[OT-UCI] Fruit 0.98a beta testers

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 09 March 2004 19:51:05:

ftp://cap.connx.com/chess-engines/new-a ... it_src.exe


my ftp site {remove http:// unless you like error messages}

[Peter Fendrich]

Geschrieben von: / Posted by: Peter Fendrich at 09 March 2004 20:29:42:
Als Antwort auf: / In reply to: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 09 March 2004 19:51:05:

ftp://cap.connx.com/chess-engines/new-approach/fruit_src.exe

What is it?
/Peter

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 09 March 2004 20:48:15:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Peter Fendrich at 09 March 2004 20:29:42:

ftp://cap.connx.com/chess-engines/new-approach/fruit_src.exe

What is it?

A UCI chess engine.
And a butt-kicker. It might be the next Ruffian.
No opening book, no tablebase.
Here is what I have so far from testing:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3253 0 0 1 100.0 % 2653 0.0 %
2 Arasanx-74a : 2653 0 0 1 50.0 % 2653 100.0 %
3 Aristarch 4.41: 2653 0 0 1 50.0 % 2653 100.0 %
4 Fruit_src : 2604 344 268 7 71.4 % 2445 28.6 %
5 31337-75 : 2053 0 0 1 0.0 % 2653 0.0 %
6 Abrok_5_0 : 2053 0 0 1 0.0 % 2653 0.0 %
7 Amateur270 : 2053 0 0 1 0.0 % 2653 0.0 %
8 Averno070 : 2053 0 0 1 0.0 % 2653 0.0 %

Just so everyone can breath a sigh of relief, the source code looks nothing like crafty or gnuchess or tscp or amy, etc.



my ftp site {remove http:// unless you like error messages}

[Steve Maughan]

Geschrieben von: / Posted by: Steve Maughan at 09 March 2004 20:51:17:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 09 March 2004 20:48:15:

Dann,

What is it?

A UCI chess engine.
And a butt-kicker. It might be the next Ruffian.
No opening book, no tablebase.
Here is what I have so far from testing:

Sounds interesting. Any more details? Who is the author? Is this their first chess program? etc
Steve

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 09 March 2004 20:59:29:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Steve Maughan at 09 March 2004 20:51:17:

Dann,

What is it?

A UCI chess engine.
And a butt-kicker. It might be the next Ruffian.
No opening book, no tablebase.
Here is what I have so far from testing:

Sounds interesting. Any more details? Who is the author? Is this their first chess program? etc

The author is Fabien Letouzey.
I do not know how much the author wants to release about the engine. He has gone home for the day.
Once people start to try it, there will be a windstorm of questions. This one is definitely going to make a stir.
I think he is from Europe somewhere, since his time zone is very different from mine.
Aaron Gordon has worked with him before.



my ftp site {remove http:// unless you like error messages}

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 09 March 2004 21:11:25:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 09 March 2004 20:59:29:

The author is Fabien Letouzey.

I have exchanged a few e-mails with Fabien the last couple of days. He
appears to be a very nice and knowledgeable person, and your initial results
for his engine makes it clear that he is the most modest chess programmer
I have ever had the pleasure of communicating with. Fabien claims that his
engine is still "pretty weak", and that it has "almost no knowledge".
He is also confident that it will improve very quickly, which means that we
should all be really afraid.
Tord

[Peter Fendrich]

Geschrieben von: / Posted by: Peter Fendrich at 09 March 2004 21:18:10:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 09 March 2004 20:48:15:

What made me ask was your url including "new-approach".
Either it was a chess program using new techniques or some other type of program.
Is this a new kind of approach for chess programs?
/Peter

ftp://cap.connx.com/chess-engines/new-approach/fruit_src.exe

What is it?

A UCI chess engine.
And a butt-kicker. It might be the next Ruffian.
No opening book, no tablebase.
Here is what I have so far from testing:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3253 0 0 1 100.0 % 2653 0.0 %
2 Arasanx-74a : 2653 0 0 1 50.0 % 2653 100.0 %
3 Aristarch 4.41: 2653 0 0 1 50.0 % 2653 100.0 %
4 Fruit_src : 2604 344 268 7 71.4 % 2445 28.6 %
5 31337-75 : 2053 0 0 1 0.0 % 2653 0.0 %
6 Abrok_5_0 : 2053 0 0 1 0.0 % 2653 0.0 %
7 Amateur270 : 2053 0 0 1 0.0 % 2653 0.0 %
8 Averno070 : 2053 0 0 1 0.0 % 2653 0.0 %

Just so everyone can breath a sigh of relief, the source code looks nothing like crafty or gnuchess or tscp or amy, etc.

[Roger Brown]

Geschrieben von: / Posted by: Roger Brown at 09 March 2004 21:49:13:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Tord Romstad at 09 March 2004 21:11:25:

he is the most modest chess programmer I have ever had the pleasure of communicating with.

You don't talk much with yourself, do you Tord?


Incidentally, you could look up a post entitled Tord Romstad..exposed for further details as to the history behind that comment.
You have been busted.


Later.

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 09 March 2004 21:53:37:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Tord Romstad at 09 March 2004 21:11:25:

The author is Fabien Letouzey.

I have exchanged a few e-mails with Fabien the last couple of days. He
appears to be a very nice and knowledgeable person, and your initial results
for his engine makes it clear that he is the most modest chess programmer
I have ever had the pleasure of communicating with. Fabien claims that his
engine is still "pretty weak", and that it has "almost no knowledge".
He is also confident that it will improve very quickly, which means that we
should all be really afraid.
Tord

The most modest chess programmer is not going to be confident that his engine is going to improve very quickly.
The opposite.
He is going to say that he does not expect his weak engine to improve and he is trying to improve it for fun but he will almost certainly fail because of bugs or bad ideas.
Uri

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 09 March 2004 22:28:51:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Peter Fendrich at 09 March 2004 21:18:10:

What made me ask was your url including "new-approach".
Either it was a chess program using new techniques or some other type of program.
Is this a new kind of approach for chess programs?

New approach is my new approach. Instead of putting everything in its own separate directory, I put everything in a single directory.



my ftp site {remove http:// unless you like error messages}

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 09 March 2004 22:32:54:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Uri Blass at 09 March 2004 21:53:37:

The author is Fabien Letouzey.

I have exchanged a few e-mails with Fabien the last couple of days. He
appears to be a very nice and knowledgeable person, and your initial results
for his engine makes it clear that he is the most modest chess programmer
I have ever had the pleasure of communicating with. Fabien claims that his
engine is still "pretty weak", and that it has "almost no knowledge".
He is also confident that it will improve very quickly, which means that we
should all be really afraid.
Tord

The most modest chess programmer is not going to be confident that his engine is going to improve very quickly.
The opposite.
He is going to say that he does not expect his weak engine to improve and he is trying to improve it for fun but he will almost certainly fail because of bugs or bad ideas.

