Arimaa Forum - Print Page


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

        Arimaa >> General Discussion >> random thoughts on arimaa from a relative newbie
        
(Message started by: ginrunner on Nov 28^th, 2010, 12:04pm)

Title: random thoughts on arimaa from a relative newbie
Post by ginrunner on Nov 28^th, 2010, 12:04pm

Let me preface this post with a few things. 1) As far as arimaa skill goes I am not as good as any of the elite players, but ive only been playing for a few months now. If there is something wrong in my post I would not be offended but rather enjoy hearing a different and probably more experienced point of view 2) In strategy and tactics games I generally favor an offensive play style unless playing defense has a definite advantage. 3) My favorite online strategy board game was tacticsarena.com where I was one of the top players until time got the better of it.

Now onto my random thoughts on arimaa.

Arimaa is a great game but I feel it may suffer a similar fate as tacticsarena suffered. Eventually the games will become almost clockwork as there will almost definitely be a "proper" gold setup and more than likely a "proper" silver setup. I will use tacticsarena as my example in this case. Arimaa has 16 pieces to arrange however one wishes on 16 spaces. Tacticsarena had many more options as far as a setup is concerned and much more complex systems of movement and attacking and yet near what I consider the end of tacticsarena there became more or less a consistent setup and play style. Once people had the setup made they also played a consistent set of opening moves depending on whatever of the limited scenarios they encountered from the start. The same thing happens in chess to a much lesser extent (white almost always opens with the same thing). I feel that in chess the reason the games don't become stale is because it is hard to gain a huge tactical advantage within 2 opening moves (I am not the best chess player either) and so each side has a bit more time to develop an interesting and unexpected attack or defense. In arimaa a player can move the gold elephant all the way down to the opposing silver line and must be accounted for to an extent which is a tactical advantage of gold (note that this may not be the best available move but it does make my point I feel). Mark my words, within a few years (if it hasn't happened already), there will be discussions based solely on setups and eventually the "best" setup will emerge and have the fewest initial tactical weaknesses AND best available opening moves. Then, with the best setups found the opening moves will become a stale clockwork. The middle game and end game will still be unique to an extent as long as both players are relatively equal skill and one player doesn't gain a game winning advantage from the opening moves.

Next I feel that the eventual dominant play style will be what I call a slow and solid offense rather than any kind of defense or a bomb style offense. I consider a bomb as rushing pieces forward from the start (swarming but way to early) in hopes of gaining a positional advantage through smothering. This style usually relies upon the opponent being caught off guard and making blunders; or relying on luck to a degree which in a game of skill such as arimaa I don't feel a bomb is possible. Second, a defensive play style shouldn't work in the long run if we look at how one wins. Either a player rabbit goals or takes the last rabbit. Taking rabbits is neither offensive or defensive in nature but rabbit goaling is completely offensive. If a player plays completely as a home player it allows the opponent to slowly and safely move rabbits up and smother the home player and create to many goal threats to be dealt with.
Currently, I use a tactic which I call "flip and feed." I've seen it used before and I am sure there is another name for it. Basically it involves me being a home player, flipping low level pieces to my horses and dogs and moving those into my home traps. I know the weaknesses of it (I have 4 bots remaining on the ladder and they take advantage of it lol) and have been trying to be more offensive and swarm other traps but have yet to master the new strategy. I want to buy Fritzlein's book but I can't justify being a college student and buying a book on a game when I have trouble buying textbooks lol.

Thank you for reading, these are just some initial thoughts and possibly things to watch out for in the future.

Title: Re: random thoughts on arimaa from a relative newb
Post by rabbits on Nov 28^th, 2010, 2:19pm

Thanks for sharing your thoughts Ginrunner!

I'm skeptical about your claim that a single setup will become the de facto best. I have a favorite setup, but Megajester called it flamboyant during the online festival :) There are discussions about whether the balanced (99of9) or the unbalanaced (mine, for example) setups are better. But that question is completely up in the air, and there are hundreds of good variants of each of these two categories of setups.

I doubt that one setup will be played by everyone for a few reasons:
* It's fun to have a signature setup. I take pride in my setup and what it can do :)
* Using nonstandard setups adds an element of surprise that the opponent may not deal with properly.
* Small changes in a setup cause unpredictable differences in middle-games, and so it would be very difficult to argue that one is better or worse than the other.

