[6] [Editor's Note: In the following written in 1994, ARPANET pioneer Keith Lynch recalls his early days online and comments on some of the challenges to the future of the Net.] History of the Net is Important by Keith F. Lynch kfl@clark.net Well, originally it was just "The ARPANET". In 1977 friends introduced me to it. We used a TI [Texas Instrument]Silent 700 terminal. This was a printing terminal which used thermal paper and built-in 300 baud acoustic coupled modem. One would dial a local "TIP". For instance there was one at Mitre, a nearby company. One would then type "@L134" to connect to host 134, or whatever. There was no TIP (later TAC) login at that time. Host numbers were always a single number of up to three digits. No dots. Host names were always short and uppercase, and also had no dots. A TIP was a machine which did nothing except allow dial-up users to connect to other machines. Later they were renamed TACs. There was no security on them. Not only was no password needed, but you could issue commands to other sessions on the TAC! Everyone was expecting that TAC login was imminent, but it wasn't installed for a long time. Not until 1986, I think. I've heard of guest users being asked not to use a TAC because all its lines were busy who resolved the problem by paying for an extra phone line and modem to be installed at the TAC! TACs had some little-known features, for instance a way to link to a user dialed into another TAC, so you can have a real-time conversation without connecting to a computer. This was handy during hours when guest users weren't allowed to log in on the ITS systems at MIT. If you were both good typists, you could disable echo, so that when either of you typed, only the person not typing saw it. Which meant you could both type at the same time without stepping on each other. A couple times, I would dial into a TAC from a printing terminal at work, and just leave it dialed in. Then, from home, I would tell that TAC port to con nect to an ITS machine. Then, I would get on ITS from home and link to the newly appeared job, log it in, and have it list various files, so that they would print out at work for me. One time I dialed into a TAC from a microcom puter running CP/M at work. (CP/M was a very simple OS for eight bit micros, before the 16 bit IBM PC and MS/DOS came out. It didn't even support hard disks, or tree structured directories.) Then I could connect to it via the net from home. I told my net-friends that we had a machine on the net at work. A machine running CP/M. I showed them how to connect to it, and they did so. This was considered a great lark. I can't easily convey how ridiculous the idea of a small machine on the net was in those days. I think this was in 1981 or 1982, when connection required a government contract and a refrigerator-sized quarter million dollar IMP. The most popular machines on the net were the ITS machines at MIT. There was DM (77), AI (134), ML (198), and MC (236). DM had Zork on it. Zork was a text-only adventure game played in woods, caverns, dungeons, etc, which contained treasure to be brought back. (Infocom later marketed a modified version of Zork for various micros.) MC had Macsyma, a program for solving equations. (Macsyma was later marketed by Symbolics.) All machines had EMACS, the screen editor written by Richard M. Stallman et al, which gave rise to the later commercial EMACS written and marketed by Gosling, and the GNU EMACS again written by Richard M. Stallman, who later won a MacArthur foundation quarter million dollar genius grant for it and for related work. The ITS EMACS was the original EMACS, and was written in TECO, a character-based editing language. ITS stood for the Incompatible Time-sharing System, an obvious take-off on CTSS, the Compati ble Time Sharing System. (Just as Unix is a take-off on the earlier TENEX, TWENEX, and MULTICS.) All four ITS machines also had UNTALK, a split-screen conferencing program similar to the later "talk" on Unix and PHONE on VMS. I was told it was written by a user whose ITS username was UNCOLA and who had committed suicide. I don't know if it was the first program of that type, but it was the first I had seen. ITS was a strange operating system. Commands took effect without one's needing to type . There was a semi-hierarchical file system, suppos edly hacked together in one weekend by David A. Moon. Files on other ITS systems were transparently available through the "Chaosnet" (a predecessor of Ethernet, and probably an inspiration for it) simply by prefacing the filename with the name of the machine it was on. Similar ideas later appeared in VMS/DECNET and Unix/NFS. Eventually (1979?), ITS instituted passwords. Fortunately for me, they allowed guest users. Even without an account, one could get in fairly easily. I'll explain how, as it helps give a flavor for the system: Users who weren't logged in still got a prompt. They just couldn't do much with it. One thing they could do was see who's logged in. Another was use the SEND command to send a real-time message to anyone who was logged in at the time. Anyhow, when one wasn't logged in, one could use SEND to send to someone else who wasn't logged in. The SEND command would then automatically invoke the MAIL command. And from within the mailer one could do "E" to invoke EMACS (just as today in the Unix mail command, one can do "~e" to do the same thing). And from within EMACS, one uses "^X^V" to load DDT (the exec) and "^X^W" to