Scan - new version

Discussion about development of draughts in the time of computer and Internet.
Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Scan - new version

Post by Krzysztof Grzelak » Thu Sep 17, 2020 09:15

I have a huge and heartfelt request to Fabien Letouzey. Can you make a new version of the program Scan. I am asking you to include a draw in the game, so:

- if during the next 25 moves only kings are rearranged, and none of the players moves with an ordinary pawn or captures;
- if the same position / setting on the board has appeared a third time.

And also that you improve the base of endings along with the kings.

Thank you very much in advance.

Fabien Letouzey
Posts: 299
Joined: Tue Jul 07, 2015 07:48
Real name: Fabien Letouzey

Re: Scan - new version

Post by Fabien Letouzey » Fri Sep 18, 2020 12:53

Scan is an engine, and that means it is simplified in various ways. Engines like Scan do not know that they are playing a game; you give them a position, and they answer a move. So it's up to the interface or the user to interpret the result; Scan will always play a move no matter what (not offer draw or anything else).

Now about the king rule. When I first wrote Scan, I asked Jaap Bus about that rule. He said it wasn't used in our (real life) tournaments, and it didn't make sense for computers. So it would be dangerous for a program to always apply the rule. Presumably, some endgames need more moves to win (there will always be an endgame with even more moves, so no arbitrary limit can be correct).

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Fri Sep 18, 2020 19:55

And what will happen if the engine repeats the same position three times.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Fri Sep 18, 2020 20:07

Another argument is that Jaap Bus is also not entirely right. He does 20 minutes or 15 out of 75 moves. Imagine Jaap Bus is doing a 3-minute tournament of 75 moves. In this case (if someone insisted - or one of the programmers would play to the end) there are no more agreements between the programmers - the program would have to play to the end.

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Scan - new version

Post by Sidiki » Fri Sep 18, 2020 23:19

Hi,

I think that excepted, some more complicated positions including King, any 6 endgame or any engine will show the score 00 for Scan or 3 for Kingsrow. Also any human with a minimum level can know a draw position.

Sorry but, i don't see the interest to do a problem with this. Kingsrow, according to an option implemented declare a game as draw when a certain position appear, i don't know if it's why you insist.

I am sure of one thing, it's that any program, even the more weaker will show an evaluation score under 5 to show that a game it's draw. As operator, also, you can see with the position and the scores of both programs without forget your draughts level that it's a draughts and stop the game.
Friendly, Sidiki.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 08:50

Sidiki wrote:
Fri Sep 18, 2020 23:19
Hi,

I think that excepted, some more complicated positions including King, any 6 endgame or any engine will show the score 00 for Scan or 3 for Kingsrow. Also any human with a minimum level can know a draw position.

Sorry but, i don't see the interest to do a problem with this. Kingsrow, according to an option implemented declare a game as draw when a certain position appear, i don't know if it's why you insist.

I am sure of one thing, it's that any program, even the more weaker will show an evaluation score under 5 to show that a game it's draw. As operator, also, you can see with the position and the scores of both programs without forget your draughts level that it's a draughts and stop the game.
Friendly, Sidiki.
I will ask you a few questions.

1. whether you know what will happen as three repeated a Scan 3.1 program will repeat the position - they thought as the draw.
2. a game of two programs is played - game time 3 minutes for 75 moves. One of the developers I want to play until the end of the game.
And with the 25-move program, I can't win the game.
3. There is no way the operator will stop the game.
4. Programs must play from the very beginning to the end by themselves and not rely on the fact that one of the programmers says yes, "surrenders the game - you won".
5. Here is an interesting example.

Image

And the game is played like this:

1. 17-44 32-37
2. 17-44 32-37
3. 17-44 32-37

And imagine that the program has games for completing about 30 seconds.

Fabien Letouzey
Posts: 299
Joined: Tue Jul 07, 2015 07:48
Real name: Fabien Letouzey

Re: Scan - new version

Post by Fabien Letouzey » Sat Sep 19, 2020 08:54

Krzysztof Grzelak wrote: And what will happen if the engine repeats the same position three times.
It's up to the user or interface to interpret the result, not the engine. The engine will just keep playing moves. Its job is to play, not organise tournaments.

