This is the next file to do. Nothing done yet. From bjones Thu Oct 25 06:55:51 1990 To: usenet.hist Subject: Technical links and a growing volume Status: R There is one kind of underdeveloped string in this discussion of the net's history that I would like to ask people to develop more fully. (Actually there are about 50 such strings but this one piques my interest most.) What I am curious about is how various systems developed to deal with the ever-expanding volume of the net. Steve Bellovin says that at the beginning the total daily volume over the net was expected to be one or two articles a day. Obviously that didn't last long. I remember when I first began to read the news in late 1984 the people here at UCSD who ran the main machine for news on campus -- sdcavax -- were already talking about turning off news. They complained that UUCP was tying up one of the ports five or six times a day, just to download news. Sdcsvax was the Computer Science faculty research machine and the faculty was getting unhappy with the competition for resources. Two years later when I first sat on the campus committee on computing the issue was still there. The committee recommended that the campus but a machine strictly for news and mail, which is what happened in 1988 or so. If I understand what has been said correctly, this whole thing began with UUCP -- "a poor man's ARPANET"* -- at Duke and UNC-Chapel Hill. What I would like to understand better is how things moved from that start to the present means of handling the news. There has been mention of siesmo as the main news machine for the net, UUNET, the Telebit Trailblazer modem, and the Internet. How did each of these come to be involved in the net and how did the net change as a result? What other kinds of transmission links were there and how did each of these address the growing problem of volume? "The whole world wants to know" :-) bj * "USENET - A General Access" : Daniel, Ellis and Truscott, no date From looking!brad@watmath.waterloo.edu Thu Oct 25 08:30 PDT 1990 Subject: Re: Technical links and a growing volume To: bjones@ucsd.edu (Bruce Jones) Date: Thu, 25 Oct 90 11:13:48 EDT From: Brad Templeton Cc: X-Mailer: ELM [version 2.2 PL0] Status: R I posted an article a few years ago noting that net growth had almost excactly matched the decrease in cost of bandwidth. It started with 1200 bps, unbatched, uncompressed (300 bps for some) in a day of higher LD charges. Modems got faster, news got batched, news got compressed, modems got even faster, LD got even cheaper. Today's uucp with Telebit sends 40 times as much news for the same dollar as the original in 1980. Since then volume has grown even more, but the new culprit is NNTP. I expect the number of LD telco links that do a full feed to stay more level or even reduce. Compression can improve no more than about 60% in practice. Modems can't get much better on voice lines. Dynamic feeding can only help non-backbones. We are close to the physical limit on voice lines. But other things, such as switched data service and the internet allow that limit to be surpassed. This is why "imminent death of the net predicted" became so common. People predicted the net would collapse under its own weight in 82 and everybody believed it. But each time a new technology saved it. If there is a gigabit network with bandwidth to spare that is willing to carry USENET, it has plenty more growth left. From rsalz@bbn.com Thu Oct 25 09:02 PDT 1990 From: Rich Salz Date: Thu, 25 Oct 90 12:01:58 EDT To: bjones@ucsd.edu, usenet.hist@weber.ucsd.edu Subject: Re: Technical links and a growing volume Status: R Seismo was "everyone's favorite mail relay" as Rick Adams puts it. It really didn't do large amount of news. At least not as large as you'd think it might. :- /r$ From bjones Thu Oct 25 09:12:00 1990 To: bjones@ucsd.edu, looking!brad@watmath.waterloo.edu Subject: Re: Technical links and a growing volume Status: R >From looking!brad@watmath.waterloo.edu Thu Oct 25 08:30 PDT 1990 > >Since then volume has grown even more, but the new culprit is NNTP. What is "NNTP"? A little bit out of it today, bj From looking!brad@watmath.waterloo.edu Thu Oct 25 10:30 PDT 1990 Subject: Re: Technical