HAL doubles the stakes

Here's an interesting article about the development of computer programs to play backgammon. It's interesting that while the best chess playing programs utilize "brute force" calculation, world class backgammon programs use neural nets and "...are entirely self taught using the technique of reinforcement learning...".