write it over SEND. Then one aborts out, and invokes SEND a second time. Only since SEND had been replaced with a copy of DDT, you'd be in the exec, fully logged in. Unfortunately, the machines (PDP-10s) were usually so heavily loaded that guests were often restricted to using them after midnight. During slack periods, they were allowed on as early as 8 pm. And sometimes all day on weekends. File space was quite restricted. And guests didn't get personal directories. Also, there was no file protection. Anyone could read or alter any file on the system. And anyone could spy on anyone else's session, and even link to their exec and issue commands to it. This is something I really miss in Unix and VMS when a user needs assistance it would be very handy to be able to look over their shoulder and to type commands for them while they watch, remotely. Guests were allowed to, and even encouraged to, modify the system. If people didn't like the modifica tions, they were taken out again. The ITS convention was that it was O.K. to read other people's mail. Eventually, this collided with the net-wide convention that this wasn't O.K., with some unfortunate results, which included at least one di vorce, that of Marty and Nancy Conner, who had married after meeting on the Bandykin mailing list. The Bandykin list was originally set up for the friends of Bandy (Andrew Scott Beals) to console him for the loss of his girlfriend. I think this was in 1984 or so. It was alluded to, not by name, in Quarterman and Hoskins' "Notable Computer Networks" (CACM, October 1986 please don't try to write a history of the net until you've read this paper). It was later renamed to Kin, when Bandy wished to be dissociated from it. Before dying, it spawned off a number of other lists, including Elbows, Lectroids, TANSTAAFL, and Info-Frobkin. That last list gave rise to FTP Software, a thriving Cambridge firm with which the company I work for has recently done business. (FTP Software was presumably named after the net's File Transfer Protocol, which of course greatly predated it.) The Kin list died because Marty Conner reserved the right to add anyone and everyone to the list. The new lists were constituted without him, and with strict rules about who could join. It wasn't until 1981 that I had fairly consistent access from home, using a borrowed 300 baud modem and H19 terminal. Prior to that, I had often gone months or sometimes years between access. After 1981, I have never been offline for more than a month. I missed a month in 1986 due to TAC logins finally being installed. And another month in 1993, when I was installing computers overseas. (Ironically, as of last month those overseas comput ers are now on the net!) In 1982 I got my own Heathkit H19 terminal and assembled it. I used it until I got a 286 PC in 1986. I'm still using that PC. I'm currently using a 2400 baud modem I borrowed from work three years ago. Prior to that, I was using my own 1200 baud modem. Early this year I rescued a TI Silent 700 terminal from the trash can at a hamfest, mostly just for old times' sake. (The TI had been marked $15, but nobody bought it. They cost about $1000 new in the late 70s.) In 1986, I started using a service called PC Pursuit. It allowed one to make off-hours long distance computer calls to about 30 cities in the US, including Boston. I used it not just to get onto ITS, but also onto various BBS systems around the country. In 1986, 1987, 1989, 1989 again, and 1990, I visited MIT in person. In May 1990, the last ITS machine was shut down. But I also had guest accounts on Unix systems at MIT by then. It was one of those on which I first used Usenet newsgroups, perhaps in 1987 or so. Previously, most of my activity had been reading and posting to mailing lists, having real-time chats, and downloading various text files. I recall one four-way real-time chat which included people in Virginia, Norway, the Philippines, and Missouri. In 1991 I switched from using a Unix system at MIT to using Digex, a Unix system in Maryland, a local call from here. Not long after, I dropped PC Pursuit. PC Pursuit was nice at first, but they changed from allowing unlimited off-hours usage to one hour a day, while increasing their rates from $20 a month to $30 a month. Also, their local number was busy most of the time, and connections were sluggish, and frequently punctuated with the notorious "** POSSIBLE DATA LOSS 00 55 **" which invariably meant several pages had been discarded. I probably would have dropped it anyhow, as there were only two long distance BBSs I called regularly, and one had shut down, while the other had moved out of a PC Pursuit area (and has since shut down). Digex was founded, and is headed by, Doug Humphrey, whom I first met in person at a convention called WATS-80 which he hosted in Washington DC in 1980. Oddly, instead of using his real name there, he called himself "Aubrey Philipsz" after a character in James Hogan's 1978 novel The Genesis Machine. I may have met him online earlier. He was DIGEX on the ITS machines. In those days, he had a large DEC-10 in his small apartment. He had bought it for scrap prices. He used to wear the key to it around his neck as jewelry. In 1989 he had an ITS system in his apartment, which was only one of two not at MIT (the other was in Scandinavia somewhere). I don't think he still has it. (I wonder if there's a law against killing an endangered operating system.) I remember his mentioning ARPANET, and how