(;FF[4]
CA[UTF-8]
GM[1]
GN[KataGo 7 komi self-play - game 105]
PC[https://online-go.com/review/992450]
PB[KataGo]
PW[KataGo]
BR[9p]
WR[9p]
OT[Error: time control missing]
RE[?]
SZ[19]
KM[7]
RU[NZ]

;B[qd]
;W[dd]
;B[pq]
(;W[cp]C[This one's a little interesting, simply because you don't want to let your opponent approach from the side they have support on, but it's not a huge deal.

[All comments on this game by Clossius (AGA 3d), except for bracketed comments by Maharani.\]]
;B[eq]
;W[oc]C[This is pretty common. Stopping the bottom right shimari is also quite big.]
;B[do]C[This is an urgent move.]
;W[co]
;B[dn]
;W[cm]
;B[pn]TR[pe]C[We get the classic two-space high. Next, I expect white's urgent move at Q15.]
;W[pe]LB[dm:A]C[[In game 101, white continued at A instead.\]]
;B[qe]
;W[pf]
;B[qg]
;W[dm]C[Now white is pushing, and it's a moyo versus moyo game.]
;B[pg]C[This is an interesting little setup here.]
;W[cf]TR[em]C[This is pretty much the last move for the opening, so now we should expect middle-game shenanigans. If I'm black, it's gotta be the top or the bottom. That's a tough choice. Maybe a hane at E7?]
;B[kp]C[Black chooses the bottom.]
;W[fm]
;B[gn]C[Black is just building up his framework.]
;W[gm]C[And white is doing the classic "reducing while building".]
;B[hn]
;W[il]C[White is getting her moyo, while also pushing and limiting black's moyo.]
;B[jm]C[Black will have to go up top at some point! He is getting a lot of solid cash, though. But white's influence is very big, so I'm not saying she is down by any means.]
;W[jl]
;B[km]
;W[lk]C[Okay, black has GOT to go up top.]
;B[fc]TR[mc]C[There we go! Classic approach. I would have expected something around N17, but maybe white doesn't care about her three stones as much as her moyo.]
(;W[hc]
;B[fe]
;W[he]
;B[db]C[Black gets some eye-space.]
;W[qo]TR[cc]C[Uhhh... okay, this is weird. The AI is really good at invading, so I'm not going to comment too much about this; it requires a lot of reading. But the timing is very odd.

I would expect white C17 instead. But I guess as soon as black is alive on the top (and from here, we could read that black is going to live at some point), maybe if white doesn't take black's territory now that black has taken white's territory, white's gonna be behind? So I guess that's the timing. It's just lots of reading.]
;B[qn]
;W[po]
;B[oo]
;W[rn]C[White is just going for life.]
;B[pp]
;W[rp]
;B[ql]C[Black is choosing a simple surrounding sequence.]
(;W[on]C[I'm not sure about this cut. What is white trying to do here?]
;B[rq]
;W[pl]C[This is the bamboo shape point.]
;B[qm]
;W[qk]C[So white is going for a squeeze more so than life; interesting. But white hasn't gained anything yet, so I'm not quite seeing her plan yet.]
;B[rk]
;W[no]
;B[op]
;W[om]C[I understand what the white moves do, but I don't quite see the future that white's seeing yet.]
;B[rj]
;W[np]
;B[nq]
;W[mq]TR[rn]TR[po]TR[qo]TR[rp]C[It looks like we're going to be breaking in and getting the cash right here instead of the cash over in the corner, but that's fair.]
;B[nr]
;W[fn]
;B[fo]
;W[go]C[Ooh! If something's here, that's big.]
;B[hm]
;W[hl]
;B[gp]
(;W[kq]LB[mn:B]TR[no]LB[lp:A]TR[np]C[There's lots of reading here. White's simple answers would be A or B instead to fix her cutting points. There is so much reading involved here, but there is a lot of aji.]
(;B[lp]
;W[ho]
;B[jo]C[The ladder doesn't work for black, so he just does this instead.]
;W[hp]
;B[gq]
;W[hq]TR[rn]TR[po]TR[qo]TR[rp]C[So basically, white made miai of either destroying the points around R5 or breaking into the H3 area. The AI is just really good at reading, so it's quite difficult to follow.]
;B[gr]
;W[hr]
;B[jq]
;W[dr]C[Black is still not alive! There's no way white wins this capturing race.]
;B[er]
;W[eo]
;B[fp]
;W[dp]
;B[cr]
;W[dq]
;B[ds]
;W[br]
(;B[fg]MA[hm]MA[jm]MA[km]MA[gn]MA[hn]TR[fo]SQ[go]SQ[ho]MA[jo]TR[fp]TR[gp]SQ[hp]MA[kp]MA[lp]TR[eq]TR[gq]SQ[hq]MA[jq]SQ[kq]SQ[mq]TR[cr]TR[er]TR[gr]SQ[hr]TR[ds]C[Um, what? I have no idea what's going on. I don't know what to tell you.

