Path: utzoo!mnetor!uunet!husc6!uwvax!dogie!uwmcsd1!ig!agate!garnet!weemba From: weemba@garnet.berkeley.edu (Obnoxious Emacs Weenie) Newsgroups: news.misc Subject: Re: Take a sniff of Gnews Message-ID: <7961@agate.BERKELEY.EDU> Date: 24 Mar 88 08:53:52 GMT References: <7928@agate.BERKELEY.EDU> <25590@yale-celray.yale.UUCP> <40@ncar.ucar.edu> Sender: usenet@agate.BERKELEY.EDU Reply-To: weemba@garnet.berkeley.edu (a small gnews-ance) Organization: Brahms Gang Posting Central Lines: 208 Supersedes: <7957@agate.BERKELEY.EDU> In-reply-to: woods@ncar.ucar.edu (Greg Woods) In article <25590@yale-celray.yale.UUCP>, lisper-bjorn@CS (Bjorn Lisper) writes: >the user in the form of a question: "tell me more about your desire to post >to comp.foo.bar...." There are so many newsgroups, and so many possible >keywords for each newsgroups. But not all of them have repeated problems caused by naive users over and over again. If a problem is annoying enough in a certain newsgroup, then by the very fact that it is annoying, someone will be motivated to work on it. If no one is willing, then it musnt' have been annoying to begin with. > Besides, just looking for keywords isn't >enough since they can mean different things in different context. A posting >to comp.parallel that mentions "abortion" 17 times doesn't have to be a >misdirected talk.abortion posting for instance, it could very well deal with >how to get rid of unwanted processes in a multiprocessor environment. Articles are not blocked, they are queried. Just say no if you want. >Language is another issue, discussions doesn't necessarily have to be in >English. Postings can be restricted to certain geographical areas and a >posting restricted to, say, West Germany could very well be in German, even >in a newsgroup where the discussion is primarily in English otherwise. SO WHAT? I couldn't care less about what's happening to a West German only newsgroup. The small size of it mitigates the need for much anyway. > This >IS serious AI and I don't think that even e-lisp will make such a screening >system feasible. No, it isn't serious AI. If it goofs, nothing worse than a little annoy- ance has occurred. So what? When it works, 1000s of little annoyances are prevented. And if it doesn't translate German into English to check if "abortion" is being discussed, so what? Really, you're exaggerating the difficulties way out of proportion. >Even if such a screening mechanism WAS possible I would object to it. The >reason is that I think those participating in the discussion of a newsgroup >should decide the topic by themselves. And it's decided year in and year out that Velikovsky belongs in talk.origins. > Every newsgroup has a brief >categorization of suitable topics for discussion but what's interesting >within a certain category changes with time. A set of screening rules would >make everything much more rigid. And a set of screening rules could include a definitive answer/explanation to the Monty Hall problem which is hitting sci.math, rec.puzzles for the third time in the last six months. > The wonderful thing with usenet is that it >is what we make it to. OK. You tell me how the readers of sci.astro can make it Velikovsky-free. > A set of detailed screening rules, expressing what >somebody sometimes decided should be allowed or not allowed, will hamper >this basic principle. I'd like to see some higher quality newsgroups around here. And I'd like to see it done without moderation. The most basic screening would serve as a "Hey you, think before posting", or "Sorry, bub, but this belongs in a different newsgroup." Why does that bother you? I don't know of anyone who's going to miss "I vote ..." articles in news.groups. Do you? >Finally, who wants to get kicked around by a machine? It is bad enough with >the little message I get every time I try to post something: > >"This program may post news to many machines. > Are you absolutely sure that you want to do that? [ny]" Hahaha. No way is this in Gnews. The rules are only brought into play when a certain condition is met. Most postings should cause no problems. >Whenever I run a program I want to be the one that is in charge, not the >other way around. Having to fight my way through a set of well intended but >obnoxious rules questioning the contents of my postings would either give me >an ulcer or force me off usenet. Well if asking you to post a naive Unix question to comp.unix.questions only, instead of cross-posting to comp.unix.wizards, just ruins your di- gestion, then I think maybe you need to relax a little. Of course, the same could be said for the person who gets upset at such crosspostings in the first place. >I do recognize the problem with junk postings that this screening technique >is supposed to solve. Good. > But I think this is the wrong way to do it. Those in >charge to take care of this problem are the other readers of the affected >newsgroups. It is possible to send e-mail to people generating too much >noise and ask them to stop. That's what I usually do. Along with KILL files. What's wrong