(;FF[4]
CA[UTF-8]
GM[1]
GN[ELF OG v2 + KataGo self-play]
PC[https://online-go.com/review/433241]
PB[ELF OpenGo v2 and KataGo]
PW[ELF OpenGo v2 and KataGo]
BR[9p]
WR[9p]
OT[Error: time control missing]
RE[?]
SZ[19]
KM[7.5]
RU[Chinese]

;B[qd]
;W[dc]C[After one million playouts by LeelaZero v0.17 with converted ELF OpenGo v2 weights (thanks, mark5000) at move 2, the sequence from move 3 - 26 is the principal variation. At move 27, high-playout KataGo takes over. All comments on this game by Anton Christenson, EGF 3d.
]

(;B[cd]SQ[cc]TR[dd]TR[ce]TR[de]TR[cf]TR[df]C[There are four standard approach moves here. In modern times, we also sometimes see the attach move in some situations, but that's basically it. It's super rare to see a move like C16 because it seems like a really bad exchange when white just responds at C17.
]

;W[cc]
(;B[dq]
;W[qp]
;B[oq]
;W[pn]
(;B[pc]TR[nc]TR[oc]TR[nd]TR[od]C[There are only four "reasonable" enclosures we normally see. Instead, we have this super rare micro-enclosure, which we'd normally say is too slow. It's clearly more solid and says, "I really want this territory and I'm not giving it up for any reason", but it seems a bit inefficient - so small! So we never see pros play something like this. Turns out, maybe it's not so bad.
]

(;W[cf]TR[cc]TR[cd]C[If we remove the earlier exchange, this is kind of a large knight's enclosure. It seems like almost the only option white has where C16 is not clearly a good exchange for black, so C16 is probably more playable than you would think. You can probably play it in your game and it won't be too bad.
]

;B[cp]TR[pc]TR[cd]C[Black even does it another time! White has played all normal moves and black has played three dubious moves, but according to KataGo, black is still only 4 or 5 points behind (compared to 1 or 2 points if black had also played only normal moves), so it's very minor. I wouldn't be able to tell the difference in a game at my level, certainly, so I might start trying this out to annoy my opponents a bit. It looks like fun!
]

;W[kq]
;B[qq]C[Black settles in the corner immediately.
]

;W[rq]
(;B[qr]TR[pp]C[The drawback is that there is a lot less eye-shape for black in the game after white responds.
]

;W[pp]
;B[mq]C[So black takes the extension.
]

;W[np]C[White continues putting pressure with this peep.
]

;B[nq]C[This group will have no trouble living, but it is a little bit uncomfortable. Black has to keep an eye on the group's eyes.
]

(;W[qj]TR[kq]C[But white just leaves it for now and makes an extension which focuses more on making points here, but maybe it leaves a little bit more weakness at L3.
]

;B[hp]TR[kq]C[Already, black sort of pincers the stone.
]

;W[ko]C[When white jumps out, white is attacking black more than the other way around. I think it makes sense for white to play in this way. White has made a wall and a nice extension.
]

;B[op]C[Black pushes to make sure that this group has eyes.
]

(;W[nn]TR[np]C[So it's better for white to jump lightly like this. Don't care about the O4 stone!
]

(;B[dg]
;W[cg]TR[pc]TR[cd]TR[cp]C[White 57 % (ELF OpenGo v2)
White 65 % (KataGo)
White 71 % (LeeLa Zero v0.17)
]

;B[di]TR[cg]TR[dg]CR[cp]CR[hp]SQ[op]CR[dq]SQ[mq]SQ[nq]SQ[oq]SQ[qq]SQ[qr]C[This is where high-playout KataGo takes over. Moves 25 - 27 are a very standard sequence. Black is starting to build some moyo in the lower left and ignores the lower-right group.
]

;W[df]
;B[jc]
;W[pe]TR[pc]TR[qd]C[This move might make you reconsider how much you like the black enclosure. It's quite a nice shape point for white to get here.
]

(;B[qe]C[Black is sort of obliged to play here.
]

;W[pg]LB[pc:1]LB[qd:5]LB[pe:2]LB[qe:3]LB[qf:4]LB[pg:6]TR[di]C[White jumps away at 6 similarly to D11. If white gets a move at R14, this would remind me of something similar to the numbered joseki. That's a joseki where white got to approach immediately after black played at 1. That we are so close to an even joseki even though black got the first two moves in the corner is a small indication for me that black Q17 was a little bit suboptimal, but white still needs another move to get there so it's still fine for black.
]

;B[fd]TR[cd]SQ[dd]C[One fun part of this game is that we're gonna see black using this C16 stone again and again. You often have this theme in AI games that they are so good at using every bit of aji from their stones and we're gonna see that very clearly with this stone, I think. Now black is threatening to just pull it out with a stone at D16.
]

;W[ee]TR[cd]LB[fd:A]LB[ee:B]C[So white has to add a move at B to keep it dead, and I think looking at this exchange, it's pretty clear that A is more useful for black than B is for white, so this feels nice for black.
]

;B[pk]TR[rr]C[For now, black can take the time to reduce the right side which was starting to look quite interesting for white. The lower right corner is still fine for now but it's going to change once white plays S2, which will happen pretty soon.
]

;W[pj]
;B[nk]CR[qe]CR[cg]CR[pg]CR[di]CR[pj]C[The third time in the game that we have this push and jump shape.
]

;W[rr]SQ[nk]SQ[pk]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[qr]C[If black is going to have to struggle a little bit here to live with this group then white is going to have a stronger attack on the two outside stones, so this is nice timing for white to start here.
]

(;B[iq]XX[no]XX[np]SQ[op]SQ[mq]SQ[nq]SQ[oq]SQ[qq]TR[kr]SQ[qr]C[Instead, black comes from the outside like this, threatening to connect up under with something like L2. This is an indirect way of helping the group which is better for points than just directly taking one stone or something like that.
]

;W[ll]SQ[nk]SQ[pk]TR[nn]TR[ko]C[White plays a loose move on the outside. This is taking care of the marked connection, making sure there's no cutting for black and also putting pressure on the two stones.
]

;B[oj]TR[nk]TR[pk]C[Black moves away with these stones.
]

;W[lr]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[qr]C[White puts pressure on this group. I would be scared of dying here as black, but there's just enough weakness in white's outside that that's not going to happen.
]

;B[no]
(;W[mn]TR[np]C[White is not going to care about the O4 stone.
]

;B[mo]C[White just lets black make an eye here, but it's still a bit unclear if black is going to get another eye on the edge.
]

(;W[qk]TR[mo]TR[no]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[qr]C[It seems like black is also slightly unwilling to go for a cut immediately while still not being settled with the marked group.
]