easy it was to get onto it, during a talk he gave at WATS-80 in 1980. The implication was that we were all unauthorized users, but that nobody really minded yet. I don't think Usenet was mentioned at that convention. WATS-80 was mentioned in the Washington Post. I'm sure I still have the newspaper clipping some where. (I always save everything forever, but often have a hard time finding it later, since it's mixed in with everything else I've saved.) As you can see from my header, I'm still on Digex. [That was in 1994 -ed] It's grown a lot since I first logged on here, from a SUN-3 with an "MX record" (not directly on the net) with about 1000 newsgroups, to several large SUN-4s linked to the Internet backbone with a T1 line, carrying about 9000 newsgroups. I still have an account on a Unix machine at MIT, too, which I can telnet into, but I seldom use it. > one of the questions I am most interested in sorting out is > "What was the degree of Usenet/Internet overlap at various > times"? That's hard to answer. I can give you my impres sions. ITS was never part of Usenet. The idea of a newsgroup is a fairly obvious one, given mailing lists. I recall commenting in 1979 or 1980, that it was silly to mail a copy of the same thing separately to lots of people on the same machine, rather than mailing a pointer to it, and having one copy in a common area. In fact, the SF-Lovers digest was set up that way for some users for a while in 1980 instead of being mailed the digest, they had the option of being mailed a notification that there's a new digest, so they can read it from the online archives. This was discontinued after a year or two, probably because it was only practical when most readers were on ITS, which is where the list origi nated. Almost all mailing lists originated from ITS, since it had the most advanced mailer software. Rich Zellich maintained a "list of lists" which could be ftp'd from SRI-NIC.ARPA. For all I know, he still does. But it was hopelessly out of date by 1983 or so, as there was no formal procedures for information on new lists, or on changes in old lists, to be conveyed to him. I gradually became aware of Usenet via references in SF-Lovers, Human-Nets and other mailing lists. It became clear that some people didn't see something called the "SF-Lovers Digest," but instead read something called "fa.sf-lovers". I became aware of what newsgroups were, and that they all began with "net." except the ones which were aliased to an ARPANET mailing list, which began with "fa.". Nothing began with alt. or misc. or rec. or sci. or soc. in those days. Speaking of SF-Lovers, Brad Templeton put the first few years of archives (starting in 1979) on a CD-ROM last year, along with lots of recent SF novels and short stories. My brother has a copy. It's easy to scan these archives, unlike my personal archives which are on thousands of five inch dis kettes, mostly unlabeled, in no particular order. It was fun to see my own postings, older than some current net users, now immortalized in plastic and tinfoil. (I just checked that disc, and found that the first mention of fa.sf-lovers in the SF-Lovers digest was in August 1982, in a message which also mentions net.sf-lovers. I don't know if those were two different newsgroups. I can forward that message to you if you like.) Actually, SF-Lovers didn't begin in 1979. It had an earlier incarnation, whose archives apparently haven't been preserved anywhere. It was shut down after Senator William Proxmire gave the ARPANET his golden fleece award for wasting taxpayers' money, citing SF-Lovers and the wine lovers mailing lists as examples. (I don't know when this was, but it should be easy to look up.) The wine lovers mailing list never came back. Usenet people also participated in mailing lists. They always had addresses in the form foo!bar!baz!zoo!yar!yaz where foo and yaz were the starting and ending points, or perhaps the other way around. ARPANET addresses were always in the form FOO@BAR, or if they were on some kind of subnet FOO%BAR@BAZ. Traffic which had traversed the nets would look like foo!bar!baz%ZOO@YAR. It wasn't always clear which way to parse this. I definitely had the impression that ARPANET (later, Internet) and Usenet were two very different things, and that mail got from one to the other only because one or two machines happened to be on both networks. These gateway machines which were on both networks kept changing, presumably because once word got out that one was acting as a gateway, it quickly became overloaded, and soon refused to act as a gateway anymore. My impression (which may have been wrong) was that the Usenet mailing lists were completely different from the ARPANET mailing lists, although some adventurous Usenetters were subscribed to the latter via a gateway. There was a Usenet map file, consisting of several pages of ASCII line drawings meant to be connected together, which showed all the systems on the Usenet, and which ones talked (via uucp) to which other ones. I may still have a hardcopy of this somewhere. I recall that only one or two machines on the map was also an ARPANET host. But it was hard to tell, since a host's Usenet name and ARPANET name could be (and usually were) completely different. Today, I have the impression that Internet and Usenet are essentially the same thing. And that the overwhelming majority of newsgroup traffic flows via TCP/IP over the Internet, rather than via uucp over dial-up modems. Trying to separate