Yet another on line chess program

Any feedback on how the new Fun Chess 17.3 is playing ?

If anyone can just give it a game on the default level and feedback on how it plays, approx rating, anything I should add etc - much appreciated.

It has got too strong for me to play casually now. I can't seem to win anymore, playing as fast as it does. A real achievement, well played Sir.

Is the source code available? It would be really cool for me too be able to see it.

I really appreciate people giving me feedback. The program currently outplays me so it's difficult for me to see if (and where) it's going wrong. I can see all the down sides of Rxg1 but what would be best? Rg6 might allow the pawns to undouble but still leaves white with better mobility for his rooks and queen to attack blacks king. Oddly the program wasn't showing any concern for king safety at this point so I suspect I need to tweak something here. Also I didn't like it placing it's queen out of the action on b6.

So what is the rough playing strength now? 1800-2000? I'm also wondering if I should make the default level easier ie level 5 is 3 secs thinking time.

It's quite a complicated program and I didn't write it with readability in mind. It's written in Java and uses: alpha beta tree pruning, stand pat, itererative deepening, MVV/LVA move ordering, killer moves and a small opening book. The board evaluation is a mixture of piece values (which change depending on the board situation), positioning, king safety and various special set situations (mostly various endgames). You'd do best looking at the source code for a public source chess program - I believe Shredder is open source and it certainly plays better than my little program.

Have you played a game against my program?

First I am impressed you wrote this program mike, congrats!

Here is the game I played on its default settings a little bit ago as a casual game, I played similar speed as your program (I think it is 5 seconds/move?).

It seemed to misplay the ending completely and allow me to get connected passed pawns down for an easy win, but I thought it played decently throughout the opening and middle game. I am not sure I would know how to rate it and maybe later this evening I will try to put it on it's more tough settings and see how I fare against it then. From my game to judge from I would say it doesn't understand pawn play well or at least in end games, for example by ignoring the significance of my protected c3 pawn and then allowing the b file to open up so I could quickly roll on to a promotion and seemed to "lack a plan" towards the end before the game completely went south for your program. As it plays now though I think a tweak in it's end game play would make it strong enough to give the casual player a good game. Perhaps at the higher levels these issues are remedied though, when I get a chance to play it at the higher level I will give you feedback on that, and the game.

Good job!

It's not a trap, nor does it loose, since it does not drop anything. On the other hand, it has little to none real value. Black has at least half a dozen of perfectly adequate replies, including some provocative ones:

That "your" move has very little value, unless Moro is on either side of the board (he has played 2.Qe2 quite a few times himself).

In that case, you'd better find a better teacher. GM Roman is a former good player, currently selling junk for gold. Actually he does little to no work in his Roman's Labs DVD's- almost all work is done by Perelsteyn, and the standard Perelsteyn practice is deliberately omitting "annoying" lines (his very own games included!).

I played your program again on the level 9 setting (30seconds/move), didn't have time for the 1 minute/move setting this evening unforunately. Anyway, this game became much more tactical than the first game and resulted in the program eventually misjudging the position allowing for a tactical sequence which resulted in mate. It was a fun game and perhaps on the highest level it wouldn't misjudge the position and tactical possibilities like that. If I get enough time I will try to play it again on the highest setting. Well done on the program though, it is fun to play and for me challenging enough to entertain me and I'm sure you can continue to improve its play more. If you can improve the program significantly more you should make the highest setting able to think for 2-4 minutes at a time and I can use it for training games! :)

 I've played it several times. It's a nice program. I realize I could get a 10,000 line source for a chess engine online, but I would surely prefer a less complicated one.

The only reason I want it is for personal learning purposes. It might help me give you some feedback, but probably not. I don't know very much Java, but I have enough general programming knowledge that I could hopefully understand the code. I've been wanting to make an engine myself, so it might be helpful to see a basic engine.

Welllll,I played it and lost!But was playing very quickly.

RoseQueen : Rxg1??? giving up the open g file is something not even a 1500 would do.
I agree, this move did look bizarre. It appeared to be abruptly turning off king safety checks once it got to an end game. I've fixed this now.