Modesty is almost never a negative emotion. On the other hand, overconfidence ia almost always a negative.
With no opening book, this is the result so far:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3224 0 0 1 100.0 % 2624 0.0 %
2 Crafty-1910 : 3224 0 0 1 100.0 % 2624 0.0 %
3 DeepSjeng : 3224 0 0 1 100.0 % 2624 0.0 %
4 Delfi-440 : 3224 0 0 1 100.0 % 2624 0.0 %
5 Dragon_45 : 3224 0 0 1 100.0 % 2624 0.0 %
6 ELChinito 3.25 : 3224 0 0 1 100.0 % 2624 0.0 %
7 Glc300 : 3224 0 0 1 100.0 % 2624 0.0 %
8 Arasanx-74a : 2624 0 0 1 50.0 % 2624 100.0 %
9 Aristarch 4.41 : 2624 0 0 1 50.0 % 2624 100.0 %
10 GnuChess-507-Net : 2624 0 0 1 50.0 % 2624 100.0 %
11 Fruit_src : 2583 120 164 26 67.3 % 2458 11.5 %
12 31337-75 : 2024 0 0 1 0.0 % 2624 0.0 %
13 Abrok_5_0 : 2024 0 0 1 0.0 % 2624 0.0 %
14 Amateur270 : 2024 0 0 1 0.0 % 2624 0.0 %
15 Averno070 : 2024 0 0 1 0.0 % 2624 0.0 %
16 Butcher-142c : 2024 0 0 1 0.0 % 2624 0.0 %
17 Chessterfield-15a: 2024 0 0 1 0.0 % 2624 0.0 %
18 DelphiMax28 : 2024 0 0 1 0.0 % 2624 0.0 %
19 Djinn848 : 2024 0 0 1 0.0 % 2624 0.0 %
20 ExChess403i60 : 2024 0 0 1 0.0 % 2624 0.0 %
21 Geko_043 : 2024 0 0 1 0.0 % 2624 0.0 %
22 Gerbil-02 : 2024 0 0 1 0.0 % 2624 0.0 %
23 Ghost-13 : 2024 0 0 1 0.0 % 2624 0.0 %
24 Gnuchess4TM : 2024 0 0 1 0.0 % 2624 0.0 %
25 Golem01 : 2024 0 0 1 0.0 % 2624 0.0 %
26 Gully2 : 2024 0 0 1 0.0 % 2624 0.0 %
27 Hermann : 2024 0 0 1 0.0 % 2624 0.0 %

A nice, clean compile. It took me just a few minutes to get it working on UCI on Windows, and he only compiles on UNIX. That's pretty remarkable.
He has nice looking code, that is easy to read. Those programs usually improve very quickly.
I expect instant stardom.



my ftp site {remove http:// unless you like error messages}

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 09 March 2004 23:53:13:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 09 March 2004 22:32:54:

The author is Fabien Letouzey.

I have exchanged a few e-mails with Fabien the last couple of days. He
appears to be a very nice and knowledgeable person, and your initial results