;B[mr]C[So black makes two eyes first, threatening to go for some cut.
]

;W[lo]SQ[mn]SQ[nn]SQ[pn]SQ[ko]TR[mo]TR[no]TR[op]SQ[kq]TR[mq]TR[nq]TR[oq]TR[qq]SQ[lr]TR[mr]TR[qr]C[White gets to protect here with a very slow-looking gote move that is not threatening to kill anything. A very peaceful result in the end. Black got to live and white got to make a strong connection shape on the outside.
]

;B[dl]SQ[oj]SQ[nk]SQ[pk]TR[mo]TR[no]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[mr]TR[qr]C[But it's still a bit surprising that black just plays away. It felt like white had a sort of dual splitting attack going. One group managed to live, the other group looks... okay.
]

(;W[oo]TR[lo]C[It's nice that white gets this little bit of extra thickness. Two moves for the price of one in gote or something like that.
]

;B[mp]
;W[mj]TR[oj]TR[nk]TR[pk]SQ[dl]C[White is following up attacking these stones. Black got D8 instead.
]

;B[jm]TR[oj]TR[nk]TR[pk]CR[dl]SQ[lm]XX[in]SQ[ln]XX[io]CR[cp]CR[hp]CR[dq]CR[iq]C[Black still doesn't save the stones directly and plays from the other side, building the left and poking a little bit at the cutting weakness, but black's connection isn't really in good shape.
]

(;W[io]XX[pc]XX[qd]XX[pe]SQ[dl]TR[jm]SQ[cp]SQ[hp]SQ[dq]SQ[iq]C[So white can just poke here with this very nice move. It's separating the black stone and also moving into black's area. Also, this is the same vital point as in the Q15 shape that's always a nice shape point locally to take away the tiger's mouth.
]

;B[om]C[White doesn't respond to this peep directly.
]

;W[ho]C[White just pushes here.
]

(;B[jp]
;W[jo]SQ[hp]SQ[jp]TR[kp]TR[lp]SQ[iq]C[White has to connect on this side. There's sort of a cut here, but it's going to be very hard for black with this weak shape, so it's not really a concern at the moment.
]

;B[oh]TR[oj]TR[nk]TR[pk]TR[om]SQ[on]XX[hp]XX[jp]XX[iq]C[Black continues moving away with this group, now actually threatening to cut soon because it's doing more of anything if black is actually saving these stones. There's no point in cutting if they're all going to just get swallowed up anyway. But we notice that black took sente and left a very weak shape on the left.
]

;W[gp]C[So white gets to hane like this.
]

;B[gq]
;W[fp]
;B[fq]
;W[on]
(;B[mk]TR[oh]TR[oj]TR[nk]TR[pk]TR[om]C[For now, black is going to continue taking care of this group.
]

;W[lk]
;B[lj]C[Black does so by cutting here, creating sort of another weak group.
]

;W[mi]SQ[oh]TR[mj]SQ[oj]SQ[mk]SQ[nk]SQ[pk]SQ[om]C[The idea here is that white right now also has to worry about this stone. Black's group has a running buddy which also has to fear for its life.
]

;B[og]TR[pe]SQ[pf]TR[pg]C[We get a very interesting shape here with this jump. Black is threatening to wedge. We very rarely see something like this.
]

;W[kd]SQ[og]SQ[oh]TR[mi]TR[mj]SQ[oj]SQ[mk]SQ[nk]SQ[pk]SQ[om]C[This is sort of a leaning attack, borrowing some strength here for white's weak group, setting up to attack the marked black group.
]

;B[jd]C[Black has to respond here locally.
]

;W[kf]TR[kd]TR[pe]TR[dg]SQ[og]TR[pg]SQ[oh]TR[di]SQ[oj]SQ[mk]TR[nk]TR[pk]SQ[om]C[We get our fourth jump like this in this game. In this case, it helps to surround the black group.
]

;B[pi]
;W[qi]
;B[pf]TR[qd]TR[qe]TR[og]XX[qg]TR[oh]SQ[ph]SQ[oi]C[Very interesting snake-like shape, going through like this. First of all, white wants to exchange the atari. Otherwise black is threatening to play R13 which would be great shape for black.
]

;W[ph]
;B[oi]TR[of]TR[qf]C[What I find really cool about this shape is that it's really tempting now to play another atari, but either one would actually be a really bad exchange.
]

(;W[je]SQ[kd]SQ[kf]XX[of]TR[pg]TR[ph]XX[kj]C[At this point, white has to sort of leave it. There's no way for black to capture the two stones anyway, so white takes care of the shape here. It sort of looks like black is really well connected, but there is still a potential to cut here in the future. K15 is also threatening a ladder at L10.
]

(;B[kj]C[So it makes sense that black chooses this moment to pull out.
]

;W[lh]SQ[kd]SQ[je]SQ[kf]TR[mi]TR[mj]C[White saves these two stones and stays connected.
]

;B[hd]TR[jc]TR[jd]C[Black helps this group.
]

;W[kc]
;B[ec]TR[cd]C[This is again using the C16 move. It might look like white can just ignore this, and it wouldn't be the end of the world if white does.
]

(;W[eb]C[Luckily for white, it can actually handle this in sente.
]

;B[fb]
;W[ed]
;B[fc]TR[eb]SQ[fb]SQ[ec]SQ[jc]SQ[fd]SQ[hd]SQ[jd]C[White keeps sente but black got some extra strength for free here. Most importantly, black can capture E18 to make this group strong, which will be important in a minute.
]

;W[ep]TR[kp]TR[lp]C[This point is really huge locally for the shape because if black got it, the lower left would be very solid black territory. When white plays here, it activates some aji with the push and cut and also sort of opens up the left, so black will need another move here if it wants to keep its points. This makes it even more unreasonable for black to think about the cut.
]

;B[le]
;W[ke]
;B[kh]
;W[lg]
(;B[rg]SQ[of]TR[pg]XX[ch]TR[ph]C[This is a nice local shape move to threaten to capture two stones, but it feels a little bit bad to me. It's sort of just a big move on points, I guess. Not quite sure what to think about it. It feels like if black wanted to play in this area, black should have fixed the cut in some way even though that would probably be just as slow. There are still points like C12 on the board, for instance, but I'm not as strong as KataGo.
]

(;W[eq]
;B[er]
;W[pl]
;B[hi]XX[kc]XX[kd]XX[je]XX[ke]XX[kf]XX[lg]TR[kh]XX[lh]SQ[di]XX[mi]TR[kj]TR[lj]XX[mj]SQ[dl]SQ[jm]C[This is sort of a nice multi-purpose move. It helps take care of the stones on the right and it starts to maybe make a few (if any) points in the center in the future which could be a nice bonus. Mainly, it's putting pressure on the white group. The whole stick doesn't really have any eyes.
]

