Since our implementation of Dice of Doom will include an AI player, we’re going to start with an extremely humble size for our game board. As you probably know, AI code can be very computationally intensive. In our early, very naive version of this game, any board larger than a 2-by-2 grid of hexagons would bring CLISP to its knees!

Here is a complete game, played on a puny 2-by-2 board:

At the beginning of the game, player A (indicated with black hexagons) possesses the top two hexagons, with three dice on each. Player B occupies the bottom row (indicated by the white hexagons), with three dice and one die, respectively. Player A attacks the lone die with one of his piles. After the attack, one of player A’s dice remains behind, while the others move to the conquered spot. Then player A passes the turn.

Player B now attacks player A’s two dice with a pile of three. Player B then passes. At this point, player B receives a single reinforcement die on her left hexagon. This is because she killed two of player A’s dice. The reinforcements, as per the rules, consist of the number of dice killed, minus one.

Player A now attacks with three of his dice and passes. Also, he gets a reinforcement die.

Player B now has only one legal move, attacking two against one.

Player A now has the upper hand, killing all of player B’s remaining dice. As you can see, player A is permitted to perform multiple attacks on his turn before passing. The game has ended with player A as the winner.