Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old January 1st 10, 03:03 AM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jul 2006
Posts: 625
Default Hej, a sanny thread

Hej,

no living Sanny thread? I felt the need to do a game on his ugly site,
anyway.

Here it is, Rybka 2.2 32bit with 1 ply search depth against Getclub easy
level:

[Date "2010.01.01"]
[White "Rybka 2.2 32 bit"]
[Black "Getclub easy"]

1. Nc3 {(Nc3) +0.12/1 0} d5
2. Nf3 {(Nf3) +0.18/1 0} Nc6 { out of book here, really, on second move }
3. d3 {(d3) +0.08/1 0} Nf6
4. Bf4 {(Bf4) +0.13/1 0} Bf5
5. Nb5 {(Nb5) +0.12/1 0} e5 { why would black attempt this exchange? }
6. Nxe5 {(Nxe5) +0.31/1 0} Nxe5
7. Bxe5 {(Bxe5) +0.35/1} Bd6
8. Bxd6 {(Bxd6) +0.43/1 0} cxd6 { isolated doubled pawn! }
9. Nd4 {(Nd4) +0.19/1 0} Bg4 { why move once, when you can move twice? }
10. h3 {(h3) +0.40/1 0} Bd7 { what was the job of that bishop on the kingside? }
11. Qd2 {(Qd2) +0.17/1 0} O-O
12. Qb4 {(Qb4) +0.21/1 0} Qc7
13. g4 {(g4) +0.22/1 0} Rac8 { oh, this power! }
14. Bg2 {(Bg2) +0.34/1 0} Rfe8
15. O-O {(0-0) +0.40/1 0} Re5 { yippie! }
16. f4 {(f4) +0.39/1 0} Re3
17. g5 {(g5) +0.39/1 0} a5
18. Qd2 {(Qd2) +1.83/1 0} Rg3 { "but i want to lose this rook!" }
19. gxf6 {(gxf6) +1.77/1 0} Bxh3 { this put-your-pieces-near-the-opponent's-king-strategy again? }
20. Kh2 {(Kh2) +2.14/1 0} Bxg2
21. Kxg3 {(Kxg3) +2.24/1 0} Bxf1
22. Rxf1 {(Rxf1) +2.42/1 0} gxf6 { two isolated doubled pawns }
23. Rg1 {(Rg1) +2.74/1 0} Qc5 { the action is on the other side, girl! }
24. Qe3 {(Qe3) +2.79/1 0} Rb8 { let's protect this unattacked backward pawn }
25. Kh4+ {(Kh4+) +3.12/1 0} Kh8 { king's best place: h8! }
26. Qg3 {(Qg3) +319.00/1 0} h6 { this is the place, where Getclub normally gives up }
27. Qg7# {(Qg7+) +319.00/1 0} { THE END }
1-0

Just let Rybka have white, and it's a quick win in 27 moves. Quick for
Rybka, it took less than a second, while Getclub took around 40 minutes.

And Rybka actually won, Getclub actually lost. It didn't give up at the
last moment. It's mate!

Getclub just walked into it, and maybe Sanny fixed this issue. But then
again, when Getclub is stupid enough to walk into a mate, why would it
find it on its last move?

I won't post a URL, someone might complain, and the list of moves is
****ed up, again.

mfg, simon .... 4 at night, all quiet!
  #2   Report Post  
Old January 1st 10, 07:37 AM posted to rec.games.chess.computer,rec.games.chess.misc
Banned
 
First recorded activity by ChessBanter: Dec 2008
Posts: 2,402
Default Hej, a sanny thread

Getclub just walked into it, and maybe Sanny fixed this issue. But then
again, when Getclub is stupid enough to walk into a mate, why would it
find it on its last move?


The work is in progress. Lets hope my computer gets repaired this
week.

I won't post a URL, someone might complain, and the list of moves is
****ed up, again.



This problem will be sorted out as soon as my computer is repaired.

I have a few ideas for GetClub. That may make GetClub 3-4 times
stronger than now.

But It will take 1-2 weeks to bring the changes.

Thanks, for playing game. I will see that the problem of false moves
recording is sorted out as soon as possible.

Bye
Sanny

--
Play Chess at: http://www.GetClub.com/Chess.html

Why Play at GetClub.com Chess.
1. Its free to play. And view Animated Games. [NO DOWNLOADS]
2. Your games are recorded.along with time taken by you.
3. Your games are analyzed by higher Levels
4. You win Prizes every month. Also get a RATING
5. Earn $1.00 for playing 30 games every month. [Win]
6. Only those games in which you win are counted for $1.00 every
month.