with doing it once and for all times with certain standard no-nos? I cited 3 or 4 examples, none of which you mentioned. Tell me what was wrong with them. > Sometimes flames ARE appropriate (although many >of them belong to the "noise"). I couldn't agree more! >PS. The original posting contained some advertising for Gnews. In fact, that advertising was the primary reason I posted in the first place. I think Emacs+Gnews really makes for a wonderful development environment. We We hear people make suggestions about do-this, do-that over and over again, but as very few have the time, they just keep getting proposed over and over again. To me, having a two-layer implementation method, whereby the first version is written in ELisp and tested on Gnewsers, and then it gets recoded in C for rn/Pnews, allows not just for speed in development, but trial runs to see whether the idea is practical or not. Maybe the rules-based screen can work. Maybe it can't. Maybe it will be too obnoxious, or maybe people will go along with the flow. I don't know, you don't know, NOBODY knows. Gnews offers a simpler and quicker way to test this. If the method turns out to be a failure, then whoever worked on it hasn't wasted as much time as he would have had he worked on it in C. > I'm in no >way opposed to Gnews, as long as people with itching fingers don't try to >hack in their favorite prejudices about the subjects of certain newsgroups. How about, "the standard prejudices that have been settled on repeatedly"? -------------------------------------------------------------------------- In article <40@ncar.ucar.edu>, woods@ncar (Greg Woods) writes: >In article <7928@agate.BERKELEY.EDU> weemba@garnet.berkeley.edu (a small gnews-ance) writes: >>Thus, looking for "Guin?ess" in articles crossposted to 4 or more news- >>groups, or counting the number of times "abortion" or "fetus" shows up in >>a non talk.abortion article, etc. are not terribly difficult tasks. There >>are 100s of these kludges that could be tried. > > Yes, there are. My question is: how useful is that information? I say: >not very. And I say: nobody knows. > Someone might post a line in a comp.lang.c article like "I wrote >a really bad subroutine to do this; it was truly an abortion". Say then requiring "abortion", "fetus", "person" to show up four times between them--and not quoted as I'm doing now either--would be a better requirement. I'm not trying to write any definitive rules right now. >Counting words is NOT GOOD ENOUGH to do classifications. I'm NOT proposing classifications. I'm proposing half-hearted, lukewarm classifications. The user is free to answer "no" to the "do we understand each other?" question, ie, "No, Mr Computer Sir, you didn't understand me. Shut up and let me post." > You have to >determine if the word is really used in the context that would place it >into the classification for it (for example, the above article would >certainly not belong in talk.abortion). And THIS is what would require >"serious AI". NO NO NO. If the rough-and-ready rule is bad, it will just mean a little more annoyance. But it doesn't require "serious AI". The user always has the final say. > As for the "fairy stories", I don't think it would be possible to keep >up-to-date with this sort of thing. I see that you are not rebutting these stories, but are pointing out an external difficulty. Is this a concession, then, that you agree that such fairy stories, if implemented, would have a beneficial effect? I'll agree that it's possible to come up with monster stories instead--but I think that it has to be agreed upon first whether or not in principle this idea of Brad's is worthwhile. Bjorn Lisper explicitly said he didn't think so --how about your opinion? I can do no more than say to Bjorn, "I disagree, here's the reason". But when it comes to questions of actually implementing something that you and other widely respected sysadmins think is OK in principle, but just happen to believe is a Lisper's pipe-dream, then only trying it out will tell us who's correct. And there's no point in trying it out if the re- action from the backbone is a loud "hell no". Remember junker? > Remember, lots of news admins are doing >that job as a 10th or 11th priority item. Maintaining a program as complex >as a GOOD keyword-based news system cannot, in my opinion, be done as a >10th priority item. Huh? This makes no sense. Why is keeping up to date for one kind of pro- gram any harder than any other? There's a database of rules, and responses. If it's out-of-date, then some will be missed. So what? And deciding just how much work this will be, based on your assumption that it's going to be a very complicated AI monster, isn't really a fair criterion yet. Remember: the idea is to encourage better traffic, not create a Usenetopia. Will it work? Hell if I know. And hell if you know either. ucbvax!garnet!weemba Matthew P Wiener/Brahms Gang/Berkeley CA 94720 They can all go to hell. Of course, some should go before others. One has a responsibility to make discriminations. --Simon Lacerous