Jonathan Palfrey -> RE: AI question (9/6/2004 2:35:13 PM)
|
I think only Frank Hunter could explain the game's AI properly. As a player, I've found that the game makes a weak opponent, suitable only for learning the mechanics. When you know how to play, then you try for real opposition: playing a human by e-mail. Example: in one of my games against the program, I was playing the French, and sent one corps sneaking along the bottom of the map. It was never intercepted, and reached Vienna without encountering any opposition. The Austrians concentrated their forces in the centre; they outnumbered me but somehow managed to lose the fight in the centre too. I don't think I've lost a game yet, though I've had one or two draws. The scoring system seems rather bizarre and could maybe do with re-examination. I think the trouble is that it counts score each turn and adds everything up at the end; it would be more conventional and intuitive to look only at the final result and score that. For instance, if we have a big struggle in the centre of the map, which ends up with the enemy completely defeated, shattered, and running away, this can be counted as a draw or a marginal victory, because not enough progress was made in the early stages. Maybe there is some historical justification for this, but I'm sceptical. To put this in proportion, I'm talking here only about the AI and the scoring system; the game itself seems to work fine. However, if you're looking for good AI, I'm afraid you won't find it here. Good AI must be very hard to do without cheating; and Frank doesn't believe in cheating. Neither do I, so I have sympathy for him on this.
|
|
|
|