I'm putting my life back together, going back to school, finishing at least one of my degrees. So, I'll not be here so often anymore. I'll still be around, just, much less active, especially in a month or so when I start work on my (third attempted) senior thesis. If I need alpha/beta testers though for it, I'll be sure poke you people

Take care Shorlen, will be nice to see you back again. Good luck with the degree

Eew, thesis as in a huge book on some obscure subject? No thank you on that alpha testing offer.

And I'm scared of college now too.

My thesis was on exercise therapy in behavorial and psychiatric medicine Is that an obscure subject?

I've heard of worse . My friend's boyfriend's thesis was on some obscure chemical marker technology that is not in widespread use but which he could possibly become an expert on. It hurt my brain to have it explained to me.

Come back soon Shorlen.

Yeah, that's obscure enough in my book. My 7th grade science teacher's was on underwater basket weaving.

That's my senior class's favorite sport!

I think we're even going to get it put on our class shirts.

I did my senior thesis on games for cellphones.. they can be fun.

I didn't leave yet I've just been trying to arrange everything for a month or two from now.

My thesis is going to be something where I can incorporate the parts of computer science I'm interested in studying - scalability, database optimization, event handling, simple AI... in other words, a MUD. My project partner and I are going to play with some non-standard spacial models, generated writing for dynamic room descriptions and other things, a text version of 'deformable terrain,' and what a friend of mine named 'on-the-fly user-initiated world modification.' We're also going to play with genetic algorithmns for the AI of certain mobs, so they literally evolve as you fight them, learning what to do and not to do.

Despite having barely started the design, much less the coding for the game, and not even having the project approved yet, I so want to start the coding of this. I've never had a chance to play with genetic algorithmns before

We're still discussing the basic issues, such as how to curb/limit griefing in a game where everything is destroyable. The answer to that one was actually quite simple - insurance (it's a sci-fi setting).

Of course, the thesis is not a finished game, the thesis is a 100-150 page paper about what we learned while coding as much of it as we could in 21 weeks. The finished game is what would come later, if we felt like finishing it, and after resolving the copywrite issues of who owns what when something is made as a school project.

i'm not sure how the Thesis thing works really, if I was to post a link to a very interesting implementation of darwinian AI in a computer game created by a university in texas it wouldn't mess you up with sources and ideas in that?

I know sounds silly but I have to be very careful in graphic design about where my ideas come from.

You people are queer. My current history professors are nagging at me to go for a History Major, and right my thesis on.. What was it? World Population and the Positive Aspects of the Holocaust and Bubonic Plague. Something like that. It all stems from me mentioning in my medieval Euro class that the Bubonic Plague was followed closely by a drought, and that Hitler's mass genocide prevented far worse conditions in today's world, as far as food and energy shortages go.

Isn't it a rather sick topic? Now both my Medievial Euro, my Eastern Civ, and my Ancient Euro profs are riding me to do it. But I've time to decide! Muaha!

Oh, sources are fine. I'll probably be looking up papers on genetic algorithmns anyway. Algorithmns filled with random elements that 'evolve' through heuristic evaluation of how well they did, the next 'generation' of algorithmns taking pieces of the algorithmns that did best, splicing them together in a certain way, evaluating them to see which work best, etc... :hum:

Shorlen, I can help beta-alpha testing or whatever, but I'd be very interested in reading the tesis when done.

It's a subject I've been studiying for quite a while (I am into computer engineering and programming too) and always wondered if it would be implementable in a MUD due to the complexity of finding the right rules to decide wheter the algorithm is good or not, and the resource use for it.

A friend of mine is on his 5th year of a Linguistics phd about the pronunciation shift of a specific vowel sound in 12th-14th century Russia. He's fantastically bored, wildly in debt, and his phd is going to be of absolutely no use to anyone... unless they tape him reading it and market it as a cure for insomnia.. (I only mention him because it's his birthday today)

P.S good luck, Shorlen.

Then you may find this interesting. It's called NERO. Neuro Evolving Robotic Opperatives. Basicly it's a computer game based around NEAT real time learning algorythm which allows for the training of a computer system. maybe it would make more sense if I explained it.

In the game you are given a heap of AI soldiers, when you first get them their minds are blank. Through training and positive and negative reinforcment you can develope their brains to do a specific task. For example you want your little soldiers to run accross a field go around a wall and shoot the bad guy.
You set the award to shooting the guy. The first generation is spawned and all these little robots just run around the screen, the ones who make it to the bad guy get awarded and thus their brains survive to be sent to the next generation, the ones that don't have their brains destroyed. next generation is spawned with a combination of the new brains and the process continues like a darwinian sieve till you have a bunch of finely honed brains.

Nero game

Edit: fixed the address

Crazy thesis project aside, Shorlen going inactive = I fall inactive too (we live together -> he got me into the game, -> less fun for me without him)
And I've also started classes back up, and in addition to classes i'm finding that I have a real social life (albeit with the kind of crazy geeks who come up with crazy projects like this one), and although I'm not nearly as close to graduating as he is and thus won't be working on that crazy MUD project.. I still won't be around so much anymore

A bit more on the current topic, I futzed about many years ago with a neural net style Quake2 AI mod with some silly results - it worked similarly to the AI 'brains' eyod described, but in a quake 2
The learning curve became a little too steep - basically, once a couple of the brains figured out 'If i see someone in my field of 'vision' point at them, if i'm pointing my railgun at someone, shoot it immediately, otherwise move around randomly' fairly quickly (well, quite a lot of generations, but it only took like an hour or two of 5ish minute games at 5x normal gamespeed or so) for the bots to go from completely inept to bots which were impossible to compete against - they would walk around a bit, but mostly spin around spastically, and if they ever saw you, you instantly died to a railgun. (weapon pickup wasnt implemented, so they just all had railguns, this may have been a problem)

I have a feeling that if weapon pickup *was* implemented, the AI would figure out how to pathfind to the railgun, get railgun, repeat earlier scenario...

but yeah, this was a fairly extreme case, and not a very good example.