them today seems about as productive as distinguishing the Angles from the Saxons today. I recall that Usenet users were considered some how lower class. For instance there was a message on the Bandykin list suggesting that Usenet people be banned from the list. I wrote a reply, replacing "Usenet" with "black", and "Internet" with "white," showing that "netism" (as I then named it) is as bad as racism. (I'm sure I still have a copy of these messages.) Today, on some newsgroups there's similar, but lesser, netism toward AOL, Delphi, and/or Fidonet users. > And I would love to know about the 1980 ARPANET crash - that's > just after Usenet started (when in 1980 was the crash?) October? I don't recall the cause, except that it came as an enormous surprise, as the ARPANET was supposed to be crash-proof. Some kind of self- propagating host table update had a bug in it, I think. It was definitely an accident, not malicious, not an attempt to crash anything. > Have you seen any history work done on Usenet and ARPANET > history? I don't think so. Not until the past year have I noticed lots of books being available, describing what the net is like now, and how to do things with it. It makes sense that such books would appear before books that describe how it came to be that way, and what it was like earlier, the latter being of lesser immediate practical use. The net's history is very small, measured in person-years. Perhaps 50 million? Compared to about 20 billion person-years of US history, and a similar number of person-years for the Roman Empire, that isn't very much. Thus one might expect one net history book for every 400 US history books. > ... and when it is often written about, the details are > often wrong (when it is written about by the press, etc.) I've noticed that the press tends to be quite accurate, except when they're writing on a subject I know something about. :-) Concerning quoting styles, the ARPANET style was to indent the text being quoted, the Usenet style (which I've long since adopted) was to quote messages with a ">" character at the beginning of each quoted line, and the Fidonet style was to quote messages with the person's initials followed by a ">" character at the beginning of each quoted line. All three styles are now found on all three nets, as are various other styles, many of them nearly unreadable. Often, the ">" is replaced by some other character such as "|", probably to get around software that puts limits on quoted text. The earliest mailing list I'm aware of is MSGGROUP, a list for discussing e-mail and related issues. I've recently seen some online archives of it dating back to 1975, and I downloaded the earliest parts of it as a souvenir. The first digestified mailing lists were SF-Lovers and Human-Nets, which became digestified in January 1980, because the daily volume became too great for the ITS mailer to handle overnight. With digestification came de facto moderation, since there was no automatic software for digestification. These may have been the first mailing lists to be moderated. The first *automatic* digestification, at least among the lists I read at the time, was on the Space Digest. I remember being very surprised by it. This was probably around 1982. I think I first saw smileys in 1981 or 1982. The original one was :-). FTP, telnet, and mail date back to the beginning of the ARPANET, though they changed somewhat when NCP was replaced by TCP/IP (in 1982?). IRC, WWW, Archie, and Gopher are quite recent. I used something just like IRC on the BITNET in 1987 or so, and I'm pretty sure there was no IRC at that time, though there were MUDs. I used something just like a one-channel IRC on an HP-2000 (not on any net) in 1977. I'm not sure when FAQs started, though I'm pretty sure they came from Usenet, not Internet. GIFs, I'm pretty sure originally came from CompuServe. > > (I do hope newsgroups have been, and are being, totally > > archived.) > They were by Henry Spencer at the university of Toronto - but > he gave his tapes last summer to someone who claimed they would > make a CD-ROM > of them ... Make a CD-ROM of the complete archives of Usenet? I believe the current volume is about equal to one CD-ROM per *week*. > But also some of the research I have done in the > past is available from wuarchive.wustl.edu in > directory /doc/misc/acn/netbook I'll get that file as soon as I finish writing this. (I don't want to bias my recollections, and feed back information already in the file to you.) Until 1990 or so, my perception was that the net, or at least my access to it, was likely to go away soon. TAC login was coming soon. Guest users at MIT were always becoming more numerous and weren't as well behaved as in the "good old days," thus were likely to soon all be flushed. The net often became unusably slow (i.e. five or ten minutes for what I type to echo sometimes I'd type ahead a whole session, including the logout, before getting the password prompt) and it was obvious that guests would be flushed since the capacity was now being exceeded. Later came the infamous FCC "modem tax" threat. The outrageous idea was that the net, PC Pursuit, etc, were underselling the phone company and the post office, and that this was unacceptable. Thus, whenever information crosses a state line electronically, it would be charged as much as it would cost to send via a regular modem over a regular long distance phone line. (This was when regular modems didn't exceed 1200 BPS.) This