Most of today's work has been figuring out precisely what I didn't understand about the system previously, such as what constitutes a rule and what doesn't. It's slightly hampered by Kosko's seeming unwillingness to provide mathematical symbols (and while the effort is appreciated in places, in the example it's a real problem -- laborious sentence structures representing underlying geometric structures mean lots of rereading before even a few lines of code can be written.)
Of course, I'm not really complaining -- much -- I mean, I think it's interesting enough to attempt to produce a module that approximates its methods. I don't know how far this project will get, mostly because today was (oddly enough and completely out of the blue) a completely free day, and I don't find many of those lately. Tomorrow, for example, I have a meeting with the rest of the volunteers and IT people at the suicide boards, and I'm going to try to convince the guy running the IT end of things that IRC is a good bet. It's going to be uphill running all the way, I suspect.
Anyway -- more research into IRC server bandwidth stats tomorrow, maybe a little more code on the fuzzy system. Hopefully lots of coffee.
Also, I'm especially proud of myself because I'm taking the extra effort to work in pod for the first time. It's not pretty, but it's functional, and I change the documentation before I make code changes, which keeps me from half-implementing half-baked ideas because I have to explain them in English to myself a few times first, trying to determine if they're understandable. If I toss it around enough, usually I see what's wrong with the design before I get more code written.
So far I think most of it's a fairly faithful and reasonable attempt to code the damn thing, but I may be way off base. When I say that I'm really just farting around here, I exaggerate to no degree.
The closest thing I have to a way to check it (again, due to these damnable nonexistent symbols and formulas and whatnot) is to build the system, run the examples Kosko provides through it, and see what I get for results. Not a perfect way to check, but then perhaps I can compare it to similar efforts, or maybe other fuzzy systems.