Do you remember the thief, dear reader?
I chose him to focus on first, thinking that the background of a murderous thief should be more easily defined, and simpler, than the pompous nobleman character.
I have defined him as a person that suffered greatly in his childhood at the hands of his father, causing him to grow a great hatred toward not only his father, but people in general.
After killing his father at the age of fourteen, he scrounges for a living, stealing what he can to survive. He ramps this up over the course of his life, until he is whirling dervish of pain and death, taking what he wants and killing when he needs.
After defining him so, I have implemented some very simple generic A.I., which takes a thing that the person feels most strongly towards, and chooses a goal and methods based on that.
I didn’t realise it, but as soon as I implemented this and ran it, I saw these traces:
Love to person(father) was -100, goal chosen was kill target
Barr is choosing a method to reach his goal!
I could [kill [it]]...
Alright, I'll do it. I'll [kill him].
Method 'kill target' was chosen.
This was not to plan, of course, and took me somewhat by surprise… apparently, my thief has a lot to learn before he does what I want him to do. He should, ideally, be coming into this situation having already attempted (and failed) to find his father; therefore, even though he may feel the strongest hatred toward his father, he won’t choose to drop everything and keep attempting to find him.
It’s also made me consider a fear table… should he fear his father more than hating him? And if so, he may want to kill him, but would probably be too fearful to actually seek him out…
In any case, this glimpse is a beacon of hope, that I am perhaps heading in the correct direction… and if so, there should be a lot more of these experiences to come.
Are, and how are, you weighting the different methods?
To further muddy your cognitive waters, I’d imagine that no individual weighting scheme would suffice for your ends… A simplistic model that springs to mind is that there are a set of weighting factors, but a character’s past and relation with that other character’s affects them.
For example, killing someone might be negatively weighted due to a character’s aversion to violence, but under the right circumstances this would be a more acceptable approach inspite of this.
I wonder if the current weakness of the Aussie dollar increases the street value of my random thoughts above $0.02…
At the moment, methods are chosen by prerequisites; ie. a character won’t choose to kill someone unless their hatred toward that person is over a set amount.
A character also remembers their experience using a method – if they’ve had a positive experience using that method in the past (or if they’ve been *taught* by an superior that a method is good to use), they’ll be more likely to use that method.
Finally, methods will be chosen based upon convenience – the easier and more straightforward methods will be chosen over methods that the character doesn’t have as much information about.
I have more of a weighting system on goals – the closer your attributes match up to a goal, the more likely you will be to choose that goal.. and then you choose methods based on that goal.
The idea is that you may choose a goal, but you may not be able to carry out the method for that goal.. (ie. you decided that you wanted to kill someone, and you create the situation in which you have a chance to kill them, but you can’t bring yourself to actually execute the action – to deal the death blow.)
Also, I think we’ll see growth factor in the street value of your random thoughts in the coming quarter!
Pingback: Me :-)