By the way, tonight I decided to play a 1-pt match with her, playing
her own style. I managed to make a 6-pt prime in her board and she did
the same in my board. After almost 2 hours, I just got tierd and
finished her off. But my interest was to see how gnubg whould analyze
some strange positions in our match.
The first interesting point was that it took almost 1 hour for gnubg's
2-ply to complete analyzing our 1-pt match, when cube-handling was not
even considered.
The second strage point was that in middle of the match, gnubg gave
both of us HUGE POSITIVE EQUITIES. As far as I know, the total sum of
equities should be zero, but in a position in our match, it was about
2 !!
I have the match log and if someone is eager to see, I will e-mail it
for him, by the way this is the oldmoves of match from FIBS:
Score is 0-0 in a 1 point match.
yeti_yew is X - PersianLord is O
X: (2 4) 12-16 12-14
O: (2 5) 8-3 6-4
X: (5 3) 17-22 1-4
O: (5 1) bar-20 6-5
X: (2 5) 1-3 16-21
O: (1 2) bar-23 23-22
X: (1 1) bar-1 4-5 19-20 21-22
O: (4 6) bar-21
X: (2 2) 19-21 19-21 21-23 14-16
O: (3 5) bar-20 bar-22
X: (5 4) bar-4 bar-5
O: (3 3) bar-22 24-21 6-3 24-21
X: (4 1) bar-1 bar-4
O: (5 3) 21-16 13-10
X: (6 3) bar-3 4-10
O: (2 1) bar-24 bar-23
X: (1 3) bar-3 10-11
O: (3 4) 6-2 23-20
X: (4 1) 1-2 2-6
O: (2 5) bar-20 bar-23
X: (5 3) 11-16 12-15
O: (6 4) bar-21 13-7
X: (4 2) 5-7 19-23
O: (3 3) bar-22 bar-22 22-19 22-19
X: (2 3) bar-2 15-18
O: (1 4) 8-4 24-23
X: (1 1) bar-1 bar-1 3-4 7-8
O: (6 3) bar-22 bar-19
X: (2 3) 1-4 4-6
O: (3 1) 22-19 19-18
X: (3 2) bar-3 16-18
O: (2 2) bar-23 20-18 13-11 11-9
X: (6 6) bar-6 3-9 12-18 6-12
O: (5 2) bar-23 bar-20
X: (6 6) 8-14 9-15 6-12 1-7
O: (2 2) 20-18 18-16 16-14 13-11
X: (6 4) bar-4 bar-6
O: (4 6) 13-7 14-10
X: (3 6) bar-3 4-10
O: (2 1) bar-24 7-5
X: (6 3) bar-3 2-8
O: (3 5) 19-14 11-8
X: (1 6) bar-6 4-5
O: (1 5) bar-24 23-18
X: (5 2) 12-14 3-8
O: (6 1) bar-24 bar-19
X: (4 6) 14-18 8-14
O: (6 6) bar-19 24-18 19-13 13-7
X: (6 5) bar-5 1-7
O: (4 2) bar-21 18-16
X: (3 6) 10-16 6-9
O: (4 3) bar-21 21-18
X: (1 6) 17-18 5-11
O: (5 1) bar-20 19-18
X: (1 4) bar-4 17-18
O: (2 4) bar-21 21-19
X: (4 5) 12-17 6-10
O: (1 4) 21-17 19-18
X: (2 6) bar-6 bar-2
O: (2 3) 20-17 18-16
X: (4 4) bar-4 12-16 4-8 8-12