Jordan_G : From my game to judge from I would say it doesn't understand pawn play well or at least in end games
It should - I spent long hours trying to work this stuff out. Your game pointed out a bug though and I've fixed this.

Jordan_G : If you can improve the program significantly more you should make the highest setting able to think for 2-4 minutes at a time and I can use it for training games!
Allowing it to think for 2-4 mins would help a little but it needs approx 10x the time to add another level of search. I suspect I'd do better adding more "clever stuff" into it's thought processes would have more benefit.

g-levenfish : Welllll,I played it and lost!But was playing very quickly.
Honesty is something to be very proud of - looking at the logs it appears you're not alone  Better luck next time.

I've released Fun Chess 17.4 (same URL as above) which fixes a few of the positional play issues. Hopefully it will win a few more games. I'd appreciate it if you could swallow your pride and say whether you win or loose as this allows me to rate the program. 

Just played a game fragment that indicates a bug:

Something wrong with its opening book?

Yes, level 6.

I've been back and forced it to play the same position again about forty times. Each time it plays Bd7 instead of Nxe4, although it did play Nxe4 the first time.

PrawnEatsPrawn : At this point the screen went black.... Java crash?

Something wrong with its opening book?

Java should run in a "sandbox" on your computer. The people who designed Java wanted to ensure that nothing you run on the web can harm your computer. I'd be ammazed if (any) Java program could crash your computer.

The new version just fixes some small bugs in the program regards when to check for king safety and what scores to apply to passed and weak pawns. None of this should alter the stability of the program. I made no changes to the openings.It's had about 1000 games on the web against you guys and this is the first stability issue reported.

It's odd it played Nxe4 as this move scores very poorly at all levels though this is the 1st move it will look at. My guess is that there was another program on your PC using up all the CPU and so the program never had the time to look at any other moves. I think this program then died a short while later and caused your computer to crash.

I'd suggest you restart your computer anyway. Do a virus scan first for peace of mind and then check for any updates (assuming it's windows). If anyone else experiences a similar issue then I'll rollback the changes but I'm fairly confident the issue is more isolated to your computer. Good luck.

Mike

My computer never crashed, the interface for your program did. Only the board went black, everything around it was unaffected.

Yes, I was running a resource hungry program in the background... your explanation makes perfect sense. My PC is unaffected, don't worry.

Sorry for the scare.

Chrome error (Firefox works allright):

Java Plug-in 1.6.0_27

Using JRE version 1.6.0_27-b07 Java HotSpot(TM) Client VM

User home directory = *****************************

----------------------------------------------------

c:   clear console window

f:   finalize objects on finalization queue

g:   garbage collect

h:   display this help message

l:   dump classloader list

m:   print memory usage

o:   trigger logging

q:   hide console

r:   reload policy configuration

s:   dump system and deployment properties

t:   dump thread list

v:   dump thread stack

x:   clear classloader cache

0-5: set trace level to

----------------------------------------------------

load: class chess.class not found.

java.lang.ClassNotFoundException: chess.class

 at sun.plugin2.applet.Applet2ClassLoader.findClass(Unknown Source)

 at sun.plugin2.applet.Plugin2ClassLoader.loadClass0(Unknown Source)

at sun.plugin2.applet.Plugin2ClassLoader.loadClass(Unknown Source)

at sun.plugin2.applet.Plugin2ClassLoader.loadClass(Unknown Source)

at java.lang.ClassLoader.loadClass(Unknown Source)

at sun.plugin2.applet.Plugin2ClassLoader.loadCode(Unknown Source)

at sun.plugin2.applet.Plugin2Manager.createApplet(Unknown Source)

at sun.plugin2.applet.Plugin2Manager$AppletExecutionRunnable.run(Unknown Source)

at java.lang.Thread.run(Unknown Source)

Exception: java.lang.ClassNotFoundException: chess.class

program seems way too interested im pawns (i am white) :)

I think your recent tweaks has affected it in a negative way somehow here is a short game (i have to go to work and was a piece up) I played with it on level 8;