threat later came back as a recurring "urban legend," but it was quite real the first time. Fortunately, the FCC received more letters opposing it than they had received in all history on all other issues combined, so they reluctantly backed down. Packet nets such as the Internet and PC Pursuit are inherently much cheaper than a dedicated phone line. It's like the difference between sharing a lane on the road, and having a whole lane dedicated to you for the duration of your trip. Naturally, the latter costs much more. A dial-up phone line is exactly equivalent to ftping a 64KB file every second, plus another one at the same time in the opposite direction, for the duration of one's session. This "modem tax" would have been an extreme and senseless distortion of the marketplace, roughly equivalent to putting a one million percent tax on trucks driven forwards, but not on those driven in reverse gear. There's long been a lot of commonality between people on the net and people at Science Fiction con ventions (cons). Not only are SF cons discussed a lot on the net, but SF cons have had "@ parties," or "@! parties" since at least 1986. There are also often parties associated with a given mailing list or newsgroup. I'm not sure whether in general people discover the net at cons, or cons on the net, or whether, like me, they discover both independently. Also, either a disproportionate number of libertari ans are on the net, or just as likely the news media are lying to us about how many libertarians are in the general population. There's also a lot of overlap with ham radio types. The net is the exciting electronic frontier that I thought I had permanently missed when reading amateur radio magazines from the 1910s. I used to have a ham radio licence, but let it lapse when I discovered the net. I couldn't combine the two hobbies, as ASCII wasn't allowed on the air until 1980. And packet ham radio came much later. (It's interesting to note that the American Radio Relay League was founded in 1916 by hams to organize networks of hams to relay messages (their own and messages from the general public) across the country, and, ten years later, across the world, using Morse code. It still exists, and I was a member for a while.) An early mailing list was Human Nets. It was for the discussion of "Worldnet," a hypothetical future worldwide computer network. The list is long gone, but I hope the archives are available online somewhere. They'd make valuable reading for you, since by read ing them "backwards" you can get a good image of what the net was like at that time, just as the best way to see what was considered bad about a time and place is to read a utopian novel written then and there, since a utopia is always fairly similar to what the author is accustomed to, with the bad features removed or reversed. One April Fool's Day sometime in the early '80s, there was a hoax posting from KREMVAX, which purported to be a VAX in the Kremlin in the USSR. This was considered quite hilarious, since the ARPANET was for US defense, and the USSR was our enemy. At that time, there were hosts at US bases overseas, but nowhere else outside the US. Much later, after Russia was on the Internet, someone in Russia became aware of this prank, and named their Internet host KREMVAX as a lark. > Thanks for writing. Would you like to say something about > today? Today there are more systems on the net in our computer room where I work, than were on the whole net in 1977. Some of these systems are a single circuit board that could fit in my shirt pocket. And tomorrow? ----------------- [The above e-mail message was written over three years ago. The author's web site is at http://www.clark.net/pub/kfl/. In a recent e-mail message Keith Lynch updated this e-mail exchange: "Note that in 1994 I saw WWW as just another random service on the net, along with Archie, Gopher, and IRC, rather than the 800 pound gorilla it has become. And spam was such a minor issue in those days that I didn't even mention it, while today it takes up the majority of my online time. I believe spam is the greatest threat the net has ever known. I'm against Usenet 2, or any other retreat due to spam. I don't discard any e-mail unread, or munge my address, or cease posting helpful messages to Usenet, or move to Usenet 2, or register with remove lists, or do anything else to surrender any part of the net to spammers, or to imply their legitimacy. Let THEM built a second Usenet or a second Internet. I won't let them drive me off this one. I wish everyone felt the same. I wish every spam to ten million victims was met with ten million strongly worded complaints. We made AGIS back down. We drove Spamford off the net, along with Nancynet, Walt Rines' Quantcom, and a dozen other rogue domains. Spammers are on the defensive now. See http://www.clark.net/pub/kfl/ftc.html for my coverage of the FTC spam hearings six months ago, which I attended. See http://www.clark.net/pub/kfl/ toll.html for my list of toll-free numbers seen in recent e-mail spam. Also see http://www.clark.net/pub/kfl/timeline .html, which should be of interest to every Internet historian."] ------------------------------------------------------------------------------- Reprinted from the Amateur Computerist Vol 8 No 1 Winter/Spring 1998. The whole issue or a subscription are available for free via email. Send a request to jrh@ais.org -------------------------------------------------------------------------------