reid@Glacier.ARPA (Brian Reid) (11/08/85)
net.news.group is so full of flamage right now that I'm reluctant to post this, not because it's a flame, but because I'm sure that many reasonable people will have stopped reading net.news.group. Six months ago, or so, I posted sent Gene Spafford a categorization of newsgroup styles, which he later posted, in which I developed a taxonomy of technical groups, hobby/affinity groups, and soapbox groups. I now realize that there are in turn 3 different kinds of technical groups: 1) Software distribution. net.sources 2) Access to experts; information exchange. Technical exchange based on demonstrable facts. I am thinking of these as ``technology'' groups. net.bugs is probably the most pure of these, but fa.unix-wizards, net.lan, and so forth are normally used for exchange of technical information. 3) Research discussion. In a certain sense these are technical soapbox groups. There are a lot of areas of computer engineering where there has not ever been a consensus as to the truth. What is the best operating system? What size should memory pages be? Are RISC architectures a good idea? There is a lot of research discussion and idea exchange that takes place outside of USENET, in every field from anthropology to zoology, and in all of these fields, research discussions are "moderated". In truth, they are normally "refereed", which means not only that a contribution must be approved by a moderator, but that it must be approved by all of the moderators. USENET technology groups are invaluable; I believe that they are the only reason why most sites are willing to articipate in USENET. USENET software distribution groups are arguably valuable: they perform a worthwhile service, but at a very high cost-to-benefit ratio. However, I have never seen a USENET research discussion group that I thought was worth reading for very long. The reason for this is that about 80% of the people who participate in a research discussion group don't know what they are talking about, but dive right in with opinions anyhow. Speaking for myself, I don't care at all what Joe XYZ of the Glukkomatic Corporation, known netwide as glukko!dragon-of-darkness, thinks about page size in a virtual memory system. I have no evidence that he knows anything about virtual memory, and my time is too precious to wade through all of the crap and decide whether or not he is to be taken seriously. But I care a lot what John Mashey of MIPS computer or Eugene Brooks of Livermore or Dick Dunn of NBI Research think. I've never met any of them as far as I know; I have observed that through the months and years on the network, they have consistently been well-informed and reasoned and reasonable. I also will always take the time to read what Joe Falcone of DEC has posted (nothing recently), even though I invariably disagree with him, because he usually supports his statements with measured data or reference to other people's measured data. And so forth. What I want to read are unmoderated newsgroups for the exchange of technical information, and moderated (refereed) newsgroups for the exchange of research information and discussion. The moderated newsgroups for the exchange of research information will take care of filtering the amateurs out of the research discussions. I can already hear you all saying "WHOA!! Who does he think he is, deciding that certain people are amateurs or are uninformed? What about my rights to be heard?" You may or may not have rights to be heard, but I also have rights not to listen. Unmoderated newsgroups serve writers, and moderated newsgroups serve readers. The current network policies always favor the interests of writers over the interests of readers, because they count writing and not reading. As a reader of netnews I want somebody to filter out the uneducated crap for me. I would be willing to pay money for that service. When I pay money to subscribe to a technical journal, what I am really paying for is the editorial service and not the distribution service. It would be awfully nice if USENET matured into something that could offer refereed, or even just moderated, exchanges of research information in addition to the other things that it does. A corollary to this is that I oppose the creation of net.os, even though I like discussions of operating systems. If net.os fills up with traffic like "what is a remote procedure call?" or "VMS is better than Primos!!" or "anybody who doesn't like page mapping is brain-damaged", then it will be a failure. In the 5-year history of the net, I have never seen any evicdence that discussion of the quality of TOCS, or even Operating Systems Review, can take place without moderation. Groups to discuss research topics should be moderated. Groups to exchange factual information should not be moderated. -- Brian Reid decwrl!glacier!reid Stanford reid@SU-Glacier.ARPA