Galdian posted to the LG forums asking what, exactly, went wrong after move 27, which he feels is a winning position.

I agree.

I think |32.o9 was the turning point.

|32.o13 would have led to victory for Black. The actual |32.o9 closes a weak gap that didn't need to be closed - |32.o13 is strong enough - a 3-3 parallelogram - while opening up a new threat around White's p12 peg. For example |32.o13 33.q10 34.p15 35.s14 36.r16 37.t17 38.u17 39.v16 40.s18 41.u18 42.u19.

I agree |32.O13 wins in Alan's variation. But white 31.P12 was not the best. How about |31.q10 32.o13 33.q14 34.r17 35.p16 36.p18 37.n12. So, maybe |30.q14 would be better for black than S12. For example 31.n12 32.o9 33.p13 34.s8 35.s10 36.q8 37.t6 38.s7 39.u8 40.u9

In that last variation |30.q14 31.n12 32.o9 33.p13 34.s8 35.s10 36.q8 37.t6 38.r6 39.t8 40.v6 would have been simpler. Another variation worth mentioning is |31.q10 32.q13 33.o11 34.m12 35.p13 36.l14 37.p17 38.o14 39.n14 40.n15 41.p15