Computer draughts tournaments

[BertTuyt]

Ed, replayed your game against Truus, impressive i would say.
Damage basically plays the same sequence (even the sacrifices).
Scores improve for black from around 0.2 at move 29 towards 0.9 at move 39.
But nowhere does it see a database win.
It was even difficult to get the win result, as some lines went into a draw (for whatever reason).
Guess most programs will have difficulty in realizing the win against Kingsrow with these databases.

What was again your approach, 4 computers working for 1/2 year (i think i need a discussion with my wife to increase computer budget here).

Bert

[Ed Gilbert]

Hi Bert,

Thanks for explaining the details of your multi-core search stats. I understand now. When you improve the load balancing and get the shared lockless hashtable working, you will likely be searching above 4500Mn/sec.

What was again your approach, 4 computers working for 1/2 year (i think i need a discussion with my wife to increase computer budget here).

This time you don't need to discuss it with your wife, because you already have the tools. (But I have used a few innovative ways to get another computer online without the wife noticing, I could tell you a few stories...). These are 4 old computers, one purchased in 2002, and the other 3 I built myself in 2004. The average power is about that of an Athlon 2700. Each core of your quad core is twice as powerful as one of these machines. I'm quite sure of this because I've benchmarked building databases with my core 2 duo, and its 2 cores working together can build databases as fast as all 4 of my older machines working in parallel. You could build these imcomplete 8 and 9-piece slices in 2 to 3 months on your machine. Also, the most time consuming slices to build were the 9-piece slices with 1 king on a side, and I ended up not even using those at Culemborg because they were just too large to be used effectively. I still use them sometimes when people send me endgame positions for analysis, but for a full game they are not helpful, probably because all the disk I/O slows the search too much. Those 2 slices alone consume almost 200GB of disk space.

-- Ed

[mschribr]

Congratulations Ed.
How do you think Kingsrow would do in a tournament such as the world championship in Hardenberg, Netherlands that was June 2007? See the games at wkdammen2007.nl/En/index.html
Mark

[Ed Gilbert]

Congratulations Ed.
How do you think Kingsrow would do in a tournament such as the world championship in Hardenberg, Netherlands that was June 2007? See the games at wkdammen2007.nl/En/index.html
Mark

Thank you Mark. I really have no way to predict how Kingsrow would do in that situation. I could use it to analyze the games that were played, and possibly find some errors that were made, but Kingsrow might make its own strategic errors, so who knows? Don't try to infer too much about Kingsrow from the Culemborg result. It is only 9 games, far too few to determine the true relative strengths of the participants.

I see you are in NYC. How is it that you are interested in international draughts? I haven't seen it played anywhere in the US.

-- Ed

[mschribr]

I really have no way to predict how Kingsrow would do in that situation. I could use it to analyze the games that were played, and possibly find some errors that were made, but Kingsrow might make its own strategic errors, so who knows? Don't try to infer too much about Kingsrow from the Culemborg result. It is only 9 games, far too few to determine the true relative strengths of the participants.

I see you are in NYC. How is it that you are interested in international draughts? I haven't seen it played anywhere in the US.

-- Ed

Hi Ed,
But would you like Kingsrow to compete in a world championship to see how strong it is? Do you think such a competition has a limited window? In about 5 years a computer would never lose to a human the same way checkers was about 8 years ago.

I play international draughts among the other versions of Draughts. My main interest in international draughts is from the few man versus computer competitions. I think the computer is already unbeatable in draughts. It just a matter of proving it. I think the only other popular classic strategy games where the world champion could easily beat the computer are shogi and go.

The Federations for Draughts in the US is the AICS - American International Checkers Society. They are small. They are the only one in the US I have ever heard even talk about international draughts. You can find contact information at fmjd.org. Click Federations then America then American International Checkers Society. I was a member. They sent out a 4 page quarterly newsletter. They try to regularly hold national tournaments to crown the United States Champion. They have had tournaments in conjunction with the ACF. They were looking for ways to increase membership by increasing the popularity of draughts. I suggested a US Champion vs computer tournament would get publicity. This happened in the Tinsley vs Chinook match. My suggestion was not embraced.
Mark

[Ed Gilbert]

But would you like Kingsrow to compete in a world championship to see how strong it is? Do you think such a competition has a limited window? In about 5 years a computer would never lose to a human the same way checkers was about 8 years ago.