;W[nc]TR[of]C[So white sort of has to play an extension here to take care of that. But now suddenly the cut is activated again.
]

(;B[ol]TR[og]TR[oh]TR[oi]TR[pi]TR[oj]TR[mk]TR[nk]TR[pk]TR[om]C[This is indirectly creating some other options for this group, but black is sort of playing with fire now.
]

(;W[dr]C[I find this timing interesting.
]

(;B[cr]C[Maybe this leaves slightly fewer ko threats (don't quote me on that, just my wild guess).
]

;W[pm]TR[lq]C[White connects peacefully like this. Now it's completely clear that the cut is not working.
]

(;B[lm]XX[of]SQ[og]SQ[oh]SQ[oi]SQ[pi]SQ[oj]SQ[mk]SQ[nk]SQ[pk]SQ[ol]TR[jm]SQ[om]C[Ever since 53. K7, this has been the shape follow-up, but black is asking for a free forcing move while the right group is quite weak with the potential cut.
]

(;W[jl]C[White answers by fighting back.
]

;B[jk]
;W[km]
;B[il]SQ[og]SQ[oh]SQ[oi]SQ[pi]SQ[oj]SQ[mk]SQ[nk]SQ[pk]TR[jl]SQ[ol]TR[km]SQ[om]C[A really fighting spirit sequence from white, trying to make some quite bad exchanges actually, leaving one stone in atari. Quite a bad shape locally, but it's going to be worth it if white gets a nice attack on the black group.
]

;W[of]
;B[qf]
;W[qh]SQ[og]SQ[oh]SQ[oi]SQ[pi]SQ[oj]TR[lk]SQ[mk]SQ[nk]SQ[pk]TR[ll]SQ[ol]SQ[om]C[We're now in a situation where black is cut off. It can't capture the lower cutting stones and has no eyes.
]

(;B[nf]TR[pe]TR[of]C[The forcing moves against these cutting stones are black's only redemption in this position.
]

;W[oe]
;B[ne]
;W[od]
;B[mg]LB[mf:A]LB[ng:B]C[It's not two eyes yet, but we're starting to see where the eyes might come from if black gets two moves in a row.
]

;W[jb]TR[mb]SQ[jc]TR[kc]TR[mc]TR[nc]SQ[hd]SQ[jd]TR[kd]TR[od]TR[je]TR[ke]SQ[le]SQ[ne]TR[oe]TR[pe]TR[kf]SQ[nf]TR[of]TR[lg]SQ[mg]SQ[og]SQ[kh]TR[lh]SQ[oh]SQ[hi]TR[mi]SQ[oi]SQ[pi]TR[mj]SQ[oj]SQ[mk]SQ[nk]SQ[pk]SQ[ol]SQ[om]C[Thankfully for black, white actually has to take care of this group and makes a safe move. White can't really afford to go all in on an attack right now when there are all these weaknesses like N17 or N18. This whole group might not have any eyes and black is sort of strong on the outside.
]

;B[kl]TR[jb]C[Black ignores the hane.
]

;W[ml]SQ[le]SQ[ne]SQ[nf]SQ[mg]SQ[og]SQ[kh]SQ[oh]SQ[hi]XX[mi]SQ[oi]SQ[pi]SQ[kj]SQ[lj]XX[mj]SQ[oj]SQ[jk]TR[lk]SQ[mk]SQ[nk]SQ[pk]SQ[il]SQ[kl]TR[ll]SQ[ol]SQ[jm]SQ[om]XX[ln]C[What's the point of saving here if black is already going to live on the right and super strong on the left? In the end, black has the option to capture the N10 stones, anyway, which looks about the same size. Perhaps the point is that white will need the connection at M6, anyway.
]

;B[kn]TR[jl]TR[km]C[Black gets a double ponnuki.
]

;W[ln]TR[lk]TR[ll]TR[ml]
;B[lf]
;W[kg]
;B[mh]TR[li]SQ[mi]SQ[mj]C[At this point, black is already completely alive and white has to tenuki. It's too small to save the two stones.
]

(;W[hc]SQ[fb]TR[jb]SQ[ec]SQ[fc]SQ[jc]TR[kc]TR[nc]SQ[fd]SQ[hd]SQ[jd]TR[kd]TR[od]TR[je]TR[ke]TR[oe]TR[pe]TR[kf]TR[of]TR[kg]TR[lg]TR[lh]C[This is helping the white group settle and also putting some pressure on the black group.
]

;B[gn]C[Black gets this peep in before coming back to the top.
]

;W[go]
;B[hb]
;W[ib]
;B[gc]LB[ic:A]LB[id:B]C[Would you play at A or B in this position?
]

(;W[ic]TR[ha]SQ[jc]SQ[jd]C[It's better to just connect because now white is threatening H19 in the future to take away the eye. Black is not going to get to save the two stones, anyway, as we will see in the game.
]

;B[db]
;W[cb]
;B[ea]
;W[ob]
(;B[ch]
;W[li]TR[ok]C[This sets up the ko.
]

;B[pb]TR[oc]C[Black can push at P17 later.
]

;W[eg]TR[ea]SQ[ha]TR[db]TR[fb]TR[hb]TR[ec]TR[fc]TR[gc]TR[fd]TR[hd]C[This is setting up to bully the black group a little more. For now, it looks like it's pretty strong, but once white gets the H19 atari, it won't have two clear eyes and struggle a little bit to live, which will give white some free endgame in the process.
]

;B[dn]SQ[co]TR[ep]C[Protecting against the weaknesses that 88. E4 was setting up. An attachment at C5 for instance would have been a nice move for white, maybe. Black puts an end to that.
]

;W[ha]C[White is not going to kill this group, but is going to bully it a little bit.
]

;B[gf]
;W[qs]TR[mo]TR[no]TR[mp]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[mr]TR[qr]C[White bullies this group a little bit, too.
]

;B[jr]SQ[kp]XX[fq]XX[gq]TR[qq]SQ[kr]TR[qr]C[Black could have just lived locally, but is saying that if white takes the two stones, black is in some way either going to cut or connect. Black may even be willing to sacrifice the G3 stones at this point.
]

(;W[ei]TR[di]TR[hi]TR[dl]TR[il]TR[dn]TR[gn]C[White reduces the black area.
]

;B[bg]
;W[dh]
;B[ci]
;W[bf]
;B[fh]
;W[eh]
(;B[ld]C[Black doesn't need a move to live here. I'm not quite sure of the reasoning behind M16. It does feel sort of sente, but it's confusing because white doesn't respond right away.
]

