Computer Ponanza gets highest rating on Shogi Club 24

[Mark Schreiber]

Shogi Club 24 rated computer program Ponanza 3211 on May 14. I believe 3211 is the highest Shogi Club 24 rating. Its current rating is 3110 putting Ponanza in 7th place on Shogi Club 24. 3110 is at 102 games. Has the computer reached high professional level?

[Bernhard]

Hi,

still I know, that 7 games are to less to confirm anything, but
Ponanza became "only" 5th rank on last WCSC.

Is the newest version of Ponanza running on SC24?
And are the other engines (Bonkras - 1st, Bonanza V6 - 2nd, GPS Shogi
- 6th) also running on SC24?

Thanks,

Bernhard

[Bernhard]

And I forgot Blunder - 7th rank on WCSC 2011.
Also a free engine.

[Gergely Buglyo]

I think the reason why shogi programs are doing so well on SC24 is because they rate them in quick games (1 minute + 30 seconds byo-yomi). In such a game, they might indeed be a match for even a high-ranking professional, but computers are still thought to be much weaker in games with longer time limits. Best regards:
 
Gergely
 

---

Date: Fri, 20 May 2011 01:23:31 -0700
From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Computer Ponanza gets highest rating on Shogi Club 24

Shogi Club 24 rated computer program Ponanza 3211 on May 14. I believe 3211 is the highest Shogi Club 24 rating. Its current rating is 3110 putting Ponanza in 7th place on Shogi Club 24. 3110 is at 102 games. Has the computer reached high professional level?

--
You received this message because you are subscribed to the Google Groups "SHOGI-L" group.
To post to this group, send email to sho...@googlegroups.com.
To unsubscribe from this group, send email to shogi-l+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/shogi-l?hl=en.

[Mark Schreiber]

Can we look at the game records and see if they are quick or slower games? Or ask the players?

[Gergely Buglyo]

Sure, you can just open the kifu browser on the SC24 website, and search for "Ponanza". If you look at the time each match lasted, it's easy to see that most of them were short games.
 

---

Date: Fri, 20 May 2011 08:09:10 -0700
From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Re: RE: Computer Ponanza gets highest rating on Shogi Club 24

Can we look at the game records and see if they are quick or slower games? Or ask the players?

[lkau...@comcast.net]

Does anyone know whether "Ponanza" is a new version of Bonanza or an unrelated program usurping the name with a one-letter change?

 

Larry Kaufman

[Gergely Buglyo]

Here's a PDF on Ponanza: http://www.computer-shogi.org/wcsc20/appeal/Ponanza/Ponanza.pdf
 
According to it, the program uses the Bonanza method, but is much weaker (at least it was at the time), that's why it was named like that, as a parody of Bonanza. Apart from the Bonanza method (which I think is some kind of technical term in programming), it doesn't use anything from the original Bonanza program. The maker of Bonanza seems to have permitted the makers to name their program "Ponanza".
 

---

Date: Sat, 21 May 2011 01:15:27 +0000
From: lkau...@comcast.net
To: sho...@googlegroups.com
Subject: Re: Computer Ponanza gets highest rating on Shogi Club 24

[h.g. muller]

> Apart from the Bonanza method (which I think is some kind of technical
> term in programming), it doesn't use anything from the original Bonanza
> program.

The 'Bonanza method' is a method for learning evaluation parameters based
on results of a minimax search as used in chess programs. AFAIK the method
was invented by the author of the Bonanza Shogi engine (Kunihito Hoki)
especially for the purpose, and derives its name from the engine, rather
than the other way around.

The reference is:
[1] K. Hoki, "Optimal control of minimax search results to learn positional
evaluation", Game Programing Workshop, Hakone, Japan 2008.

[Mark Schreiber]

Another way to know Ponanza is not a new version of Bonanza. If Ponanza were a new version of Bonanza then both Bonanza and Ponanza could not have played in the World Computer Shogi Championship. The rules of the World Computer Shogi Championship say only one program from each developer. In the World Computer Shogi Championship Ponanza scored 5th place. Bonanza scored 2nd place. On the floodgate server, Bonanza is 139 points higher than Ponanza. Ponanza is now 6th place on Shogi club 24.