Start a game: http://www.GetClub.com/Chess.html
  #3   Report Post  
Old January 1st 10, 08:23 AM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jul 2006
Posts: 625
Default Hej, a sanny thread

* Sanny (08:37) schrieb:

Getclub just walked into it, and maybe Sanny fixed this issue. But then
again, when Getclub is stupid enough to walk into a mate, why would it
find it on its last move?


The work is in progress. Lets hope my computer gets repaired this
week.


Your computer is broken? Sorry to read that.

I have a few ideas for GetClub. That may make GetClub 3-4 times
stronger than now.


Maybe then, Vas' work on the search algorithm will be worth something
against Getclub.

mfg, simon .... l
  #4   Report Post  
Old January 1st 10, 12:04 PM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jul 2006
Posts: 1,015
Default Hej, a sanny thread

Simon Krahnke wrote:
Hej,

no living Sanny thread? I felt the need to do a game on his ugly site,
anyway.

Here it is, Rybka 2.2 32bit with 1 ply search depth against Getclub easy
level:

[Date "2010.01.01"]
[White "Rybka 2.2 32 bit"]
[Black "Getclub easy"]

1. Nc3 {(Nc3) +0.12/1 0} d5
2. Nf3 {(Nf3) +0.18/1 0} Nc6 { out of book here, really, on second move }
3. d3 {(d3) +0.08/1 0} Nf6
4. Bf4 {(Bf4) +0.13/1 0} Bf5
5. Nb5 {(Nb5) +0.12/1 0} e5 { why would black attempt this exchange? }
6. Nxe5 {(Nxe5) +0.31/1 0} Nxe5
7. Bxe5 {(Bxe5) +0.35/1} Bd6
8. Bxd6 {(Bxd6) +0.43/1 0} cxd6 { isolated doubled pawn! }
9. Nd4 {(Nd4) +0.19/1 0} Bg4 { why move once, when you can move twice? }
10. h3 {(h3) +0.40/1 0} Bd7 { what was the job of that bishop on the kingside? }
11. Qd2 {(Qd2) +0.17/1 0} O-O
12. Qb4 {(Qb4) +0.21/1 0} Qc7
13. g4 {(g4) +0.22/1 0} Rac8 { oh, this power! }
14. Bg2 {(Bg2) +0.34/1 0} Rfe8
15. O-O {(0-0) +0.40/1 0} Re5 { yippie! }
16. f4 {(f4) +0.39/1 0} Re3
17. g5 {(g5) +0.39/1 0} a5


It is surprising here that GetClub could not find
17. ... Nh5 which moves the threatened piece and puts it back on the
path to equality. 18. Bxd5 is punished by a decent kingside attack.

What black cannot afford to do is allow Qd2 and then move the rook into
immediate peril, but that is how the game proceeds. It is all over bar
the shouting at this stage.


18. Qd2 {(Qd2) +1.83/1 0} Rg3 { "but i want to lose this rook!" }
19. gxf6 {(gxf6) +1.77/1 0} Bxh3 { this put-your-pieces-near-the-opponent's-king-strategy again? }
20. Kh2 {(Kh2) +2.14/1 0} Bxg2
21. Kxg3 {(Kxg3) +2.24/1 0} Bxf1
22. Rxf1 {(Rxf1) +2.42/1 0} gxf6 { two isolated doubled pawns }
23. Rg1 {(Rg1) +2.74/1 0} Qc5 { the action is on the other side, girl! }
24. Qe3 {(Qe3) +2.79/1 0} Rb8 { let's protect this unattacked backward pawn }
25. Kh4+ {(Kh4+) +3.12/1 0} Kh8 { king's best place: h8! }


This is quite incredible. It chose the forced mate line against the only
alternative Kf8 which holds out much longer. It is more like a helpmate.

26. Qg3 {(Qg3) +319.00/1 0} h6 { this is the place, where Getclub normally gives up }


Perhaps a "bug" in the code prevents it spuriously giving up if every
possible outcome is mate #1.

27. Qg7# {(Qg7+) +319.00/1 0} { THE END }
1-0

Just let Rybka have white, and it's a quick win in 27 moves. Quick for
Rybka, it took less than a second, while Getclub took around 40 minutes.


Is that less than a second in aggregate timing?

And Rybka actually won, Getclub actually lost. It didn't give up at the
last moment. It's mate!


Seems like it didn't see the mate coming which is hard to understand.

Happy New Year!