;W[jq]TR[mo]TR[no]TR[mp]TR[op]TR[mq]TR[nq]TR[oq]TR[mr]C[This doesn't seem like it's sente, but the black group is relying on the potential connection here.
]

(;B[ir]
;W[em]
;B[en]
;W[ip]
(;B[hq]TR[mo]TR[no]SQ[hp]TR[mp]TR[op]SQ[fq]SQ[gq]SQ[iq]TR[mq]TR[nq]TR[oq]TR[qq]SQ[ir]SQ[jr]TR[mr]TR[qr]C[Both sides live for now.
]

(;W[dp]
;B[cq]
;W[co]C[This is a really fun part.
]

(;B[cn]
(;W[nd]C[White protects the upper side. There's still a lot of weird stuff happening here; let's not worry about the details.
]

;B[bh]
;W[pa]
;B[qa]
;W[oa]
;B[rb]
;W[fr]
;B[ds]
;W[fj]
;B[ek]
;W[hg]
;B[hf]
;W[ig]TR[ea]TR[db]TR[fb]TR[hb]TR[ec]TR[fc]TR[gc]TR[jc]TR[fd]TR[hd]TR[jd]TR[gf]TR[hf]C[White still keeps black worrying a little bit about this group.
]

;B[gb]LB[fa:A]LB[eb:C]LB[ge:B]C[Black makes miai of making an eye at A or around B in addition to having the eye at C.
]

;W[ca]C[White gets to descend in sente.
]

;B[da]
;W[ej]
;B[dj]
;W[id]SQ[ca]TR[da]TR[ea]TR[db]TR[fb]TR[gb]TR[hb]TR[ec]TR[fc]TR[gc]TR[fd]TR[hd]TR[gf]TR[hf]C[Even capturing here is sente. Very nice how white bullied the black group and got a nice endgame move at C19.
]

;B[ga]
;W[fk]
;B[gl]
;W[kr]TR[mo]TR[no]TR[mp]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[mr]TR[qr]C[Black needs to live now. How would you do it? There are at least two ways, but one is slightly better on points.
]

(;B[os]TR[pr]C[Black makes around 3.5 points compared to exactly 2 points in the other variation.
]

(;W[fl]
;B[sj]
;W[si]
;B[ri]
;W[rh]
;B[sh]
;W[rj]
;B[si]
;W[sk]
;B[sg]SQ[ok]TR[rl]C[Black has a follow-up at S8 once the ko is connected.
]

;W[hj]
;B[fn]
;W[hm]
;B[hk]
;W[ok]
;B[rl]
(;W[do]C[This looks like a ko threat to me, but white plays it just before starting the ko.
]

;B[bo]
(;W[nj]
;B[pk]
;W[pq]
(;B[pr]
;W[ok]
(;B[nh]TR[ld]TR[le]TR[ne]TR[lf]TR[nf]TR[mg]TR[og]TR[mh]TR[oh]TR[oi]TR[pi]TR[oj]C[I find this really interesting. This is not a ko threat. Black backs off because the ko is really heavy for black. If white wins it, the follow-up would kill the whole black group.
]

;W[rk]SQ[pk]TR[ql]C[White also backs off, but white will need to connect again after black captures the ko.
]

;B[pk]
;W[dm]
;B[cm]
;W[ok]
;B[rp]
;W[ro]
;B[pk]
;W[el]
;B[dk]
;W[ok]
;B[qo]
;W[sp]
;B[pk]
;W[ms]
;B[ns]
;W[ok]
;B[md]C[At first glance, it's not obvious to me that this is a ko threat. It looks like even if black plays next here, white can still connect.
]

;W[gr]C[How would black follow up on the top?
]

(;B[hs]TR[gr]C[I have no idea about this exchange.
]

;W[mb]C[White defends.
]

;B[pk]
;W[ij]TR[kj]TR[lj]TR[hk]TR[jk]TR[gl]TR[il]TR[kl]TR[jm]TR[lm]TR[kn]C[Instead of making points, black is just barely alive with the center group after all the moves white got here.
]

;B[hn]TR[gm]C[Black protects here; otherwise, white would cut at G7 next.
]

;W[ok]
;B[lp]
;W[kp]
;B[pk]
;W[gm]C[This is another point-losing ko threat, or at least it seems so.
]

;B[fm]
;W[ok]
;B[af]
;W[ae]
;B[pk]
;W[gk]TR[gm]C[Maybe G7 didn't lose points after all. Since the ko is so complicated, it's very hard to say.
]

;B[ag]
;W[hl]C[This is white sente.
]

;B[in]
;W[be]
;B[ok]
;W[ql]XX[pg]XX[ph]XX[qh]XX[rh]TR[ei]XX[qi]TR[ej]TR[fj]XX[pj]XX[qj]XX[rj]TR[fk]TR[gk]SQ[mk]SQ[nk]SQ[ok]SQ[pk]XX[qk]XX[rk]XX[sk]TR[el]TR[fl]TR[hl]SQ[ol]TR[dm]TR[em]TR[gm]TR[hm]SQ[om]C[Very peaceful solution to such a complicated situation. White got some profit in the center, black got to connect back and white got to save all the points on the right.
]

;B[gg]C[This is a cool little tesuji. How would you play as white to keep as many points as possible in the center?
]

(;W[gh]C[This tesuji gains two points in comparison. A nice move to appreciate compared to all the complicated ko fighting that goes way above my head.
]

(;B[fg]C[The best black can do.
]

;W[fi]TR[hh]TR[gi]C[White has made two extra points compared to the other variation.
]

;B[ki]
;W[jh]
;B[ik]
;W[rs]
;B[ps]C[We're basically at the end of the game, but it's still super close.
]

;W[he]
;B[ge]TR[da]TR[ea]TR[ga]TR[db]TR[fb]TR[gb]TR[hb]TR[ec]TR[fc]TR[gc]TR[fd]TR[gf]TR[hf]TR[fg]TR[gg]TR[fh]C[This group is completely alive.
]

(;W[if]
(;B[mc]
;W[gd]SQ[ge]SQ[gf]SQ[hf]TR[if]SQ[fg]SQ[gg]SQ[fh]C[J14 looks a bit wasted if white wants to kill these stones now.
]

;B[lb]
;W[lc]
;B[nb]
;W[na]
;B[oc]TR[nc]TR[nd]TR[od]TR[oe]TR[pe]TR[of]C[Black has killed these stones.
]

;W[fe]C[As compensation, white gets to start a ko. Looks like a desperate lost attempt.
]

;B[hd]
;W[me]
;B[mf]
;W[gd]
;B[dd]C[This is an interesting point where white needs to choose carefully how to minimize the ko threats.
]

