In another, more recent thread I've posted some interesting statistics about how Arimaa differs from chess in terms of the usual results of a game.
I think it is debatable wether you like or dislike the possibility to draw in chess.
It is a very peacable result for both sides afterall.
I still slightly favour the drawless nature of Arimaa, since this makes the game more and more tense towards the end.
But where I feel its clearest advantage lies is in the much better color-balancing.
The first-move-advantage of one player is counteracted with the second-setup-advantage of the other. And so far statistics demonstrate that neither side can claim an advantage distinctive enough to actually make a difference.
The by far biggest disadvantage for Arimaa is that it is known by so few who would potentially enjoy it. Thus me reviving ancient threads so that one or the other might become intrigued enough to check it out. ^^
Oh, and the 10,000$ price for writing an AI to beat a selection of defenders of mankind still has not been claimed by anyone.
When I was playing a game of chess with asmael, I noticed a reference in his blog to Arimaa. Arimaa is a game that was created by a computer scientist and by design it is difficult for computers to play well. Arimaa has a high branching factor (makes brute force search difficult) , has a randomized setup (makes opening book less effective), and games have relatively few captures (makes endgame tables less relevant). I've not played Arimaa yet, but I love the theory behind it. Incidentally it's playable with a standard chess board/set.
Here's a pointer to the wikipedia entry:
http://en.wikipedia.org/wiki/Arimaa
The official website has some cool papers if you want to study the theory:
http://arimaa.com/arimaa/
Kudos to asmael for pointing this out.... he's gone now so I'm trying to keep the fire alive for arimaa.