Register
to add
comments.
Game number: Main Page
White: TwixtBot    Black: Florian Jamain
This game (LG) | Download JTwixt file
On 2018-12-18 at 21:02, twixter (info) said:
Vongratulations to Florian for making it look easy!

Maybe |9.p16 would have been better? For example, just to illustrate the threat, 10.o16 11.n15 12.q17 13.r17 14.s18 15.t18 16.u19 17.v19 18.w20 19.l16 20.k16 21.h14

But then black might play 12.m15 13.r15 14.o7 or many other possible moves.

|13.j11 looks better than R12.

On 2018-12-18 at 21:03, twixter (info) said:
I meant Congratulations of course. How embarrassing.

On 2018-12-18 at 21:51, BonyJordan (info) said:
I don't know why 36.e12 is such a surprising move, but that is the point at which the evaluation dramatically changed. After 35.b3 the bot evaluated itself at +0.443 and after 37.j7 at -0.423 and straight down from there.

We can tell that it saw that things weren't so good because instead of making the straightforward block of 35.b4 or 35.d4 it plays the weird looking 35.b3. To me that implies that it ran through the variations after b4 and decided those were no good, so it tried the variations after d4, and decided those were no good either, and ran out of searching time before it could conclude that the variations after b3 were no good.

On 2018-12-18 at 21:57, Florian Jamain (info) said:
First, I don't know about the very first move... 1.s8 2.swap 3.d19

He played on the corner but I still have the Medcalf (|4.d21 5.f20 6.i21 7.g22 8.e16) here... seems not very good for me to let this after playing in a corner, but maybe it is for some advanced reasons, I can consider it as some very strong move but I am not convinced actually.

Moreover, during the game if you wanna try say this : |13.j11
You need to take care about the following |13.j11 14.i21 15.h12 16.f18 17.e17 18.d23 which is working for black locally.
Of course you can expect something like |13.j11 14.i21 15.h12 16.f18 17.e17 18.d23 19.g22 20.f22 21.i23 22.h21 23.p15 24.r19 25.s18 26.t20 27.q19 28.p20 29.o20 30.n21 31.m21 32.l22 33.k22
But I will never let something like this happen to be honest, I knew this kind of lines and I just won't play it.

In the end the IA let me built some walls but I am sure it knew that it was useless for me cause 37.j7 is just okay.

Now let's talk about the point where something could be different.

After 21.u13 I knew that the game was won.

But I was ready to spend hours in this if it played |21.t12, what would you like to answer to this?

Maybe |21.t12 22.q14 ?

Not sure actually.

What the Bot is thinking about this in a game between himself and himself?

The variations are so... complicated, my head was burning, the IA does not care and play easily in 5 minutes, there are no difference between 2 losing lines probably... but let me know about this |21.t12, why he prefered 21.u13?

On 2018-12-18 at 22:05, Florian Jamain (info) said:
David, in your variation, even... |13.j11 14.k7

No?

On 2018-12-19 at 02:29, BonyJordan (info) said:
It is weird the way the bot likes to play far to the sides in early games. I may try to give it an opening book at some point which may help.

The bot rates 21.t2 22.u10 as -0.532 and 21.u13 22.u10 as -0.488, which is to say it agrees that 21.t12 is a better move. I guess in the heat of battle u13 looked better? Anyway, it self plays to |21.t12 22.u10 23.u9 24.w11 25.l8

On 2018-12-19 at 07:15, Florian Jamain (info) said:
|21.t12 22.u10 23.v11 24.w11 25.l8

Since black won the game, the question is how can I still win in this variation, what black would play, especially on bottom right if the Bot was playing black?

On 2018-12-19 at 10:19, Florian Jamain (info) said:
Maybe I just had to play... |14.r8

And win easily... for instance: |14.r8 15.j11 16.i13 17.p11 18.k7

On 2018-12-20 at 04:03, BonyJordan (info) said:
bot self-plays: |21.t12 22.u10 23.v11 24.w11 25.l8 26.p17 27.i14 28.i15 29.t17 30.q15 31.c12 32.g14 33.d14 34.e8 35.e7 36.c9 37.b10 38.a10 39.f9 40.f10 41.l17 42.j16

On 2018-12-20 at 06:16, Florian Jamain (info) said:
Yes of course!

The same variation is winning.

Was tired, missed the possible connection in Q11.

So the Bot was crushed earlier in the game already.

On 2018-12-20 at 06:55, Florian Jamain (info) said:
Wait a second... No.

All your variation did not solve the fact that white can still play |21.t12 22.u10 23.v11 24.w11 25.l8 26.p17 27.i14 28.i15 29.t17 30.q15 31.c12 32.g14 33.d14 34.e8 35.e7 36.c9 37.b10 38.a10 39.f9 40.f10 41.l17 42.j16 43.p11

How do you connect with black?

Gonna try s21?

Go almost untill the end, only the connection on bottom right is interesting, the rest on the left and middle is useless please!

Actually he considers that black is winning or white???

Let see the right side please after p11 for white.

On 2018-12-20 at 13:04, BonyJordan (info) said:
|21.t12 22.u10 23.v11 24.w11 25.l8 26.p17 27.i14 28.i15 29.t17 30.q15 31.c12 32.g14 33.d14 34.e8 35.e7 36.c9 37.b10 38.a10 39.f9 40.f10 41.l17 42.j16 43.p11 then 44.u17 45.v16 46.s18

On 2018-12-20 at 14:13, Florian Jamain (info) said:
Seems correct, so... Why the Bot thought he was winning and now is okay to admit that it was not...

It seems really not normal that e12 is something it is just an obvious move. Special.

On 2018-12-21 at 00:20, twixter (info) said:
How about |21.t12 22.u10 23.v11 24.w11 25.l8 26.p17 27.i14 28.i15 29.v16

On 2018-12-21 at 01:21, twixter (info) said:
Uh never mind, |21.t12 22.u10 23.v11 24.w11 25.l8 26.p17 27.i14 28.i15 29.v16 30.s18 looks crunchy

On 2018-12-21 at 02:41, BonyJordan (info) said:
|21.t12 22.u10 23.v11 24.w11 25.l8 26.p17 27.i14 28.i15 29.v16 30.s18 31.s19 32.r20 33.u20 34.u17 35.m18 36.n16 37.u18 38.v15 39.w14 40.w13 41.o17 42.q19 43.e12 44.g14 45.e14 46.f10 47.d10 48.e8 sayeth the bot

You must be logged in to add comments.