(;W[bc]
(;B[hd]
;W[ks]C[This is sente to win the capturing race.
]

;B[fs]
;W[gd]
;B[mm]
;W[nm]
;B[hd]
;W[nr]
;B[or]
;W[gd]
;B[km]TR[lk]TR[ll]TR[ml]C[Black connects, threatening to capture three stones.
]

(;W[jj]SQ[ge]SQ[gf]SQ[hf]SQ[fg]SQ[gg]SQ[fh]TR[ji]C[This makes no sense at all to me. I think this must be a crazy move because everything is losing for white. Its only redeeming quality is that now there is a ko threat at K11, but white had the opportunity to win the ko and take the stones if it had wanted to, anyway.
]

;B[hd]
;W[qb]
;B[qc]
;W[gd]
;B[nl]TR[lk]TR[ll]TR[ml]C[Black just takes the stones.
]

;W[ji]C[White doesn't even use this as a ko threat.
]

;B[hd]
;W[ra]
;B[kk]
;W[qb]
;B[pd]
(;W[gd]
;B[nb]
;W[nc]
;B[nd]C[Black 95 % (KataGo)

-- chat --
Maharani: Video review by Anton Christenson, EGF 3d: https://www.youtube.com/watch?v=RQ1piEevIp8

]

)(;W[nb]
;B[qa]C[White can't save the stones.
]

))(;W[ff]C[White should just finish the ko.
]

;B[jj]
;W[js]
;B[is]
;W[qg]
;B[rf]
;W[ji]
;B[nl]
;W[la]
;B[pd]
;W[ia]
;B[nb]
;W[nc]
;B[nd]
;W[ma]
;B[ls]
;W[ka]
;B[ms]
;W[gl]C[KataGo gives white -1 pp for 288. F14
]

)(;W[nl]C[Or white could save the three stones.
]

;B[hd]
;W[js]
;B[is]
;W[gd]
;B[ii]
;W[ji]
;B[hd]
;W[qg]
;B[rf]
;W[gd]
;B[jj]
;W[gj]
;B[hd]
;W[bn]
;B[bp]
;W[gd]
;B[po]
;W[rp]
;B[hd]
;W[qb]
;B[qc]
;W[gd]
;B[la]
;W[kb]
;B[hd]
;W[gs]
;B[hr]
;W[gd]
;B[jg]
;W[ih]
;B[hd]
;W[ra]
;B[gd]
;W[qb]
;B[pd]
;W[qa]
;B[nb]C[KataGo gives white -0.6 pp for 288. O8
]

))(;B[de]
;W[ff]C[White could just capture here anyway.
]

))(;W[bd]
;B[bc]C[Black will get one more ko threat however white responds.
]

))(;B[ie]
;W[mm]
;B[js]
;W[ks]
;B[fs]
;W[km]
;B[dd]
;W[bc]
;B[lm]
;W[qg]
;B[rf]
;W[km]
;B[im]C[KataGo gives black -10 pp for 263. J15
]

))(;W[gd]
;B[dd]
;W[de]
;B[hd]
;W[qg]
;B[rf]
;W[gd]
;B[fe]
;W[if]
;B[js]
;W[ks]
;B[fs]
;W[ia]
;B[hd]
;W[gl]
;B[ie]
;W[mm]
;B[km]
;W[im]
;B[jn]
;W[kk]
;B[jj]C[KataGo gives white -2 pp for 262. G16
]

))(;B[hh]
;W[gi]C[This doesn't do anything for black.
]

)(;B[gi]
;W[hh]C[This doesn't do anything for black.
]

;B[ki]
;W[jh]
;B[fg]
;W[fi]
;B[ik]
;W[he]
;B[ge]
;W[gd]
;B[fe]
;W[qg]
;B[rf]
;W[js]C[KataGo gives black -10 pp for 253. G11
]

))(;W[gi]
;B[gh]
;W[hh]TR[hi]C[Killing this stone is not as big.
]

))(;B[lb]
;W[lc]
;B[mc]
;W[mb]
(;B[nb]
;W[na]
;B[oc]
;W[la]
;B[pd]TR[nc]TR[nd]TR[od]TR[oe]TR[pe]TR[of]C[Black gets to capture the tail.
]

(;W[nb]
;B[ma]C[White can't connect.
]

)(;W[kb]
;B[nb]
;W[nc]C[Black gets quite a few points here.
]

))(;B[hs]
;W[js]
;B[fs]
;W[ks]
;B[is]
;W[ij]
;B[hn]
;W[nl]
;B[gi]
;W[gk]C[KataGo gives black -33 pp for 229. M18
]

))(;B[mb]C[It's slightly better on points to play this one which is working in a different way.
]

;W[mc]
(;B[lc]C[Black should not play here.
]

;W[lb]C[This is completely not working for black.
]

;B[la]
;W[kb]
;B[fs]
;W[ij]
;B[hn]
;W[nb]
;B[pk]C[KataGo gives black -38 pp for 229. N18
]

)(;B[nb]C[Black should start here. Such a weird tesuji!
]

(;W[oc]
;B[lc]TR[oa]TR[pa]TR[ob]TR[mc]TR[nc]TR[oc]TR[nd]TR[od]TR[oe]TR[pe]TR[of]C[The white group is dead.
]

)(;W[lc]C[White has to just connect at this time.
]

;B[oc]C[Black gets to capture. Black makes around 1 more point than in the other variation.
]

))))(;B[ni]C[This move makes one extra point, but leaves one extra ko threat.
]

(;W[nh]
;B[ng]
)(;W[rk]
;B[pk]
;W[ij]
;B[hn]
;W[ok]
;B[af]
;W[ae]
;B[pk]
;W[rs]
;B[ms]
;W[ok]
;B[ag]
;W[be]
;B[pk]
;W[dm]
;B[cm]
;W[ok]
;B[dd]
;W[bc]
;B[pk]
;W[el]
;B[dk]
;W[ok]
;B[md]
;W[mc]
;B[pk]
;W[gr]
;B[fs]
;W[ok]
;B[lc]C[KataGo gives black -11 pp for 209. O11
]

)))(;B[ok]TR[pg]TR[ph]TR[qh]TR[rh]TR[qi]TR[pj]TR[qj]TR[rj]TR[qk]TR[sk]C[This would threaten the white group on the right.
]

;W[or]
;B[ql]SQ[pg]SQ[ph]SQ[qh]SQ[rh]SQ[qi]SQ[pj]SQ[qj]SQ[rj]SQ[qk]SQ[sk]TR[mo]TR[no]TR[mp]TR[op]TR[mq]TR[nq]TR[oq]TR[mr]TR[os]C[This is a good trade for white, I guess.
]