I think black is dead on the bottom? That's the conclusion that I'm coming to, but black has all these kos and stuff, so I guess he's just saving it for later.

But then I also have to ask: If the H5 stones are alive, are the K5 stones alive? I guess there's plenty of aji. So much reading! White's marked stones cannot all be dead with no consequence though; that would be horrible.]
(;W[fs]C[White fixes it immediately; she's not even gonna worry about a ko.]
;B[gs]
;W[fr]
;B[ep]
;W[en]
;B[bs]
;W[cq]
;B[cc]C[There's STILL a ko on the bottom!]
;W[dh]
;B[eh]
;W[di]TR[qd]TR[rd]TR[sd]TR[qe]SQ[af]SQ[bf]SQ[cf]TR[pf]SQ[cg]SQ[dg]TR[qg]SQ[dh]TR[qh]TR[qi]SQ[dj]TR[rj]TR[sj]SQ[dk]TR[rk]SQ[dl]TR[ql]SQ[dm]TR[pm]TR[qm]SQ[en]TR[on]TR[pn]SQ[do]SQ[eo]TR[fo]TR[go]TR[oo]SQ[ap]SQ[bp]SQ[cp]SQ[dp]TR[ep]TR[gp]TR[op]MA[bq]TR[cq]TR[dq]TR[gq]TR[nq]TR[ar]TR[br]TR[gr]TR[hr]TR[nr]TR[hs]TR[ns]C[The score is probably pretty close. The right is solid territory. The lower left is probably gonna die, but there's a still a stupid ko at B3, so I can't even count this.

