Whilst I quite like the method that the Dungeons & Dragons Adventure System uses to handle enemy movement and action, it becomes very easy from a player point of view to exploit when you always know exactly what your enemy will do in any given situation.
It's like attracting your enemies in a first person shooter and then camping around a corner. Knowing full well that they will just walk on over blindly, only to have both barrels of a sawed-off shotgun pumped into them at point blank range.
It's just not realistic and, it's certainly nothing like playing against another human being.
Then again, the only real alternative card driven A.I out there, found in Gears of War may be more unpredictable but, it's random and, random A.I is really no kind of A.I at all.
Both of them also really fail in taking into account the mind of a gamer too.
Strategies change based on who you're playing against, how experienced you are and, the situation at hand. All of which should be considered when designing the games A.I. Even if there is an optimal strategy, there is no guarantee it will be used or even spotted by a player or, that it will be relevant to every single game. Maybe not even the next game.
All of this as you can imagine is proving quite interesting food for thought. Can a board game ever truly have a functional and realistic A.I?
Argh...good question. I would think that you would have to have some form of tree that would ask the player 'what is your current status'?
ReplyDeleteAre you being overwelmed - then the AI will keep coming the way they are.
Are you easily handling the wave - then the AI will try to flank or pull something else.
As you said - many variables to consider to make challenging and unpredictable. I would have much fun helping you come up with something here :)
Granted - I have no clue about the Deathwatch Expansion and what kind of solitaire rules it had in it...
ReplyDeleteIn fairness Deathwing did very little for solitaire play beyond generating a random hulk, which we have more than tied up with the mechanics in the Adventure System.
DeleteYour observation of a logic tree is very apt too and most likely the way forwards.
The trick will be in breaking down an extremely complex tree into only a few lines of instructions that best handle the situation at hand.
What I am currently thinking is that there needs to be two processes at work in tandem. One to drive the AI towards the optimum situation and, a second to best take advantage of it.