))(;W[bn]TR[do]C[Maybe this follow-up is a bigger ko threat than D5? It wasn't played in the game, though, so I have no idea what's going on.
]

))(;W[nj]C[It seems like white should start the ko now.
]

;B[pk]
;W[dm]
;B[cm]
;W[ok]
;B[ni]
;W[do]
;B[bo]
;W[rk]
;B[pk]
;W[ij]
;B[hn]
;W[ok]
;B[lp]
;W[kp]
;B[pk]
;W[in]
;B[im]
;W[ok]
;B[md]
;W[mc]
;B[pk]
;W[pq]
;B[pr]
;W[ok]
;B[rp]
;W[ro]
;B[pk]
;W[fm]
;B[gm]
;W[ok]
;B[qo]
;W[sp]
;B[pk]
;W[rs]
;B[ms]
;W[ok]
;B[po]
;W[qn]
;B[pk]
;W[ps]
;B[or]
;W[ok]
;B[af]
;W[ae]
;B[pk]
;W[ck]
;B[dk]
;W[ok]
;B[ag]
;W[be]C[KataGo gives white -3 pp for 202. O10
]

))(;W[pq]
;B[pr]
;W[ms]C[This is gote for white.
]

;B[ns]
;W[ls]
))(;B[pr]C[This does live.
]

;W[ns]
;B[os]
;W[ms]
;B[nr]
;W[ls]
;B[pq]C[Black lives with two points.
]

;W[fl]
;B[sj]
;W[si]
;B[ri]
;W[rh]
;B[sh]
;W[rj]
;B[gk]
;W[ok]
;B[md]
;W[mc]
;B[pk]
;W[rs]
;B[ps]
;W[ok]
;B[af]
;W[ae]
;B[ag]
;W[bd]
;B[pk]
;W[do]C[KataGo gives black -5 pp for 185. Q2
]

))(;W[do]
;B[bo]TR[co]C[White is going to make this exchange later.
]

;W[nd]
;B[bh]
;W[pa]
;B[qa]
;W[oa]
;B[qb]
;W[fj]
;B[ek]
;W[hh]
;B[gi]
;W[fg]
;B[gg]
;W[gh]
;B[ih]
;W[hg]
;B[if]
;W[ig]
;B[jg]
;W[fi]
;B[hf]
;W[fh]C[KataGo gives white -1 pp for 162. D5
]

))(;B[do]TR[co]C[The bad shape move makes more points. It turns C5 into a mistake.
]

(;W[dm]
;B[cm]
;W[cl]
(;B[ck]
;W[cn]
;B[bl]
;W[fn]C[This would be a disaster for black.
]

)(;B[fm]
;W[dk]
;B[el]
;W[bk]TR[bg]TR[ch]TR[ci]TR[di]C[White is starting to separate the black group.
]

;B[ej]TR[bk]TR[dk]TR[cl]C[Black cuts off the white stones.
]

;W[dj]
;B[bj]
;W[bm]
;B[bn]
;W[ek]
;B[fj]
;W[fk]
;B[gk]
(;W[gj]
;B[fi]
;W[gl]
;B[hk]
;W[fl]
;B[gm]TR[dj]TR[bk]TR[dk]TR[ek]TR[fk]TR[cl]TR[fl]TR[gl]TR[bm]C[White is surrounded.
]

)(;W[fi]C[White can't save this as a ko threat.
]

;B[gj]
;W[ak]
;B[fl]
;W[ck]
;B[ah]
;W[aj]
;B[ai]
;W[af]
;B[bh]
;W[cj]
;B[al]C[We get a direct ko.
]

;W[bi]
;B[dd]
;W[bd]
;B[bj]TR[od]TR[oe]TR[pe]TR[of]LB[fi:A]LB[gj:B]LB[bl:C]C[A would not be a valid ko threat right now. Black would just capture at C. This ko looks like a disaster for black, but it's still an even game! Black may get the four marked white stones in compensation for losing it.
]

)))(;W[fr]
;B[ds]
;W[dm]
;B[cm]
;W[cl]
;B[fm]
;W[dk]
;B[el]
;W[ck]
;B[bj]
;W[bk]
;B[ej]
;W[dj]
;B[ah]
;W[ek]
;B[fj]
;W[fk]
;B[gk]
;W[ak]
;B[af]
;W[aj]
;B[ai]
;W[ae]
;B[be]C[KataGo gives black -1.4 pp for 161. D5
]

)))(;W[pq]
;B[or]
;W[pr]C[This is still gote for white.
]

;B[bh]
;W[ns]
;B[ms]
;W[os]
;B[kr]C[Black connects like this.
]

))(;B[bh]
;W[fr]
;B[ds]
;W[hq]
;B[gr]
;W[hr]
;B[fs]TR[mo]TR[no]TR[mp]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[mr]TR[qr]C[Allowing white to cut would be troublesome while the marked group still doesn't have two eyes.
]

))(;B[bh]
;W[fr]
;B[ds]
;W[gr]
;B[hq]
;W[ip]
;B[ir]
;W[kr]C[This is still sente.
]

;B[hs]C[Black needs a move to win the semeai. White has cut black off in sente.
]

;W[os]TR[mo]TR[no]TR[mp]TR[op]TR[mq]TR[nq]TR[oq]TR[qq]TR[mr]TR[qr]C[Now, any move here would suddenly just kill this group.
]

))(;B[ej]
(;W[me]C[This is not going to kill anything.
]

;B[md]
;W[ld]
;B[nd]
;W[oc]
;B[mf]C[Black is alive.
]

)(;W[jq]
;B[ir]
;W[fi]
;B[hg]
;W[ip]
;B[hq]
;W[he]
;B[ge]
;W[bh]
;B[bi]
;W[ag]
;B[fj]
;W[pa]
;B[qa]
;W[oa]
;B[qb]
;W[fn]
;B[sj]
;W[si]
;B[ri]
;W[rh]
;B[sh]
;W[rj]
;B[pr]
;W[si]
;B[sg]
;W[ai]
;B[bj]C[KataGo gives black -3 pp for 151. E10
]

)))(;W[pq]
(;B[or]TR[qq]TR[qr]C[It might be completely gote for white to take the two stones.
]

)(;B[kr]C[Black won't respond like this.
]

)))(;B[li]
;W[ok]C[White can no longer do anything with this ko.
]

;B[nl]TR[nj]C[White can no longer atari from the inside.
]

))(;W[id]TR[jc]TR[jd]C[Wouldn't capturing the two stones be better for points?
]