O: (5 6) bar-19 19-14
X: (4 4) bar-4 10-14 6-10 4-8
O: (4 4) bar-21 21-17 17-13 13-9
X: (2 4) bar-2 5-9
O: (1 5) bar-20 17-16
X: (2 1) bar-1 14-16
O: (5 4) bar-20 20-16
X: (4 1) bar-1 12-16
O: (6 4) bar-19 19-15
X: (5 4) bar-5 11-15
O: (1 1) bar-24 17-16 16-15 24-23
X: (3 5) bar-5 bar-3
O: (6 4) 23-17 20-16
X: (2 4) 9-13 13-15
O: (6 4) bar-19 19-15
X: (3 2) bar-3 10-12
O: (2 6) 16-10 15-13
X: (3 3) 3-6 7-10 10-13 13-16
O: (2 4) bar-21 bar-23
X: (5 1) 12-17 3-4
O: (1 1) bar-24 23-22 21-20 24-23
X: (3 4) 8-12 12-15
O: (6 6) 20-14 22-16 23-17 17-11
X: (3 5) bar-5 bar-3
O: (6 2) 16-10 14-12
X: (1 5) 5-10 10-11
O: (3 4) bar-21 bar-22
X: (6 6) 3-9 9-15 6-12 12-18
O: (5 6) bar-19 22-17
X: (1 1) 15-16 16-17 15-16 11-12
O: (2 1) bar-23 21-20
X: (3 5) 6-11 12-15
O: (1 2) 20-18 19-18
X: (1 1) bar-1 11-12 12-13 13-14
O: (1 1) 18-17 17-16 16-15 15-14
X: (1 2) bar-2 bar-1
O: (4 3) 23-19 19-16
X: (5 1) bar-1 bar-5
O: (3 1) 18-15 14-13
X: (5 5) 1-6 5-10 10-15 1-6
O: (6 4) bar-19 19-15
X: (2 2) bar-2 2-4 4-6 6-8
O: (5 2) 15-10 16-14
X: (2 5) 8-10 2-7
O: (2 6) bar-19 19-17
X: (2 6) 7-13 1-3
O: (5 5) bar-20 20-15 17-12 15-10
X: (1 1) bar-1 13-14 3-4 4-5
O: (4 3) bar-21 21-18
X: (6 4) 6-12 6-10
O: (2 6) bar-19 bar-23
X: (6 4) 12-18 10-14
O: (5 5) bar-20 23-18 20-15 18-13
X: (3 5) bar-5 5-8
O: (4 2) 19-15 15-13
X: (6 3) 5-11 1-4
O: (5 3) 15-10 13-10
X: (1 3) 4-7 14-15
O: (4 3) 13-9 10-7
X: (2 4) bar-4 5-7
O: (3 4) bar-21 21-18
X: (6 6) 4-10 3-9 11-17 10-16
O: (1 3) bar-22 bar-24
X: (2 3) 16-18 9-12
O: (5 2) bar-20 20-18
X: (3 6) bar-3 12-18
O: (1 3) bar-22 22-21
X: (1 3) 15-16 8-11
O: (3 1) 22-19 21-20
X: (6 3) 7-13 11-14
O: (4 2) 24-20 20-18
X: (3 4) bar-3 14-18
O: (3 1) bar-22 22-21
X: (1 6) 14-15 3-9
O: (5 6) 24-18 23-18
X: (6 5) bar-5 17-23
O: (6 2) bar-23 23-17
X: (2 5) bar-5 15-17
O: (6 3) bar-22 24-18
X: (4 1) 5-6 9-13
O: (6 3) 22-16 20-17
X: (4 6) bar-6 bar-4
O: (1 5) 21-16 18-17
X: (5 1) 4-9 13-14
O: (6 2) 19-13 13-11
X: (4 6) bar-4 5-11
O: (1 3) bar-22 22-21
X: (6 4) 9-15 4-8
O: (5 2) 21-16 16-14
X: (6 4) bar-4 8-14
O: (4 5) bar-20 20-16
X: (2 1) 4-5 11-13
O: (4 5) 22-17 22-18
X: (5 5) 6-11 5-10 4-9 3-8
O: (6 4) 21-15 19-15
X: (3 6) bar-3 13-19
O: (4 1) bar-21 20-19
X: (1 4) bar-4 11-12
O: (3 1) 17-14 20-19
X: (6 4) bar-6 10-14
