|
Advanced AI in MUDs
The opening chapters of Mat Buckland's excellent book Programming Game AI By Example implements a state machine for a miner and his wife in a mud like world. The miner goes and gets gold, getting a little more thirsty each time, when he's thirsty he goes to the bar and at the end of the day he comes home and sleeps. His wife cooks, cleans and, well, pees. The source code and binaries for his book can be found here:
His idea here is quite simple, yet the results are quite convincing, especially in a text based world where animations and sounds don't get in the way of the simulation. Yet in MUDs, the behaviour of NPCs has been nothing but wandering around a fixed set of rooms. Of course, for the most part MUDs focus on player to player interaction, but I think that strong AI would at least serve to make the backdrop more convincing.
I close this post with a couple of questions:
To the developers, why do MUDs generally not have good AI? Is it something that players don't care about? Is it too difficult to implement on a large scale? Are these techniques not known or not interesting to MUD developers?
And to the players, do you think that more AI would add significantly to your favourite MUD, or would it just be extra noise?
Finally, if I am mistaken about AI in MUDs being lacking, I would love to be pointed to good examples of it in the wild.
|