Of course, you may be right. In that case, I will be happy to play Arimaa until a "best" setup is recognized... It'd be fun to play test it and maybe even "prove" that it's the best. And, I would probably be happy to play Arimaa even after the best setup is found. As you mention, Chess has standard openings, but a lot of people still love playing that game :)

There has been a recent shift towards more attacking playing styles, but I don't know that I would be comfortable saying that any one strategy is the best. Too much depends on what your opponent does... Swarming is good sometimes and bad other times... Framing a piece is good sometimes and bad other times... etc.

I bought Fritzlein's book as a college student. (Now I'm a graduate student.) I would recommend it. I mean, if you have the motivation to read it, then it's probably worth buying. (I, too, struggle to have proper motivation to read my textbooks.....) Chessandgo is working on a book too, by the way :)

Title: Re: random thoughts on arimaa from a relative newb
Post by Fritzlein on Nov 29^th, 2010, 8:56am

Welcome, ginrunner. Thanks for sharing your thoughts.

I agree that there is a possibility that Arimaa openings will become stereotyped and stale, but I strongly disagree that it is inevitable as you assert. This situation is rather more subtle than you are giving it credit for.

The key factor in the freshness and unpredictability of Arimaa so far has very little to do with the branching factor. As you point out in the case of tacticsarena, a game can have a huge branching factor and nevertheless become stale. Instead, the critical point is that strategic understanding of Arimaa is still deepening. We have already had a couple of revolutions in Arimaa opening theory; these were not caused by someone slapping his forehead and saying, "I never thought of setting up my pieces this way before," but rather by a changing understanding of middlegame positions, such as, "I didn't realize advancing a horse up the flank would be good."

As of today, eight years into on-line play of Arimaa, there is still no obvious way to gain an advantage in the opening. Capturing a piece would be an obvious advantage, but we have discovered no way to force an early capture that the other player can't prevent. You mention a "flip-and-feed" strategy, but the opponent can easily avoid losing any pieces to this sort of attack. It's not so easy.

Given that there is no way to force quick captures, we have to rely on other positional landmarks for who is winning. The exciting thing is that we aren't even sure yet whether you are right or wrong that a slow, rolling advance is best. Conventional wisdom for many years has been that you are wrong, and that if you voluntarily advance pieces as you are suggesting, they will become hostages, which ties up your position and eventually leads to your losing material in one opposing trap or the other. However, that is not a foregone conclusion, and I caution you against betting heavily that you have hit on what will be the dominant style of play among future Arimaa grandmasters.

In 2005, new players like robinson upset the apple cart by showing that it could be good to give up a horse hostage, and sometimes even good to give up a camel hostage. We didn't know that before. Obviously that changed the way openings were played. But it didn't solve the question of what the "right" opening is, because sometimes giving up a camel hostage is still bad.

In 2009, it started to become clear that EHH could frame an opposing H without the help of the friendly camel. This middlegame insight suddenly made an unbalanced opening look attractive. That again radically changed opening theory, and we are still working out the consequences.

For you to say that opening theory will settle down on a single setup and/or settle down into a single dominant style of play entails that we stop discovering new middlegame strategies. I admit that your prediction is one possible future, because we might have already discovered all of Arimaa strategy that there is to know. I think, however, that that is an unlikely future. My belief is that some opening positions which we think of as favorable today will in the future be understood to be unfavorable. When that deeper understanding arrives, then popular Arimaa setups will again undergo a transformation.

In short, the argument isn't really about Arimaa openings, it is about how deep Arimaa is strategically. Your claims amount to saying that Arimaa is not very deep, or at least not much deeper than we have already discovered. I beg to differ. :)

Title: Re: random thoughts on arimaa from a relative newb
Post by ginrunner on Nov 30^th, 2010, 5:55pm

Thank you guys for the comments. Honestly I think Arimaa is about as deep as chess but in a different way. The reason for my thoughts on a best setup are because even though you can come up with something flashy or a signature setup there will always be a fall back setup that can be deemed the "best." Back to tacticsarena, the "best" setup changed many times and even went through phases as people learned more about the opening, middle, and end game. The final "best" setup would have been laughed at in earlier phases and discounted. (I was known for having signature setups as well and would log on under a different name only to be told I was using "imaginations rush" or something but the vast majority used the same thing or a slight variation). I would rank tacticsarena just as deep as chess and arimaa. When there is an option to change how one starts and once the grandmasters gain a basic knowledge of all or most of the plausible strategies available there will be a setup found that has the least amount of starting weaknesses.

