Well, I'm out of town now, but as soon as I get back.

New version still cheats, but good news is so do I. I let the computer take a line and after a few iagonal plays I got those too.

Anyway, my basic algo is based on a 8x8 board, each one has a "value", with center being zero and corners being 100. This allows prioritising play. This is level one.

At level two, game tries to play all his valid moves, then all my next moves. Once this is done, it sorts solutions by the most advatageous move. E.g., if a play of his will open me to take a corner, that solution get him -100, because it loses the hundred peg.

This allows it to not play moves that allow me to get corners, unless it's the last move.

The higher the level, the more it tries to predict.

With enough heuristcs, it will become a war of limitation of damage. As a result, and ideal player will exhaust my safe moves before I exhaust his, evntually forcing me to play a 22 27 72 or 77, meaning he gets corners.

Because of value grid, it will first try to get a foothold on lines, the semicorners, then corners. A few exceptions apply, such as black 13 14, white 15, his move. This should be a 99 move for 12, not because of position, but because it kicks me off the line.

Another exception is black 13 15, if it can it should take 14. Because if I ever play up to 12 or 17, it's his corners and his line.

Once it predicts the whole game and has a complete list of exception, I'll pit it against my algo and, if it beats that, me.

Standard N900 post disclamer.