| |
Main
Date: 28 Dec 2008 00:12:32
From: Sanny
Subject: Why GetClub is Silent?
|
Many players feel why there was no post by me last week. Its because
now GetClub Chess is under close examination.
A new technique has been evolved and it is as complicated as the
GetClub Chess program itself.
This new technique will be completed in a week time. One week has
already been spent.
After 2 weeks it will be tested to see there is no bug. Since it is a
huge modification 100s of new bugs will come out when new technique is
added to program.
But if the new technique improves the game speed 2-4 times then that
work is worth doing.
In-case the new approach do not give good results. The old program
will remain as it is.
It is very tough time for GetClub. As we do not know what may happen.
And new improvement is so big that 10,000 lines of new code is
generated with 100s of Bugs coming daily.
If this works as predicted GetClub will play 2-4 times faster.
Lets hope we complete this job in 1-2 weeks. And the GetClub Chess
speeds up.
Today I saw Help Bot was able to win against the Normal Level. Well
played by Help Bot. So even after improvement Help Bot will be a
strong challenger to GetClub Program.
The new approach is very complex and we need to sort out 100s of Bugs
in 10,000 lines of code. So It will take around 2 weeks to implement
it fully.
I will be hearing lots of complains of Bug if the testing is not done
correctly. So we will test well before implementing it.
Lets hope the project is successful and we are able to make Getclub
Chess 2-4 times faster.
I remember old days when every week we used to improve the GetClub
speed by a factor of 2. This time it will take 1 month time to do
that.
Its very complex and very challenging to use the new approach. I will
inform you all when the new approach is completed. It is still not
clear this new approach is faster or not. If yes that will be great
for GetClub.
Bye
Sanny
Play Chess at: http://www.GetClub.com/Chess.html
|
|
| |
Date: 29 Dec 2008 21:31:06
From: [email protected]
Subject: Re: Why GetClub is Silent?
|
> No, that would still be beyond his skill set, since it involves being
> accurate to the penny. =A0 I think he needs to try the "unbeatable
> Tic-Tac-Toe" program.
We can see it now. "GitClub twice improved and now plays better moves
with O side. Please look at game and analyze where GitClub make
mistake and lose with X. Big bug will be removed and now only top
player can win him in Tic Tac Toe. Human player will always loose."
|
| | |
Date: 30 Dec 2008 13:47:09
From: Bob Campbell
Subject: Re: Why GetClub is Silent?
|
<[email protected] > wrote in message
news:[email protected]...
>We can see it now. "GitClub twice improved and now plays better moves
>with O side. Please look at game and analyze where GitClub make
>mistake and lose with X. Big bug will be removed and now only top
>player can win him in Tic Tac Toe. Human player will always loose."
LOL, yeah. You forgot that "GitClub Tic-Tac-Toe now plays at 2400 level
with 10 second response"!
|
| |
Date: 29 Dec 2008 23:14:54
From: Simon Krahnke
Subject: Re: Why GetClub is Silent?
|
* Sanny <[email protected] > (16:48) schrieb:
> Yes when attacking pawns near king the program gives +1 advantage.
> This was done and found that it helps delay the Mate against strong
> players.
Who are you? Sanny never talks about any internals.
mfg, simon .... l
|
| |
Date: 29 Dec 2008 10:09:37
From: [email protected]
Subject: Re: Why GetClub is Silent?
|
> A lot more work need to be done. I hope the new improvement will be
> successful and the program plays stronger after two weeks of hard
> work.
Sanny, I know you are working very hard on this program but after many
months of hard work it is still not a good chess program. Maybe you
and your team should turn your talents in another direction. Perhaps
you should write an Accounts Receivable program in Cobol as an
alternative activity.
|
| | |
Date: 29 Dec 2008 20:02:41
From: Bob Campbell
Subject: Re: Why GetClub is Silent?
|
<[email protected] > wrote in message
news:[email protected]...
> Perhaps you should write an Accounts Receivable program in Cobol as an
> alternative activity.
No, that would still be beyond his skill set, since it involves being
accurate to the penny. I think he needs to try the "unbeatable
Tic-Tac-Toe" program.
|
| |
Date: 29 Dec 2008 07:48:41
From: Sanny
Subject: Re: Why GetClub is Silent?
|
> =A0 Oner thing I don't quite understand is how
> the GetClub program can sometimes be in
> book for several moves, while at other times
> it seems to have no openings book at all,
> going into a deep think at move one.