[Mark Schreiber]

I see Ponanza played white and won against ginkanmuri7 on May 16, 2011. The game stared at 9:26:46 AM. The ended at 10:28:03 AM. The game was 1 hour, 1 minute and 17 seconds. The game had 100 moves. Ginkanmuri7 had a rating 2845. If the game was 1 minute + 30 seconds byo-yomi then the maximum time would be less than 52 minutes. The game had to be longer than 1 minute + 30 seconds byo-yomi.

[Gergely Buglyo]

I agree, it had to be. Also, it was a very close match. Still, the majority of games played by Ponanza seem to have been quick games.
 

---

Date: Sun, 22 May 2011 23:41:21 -0700

From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Re: RE: Computer Ponanza gets highest rating on Shogi Club 24

I see Ponanza played white and won against ginkanmuri7 on May 16, 2011. The game stared at 9:26:46 AM. The ended at 10:28:03 AM. The game was 1 hour, 1 minute and 17 seconds. The game had 100 moves. Ginkanmuri7 had a rating 2845. If the game was 1 minute + 30 seconds byo-yomi then the maximum time would be less than 52 minutes. The game had to be longer than 1 minute + 30 seconds byo-yomi.

[Mark Schreiber]

I do not know the version of ponanza. I do not see GPS Shogi or Blunder on SC24. I see there are currently 5 higher rated players than ponanza on SC24. Not one of them is a computer. I see Bonanza, YSS and Otsuki Shogi played on SC24 before 2009. Therefore, I do not see their games. Otsuki was rated 2632. Bonanza was rated 2640. YSS was rated 2744 in 2007. I do not see other strong computers that played on SC24.

[BCM]

Hi,

On 22 Mai, 13:26, Mark Schreiber <mschr...@gmail.com> wrote:
> Another way to know Ponanza is not a new version of Bonanza. If Ponanza were
> a new version of Bonanza then both Bonanza and Ponanza could not have played
> in the World Computer Shogi Championship. The rules of the World Computer
> Shogi Championship say only one program from each developer. In the World

but for example in 19th WCSC 2009, there played Bonanza and Monju,
where later is nothing else than six Bonanza running on 3 machines.
Or maybe this is a new rule, because I can't find Monju on 21th WCSC
2011.

By the way: what about bonkras?
Has this one something to do with Bonanza?

Is bonkras free software?
On the given link (http://aleag.cocolog-nifty.com/) you can download
bonasse. And what's this?
Is bonasse the same than bonkras?
Downloading bonasse, it seems to be bonanza, but what version is this
V1.2.2?

Thanks for all answers to this many questions,
Bernhard

[Mark Schreiber]

The rule of only one program from each developer existed at the 19th WCSC. I do not know why Monju was allowed, maybe because Monju is from another developer. Maybe the developer of Monju added voting to Monju so there are enough changes to make it different. Your questions are good. I tried to read the program’s description with Google translate. That did not work.

Bonkras is not Bonanza. The programmer of bonkras is Hideki Itou. Hideki Itou is the programmer of Deep Purple which he entered in 1999. Shogi program A-Class League Move #1 is the successor to Deep Purple. Bonkras is the successor to Shogi program A-Class League Move #1. 

One link for bonkras goes to a cartoon at http://www.computer-shogi.org/wcsc21/appeal/bonkras/WCSC2011appeal_bonkras.pdf . The cartoon cannot be translated with Google because it is a picture. The cartoon has a blue haired girl and red haired girl talking. Can someone explain this cartoon?

Thank you. Mark.

[BCM]

Thanks for your answer.

