bobvan (12/04/82)
The more I think about this "survey survey" idea, the more I come to realize that there are many things that should be on the list that really aren't surveys. The only one I had on the prototype list was the source for the netiquette document. Since then, it has occurred to me that both the logical and physical (?) net maps should be listed. Listing net addresses for the DECWARS and LBJ anthologies would certainly cut down on flames. Also, the joke decryption program and a recent index of net.sources should be posted periodicly. This would minimize requests like this one recently seen in this newsgroup: A little while ago, someone posted an article advertising an updated version of expire that would, among other things, rebuild the history file. Now I need it. ... Could someone send this stuff to me or re-post it to net.sources? How many times have your found yourself in this situation? I know it has happened to me. We have used never-expiring articles to archive articles of great importance (such as the netiquette article). An inherent problem with this idea is that sites that join the net since the article was posted don't get a copy. It is ironic that these are the sites that need it the most. (Perhaps someone should be in charge of mailing it to new sites. It could be automaticly mailed to user "usenet" at the new site by a simple shell script invoked from some site's sys file.) In the two weeks since the netiquette article was posted, we've received nine new site announcements. If this growth continues, over 1/4 of the sites will not have the article in less than a year. Don't get me wrong; I'm not saying that we shouldn't use never-expiring articles -- I'm just pointing up a shortcoming of the idea. When discussing news archiving recently, someone on the net suggested a "distributed archive" -- each site is responsible for keeping an archive of all the news it sends out. The idea was shot down in flames by sites that are gateways. I claim that there is a sort of "natural distributed archive" out there already. Anyone who posts something of value generally keeps a copy. This would certainly seem true in the case of net.sources. If you've got the source on-line, it is a trivial matter to mail out a copy whenever someone asks for a copy. I propose that we (the net) regularly post an index for net.sources. Much like the "survey survey", the index would consist of the name and net address of the author, the posting date, and a one line description of the program. With the current frequency of net.sources postings, it wouldn't take much effort to maintain such an index. Requesting source directly from the author only when you are ready to use it means that you'll have the lastest and greatest version. If all else fails, you can always ask the net. So, to summarize and conclude, I suggest that the regularly posted "survey survey" also include at least a net.sources index and perhaps a few other pointers to other things of general interest. I invite your comments on these ideas. Thanks, Bob Van Valzah (...!decvax!ittvax!tpdcvax!bobvan) P.s. The current "accurate" tally of news readers per site has dropped to about 17.5, with about 20% of the sites reporting in. That puts the net total at about 6,400. The count derived from wc /usr/lib/news/users remains at about 18,000. These figure are subject to the same caveats as before. I am compiling a rather extensive list of reasons why they are likely to be in error.