(;B[ic]
;W[hc]
)(;B[db]
;W[cb]
;B[ea]
(;W[ha]TR[hc]C[This is a little less effective if H17 is in atari.
]

;B[ic]
;W[hc]
;B[ic]
;W[ga]
;B[ie]TR[ic]C[Since black captured back, white didn't gain that many points, anyway.
]

)(;W[ob]
;B[bh]
;W[li]
;B[pb]
;W[nd]
;B[ld]
;W[pa]
;B[qa]C[KataGo gives white -12 pp for 130. J16
]

))))(;W[li]TR[mi]TR[mj]SQ[jl]SQ[km]C[Even if white were to the save the two cutting stones, black could now tenuki. This raises an interesting question of whether white should have tenuki'd earlier. It looks like black got a good deal with the ponnukis.
]

;B[ib]
(;W[me]C[It might look like there's a little bit of aji here.
]

;B[md]
;W[ld]
(;B[mf]
;W[nd]C[We can go for something like this.
]

)(;B[nd]C[But of course this move is black sente.
]

;W[oc]
;B[mf]C[It's not working for white.
]

))(;W[ob]
;B[pb]
;W[kb]
;B[gn]
;W[go]
;B[bh]C[KataGo gives white -30 pp for 124. N11
]

)))(;B[kl]TR[lk]TR[ll]C[The magic here is that black can capture one stone like this, but it's not possible to capture the main cutting stones.
]

;W[ml]C[That's what white was doing.
]

;B[nf]
;W[oe]
;B[kn]
;W[ln]
;B[ne]
;W[od]C[KataGo gives black -30 pp for 111. L8
]

)(;B[kn]TR[kl]C[White is not going to connect at L8.
]

;W[ml]
;B[kl]
;W[ln]C[Instead, white can just play like this to save the main stones.
]

))(;W[mm]SQ[ml]TR[lm]C[White could just block like this, but this is sort of allowing black to get a nice forcing exchange in. Black can't cut right now, though.
]

(;B[kl]C[Maybe this is the follow-up later on, I'm not quite sure.
]

;W[ml]
;B[gn]
;W[go]
;B[jq]C[KataGo gives white -26 pp for 104. N7
]

)(;B[km]TR[ml]TR[ln]SQ[ho]SQ[io]SQ[jo]SQ[ko]SQ[lo]SQ[ep]SQ[fp]SQ[gp]SQ[eq]SQ[kq]SQ[lr]C[Could black even connect here, threatening two cuts? We have to remember that the whole white group is a bit eye-less, so white really needs the M6 connection, actually. Not sure.
]

))(;W[of]
;B[qf]
(;W[qh]
;B[kl]TR[of]SQ[lk]SQ[ll]C[Right now if white goes for the cut, black can just in some way capture these stones.
]

)(;W[km]
;B[nf]
;W[jk]
;B[jj]
;W[ik]
;B[kb]
;W[lb]C[KataGo gives white -15 pp for 104. P14
]

)))(;B[kp]
;W[lp]
;B[lq]
;W[jq]
;B[ip]
;W[ir]
;B[hq]
;W[fr]
;B[ds]
;W[gr]
))(;B[fr]TR[dr]C[Maybe D2 is a probe, saying that if black connects on this side, that leaves some bigger ko threats. Maybe in this case, white would fight the ko.
]

(;W[ok]
)(;W[of]
;B[qf]
;W[ok]
;B[lm]
;W[ml]
;B[pk]
;W[qh]
;B[pm]
;W[rm]
;B[ql]
;W[rl]
;B[nf]
;W[oe]
;B[ne]
;W[od]C[KataGo gives black -3 pp for 101. G2
]

)))(;W[ok]LB[pk:1]LB[pm:2]C[There's a potential for a ko here. It might seem like there's no harm in capturing once and then connecting.
]

(;B[lm]C[But black has this move and stuff.
]

)(;B[qh]
;W[of]
;B[qg]
;W[co]
;B[lm]
;W[ml]
;B[pk]
;W[dp]
;B[cq]
;W[dr]
;B[fr]
;W[bo]
;B[bq]
;W[fk]
;B[em]
;W[gl]
;B[gm]
;W[hm]
;B[gn]
;W[ip]C[KataGo gives white -0.4 pp for 100. P9
]

))(;W[pm]
;B[dn]
;W[dr]
;B[fr]
;W[bi]
;B[gn]
;W[go]
;B[lm]C[KataGo gives white -3.5 pp for 100. Q7
]

))(;B[oe]C[Of course, black doesn't want to just play a slow move here.
]

))(;W[qh]TR[rg]C[Black has just made a nice exchange.
]

;B[of]
;W[he]
;B[lm]
;W[km]
;B[hi]
;W[jk]
;B[jj]C[KataGo gives white -11 pp for 94. R12
]

)(;W[of]
;B[qf]
(;W[qh]TR[of]C[White still has the option of playing this atari and connecting here.
]

;B[nf]
;W[oe]
;B[ne]TR[oe]TR[pe]TR[of]C[Right now, the cut is not working because these white stones are not going anywhere, but that's going to change very soon.
]

)(;W[pl]
;B[hi]
;W[co]
;B[nf]
;W[he]
;B[ge]
;W[hf]
;B[gf]
;W[ok]
;B[nj]
;W[nc]
;B[bo]
;W[eq]C[KataGo gives white -1 pp for 94. P14
]

)))(;B[qg]
;W[qh]TR[rg]C[This wouldn't be as good because after white responds here, white still has a good move.
]

)(;B[eq]
;W[of]
;B[qf]
;W[rg]
)(;B[oe]C[Black could take away the cut directly like this, but it would be too slow.
]

))(;W[lq]
;B[dd]
;W[ed]
(;B[bc]C[It wouldn't quite work for black to do something like this.
]

;W[eb]
;B[fc]
;W[bb]C[This is actually fine for white.
]

)(;B[eb]C[Black can just go down here.
]

(;W[bd]
;B[bc]TR[bd]SQ[cd]SQ[dd]C[If white captures, black has this move later on.
]

;W[bb]
;B[cb]
;W[ac]
;B[db]
;W[bc]C[Black squeezes like this.
]

)(;W[bc]
;B[db]TR[bc]C[If white instead plays a slow move like this, later on black can push here.
]

(;W[bd]C[White might need to play here to save and is losing a lot of territory, actually.
]

)(;W[cb]
;B[bd]
;W[ab]
;B[ba]
;W[ca]
;B[da]
)))))(;B[qh]
;W[qg]
;B[rh]
;W[rg]TR[qh]TR[rh]C[This just doesn't work.
]

))(;W[of]
;B[qf]TR[qh]C[Now black is threatening R12.
]

