wu :: forums - Print Page


    
      
        wu :: forums
        (http://www.ocf.berkeley.edu/~wwu/cgi-bin/yabb/YaBB.cgi)
      

        riddles >> cs >> Get revenge on your computer
        
(Message started by: Scott on Sep 21^st, 2004, 11:23pm)

Title: Get revenge on your computer
Post by Scott on Sep 21^st, 2004, 11:23pm

When was the last time you won a chess game against your computer. Probably never. Now you might have a chance. A new game called Arimaa shows that computers still have a long way to go before they can match human creativity and strategic planning. The game is played with a standard chess set, but using very different rules. Unlike chess the game has virtually no chance of draws and is very exciting to watch, because a material advantage does not always result in a victory; strategic positioning matters much more. The games inventor is offering a reward of $10,000 for anyone who can produce a program to defeat a top ranked player before the year 2020. More information about the game can be found at:

http://arimaa.com/

Title: Re: Get revenge on your computer
Post by towr on Sep 22^nd, 2004, 2:59am

There's plenty of other games computers are ill-equiped to play.. They don't fare well against good go players either, for example. A high branching factor helps..

Title: Re: Get revenge on your computer
Post by John_Gaughan on Sep 22^nd, 2004, 6:55am

Is that because computers are ill-equipped in theory or in practice? For example, another thread in one of the forums here talked about Go programs. When restricted to a smaller board, computers fared well, but lacked the physical time and space to play well on a normal sized board.

Given the sheer number of possibilities, and that each successively larger board size increases the number of options, this looks like an NP problem. That makes sense. NP does not mean impossible, clearly such problems are quite possible to solve satisfactorily with a sufficiently small data set (or board size), while practically impossible on larger data sets.

I think the solution to a problem such as the one the OP presented will not be a brute-force, "try every solution, rank them, take the best one" style of solution, but one that uses other aspects of computer science or even computer engineering to solve. Solutions that involve creative applications of other technology, major jumps in computer science, not minor advancements. Thinking outside the box.

These are the ideas that get me excited. This is the reason I study computer science. Not to implement a karnaugh map or binary tree, but to apply these concepts to develop new AI, more efficient algorithms, and ultimately, a computer game where the bad guys aren't stupid.