A Brief History of MoksaA Brief History of Moksa(Unfortunately, the history isn't brief, but my description will have to be)
I started playing RPGs years and years ago: my first ventures into the field were as far back as '83 (which might not be all that long ago to old-timers in the gaming community, but it seems long ago to me). I played Basic Dungeons & Dragons, moving along to Advanced D&D, and recently branching out into other gaming systems.
Since the beginning, however, my interest in computers forced me to think that there had to be some way to automatically generate characters with the computer. Typically, this is a process which can take hours, and will likely consume your entire first gaming session (A character that you expect to play for years isn't a trivial thing, and you want to be sure you don't make mistakes that'll bite you later). It's essentially applying deterministic rules, to generate stats, and looks automatable (don't get me wrong: I'm a stong believer in detailed character histories, personalities, and the like, but I'd rather not spend days figuring out what the correct bonuses for a 16 strength are this week, if I'm a dwarf).
So, naturally, I wrote one. It looks pretty easy. My first attempt was a really gross Pascal program written with the help of my friend, Christian Schlosser. It was buggy, and was never really finished, and made liberal use of massive global arrays (we actually had to increase the static allocated memory the compiler gave the program to its maximum value, and still were running into problems). It was then that I discovered that it wasn't nearly as easy as I'd hoped. (I don't even know where the code for that program went, now)
Since then, I've worried on the problem (for those who are interested, that was 1989-90). I came up with the basic ideas behind what I wanted out of (what was then called) TAGS a year or two ago, but the way that I was thinking about it was wrong, and wouldn't work.
Sometime in late '94, early '95, I had the basic idea down pat. I wasn't sure it would work (what about nondeterminism??), but I was on the right track, and actually cornered one of my statistics profs at the University of Waterloo to help me work through some of the more technical math issues. The name came around then, too, when I was thinking about characters as collections of tagged values (which are now termed 'variables', and 'tags' are something else altogether...but I figured I'd keep the word around somewhere, since the project is named that). I had a nondeterministic algorithm in my head for the core char generation routine, which evolved into the one used today. About a month after this, however, I gave up my idea as hopeless, because I thought it couldn't be done.
In November or December of '95, however, I picked it up again when I found some slides that I'd used for a presentation of TAGS to my Independent Studies class earlier in the year. I was explaining it to a friend (either Victor Wiewiorowski or John Paul Vrolyk, I don't remember) and couldn't figure out why it was impossible. After careful consideration, I decided that it wasn't, actually, and I had just made a mistake in my reasoning. So it came back to life.
I worked through most of the details of the previous incarnation of TAGS with my roommates (Victor, John Paul, and Chris Thiessen), and it was actually JP that came up with the amazing name (the Totally Automated Generation System) that TAGS was the acronym of.
I worked on this C++ version for a while, before giving up on it. It wasn't quite right. I had the idea that I needed to propogate deterministic sets around, and stuff like that, but I didn't understand exactly how. I actually got this version sort of working, limping along, but finally gave up, getting engrossed in school and work, and not having amazing success near the end.
Then one day I heard the magical words: "Constraint Satisfaction Problem", and my life has never been the same since. It turns out that hundreds of AI researchers have been working on this problem for years! So, I read up on it, and finally took a course in AI, and one lecture dedicated to CSP's did the trick. The concepts of TAGS crystallized in my mind, and I hacked out a quick version in three days in Java.
Over the next couple of weeks, I worked on it, and finally version 0.2 was released on the Internet.
Some months after playing around with Constraint Satisfaction Problems as a way to generate random content, I discovered some fatal flaws. TAGS was turning into a programming language, but it wasn't a very nice programming language, as languages go. It was weak and ineffectual. So, this made me rethink what I was doing. I read and pondered, and gave up, and tried again. Eventually, I settled on something called "Constraint Logic Programming". This is a programming paradigm based on constraints: just like TAGS was a programming paradigm based on constraints.
At about the same time, I was taking a Religions of the East course, and discovered an amazing term: Moksa. It means "To escape from the cycle of death and rebirth." This is just what I wanted to do, so I've changed the name. So TAGS is now Moksa.
Moksa is a logic programming language, based roughly on Prolog,
or CLP(FD). The only thing you can do in a language like this is
specify primitive constraints, like X=3, or A = B
+ C, and define new user constraints, such as:
grandfather (G, X) :-
father (G, Y), father (Y, X).
(This says that 'G' is the grandfather of 'X' if 'G' is the father of 'Y' and 'Y' is the father of 'X').
Using rules like this, we can build up complex constraint networks, that constrain the random content.
Finally, after writing a complete working version based on naive approaches to Prolog implementation, I bit the bullet and started to write a WAM-based version.
Relatively quickly, I had a working simple Prolog environment, and I'm working towards making it robust, clean, simple, fast, and ISO-compliant. After that, it will be a simple matter to add the features that will create the greatest random gaming content utility of all time.
Well, possibly not the greatest, but certainly the most over-engineered. I just hope somebody uses it, in the end.
And Moksa is becoming reality.
And God saw that it was Good.
Moksa can be found at http://www.svincent.com/moksa/
Comments, Criticisms, Critters, Complaints and Commendations to: svincent@svincent.com