Black has more cash while the lower left is undecided, but white should have more cash if it dies, so it's hard to say.]
(;B[jc]C[This is the next big thing to do if the bottom's irrelevant.]
;W[lc]
;B[ld]
;W[mc]
;B[ng]C[This is just some annoying sabaki.]
;W[if]
;B[je]
;W[cd]
(;B[lf]TR[lc]TR[mc]TR[oc]TR[pe]TR[pf]C[Now white's getting surrounded, but we all know the AI can live pretty nicely.]
;W[dc]
;B[ih]C[I like this. Yeah, the top left corner is big in points, but if white dies...]
;W[jf]
;B[kg]C[Surrounding things is important. Black can always come back to live in the top left.]
;W[li]
;B[ki]
;W[gh]
;B[hg]
;W[lh]
;B[gd]C[I swear the AI is the master of miai.]
;W[hd]
;B[kd]
(;W[eb]
;B[fb]
;W[me]TR[db]TR[cc]SQ[hc]SQ[hd]SQ[he]SQ[if]SQ[jf]C[Black traded the corner for H15.]
;B[kf]
;W[kh]
;B[jh]TR[hc]TR[hd]TR[he]TR[if]TR[jf]C[Now black's gonna squeeze it.]
;W[qb]C[And white makes life!]
;B[bq]C[There's that ko! Alright, I guess it's finally big enough.]
;W[cs]
;B[kk]
;W[kl]
;B[cr]
;W[ar]
;B[aq]
;W[cs]C[Black needs a lot of moves in a row to win this ko, though.]
;B[ll]
;W[mk]
;B[cr]
;W[es]C[[At this point, KataGo gives black 65.9 % winrate and the score as B+2.\]]
;B[fq]
;W[cs]
;B[ml]
;W[kj]
;B[nk]
;W[fs]C[I believe black is just dead now.]
(;B[ir]
;W[hs]
;B[gg]SQ[nk]TR[qk]SQ[ll]SQ[ml]TR[pl]SQ[hm]SQ[jm]SQ[km]TR[om]SQ[gn]SQ[hn]TR[on]SQ[jo]TR[no]SQ[kp]SQ[lp]TR[np]SQ[jq]TR[kq]TR[mq]SQ[ir]C[But black cut the P6 stones off in the exchange. Black's L7 group is also not one hundred percent alive, though. So complicated!]
;W[oi]C[This is sabaki again.]
;B[pj]
;W[oj]
;B[ok]
;W[ef]
;B[ff]
;W[pk]
;B[nj]
;W[oh]C[Fancy is fancy.]
;B[mi]
;W[mh]
;B[nh]
;W[og]
;B[pi]TR[kh]TR[lh]TR[mh]TR[li]TR[kj]TR[lk]TR[mk]C[White is still dead.]
;W[qf]
;B[rf]
;W[nf]
;B[ni]
;W[mg]C[Black's got to fix.]
;B[mf]
(;W[rg]TR[mg]TR[kh]TR[lh]TR[mh]TR[li]TR[kj]TR[lk]TR[mk]C[This is looking for some points to make up for the dead group.]
;B[rh]
;W[ji]TR[ng]TR[nh]TR[mi]TR[ni]TR[nj]TR[nk]TR[ok]TR[ll]TR[ml]TR[hm]TR[jm]TR[km]TR[gn]TR[hn]TR[jo]TR[kp]TR[lp]TR[jq]TR[ir]C[This threatens black's marked group. The timing is weird, but whatever.]
;B[nl]
;W[re]
;B[sg]
;W[cb]C[This is just points.]
;B[rd]
;W[lm]TR[ng]TR[nh]TR[mi]TR[ni]TR[nj]TR[nk]TR[ok]TR[ll]TR[ml]TR[nl]TR[hm]TR[jm]TR[km]TR[gn]TR[hn]TR[jo]TR[kp]TR[lp]TR[jq]TR[ir]C[And now we're back to black's group. KataGo is switching between endgame and threatening this group.]
;B[mn]
;W[md]SQ[gj]TR[pk]TR[qk]TR[pl]TR[om]TR[on]TR[no]TR[np]TR[kq]TR[mq]C[I'm pretty sure white's marked group is dead. If that is the case, she has to have enough territory to make up for it.

[KataGo predicts the score as B+1.4 at this point.\]

Yeah, we can totally count that… The endgame does look in white's favour. Black can't make anything in the G10 section, but white can, and there are lots of forcing moves for white on the top.]
;B[jb]
;W[ln]
;B[kn]
;W[gi]C[Yup, there are the points. The AI is a master of endgame. If dan players can gain twenty to thirty points against kyu players in endgame, the AI can probably gain thirty to fifty points against dan players in endgame. It's ridiculous.]
;B[ma]
;W[qc]
;B[pd]
;W[od]
;B[rc]
;W[im]
;B[ob]
;W[nb]
;B[na]
;W[mb]
;B[ka]
;W[pb]
;B[ei]
;W[jr]C[This is fancy.]
;B[oa]C[This threatens to kill, that's important.]
;W[pc]
(;B[rb]
;W[kb]TR[la]C[Now white is threatening the throw-in at M19.]
(;B[kr]TR[jr]C[And then black responds to the K2 tesuji. The AI can't decide between endgame and sente moves that it's gonna get no matter what.]
;W[ej]
;B[kc]
;W[lb]
;B[la]
(;W[nd]TR[of]C[Surprised about this one instead of P14.]
;B[fa]
;W[in]C[This is quite big.]
;B[cr]C[This is nothing, just the AI being AI.]
;W[es]
;B[ne]C[So black is losing, because black is freaking out and can't figure out what to do to win, so he's guessing at sente moves now.