Mark, I think most players don't like to play against computers. They use them for analysis and for learning, but not for competition. I'm sure you are right that computers will eventually overtake humans in draughts (if they haven't already, I don't know), but I am primarily interested in computer vs. computer draughts competition, not computer vs. people.

[TAILLE]

But would you like Kingsrow to compete in a world championship to see how strong it is? Do you think such a competition has a limited window? In about 5 years a computer would never lose to a human the same way checkers was about 8 years ago.

Mark, I think most players don't like to play against computers. They use them for analysis and for learning, but not for competition. I'm sure you are right that computers will eventually overtake humans in draughts (if they haven't already, I don't know), but I am primarily interested in computer vs. computer draughts competition, not computer vs. people.

Hi Ed.

Thank you very much Ed for having sent my a picture from the Culemborg tournament. I tried to answer you at 'eyg@prodigy.net' address it did not work.

The question of games between computer and human is not that simple. I understand that some people would like to see such game but is it really a good idea for draughts? For the short term this could be a kind of pub for draughts but what about the long term ? The worse case would be a win of the computer. In that case I imagine that the number of new young players will dramaticaly diminish.

In addition what could be the interest of a GMI in such game ? Of course we can imagine giving a lot of money to a GMI for competing a stong program but that does not seem worthwhile for draughts.

I agree with you Ed. Program can efficiently be used by humans for analysis but let's the computers compete between them.

As far as I am concerned, if I want to know what will be the result of a game between Damy and a GMI it is very easy. I ask one of the french GMI to play a "training" game against Damy (without any spectator!!!) and then this GMI will be able to tell me what really is the strength of Damy. Of course I will never give the result of such experience !

Gérard

[Rein Halbersma]

But would you like Kingsrow to compete in a world championship to see how strong it is? Do you think such a competition has a limited window? In about 5 years a computer would never lose to a human the same way checkers was about 8 years ago.

Mark, I think most players don't like to play against computers. They use them for analysis and for learning, but not for competition. I'm sure you are right that computers will eventually overtake humans in draughts (if they haven't already, I don't know), but I am primarily interested in computer vs. computer draughts competition, not computer vs. people.

Hi Ed.

Thank you very much Ed for having sent my a picture from the Culemborg tournament. I tried to answer you at 'eyg@prodigy.net' address it did not work.

The question of games between computer and human is not that simple. I understand that some people would like to see such game but is it really a good idea for draughts? For the short term this could be a kind of pub for draughts but what about the long term ? The worse case would be a win of the computer. In that case I imagine that the number of new young players will dramaticaly diminish.

In addition what could be the interest of a GMI in such game ? Of course we can imagine giving a lot of money to a GMI for competing a stong program but that does not seem worthwhile for draughts.

I agree with you Ed. Program can efficiently be used by humans for analysis but let's the computers compete between them.

As far as I am concerned, if I want to know what will be the result of a game between Damy and a GMI it is very easy. I ask one of the french GMI to play a "training" game against Damy (without any spectator!!!) and then this GMI will be able to tell me what really is the strength of Damy. Of course I will never give the result of such experience !

Gérard

Gerard,

It's an interesting subject for speculation.

There have been 4 reasonably long and public matches of humans vs computers. The two Buggy vs Samb matches and the matches of Flits against Krajenbrink and Valneris. Buggy lost the first and convincingly won the second. Krajenbrink lost with a large margin (although he had winning chanches in two drawn games). Valneris drew 6 boring games if I recall correctly.

Arguments for computer dominance: All matches were more than 4 years ago. So I would think that with the new technology, bigger databases etc., humans will find it very hard to win unless they can devise "anti-computer play" as the chess grandmasters were able to do for a while. But even in chess, computers (Rybka e.g.) are dominating grandmasters even at pawn odds or without opening books. Only the absolute top GMs like Kramnik are not being completely destroyed.

Arguments for human equivalence(/dominance?): Perhaps the big drawing margin in draughts will keep humans from losing for the near future. Also, there has been considerably less effort been put into building the best possible draughts program than with chess. No disrepect to the many, many hours that single dedicated individuals put into their programs, but chess computers are big team efforts, often with the Also the chess community is much more active with bulletin boards (Talkchess, Winboard forums) and there are strong open source draughts program (like Fruit or Crafty) to learn from.

So who knows? A match would certainly interesting. Of course, a GM would want to be paid for it, but it could be interesting publicity for some computer company as well to have a program run on its hardware.