O: (6 4) bar-19 19-15
X: (1 4) 9-13 8-9
O: (4 3) 21-17 21-18
X: (1 6) 5-11 9-10
O: (5 5) 19-14 19-14 18-13 18-13
X: (2 1) bar-1 bar-2
O: (3 4) 16-12 15-12
X: (5 1) bar-5 2-3
O: (6 6) 17-11 17-11 16-10 16-10
X: (2 4) bar-4 bar-2
O: (3 4) 17-13 18-15
X: (5 6) 1-7 2-7
O: (2 6) 15-9 18-16
X: (1 6) 3-9 4-5
O: (4 4) bar-21 13-9 13-9 21-17
X: (3 1) bar-1 1-4
O: (6 1) 14-8 14-13
X: (2 6) 6-8 8-14
O: (5 2) bar-20 10-8
X: (6 2) 4-6 14-20
O: (5 5) bar-20 15-10 20-15 16-11
X: (1 1) bar-1 7-8 1-2 2-3
O: (5 4) bar-21 13-8
X: (5 6) bar-5 7-13
O: (2 1) bar-23 15-14
X: (4 4) 13-17 4-8 17-21 3-7
O: (3 3) bar-22 bar-22 bar-22 22-19
X: (6 5) 8-14 14-19
O: (3 6) bar-19 bar-22
X: (4 6) bar-4 7-13
O: (6 5) 23-17 22-17
X: (5 3) 5-8 8-13
O: (5 4) 17-12 15-11
X: (6 6) 13-19 2-8 8-14 14-20
O: (3 1) bar-22 11-10
X: (2 1) 21-23 23-24
O: (5 4) 17-12 12-8
X: (1 5) 2-7 7-8
O: (6 4) bar-21 22-16
X: (1 5) 20-21 13-18
O: (3 1) bar-22 16-15
X: (4 3) 19-23 21-24
O: (5 5) 22-17 22-17 22-17 15-10
X: (4 4) 3-7 4-8 3-7 18-22
O: (5 5) 17-12 17-12 17-12 9-4
X: (5 4) bar-4 4-9
O: (1 3) bar-22
X: (4 6) bar-4 9-15
O: (1 5) bar-20 12-11
X: (1 6) 15-21 21-22
O: (5 3) bar-20 20-17
X: (5 5) 4-9 9-14 8-13 13-18
O: (3 6) 17-11 20-17
X: (2 2) 18-20 7-9 20-22 14-16
O: (4 3) 17-13 13-10
X: (6 4) 9-15 16-20
O: (6 3) 10-4 12-9
X: (4 5) 5-9 9-14
O: (2 6) bar-19 19-17
X: (2 4) 15-17 17-21
O: (1 6) bar-19 19-18
X: (3 6) 1-4 7-13
O: (5 3) bar-20 18-15
X: (5 3) bar-5 1-4
O: (6 6) 20-14 15-9 14-8 9-3
X: (2 4) bar-4 bar-2
O: (3 1) 12-9 10-9
X: (1 2) 2-3 6-8
O: (1 2) bar-23
X: (1 1) bar-1 22-23 4-5 5-6
O: (1 6) bar-19
X: (4 1) 13-17 17-18
O: (2 6) bar-19 11-9
X: (5 4) 8-13 13-17
O: (6 4) 19-13 19-15
X: (4 3) 17-21 18-21
O: (2 5) 15-10 13-11
X: (4 6) 3-7 7-13
O: (5 4) 12-7 11-7
X: (3 4) 13-17 17-20
O: (5 1) 12-7 12-11
X: (6 1) 21-22 6-12
O: (2 1) 10-8 9-8
X: (1 5) 12-17 17-18
O: (2 5) 11-6 11-9
X: (5 1) bar-5 5-6
O: (5 2) bar-20 10-8
X: (6 6) bar-6 6-12 12-18 6-12
O: (2 3) 9-6 8-6
X: (1 2) 18-20 18-19
O: (5 3) bar-20 11-8
X: (6 1) bar-1 12-18
O: (5 2) 20-15 15-13
X: (5 1) 18-23 19-20
O: (6 5) 13-7 11-6
X: (6 6) 5-11 11-17 5-11 11-17
O: (6 1) bar-19 19-18
X: (2 6) 17-23 1-3
O: (2 2) 18-16 16-14 14-12 12-10
X: (6 2) 3-5 5-11
O: (5 2) 10-5 7-5
X: (2 5) 11-16 17-19
O: (2 1) 10-8 10-9