for his engine makes it clear that he is the most modest chess programmer
I have ever had the pleasure of communicating with. Fabien claims that his
engine is still "pretty weak", and that it has "almost no knowledge".
He is also confident that it will improve very quickly, which means that we
should all be really afraid.
Tord

The most modest chess programmer is not going to be confident that his engine is going to improve very quickly.
The opposite.
He is going to say that he does not expect his weak engine to improve and he is trying to improve it for fun but he will almost certainly fail because of bugs or bad ideas.

Modesty is almost never a negative emotion. On the other hand, overconfidence ia almost always a negative.
With no opening book, this is the result so far:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3224 0 0 1 100.0 % 2624 0.0 %
2 Crafty-1910 : 3224 0 0 1 100.0 % 2624 0.0 %
3 DeepSjeng : 3224 0 0 1 100.0 % 2624 0.0 %
4 Delfi-440 : 3224 0 0 1 100.0 % 2624 0.0 %
5 Dragon_45 : 3224 0 0 1 100.0 % 2624 0.0 %
6 ELChinito 3.25 : 3224 0 0 1 100.0 % 2624 0.0 %
7 Glc300 : 3224 0 0 1 100.0 % 2624 0.0 %
8 Arasanx-74a : 2624 0 0 1 50.0 % 2624 100.0 %
9 Aristarch 4.41 : 2624 0 0 1 50.0 % 2624 100.0 %
10 GnuChess-507-Net : 2624 0 0 1 50.0 % 2624 100.0 %
11 Fruit_src : 2583 120 164 26 67.3 % 2458 11.5 %
12 31337-75 : 2024 0 0 1 0.0 % 2624 0.0 %
13 Abrok_5_0 : 2024 0 0 1 0.0 % 2624 0.0 %
14 Amateur270 : 2024 0 0 1 0.0 % 2624 0.0 %
15 Averno070 : 2024 0 0 1 0.0 % 2624 0.0 %
16 Butcher-142c : 2024 0 0 1 0.0 % 2624 0.0 %
17 Chessterfield-15a: 2024 0 0 1 0.0 % 2624 0.0 %
18 DelphiMax28 : 2024 0 0 1 0.0 % 2624 0.0 %
19 Djinn848 : 2024 0 0 1 0.0 % 2624 0.0 %
20 ExChess403i60 : 2024 0 0 1 0.0 % 2624 0.0 %
21 Geko_043 : 2024 0 0 1 0.0 % 2624 0.0 %
22 Gerbil-02 : 2024 0 0 1 0.0 % 2624 0.0 %
23 Ghost-13 : 2024 0 0 1 0.0 % 2624 0.0 %
24 Gnuchess4TM : 2024 0 0 1 0.0 % 2624 0.0 %
25 Golem01 : 2024 0 0 1 0.0 % 2624 0.0 %
26 Gully2 : 2024 0 0 1 0.0 % 2624 0.0 %
27 Hermann : 2024 0 0 1 0.0 % 2624 0.0 %

