| |
Main
Date: 07 Mar 2007 22:37:05
From: Sanny
Subject: How GetClub Chess game developed so Strong.
|
GetClub game speed and strength have kept improving because of good
players playing it and beating it. But many players are not playing it
thinking it to be as weak as it was 3 months back.
Try a game: http://www.GetClub.com/Chess.html
GetClub Game was designed in May 2006
Jun 2006 its all illegal Moves were Removed.
July 2006 it was playing good enough to complete a game.
Aug 2006 Game Strength was increased
Sep 2006 Opening was added
Oct 2006 New rules were added
Nov 2006 It played taking long time 5-10 min/ move
Dec 2006 Time was reduced to maximum 1 min/ move
Jan 2007 Great Speed Jump done
Feb 2007 Game strength was increased
So one can see every month there was a bit improvement in the game. A
player who had played say in Nov 2006 has not seen the improvements
made later and he still believes the game is weak despite so many
improvements.
Still many improvements are needed but I am not getting much
suggestions out of the Box. Good players let me know if you can still
beat it's Beginner & Easy Levels.
Bye
Sanny
Play Chess at: http://www.GetClub.com/Chess.html
|
|
| |
Date: 13 Mar 2007 18:26:29
From: help bot
Subject: Re: This month a new Improvement done.
|
On 12, 11:24 pm, "Chess Freak" <[email protected] > wrote:
> At "expert level" on getclub.com, Sanny's program played
>
> 1. e4 e6
> 2. d4 d5
> 3. Nc3 Nf6
> 4. e5 h6???
>
> It even took about 400 seconds to make that killer 4. ... h6
> move!
Okay, there are a few glitches. Suppose you were to
have what appears to be a lock-up, or maybe you started
another game in the same browser as the previous one.
I have seen Sanny's program take "forever" and refuse to
move, but if I reboot my machine it suddenly decides to
move quickly, despite not having reached anywhere near
the depth it had analyzed before, when it would not move.
Strange things can happen, but this is not proof that his
program is horrible on the whole.
In the case you describe above, my first guess is that
Sanny mistyped the book moves, mixing up different lines.
For example, in one popular line the move ...h6 is played
after ...B-b4 and ...Nf6, but there would of course be a
White Bishop on g5. Now why would the program take
400 seconds if it were still "in book", I don't know. It is
possible that the tactical search is so messed up that it
just failed to see the piece hang, of course. But the vast
majority of the time -- even on its weakest levels -- it sees
the simple stuff now, though it wasn't always so.
Here's an example of how it handles tactics on Easy
level: Yesterday, I steadily managed to chalk up a full Rook
advantage (plus several pawns, to boot) and walked my
King right up the board to set up a mating net. But before
I could execute, the program resigned by stopping the
game, for it correctly foresaw that in every line it was a
mate in two coming. This seems to be the typical case
these days; like a weak human player, the program drops
material but it sees enough to put up a bit of resistance.
Recently, I have played just three games at GetClub,
and not one entailed anything like what you describe
above. Two were against Easy level, and one was vs.
(the even easier) Beginner level. All three games
progressed at a reasonable speed, too, but none so
quick that I could afford to try the top levels again on
my notebook, with its rapid battery drain rate.
Okay, here is another possibility in your game: the chess
program was "down" for repairs, so Sanny bravely stepped
in and took the reigns. ; >D
-- help bot
|
| |
Date: 12 Mar 2007 20:01:04
From: help bot
Subject: Re: This month a new Improvement done.
|
On 12, 3:26 am, "Sanny" <[email protected] > wrote:
> ch 2007: Game strength further improved.
>
> Now it will be twice tougher than the games it played in Feb 2007.
>
> If The game continues to improve like earlier, I am sure beginner
> level will reach a rating of 1800 in next 2-3 months. Currently
> Beginner plays like a 1600 Rated Player.
>
> Play 5/10 Games at:http://www.GetClub.com/Chess.html
>
> If you win those game then Tell my the moves where computer was wrong
> and What was the correct move. I will see that mistake do not repeat
> in future.
One way to improve the program would be to have it
"hang" or lock up whenever I make a move (i.e. whenever
I am likely winning) so it never loses to me. This same
strategy should also work against Taylor Kingston and
"Bob". No, wait -- that's not fair!
Okay, maybe we should make a few comments after
the games so you will have some clue as to why the
program lost. But, again, the approach whereby we
state the "correct" move for you to hardwire in is far
from any real solution, as there is a nearly *infinite*
number of possible chess positions. In short, the right
way is to "teach" the program to find better moves on
its own.