Regards,
Martin Brown
  #5   Report Post  
Old January 2nd 10, 04:53 AM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jul 2006
Posts: 625
Default Hej, a sanny thread

* (13:04) schrieb:

It is surprising here that GetClub could not find
17. ... Nh5 which moves the threatened piece and puts it back on the
path to equality. 18. Bxd5 is punished by a decent kingside attack.

What black cannot afford to do is allow Qd2 and then move the rook into
immediate peril, but that is how the game proceeds. It is all over bar
the shouting at this stage.


Would you try a game against Rybka at 1 ply per move? I would like to
know, how strong that really is.

18. Qd2 {(Qd2) +1.83/1 0} Rg3 { "but i want to lose this rook!" }
19. gxf6 {(gxf6) +1.77/1 0} Bxh3 { this put-your-pieces-near-the-opponent's-king-strategy again? }
20. Kh2 {(Kh2) +2.14/1 0} Bxg2
21. Kxg3 {(Kxg3) +2.24/1 0} Bxf1
22. Rxf1 {(Rxf1) +2.42/1 0} gxf6 { two isolated doubled pawns }
23. Rg1 {(Rg1) +2.74/1 0} Qc5 { the action is on the other side, girl! }
24. Qe3 {(Qe3) +2.79/1 0} Rb8 { let's protect this unattacked backward pawn }
25. Kh4+ {(Kh4+) +3.12/1 0} Kh8 { king's best place: h8! }


This is quite incredible. It chose the forced mate line against the only
alternative Kf8 which holds out much longer. It is more like a helpmate.


Yeah, Sanny, if you ask for a wrong move, here it is.

26. Qg3 {(Qg3) +319.00/1 0} h6 { this is the place, where Getclub normally gives up }


Perhaps a "bug" in the code prevents it spuriously giving up if every
possible outcome is mate #1.


Maybe some day I will try to look at the code again.

27. Qg7# {(Qg7+) +319.00/1 0} { THE END }
1-0

Just let Rybka have white, and it's a quick win in 27 moves. Quick for
Rybka, it took less than a second, while Getclub took around 40 minutes.


Is that less than a second in aggregate timing?


I think so. It's what Arena shows. All the moves were shown as done in
0s, and it adds them up as 0s. The responses were all below 100ms, that
it, instantly.

And Rybka actually won, Getclub actually lost. It didn't give up at the
last moment. It's mate!


Seems like it didn't see the mate coming which is hard to understand.


Selective Search. But even that makes one wonder, what it though about
on the last two moves.

Maybe at 25. ... Kh8 it thought about Kf8 for a while, found it to be
inferior and then just chose the other move.

mfg, simon .... l


  #6   Report Post  
Old January 2nd 10, 06:22 AM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jan 2006
Posts: 470
Default Hej, a sanny thread

On Jan 1, 8:53*pm, Simon Krahnke wrote:

Seems like it didn't see the mate coming which is hard to understand.


Selective Search. But even that makes one wonder, what it though about
on the last two moves.

Maybe at 25. ... Kh8 it thought about Kf8 for a while, found it to be
inferior and then just chose the other move.


Inferior to a mate in 2??? There is no reason for this at all, except
for a horrific bug.

Maybe some aggressive pruning could cause the engine to hold off
seeing the mate in 2 until depth 7 or 8, but that still shouldn't take
more than one or two seconds to see. You don't show the times for each
of GetCrap's moves, but since you said that it took 40 minutes, and
made only 27 moves, I think it's pretty safe to assume that at least
15 seconds was taken on this terrible move. Looks like GC hasn't
improved, in terms of total playing strength (i.e. the ability to
score points), in all these years if it can't even avoid a mate in 2.

jm
  #7   Report Post  
Old January 3rd 10, 12:12 AM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jul 2006
Posts: 625
Default Hej, a sanny thread

* (07:22) schrieb:

On Jan 1, 8:53*pm, Simon Krahnke wrote:

Seems like it didn't see the mate coming which is hard to understand.


Selective Search. But even that makes one wonder, what it though about
on the last two moves.

Maybe at 25. ... Kh8 it thought about Kf8 for a while, found it to be
inferior and then just chose the other move.


Inferior to a mate in 2??? There is no reason for this at all, except
for a horrific bug.


My suggestion is, that it didn't look at Kh8 enough to see it was a mate
in 2 and chose it because Kf8 is negative.

Maybe some aggressive pruning could cause the engine to hold off
seeing the mate in 2 until depth 7 or 8, but that still shouldn't take
more than one or two seconds to see.


