on 27 November 2014
I feel harsh writing this but this is not a good book.
There's a thing I call the "Hello world gulf" - it's the gap between theory and little hacked-together snippets that do stuff, and on the other side actual stuff that works real programs and whatnot. This stays firmly on the "hello world" side and only explores the base theory.
It uses for example when explaining utility (a concept you should be familiar with before even thinking about game AI) a health pack and an NPC, if it has nearly-full-health a health-pack's utility is relatively low, if it is near death that utility is quite large. Great, this is obvious!
How do we relate this utility to "should I run for the health pack" "should I take a stealthy route" and stuff like that. Now the reader knows about utility and this (fine) example, now what?
The book is like this all the way. I remember one example of a Civilisation like game, what order to build settlers and warriors, or building defence towers and it explained this fairly well, but it did not tell you anything about evaluating the "danger" for use, it just pulled a value out of no-where and talked about finding utility of this, it then produced a really specific example for the function that would choose to build a tower (or not) and it was quite literally a function implementing the if statement and the calculation on the previous page. Now you can decide to build towers and you can copy out his implementation, great, how do you find out danger and whatnot?
The book is like this constantly, it also assumes you know less than nothing, like "the gradient of a line" and has a half-page-sized diagram of a triangle, it reminded me of being in Year 8 at school (12 year old education)
It dabbles a bit in philosophy, some statistics but it's not good it at, it's really shallow and just gives results with the formulation actually missing the parts you would want (like when deciding to build a tower, I know I need to compare the not build and build options, but how do I consider things like the utility of where to place it - not addressed at all)
I urge anyone reading this not to buy it, if you want an AI book Funge's (and Millington's) is much much better. I regret buying this a lot.
It is a book printed for little more than making money from people hoping to learn about the subject. I hate books like this.