Let me give an example of what I mean. In a recent
game at RedHotPawn, the play went:
1.e4 c6
2.Bc4 d5
3.ed ed
4.Bb5+ Nc6
5.d3
Now here I noticed a tactical problem in that after
my next move the only response to not immediately
hang the Bishop is Nc3, which allows p-d4, attacking
the pinned Knight and (apart from any complications),
snaring a piece.
5. ... Qa5+
6.Nc3
Here I note that after my attack on the pinned Knight,
my opponent can easily trade his Bishop *with check*,
thereby leaving only the Knight to be reckoned with.
In addition, there is another issue: White can try Q-f3,
with tactical threats of his own which might well save
the day or perhaps even turn the tables. Nevertheless,
I proceed as planned.
6. ... d4
7.Bd2
The game is lost, for my opponent did not see the
one move which might have saved him: Qf3.
Now, suppose this game had been between the
program at GetClub and me, and I were to tell you
that in the opening, White erred with 2.Bc4. In no
way would addressing that problem help the more
serious ones which followed. You would perhaps
type in an automatic 2.d4, but the program would
not avoid making this sort of error in other, similar
positions which might arise. The real fix is to go in
and rehash the tactical search, so that the program
could "see" all the basic tactical ramifications of
the move 5.d3 using check-and-capture extensions.
------------
It's really a simple case of: give a chess program a
fish, you feed it for a day; TEACH a chess program
to fish, you feed it for life.
-- help bot
("...a font of wisdom" -- Taylor Kingston)
|
| | |
Date: 12 Mar 2007 20:24:48
From: Chess Freak
Subject: Re: This month a new Improvement done.
|
At "expert level" on getclub.com, Sanny's program played
1. e4 e6
2. d4 d5
3. Nc3 Nf6
4. e5 h6???
It even took about 400 seconds to make that killer 4. ... h6
move!
"help bot" <[email protected] > wrote in message
news:[email protected]...
> On 12, 3:26 am, "Sanny" <[email protected]> wrote:
>
>> ch 2007: Game strength further improved.
>>
>> Now it will be twice tougher than the games it played in Feb 2007.
>>
>> If The game continues to improve like earlier, I am sure beginner
>> level will reach a rating of 1800 in next 2-3 months. Currently
>> Beginner plays like a 1600 Rated Player.
>>
>> Play 5/10 Games at:http://www.GetClub.com/Chess.html
>>
>> If you win those game then Tell my the moves where computer was wrong
>> and What was the correct move. I will see that mistake do not repeat
>> in future.
>
> One way to improve the program would be to have it
> "hang" or lock up whenever I make a move (i.e. whenever
> I am likely winning) so it never loses to me. This same
> strategy should also work against Taylor Kingston and
> "Bob". No, wait -- that's not fair!
>
> Okay, maybe we should make a few comments after
> the games so you will have some clue as to why the
> program lost. But, again, the approach whereby we
> state the "correct" move for you to hardwire in is far
> from any real solution, as there is a nearly *infinite*
> number of possible chess positions. In short, the right
> way is to "teach" the program to find better moves on
> its own.
>
> Let me give an example of what I mean. In a recent
> game at RedHotPawn, the play went:
>
> 1.e4 c6
> 2.Bc4 d5
> 3.ed ed
> 4.Bb5+ Nc6
> 5.d3
>
> Now here I noticed a tactical problem in that after
> my next move the only response to not immediately
> hang the Bishop is Nc3, which allows p-d4, attacking
> the pinned Knight and (apart from any complications),
> snaring a piece.
>
> 5. ... Qa5+
> 6.Nc3
>
> Here I note that after my attack on the pinned Knight,
> my opponent can easily trade his Bishop *with check*,
> thereby leaving only the Knight to be reckoned with.
> In addition, there is another issue: White can try Q-f3,
> with tactical threats of his own which might well save
> the day or perhaps even turn the tables. Nevertheless,
> I proceed as planned.
>
> 6. ... d4
> 7.Bd2
>
> The game is lost, for my opponent did not see the
> one move which might have saved him: Qf3.
>
> Now, suppose this game had been between the
> program at GetClub and me, and I were to tell you
> that in the opening, White erred with 2.Bc4. In no
> way would addressing that problem help the more
> serious ones which followed. You would perhaps
> type in an automatic 2.d4, but the program would
> not avoid making this sort of error in other, similar
> positions which might arise. The real fix is to go in
> and rehash the tactical search, so that the program
> could "see" all the basic tactical ramifications of
> the move 5.d3 using check-and-capture extensions.
>
> ------------
>
> It's really a simple case of: give a chess program a
> fish, you feed it for a day; TEACH a chess program
> to fish, you feed it for life.
