It's my first draw, and spd_iv's 2nd real draw (he had a courtesy draw some years ago. His other real draw was against Loïc in game 854386.)

The inevitable ending we both saw was |39.q9 40.p9 41.n8.

At the end, spd_iv said "i should have played 26.q10 ;)" ... but I don't think that's the answer, because I was planning |27.r6, but just before I played it, I saw the devastating 28.q6 response, and if he had let me close the s10 gap first, that q6 reply would not have worked: |26.q10 27.s10 28.s4 29.r6 30.q6 31.t5 32.s7 33.s6. Without closing the gap, it's losing: |27.r6 28.q6 29.t5 30.s7 or |27.r6 28.q6 29.s6 30.s5 31.t4 32.u3 33.r5 34.r4 35.s3 36.s2 or |27.r6 28.q6 29.t7 30.s5 31.w4 32.u4.

Is there anything White could have done after |26.s4 to win?

Backing up a step, after |25.t8, is there a winning move for Black? I'm thinking 26.s10 would have made things more interesting, but after 27.s9 28.r8 29.t11 30.s4 31.r6, White might be winning.

Or, going forward a step to |27.r7, is it possible for either player at that point to play differently, and win? Or is the draw already sealed? If |28.q3 then 29.o4, and then... um... oh yeah, 30.n4 31.n6 32.p3 33.l5 34.o5 35.p5 36.q6 37.r6 38.s7 39.u6 40.u3 41.v4 42.w2 43.t5. If |28.p3 then 29.s5 30.u3 31.q4. If |30.q10 then 31.s9 32.t6 33.p5 34.p7 35.q8.

It seems that my early judgment that |11.r12 was probably strong enough to support a cascade attack starting at F9 was not quite right! Twixt has a lot of tough judgment calls like that in the opening moves. Spd_iv was probably thinking something I'd play something like |13.t13, continuing the windmill, but I was running out of room on both the left and the right, which I've learned to recognize as a very bad sign.

I think the suggested alternative line for black, |26.s10 27.s9 28.r8 29.t11 30.s4 31.r6, works. 32.q5 33.t5 34.u5. If instead 33.p5, 34.p6.

For white, it seems natural to try 27.q6, which I think doesn't work: |27.q6 28.q5 29.s5 30.t6 31.t7 32.o6 33.u5 34.u3; or |27.q5, which seems strong on the top, but I guess still loses to 28.s10?

I thought white's actual sequence 27.r7 28.q5 29.s5 30.t6 31.p5 was brilliant.

BTW, I can't copy from this page when I'm composing a comment. I can select, but focus jumps as soon as I try to copy. I have to click "No comment", and copy to an external editor. The problem does not occur when not in compose mode. Does this happen to anyone else?

|26.s10 27.s9 28.r8 29.t11 30.s4 31.r6, and then 32.q5 33.t5 34.u5 as you said, but then 35.v6 and White gets the final double threat conversion and wins (or, in other words, the theory is still alive that |26.s10 would have been suicide for Black). It's not a draw, either: 36.w6 37.p5 38.t9 39.u10 40.u7 41.u6 42.s7 43.s5 44.t7 45.q6.

The 2-2 crossover (here, R6) is an important pattern. In fact, I used it in my game against tasuki. It may be the first thing to consider if the link crossover (R7) doesn't work.

The reply to |27.q5 that worried me was 28.p5. Then, for example, 29.p7 30.r4.

|27.r6 was really a dilemma. Should I take the risk that my opponent won't see the non-obvious reply 28.q6? If my opponent were a lesser player, I might have tried it. Risk a loss for the chance at a win, or take the draw? The Championship is on the line! I won't comment on the other games in progress, but I will say that I analyzed them before making this choice.

I can't reproduce your copy & paste problems - what OS and browser are you using?

|26.q10 27.s10 28.s4 29.r6 30.t6 looks like a more obstinate defense, but after 31.r7 32.q3 33.o4 34.n4 35.n6 white still wins.

How about

|20.p6 21.n6 22.m7 23.p7 24.q8 25.m4 26.n9 27.o9 28.o11 29.t17 30.t15 OR

|20.p6 21.p7 22.o8 23.o9 24.n10 OR ... ?

Alan, your comments on crossovers are very useful. For whatever reason, I primarily think about the 3-2 and 3-3 in this situation.

For the C-P problem, I have had it with a variety of flavors and versions of linux, and several versions of firefox. But it's easy to work around, if I use the mouse to select Edit/Copy, instead of typing ctrl-c. I thought I had tried this before posting, but I'm so addicted to the keyboard that I apparently accidentally used the keyboard version of mouse selection (alt-E-C).

David: I'm pretty sure I was thinking, if |20.p6 then 21.t7, and that looks pretty good for White... but there is still a lot of space for Black to work with. I'm not even sure what move comes next. Got a suggestion?

I remember also thinking, if |22.m9 then 23.n12 (I have to close that up because 23.n10 24.l11 25.n12 26.k13 is disastrous). So 23.n12 and then 24.r9 25.r10, which looks good for White, or 24.p8 which I had already misjudged as a win for White, except in this scenario there is the possibility of making a 25.m8 threat thru the little gap. But now I see 24.q8, and now nothing seems to work for White. Yikes! Did I dodge a bullet?

Dushoff: Thanks. I guess I was light on the Linux testing. I have no desktop Linux at the moment. Is it just Firefox? Did you try any other popular browsers?

No, I don't have any other browsers. But it's really not much of a problem with the workaround.