I have an idea to contribute from reading this thread.

You stated earlier in the thread that you saw some interesting emergent behaviour where the agents were wandering around digging for fun. Now, i program for fun. It can be used to make money but for me its a hobby when it comes to games. Let's use a similar example. The guy likes to dig, and he digs all day until all of a sudden in his hobby of digging he comes across something of value. Some ore, or some gold or such. He pockets this and can then profit from it.

Is this something worth factoring in, that sometimes even these amusement activities that bored agents do might have a profit value attached that might weight the agents decision to do them as well as just entertainment and/or having nothing better to do? Especially if in the past the agent has had a 'lucky' windfall.
To take it further fishing may be a hobby as well as a way of sustinence, as for berry picking, a little more doubtful :-)

What do you guys think, is this a good idea to consider or did i miss something here?

nope, didn't miss anything and, in fact, what you describe is 'kinda' already taken into account: an agent chooses his actions through his 'personality filter' (as I call it), and if he 'likes to dig' then he will tend to do things that involve digging, this means everything from 'fun' activities, to activities where he is trying to earn money... but, there is no way (yet) that I can see to link up an arbitrary event (like finding gold) to the task of 'earning money' through which 'digging' would be then chosen.... restated, the agents may know that certain, predetermined, tasks 'earn money', but don't have the ability to 'add new money earning tasks' to this finite list.... very interesting!

Between all these posts on this thread, my programming time to completion in starting to be measured in decades instead of months/years.... but all very worthy ideas, makes me think 'differently'...

Call it the "hobbyist effect".

... or the ever-expanding AI programmers nightmare effect....

as an aside, I am pretty well versed in programming 2D tile games and 2.5D Isometric games, but have stayed away from the true 3D stuff. But DF & MC (Dwarf Fortress, MineCraft) have got me hooked on modifying the terrain by players/AI agents and I came up against the upper limits of what is possible with 2.5D Iso - I had a modifiable terrain, with 32 levels of height, huge scrolling map (view was 72 x 72 x 8 'cubes' in a memory map of 512 x 512 x 32), but after adding in 'height' I discovered that it just wasn't gonna work for digging caves, or when things go around behind hills/mountains/tall buildings/etc - yes, I did add the ability to 'rotate' the map into the 4 directions, which 2.5D is capable of, and this allowed me to look behind things and such... but caves were a different story....

Soooo, long story even longer, the past few days I have been forcing myself to learn OpenGL and dealing with 3D worlds. and in the process of trying to modify my sim (which was just 2D), I broke it, good and thorough, and all I had saved of past versions was just a basic 'tree' locator and basic influence map of forests testing thing.... very sad! Oh well, I probably needed to redesign my whole AI architecture from the ground up anyhow, but this time I will (A) have the sim output in 3D, and more importantly, ( try to remember to keep multiple versions , or backups of my work as I go along - probably making a copy after every programming session or so.... argh!

Ultimately, my idea is that have this single, versatile, gignormous AI engine which can be dropped into any kind of RPG game environment with minimal hooks, and have it provide interesting gameplay for the user no matter if the game interface is 2D, 2.5D, or 3D... that shouldn't matter much except to a few specific routines (like A* pathfinding, collision detection, etc)

Should be getting Mark's book anyday now... that will get me started off in the right directions!

darn! I was wrong... The dang book just shipped... haven't I been waiting for weeks already? sheesh! Guess its back to studying OpenGL crap... yuck... maybe I will just stick the AI sim into a 2.5D...arg!

Soooo, long story even longer, the past few days I have been forcing myself to learn OpenGL and dealing with 3D worlds. and in the process of trying to modify my sim (which was just 2D), I broke it, good and thorough, and all I had saved of past versions was just a basic 'tree' locator and basic influence map of forests testing thing.... very sad! Oh well, I probably needed to redesign my whole AI architecture from the ground up anyhow, but this time I will (A) have the sim output in 3D, and more importantly, ( try to remember to keep multiple versions , or backups of my work as I go along - probably making a copy after every programming session or so.... argh!

sounds like you need to to get yourself some free souce control software. I recommend http://www.visualsvn.com/ if you check in your changes everyday then you can easily rollback to an old version should you break someting.

sounds like you need to to get yourself some free souce control software. I recommend http://www.visualsvn.com/ if you check in your changes everyday then you can easily rollback to an old version should you break someting.

that sounds like something a bit too smart for me to do. LOL! thanks!

Just wanted to provide an update:

In my quest for understanding and deciding how exactly to implement such a dynamic AI, I think I have settled on the GOAP system (Goal Oriented Action Planner). I have come across quite a few articles (and thesis papers) which have been detailed enough for me to grasp the actual possible programming of one. Here is the paper that really decided it for me (Enhanced Behaviors using GOAP). I also have come across quite a few articles by you, Mr. Dave Mark, and you continue to impress me with how you are able to instruct in an inviting and casual way - keep it up!

One of the advantages of the GOAP system is that the 'actions' are independent from the AI architecture which means once you have a GOAP setup, you can easily add new actions to it and it will incorporate such actions into its possible planning process - VERY slick and allows for a continual upgrade path, AND, perhaps more importantly, allows for the possibility of re-using the AI structure between games, even in different genres. Which leads me to my next question....

instead of re-inventing the wheel, does anyone know or have a GOAP implementation to which I could just focus on adding 'actions' and the world hooks (animations, etc)? If not, then wouldn't this be an IDEAL piece of middleware to develop for the AI community and Indie developer community in general? I could see very amateur programmers being able to use such a system and easily incorporate it into their games, just like how they currently are able to create pretty decent 3-D games without evening understanding all the math and algo's that go into the engine itself. Or is this beast out there somewhere already and I just haven't found it yet?

...just askin'!

BTW: dave Mark - I have really enjoyed your Behavioral Mathematics book and it has helped me understand some things and look at problems from sometimes totally different perspectives - Thanks! worth every penny!

ALSO: interesting GOAP implementation... but I don't get how GOAP is able to 'figure out' the guard patrol routes... anyone care to explain or theorize how this was accomplished? I don't see anything within standard GOAP that allows for predictive planning of other agents. (Watch the video)

DaveMARK,
In regards to generating market pricing, supply, demand curves, etc... I came across this interesting thesis and I would like your opinion:

http://larc.unt.edu/techreports/LARC-2010-03.pdf

Also, any updates or tidbits regarding your own AI Sim Project that you have alluded to before? Any scraps of info would be great to read!

Thanks!