Arimaa Forum - Print Page


    
      
        Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> General Discussion >> No Repetition Rule
        
(Message started by: appleface on Apr 18^th, 2013, 2:04pm)

Title: No Repetition Rule
Post by appleface on Apr 18^th, 2013, 2:04pm

Disclaimer: I made this thread to talk about my asthetic issues with the rules, I do not believe this affects game balance in any significant way.

I dislike the way the curret anti-repetition rule is implemented. My point of referrence here is Go. Go also has the anti-repetition rule, but it forbids any previous game state, not just those that have been repeated twice. The repetition requirement feels like an unnecessary hold over from chess to me.

My reasoning goes like this: imagine there was a situation where repetition was the best option for both players. Under the current rules the game state would be repeated three times, then be forced down a different path. Why not skip the repetition and head down a different path immediately?

The strict no repetitions rule Go used feels cleaner for the same outcome (in so far as omniscient players would arrive at the same outcome). I would prefer to play the game that way.

Thoughts?

Title: Re: No Repetition Rule
Post by Hippo on Apr 18^th, 2013, 5:48pm

on 04/18/13 at 14:04:01, appleface wrote:

My bot uses no repetition rule and automatically punishes an opponent for not doing the same ... by repeating repeated ... but as a player I like the option to gain the extra one move time ...

Title: Re: No Repetition Rule
Post by aaaa on Apr 18^th, 2013, 8:51pm

A second occurrence of a board position with the same side to move was allowed as a theoretical courtesy to whoever would eventually be forced to deviate, namely to be able to choose where in the cycle to deviate after actually being confronted with the first repetition.

Title: Re: No Repetition Rule
Post by mattj256 on Apr 19^th, 2013, 12:06am

Since we're talking about aesthetics...
Let's say Gold and Silver both set up with all their rabbits forward:

http://i.imgur.com/D5h6odq.png

On subsequent moves, each player pushes two rabbits forwards two squares:

http://i.imgur.com/VTmLgrb.png

This is a theoretical question, not a practical one.
Any two rational humans would agree to a draw here because it's totally obvious that neither player can make progress.
What actually happens is:
(1) Both players end up making unproductive moves.
(2) Both players have to keep track of which positions have already happened, which after a while becomes impossible.
(3) Eventually one player is forced to repeat the same position for the third time and he loses by immobilization.

This is a theoretical problem, not a practical one, but I think it highlights a specific way in which the rules don't give a "fair" outcome...

Title: Re: No Repetition Rule
Post by browni3141 on Apr 19^th, 2013, 3:14am

on 04/19/13 at 00:06:50, mattj256 wrote:

Actually the game would end on score, unless there were no game time limit set, but it's the same problem of "the rules don't give a fair outcome." Silver would win because omar arbitrarily chose that silver would win the tie-breaker situation of both sides having equal number of pieces.
Most people don't care about this problem because it is so unlikely to happen. I don't believe that it has ever happened in all the games of Arimaa played where both sides were trying to win, but it has come close to happening on at least one occasion. In one of my own games the position got almost completely blockaded. I could make a break (which I didn't realize, and the win wouldn't have been trivial anyway), but my bot opponent couldn't. We shuffled until the bot made an illegal move. The game probably should have been drawn because I didn't see the way to win.

I've made a thread trying to replace the score function.

Title: Re: No Repetition Rule
Post by mattj256 on Apr 19^th, 2013, 7:48pm

on 04/19/13 at 03:14:03, browni3141 wrote:

Actually the game would end on score, unless there were no game time limit set, but it's the same problem of "the rules don't give a fair outcome."

You're right; I'm glad my "theoretical" post has been the seed for a more practical discussion. :)

These are two other "theoretical" positions.
http://i.imgur.com/wJcXGvu.png
http://i.imgur.com/Tpip7QW.png

My original example is a rock-solid draw. In these two examples the position is quite drawish but both players have the option of sacrificing a rabbit to open up the board. (In the first example, Silver can't push anything through the gap because it will be captured immediately. If Silver puts his/her elephant on f4 it prevents Gold from making any progress as well.)

Title: Re: No Repetition Rule
Post by chessandgo on Apr 20^th, 2013, 3:38am

I would agree that a no-repetiton rule would be better than a no 3rd timer rule.

For the last 2 examples, they are clearly not drawish, people will capture rabbits and play on (but I think Boo adressed that in the sister thread).

Title: Re: No Repetition Rule
Post by novacat on Apr 20^th, 2013, 6:55am

Immediate repetition is one thing, long term repetition is another.
I agree allowing immediate repetition is unnecessary, but I feel if several moves are played and the position returns to a previous point, it should be allowed so that the player may choose to deviate at any point in the series of moves before a third repetition.
It is a lot to ask a player to see ahead several moves and recognize the possibility of a repeated position.

Go does not have this problem as the board will always be different if you do not repeat the position immediately.

Title: Re: No Repetition Rule
Post by Janzert on Apr 20^th, 2013, 8:19am

on 04/20/13 at 06:55:56, novacat wrote:

Go does not have this problem as the board will always be different if you do not repeat the position immediately.

Go repetition can be arbitrarily distant although a bit less likely to be than Arimaa it seems.

Personally I prefer the current Arimaa rule.

Janzert