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

        Arimaa >> General Discussion >> Should the repetition rule favor attacker?
        
(Message started by: clojure on Nov 3rd, 2010, 6:36pm)
Title: Should the repetition rule favor attacker?
Post by clojure on Nov 3rd, 2010, 6:36pm
First time in a game I encountered the repetition rule. I thought it would favor attacker, so I played my move twice but I got notification that third time was not allowed.

How do you see that situation in this game starting from move 51g: http://arimaa.com/arimaa/gameroom/opengamewin.cgi?client=1&gameid=159186&role=v&side=w

So we have position A which goes to B.
ABAB and then A is not allowed but if we want to favor attacker it should be that ABABA is allowed but after that B is not.

(Also my intuition seems to fail me yet again, since I lost from that position.)
Title: Re: Should the repetition rule favor attacker?
Post by ocmiente on Nov 3rd, 2010, 7:36pm
It generally favors the attacker, but it may be more accurate to say that it favors the 'first mover'.  In this case, bot_zombie was the first to move on 51s.  51g undid bot_zombie's move - I think... it can be pretty confusing, so I might be wrong about that :)

Click here (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi?board=talk;action=display;num=1258086245) for an older thread that discusses the repetition rule at some length.

I've been in the same situation where I've made a move and thought I had the right of way, but found out that I couldn't continue repeating my move as I expected.  In rare cases, the offending repeated position may have happened more than a few moves prior to the spot where you can't make a move.  

It's a good thing the computer keeps track of this for us :)
Title: Re: Should the repetition rule favor attacker?
Post by clojure on Nov 3rd, 2010, 7:51pm

on 11/03/10 at 19:36:45, ocmiente wrote:
but it may be more accurate to say that it favors the 'first mover'.


Yeah I generally meant by attacking "first mover". But I'm still confused.


Quote:
 In this case, bot_zombie was the first to move on 51s.  51g undid bot_zombie's move - I think...


51g put the game into unique position which 51s tried to prevent. So I see the situation as gold was the "first mover" and actually I don't understand what you mean! :)

I looked the old thread but it didn't help yet. Either I'm tired or the rule is counter-intuitive to me, personally. Sometimes I get fixed idea on something and it's very hard to see in different light.


Quote:
It's a good thing the computer keeps track of this for us :)


It's good but it shouldn't be a requirement. :)

edit: The question I'm after is why would the current rule favor in general the attacker / first mover but not in this case? And why is the current rule best of the different options?

Can someone give example where an above mentioned rule that gives advantage to the player who creates the unique position be worse than current rule? i.e. my intuition says ABABA is permitted but ABABAB is not (where first A is unique position).
Title: Re: Should the repetition rule favor attacker?Your
Post by ocmiente on Nov 3rd, 2010, 8:20pm
If you open the game, and click on move 51s, then click on move 52s, you will see no difference in the board position.  Your move 51g only undid silver's move 51s.  So, in this case, Silver was the first mover/attacker.  

The same goes for Adanac vs. bot_marwin (http://arimaa.com/arimaa/games/jsShowGame.cgi?gid=141408&s=b), move 42s.  If you click on 42s and 43s, the board positions are identical.  Marwin's move 42g undid Adanac's move.  In this case Silver was the first mover/attacker.  

<EDIT: should have written, "Your move 52g only undid silver's move..." and in Adanac vs. marwin is was move 43g, not 42g that undid the move>
Title: Re: Should the repetition rule favor attacker?
Post by clojure on Nov 3rd, 2010, 8:31pm
How could my move 51g "undo" move that comes after, namely 51s?

If I click on 51s on the list, it means the position that comes after "my move" 51g. You are describing exactly what bothers me but you are seeing it as if silver was first mover.

I probably need to sleep over this :) Thanks for trying to explain it to me.
Title: Re: Should the repetition rule favor attacker?
Post by ocmiente on Nov 3rd, 2010, 8:35pm
52g was the move that undid Silver's move, not 51g.  I made a mistake in my earlier post - sorry.

It took me a little while to grok this too.  If my experience is any indicator, this will become very clear to you, very soon.
Title: Re: Should the repetition rule favor attacker?
Post by clojure on Nov 3rd, 2010, 8:42pm

on 11/03/10 at 20:35:56, ocmiente wrote:
52g was the move that undid Silver's move, not 51g.  


I was referring to what you said earlier:

Quote:
Your move 51g only undid silver's move 51s.



Quote:
It took me a little while to grok this too.  If my experience is any indicator, this will become very clear to you, very soon.


I probably will give up before that since the Adanac game worked as if the rule I suggested was used (though only ABAC not ABABAC). i.e. Adanac created first unique position from which the repetition started, and it gave him the advantage.

edit: In my game, the silver is first one to trying to break repetition, not create it. Gold would be first one to repeat it third time if he would be allowed.
Title: Re: Should the repetition rule favor attacker?
Post by novacat on Nov 3rd, 2010, 8:53pm

on 11/03/10 at 19:51:00, clojure wrote:
51g put the game into unique position which 51s tried to prevent.

51s is also a unique move.  Your elephant is in a different position than 50s.  It feels the same because you are still stuck with a cat in your way and a rabbit on a trap.  Thus, 52g is the first repeat.


Quote:
edit: The question I'm after is why would the current rule favor in general the attacker / first mover but not in this case?