It's possible that, even after all those years, you are still confusing Scan (the subject of your request) and Hub. Scan is the brain and plays moves; Hub displays positions on the screen and handles mouse clicks.

Hub does handle repetitions, but not the king rule (since it's not always applied). It will display "End" as the status, and no more moves will be played. During development I use 2 for the repetition counter. It seems that during the public release of version 2.1, I forgot to set it to 3 (which I never use).
Another argument is that Jaap Bus is also not entirely right. He does 20 minutes or 15 out of 75 moves. Imagine Jaap Bus is doing a 3-minute tournament of 75 moves. In this case (if someone insisted - or one of the programmers would play to the end) there are no more agreements between the programmers - the program would have to play to the end.
I don't understand what you are trying to say. Changing the total time cannot affect anything, let alone the result of a game. I remember one instance from a real-life tournament where Scan was in a winning (but difficult) position against TD King; a 4 pieces vs 2 pieces endgame. They stopped playing after move 75, and Jaap came to adjudicate it. First he asked whether the two programs knew the result. Yes, they did, as the position only had 6 pieces in it. Then he spent some time figuring out what the winning plan was, presumably to convince himself that the position was indeed winning (remember that Jaap is a strong player).

I don't understand your paragraph, but I can comment on the "play to the end" part. It is customary in other games, and chess in particular, to always play the game to the end. Tournament directors use special rules to shorten boring endgames, like king+rook vs king+rook (similar to king vs king in draughts). lidraughts for instance stops king vs king after 4 plies, since the longest possible win in this endgame is 3 plies (move the kings into opposite corners). This is not an official rule (except in Frisian draughts, I think), but it makes games much more pleasant.

Unfortunately, I don't think that playing to the end is possible in draughts. For some reason, the tradition seems to be to stop the game after a fixed number of moves. The problem is, what to do after that? It seems that you need a way to adjudicate the game; using endgame tables from another program would make sense. We can't play to the end, because old programs will refuse to continue the game I guess (or run out of time if they did, since it's not clear whether new time will be added).

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 09:40

Fabien Letouzey wrote:
Sat Sep 19, 2020 08:54
It's up to the user or interface to interpret the result, not the engine. The engine will just keep playing moves. Its job is to play, not organise tournaments.

It's possible that, even after all those years, you are still confusing Scan (the subject of your request) and Hub. Scan is the brain and plays moves; Hub displays positions on the screen and handles mouse clicks.

Hub does handle repetitions, but not the king rule (since it's not always applied). It will display "End" as the status, and no more moves will be played. During development I use 2 for the repetition counter. It seems that during the public release of version 2.1, I forgot to set it to 3 (which I never use).
Well, here are the errors of the Scan program. Item as above. Now I will write what happened in the tournament during the game. You asked me why the program stopped and didn't go on. And what really happened in the game during the tournament (from what I remember):

Scan 3.1 played 44-17. The program showed a rating of 0.00. And exactly 44-17/76or79 0.00.
At this point, the program showed a message:
d:/Scan International/cd hub
Time to finish the game from the Scan program is about 30 seconds.

At this point, it was impossible to play any more. Can you write what happened that made the program behave like this?
Fabien Letouzey wrote:
Sat Sep 19, 2020 08:54
I don't understand what you are trying to say. Changing the total time cannot affect anything, let alone the result of a game. I remember one instance from a real-life tournament where Scan was in a winning (but difficult) position against TD King; a 4 pieces vs 2 pieces endgame. They stopped playing after move 75, and Jaap came to adjudicate it. First he asked whether the two programs knew the result. Yes, they did, as the position only had 6 pieces in it. Then he spent some time figuring out what the winning plan was, presumably to convince himself that the position was indeed winning (remember that Jaap is a strong player).
I consider it a very serious mistake by the referee. The main referee can not decide the game. From it You and the Tone were. The judge can only intervene in a few cases. Which in this case should do judge. Play the game on to see who wins or loses the game. Well, unless the players themselves determined the result of the game.
Last edited by Krzysztof Grzelak on Sat Sep 19, 2020 10:11, edited 1 time in total.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 10:06