>
> =A0 I've also noticed that whenever either side
> is attacking squares near the enemy King,
> the position score gets way out-of-whack,
> crediting even useless "attacks" as equal
> to a piece, for instance. =A0Apparently, the
> program cannot tell the difference between
> a truly dangerous attack, and harmlessly
> sweeping across empty space near to the
> enemy King.
>
> =A0 It's amazing how, in spite of innumerable
> doublings in calculating speed, the figures
> shown on the Web page as move times
> are still too high-- missing the mark
> substantially. =A0 If, say, a given level says it
> takes 20-40 minutes, you can bet that it
> really averages an hour per move or more.
> Too bad these clever programmers can't
> do simple math.
>
> =A0 For some reason, I have been walking all
> over the GetClub program on various
> different levels as White, but when I have
> Black the games are a bit more difficult,
> sometimes even taking quite a few moves
> to finish up. =A0 I'm guessing that GetClub's
> opening moves as Black are not quite up
> to snuff, allowing White to gain a large
> advantage alomost effortlessly.
Good points, I will take care of them.
I find sometimes it do not play opening moves even when it knows the
opening. That error has occured sometimes. The real reasion is not yet
known.
Yes when attacking pawns near king the program gives +1 advantage.
This was done and found that it helps delay the Mate against strong
players.
Chrisf usually used to sacrifice its pawn and attack the King and win.
It was done So that GetClub do not take pawns when its King is under
attack.
This helped many times. As now GetClub tries to defend its king from
attacks and save its king at proper place.
A lot more work need to be done. I hope the new improvement will be
successful and the program plays stronger after two weeks of hard
work.
Bye
Sanny
Play Chess at: http://www.GetClub.com/Chess.html
|
| |
Date: 28 Dec 2008 23:16:43
From: help bot
Subject: Re: Why GetClub is Silent?
|
On Dec 28, 3:12=A0am, Sanny <[email protected] > wrote:
> Many players feel why there was no post by me last week. Its because
> now GetClub Chess is under close examination.
>
> A new technique has been evolved and it is as complicated as the
> GetClub Chess program itself.
>
> This new technique will be completed in a week time. One week has
> already been spent.
>
> After 2 weeks it will be tested to see there is no bug. Since it is a
> huge modification 100s of new bugs will come out when new technique is
> added to program.
>
> But if the new technique improves the game speed 2-4 times then that
> work is worth doing.
>
> In-case the new approach do not give good results. The old program
> will remain as it is.
>
> It is very tough time for GetClub. As we do not know what may happen.
> And new improvement is so big that 10,000 lines of new code is
> generated with 100s of Bugs coming daily.
>
> If this works as predicted GetClub will play 2-4 times faster.
>
> Lets hope we complete this job in 1-2 weeks. And the GetClub Chess
> speeds up.
>
> Today I saw Help Bot was able to win against the Normal Level. Well
> played by Help Bot. So even after improvement Help Bot will be a
> strong challenger to GetClub Program.
>
> The new approach is very complex and we need to sort out 100s of Bugs
> in 10,000 lines of code. So It will take around 2 weeks to implement
> it fully.
>
> I will be hearing lots of complains of Bug if the testing is not done
> correctly. So we will test well before implementing it.
>
> Lets hope the project is successful and we are able to make Getclub
> Chess 2-4 times faster.
>
> I remember old days when every week we used to improve the GetClub
> speed by a factor of 2. This time it will take 1 month time to do
> that.
>
> Its very complex and very challenging to use the new approach. I will
> inform you all when the new approach is completed. It is still not
> clear this new approach is faster or not. If yes that will be great
> for GetClub.
Oner thing I don't quite understand is how
the GetClub program can sometimes be in
book for several moves, while at other times
it seems to have no openings book at all,
going into a deep think at move one.
I've also noticed that whenever either side
is attacking squares near the enemy King,
the position score gets way out-of-whack,
crediting even useless "attacks" as equal
to a piece, for instance. Apparently, the
program cannot tell the difference between
a truly dangerous attack, and harmlessly
sweeping across empty space near to the
enemy King.
It's amazing how, in spite of innumerable
doublings in calculating speed, the figures
shown on the Web page as move times
are still too high-- missing the mark
substantially. If, say, a given level says it
takes 20-40 minutes, you can bet that it
really averages an hour per move or more.
Too bad these clever programmers can't
do simple math.
For some reason, I have been walking all
over the GetClub program on various
different levels as White, but when I have
Black the games are a bit more difficult,
sometimes even taking quite a few moves
to finish up. I'm guessing that GetClub's
opening moves as Black are not quite up
to snuff, allowing White to gain a large
advantage alomost effortlessly.
-- help bot
|
|
