to add
Game number: Main Page
White: TwixtBot    Black: David J Bush
This game (LG) | Download JTwixt file
On 2020-06-23 at 22:38, Alan Hensel (info) said:

On 2020-06-23 at 22:39, Peyrol (info) said:
I'm not humble like Alan, who beat the bot last year and didn't post any comment.

An alternate line was |31.n14 32.l14 33.l13 34.k12 35.k11 36.i11 which is why I played those pegs in the top left before K16.

I would be grateful to see how TB assessed its position during the game. Maybe it thought I was going to try to use R21, for example with |16.n21?

On 2020-06-23 at 22:54, Peyrol (info) said:
Thanks Alan!

On 2020-06-23 at 23:21, MisterCat (info) said:
Congratulations seconded! I could fathom very little of any 'strategy' by the bot in this game, but perhaps Twixter has developed the necessary techniques to beat it at it's own game.
Don't forget to personally invite it to your 'rated bot' tournament; Jordan may not have noticed yet!
And Twixter should consider a challenge vs. Twixtbotslow - if you dare!

On 2020-06-24 at 08:07, Peyrol (info) said:
Thanks MisterCat. I agree, the bot gave me too much in the opening. I wonder if 10.G15 surprised it. (BTW I am now Peyrol because I lost my password.) Sometimes it goes off the rails, which Slowbot is much less likely to do.

On 2020-06-24 at 14:19, shyryan (info) said:
Congratulations!! And thank you for including the comment about the pegs in the upper left; super helpful. : ) Well played!

On 2020-06-24 at 18:52, BonyJordan (info) said:
-0.426 |3.t20 4.u20 5.r19 6.t22 7.p20 8.o22 9.o23 10.n20 11.q22 12.q17 13.s16 14.t16 15.q15 16.s18 17.m15
-0.111 |5.g12 6.h16 7.j15 8.h8 9.p11 10.u20 11.r19 12.t22 13.p20 14.o22 15.o23 16.n20 17.q22 18.q17 19.t15 20.s16 21.s17 22.t14 23.u13
0.365 |7.e7 8.n19 9.s18 10.h15 11.q13 12.i13 13.q9 14.r15
0.386 |9.d19 10.g16 11.e15 12.f16 13.d15 14.c15 15.k16 16.n19 17.j18 18.k20
0.476 |11.f14 12.i9 13.g16 14.e19 15.d9 16.g18 17.f20 18.c18 19.k18 20.k20 21.m19 22.m21
0.657 |13.h15 14.i10 15.d13 16.e9 17.g8 18.h12
0.625 |15.l15 16.k4 17.i7 18.m17 19.o16 20.p18 21.l19
0.698 |17.o16 18.p18 19.r17 20.m21 21.d15 22.e19 23.f20 24.c18 25.n18 26.n19
0.708 |19.r17 20.n21 21.o20 22.n19 23.q15 24.u20 25.r19 26.t22
0.762 |21.d15 22.e19 23.l19 24.i10 25.k12 26.f7 27.g6 28.e9 29.f10 30.c8 31.i7 32.h8
0.678 |23.l19 24.i9 25.h10 26.k16 27.n14 28.j14 29.k11 30.n19 31.q17 32.g10 33.j9
0.200 |25.f9 26.j4 27.k17 28.e4 29.i5 30.i6 31.d5 32.c5 33.f4 34.g3 35.j3 36.i2 37.h3 38.k3 39.l4
-0.590 |27.d10 28.k16 29.j16 30.n19 31.f20 32.c18 33.r13 34.m21 35.j20
-0.479 |29.f4 30.k16 31.j16 32.n19 33.f20 34.c18 35.q17 36.n15
-0.662 |31.j16 32.n19 33.f20 34.c18 35.c16 36.a17 37.q15 38.n15 39.p13 40.o13
-0.682 |33.q15 34.n15 35.p13 36.o13 37.q10 38.r5 39.p6 40.q6
-0.767 |35.p13 36.o13 37.f20 38.c18 39.s10 40.u20 41.r19 42.s7 43.k18
-0.761 |37.r10 38.r7 39.p9 40.p6 41.d17 42.c20
-0.833 |39.p9 40.p6 41.m7 42.o9 43.f20 44.c18
-0.859 |41.v7 42.u20 43.r19 44.m21 45.k21 46.k18 47.o11

On 2020-06-24 at 19:31, Alan Hensel (info) said:
So TwixtBot felt like it was losing in the first couple of moves, but David's |6.n7, in its opinion, gave it the upper hand, and it was getting pretty darn confident until it was surprised by David's |24.h8. That |27.d10 is one of its signature bonkers moves (because no human would leave a gap like that, when |27.g11 or |27.e11 would do the same job of blocking Black's connection to the left while closing the gap).

|24.h8 is perhaps not so easy for the bot to find. Too much open space for it?

That is interesting.

David, when did you start to get confident that you were going to win?

On 2020-06-24 at 20:23, Peyrol (info) said:
I didn't really look at the top left until 23.L19. Once I realized I could make three consecutive tempo threats there, then I felt like maybe the bot is not so omniscient. Before that I was thinking there has to be a flaw in my plan. The fact that the bot took a bit longer than usual to respond to H8, pleased me greatly. So maybe, since the bot probably does not think about the game conceptually, like "3 tempo threats would be something to watch out for," just missed it among the myriad possibilities. I was surprised by E7 in the opening. Maybe |7.i7 would have won.

Thanks very much to BonyJordan! AND thanks for putting TB in my tournament!

On 2020-06-25 at 08:06, Peyrol (info) said:
Alan, you were more correct about 27.D10 being a bonkers move than I realized. Is |27.g11 winning? For example 28.k16 29.n14 30.l14 31.l13 32.k12 33.k11 34.g6 35.i10 36.f4 37.j8 38.k4 39.b4 40.d4 41.d5 42.b3 43.g2 44.f3 45.k6 46.m5 47.j4 48.i3 49.h3
or 34.i11 35.l8
or 28.g6 29.i10 30.f4 31.j8 32.k4 33.b4 is the same pattern

On 2020-06-25 at 08:21, Peyrol (info) said:
Taking a tip from the computer, maybe |27.g11 28.j4 29.i10 30.e4 31.i5 32.i6 33.d5 34.c5 35.f4 36.g3 37.j3 38.i2 39.h3 40.k3 but I did not see this line during the game, and I'm still not convinced black wins.

On 2020-06-25 at 10:08, Peyrol (info) said:
27.g11 28.j4 29.i10 30.e4 31.i5 32.i6 33.d5 34.c5 35.f4 36.g3 37.j3 38.i2 39.k5 40.l10 41.g6 42.k16 43.n14 44.l14 BUT that doesn't work because of 41.n14
28.h5 29.i10 30.f4 31.g6 32.i6 33.b4 34.d4 35.d3
doesn't work either
28.h4 29.i10 30.e3 31.g6 32.f5 33.k6 34.j5 35.m5 36.l10 37.n14 38.k7 maybe that works?

On 2020-06-25 at 22:58, BonyJordan (info) said:
Likely 27.d10 means that the bot suddenly realized it was losing and tried the best 50 moves to try to get out of it, and none of them worked, so it tried number 51, and ran out of time before it had discovered that #51 was also lost.

You must be logged in to add comments.