Rein

[mschribr]

In the long run World Champion vs Computer matches will also increase popularity of draughts. These matches will inspire more programmers to write programs. This will increase competition that will increase the level of play of the programs. The publicity will also generate more people interested in playing draughts. This will inspire programmers to write programs that teach draughts. Take fritz it is a great teaching and training program for chess. Many new draughts players would start to play if there was an easy to use program to teach them to play better draughts.

You ask what interest a GMI has in a computer match? We learn to play better by studying the moves of stronger players. If the computer will win a match 20-0 then all GMIs would study the computer moves to learn to play better.

If we look at 3 other games where world champions have decisively lost to a computer the games have not dropped in popularity. World champions have lost in othello, checkers and chess and these games are no less popular today then before the matches.
Mark

[ildjarn]

Mark, I think most players don't like to play against computers.

I've played twice against a computer in an official game (both in Geleen Open, once against Cerberus, once against Truus), and I didn't really mind playing against a computer. Although, with the current strength of computers, and my own weakness (I haven't played an official match in over 6 years now), I wouldn't do it .

[Ed Gilbert]

Thank you very much Ed for having sent my a picture from the Culemborg tournament. I tried to answer you at 'eyg@prodigy.net' address it did not work.

I discovered recently that my prodigy address does not always deliver emails to me originating from the .nl domain, and now it appears to have the same problem from your domain. I've been complaining to the support people at my ISP but they say there is no problem. If you received a reply about undeliverable mail from your attempt to email me and if it is still in your inbox or the deleted mail folder, could you forward it to me? You can use my alternate email address which has always been reliable for overseas email, it is "eygilbert at verizon.net".

-- Ed

[Hanco Elenbaas]

De Gelderlander, 4 december 2007

Debutant uit Verenigde Staten wint Open NK

CULEMBORG - De Amerikaan Ed Gilbert is zondag in Culemborg de verrassende winnaar geworden van het Open Nederlandse Kampioenschap computerdammen.

<center>
De spanning is van de gezichten af te scheppen tijdens het
Open Nederlands kampioenschap computerdammen. </center>

Gilbert, afkomstig uit New Jersey, zorgde daarmee voor een sensationale afloop van het toernooi in Culemborg.
In de damwereld is hij totaal onbekend, zijn programma Kingsrow ook. De Amerikaan heeft geen achtergrond in de damsport maar is als programmeur des te beter thuis. Dat bleek op het toernooi waarin verschillende gerenommeerde dammers met hun software elkaars rekenkracht testten. Gilbert had het beste programma: hij won vijf keer en speelde vier keer remise. Een beter debuut had de Amerikaan – die de moeite nam om zelf naar Nederland te komen om zijn vindingen live te beproeven – zich niet kunnen wensen. Kenners waren vol lof over de aanpak van de nieuwe Open Nederlandse kampioen: "Hij heeft in de gebruikte databases enkele slimmigheidjes ingebouwd waardoor Kingsrow in het middenspel duidelijk beter is dan de meeste overige programma's."

Kingsrow had in de eindstand één punt voorsprong op de Fransman Gerard Taille met zijn programma Damy en op de Nederlander Chris Jurriëns uit Eibergen met het damprogramma DIOS.

[mschribr]

Hi Hanco,
This picture is interesting. 2 people standing and maybe grinning behind a calm and confident operator as he presses the clock. While the opponent, his face half turned away, stares at his computer screen. Can someone give the names to the people in the picture? Which game is being played?
Mark

[Rein Halbersma]

Hi Hanco,
This picture is interesting. 2 people standing and maybe grinning behind a calm and confident operator as he presses the clock. While the opponent, his face half turned away, stares at his computer screen. Can someone give the names to the people in the picture? Which game is being played?
Mark

The confident operator is Jelle Wiersma (Sjende Blyn) playing against Gerard Taille (Damy). The two spectators him are Frank Mesander (Tornado) and Ton Tillemans (TD King). Next in line are Frits Luteyn (specatator) and referee Jaap Bus. Ed Gilbert (Kingsrow) is also visible behind the black monitor on the table behind the Sjende Blyn-Damy game.

[Piet Bouma]

Open French Championship Rapid Polish Draughts for Computerprograms 2008 in Gouy Sous Bellonne.

Games and results on: <a href=http://toernooibase.kndb.nl/opvraag/sta ... d=954&jr=8 target=blank>Toernooibase.</a>

also: http://hrx.free.fr/Gouy_sous_Bellonne_2 ... s_eng.html