From: Xah Newsgroups: comp.lang.lisp Subject: Re: Curious about functional programming Date: Sat, 22 Jul 2000 11:07:18 -0700 Message-ID: Kent M Pitman wrote: > No conflict with anything Barry said here--jsut wanted to add some thoughts > to underscore some of his remarks for those few who remain both unconvinced > but still open to argument on the matter... No conflict with anything Kent said here -- i just wanted to add some thoughts to underscore some of his remarks for those few who are clouded by the lure of the imperatives but are still capable of critical thinking. Kent begs: > Personally, I would say it's because things in the world DO have state and > because people develop brain hardware that directly mirrors those things > about which they reason most often. Impersonally, i would say Kent have no idea what he is talking about. First off, it's meaningless mumble jumble. On a meta or intuitive level, it's also plagued by non sequiturs. >If most things in the world were state-free what the fuck does this really mean? > I think people would reason well about state-free things. how this follows? and what the fuck is "_reason well_ about state-free things"? When not being technical, mumble jumble appears to be Kent's forte. > Certainly when discussing optimization, as you suggest, which is really about > the abstract meta-process of reasoning about a process WITHOUT reasoning > about its domain data and reasoning only about abstract identities and > transformations of known characteristic (since it's conventionally done by > programs that don't understand the domain task), there are properties of > state-free things that makes "mindless" (I'll call it) reasoning easier. > But "mindful" things--that is, things that are really about a domain and how > it works as reasoned by a trained expert--are often about systems with state > and trying to "model" them without isomorphism but instead by making a > trustable proof that transforming it to a state-free decomposition is hard > because you ahve to do not only the state-free result but also the > transformation. It's easier just to use the brain's natural hardware for > doing statistical prediction, structural modeling, simulation, and other > techniques that are directly available in wetware. Wetware my ass. For those "stateful" things as you say, then it should be modeled by automata, monads (excuse me for my ignorance for this one), neuro-networks, genetic programing, fuzzy-logic, pure objects and so on that has a rigid mathematical foundation. Most of which can be embodied in functional programing. That's the best reply i can gather for Kent's vague verbalization. Barry Margolin writes: >> (our >> hunter/gatherer forebears didn't think "Eat the animal that I cooked that I >> killed", they thought "kill the animal, then cook it, then eat it", and we >> inherited their planning thought processes). So programming practicioners, >> as opposed to academics, tend to prefer imperative programming styles. Kent Pitman apologizes: > Yes, I agree with this analysis completely. I've made similar comments about > the value of recursion in teaching children. Does anyone know someone who has > taught a kid to climb stairs by saying "climb the first stair, then climb > the stairs"? I think we all say "here are a bunch of stairs. learn how to > climb one. now do it to all of them." Huh? Excuse me for my lack of imagination, but could you be a bit precise? Are you now, paralleling analogies with mappings from the functional paradigm to the methodology domain for the injection of your opinion??? By ways of empathy, i think you know what i mean too. Barry Margolin sings: >> As Johan pointed out, Common Lisp provides many functional features, so you >> can use that style when it fits your needs, but overall program design >> tends to be mostly imperative. Kent Pitman echoes: > Precisely. Some of the world is functional, and CL models that with proper > isomorphism to functions. What the FUCK does "CL models that with proper isomorphism to functions" ^^^^^^ ^^^^^^^^^^^ mean? Here Kent Pitman is fucking fuck all fucked up about using jargons to impinge those lesser educated. > I hate working in languages where first class > functions are not available for this. However, most of the world has state > and CL models that as well. It's exactly the ability to integrate > these stylistically clashing abstractions which makes CL so able to model > a world which is likewise full full of uncoordinated abstractions, artifacts, > etc. It's hardly surprising either, since CL's roots are in early AI work > trying to mirror thought on the very hardest of representational problems. > CL has spent less time worrying about the elegance of a "hello world" program > and more on worrying about the "how world" or "why world"... I get tired of correcting sloppy iffy writings like these that's not even belles-lettres nor sophistry. Kent, maybe your technical career wrapped up, maybe you should focus on fiction writing full time. Xah xah@xahlee.org http://xahlee.org/PageTwo_dir/more.html