X: (5 6) 4-10 10-15
O: (1 2) 8-6 6-5
X: (3 2) 15-18 18-20
O: (3 2) 7-4 6-4
X: (2 1) bar-1 22-24
O: (1 3) 9-6 8-7
X: (6 3) 16-22 21-24
O: (5 4) 8-3 7-3
X: (2 3) 20-23 19-21
O: (6 1) 9-3 3-2
X: (4 1) 1-2 20-24
O: (6 6) bar-19 8-2 7-1 7-1
X: (3 2) bar-2
O: (5 3) bar-20 5-2
X: (4 5) can't move
O: (3 6) 9-3 19-16
X: (4 2) bar-2
O: (3 3) can't move
X: (1 6) can't move
O: (3 5) bar-20 20-17
X: (4 5) can't move
O: (1 2) 3-2 16-14
X: (1 1) can't move
O: (4 5) 14-9 6-2
X: (2 1) can't move
O: (6 5) 17-11 20-15
X: (3 6) can't move
O: (4 5) 15-10 11-7
X: (4 6) can't move
O: (6 4) 10-4 9-5
X: (1 5) can't move
O: (3 2) 7-4 4-2
X: (6 2) can't move
O: (2 4) 6-2 6-4
X: (2 4) can't move
O: (3 2) 5-2 2-off
X: (2 2) can't move
O: (3 1) 4-1 4-3
X: (6 3) bar-6
O: (5 3) 5-off 5-2
X: (3 6) bar-6 6-9
O: (3 4) 4-off 3-off
X: (1 3) 9-12 12-13
O: (1 2) 2-off 1-off
X: (2 3) 13-16 16-18
O: (6 1) 4-off 1-off
X: (1 4) 6-10 18-19
O: (5 1) 3-off 1-off
X: (5 1) 10-15 19-20
O: (2 1) 2-off 2-1
X: (4 4) 20-24 15-19 19-23 21-off
O: (3 5) 3-off 2-off
X: (3 3) 21-24 22-off 22-off 23-off
O: wins
You may export it to your bot and have it analyze the match.
My question is, how gnubg can think both players may have positive
equities?
Another point is that gnubg thinks that NONE of our rolls, both mine
and her, were related to luck, i.e. none of them were lucky, neither
unlucky. How can it be?
Any input would be appreciated.
Regards
Wish I could come close to that, dude.
Never could get to keep my prime that long, just have to get it
straight forward and bear-off already.
>
> You may export it to your bot and have it analyze the match.
>
> My question is, how gnubg can think both players may have positive
> equities?
>
>
> Any input would be appreciated.
>
My input is that you should supply the position id so one could just
type that into gnu instead of exporting an entire match and looking
for it.
Bob Koca
Bob Koca
lol, when I've posted all the moves, you definitely can export the
whole match into your gnubg and see it for yourself. (In FIBS oldmoves
format)
By the way, this is one of positions I was talking about:
Position ID: AIiw7TYAILJtWw ------- Match ID: cIk2AAAAAAAA
According to gnubg, my equity was about +0.9, at the same time my
opponent's was about +0.9 too.
