T2 FM: A Nice Game of Chess (2013-04-16)

Swiz on 19/4/2013 at 15:08

I've not played chess in years ( as I am terrible at it ) but isn't there some rule where if the exact same board position comes up 3 times it's officially stalemate?

Tannar on 19/4/2013 at 16:09

@ gigagooga: I had a similar experience during testing. It seems that the Fruit engine doesn't always deal with the end of the game the way it should. I'm sure Kdau will comment on this more.

@ Swiz: That's not exactly right. The rule you are thinking of is called (http://en.wikipedia.org/wiki/Threefold_repetition) Threefold Repetition. Kdau didn't implement it in this version but I'm sure he will have more to say about it.

kdau on 19/4/2013 at 20:21

Thanks for noting that, gigagooga. Between your and Tannar's experiences, this really seems to be a consistent problem. Were you playing on Easy difficulty? I just quizzed the engine directly on a position like you described (kings on light corners, black rooks at a3 and f8). On Easy (which is 1-ply evaluation), it missed the mate and tried Kb7?, but on Normal (which is 4-ply) it correctly suggested Ra2, expecting the game to finish Kg1 Kb7 Kh1 Rf1 0-1. (UCI output below.) This is probably due to Fruit 2.1 not supporting endgame tablebases, which help engines identify strategies for specific endgame position classes. I hope to find a stronger open-source engine for the next version of this mission. For now, let's call it a draw. :erg:

Code:

> position fen k4r2/8/8/8/8/r7/8/7K b - - 0 1
> go depth 9 movetime 7500
info depth 1 seldepth 4 score cp 1023 time 1 nodes 10 pv a3a1 h1g2
info depth 1 seldepth 4 score cp 1099 time 4 nodes 11 pv a3a2
info depth 1 seldepth 4 score cp 1110 time 7 nodes 28 pv a8b7
...
info depth 4 seldepth 10 score mate 3 time 28 nodes 2889 pv a3a2 h1g1 a8b7 g1h1 f8f1
...
info depth 9 seldepth 10 score mate 3 time 55 nodes 7484 pv a3a2 h1g1 a8b7 g1h1 f8f1

Swiz and Tannar, you're right that (although it shouldn't have happened in the first place) this case should have fallen under the threefold repetition rule, and I do intend to add that in the 1.2 version. In gigagooga's case, with no pawns available to advance, the fifty-move rule would have eventually kicked in, but who would want to wait that long? :cheeky:

Edit: I've found that the (http://www.computerchess.info/tdbb/phpBB3/viewtopic.php?f=9&t=665) Toga II 3.0 engine, an open-source Fruit derivative, is the cutting edge these days. Even at 1-ply without an endgame tablebase installed, it correctly suggested Ra2 for the position above. This will probably be the engine I use in my 1.2 version, though I will need to adjust the difficulty settings to compensate for its strength. If anyone would like to try it, you can download it from the post I just linked and replace the engine.ose file in my mission with toga-30-ja-release\Windows\32\toga-30-32-ja.exe from its ZIP file.

gnartsch on 20/4/2013 at 13:41

Very impressive and very well done ! :thumb:
Too bad that I suck at chess :p

Weasel on 20/4/2013 at 18:39

I agree with gnartsch. This mission is beautiful! Having three different settings puts it over the top. Each one is really well done. I know what I'll be doing any time I'm in the mood to play chess from now on (playing this version)!

SlyFoxx on 22/4/2013 at 02:31

After ten or so games I got a win. It was a bloodbath of even exchanges at first but I finally arrived at (white) queen vs. (black) two rooks. But I had a three pawn advantage with two passed pawns. I forked one rook away and took the second in a queen exchange for a promoted pawn. This left queen, king vs. king, pawn and an easy win from there. Whew...my chess is rusty.

The win was odd because my end game has always been weak. Fewer pieces is usually an advantage for the computer but the passed pawns were big and for some reason the machine bit on taking my queen while sacrificing it's last rook leaving me an easy promotion.

Oh well. You take your wins when they come!
:sly:

(http://s20.photobucket.com/user/DromEd/media/mate1_zps8d4ab89c.jpg.html)

Inline Image:

http://i20.photobucket.com/albums/b202/DromEd/mate1_zps8d4ab89c.jpg

Zontik on 22/4/2013 at 10:46

I have a problem playing on netbook under Windows 7. Every 5-10 minutes I get a message about "The engine can't play with you anymore..." and then I have to finish the game myself, making moves instead of the computer. Quick load saves my effort, but I even can't guess what causes the problem.

thiefessa on 22/4/2013 at 16:24

I'm a chess addict and this is a lot of fun to play! Really enjoying it, congratulations and thank you. :thumb:

Peanuckle on 22/4/2013 at 20:35

I love FMs that aren't Thief, but something neat with the Dark Engine. Like that one mission that was a giant arena where you pushed buttons to spawn enemies and have them fight. That was a good timewaster.
Can't wait to see how I stack up against the AI.

kdau on 23/4/2013 at 01:30

Quote Posted by SlyFoxx

... The win was odd because my end game has always been weak. Fewer pieces is usually an advantage for the computer but the passed pawns were big and for some reason the machine bit on taking my queen while sacrificing it's last rook leaving me an easy promotion. ...

Heh, another endgame deficiency with Fruit 2.1. Well, at least it takes the edge off an otherwise too strong engine. :)

Quote Posted by Zontik

Strange. If the computer makes any moves at all, the engine program must be loading correctly. So the engine must be crashing in the middle of the game. :weird: I'll PM you with some debugging information.

Quote Posted by Peanuckle

Ah yes, Creature Clash 2. I always liked to fill the arena with frogbeasts, open the gates, and let the explosions thin the herd. :sly: