From msb@sq.sq.com Wed Nov 23 07:51:03 1994 From: msb@sq.sq.com (Mark Brader) Newsgroups: comp.society.folklore Subject: Re: The Backbone Cabal Date: 18 Nov 1994 14:23:38 -0500 Organization: SoftQuad Inc., Toronto, Canada NNTP-Posting-Host: bio6.acpub.duke.edu X-Moderator-Note: comp.society.folklore is a moderated newsgroup for serious discussion of computer history and folklore. Material not related to computer history or folklore will not be approved for posting. Joel K. Furr (jfurr@acpub.duke.edu) writes: > Speaking as someone who wasn't active on the Internet when it happened, > I'd like to bring up the subject of the Backbone Cabal: what was it... For starters, the Internet is pretty well irrelevant here. As you know, the Usenet "History and Sources" periodic posting says: # Usenet came into being in late 1979, shortly after the release of V7 # Unix with UUCP. Two Duke University grad students in North Carolina, # Tom Truscott and Jim Ellis, thought of hooking computers together to # exchange information with the Unix community. Steve Bellovin, a grad # student at the University of North Carolina, put together the first # version of the news software using shell scripts and installed it on # the first two sites: "unc" and "duke." At the beginning of 1980 the # network consisted of those two sites and "phs" (another machine at # Duke), and was described at the January Usenix conference. ... (See also the signature quote.) In those days UUCP was the only communication method used on Usenet, and the Internet's ancestor the Arpanet was essentially a separate network. According to the "History and Sources" posting, it was only after NNTP appeared in 1986 that TCP/IP connections (as on the Internet) became important to Usenet. UUCP is a point-to-point connection protocol, and instead of the network looking to the user like an amorphous blob, it had the form of a graph (in the graph-theory sense). "Network maps" were produced from time to time which looked like this. (The machine names are real, but the connections are drawn at random.) utzoo ----- decvax ------ seismo ----- ihnp4 | | | ittvax harpo mcvax |\---utcsri | | ---lsuc duke ----- unc dciem | phs Now the thing is that on such a diagram you can choose to emphasize a set of lines forming a path through the hosts -- say "utzoo - decvax - seismo - ihnp4". The "backbone" was simply a group of hosts whose admins agreed to form such a connected set, and to devote whatever resources were necessary to carry all the Usenet traffic and to pass it on promptly (rather than, say, waiting for overnight when their machine was less busy, as other sites often did). In practice the backbone contained a number of redundant connections, so its graph contained circuits rather than being a tree as implied by the diagram. The idea was that in each area reached by the net, there would be one or more backbone sites, and other sites would probably form themselves into a tree around a backbone site. They way you paid for a position close to the backbone site was to agree to feed a larger number of other sites. So the backbone was important because, in a network organized that way, if the backbone administrators agreed to carry or not carry a certain newsgroup, they could enforce that decision on the net. Hence the term "cabal", which I always viewed as being used in a friendly light. The Jargon File says: :backbone cabal: n. A group of large-site administrators who pushed through the {Great Renaming} and reined in the chaos of {USENET} during most of the 1980s. The cabal {mailing list} disbanded in late 1988 after a bitter internal catfight. :backbone site: n. A key USENET and email site; one that processes a large amount of third-party traffic, especially if it is the home site of any of the regional coordinators for the USENET maps. Notable backbone sites as of early 1993 include uunet and the mail machines at Rutgers University, UC Berkeley, {DEC}'s Western Research Laboratories, Ohio State University, and the University of Texas. Compare {rib site}, {leaf site}. but although it gives a definition for "backbone site", it doesn't have the same meaning as it did in Cabal days, when there were explicit agreements as to whether a site was on the backbone or not. Incidentally, maps like that, in later days showing only the backbone sites, continued to exist until somewhere around 1984-5. The last one I saw was split between two or three sheets of paper. Around that time the Usenet Mapping Project started, and that eventually gave us got Brian Reid's PostScript maps showing the network geographically. By the way, the no- tation "city" used on the #L lines in the mapping files was invented by me. > especially, what brought it down? The story I've always heard is that > someone forced through a successful vote for a group called > comp.protocols.tcp-ip.eniac and the Cabal refused to carry it ... That did happen, but I don't think it was really an important event. What that vote mainly did was to provide a precedent for the notion that a vote could have been rigged through campaigning and therefore the result should be ignored. However, my recollections here may be wrong. The thing that you really have to understand is that being on Usenet, not all that long ago, was a low-profile sort of thing. Site administrators might well set up newsfeeds on their own initiative, using modems bought for other purposes, and might do all news maintenance on their own time. Some of them were fearful that if their management got wind of what was going on with the computers, they'd shut down the connection. If the management did want a high-profile network connection, Usenet would not be their likely network of choice. It wasn't like that at the backbone sites, but even there, they didn't want to go around attracting controversy. And then someone proposed that the discussions and stories about sex that had been appearing mostly in soc.singles should be split out into a new newsgroup about sex. Others have said that it was to have been rec.sex; my memory says soc.sex. In any case it passed something like 200-6 -- but several backbone sites refused to carry it, for the reasons just described, and it was never created. The alt.* hierarchy may already have existed at that point, I'm not sure, but what really got it going was the non-creation of {soc,rec}.sex and the creation of alt.sex in its stead. Sites that didn't mind carrying such a newsgroup carried alt.sex, and... here we are. And at about this time, the regular monthly postings that formerly said things like: | You can also send mail to "postmaster" at a backbone site. were quietly edited to: | You can also send mail to "postmaster" at one of the major Usenet sites. and, not with a bang but a whimper, the backbone was dead. > ... and the public outcry caused them to say "Fine, screw it, we quit." I saw no particular public outcry. I saw flaming and whining from a few individuals. The "internal catfight" wording suggests that some people among the backbone cabal agreed with the complainers; then, since the cabal no longer seemed able to form a consensus, and could be routed around anyway with NNTP and alt.*, they decided their time was past and quietly abdicated. The one major survival from the backbone era is the List of Active Newsgroups and the accompanying notion of voting on newsgroup creation. This write-up describes my personal perceptions of what happened. I was never on a backbone site, and there may have been important undercurrents (or for that matter postings) that I missed. Also, I'm writing all of this >from memory, except for the places where I've quoted some other text.