Question about endgame databases

Discussion about development of draughts in the time of computer and Internet.
Post Reply
Ed Gilbert
Posts: 832
Joined: Sat Apr 28, 2007 14:53
Real name: Ed Gilbert
Location: Morristown, NJ USA
Contact:

Question about endgame databases

Post by Ed Gilbert » Sun Sep 19, 2021 18:06

I have a general question about using the kingsrow endgame databases in computer draughts tournaments. At least 2 programmers have told me they are reluctant to use them because they think they will be criticized for not using all original code in their draughts engines.

I hope this is not the case. The algorithms for building the dbs have been published for a long time and in many places. It does not require any new, original work, just a lot of time and computing resources. Everyone who builds the dbs ends up with the exact same information. The correct WLD counts for the various subdb slices are published so we know this is true.

I feel that it is perfectly ok to use publicly available endgame dbs in draughts engines. I invite others to express their opinions about it here.

Krzysztof, what do you have to say about this? It seems that you are ok with it, as at least one other engine in the recent tournament uses a publicly available endgame db. Are you willing to write into the rules for your tournaments that this is acceptable?

-- Ed

BertTuyt
Posts: 1521
Joined: Wed Sep 01, 2004 19:42

Re: Question about endgame databases

Post by BertTuyt » Sun Sep 19, 2021 18:25

Ed, first of all thanks for sharing this huge effort with the draughts community.
As, although it might not be rocket science, it takes lots of resources and time.

I guess your arguments make sense.
In the chess world this is already common practice, as all programs (or many) use the Syzygy tablebases.
As DBs, in general occupy much disk space, it has an advantage if we would standardize in this area.

But it remains fair to provide the credits to the person who made this possible, which is you in this case!

Bert

Krzysztof Grzelak
Posts: 1132
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Question about endgame databases

Post by Krzysztof Grzelak » Sun Sep 19, 2021 18:32

Honestly, I never really thought about it. I know that there was a proposal somewhere that all drafts engines should use one ending database during the next tournament.

Jelle Wiersma
Posts: 60
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Question about endgame databases

Post by Jelle Wiersma » Sun Sep 19, 2021 20:38

I think it is important to notice that in the ICGA tournaments of the World Computer Chess Championship and the Computer Olympiad, you are supposed to use original work, and if not, mention those who work you used; the credits!
And correct me if I'm wrong, but I thought Ed made his endgame database + driver available a few years ago. If so, you don't have to generate the 8-piece database at all, you can just use it.
The drawback I see when every program would use the 8-piece Kingsrow database, is that there will be more draws...
Personally I will likely not use the Kingsrow database.

In general, using stuff available and mentioning credits also applies to all parts of the program.
For example, the most important part; the evaluation of the position:
I would like programmers to mention if it was based upon open source, if they used tools of other people, and if they use an evaluation of another program.

Jelle

jj
Posts: 187
Joined: Sun Sep 13, 2009 23:33
Real name: Jan-Jaap van Horssen
Location: Zeist, Netherlands

Re: Question about endgame databases

Post by jj » Mon Sep 20, 2021 17:35

Ed Gilbert wrote:
Sun Sep 19, 2021 18:06
I have a general question about using the kingsrow endgame databases in computer draughts tournaments. At least 2 programmers have told me they are reluctant to use them because they think they will be criticized for not using all original code in their draughts engines.

I hope this is not the case. The algorithms for building the dbs have been published for a long time and in many places. It does not require any new, original work, just a lot of time and computing resources. Everyone who builds the dbs ends up with the exact same information. The correct WLD counts for the various subdb slices are published so we know this is true.

I feel that it is perfectly ok to use publicly available endgame dbs in draughts engines. I invite others to express their opinions about it here.

Krzysztof, what do you have to say about this? It seems that you are ok with it, as at least one other engine in the recent tournament uses a publicly available endgame db. Are you willing to write into the rules for your tournaments that this is acceptable?

-- Ed
I agree. By the way, Maximus does not use the (6 piece) Kingsrow databases anymore, but its own.

jj
Posts: 187
Joined: Sun Sep 13, 2009 23:33
Real name: Jan-Jaap van Horssen
Location: Zeist, Netherlands

Re: Question about endgame databases

Post by jj » Mon Sep 20, 2021 17:36

Jelle Wiersma wrote:
Sun Sep 19, 2021 20:38
I think it is important to notice that in the ICGA tournaments of the World Computer Chess Championship and the Computer Olympiad, you are supposed to use original work, and if not, mention those who work you used; the credits!
And correct me if I'm wrong, but I thought Ed made his endgame database + driver available a few years ago. If so, you don't have to generate the 8-piece database at all, you can just use it.
The drawback I see when every program would use the 8-piece Kingsrow database, is that there will be more draws...
Personally I will likely not use the Kingsrow database.

In general, using stuff available and mentioning credits also applies to all parts of the program.
For example, the most important part; the evaluation of the position:
I would like programmers to mention if it was based upon open source, if they used tools of other people, and if they use an evaluation of another program.

Jelle
I agree. That is what I said last year.

Jelle Wiersma
Posts: 60
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Question about endgame databases

Post by Jelle Wiersma » Mon Sep 20, 2021 20:28

jj wrote:
Mon Sep 20, 2021 17:36
Jelle Wiersma wrote:
Sun Sep 19, 2021 20:38
I think it is important to notice that in the ICGA tournaments of the World Computer Chess Championship and the Computer Olympiad, you are supposed to use original work, and if not, mention those who work you used; the credits!
And correct me if I'm wrong, but I thought Ed made his endgame database + driver available a few years ago. If so, you don't have to generate the 8-piece database at all, you can just use it.
The drawback I see when every program would use the 8-piece Kingsrow database, is that there will be more draws...
Personally I will likely not use the Kingsrow database.

In general, using stuff available and mentioning credits also applies to all parts of the program.
For example, the most important part; the evaluation of the position:
I would like programmers to mention if it was based upon open source, if they used tools of other people, and if they use an evaluation of another program.

Jelle
I agree. That is what I said last year.
Ah, interesting, can you mention the topic? (as you probably know I'm not a frequent user of this forum)

jj
Posts: 187
Joined: Sun Sep 13, 2009 23:33
Real name: Jan-Jaap van Horssen
Location: Zeist, Netherlands

Re: Question about endgame databases

Post by jj » Mon Sep 20, 2021 20:50

Jelle Wiersma wrote:
Mon Sep 20, 2021 20:28
Ah, interesting, can you mention the topic? (as you probably know I'm not a frequent user of this forum)
viewtopic.php?f=53&t=8275&start=45#p128881

pontel
Posts: 26
Joined: Tue Jan 26, 2021 21:48
Real name: João Anselmo Pontel

Re: Question about endgame databases

Post by pontel » Tue Sep 21, 2021 01:24

BertTuyt wrote:
Sun Sep 19, 2021 18:25
Ed, first of all thanks for sharing this huge effort with the draughts community.
As, although it might not be rocket science, it takes lots of resources and time.

I guess your arguments make sense.
In the chess world this is already common practice, as all programs (or many) use the Syzygy tablebases.
As DBs, in general occupy much disk space, it has an advantage if we would standardize in this area.

But it remains fair to provide the credits to the person who made this possible, which is you in this case!

Bert
It was good that you brought up the subject Mr BertTuyt, I thought these days to create a topic to suggest something similar to your proposal, the online consultation of Ending Bases similar to Chess as the link: https://syzygy-tables.info

In TCEC-sponsored inter-engine tournaments, these online hosted bases often determine the end of the match according to the scoring mark.

Of course, I understand that to host our friend Ed Gilbert's endgame database, there would be an unfeasible fixed cost.

Jelle Wiersma
Posts: 60
Joined: Wed Aug 12, 2015 17:45
Real name: Jelle Wiersma

Re: Question about endgame databases

Post by Jelle Wiersma » Tue Sep 21, 2021 07:50

jj wrote:
Mon Sep 20, 2021 20:50
Jelle Wiersma wrote:
Mon Sep 20, 2021 20:28
Ah, interesting, can you mention the topic? (as you probably know I'm not a frequent user of this forum)
viewtopic.php?f=53&t=8275&start=45#p128881
@Jan-Jaap, I largely read the topic and like to answer your question of May 26, 2020: What do the others think? (Ed, Jelle, Michel, Harm, etc.):
In my opinion using stuff of other people (source code, tools, data, evaluation, database, etc.) is allowed, but one should mention those who work you used; the credits.
Unfortunately, I never see this kind of information on tournament sites, also not on ICGA's..., which effectively kind of nullifies my opinion.
So, in general, I understand your frustration.

clp
Posts: 123
Joined: Tue Jul 10, 2018 14:28
Real name: Kees Pippel
Location: IJmuiden

Re: Question about endgame databases

Post by clp » Tue Sep 21, 2021 09:42

Jelle Wiersma wrote:
Tue Sep 21, 2021 07:50
Unfortunately, I never see this kind of information on tournament sites, also not on ICGA's..., which effectively kind of nullifies my opinion.
So, in general, I understand your frustration.
Sorry to interrupt this conversation.

A first step could be to agree on the components (opening book, endgames, optimization models, other libraries relevant to the strength of the engines) and where and how to list them. The version information seems a logical place to me. It is then easy for a tournament organizer to publish this data.

Post Reply