chessgame in coulour-help wanted

Interesting concept! I really like the way you are trying to make peaceful, cooperative gameplay.

I have one question about the gameplay.

I may not be understanding the rules properly here, but it seems to me that the two players are cooperating, and the starting positions for pieces are always the same.

Doesn’t that make it more of a puzzle, rather than a game – if you had a tournament, as you suggest on the site, surely everyone would just play the known best solution to the problem?
I may be missing something here, though.

i managed to show things with the drag and drop system but need help with
-(1)how could i put the pieces straight on top of the carpets?
-(2)is there a way of counting the moves people make?
-(3)any help or idea to make it more interactive
-(4)i have seen somewhere a chess game in flash where the moves pieces made, where shown[/quote:9821c76def]

First off, let me say that the pieces and boards you’ve made are really beautiful… very nice :-)

It look like (and correct me if I’m wrong) you’ve tried to implement your game in HTML, on that website.
It would probably be possible to do it using raw HTML, and some server side logic, but that’s probably not the best way to do it.

As you suggest in 4) the best way to implement your game is probably using Flash, or Java.

This would make the problems 2&3 be solvable.

However, writing a game like chess isn’t an easy thing to do without some amount of previous computer programming knowledge.

So in summary, if you want to make an interactive version, you will probably need the help of someone with some computer programming skills.

It’s possible someone on the boards here would be willing to help, especially if you formulated the rules in a very easy to understand, and well specified, manner. (makes it easier to translate into a computer program).

Anyway, hope that answers your questions somewhat?

Haven’t had a chance to look at the site, might have a chance tomorrow in work.

Programming chess can be either really easy or really hard. If you’d like it to be a two player game, its very simple as there is a very strict rule set that can’t be deviated from. Simple to program.
If you’d like it to be played with the computer and having some reasonable level of AI, it gets very complex very quickly. Because of the number of possible moves every time, looking ahead several moves becomes highly complex . Don’t know enough about Flash myself to know how easy it would be implement it there, but doing it in Java or C++ wouldn’t be a problem.

Once you have logic running the game its trivial to place pieces, count number of moves, remember the moves and replay them or anything else you want.

If you want to implement a prototype, I would suggest modifying an existing program, and chucking out the AI code, rather than making a new one.

I looked briefly at http://jogre.sourceforge.net/main.htm
These guys have a chess implementation that appears to be quite cleanly written, and might make a good jumping off point.

It’d probably take very little time at all to hack the rules around and get the game working.

Of course, finding the quickest solution for the game is a different problem.

If you’d like it to be played with the computer and having some reasonable level of AI, it gets very complex very quickly. Because of the number of possible moves every time, looking ahead several moves becomes highly complex .[/quote:b90c5cdf0d]

But you aren’t _really_ writing a normal game AI*, as the two sides cooperate. You really are just looking for solutions to a puzzle, and given a few decent heuristics I guess you could prune the the search space massively.

The ‘AI’ does not have to adapt to unpredicted moves of the other player, which means it doesn’t need to do search at game time. I would think that the problem is thus closer to something like Knights Tour, than that of playing Chess well.

*I amn’t trying to start a debate on the semantics of AI, before anyone replies :-)

Implementing a non-AI solution first to play test the design should be done first, just to make sure everything seems to sit right. Low level ai wouldn’t be a big hit anyway. The ai on the chess game that put the link too is pretty basic, I managed to get check mate in 4 moves…

Er… I’m sure it’s just me getting confused here, but I don’t quite follow what you’re saying?

The ai on the chess game that put the link too is pretty basic, I managed to get check mate in 4 moves…[/quote:93700a83dd]
The AI may well be basic – but how is that relevant?
I wasn’t actually suggesting the AI code be used at all! :-) I was thinking that the other elements, such as the netcode and drawing code would be useful.

I would suggest modifying an existing program, and chucking out the AI code[/quote:93700a83dd]

The game under discussion isn’t actually chess.

You are talking a lot about chess AI in your posts. However, this game has several differences to chess, which I think make it a very different problem.

What do you think from your reading of the rules? Are you saying that the differences aren’t important, and that chess AI techniques are the way to go?

If so, I’d be curious to hear your opinion about what I was trying to get at, about it being more a puzzle than a game?

Amn’t I right in thinking that the game is basically the same as chess in that the pieces move in the same manner and each player takes it in turns to move ??
If so the ai process is going to be the exact same, the only difference is in what constitutes a good move. In this game thats going to me a lot easier to evaluate probably.

