Unofficial World Championships Of Computer Programs Rapid 2019

[Ed Gilbert]

Krzysztof, can you post the game here? Thanks.

-- Ed

[Jelle Wiersma]

It's this position (not easy to win):
https://1drv.ms/u/s!ApHwKy0s1tFKhSv-QAk ... 5?e=REWCN1

[Ed Gilbert]

Thank you Jelle.

If I have setup the position correctly ([FEN "W:W15,K35,K49:B4,K25,K33,39,K50"]), this is a black win in the kingsrow endgame db.

IMO, if you are going to run a tournament where the games are stopped after 65 moves, then you have to attempt to adjudicate games that are still in play when they are interrupted at the 65-move limit. If a game has been stopped for this reason, and end position is a clear win, loss, or draw result, then the game result should be declared as per the last position.

Since scan was in a winning position after 65 moves, I think the game should be declared as a win for scan.

-- Ed

[Krzysztof Grzelak]

I am having dinner only now. He is thinking what now with it to do.

[Jelle Wiersma]

I just gave the "lost" game to Krzysztof (he asked me for it).

[Krzysztof Grzelak]

Thank you, Jelle. After a long thought and deep analysis of the game from the 21st round of Moby Dam - Scan change the result to win the Scan program. Tomorrow morning I will post the game and exchange the result of the Moby Dam - Scan match where live games are played.

[Jelle Wiersma]

Hi Krzysztof,
I think you made the right decision.
Jelle

[Krzysztof Grzelak]

Thank you, Jelle.

[Krzysztof Grzelak]

Tournament " The Unofficial World Championships Of Computer Programs In International Draughts Rapid 2019 " ended with victory of the program Sjende Blyn by Jelle Wiersma, second and third place was taken by Scan by Fabien Letouzey and Maximus by Jan-Jaap van Horssen. Congratulations to Jelle Wiersma, Fabien Letouzey and Jan-Jaap van Horssen. Many thanks to the other participants in the tournament.

Krzysztof.

[jj]

For ease of reference:

Tournament.jpg (151.89 KiB) Viewed 10438 times

[jj]

Congratulations to Jelle for winning the tournament! Scoring 2 points against numbers 2-5, 3 points against numbers 6-10 and 4 points against numbers 11-15 is impressive. Congratulations to Fabien for taking second place and to Ed for staying also undefeated.

Also congratulations to Krzysztof for successfully conducting the tournament. Playing 210 games in 10 days is no small feat. Thank you for playing this tournament, I enjoyed many of the live games.

Maximus did well in the tournament. I will shortly post some info on the new version and some DXP match results in a separate thread, see https://laatste.info/bb3/viewtopic.php?f=53&t=8156.

Jelle, what can you tell us about Sjende Blyn? Do you also use machine learning?

We all know that a tournament, even if it has as many rounds as this one, does not definitively tell us which program is the strongest. Can I propose -following computer chess- a superfinal between numbers 1 and 2, Sjende Blyn and Scan?

Which brings me to the following point: maybe winning a strong tournament (with many opponents of varying strength) and winning a match against the strongest opponent are two separate qualities. In computer chess we see Stockfish winning the tournaments and Leela winning the matches. In human competitions we also have "tournament players" and "match players". In a tournament winning is important, in a match not losing is important. Any thoughts on this?

Jan-Jaap

[Jelle Wiersma]

Krzysztof: Thank you for organizing this tournament!
Playing this amount of games is quite an effort!

Thank you Jan-Jaap for the congratulations.
You, and Fabien, also my congratulations with the 3rd and 2nd place respectively!

@Jan-Jaap
Sjende Blyn is the only program in the top 5 of this tournament that does not use machine learning.
In my opinion Sjende Blyn is a "tournament player", not a "match player":
Originally I designed the program to play against humans,
later on the focus shifted to play against programs (more tournaments against programs, than against humans),
but this design is still present in the "DNA" of Sjende Blyn.
As such, Sjende Blyn does not necessarily select the "best" position, but may select the position with the best "opportunities" to win.

Jelle

[jj]

Sjende Blyn is the only program in the top 5 of this tournament that does not use machine learning.
In my opinion Sjende Blyn is a "tournament player", not a "match player":
Originally I designed the program to play against humans,
later on the focus shifted to play against programs (more tournaments against programs, than against humans),
but this design is still present in the "DNA" of Sjende Blyn.
As such, Sjende Blyn does not necessarily select the "best" position, but may select the position with the best "opportunities" to win.

Okay thanks for the info. Although it is a bit abstract. Interesting, this clash of different approaches. Did you improve much in the past year?

Jan-Jaap

[EricvanDusseldorp]

Jelle,
Until August Sjende Blyn was not the strongest program. How did you improve it?

[Jelle Wiersma]

I fixed some bugs and applied changes to the handling of isolated man (geisoleerde schijven) and locks (opsluitingen).
Apparently, it worked