The rule favors the last person to make a move that leaves the board in a unique position.  This is often the attacker, but not always.


Quote:
Can someone give example where an above mentioned rule that gives advantage to the player who creates the unique position be worse than current rule? i.e. my intuition says ABABA is permitted but ABABAB is not (where first A is unique position).


If you allow ABABA, it just reverses who wins.  It makes the strategy different, but the concept is the same.

Some advice for your game:
It is easier to get to goal by removing pieces than trying to push a rabbit through.  From move 50g to 53g you can kill the silver cat.
Title: Re: Should the repetition rule favor attacker?
Post by ocmiente on Nov 3rd, 2010, 8:53pm

on 11/03/10 at 20:42:40, clojure wrote:
I was referring to what you said earlier:


Yes, sorry.  Please see apologies and edits above.  I left the original text to that the thread doesn't get totally confusing (I hope).  

If you go through the games and click on the first move, then the third, then the second then the fourth (i.e. two steps forward, one step back, e.g. 1g, 2g, 1s, 2s, 2g, 3g, 2s, 3s, ...) you can easily see when the first repeated position happens.  The board won't change between your clicks.  

In your game, this is between moves 51s and 52s.  In Adanac's game, between moves 42s and 43s.  Note that Gold actually created the first unique position, not silver.  This position that could not be repeated three times was created in move 51g in your game, 42g in Adanac's.  One critical thing to understand that it has to do with position, not with movement.  In fact, repeated positions can be created by completely different moves.  

The illegal part is that a player cannot move so that the position and player to move repeats a third time.  Undoing a move has the effect of causing the position and player to move to repeat.  This is the case in moves 52g in your game, and 43g in Adanac's game.  This caused a second repetition.  The computer prevented you from making a third. 

I think I've beaten this horse enough for now... I'll check back on this thread tomorrow.  Good luck figuring that out!
Title: Re: Should the repetition rule favor attacker?
Post by clojure on Nov 3rd, 2010, 8:59pm

on 11/03/10 at 20:53:02, novacat wrote:
51s is also a unique move.  Your elephant is in a different position than 50s.  It feels the same because you are still stuck with a cat in your way and a rabbit on a trap.  Thus, 52g is the first repeat.


Yes. 52g is the first repeat and this is what I meant. Gold is first to create the repetition, thus the one who initiated it, and should be given the advantage, in my naive view.


Quote:
The rule favors the last person to make a move that leaves the board in a unique position.  This is often the attacker, but not always.


I agree what you say but I don't see concretely why the current rule favors attacker more often my suggestion.

Thanks for the playing advice. I thought that 3 rabbits were enough in that position and I was banging my head against the wall. I'm not good at fast games so I just tried to stumble upon the right tactic...
Title: Re: Should the repetition rule favor attacker?
Post by clojure on Nov 3rd, 2010, 9:15pm

on 11/03/10 at 20:53:43, ocmiente wrote:
Yes, sorry.  Please see apologies and edits above.  I left the original text to that the thread doesn't get totally confusing (I hope).


No problem :)


Quote:
Note that Gold actually created the first unique position, not silver.  One critical thing to understand that it has to do with position, not with movement.  In fact, repeated positions can be created by completely different moves.


Yep. I was thinking in positions.


Quote:
I think I've beaten this horse enough for now... I'll check back on this thread tomorrow.  Good luck figuring that out!


Thanks for the beat! Now I realize that there was only one problem. To me it felt as if player who starts a repetition cycle should be given advantage, and the current rule is favoring the guy who leaves the game into unique position.

So my confusion arises from the idea that if player makes a unique position, he should be able to maintain it if he can and the current rule is saying that you shouldn't be able to maintain it. How does this effect the play in concrete terms, I'm not sure, with only these two examples.
Title: Re: Should the repetition rule favor attacker?
Post by 722caasi on Nov 3rd, 2010, 10:16pm
In this case, the last player to create a new position (the bot, after move 51s) forced the opponent to deviate by maintaining the position. The rule favors the last player to create a unique position. That player will never be forced to deviate, and indeed, the bot was not forced to deviate.
Title: Re: Should the repetition rule favor attacker?
Post by novacat on Nov 3rd, 2010, 11:27pm

on 11/03/10 at 20:59:31, clojure wrote:
I agree what you say but I don't see concretely why the current rule favors attacker more often my suggestion.


I would say it favors the attacker because the attacker usually sets up the situation.  For example, if you had used your last step of 49g to put the elephant on d6 instead of c5 (and the opponent responded with the same 49s), you would have won the repetition rule.  The defender must therefore avoid the repetition challenge or lose.

If you prefer ABABA, putting the elephant on c5 was the correct move.  Under either system  the attacker simply needs to set up correctly.  I don't know if there is an advantage with ABAB versus ABABA.  (ABAB seems easier to me, but I have no proof.)
Title: Re: Should the repetition rule favor attacker?
Post by rbarreira on Nov 4th, 2010, 4:02am
I think this is the rule of thumb: do not undo your opponent's move, unless he has already done so.
Title: Re: Should the repetition rule favor attacker?
Post by omar on Nov 5th, 2010, 2:45pm
Sometimes it is hard to define who is the attacker. But in terms of coding all we are doing is looking through the history of positions (which also includes side to move) and making sure that the move does not put the game into a position that has already occurred before at least two times.


Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.