And here's Snowie's analysis of the same position:
http://www.sequit.net/Content/Snowie%20Matches/200805180102_yeti_yew_data/gexprt01.html#Move_54Green
Regards
Position ID: AIiw7TYAILJtWw, Match ID: cIk2AAAAAAAA
>
> lol, when I've posted all the moves, you definitely can export the
> whole match into your gnubg and see it for yourself. (In FIBS oldmoves
> format)
>
> By the way, this is one of positions I was talking about:
>
> Position ID: AIiw7TYAILJtWw ------- Match ID: cIk2AAAAAAAA
>
> According to gnubg, my equity was about +0.9, at the same time my
> opponent's was about +0.9 too.
Not seeing what is so funny. I never disagreed that one couldn't
export the whole match into gnubg and see for oneself. But if you want
someone to answer your questions it makes sense to make it easy for
them.
I think that an odd even effect combined with gnubg having no clue
about how to evaluate the position explains your observation.
Bob Koca
Agreed. Anyway, Gnubg has no clue here:
GNU Backgammon Position ID: QBjAbTsAANy2Ww
Match ID : cIk2AAAAAAAA
+24-23-22-21-20-19------18-17-16-15-14-13-+ O: yeti_yew
| X X X X X X | | O O | 0 points
| X X X X X X | | O |
| X X | | |
| | | |
| | | |
| |BAR| |v 1 point match (Cube:
1)
| | | |
| | | |
| O O | | |
| O O O O O | | | Rolled 55
| O O O O O | X | | 0 points
+-1--2--3--4--5--6-------7--8--9-10-11-12-+ X: PersianLord
X has a closed board -- in O's homeboard! X's homeboard is the lower
left quadrant. The pipcount is X 325, O 293.
1. Cubeful 2-ply bar/20 24/19 23/18* 21/16 Eq.: +0.975
0.988 0.883 0.055 - 0.012 0.005 0.002
2-ply cubeful prune [world class]
X is winning 98.8%. And on O's next turn:
GNU Backgammon Position ID: AJBw2zYADOC2XQ
Match ID : MAEgAAAAAAAA
+-1--2--3--4--5--6-------7--8--9-10-11-12-+ O: yeti_yew
| X X X X X X | O | X O X | 0 points
| X X X X X X | | O | On roll
| X | | |
| | | |
| | | |
| |BAR| |^ 1 point match (Cube:
1)
| | | |
| | | |
| O O | | |
| O O O O O | | |
| O O O O O | | | 0 points
+24-23-22-21-20-19------18-17-16-15-14-13-+ X: PersianLord
1. Cubeful 2-ply bar/22 24/19 Eq.: +0.953
0.977 0.803 0.111 - 0.023 0.007 0.004
2-ply cubeful prune [world class]
Now O is winning 97.7%.
3-ply evaluation doesn't help:
1. Cubeful 3-ply bar/20 24/19 23/18* 21/16 Eq.: -0.957
0.021 0.010 0.008 - 0.979 0.781 0.074
3-ply cubeful prune [grandmaster]
and
1. Cubeful 3-ply bar/22 24/19 Eq.: -0.976
0.012 0.006 0.003 - 0.988 0.825 0.040
3-ply cubeful prune [grandmaster]
Now X is winning 2% and O is winning 1%.
Thank you Raccoon for the input. I think that GNUBG, and perhaps other
bots as well, still has difficulties in analyzing the massive
backgames and also the strength of rolling primes in the outfield when
1 or 2 checkers are behind them. As you may have noticed, Snowie 4 is
as clueless as GNUBG regarding this position.
Regards
> And here's Snowie's analysis of the same position:
>
> http://www.sequit.net/Content/Snowie%20Matches/200805180102_yeti_yew_data/gexprt01.html#Move_54Green
Aaaah, a clean and readable output (I prefer to send it like this to
people too in rare cases).
Awesome. I need to go and post this to our locals... Some might even
understand that bots and backgames are not to be trusted.
Eskimo
--
//------------------------------
//Remove tämä all the way to and including soomee to mail directly.
//Ascended:W,V (genopolywish),P(ill ath), T,K,H,S,B,C,P,W
(naked),Ro,Ra,A,W,almost pacifist A
//In progress:PAIN