The reason for my idea on stale openings is because of the ability to gain such an advantage positionally that your opponent has to react immediately. Yes, gaining a hostage or even capture immediately is extremely attractive but isn't necessary in a game that is based so heavily on positioning. If someone can make their opponent react in a predictable and almost planned way based on what is deemed as the most logical way out of a situation then openings do become stale and repetitive (and not even on the first 2 or 3 moves but maybe up to the first 10 or so). This is also another reason for my bet on a slow rolling offense, it doesn't require the taking of any pieces.

I absolutely love developing games such as arimaa and I think I will be here for the long haul as I am already addicted. My prediction for the final best strategy could be (and may probably be) completely wrong but I have seen merits for it versus some of the top bots and would have had some easy wins had I had more experience with certain situations. I had what I considered a perfect advantage with a full board block with my all of my rabbits 1,2, and 3 ranks away from goaling before I made blunders and let an elephant or camel become the strongest free piece behind my line in a few games but have yet to actually win with it. I will definitely keep on playing and hope to see arimaa become a very popular game.

Title: Re: random thoughts on arimaa from a relative newb
Post by Fritzlein on Nov 30^th, 2010, 6:07pm

If you keep on playing Arimaa long enough to be proven wrong as many times as I have been proven wrong, you will be here quite a while. :)

Title: Re: random thoughts on arimaa from a relative newb
Post by rozencrantz on Nov 30^th, 2010, 7:33pm

I think the biggest thing Arimaa has taught me is just how wrong I can be. A lot of what you're saying is similar to things I've said about games, including Arimaa, and the Great Camel just sort of snickered and said "you think so?"

And I've only been here a few months.

I will say one thing about openings though. Laugh if you like, but the game Street Fighter is a fairly deep and sophisticated game that some people have taken as far as they can. Opening theory in SF is pretty simple, and fairly well understood: for any opening, it's known which openings win what % of the time.

And the biggest thing I learned from that game is: it doesn't matter. It's well known that "T. Hawk" is by far the worst opening in the version I play, and people still win with it. Sometimes it puts you at an 80% disadvantage, but that just means you have to be 80% better than your opponent. And playing the opening you are most comfortable with is a far more sound strategy than playing the opening that is statistically superior.

Arimaa has far more different openings, and is "shotgun balanced," offering a huge variety of openings rather than a finely tuned set of 10 or so. My guess is that there are unwinnable openings (all rabbits forward, perhaps?). But I will also be so bold as to guess there are at least 6 openings that you could use exclusively to win a tournament, if you stepped up your game that much.

But I've only been playing a few months, so I hope you have some salt handy.

Title: Re: random thoughts on arimaa from a relative newb
Post by ocmiente on Dec 1^st, 2010, 12:50pm

on 11/30/10 at 19:33:17, rozencrantz wrote:

My guess is that there are unwinnable openings (all rabbits forward, perhaps?).

Probably unwinnable for most of us, but not Fritzlein: Fritzlein vs. Bot_Bomb2005CC (http://arimaa.com/arimaa/games/jsShowGame.cgi?gid=17405&s=w)

Title: Re: random thoughts on arimaa from a relative newb
Post by omar on Dec 2^nd, 2010, 3:33pm

Quote:

Eventually the games will become almost clockwork as there will almost definitely be a "proper" gold setup and more than likely a "proper" silver setup.

I have to agree with you on this in an absolute sense because technically there is always a best move for any given position; even for the first move. But weather we will discover what that is without having access to the full game tree will depend on how computationally reducible Arimaa is. That is, can we find the complete set of heuristics that will allow us to stay on the winning line without have a lookup table. We are starting to discover these heuristics (capture an enemy piece, take camel hostage, don't start with elephant in the back row, etc), but they tend to be riddled with exceptions; and exceptions to exceptions and so on; creating layers of complexity. Sometimes you are better off just building a lookup table than trying to encode the heuristics (as with chess end games). But getting back to the possibility of discovering the "proper" setup. It would basically mean that one setup gives such a noticeable advantage over all other setups that we can't help but have to use it. In actuality though I think the differences in advantage between a large number of setups is very subtle and would require an incredibly high level of play to distinguish it. So even though there is a "proper" setup, recognizing it might not be easy. But you never know until you try :-)