[At this point, KataGo gives white's winrate as 60.4 % and the score as W+0.6.\]]
;W[oe]
;B[ii]
;W[ij]
;B[fr]
;W[cs]
;B[ph]C[[At this point, KataGo gives white's winrate as 66.8 % and the score as W+0.8.\]]
;W[of]
;B[es]C[[Now, KataGo gives white's winrate as 67.9 % and the score as W+0.9.\]

The AI as black is just being a bit stupid now. I'm not surprised that after a certain point, things just drastically get worse, because the AI doesn't care about the score, whether it's +25 or +1. All it cares about is the winrate. But we can also see that the AI as white is a master of miai.]
;W[fs]
;B[iq]
;W[ec]
;B[ee]
;W[eg]
;B[fh]
;W[fi]
;B[hi]
;W[hj]
;B[dj]
;W[fj]
;B[ea]
;W[da]
;B[lj]
;W[jk]
;B[de]
;W[ce]
;B[df]
;W[dg]
;B[mj]
;W[le]
;B[ke]
;W[hb]C[No way! Really? Really?!]
;B[jg]
;W[ed]
;B[ge]
;W[rl]
;B[rm]
;W[qq]C[Is the AI playing timesujis? White is losing ko threats for no good reason.]
;B[qp]
;W[ro]
;B[qr]
;W[sm]
;B[sl]
;W[hf]
;B[lg]
;W[hh]
;B[ig]
;W[ki]
;B[pa]
;W[qa]
;B[cg]C[Black's freaking out.]
;W[ch]LB[fb:A]LB[gb:A]LB[fc:A]LB[gc:A]LB[fd:A]LB[gd:A]LB[fe:A]LB[ge:A]LB[cf:B]LB[ff:A]LB[gf:A]LB[cg:B]LB[fg:A]LB[gg:A]LB[dh:B]LB[di:B]LB[dj:B]LB[dk:B]LB[dl:B]LB[nl:D]LB[ol:D]LB[pl:D]LB[sl:C]LB[mm:D]LB[nm:D]LB[om:D]LB[pm:D]LB[qm:C]LB[rm:C]LB[sm:C]LB[mn:D]LB[nn:D]LB[on:D]LB[pn:D]LB[qn:C]LB[rn:C]LB[eo:E]LB[fo:E]LB[go:E]LB[ho:E]LB[mo:D]LB[no:D]LB[oo:D]LB[qo:C]LB[ro:C]LB[dp:E]LB[ep:E]LB[fp:E]LB[gp:E]LB[hp:E]LB[ip:D]LB[jp:D]LB[kp:D]LB[lp:D]LB[mp:D]LB[np:D]LB[op:D]LB[qp:C]LB[rp:C]LB[dq:E]LB[eq:E]LB[fq:E]LB[gq:E]LB[hq:E]LB[iq:D]LB[jq:D]LB[kq:D]LB[lq:D]LB[mq:D]LB[nq:D]LB[cr:E]LB[dr:E]LB[er:E]LB[fr:E]LB[gr:E]LB[hr:E]LB[ir:D]LB[jr:D]LB[kr:D]LB[lr:D]LB[mr:D]LB[nr:D]LB[cs:E]LB[ds:E]LB[es:E]LB[fs:E]LB[gs:E]LB[hs:E]LB[is:D]LB[js:D]LB[ks:D]LB[ls:D]LB[ms:D]LB[ns:D]C[Master of Miai. Black destroyed the A area, but white got points in the B area. White started at C, then sacrificed it to get to D, then sacrificed it to get all of E. "As long as I get something, I don't care!" The bottom side was crazy.]
;B[bg]
;W[bh]TR[hb]TR[hc]TR[hd]TR[he]TR[hf]TR[if]TR[jf]C[The top is an interesting go problem.]
(;B[ja]
;W[id]C[Black's whole group dies?]
;B[ha]
(;W[jn]C[[White 99 % (KataGo)\]

-- chat --
Maharani: Video review by Clossius (AGA 3d): https://www.youtube.com/watch?v=-BYb1Hj8H8Y
]
)(;W[ib]C[[KataGo gives this move -0.1 pp.\]]
(;B[ia]
;W[gb]
;B[ga]
;W[gc]
;B[fd]C[This is a seki! Like I said, a very interesting go problem here. But the winrate doesn't change, so it doesn't really matter for the AI.]
)(;B[bi]C[[KataGo variation.\]]
;W[bf]
;B[ag]
;W[ah]
;B[kk]
;W[lk]
;B[ia]
;W[gb]
;B[ga]
;W[fd]
;B[gc]
;W[ra]
;B[sb]
;W[jd]
;B[ip]
;W[gf]
)))(;B[ia]C[What if black gives up the tail?

