stevenm (12/05/82)
The Problem ----------- I am very worried about the proliferation of extremely-special-interest newsgroups on the network. Many/most special interest groups are "flashes in the pan", where discussion is heavy to moderate for a few weeks before interest dies off to nothing. We have recently been lamenting about this proliferation, and discussing the need for 'spring cleaning' of newsgroups. This is the sort of newsgroup that will need removing in a few months. The Symptoms ------------ I am annoyed that anyone who gets 10 or 20 replies to a "let's create a new group" query feels that he or she is entitled to create a new group, and broadcast special-interest information to a network of several hundred sites. Take for example a recent article: I've received 22 replies. They break down into: 18 - in agreement, ranging from mild curiosity to enthusiasm 3 - suggested using net.misc (2 of whom referred to a previous discussion about a net.psi) 1 - strongly against, stated that such a group has "no basis for reasonable discussion" I imagine the number of yes votes is of the same order of magnitude as that received by other groups voted into existence so we should go ahead and create the group. I feel that a response of 22 out of a network of 300 sites (and possibly 6000 readers) is insignificant, and that the creation of a new newsgroup on the basis of this response is absurd. For a reader population of 22, a mailing list is a much more reasonable approach that a network newsgroup. Solutions --------- There are several possible solutions to this problem. Unfortunately, most of them require changes in netnews software. 1) Allow only selected sites (ucbvax,cbosg,decvax,etc) to create newsgroups. Appoint network administrators at these sites who are resonsible for creating groups when they are called for. 2) Change netnews software to allow a configuration option which does respond to new group creation requests directly, but forwards them to the local administrator for consideration. This would be coupled with mods to the software to discard articles to groups which have not been authorized. 3) Change to network 'policy' for new group creation to mandate that some minimum number os ***sites*** must agree before a new group can be created. I would suggest that this number be about 100. Alternately, raise the limit on the number of readers to (say), 200. Comments -------- Comments may be addressed to me personally. I will digest them for net.news. Please limit your responses to rational discussion of the (possible) problem, and suggestions for other solutions. Flames will be discarded. S. McGeady Tektronix, Inc. {ucbvax,decvax,zehntel,cbosg}!teklabs!stevenm UUCP stevenm.tektronix CSNET
sjb (12/05/82)
Your concerns are genuine and very good, however your solutions are not so hot (an opinion only): 1) Only allowing certain sites to create groups wouldn't work, since ANY kind of netnews article can be forged to look as though it came from ANYWHERE. 2) This is the best of the three, and I do like part of it. The idea of not creating new groups that wonder in without the proper control message (i.e. an article to a group that doesn't exist on a system) is good and I hear it will be in 2.10 news. However, not creating a group when a control message comes in, opting instead to just tell the news admin, puts an extra burden on the admin, for he/she him/herself has to create it, possibly on many sites. 3) How would the software know how many people voted on it? I agree that, as in your example, 22 people is really not enough. Others disagree, and no one is really wrong. However, if we do keep ''spring cleaning'' as I put it, I think the problem will subside a bit. We need to take the attitude that groups must be removed when they go unused for a period of a few months. This way, the dead groups will die after they have served their purpose (and they will have been given a chance to serve their purpose), they can always be just as easily recreated if they are needed again, and the live groups will stay until they die -- a natural life-span on groups, if you will.
goutal (12/07/82)
This is just another vote in favour of letting groups be born of the common people and die of natural causes. Perhaps someone should explain what's bad about having old groups hanging around. Of course, I can see two things that happen that ARE disadvantageous: (1) Duplicate newsgroups get created from time to time, either because the poster typed the group name wrong or because they didn't know such a group existed at all. Clearly, spring cleaning or better posting software is needed. (I think I'll think about the latter clause and make some more concrete suggestions later.) (2) Old articles don't expire, they just clutter up disks. Some posting software makes it easier to remember to put an expiration date on stuff. I gather that if there isn't one, it just stays around forever, depending on the system and the whim of its manager. In many ways, the second problem is not really a problem (depends on how much room the system has for archival purposes!). Someone in HUMAN-NETS Digest (I think it was) was pointing out that there is no formal archival system for all this craziness. To be sure, it would be a tremendous task, but it might avoid all the redundancy we presently see in the groups. It's nice to suddenly become interested in some area, and start reading the old news before beleaguering the net with questions that were answered last month. Hmmmm, now I'm wandering into the area touched on a week or two back about 'experts', being people who collect and retain surveys. I think we'd better discuss this surveying/archiving issue at greater length, and probably elsewhere. Anyhow, I think spring cleaning, manually or electronically, should be restricted to pruning truly spurious groups, and should not be applied to groups that have only had light traffic or have been inactive for a while. -- Kenn (decvax!)goutal
sjb@sri-unix (12/08/82)
The main disadvantage of having old groups around is that each group active adds to the overhead produced by all of the news programs, due to the search of the active file for each group. The fewer groups, the less time it takes. It's that simple. Also, and I point this out again, there *IS* a limit to the number of groups allowed. Once you reach a certain number of lines in your .newsrc file, you can't read news until you prune it. The more groups there are, the more lines there will be. Getting rid of dead groups helps to solve these (and other) problems. Groups can always be recreated when the need arises.
mike (12/10/82)
Wouldn't the real fix be to create a program like expire or an extension of expire to retire a a newsgroup after a certain threshold of inactivity? We all don't work for the same establishment. Why not consider adding "retire" to the repetoire of netnews support utilities. Its a solvable problem. And probably a lot easier to do than "appoint" an adminitrator to remove and create newsgroups. Also, what about adding a cleanrc function that will take a .newsrc file and remove the entries that have been retired? Enough said - Michael Wended General Instrument R&D ucbvax!menlo70!sytek!gi!mike decvax!sytek!gi!mike