links and a growing volume To: bjones@ucsd.edu (Bruce Jones) Date: Thu, 25 Oct 90 12:59:33 EDT From: Brad Templeton X-Mailer: ELM [version 2.2 PL0] Status: R The internet news transfer protocol, developed at UCSD. From bjones Thu Oct 25 11:25:43 1990 To: looking!brad@watmath.waterloo.edu Subject: Re: Technical links and a growing volume Status: R Thanks Brad. I suppose that I should know about NNTP, given that it was developed here. Oh well, it's been one of those weeks. bj From vortex!lauren@rand.org Thu Oct 25 12:40 PDT 1990 Date: Thu, 25 Oct 90 11:37:02 PDT From: lauren@vortex.COM (Lauren Weinstein) Subject: mail relays To: usenet.hist@weber.ucsd.edu Status: R Going back even farther, let's not forget the tremendous mail volumes handled for the net by ihnp4 at IH, and before that by harpo at Whippany. One of my favorite arguments, of course, is that the net has tended only to address the traffic *volume* issues, while as a whole not wanting to worry about the "quality" issues. It's the increase in volume, with if anything a continuing decay of the signal/noise ratio, which has driven many of the "old-timers" (including myself) into much more restricted reading of and participation in the net than some years ago. Another largely ignored issue all along has been the legal ramifications of much of the material flowing around. This has become more acute, with sexually oriented pictures (clearly pornographic under various local and state laws) being moved around on the NSF backbone and passed through and in some cases archived on highly visible sites. Remember that serious charges were recently brought against GTE, even though they are an *official* common carrier, for carrying the now defunct American Exxxstacy (sic) X-rated movie service--even though they were simply a third party carrier with no control over the content. They got out of it by dropping the service from the satellite using an escape clause in their contract--in the process destroying not only the X-rated service but other R and PG/G rated services run by the same firm (in fact, they drove the firm out of business). This was all triggered by ONE politically ambitious district attorney (in Alabama I believe), who decided to apply his local community standards to a national service. The vulnerability of the net in this area, or more specifically of major sites that support the net, would seem to be growing. --Lauren-- From bjones Thu Oct 25 13:00:42 1990 To: usenet.hist@weber.ucsd.edu Subject: Re: mail relays Status: R > >From vortex!lauren@rand.org Thu Oct 25 12:40 PDT 1990 > >One of my favorite arguments, of course, is that the net has tended >only to address the traffic *volume* issues, while as a whole not wanting >to worry about the "quality" issues. While my earlier message came out of my curiosity about technical solutions to the volume of traffic, I am also curious about what kinds of attempts have been made address Lauren's issue -- quality. I know that this was part of the impetus to the renaming but have there been other changes over time that tried to address the quality issue? (Aside from things like kill files etc.) bj - From karl_kleinpaste@cis.ohio-state.edu Thu Oct 25 13:19 PDT 1990 From: karl_kleinpaste@cis.ohio-state.edu Date: Thu, 25 Oct 90 16:18:41 -0400 To: bjones@ucsd.edu (Bruce Jones) Cc: usenet.hist@weber.ucsd.edu Subject: Re: mail relays Status: R Date: Thu, 25 Oct 90 13:00:44 pdt From: bjones@ucsd.edu (Bruce Jones) have there been other changes over time that tried to address the quality issue? 4-line limitation on .signature. It can't prevent the user who reads in his monster .signature by hand, but it stops a _lot_ of trash. "included more text than new text" error to prevent excessive quoting. (I have this memory of Brad chastising Larry Wall fairly severely in 1985 or so for creating F/R commands in rn.) Default Followup-To: to only the first newsgroup specified for many-newsgroup postings; only in the face of no Followup-To: supplied by the user. All B News-only, I believe. That's unfortunate, IMHO. Little things, but they help. --karl From smb@ulysses.att.com Thu Oct 25 14:12 PDT 1990 From: smb@ulysses.att.com To: bjones@UCSD.EDU (Bruce Jones) Cc: usenet.hist@weber.ucsd.edu Subject: Re: mail relays Date: Thu, 25 Oct 90 17:07:17 EDT Status: R >From vortex!lauren@rand.org Thu Oct 25 12:40 PDT 1990 > >One of my favorite arguments, of course, is that the net has tended >only to address the traffic *volume* issues, while as a whole not wanting >to worry about the "quality" issues. While my earlier message came out of my curiosity about technical solutions to the volume of traffic, I am also curious about what kinds of attempts have been made address Lauren's issue -- quality. I know that this was part of the impetus to the renaming but have there been other changes over time that tried to address the quality issue? (Aside from things like kill files etc.) bj - There have been two primary ways to deal with quality. First, moderated groups were created for just that reason. Even with moderators, I'm sure it comes as no surprise that quality still varies widely, depending on how good the moderator is. Second, there has been a return to mailing lists for some news-like traffic. That also acts as a volume filter, simply because the knowledge of the lists' existence is limited. From henry@zoo.toronto.edu Thu Oct 25 14:44 PDT 1990 From: henry@zoo.toronto.edu Date: Thu, 25 Oct 90 17:31:59 EDT To: usenet.hist@weber.ucsd.edu Subject: Re: mail relays Status: R >Second, there has been a return to mailing lists for some news-like traffic. >That also acts as a volume filter, simply because the knowledge of the lists' >existence is limited. This rather brings to mind a syndrome that sometimes occurred when an old Arpanet mailing list got gatewayed to a Usenet newsgroup. After a little while, some Usenet novice would commit some faux pas, a furious Arpanaut would flame violently about how Usenet was nothing but bozos who didn't understand How Things Should Be Done and the gatewaying was a stupid idea, and the list maintainer would quietly observe that the gatewaying had indeed produced problems, but it was also producing most of the real content... Henry Spencer at U of Toronto Zoology henry@zoo.toronto.edu utzoo!henry From thomas%mvac23.uucp@udel.edu Thu Oct 25 17:08 PDT 1990 Date: Thu, 25 Oct 90 19:21:02 EDT From: Thomas Lapp Subject: RE: IHNP4 To: usenet.hist@weber.ucsd.edu Status: R Lauren Weinstein writes: > Going back even farther, let's not forget the tremendous mail volumes > handled for the net by ihnp4 at IH, and before that by harpo at > Whippany. This is showing my net.age, I know, but I started networking over internet around the same time that inhp4 was "retired" from being THE way to route traffic. (I was a BITNETer for several years before that). I had thought that it was a machine at Bell Labs in Columbus, OH. But you mention IH (International Harvester?). Who, or what was, inhp4? - tom -- internet : mvac23!thomas@udel.edu or thomas%mvac23@udel.edu (home) : 4398613@mcimail.com (work) uucp : {ucbvax,mcvax,psuvax1,uunet}!udel!mvac23!thomas Location : Newark, DE, USA Advise: : Be slow to anger, and quick to praise From spaf@cs.purdue.edu Thu Oct 25 17:26 PDT 1990 To: Thomas Lapp Cc: usenet.hist@weber.ucsd.edu Subject: Re: IHNP4 <9010251921.0.UUL1.3#5131@mvac23.UUCP> Date: Thu, 25 Oct 90 19:19:26 EST From: Gene Spafford Status: R inhp4 was at AT&T Indian Hill, Illinois (outside Chicago - Naperville). Some of the other list participants can probably explain this better, but most of the AT&T machines were named with a location code, a function code, and a unique id to differentiate multiple machines. At least, that was the idea, although to an outsider it didn't always seem to be that way. Thus, ihnp4 was Indian Hill Network Processor #4 mh was Murray Hill. ak was the Atlanta Wire Works, sb was Southern Bell, cb was Columbus (Mark Horton was mark@cbosgd for a long time) plus others. Then there were the machines in the lab that had (and have) names like bonnie, clyde, ulysses, research, allegra, lento, harpo, chico, etc. From smb@ulysses.att.com Thu Oct 25 17:56 PDT 1990 From: smb@ulysses.att.com To: Gene Spafford Cc: Thomas Lapp , usenet.hist@weber.ucsd.edu Subject: Re: IHNP4 Date: Thu, 25 Oct 90 20:48:42 EDT Status: R Thus, ihnp4 was Indian Hill Network Processor #4 mh was Murray Hill. ak was the Atlanta