It seems that you test too a lot against weak programs.
Movei with no book probably also can do similiar results.
It may be interesting if you test movei with no book against the same programs
and compare results(note that movei use book for the first move even without external book but I do not think that it changes much).
It is possible that there is going to be a significant improvement but
I do not think that book is going to be a reason for a big improvement.
I guess that good book can maybe give you 50-100 elo in this kind of test relative to no book(I guess also that movei has not a good book)
Uri

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 10 March 2004 00:10:52:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Uri Blass at 09 March 2004 23:53:13:

[snip]

It seems that you test too a lot against weak programs.
Movei with no book probably also can do similiar results.
It may be interesting if you test movei with no book against the same programs
and compare results(note that movei use book for the first move even without external book but I do not think that it changes much).
It is possible that there is going to be a significant improvement but
I do not think that book is going to be a reason for a big improvement.
I guess that good book can maybe give you 50-100 elo in this kind of test relative to no book(I guess also that movei has not a good book)
Uri

I chose a broad spectrum. I think the win against List 5.12 is impressive:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3252 0 0 1 100.0 % 2652 0.0 %
2 Crafty-1910 : 3252 0 0 1 100.0 % 2652 0.0 %
3 DeepSjeng : 3252 0 0 1 100.0 % 2652 0.0 %
4 Delfi-440 : 3252 0 0 1 100.0 % 2652 0.0 %
5 Dragon_45 : 3252 0 0 1 100.0 % 2652 0.0 %
6 ELChinito 3.25 : 3252 0 0 1 100.0 % 2652 0.0 %
7 Glc300 : 3252 0 0 1 100.0 % 2652 0.0 %
8 Kke-253 : 3252 0 0 1 100.0 % 2652 0.0 %
9 Ktulu : 3252 0 0 1 100.0 % 2652 0.0 %
10 Arasanx-74a : 2652 0 0 1 50.0 % 2652 100.0 %
11 Aristarch 4.41 : 2652 0 0 1 50.0 % 2652 100.0 %
12 GnuChess-507-Net : 2652 0 0 1 50.0 % 2652 100.0 %
13 Holmes-0820m : 2652 0 0 1 50.0 % 2652 100.0 %
14 Horizon : 2652 0 0 1 50.0 % 2652 100.0 %
15 Fruit_src : 2603 88 140 40 71.2 % 2446 12.5 %
16 31337-75 : 2052 0 0 1 0.0 % 2652 0.0 %
17 Abrok_5_0 : 2052 0 0 1 0.0 % 2652 0.0 %
18 Amateur270 : 2052 0 0 1 0.0 % 2652 0.0 %
19 Averno070 : 2052 0 0 1 0.0 % 2652 0.0 %
20 Butcher-142c : 2052 0 0 1 0.0 % 2652 0.0 %
21 Chessterfield-15a: 2052 0 0 1 0.0 % 2652 0.0 %
22 DelphiMax28 : 2052 0 0 1 0.0 % 2652 0.0 %
23 Djinn848 : 2052 0 0 1 0.0 % 2652 0.0 %
24 ExChess403i60 : 2052 0 0 1 0.0 % 2652 0.0 %
25 Geko_043 : 2052 0 0 1 0.0 % 2652 0.0 %
26 Gerbil-02 : 2052 0 0 1 0.0 % 2652 0.0 %
27 Ghost-13 : 2052 0 0 1 0.0 % 2652 0.0 %
28 Gnuchess4TM : 2052 0 0 1 0.0 % 2652 0.0 %
29 Golem01 : 2052 0 0 1 0.0 % 2652 0.0 %
30 Gully2 : 2052 0 0 1 0.0 % 2652 0.0 %
31 Hermann : 2052 0 0 1 0.0 % 2652 0.0 %
32 Hoplite-20-net : 2052 0 0 1 0.0 % 2652 0.0 %
33 Jester : 2052 0 0 1 0.0 % 2652 0.0 %
34 Kanguruh62 : 2052 0 0 1 0.0 % 2652 0.0 %
35 Knightx182 : 2052 0 0 1 0.0 % 2652 0.0 %
36 Leila_053h : 2052 0 0 1 0.0 % 2652 0.0 %
37 List512 : 2052 0 0 1 0.0 % 2652 0.0 %
38 Monarch : 2052 0 0 1 0.0 % 2652 0.0 %
39 Monsoon : 2052 0 0 1 0.0 % 2652 0.0 %
40 NagaSkaki : 2052 0 0 1 0.0 % 2652 0.0 %
41 Needle : 2052 0 0 1 0.0 % 2652 0.0 %