About bonasse I've found the following, translated with google-
translation.
Hard to understand, but the main thing should be clear: it's bonanza,
which is optimized for speed

>>
bonasse (Bonanza faster version) v1.2.2 Release Notes
Itou Hideki 5/30/2010

- Revision History
5/30/2010 v1.2.2 conditions of the document and change the
description of gcc (the source no-change)
5/25/2010 v1.2.1 environmental enhancement compatible with various
versions of the show change at chess
5/22/2010 v1.2 Feliz support measures Inaniwa faster
5/15/2010 v1.1 Windows compatible, affinity command added
Initial 5/11/2010 v1.0


• About this package
Bonanza is the speed. Parallel cluster (Bonkurazu) is a different
thing.
Bonanza Feliz 0.0 is based on.
Bonanza until ※ v1.1 v4.1.3 (some 4.1.2) was based on, v1.2 with
Changed
...
<<

[Mark Schreiber]

Thanks for the information.

So if I understand correctly, bonasse is a faster version of Bonanza. CSA winner Bonkras (Bonkurazu) is not related to bonasse or to Bonanza. Bonasse was not entered in the CSA tournament. But Bonasse plays on the floodgate server.

Mark

[Jun Koda]

Hi,

This is what find about bonkras by browsing around.

Yes, bonasse is a Bonanza optimized by Hideki Itou, who is also the
developer of Bonkras. Bonasse is a serial program, not usable for
computer clusters.

Bonkras is a parallelized version, originally used Bonanza or bonasse
as a search routine [therefore, bonkras is said to be an abbreviation
of "Bonanza Clusters" (clu ~ kra, same in Japanese)], but that bonasse
part was completely rewritten this year to avoid the regulation of
"one program from each developer." Itou writes that he closely
imitated bonasse (*1), so I think the serial search part of bonkras is
pretty similar to bonasse or Bonanza.

The comic for bonkras is basically saying that their parallelization
scheme is different from others. Most program based on "young brothers
wait concept," but bonkras is not (I don't know that technical part,
and its not explained in the comic). According to another entry by the
programmer (*2), bonkras is using MPI (message passing interface)
parallelization; one master process is assigning other slave processes
to search & evaluate different positions.

(*1) http://aleag.cocolog-nifty.com/blog/2011/05/post-39b1.html
(*2) http://aleag.cocolog-nifty.com/blog/2010/01/post-6445.html

Jun

[Mark Schreiber]

Jun, Thanks for the explanation.  Mark

[Yoshiki Hayashi]

Hi,

Ponanza in WCSC 20 was an original program but the one in WCSC 21 was
based on Bonanza 4.1.3. WCSC 21 allowed maximum of two programs not
including originals using the same library to advance to the second
round and only one program to the final round. Bonanza 4.1.3 is one
of the registered libraries that can be used in WCSC. From what I've
heard, Ponanza optimized search speed using magic bitboard and it hand
modified some values of tables used in evaluation function to improve
entering king game. That's pretty much all I know but we'll have
better picture of what Ponanza had improved when Computer Shogi
Association releases documents from authors describing features of
their programs.

The Bonanza program participated in WCSC 21 was released as Bonanza
6.0 and can be downloaded from http://www.geocities.jp/bonanza_shogi/
(follow the link that says 40MByte). The improvement from Bonanza
4.1.3 is described in src/client/readme.txt in English if anyone is
interested. According to the author of Ponanza, Ponanza faired pretty
well against Bonanza 4.1.3. As far as I know, whether Ponanza is
better than Bonanza 6.0 or not is unknown.

--
Yoshiki

[Mark Schreiber]

Thank you, Yoshiki.

What are registered libraries that are used in WCSC?

Did the CSA release documents from the authors describing features of the 2010 programs?

Mark

[Yoshiki Hayashi]

The registered library is listed at the bottom of http://www.computer-shogi.org/library/
Look for 第21回コンピュータ将棋選手権使用可能ライブラリ and below. Libraries used by each
team is shown at the next to the last column of http://www.computer-shogi.org/wcsc21/team.html
I hope Google Translate or other machine translation service could do
a decent job translating those web pages.

