.])) (;W[hg]C[analysis prefers this but the variation is very messy.] ;B[qr] ;W[rr] ;B[rq] ;W[rp] ;B[rs] ;W[jj] ;B[hp] ;W[sq] ;B[mj] ;W[kq] ;B[ko] ;W[gi] ;B[jk] ;W[ik] ;B[fj] ;W[lp] ;B[ml] ;W[mq])) (;B[qr] ;W[rr] ;B[rq] ;W[rp] ;B[rs] ;W[kq] ;B[kr] ;W[mj] ;B[sr] ;W[ne] ;B[nc] (;W[jj] ;B[kh] ;W[oe] ;B[gi] ;W[hg] ;B[hf] ;W[mg] ;B[lh] ;W[ni] ;B[nh] ;W[kp] ;B[jr] ;W[fb] ;B[ed] ;W[da] ;B[ea] ;W[eb] ;B[dc]) (;W[mg] ;B[ik] ;W[gf] ;B[ge] ;W[he] ;B[fd] ;W[hf] ;B[hd]))) (;W[ml] ;B[ll] ;W[lm] ;B[qr])) (;B[kn]C[analysis says] ;W[lm] ;B[ec] ;W[de] ;B[mk] ;W[ml] ;B[ll] ;W[gi] ;B[ik] ;W[lk] ;B[qr] ;W[rr] ;B[ll] ;W[ko] ;B[ln] ;W[mn] ;B[jm] ;W[mp] ;B[mq] ;W[pr] ;B[or] ;W[km] ;B[jn] ;W[il] ;B[hf] ;W[mj] ;B[ps] ;W[kq] ;B[kr] ;W[kp] ;B[jq] ;W[lk] ;B[kh] ;W[ne] ;B[nc] ;W[oe] ;B[lf] ;W[fb] ;B[eb] ;W[fe]) (;B[cf]C[analysis says this is 2 points worse. Consider that in the game black maintained the threat of forcing white to add another move to capture D12 (the centre group didn't have eyes, and the D14 atari with D15 not being completely solid. Also, this black group isn't very strong so going deeper into white's area is heavy.] ;W[ci] ;B[df] ;W[ec] ;B[gg] ;W[gi] ;B[eb] ;W[gd] ;B[hc] ;W[be])) (;W[fb]C[doesn't really work] ;B[ec]) (;W[kq]C[doesn't work yet] ;B[kp] ;W[lp] ;B[mq] ;W[im] ;B[ko] ;W[kn] ;B[ln] ;W[ne] ;B[mm] ;W[jj] ;B[jk] ;W[jn] ;B[mo])) (;B[gg]C[analysis] ;W[gi] ;B[ce] ;W[ed] ;B[ge] ;W[fb] ;B[gb] ;W[ne] ;B[df] ;W[ci] ;B[ik] ;W[lj] ;B[ki] ;W[li] ;B[lh])) (;B[fe] ;W[fb] ;B[ec] ;W[de] ;B[fc]) (;B[pn] ;W[pm] ;B[km] (;W[gd] ;B[hc] ;W[lm] ;B[jl] ;W[ll]) (;W[jl] ;B[lm] ;W[ll] ;B[om] ;W[mm] ;B[mn] ;W[ml] ;B[pl] ;W[qm] ;B[no]))) (;W[cf]C[this doesn't help later black has C15, B15 in sente, greatly threatening the C18 group.])) (;W[ef] ;B[fg] ;W[fi] ;B[bh] ;W[bf] ;B[fj] ;W[hg] ;B[gg] ;W[gh] ;B[gf] ;W[ne] ;B[nc] ;W[hf] ;B[ge] ;W[lf]) (;W[df] ;B[gh] ;W[fg] ;B[gi])) (;W[jl] ;B[ce] ;W[de] ;B[cf] ;W[df] ;B[ci] ;W[bd] ;B[bj] ;W[bk] ;B[bl] ;W[am] ;B[be] ;W[ah] ;B[ai] ;W[aj]) (;W[ne] ;B[lj] ;W[jl] ;B[pn] ;W[pm] ;B[mn] ;W[mo] ;B[lm] ;W[mq]) (;W[ed] ;B[pn] ;W[pm] ;B[km] ;W[lc] ;B[nc] ;W[kd] ;B[lf] ;W[hc])) (;B[eg] ;W[fg] ;B[ef] ;W[bh] (;B[db] ;W[ed]C[black not great]) (;B[ff] ;W[fd] ;B[gd] ;W[fc])) (;B[ec] ;W[ed] ;B[fd] ;W[eb] ;B[eg] ;W[fg] ;B[ef] ;W[bh] ;B[fc] ;W[db] ;B[ff] ;W[jl] ;B[cf] ;W[ci] ;B[df] ;W[im] ;B[qr])) (;W[] ;B[rh] ;W[sh] ;B[si] ;W[sg] ;B[se] ;W[sd] ;B[rd] ;W[ri] ;B[sc] ;W[sj] ;B[rf]C[later] ;W[sf] ;B[oc] ;W[sd] ;B[oe])) (;B[oj]C[analysis prefers this] ;W[ni] ;B[lj] ;W[gd] ;B[rh])) (;W[lc] ;B[kd] ;W[re]) (;W[mg] ;B[nd] ;W[ne] ;B[nc] ;W[oe] ;B[mj] ;W[re])) (;B[jl]C[responding is small local loss] (;W[jm]C[this means that black still has an important cutting point when O13 is not alive.]) (;W[km]C[even here is something to consider]))) (;B[mi]C[analysis prefers] ;W[kf] ;B[ll] ;W[ml] ;B[mm] ;W[lm] ;B[kl] ;W[km] ;B[ec] ;W[kq] ;B[kp] ;W[lp] ;B[lr] ;W[ni] ;B[mh])) (;B[rf]C[not good as white has many local threats and this ko is too heavy on black.] ;W[re] ;B[qe] ;W[lc])) (;B[rj]C[not great] ;W[rh] ;B[ri] ;W[pf] ;B[mi] ;W[li] ;B[lj] ;W[ni] ;B[nj] ;W[oj] ;B[mj] ;W[nh] ;B[pn] ;W[om] ;B[pm])) (;B[pf]C[analysis prefers] ;W[of] ;B[rf] ;W[pe] ;B[re])) (;W[of]C[analysis prefers] ;B[rj] ;W[me])) (;W[ng] ;B[pi] ;W[nc] ;B[lb])) (;W[po] ;B[pi] (;W[qj]) (;W[og]) (;W[ph] ;B[pg]))) (;W[rd] (;B[pn] ;W[ro] ;B[po] ;W[ll]) (;B[pi] ;W[oi] ;B[ph] ;W[qj] ;B[oh] ;W[rh] ;B[me] ;W[po] ;B[oo] ;W[pn] ;B[nn]))) (;B[gr] ;W[cl] ;B[gm] ;W[ip] ;B[io] ;W[jp] ;B[hp] ;W[hq] ;B[jq] ;W[kq] ;B[hr] ;W[jo] ;B[in] ;W[km] ;B[kr] ;W[im] ;B[jn] ;W[kn] ;B[hm] ;W[jm] ;B[hn] ;W[qp] ;B[qq] ;W[qo])) (;B[fn] ;W[br] ;B[bs] ;W[ar] ;B[fr] ;W[ip] ;B[gr] ;W[cm] ;B[em] ;W[cl] ;B[hp] ;W[hq] ;B[io] ;W[jp] ;B[jq] ;W[kq] ;B[ir] ;W[hm] ;B[in] ;W[ek])) (;B[en]C[analysis prefers this]))