I think 50 to 100 Elo is probably a pretty good guess.
This is the initial release of his program (in fact still in beta stage). Compare this result with the first release of your program. Are you impressed yet?



my ftp site {remove http:// unless you like error messages}

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 10 March 2004 09:38:23:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 10 March 2004 00:10:52:

[snip]

It seems that you test too a lot against weak programs.
Movei with no book probably also can do similiar results.
It may be interesting if you test movei with no book against the same programs
and compare results(note that movei use book for the first move even without external book but I do not think that it changes much).
It is possible that there is going to be a significant improvement but
I do not think that book is going to be a reason for a big improvement.
I guess that good book can maybe give you 50-100 elo in this kind of test relative to no book(I guess also that movei has not a good book)
Uri

I chose a broad spectrum. I think the win against List 5.12 is impressive:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3252 0 0 1 100.0 % 2652 0.0 %
2 Crafty-1910 : 3252 0 0 1 100.0 % 2652 0.0 %
3 DeepSjeng : 3252 0 0 1 100.0 % 2652 0.0 %
4 Delfi-440 : 3252 0 0 1 100.0 % 2652 0.0 %
5 Dragon_45 : 3252 0 0 1 100.0 % 2652 0.0 %
6 ELChinito 3.25 : 3252 0 0 1 100.0 % 2652 0.0 %
7 Glc300 : 3252 0 0 1 100.0 % 2652 0.0 %
8 Kke-253 : 3252 0 0 1 100.0 % 2652 0.0 %
9 Ktulu : 3252 0 0 1 100.0 % 2652 0.0 %
10 Arasanx-74a : 2652 0 0 1 50.0 % 2652 100.0 %
11 Aristarch 4.41 : 2652 0 0 1 50.0 % 2652 100.0 %
12 GnuChess-507-Net : 2652 0 0 1 50.0 % 2652 100.0 %
13 Holmes-0820m : 2652 0 0 1 50.0 % 2652 100.0 %
14 Horizon : 2652 0 0 1 50.0 % 2652 100.0 %
15 Fruit_src : 2603 88 140 40 71.2 % 2446 12.5 %
16 31337-75 : 2052 0 0 1 0.0 % 2652 0.0 %
17 Abrok_5_0 : 2052 0 0 1 0.0 % 2652 0.0 %
18 Amateur270 : 2052 0 0 1 0.0 % 2652 0.0 %
19 Averno070 : 2052 0 0 1 0.0 % 2652 0.0 %
20 Butcher-142c : 2052 0 0 1 0.0 % 2652 0.0 %
21 Chessterfield-15a: 2052 0 0 1 0.0 % 2652 0.0 %
22 DelphiMax28 : 2052 0 0 1 0.0 % 2652 0.0 %
23 Djinn848 : 2052 0 0 1 0.0 % 2652 0.0 %
24 ExChess403i60 : 2052 0 0 1 0.0 % 2652 0.0 %
25 Geko_043 : 2052 0 0 1 0.0 % 2652 0.0 %
26 Gerbil-02 : 2052 0 0 1 0.0 % 2652 0.0 %
27 Ghost-13 : 2052 0 0 1 0.0 % 2652 0.0 %
28 Gnuchess4TM : 2052 0 0 1 0.0 % 2652 0.0 %
29 Golem01 : 2052 0 0 1 0.0 % 2652 0.0 %
30 Gully2 : 2052 0 0 1 0.0 % 2652 0.0 %
31 Hermann : 2052 0 0 1 0.0 % 2652 0.0 %
32 Hoplite-20-net : 2052 0 0 1 0.0 % 2652 0.0 %
33 Jester : 2052 0 0 1 0.0 % 2652 0.0 %
34 Kanguruh62 : 2052 0 0 1 0.0 % 2652 0.0 %
35 Knightx182 : 2052 0 0 1 0.0 % 2652 0.0 %
36 Leila_053h : 2052 0 0 1 0.0 % 2652 0.0 %
37 List512 : 2052 0 0 1 0.0 % 2652 0.0 %
38 Monarch : 2052 0 0 1 0.0 % 2652 0.0 %
39 Monsoon : 2052 0 0 1 0.0 % 2652 0.0 %
40 NagaSkaki : 2052 0 0 1 0.0 % 2652 0.0 %
41 Needle : 2052 0 0 1 0.0 % 2652 0.0 %

I think 50 to 100 Elo is probably a pretty good guess.
This is the initial release of his program (in fact still in beta stage). Compare this result with the first release of your program. Are you impressed yet?

I could also wait until my engine becomes strong before releasing it.
I also do not consider myself to be a good programmer and if I implemented 1/10 of what I thought to implement my program could be certainly better.
Uri

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 10 March 2004 09:47:34:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Uri Blass at 10 March 2004 09:38:23:

[snip]

It seems that you test too a lot against weak programs.
Movei with no book probably also can do similiar results.
It may be interesting if you test movei with no book against the same programs
and compare results(note that movei use book for the first move even without external book but I do not think that it changes much).
It is possible that there is going to be a significant improvement but
I do not think that book is going to be a reason for a big improvement.
I guess that good book can maybe give you 50-100 elo in this kind of test relative to no book(I guess also that movei has not a good book)
Uri

I chose a broad spectrum. I think the win against List 5.12 is impressive:

Program Elo + - Games Score Av.Op. Draws
1 Amy-net-087 : 3252 0 0 1 100.0 % 2652 0.0 %
2 Crafty-1910 : 3252 0 0 1 100.0 % 2652 0.0 %
3 DeepSjeng : 3252 0 0 1 100.0 % 2652 0.0 %
4 Delfi-440 : 3252 0 0 1 100.0 % 2652 0.0 %
5 Dragon_45 : 3252 0 0 1 100.0 % 2652 0.0 %
6 ELChinito 3.25 : 3252 0 0 1 100.0 % 2652 0.0 %
7 Glc300 : 3252 0 0 1 100.0 % 2652 0.0 %
8 Kke-253 : 3252 0 0 1 100.0 % 2652 0.0 %
9 Ktulu : 3252 0 0 1 100.0 % 2652 0.0 %
10 Arasanx-74a : 2652 0 0 1 50.0 % 2652 100.0 %
11 Aristarch 4.41 : 2652 0 0 1 50.0 % 2652 100.0 %
12 GnuChess-507-Net : 2652 0 0 1 50.0 % 2652 100.0 %
13 Holmes-0820m : 2652 0 0 1 50.0 % 2652 100.0 %
14 Horizon : 2652 0 0 1 50.0 % 2652 100.0 %
15 Fruit_src : 2603 88 140 40 71.2 % 2446 12.5 %
16 31337-75 : 2052 0 0 1 0.0 % 2652 0.0 %
17 Abrok_5_0 : 2052 0 0 1 0.0 % 2652 0.0 %
18 Amateur270 : 2052 0 0 1 0.0 % 2652 0.0 %
19 Averno070 : 2052 0 0 1 0.0 % 2652 0.0 %
20 Butcher-142c : 2052 0 0 1 0.0 % 2652 0.0 %
21 Chessterfield-15a: 2052 0 0 1 0.0 % 2652 0.0 %
22 DelphiMax28 : 2052 0 0 1 0.0 % 2652 0.0 %
23 Djinn848 : 2052 0 0 1 0.0 % 2652 0.0 %
24 ExChess403i60 : 2052 0 0 1 0.0 % 2652 0.0 %
25 Geko_043 : 2052 0 0 1 0.0 % 2652 0.0 %
26 Gerbil-02 : 2052 0 0 1 0.0 % 2652 0.0 %
27 Ghost-13 : 2052 0 0 1 0.0 % 2652 0.0 %
28 Gnuchess4TM : 2052 0 0 1 0.0 % 2652 0.0 %
29 Golem01 : 2052 0 0 1 0.0 % 2652 0.0 %
30 Gully2 : 2052 0 0 1 0.0 % 2652 0.0 %
31 Hermann : 2052 0 0 1 0.0 % 2652 0.0 %
32 Hoplite-20-net : 2052 0 0 1 0.0 % 2652 0.0 %
33 Jester : 2052 0 0 1 0.0 % 2652 0.0 %
34 Kanguruh62 : 2052 0 0 1 0.0 % 2652 0.0 %
35 Knightx182 : 2052 0 0 1 0.0 % 2652 0.0 %
36 Leila_053h : 2052 0 0 1 0.0 % 2652 0.0 %
37 List512 : 2052 0 0 1 0.0 % 2652 0.0 %
38 Monarch : 2052 0 0 1 0.0 % 2652 0.0 %
39 Monsoon : 2052 0 0 1 0.0 % 2652 0.0 %
40 NagaSkaki : 2052 0 0 1 0.0 % 2652 0.0 %
41 Needle : 2052 0 0 1 0.0 % 2652 0.0 %

I think 50 to 100 Elo is probably a pretty good guess.
This is the initial release of his program (in fact still in beta stage). Compare this result with the first release of your program. Are you impressed yet?

I could also wait until my engine becomes strong before releasing it.
I also do not consider myself to be a good programmer and if I implemented 1/10 of what I thought to implement my program could be certainly better.
Uri

To be more correct we cannot talk about % but I believe that with good use of hash tables and internal iterative deepening and some non standard ideas that I thought about some years ago but did not implement, my program can be significantly better and I am talking about 200 elo improvement only thanks to search improvements(even without changing the static evaluation .
A good programmer instead of me who is not going to be tempted to look too much in games could probably get it in a few weeks.
Uri

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 10 March 2004 09:52:11:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Uri Blass at 10 March 2004 09:47:34:

[snip]

I could also wait until my engine becomes strong before releasing it.
I also do not consider myself to be a good programmer and if I implemented 1/10 of what I thought to implement my program could be certainly better.
Uri

To be more correct we cannot talk about % but I believe that with good use of hash tables and internal iterative deepening and some non standard ideas that I thought about some years ago but did not implement, my program can be significantly better and I am talking about 200 elo improvement only thanks to search improvements(even without changing the static evaluation .
A good programmer instead of me who is not going to be tempted to look too much in games could probably get it in a few weeks.