Yes, CSA released those documents submitted for WCSC 20. It's linked
from the last column of the following page: http://www.computer-shogi.org/wcsc20/team.html

--
Yoshiki

[Mark Schreiber]

What is the purpose of the registered libraries? I do not understand how the registered libraries are used. Is the Bonanza code in the registered library? Can a person copy the Bonanza code into his program? Then he is running Bonanza in WCSC?

Thank you. Mark

[Yoshiki Hayashi]

There's official statement from CSA why libraries are allowed at
http://www.computer-shogi.org/library/ (in Japanese). My
understanding is that the library is there to encourage participation
of new programmers. It is not an easy task to write computer shogi
program from scratch. The registered libraries are the only piece of
code that you can use from other authors. WCSC rule says all other
part of the program must be written by yourself or the members of your
team. Without registered library system, even if someone had a
brilliant idea, he/she has to write all the mundane code to
participate in the WCSC which is likely to discourage them to even try
out their new ideas.

All source code of Bonanza 4.1.3 is part of a registered library. So
someone can copy Bonanza's code to his/her program if he/she states
the fact and is willing to be the subject of the next round advance
restriction. But there's another rule that says there must be some
originality in your program so you can't participate in WCSC with a
dead copy of Bonanza. From what I've heard, Ponanza was using
Bonanza's search + original shogi state representation more efficient
than Bonanza + slightly modified Bonanza's evaluation function.
Shibaura shogi was using Bonanza's search + original evaluation
function.

--
Yoshiki

[Mark Schreiber]

Thanks Yoshiki. Google translate is little helpful. Good computer translation is taking longer than beating the Meijin. We need computer translation tournaments.

It is interesting that computer chess tournaments allow only original programs. If registered libraries get more people writing Shogi programs, then there will be more innovation.

Mark

[Mark Schreiber]

Gergely,

   What rating would you give the computer in games of 30 minutes and 50 seconds byo-yomi?

What do you rate the game between computers Bonkras and Bonanza? Bonkras won after 140 moves in round 3 of the 21st World Computer Chess Championship on 5 May 2011. This was a 25 minute game.

   I saw there will be human against computer games. It looks like Bonanza will play 2 amateurs. Then the 2 amateurs will play Bonkras. Each game will be for 2 hours. The games will be broadcast live the morning of Sunday 24 July. How do these amateurs compare to Shimizu Ichiyo? Details at http://entcog.c.ooco.jp/entcog/event/event2011_comvshum.html

Mark

[Gergely Buglyo]

Dear Mark,
 
Good question, I have no idea. All I know is that a few years ago, top professional players seemed to agree that the best computer programs are at Shoreikai 3 Dan level. They may or may not have made a significant progress since then, I really don't know. Maybe another great human vs computer match would be just what is needed to bring the computers' strength into focus again. (I mean a game like the Watanabe vs Bonanza match, not a female pro playing a computer like Shimizu vs Akara last time. I don't mean to sound sexist here, but female pros, although they have gotten much stronger in the past years, are still a fair deal weaker than male pros. It is said that the top female pros are about Shoreikai 1 Dan level.)
 
I'm sorry I couldn't answer your question, but hopefully it will be answered soon. Best regards:
 
Gergely
 

---

Date: Thu, 2 Jun 2011 16:11:57 -0700
From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Computers Bonkras and Bonanza Will Play 2 Amateurs

--
You received this message because you are subscribed to the Google Groups "SHOGI-L" group.

To view this discussion on the web visit https://groups.google.com/d/msg/shogi-l/-/N3JSU0lKUzJCZHdK.

[Mark Schreiber]

Dear Gergely,

    I think the problem is the Japan Shogi Association will not allow professionals to play computers because they might lose. I have 3 ideas to measure the computer’s rank without a professional. Computers play 30 minute games on Shogi Club 24. Computers play games of 4 hours with amateurs. Handicap computers when playing amateurs. I think another problem is the programmers are not interested in playing humans.

It is unfortunate the JSA forbids pros. We see from chess that computers help players and attract players to the game.

