[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: computer-go: new strategy game with $10,000 prize




His goal was to make a game played with chess pieces on a chess board
that was very different from chess and still fun to play. I've tried a few games,
and I think he succeeded in creating an enjoyable game. The game certainly is
less tactical than chess. Because each side has 4 moves in a row, it's very hard
to set up any tactical sequence. It's always too easy for the opponent to escape.

I haven't played enough to tell how much strategy there is. All I do it try to
use my elephant to control one of the advanced traps, and move up pawns on that side
of the board.

David

At 05:04 PM 1/21/2003 -0800, Bob Myers wrote:
From: "David Fotland" <fotland@xxxxxxxxxxxxxxxxx>
> http://arimaa.com/arimaa/
>
> He thinks his game is 100x more complex than go....

Begging the question of what the meaning of complexity is, of course.  I
don't doubt that the total number of possible Arimaa games, or some similar
metric, is several orders of magnitude greater than for chess.  But it is
highly misleading to equate such a metric with complexity.  A
counter-example of a game with a huge branching factor but obviously low
complexity level would be trivial to construct.  And of course what is
"complex" for human beings may not be "complex" for computers, and vice
versa.

My unsubstantiated hunch is that this game might actually be substantially
easier to program a computer to win at than chess is.  The $10,000 prize
should take more like three years to win than eighteen, if anybody bothers,
that is.

My sense is that fellow that invented this game, with all due respect,
actually has virtually no understanding of what makes a game hard to solve.
He doesn't seem to even understand the way chess programs work, apparently
believing that they enumerate all positions.  If all he has really done is
added three orders of magnitude to the size of the game tree, then we can
easily ask what the strength is of a chess program which does 1/1000 of the
computation done by the current strongest programs.  Certainly it is not
only capable of beating an "average human" but in fact very strong humans.

Having freedom to place your pieces at the beginning of the game as Arimaa
allows certainly changes the nature of the opening book, but hardly makes it
irrelevant.  Certain categories of starting positions will be identified as
most favorable, and standard openings will grow up around them.  And of
course Go already has this characteristic -- with the freedom to place your
pieces anywhere you like starting at the first move, which doesn't seem to
have prevented the development of thousands of joseki.

His claim that Arimaa is more positional and less tactical than chess seem
impossible to verify, but as another poster pointed out, it seems unlikely
given the 8x8 size of the board, the different types of pieces, etc.  It is
hard to imagine a game more positional than Go with its large board size and
homogenous playing pieces.

--
Bob Myers