;W[rg]TR[of]SQ[og]SQ[oh]SQ[oi]SQ[pi]SQ[oj]SQ[mk]SQ[nk]SQ[pk]SQ[om]C[If black can't capture P14, the group might be cut and in trouble, so this cut would be a recurring theme here.
]

)(;W[qf]
;B[of]
(;W[je]
;B[qh]
;W[qg]
;B[rh]C[Black can also cut here.
]

;W[rg]TR[qf]C[R14 is actually really badly placed, removing one of white's own liberties.
]

)(;W[rg]C[This would take gote here unnecessarily.
]

)))(;B[kp]
;W[lp]
;B[lq]C[Black wouldn't want to play this cut right away. For starters, it's kind of small, it's just a little bit of territory.
]

;W[ip]
;B[jq]
;W[hq]
;B[hr]
;W[gr]C[But there's also something like this cut here.
]

;B[ir]C[Let's ignore the ko for now and say that black has to connect here.
]

;W[eq]TR[hp]TR[hq]SQ[lq]C[Then white gets to wedge over here. Black will not be able to keep winning the ko, it's quite a heavy ko for black. So black will give up some points as a result of the cut, but both sides will have to keep reevaluating that as things change on the bottom.
]

))(;B[on]
;W[po]TR[oj]CR[pj]CR[qj]TR[nk]TR[pk]CR[qk]TR[om]SQ[mn]SQ[nn]TR[on]CR[pn]SQ[ho]SQ[io]SQ[ko]SQ[lo]CR[oo]CR[po]CR[pp]CR[qp]SQ[kq]CR[rq]SQ[lr]CR[rr]C[If black cuts here now, white can respond and will have no trouble on either side. Even though black got the cut in sente, it's not clear that this exchange is good for black. If the group is going to die anyway, black just reduced its own liberties, so black also leaves it for now.
]

;B[oh]
;W[gp]
;B[gq]
;W[ip]
;B[hq]
;W[fp]
;B[fq]
;W[mk]
;B[hm]
;W[qs]
;B[os]
;W[gm]
;B[hl]C[KataGo gives black -1 pp for 57. P6
]

))(;W[nh]
(;B[ln]C[It would not work for black to cut here immediately.
]

;W[km]TR[ln]C[The cutting stone just dies completely.
]

;B[kp]
;W[lp]C[There's not even any fanciness here, it's just completely connected for white.
]

)(;B[lm]C[It's a bit complicated, different things can happen here. It's going to depend on the surroundings, but there's some stuff here to worry about for white.
]

(;W[ml]TR[ln]C[If white pulls back like this, there is the cutting point.
]

)(;W[mm]
(;B[ln]
;W[km]TR[mm]SQ[ln]C[If white wants to protect like this, it needs to make sure that this doesn't work immediately.
]

)(;B[kl]C[Then there might be some other moves like this later.
]

)))))(;W[mj]TR[oj]TR[nk]TR[pk]C[Black is allowing white to follow up towards this group, but white doesn't do it right away.
]

;B[im]
;W[oo]
;B[mp]
;W[kd]
;B[jd]
;W[ke]
;B[kl]
;W[kk]
;B[lm]C[KataGo gives white -2 pp for 50. N10
]

))(;W[os]C[It would probably be an overplay for white to take away the right now because of the cutting points in the white shape on the outside.
]

;B[ns]
;W[mr]
;B[nr]
;W[qs]
;B[kp]
;W[jp]
;B[lp]
;W[jq]
;B[jo]
;W[ip]
;B[io]
;W[hq]C[KataGo gives white -22 pp for 46. P1
]

))(;W[mp]TR[np]C[Just running this stone out wouldn't be a good idea for white.
]

(;B[on]C[Maybe black should wedge here.
]

;W[om]
;B[oo]C[There are so many weaknesses in white's shape.
]

)(;B[lp]
;W[mo]
;B[lq]C[Multiple holes in white's shape.
]

)))(;B[no]TR[np]C[Black doesn't immediately go for this stone (or something like this).
]

;W[mo]
;B[mp]
;W[mn]
;B[oo]
;W[on]
;B[ql]
;W[lr]
;B[qn]
;W[rn]
;B[qm]
;W[qs]C[KataGo gives black -3 pp for 39. O5
]

))(;B[iq]
;W[qe]C[This is such a nice shape for white.
]

))(;B[no]SQ[qj]SQ[nn]SQ[pn]SQ[ko]TR[np]C[Black can capture it, but white is more interested in making good shape on the outside.
]

))(;W[no]C[I think this would be a small mistake for white.
]

;B[oo]
;W[on]
;B[nn]C[Black can push and cut here.
]

;W[lp]
;B[mn]
;W[lo]C[Nothing is collapsing for white on the outside.
]

(;B[po]
;W[qo]
;B[qn]
;W[qm]
;B[rn]
;W[rm]C[For instance, it's not like this is working immediately for black.
]

)(;B[rr]C[But black will just live in the corner easily.
]

;W[ll]TR[mn]TR[nn]C[This extra aji on the outside is a little bit annoying for white.
]

)))(;W[lo]C[White could surround this group immediately. This would be quite a straightforward way to play.
]

))(;B[pp]C[You could reasonably think about this normal move, and this is also playable.
]

(;W[qo]
;B[qr]LB[mq:A]LB[rr:B]C[You get something like this. Black his miai of making extension or settling in the corner.
]

)(;W[qr]C[White can of course also resist with this atari
]

;B[qo]
;W[pq]
;B[rp]
;W[qq]
;B[po]
;W[nq]TR[kq]C[White can actually clamp here with the support of L3. This is slightly pleasant for white maybe, still an even result, but maybe this is what black wanted to avoid.
]

)))(;W[de]TR[cc]TR[cd]C[This looks like really bad shape here. The earlier exchange is good for black. C16 has quite a lot of aji.
]

)(;W[ce]TR[cc]TR[cd]C[Again, it seems like this is a good exchange for black.
]

))(;B[oc]C[Otherwise, this is the most solid you get.
]

;W[rd]
;B[qe]
;W[qc]
;B[pc]
;W[qb]TR[oc]C[In some situations, we get white living in the corner even after that one.
]

))(;B[dd]
;W[ed]C[This would be really bad shape for black.
]

)(;B[de]TR[cc]TR[cd]C[Black should have just played at D15 to begin with.
]

))(;B[pq]
(;W[cf]
;B[cd]
;W[cc]TR[cd]C[Maybe white would respond here now anyways, and C16 would not be an extra bonus for black.
]

)(;W[ce]
;B[cd]
(;W[cc]TR[cd]C[This seems clearly good for black to me.
]

)(;W[dd]C[Normally, white would respond like this.
]

))))