Mark

[Gergely Buglyo]

Dear Mark,
 
That's right, Nihon Shogi Renmei doesn't normally allow professional players to play against computers, but people think there may be more highly advertised matches between top humans and top computers in the future. Let's not forget that such matches are also highly sponsored, which should be considered in NSR's current state of financial trouble.
 
Regards:
 
Gergely
 

---

Date: Fri, 3 Jun 2011 14:54:39 -0700
From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Re: RE: Computers Bonkras and Bonanza Will Play 2 Amateurs

--
You received this message because you are subscribed to the Google Groups "SHOGI-L" group.

To view this discussion on the web visit https://groups.google.com/d/msg/shogi-l/-/UER2b29MNXBHRzhK.

[Mark Schreiber]

Dear Gergely,

  You are right people may think there will be matches between humans and computers in the future. However, remember people have been thinking this for 4 years and there has not been a match between a male pro and a computer. Its true NSR has financial trouble but think who will pay the NSR price of 100s of millions of yen especially in today economy.

Mark

[Gergely Buglyo]

Dear Mark,
 
After the famous Shimizu-Akara match, NSR president Yonenaga was asked if there would be a male pro vs computer match soon. He said: "It was a close game that either side could have won, but Ms. Shimizu seems to have had some time trouble. We're not considering a male professional vs computer match at all for next time: it's only fair that Ms. Shimizu should get her chance at a rematch." While this does show NSR's reluctance to send a male pro to the ring, it also seems to imply they will do so if Ms. Shimizu loses again. However, I'm not really the right person to answer your questions, as I don't know anything apart from what can be read on the internet. Maybe someone (perhaps from ISPS?) with a closer connection to NSR or president Yonenaga could offer some more valuable insight.
 
Best regards:
 
Gergely
 

---

Date: Sun, 5 Jun 2011 14:47:42 -0700

From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Re: RE: Computers Bonkras and Bonanza Will Play 2 Amateurs

Dear Gergely,

  You are right people may think there will be matches between humans and computers in the future. However, remember people have been thinking this for 4 years and there has not been a match between a male pro and a computer. Its true NSR has financial trouble but think who will pay the NSR price of 100s of millions of yen especially in today economy.

Mark

--
You received this message because you are subscribed to the Google Groups "SHOGI-L" group.

To view this discussion on the web visit https://groups.google.com/d/msg/shogi-l/-/MVNWSFlqcTIzU1FK.

[BCM]

Hi,

my three cents about JSA don't allow professionals play against
computers:

there can be some resons, like fearing the pros could lose (stupid
thing - sometimes in the future this will happen).
Or a better argument is, that's all rotating about money. They like to
be sure, that all such games will bring a lot of it.
(told already.)

But in my oppinion they are waiting too long.
NOW is the point, where games of the top humans against computers are
interesting.
There still are not realy many people, being able to beat computers.
In a few years (how many???) this chance will be gone.

The JSA is missing increasing there income to maximum.

Bernhard

[Mark Schreiber]

Dear Gergely,

    I also wish more people commented. A Shimizu Akara rematch should have played already. A rematch after a year gives Ms Shimizu less chance to win because the computer is then stronger. Ms Shimizu should not have played Akara in 2010. In 2007, Habu said Bonanza is about 2 dan shoreikai. So Ms Shimizu would have been a good match for Bonanza in 2007. After 3 years of improvements, computers are too strong for Ms Shimizu. Of course, the NSR president Yonenaga is going to say Shimizu-Akara match was a close game that either side could have won. Yonenaga is trying to save the reputation and dignity of Ms Shimizu. Yonenaga would not say the game was a devastating and embarrassing loss for Ms Shimizu? Therefore, I do not think we will see Shimizu Akara rematch.

Dear Bernhard, I agree with you. In addition, a computer versus professional will give Shogi lots of publicity.

Mark

[Reijer Grimbergen]

Too busy to really enter this conversation, but I would like to add that
the biggest problem for having matches between computers and
professional players is (as always) money. The NSR doesn't want to risk
the reputation of its players unless there is a considerable amount of
money offered up. The problem the CSA has is that sponsors don't seem to
be really interested in this. This has only gotten worse since the
tsunami disaster.