Selective Search leaves some moves totally unevaluated, you can easily
miss mates that way.

You don't show the times for each of GetCrap's moves,


I don't have them, Arenas times include my operator time, and the list
of moves on site is ****ed up.

but since you said that it took 40 minutes,


That's my estimate based on Arena's total time for black and a guess on
my operator time.

and made only 27 moves, I think it's pretty safe to assume that at
least 15 seconds was taken on this terrible move.


Getclubs list mentions "32. Kg8-h8 (76)", I guess that means it spent 76
seconds on this move.

Looks like GC hasn't improved, in terms of total playing strength
(i.e. the ability to score points), in all these years if it can't
even avoid a mate in 2.


That obvious from its unability to win any games.

mfg, simon .... l
  #8   Report Post  
Old January 4th 10, 05:00 PM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jan 2006
Posts: 470
Default Hej, a sanny thread

On Jan 2, 4:12*pm, Simon Krahnke wrote:
* (07:22) schrieb:

On Jan 1, 8:53*pm, Simon Krahnke wrote:


Seems like it didn't see the mate coming which is hard to understand.


Selective Search. But even that makes one wonder, what it though about
on the last two moves.


Maybe at 25. ... Kh8 it thought about Kf8 for a while, found it to be
inferior and then just chose the other move.


Inferior to a mate in 2??? There is no reason for this at all, except
for a horrific bug.


My suggestion is, that it didn't look at Kh8 enough to see it was a mate
in 2 and chose it because Kf8 is negative.

Maybe some aggressive pruning could cause the engine to hold off
seeing the mate in 2 until depth 7 or 8, but that still shouldn't take
more than one or two seconds to see.


Selective Search leaves some moves totally unevaluated, you can easily
miss mates that way.


True, you can miss LONGER mates. But for a mate in 2 that move would
only remain unevaluated for early search depths, as I mentioned. With
only two legal moves, this should take no time at all.

My engine is rated around 1950, and it avoids Kh8 in a quarter of a
second at depth 8 (as I said, it has some aggressive pruning that puts
off seeing the mate for a few search depths). Let it think for 76
seconds (on my P4-2.4GHz) and it's already evaluating depth 14.

I doubt there is a single decent engine (and by decent, I mean rated
at least 1600-1700) that would play Kh8 even if it had only a few
seconds to think. Of course, GetClub is not decent, so it played it.

jm
  #9   Report Post  
Old January 4th 10, 06:02 PM posted to rec.games.chess.computer,rec.games.chess.misc
Banned
 
First recorded activity by ChessBanter: Dec 2008
Posts: 2,402
Default Hej, a sanny thread

I doubt there is a single decent engine (and by decent, I mean rated
at least 1600-1700) that would play Kh8 even if it had only a few
seconds to think. Of course, GetClub is not decent, so it played it.


This problem was sorted out today.

Now GetClub will see such mates.

Just play a new game and see how good GetClub plays now.

Bye
Sanny

Play Chess at: http://www.GetClub.com/Chess.html


  #10   Report Post  
Old January 4th 10, 08:48 PM posted to rec.games.chess.computer,rec.games.chess.misc
external usenet poster
 
First recorded activity by ChessBanter: Jan 2006
Posts: 470
Default Hej, a sanny thread

On Jan 4, 10:02*am, Sanny wrote:
I doubt there is a single decent engine (and by decent, I mean rated
at least 1600-1700) that would play Kh8 even if it had only a few
seconds to think. Of course, GetClub is not decent, so it played it.


This problem was sorted out today.

Now GetClub will see such mates.

Just play a new game and see how good GetClub plays now.


I doubt it. I'll let other people who are much more masochistic than I
am play on your site.

jm
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Terry Winchester Locks Sam Sloan Thread AGAIN samsloan rec.games.chess.politics (Chess Politics) 3 June 21st 09 11:22 PM
The Sanny Loop MeaningWhat rec.games.chess.computer (Computer Chess) 0 December 12th 08 01:38 PM
Sanny vs Ivan , computer , ELO I700 to I800 ELO . will be a goodgame SAT W-7 rec.games.chess.misc (Chess General) 25 February 23rd 08 07:37 AM
not to worry; I have a copy of the thread through 2:37 p.m. Saturday [email protected] rec.games.chess.politics (Chess Politics) 0 January 1st 07 08:22 AM


All times are GMT +1. The time now is 10:54 AM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Copyright 2004-2019 ChessBanter.
The comments are property of their posters.
 

About Us

"It's about Chess"

 

Copyright © 2017