Although ptolemycasio managed to get a draw in 2020, I think I'm the first player from Japan who actually won against TwixtBot.
It was a very interesting game for me, although I feel like I got very lucky with the first move! I also don't really understand the sequence from 13.h13 to 22.c22 - the exchange doesn't seem to benefit White that much, yet I gain a connected wall facing the right. According to my calculation 24.o8 barely connects even in gote, making White's situation quite desperate.
What I'm most curious about is whether Black could connect though the bottom given first move? Let's say something like |31.N1132.P17
Is there any way for White to save the bottom right corner even if they had a move that works on the top? I can't find anything but feel like there must be something.
Any other comments or feedback is welcome as always!
On
2021-02-08 at 07:43,
mmKALLL
said:
For easy reference a few simple variations for 24.o8 - I can't find much room for improvement for White due to how threatening the right-side pegs are:
If I were playing white after black Q7, I might try to find a way to connect R11 to D10. I would have to find some tempo move near D10, which threatens to win along the left, and at the same time serves as a negative ladder catcher for the sequence |13.p1014.o815.n916.m9. But 17.f11 for example does not make a threat down the left side. |13.d14 looks better in that respect, but then black has 14.f12 which blocks the ladder and also threatens to connect to the left in two different ways.
So MAYBE what happened was, several moves previously the bot thought it had some winning move in the vicinity of D10, but realized too late that it didn't work. Then it did the same thing it did against me, as BonyJordan described. It looked at maybe 49 other moves that also didn't work, and was working on choice 50 when it ran out of time. Since it hadn't realized at that moment that choice 50 was as bad as the others, that's the move it went with. I hope BonyJordan can show us the machine evaluation of its position as the game progressed.
It was a very interesting game for me, although I feel like I got very lucky with the first move! I also don't really understand the sequence from 13.h13 to 22.c22 - the exchange doesn't seem to benefit White that much, yet I gain a connected wall facing the right. According to my calculation 24.o8 barely connects even in gote, making White's situation quite desperate.
What I'm most curious about is whether Black could connect though the bottom given first move? Let's say something like
|31.N11 32.P17
Is there any way for White to save the bottom right corner even if they had a move that works on the top? I can't find anything but feel like there must be something.
Any other comments or feedback is welcome as always!
|31.j12 32.k12 33.n9 34.m9 35.l10 36.k10 37.j11 38.i11
|31.n11 32.m9 33.j11 34.k10 35.l12 36.i11 37.j12 38.k12
|31.n11 32.p17 33.u9 34.u10 35.v11 36.v12 37.v15 38.w14 39.u17
If I were playing white after black Q7, I might try to find a way to connect R11 to D10. I would have to find some tempo move near D10, which threatens to win along the left, and at the same time serves as a negative ladder catcher for the sequence |13.p10 14.o8 15.n9 16.m9. But 17.f11 for example does not make a threat down the left side. |13.d14 looks better in that respect, but then black has 14.f12 which blocks the ladder and also threatens to connect to the left in two different ways.
So MAYBE what happened was, several moves previously the bot thought it had some winning move in the vicinity of D10, but realized too late that it didn't work. Then it did the same thing it did against me, as BonyJordan described. It looked at maybe 49 other moves that also didn't work, and was working on choice 50 when it ran out of time. Since it hadn't realized at that moment that choice 50 was as bad as the others, that's the move it went with. I hope BonyJordan can show us the machine evaluation of its position as the game progressed.