Reijer

> --
> You received this message because you are subscribed to the Google
> Groups "SHOGI-L" group.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/shogi-l/-/anZ1a0FuQmVNcEVK.

> To post to this group, send email to sho...@googlegroups.com.
> To unsubscribe from this group, send email to
> shogi-l+u...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/shogi-l?hl=en.

--
Reijer Grimbergen
School of Computer Science, Tokyo University of Technology
1404-1 Katakura, Hachioji, Tokyo, 192-0982 Japan
Tel: +81-(0)42-637-2591 FAX: +81-(0)42-637-2112
http://www.teu.ac.jp/gamelab/

[Mark Schreiber]

Reijer thanks for your comment. I miss your computer tournaments report. Will you be entering the next computer tournament? Even if you do not play, a report would be nice to read. Especially at this time, now that computers play are at the professional level. 

How is NSR risking the reputation of its players when they play computers? It is obvious that computers would beat some professional players today. We know in the near future even the strongest professionals will lose to computers. This has already happened in chess and I believe Xiangqi. It will happen in Shogi and even in Go. Even great players like Watanabe admit to this. There is only disagreement about when. Will it be in 5 years or 10 years? Losing a game to a computer does not change their reputation because this is expected.

Mark

[Fred]

Hello Mark,
the reluctance of Japanese shogi organizations to permit professional male players to play against computers is probably based on a cultural sensitivity we Westerners will never be able to fully comprehend... Maybe Reijer has some more insights on this subject?

Warm regards,

Fred

[BCM]

Hi,

I also believe, that it's depending of some kind of cultural things.

I also believe, the best way to understand it for westeners is too
compare it with religious things.
Especially compare it with our history, where people thought the earth
is a disc and not a ball.

Even when people START believe that it's a ball, it nevertheless
needed
a long time to ACCEPT that.

BCM

[Mark Schreiber]

    I understand the NSR banning professionals playing computers is irrational. However, why are there so few amateur players versus computer games? There have not been any amateur versus computer games in almost a year. Do the Shogi programmers want to show how strong their programs are? Would a Shogi program beating amateur players increase sales?

Mark

[Mark Schreiber]

 Does the Shogi Club 24 prohibit computers? What does http://www.shogidojo.com/etc/softorii/index.htm say?

Thank you. Mark Schreiber

[Gergely Buglyo]

Dear Mark,
 
It says that it's prohibited for regular members to play games with the help of a computer - such an act is punished by banning the account of the individual in question. Using a program is only allowed for software developers and their affiliates.

Best regards:
 
Gergely

---

Date: Thu, 7 Jul 2011 19:55:53 -0700
From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Re: Computer Ponanza gets highest rating on Shogi Club 24

 Does the Shogi Club 24 prohibit computers? What does http://www.shogidojo.com/etc/softorii/index.htm say?

Thank you. Mark Schreiber

--
You received this message because you are subscribed to the Google Groups "SHOGI-L" group.

To view this discussion on the web visit https://groups.google.com/d/msg/shogi-l/-/cmqJ96tPslcJ.

[Mark Schreiber]

Dear Gergely,

Thank you for the explanation.

I wonder when Shogi club 24 first made this announcement. How strong did the Shogi programs need to be so that many people used them to cheat?

Mark

[Gergely Buglyo]

Dear Mark,
 
I don't know - the warning has been around for many years. Still, I don't think there are too many cheats on Shogi Club 24 (at least in my experience).
 
Best regards:
 
Gergely
 

---

Date: Fri, 8 Jul 2011 13:33:43 -0700
From: msch...@gmail.com
To: sho...@googlegroups.com
Subject: Re: RE: Computer Ponanza gets highest rating on Shogi Club 24

--
You received this message because you are subscribed to the Google Groups "SHOGI-L" group.

To view this discussion on the web visit https://groups.google.com/d/msg/shogi-l/-/UcsSw6H_n5IJ.