>
> -- help bot
>
> ("...a font of wisdom" -- Taylor Kingston)
>
>
>
>
>
>
>
>
|
| |
Date: 12 Mar 2007 00:26:19
From: Sanny
Subject: This month a new Improvement done.
|
On 8, 11:37 am, "Sanny" <[email protected] > wrote:
> GetClubgame speed and strength have kept improving because of good
> players playing it and beating it. But many players are not playing it
> thinking it to be as weak as it was 3 months back.
>
> Try a game:http://www.GetClub.com/Chess.html
>
> GetClubGame was designed in May 2006
>
> Jun 2006 its all illegal Moves were Removed.
> July 2006 it was playing good enough to complete a game.
> Aug 2006 Game Strength was increased
> Sep 2006 Opening was added
-----
------
> Dec 2006 Time was reduced to maximum 1 min/ move
> Jan 2007 Great Speed Jump done
> Feb 2007 Game strength was increased
>
Now to add this month improvements to the list
ch 2007: Game strength further improved.
Now it will be twice tougher than the games it played in Feb 2007.
If The game continues to improve like earlier, I am sure beginner
level will reach a rating of 1800 in next 2-3 months. Currently
Beginner plays like a 1600 Rated Player.
Play 5/10 Games at: http://www.GetClub.com/Chess.html
If you win those game then Tell my the moves where computer was wrong
and What was the correct move. I will see that mistake do not repeat
in future.
Say 20. N-f3 was wrong because ..... And Correct move is 20. P-f3
Bye
Sanny
Play Chess at: http://www.GetClub.com/Chess.html
|
| |
Date: 08 Mar 2007 13:18:41
From: Kenneth Sloan
Subject: Re: How GetClub Chess game developed so Strong.
|
Sanny wrote:
> GetClub game speed and strength have kept improving because of good
> players playing it and beating it. But many players are not playing it
> thinking it to be as weak as it was 3 months back.
>
> Try a game: http://www.GetClub.com/Chess.html
>
OK...here goes...
a) no move listing
b) 20-second delay, every so often, between when the user drops a piece
on a new square and when the piece actually is displayed on that square
c) program oversteps the time control on every move - according to it's
own clock
d) the clock is inaccurate
e) 20-second delay, every so often, between when the clock stops, and
when the board is updated
f) program drops a piece for nothing around move 15
g) I intentionally give back the piece and trade down to a R+2P ending,
allowing the program the distant passed pawn
h) program drops the other pawn (giving me two connected passed pawns
protected by - and protecting - my K)
i) programs distant passed pawn is never pushed
j) I queen
k) it's now mate in one, and the program reports that it has taken 1060
seconds searching for a way out. the program has 3 legal moves, and my
response to all of them is the same: Qh8++. The program claims to be
"Thinking on King (7)"
You are right - this version is much better than the last version.
--
Kenneth Sloan [email protected]
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/
|
| | |
Date: 08 Mar 2007 11:25:58
From: Mike Murray
Subject: Re: How GetClub Chess game developed so Strong.
|
On Thu, 08 2007 13:18:41 -0600, Kenneth Sloan
<[email protected] > wrote:
>You are right - this version is much better than the last version.
Maybe we start worrying about somebody planting GetClub on a MonRoi ?
Heh, heh, heh.
|
| |
Date: 08 Mar 2007 07:22:04
From: Ron
Subject: Re: How GetClub Chess game developed so Strong.
|
I see another timeline:
Jun 2006: Sanny lies about the strength and speed of his program.
July 2006: Sanny lies about the strength and speed of his program.
Aug 2006: Sanny lies about the strength and speed of his program.
Sep 2006: Sanny lies about the strength and speed of his program.
Oct 2006: Sanny lies about the strength and speed of his program.
Nov 2006: Sanny lies about the strength and speed of his program.
Dec 2006: Sanny lies about the strength and speed of his program.
Jan 2007: Sanny lies about the strength and speed of his program.
Feb 2007: Sanny lies about the strength and speed of his program.
|
| | |
Date: 09 Mar 2007 18:59:30
From: Terry
Subject: Re: How GetClub Chess game developed so Strong.
|
"Ron" <[email protected] > wrote in message
news:[email protected]...
>
> I see another timeline:
>
> Jun 2006: Sanny lies about the strength and speed of his program.
> July 2006: Sanny lies about the strength and speed of his program.
> Aug 2006: Sanny lies about the strength and speed of his program.
> Sep 2006: Sanny lies about the strength and speed of his program.
> Oct 2006: Sanny lies about the strength and speed of his program.
> Nov 2006: Sanny lies about the strength and speed of his program.
> Dec 2006: Sanny lies about the strength and speed of his program.
> Jan 2007: Sanny lies about the strength and speed of his program.
> Feb 2007: Sanny lies about the strength and speed of his program.
LOL
|
|