Fabien Letouzey wrote:
Sat Sep 19, 2020 08:54
I don't understand your paragraph, but I can comment on the "play to the end" part. It is customary in other games, and chess in particular, to always play the game to the end. Tournament directors use special rules to shorten boring endgames, like king+rook vs king+rook (similar to king vs king in draughts). lidraughts for instance stops king vs king after 4 plies, since the longest possible win in this endgame is 3 plies (move the kings into opposite corners). This is not an official rule (except in Frisian draughts, I think), but it makes games much more pleasant.
In draughts, it should also be no matter what the endgame base a given program has.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 10:18

Fabien Letouzey wrote:
Sat Sep 19, 2020 08:54
Unfortunately, I don't think that playing to the end is possible in draughts. For some reason, the tradition seems to be to stop the game after a fixed number of moves. The problem is, what to do after that? It seems that you need a way to adjudicate the game; using endgame tables from another program would make sense. We can't play to the end, because old programs will refuse to continue the game I guess (or run out of time if they did, since it's not clear whether new time will be added).
You write strange things. Every single program is to indicate to me whether the program is failing or not. During the tournament, I can only rely on what program is playing, not follow any other program. There will be misunderstandings and absurdities.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 10:23

In Aurora Borealis Professional, if the position is repeated 3 times, the program shows a window with the word "Draw" and the program does not play any further. So it should look like.

Sidiki
Posts: 313
Joined: Thu Jan 15, 2015 16:28
Real name: Coulibaly Sidiki

Re: Scan - new version

Post by Sidiki » Sat Sep 19, 2020 10:37

Hi Krzysztof,
Concerning program writing, i think that any programmer is most placed to explain anything to you. I program too, and what Fabien try to tell to you, and i done it before. You don't understood because you stay focalised on what it's in your mind.
IF YOU ARE REALLY A GOOD DRAUGHTS PLAYER, YOU DON'T NECESSARY NEED THAT AFTER A REPETITION OF 3 OR 4 MOVES, THE ADVICE OF ANY PROGRAM TO ADJUCATE A DRAW GAME ISSUE.

Answer to this question :
Because, you are writing to play until the end of the game.
WHEN YOU DO SCAN VS SCAN, YOU NEVER SEEN THE GAME STOPPED WITH A DRAW SCORE OR MESSAGE ?

Answer sincerly please.

Sidiki.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 10:48

Sidiki wrote:
Sat Sep 19, 2020 10:37
Hi Krzysztof,
Concerning program writing, i think that any programmer is most placed to explain anything to you. I program too, and what Fabien try to tell to you, and i done it before. You don't understood because you stay focalised on what it's in your mind.
IF YOU ARE REALLY A GOOD DRAUGHTS PLAYER, YOU DON'T NECESSARY NEED THAT AFTER A REPETITION OF 3 OR 4 MOVES, THE ADVICE OF ANY PROGRAM TO ADJUCATE A DRAW GAME ISSUE.
It is not the player that matters, but the program that is playing. The player is not just a game program.
Sidiki wrote:
Sat Sep 19, 2020 10:37
Answer to this question :
Because, you are writing to play until the end of the game.
WHEN YOU DO SCAN VS SCAN, YOU NEVER SEEN THE GAME STOPPED WITH A DRAW SCORE OR MESSAGE ?

Answer sincerly please.

Sidiki.
I have never played a Scan vs. Scan game.With the notice that the game was stopped, I NEVER HAD CONTACT.

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 10:56

Of course, if the same position was repeated 3 times, I gave it a draw. But when I was playing Scan, this was the case, and I was surprised it couldn't be played anymore.

@Sidiki

Another thing is that you count too much on the player, and let the player make a mistake, what then?

Krzysztof Grzelak
Posts: 1307
Joined: Thu Jun 20, 2013 17:16
Real name: Krzysztof Grzelak

Re: Scan - new version

Post by Krzysztof Grzelak » Sat Sep 19, 2020 11:21

I wonder how the user of the Scan 3.1 program is informed that the same position has occurred three times.

Post Reply