Probably, some of that went on here.
Your program is already very, very strong.
Once you have implemented all of your ideas, your program will be as strong as commercial programs.



my ftp site {remove http:// unless you like error messages}

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 March 2004 10:54:03:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Uri Blass at 09 March 2004 21:53:37:

The most modest chess programmer is not going to be confident that his engine is going to improve very quickly.
The opposite.

The phrase about "improving very quickly" was taken out of context, and I can
understand that it looks less than modest when seen alone. It would probably
have been better if I had left it out of my post, or if I had refrained from
posting at all. My apologies to Fabien.
Tord

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 March 2004 11:04:54:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Roger Brown at 09 March 2004 21:49:13:

he is the most modest chess programmer I have ever had the pleasure of communicating with.

You don't talk much with yourself, do you Tord?


Incidentally, you could look up a post entitled Tord Romstad..exposed for further details as to the history behind that comment.
You have been busted.

Ah, I missed that one. Apparently you posted during my holiday, and I didn't read
all the old posts when I came back. Thanks for bringing it to my attention. Have
you, by any chance, kept the games? I would be happy to receive them.
Not at all.
Seriously, I think I have rather accurate and realistic views about my engine,
its strong and weak sides, and its potential. It is neither among the strongest
nor the weakest of the free engines, but its unbalanced, aggressive and often
suicidal style makes it capable of winning or losing against any computer
opponent, regardless of strength.
Tord

[Fabien Letouzey]

Geschrieben von: / Posted by: Fabien Letouzey at 10 March 2004 12:00:20:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Tord Romstad at 10 March 2004 10:54:03:

The most modest chess programmer is not going to be confident that his engine is going to improve very quickly.
The opposite.

The phrase about "improving very quickly" was taken out of context, and I can
understand that it looks less than modest when seen alone. It would probably
have been better if I had left it out of my post, or if I had refrained from
posting at all. My apologies to Fabien.
Tord

No worries Tord,
Yes I am confident I can improve Fruit easily for the next version (but this one is not even release-ready yet, still working on UCI support).
But maybe that's only going to get it 20 elo points, I don't know!
I base that on knowledge I know is missing and that I see affecting games.
Also after that, I will need enormous computing time to improve the program (e.g. search parameter tuning). I can't do that on my own, I need the help of the internet community.
Fabien.

[Fabien Letouzey]

Geschrieben von: / Posted by: Fabien Letouzey at 10 March 2004 12:08:02:
Als Antwort auf: / In reply to: Re: [OT-UCI] Fruit 0.98a beta testers geschrieben von: / posted by: Dann Corbit at 09 March 2004 20:59:29:

The author is Fabien Letouzey.
I do not know how much the author wants to release about the engine. He has gone home for the day.
Once people start to try it, there will be a windstorm of questions. This one is definitely going to make a stir.
I think he is from Europe somewhere, since his time zone is very different from mine.
Aaron Gordon has worked with him before.

I am french, and live in England.
I'd like to have "European" on my ID card!
Aaron was my hardware partner about two years ago.
He is a hardware/overclocking specialist, and he knows a lot about compilers as well, the perfect partner for me!
I had pitiful hardware at the time, and tried to make my engine (Chessy II) play on FICS.
It got crushed tactically on almost every game.
I thought the engine must have been really bad.
Then I "met" Aaron, and he created an account for me on one of his "slow" machines.
I got much better results instantly, like the engine got past some sort of "tactical barrier".
I am indebt to Aaron and I hope we can work together on Fruit, but he seems inactive at the moment.
Although I am completely unknown in the computer-chess community, I have written many Othello programs. "Turtle" was strong in its time. More recently I wrote a freeware program for Palm called "PilOth". If you enjoy losing all games like 60 to 4, this is the program for you!
Fabien.