simon (12/06/82)
We were recently warned of a vile disease threatening to invade the network, the problem - proliferating newsgroups. As one of the 'symptoms' of this problem, apparently my proposal triggered the warning, I suppose I should respond. Unfortunately the problem has been misidentified, it is not proliferating newsgroups but proliferating users (please don't take this in the wrong sense). The opinion of those worried about 'viral' newsgroups seems to be that a newsgroup (and hence the articles submitted to it) should be allowed only if read by a substantial portion of the network community. While this sounds reasonable, even a simple analysis shows a network based upon such a policy is not workable. Suppose we had a network where an article would be accepted only if it would be viewed by x% of the user population. It is easy to see that, assuming the average frequency of articles submitted per user stays constant, the amount of time we spend reading the news will grow with the user population. The only way a user can stop this bombardment of news articles is by using a more selective 'filter'. This is where newsgroups come in, I think we should view them as being like keywords in an information retrieval system. And just as when an information system grows one needs more refined keywords so do we need more (and more selective) newsgroups as the network grows. So it would seem that the proliferation of newsgroups is not the disease but the cure. Simon Gibbs, Dept. of CS, U. of Toronto