Wire Works, sb was Southern Bell, cb was Columbus (Mark Horton was mark@cbosgd for a long time) plus others. Yup, Columbus Operating Systems Group D, as I recall. Then there were the machines in the lab that had (and have) names like bonnie, clyde, ulysses, research, allegra, lento, harpo, chico, etc. Ulysses was once mhb5b for email. But we called it ulysses. Shortly after I got there, I got tired of mhb5c being circe, mhb5p being penelope, but mhb5b not being something with a 'b'. For that matter, I didn't think that Murray Hill Building 5 made a great prefix for a computer name, so I rammed through a name change. Folks with very old pathalias databases will find that the aliases remain. From vortex!lauren@rand.org Thu Oct 25 18:16 PDT 1990 Date: Thu, 25 Oct 90 17:59:37 PDT From: lauren@vortex.COM (Lauren Weinstein) Subject: RE: IHNP4 To: thomas%mvac23.uucp@udel.edu Cc: usenet.hist@weber.ucsd.edu Status: R The BTL Columbus machines, in general, used (and use) the cb* convention for naming. BTL Indian Hill (just west of Chicago) used ih* naming. So "ihnp4" was "Indian Hills Network Processor 4". Its rise to fame was very much the work of Gary M. When he left IH, ihnp4 dropped its major role in the network very quickly. This brings up an interesting point, which is that historically, in many locations, the presence of a major site providing lots of network mail forwarding functions would be dependent on the work of one or two people. When those people leave, the machine in question will often return to relative (or total) obscurity. Another example of this is Brian Redman and harpo at BTL Whippany... --Lauren-- From vortex!lauren@rand.org Thu Oct 25 18:17 PDT 1990 Date: Thu, 25 Oct 90 17:51:15 PDT From: lauren@vortex.COM (Lauren Weinstein) Subject: Re: mail relays To: smb@ulysses.att.com Cc: usenet.hist@weber.ucsd.edu Status: R It's worth noting though, that even though some of us did push hard for moderated groups, we were never able to really get them accepted the way one might have hoped. There would always be the kicking and screaming and totally inaccurate "censorship" cries. I'd find myself spending hours writing messages to try to explain why moderation was important and why this wasn't censorship, etc. I finally got tired of fighting that fight. Sure, there are some moderated groups, but without lots more of them, attracting readers who don't read the unmoderated versions (when they exist) their utility will remain limited. But I still think they can help a lot. As for the return to mailing lists... you're right of course. But it's a sad commentary that we need to rely on "limited membership", virtually "hidden" mailing lists to keep things under control. Take this list for example. If it were an unmoderated newsgroup instead of a limited list, I think we all know what kind of shape it would probably be in by now... --Lauren-- From spaf@cs.purdue.edu Thu Oct 25 18:49 PDT 1990 To: usenet.hist@weber.ucsd.edu Subject: Re: IHNP4 <9010251759.0.UUL1.3#1@vortex.COM> Date: Thu, 25 Oct 90 20:42:39 EST From: Gene Spafford Status: R Lauren's comments about people is right on. "Backbone" sites were usually the result of a specific person at a site where he/she had fairly extensive control over a machine with sufficient capacity to handle news and mail load, and where he actively pursued connectivity. A list of machine/individuals is perhaps worth note (and nostalgia). I'll start it with what I remember: Machine Person Login Rick Adams seismo seismo!rick (later uunet) Larry Auton clyde clyde!lda Fred Avolio decuac decuac!avolio Steve Bellovin ulysses ulysses!smb Scott Bradner husc6 Lindsay Cleveland akgua akgua!glc Erik Fair ucbvax Greg Fowler hplabs hplabs!fowler Ron Heiby cuae2 cuae2!heiby Peter Honeyman allegra Mark Horton cbosgd cbosgd!mark Curtis Jackson burl burl!rcj (later moss!rcj) Robert Lake alberta alberta!lake Gary Murkamani ihnp4 Mel Pleasant rutgers rutgers!pleasant Gene Spafford gatech gatech!spaf Henry Spencer utzoo utzoo!henry Armando Stettner decvax Sid Stuart linus linus!sid Greg Woods hao hao!woods mcnc I'm sure I've missed others, but these were all important folks/machines Who wants to make the next pass at completing this list?