Or maybe i’m completely lost due to being in work on my holiday…

Amn’t I right in thinking that the game is basically the same as chess in that the pieces move in the same manner and each player takes it in turns to move ??[/quote:7482671a4a]

Well, yeah, whatever logic determines the list of acceptable moves will be the same, and the rules engine is largely the same. That’s why I suggested basing off an open source chess game.
But thats not _really_ the AI code – I least I wouldn’t say so. I guess you could look at it either way :)

If so the ai process is going to be the exact same, the only difference is in what constitutes a good move. [/quote:7482671a4a]

Well… yeah, but if what constitutes a good move is different, wouldn’t you say that makes the AI different too? Also, the fundamental fact that the players are _cooperating_ rather than competing changes things a lot, no? I mean, even from a game theory point of view, its a very different game?

Also, the change in the rules isn’t just one you can compensate for by altering some weights, and using the same AI code… (well, maybe in extreme cases, but I think it’s highly unlikely – correct me if im wrong)

If I’m not mistaken, what makes chess AIs good is not the way they brute force through search space. It’s the way they cleverly prune parts of the search space early.
Surely that bit of the AI would be totally different with the change of rules in this game?

One way of looking at the difference is this:
The number of permutations in chess are way too big to store some sort of lookup.
Therefore, the chess AI has to evaluate board states and future moves on the fly. (obviously it can recognise certain subpatterns on the board that it’s been trained to, but overall, it’s evaluating the board dynamically).

Your AI needs to search enough of the potential space that could result from it’s move in the couple of seconds or so that the player will wait. (It can’t precompute this, and store a lookup table, as that’s way too big). If the player then makes an ‘unexpected’ move, it has to recalculate a new strategy. So everything must be done quick enough to not keep the player waiting – this imposes strict computational limits.

However, the problem is very different when the AI controls both sides, and is trying to cooperate with itself, as appears is the case in this game.
The computation of the most effective strategy (that which resolves the game in the shortest sequence of moves) can all be done offline.
Granted, the search space is probably still (way) too big to brute force, but the point is that you are only looking for one good solution.
Any time you come across a solution that is longer than the shortest one you’ve seen before, you can just discard it, and discard that entire area of search.

Also, from what I read of the game, eliminating a pair of pieces from the board is _almost always a good thing to do_.

By using some heuristics (probably like the one above, or perhaps trying to eliminate first whatever pieces require the least moves to eliminate) you could possibly find very good solutions.

Thus, overall, the problem is much easier than chess.

As another reason for why I say that consider the following:
I have read that in chess, AIs frequently do really well in the opening stage of the game because they have prior knowledge of most clever openings, and best response to them.
Sometimes, when playing a chess AI, the best opening to use is one that’s completely unorthodox, as the AI does not have a best response hard coded in.

From the AI point of view, the enemy player doing the unexpected thing is sometimes very hard to deal with.
All those problems with disappear with a game like this, where you are playing against yourself… that’s why I say it’s more of a puzzle than a game, and more like knights tour than chess.

Although I’m sure many of the advanced techniques employed would be similar (they are both problems of search (although some people say that’s all AI is anyway :-)) I think the heuristics and special case logic used, which are very important to the success of a chess AI, are very different.

Oh well, maybe not! Excuse the ranting :-) I should state that I havn’t really thought about this, so I hope I’m not either stating the obvious or not making sense! :) (Do tell me if so)

Well… yeah, but if what constitutes a good move is different, wouldn’t you say that makes the AI different too?[/quote:6a7ebb9a28]
When I say the ai is basically the same I’m talking about the pruning, the only difference is the heuristic, which is a simple calculation. Cooperation doesn’t make a difference, its still I move one piece, other person moves one piece, I move one piece…. Code-wise, its making no difference.

I have read that in chess, AIs frequently do really well in the opening stage of the game because they have prior knowledge of most clever openings, and best response to them.[/quote:6a7ebb9a28]
There are also much fewer moves available at the beginning, so its easier to look further ahead.

Deedebee, how are you wanting to present this game, as a flash web game, java applet, stand alone application?

Hey deedebee,

I’d advise going for Flash, it’s a bit more designer friendly than Java ( you might even be able to get someone to design the core, and you could update it from there ).

You’ll find a wealth of people ( even students ) who would have enough Flash skills to put something together for you, for a prototype. There are also utilities out there that allow you to create a professional looking stand-alone app from your .swf file ( rather than using the Flash Player style app )

Regards…
Mal

i’m good at designing everything but would greatly appreciate if somebody could help me out with the programming wether in flash or java[/quote:728bbd43b8]

You will need a complete specification of the rules, if you want someone to help you build a prototype.

Some questions that might come up could include:

Are piece promotions allowed?
Is castling allowed?
What happens when Kings come into check? Is there still the idea of check in your game?

It would take some time to build the game, even with a concrete specification.

I don’t mean to be discouraging – it’s not a large task, but it can be hard to find someone that’s got the disposable time to voluntarily see such a project through.