[KataGo gives this move -2.6 pp.\]]
(;W[ha]
;B[id]C[Black makes the eye.]
;W[ja]
;B[ka]
;W[la]
;B[ja]
;W[fd]
;B[gc]TR[ma]TR[na]TR[oa]TR[pa]TR[ob]C[In this variation, it's an eye versus no eye. So black's alive, but has lost a big chunk. Giving up the stones and keeping white dead is worth around ten points for black.]
)(;W[id]C[[KataGo variation.\]]
;B[kk]
;W[lk]
;B[ja]
;W[ha]
;B[ra]
;W[ip]
;B[jp]
)))(;W[of]TR[nd]C[I guess white's gonna have to play O16 later, anyway.

[KataGo gives this move -4.7 pp.\]]
(;B[ph]C[Because if black plays here, white has to respond.]
;W[ss]
;B[ne]
;W[oe]
;B[nd]C[Otherwise, white dies. So it's three points either way. Okay, cool, the AI is just better at vital points, don't mind me.]
)(;B[ne]C[[KataGo variation.\]]
;W[nd]
;B[fa]
;W[ii]
;B[hh]
;W[hi]
;B[da]
;W[ec]
;B[ea]
;W[ee]
;B[fd]
;W[in]
;B[eg]
;W[js]
;B[ks]
;W[is]
)))(;B[ss]
;W[la]
;B[lb]
;W[pa]C[It's not even a ko, just squeezing black's group for all it's worth.]
))(;B[ne]C[Black can take out the eye here.

[KataGo gives this move -1.5 pp.\]]
;W[of]
(;B[nd]
;W[pa]C[But then there's an atari.]
;B[la]
;W[rb]C[And then there's an eye here.]
)(;B[ph]C[[KataGo variation.\]]
;W[oe]
;B[rb]
;W[nd]
;B[kr]
;W[ej]
;B[ii]
;W[ij]
;B[fa]
;W[ec]
;B[fj]
;W[fi]
;B[ee]
;W[js]
;B[in]
;W[lq]
;B[lr]
;W[mp]
;B[pm]
;W[jp]
;B[ip]
;W[io]
)))(;W[nl]
;B[lj]TR[nl]C[If white cuts, this tesuji exists — got it. I understand now.]
))(;B[cr]C[It's not a ko anymore.

[KataGo gives this move -35 pp.\]]
;W[es]C[Black loses the capturing race.]
;B[fr]
;W[es]
;B[fs]
;W[hs]C[And there's no life in the corner.]
))(;W[jb]TR[kd]C[The idea of black's tiger's mouth is to fix the wedges.]
;B[ib]
;W[ic]
;B[kb]
;W[jd]
)(;W[fb]TR[kd]
;B[eb]
;W[fd]C[There's no wedge.]
))(;B[ss]
;W[dc]
;B[cb]
;W[eb]
;B[ec]
;W[fb]
;B[gb]
;W[gc]
;B[fa]
;W[fd]TR[cd]C[I believe C16 is looking at some aji around this cut.]
))(;B[bq]C[There's still this annoying ko! Is this not anything to worry about? I don't know, I guess save it for later.

[KataGo gives this move -3.2 pp.\]]
;W[cs]C[[KataGo variation.\]]
;B[dg]
;W[cg]
;B[cr]
;W[qq]
;B[qp]
;W[cs]
;B[ir]
;W[hs]
;B[cr]
;W[es]
;B[fq]
;W[cs]
;B[mk]
;W[ll]
;B[cr]
;W[fs]
;B[mc]
;W[qf]
;B[rf]
;W[rg]
;B[re]
;W[og]
;B[oh]
;W[ng]
;B[nh]
;W[mg]
;B[mh]
;W[lg]
;B[pd]
;W[ph]
;B[qh]
;W[pi]
;B[of]
;W[lh]
))(;W[cq]C[Let's just play this out.

[KataGo gives this move -28 pp.\]]
(;B[cs]
;W[bs]
;B[es]
;W[gs]
;B[fs]
;W[hs]C[So black can die.]
)(;B[hs]C[[KataGo variation.\]]
;W[is]
;B[gs]
;W[ir]
;B[fr]
;W[mp]
;B[jr]
;W[mr]
;B[lr]
)))(;B[bq]C[If black plays here, there's at least a ko.

[KataGo gives this move -26 pp.\]]
;W[cq]
;B[bs]
)(;B[bs]C[What if black starts taking liberties?

[KataGo gives this move -9.1 pp.\]]
(;W[bq]
;B[io]
;W[ep]
;B[ir]
;W[hs]
(;B[is]C[This should be "eye versus no eye".]
;W[fs]
;B[es]
;W[fr]C[Maybe it's not?]
)(;B[gs]C[I don't know!]
))(;W[cq]C[[KataGo variation.\]]
;B[fg]
;W[cs]
;B[dg]
;W[cg]
;B[cr]
;W[fs]
;B[gs]
;W[fr]
;B[df]
;W[de]
;B[cc]
;W[ch]
)))(;B[mp]SQ[no]SQ[np]TR[kq]C[Maybe L3 can say that white doesn't care about the O4 stones.

[KataGo gives this move -16 pp.\]]
(;W[lp]
;B[mo]
;W[lo]
;B[nn]
(;W[iq]SQ[no]SQ[np]TR[kq]C[Because white can make life here.]
)(;W[jp]
;B[ko]
;W[kn]
;B[jo]C[Or maybe white can squeeze this somehow.]
)(;W[io]C[Or maybe white can peep?]
;B[jp]
;W[ho]
;B[jn]
;W[fp]
;B[eo]
;W[gq]C[There's so much to read!]
))(;W[ho]C[[KataGo variation.\]]
;B[io]
;W[hp]
;B[ip]
;W[hq]
;B[iq]
;W[gq]
;B[ir]
;W[er]
;B[fr]
;W[fq]
;B[dr]
;W[ep]
;B[dq]
)))(;W[lp]C[[KataGo gives this move -3.3 pp.\]]
;B[kq]C[[KataGo variation.\]]
;W[og]
;B[oh]
;W[nh]
;B[oi]
;W[ng]
;B[fg]
;W[cc]
;B[hg]
;W[fb]
;B[ed]
;W[eb]
;B[ec]
;W[gc]
;B[dc]
;W[de]
;B[dg]
;W[cg]
;B[dh]
;W[ch]
)(;W[mn]C[[KataGo gives this move -14 pp.\]]
))(;W[qr]C[Life in the corner should be pretty easy.

[KataGo gives this move -5.8 pp.\]]
(;B[pr]
;W[rm]
;B[rl]
;W[rr]C[Hard to say, maybe here?]
;B[qp]C[Reduce.]
;W[ro]
;B[qs]C[Reduce.]
;W[rs]C[And this should be alive.]
)(;B[rq]C[[KataGo variation.\]]
;W[pl]
;B[qm]
;W[cc]
;B[ed]
;W[kq]
;B[lq]
;W[lr]
;B[jq]
;W[kr]
;B[mq]
;W[jp]
;B[iq]
)))(;W[ec]C[Another option to consider for white is to kick, followed by a two-space high pincer.

[KataGo gives this move -8.8 percentage points winrate (pp).\]]
;B[fd]
(;W[id]C[This is a common variation.]
)(;W[ff]C[[KataGo variation.\]]
;B[jd]
;W[lc]
;B[kc]
)))(;W[dq]C[Normally it would be this one.

[Games 56, 57, 61, 95 and 96 have continued here.

All comments on this game by Clossius (AGA 3d), except for bracketed comments by Maharani.\]]
))