Chapter Three: History of Electronic Mail
Richard T. Griffiths (Leiden University)
Back to homepage
This section relies heavily on information gleaned from:
- Ian R. Hardy,
Evolution of ARPANET email (1996)
ARPANET developers had always considered information and resource sharing as one of their primary goals. Indeed the construction of ARPANET had been undertaken to that very end (see Chapter One) but it would be fair to say that their horizons were limited to the exchange of scholarly, technical papers and programs via ftp. For example, direct human-human contact had little place in Licklider's original concept of a 'galactic network'... probably because it appeared too trivial an application in a world where machines still occupied whole rooms rather than the tops of desks. Yet, like the internet itself, it was invention waiting to happen. Almost as soon as terminals in different rooms could be linked to the same 'host' computer through 'time-share' operating systems, it became possible to leave messages for one another within the same system. Such applications began to appear from 1961 onwards and immediately proved popular among users. However their limitation was that their use was restricted to the users of a single computer. So why not between different computers in different locations? So, once ARPANET came into being (linking four centres in 1969, 15 by 1971) scientists began to consider sending direct messages over the same medium.
In 1971 Ray Tomlinson of ARPANET sent the World's first e-mail, by adapting an existing, popular, time-share internal mail program and linking it to the new network file transfer technology that underpinned ARPANET's further activities. The first message was simply addressed to himself, sent from one computer to another, with the text 'Testing 1-2-3'. The next thing he did was to address a message to all ARPANET users explaining the availability of 'electronic mail' and giving instructions on how to address mail to another user using the convention - users' log-in name @ host computer name - which is still the basis of e-mail today.
That was a start, but it was still unbelievably crude. The only way to know what was in a message was to open it in its entirety. The messages had to be read in the order they were received. They read as pieces of continuous text (rather like a teletext message today). The reading process and sending process took place via two completely separate programs. What was needed was a system to make the entire operation more user-friendly... a piece-of-cake given the computer expertise concentrated around ARPANET. Innovations followed thick and fast. More or less in chronological order there was developed:
As a result of these developments, by the summer of 1972 (within twelve months of the first message) most of the facilities that we recognise in current e-mail programs were already in place.
It did not take long for e-mail to establish its own particular style. It was more like a post-card than a letter, or perhaps more accurately, more like an office memorandum (with its headings "To:", "From:", "Subject:" and "cc:". Anyway, whatever the analogy chosen, e-mail offered:
The style of e-mail communication was one reason behind the success of the new medium. We are in the 1970s, when the stuffiness and hierarchy associated with society in the 1950s had already been swept away. The 'bluntness' of the medium was no longer seen as threatening but, instead conveyed a feeling of intimacy and immediacy. But there were other advantages. For example, as long as the cost of keeping computer links open were carried by the computing centre, it was a much cheaper medium than the telephone. Moreover, unlike a telephone, one could always keep a copy of the communication. And, finally, although communication was virtually simultaneous, the recipient did not actually have to be present to receive the message.
If you are not familiar with e-mail, the links below give a guide to some of the more common programs in use. Browsers & Mail
E-mail was an instant success.... although, for the first five years, its development was largely ignored, and left unrecognised, by the ARPANET administration. By that time various discussion groups devoted to specific topics (whereby you put yourself on a mailing-list to receive all posts on a particular topic) had sprung into existence... and electronic mail was set to overtake file transfer in the volume of traffic over ARPANET. By the end of the 70s there were 17 groups in existence, by 1982 there were 44. The most popular among the subjects of these early lists were science fiction and Human-nets. The latter, which appeared around 1975, was devoted to the social implications of the e-mail medium itself, and it helps define the moment when the e-mail users began to realise the full implications of the communication tool they were using. But ironically, while ARPANET scientists were beginning to philosophise about the brave new world they were about to enter, they were the only ones with access to it.... a group of elite defence and communication scientists in institutes whose membership of the net was dependent their role as ARPA (sub-)contractors. Before it could reach the broader public, e-mail itself had to be re-invented.
As mentioned at the start of this section, time-sharing (and message exchange) on a single computer had been developing since the early 1960s - the bottleneck lay in repeating the trick between different computers. In 1964 collaborative project was called Multics [Multiplexed Information and Computing Service] started by General Electrics, MIT, and AT&T (http://www.att.com/technology), to construct a time-share program to be implemented on the GE 645 computer. In 1969 AT&T Bell Laboratories withdrew from the project and began work on their own. The operating system they developed was known as The Unix Programming Environment. From the start the system was made an "open" one, in the sense that all Bell employees had free access to it, and so too did universities. This made for several advantages - it guarantied the spread of the system, it provided designers with more feedback on glitches in the system and possible applications, and it allowed university scientists (and their students) to design their own special applications. In a way, universities were forced to gain hand-on experience with the system since while Bell Labs provided the tapes and the manuals for the system, they offered no back-up or support services. The system was widely adopted throughout universities, popular because it was free and because it could be run on computers far smaller than the room-huggers that dominated the ARPA network. In 1981, the spread of Unix run software received a further flip when it was installed as the basic program environment on IBM's mainframe computers (starting with the 3033AP).
Given Unix's 'open' nature of the system it is not surprising that there also developed an equally wide community of Unix-users with expertise in running the system and developing new applications for it. In 1986, computers using the system were able to link-up to each other through 'dial-in' facilities, but the operation was still primitive and painfully slow. The system was never integrated into the 'Internet' since it was incompatible with the file transfer programs eventually adopted. Nevertheless, it did allow mutual accessing of files and programs and it offered facilities for 'computer conferencing'... and computer chess.
Two post-graduate students at Duke University, Tom Truscott and Jim Ellis, were responsible for the concept and early development of Usenet. They had met through a joint interest in computer chess and had designed their own program. They were regular participants at computer chess competitions, which utilised the 'conferencing' facility of the Unix system (with the competition organisers footing the huge telephone charges). It was through these competitions that they had met people from Bell Labs and had worked there in the summer of 1979. It was also in the summer of 1979 that they attended their first Unix-users conference. Returning to Duke they conceived the idea of creating a network for Unix users and started to design the programming to do it. Their ambition was to make a little contribution to Unix software and "hopefully achieve some minor level of fame." In 1979 they had succeeded and sharing the software with friends at the University of North Carolina, they established the first 'Usenet'. Perhaps not surprisingly, one of the first groups established was NET.chess. In January 1980 Truscott and Ellis presented their system at a meeting of the Academic Unix Users Group and handed out a five page 'Invitation to a General Access UNIX Network'. It explained:
The Usenet concept differed from that used by Arpanet. Whereas to join an Arpanet group, you subscribed and were sent the e-mails, in Usenet the messages were (temporarily) stored in the computer and could be accessed at will by the reader - the difference between having a postcard land on your doormat or reading on a pin-board in a secretarial office.
One couldn't really say that after the 'Invitation' things took off with a bang. By mid-summer 1980 usenet like this:
The presence of two of these partners is worth further comment. Bell Labs was important because they agreed to pickup the telephone bills for the operation. This was not pure altruism since Bell was still interested in the potential of the system for its own communication problems. They also continued to help the improvement of Usenet software. The other interesting new partner was Berkley since it was also a member of ARPANET. Indeed, it was only after meeting Mark Horton, a graduate student in Berkley, that Truscott and Ellis became aware of the developments in ARPANET. Horton began feeding ARPANET news items (labelled FA.) into the Usenet network, which at least generated some traffic and rationale for the infant network, including the popular sci-fi and human-net groups (though on a read-only basis... the traffic from ARPANET and Usenet was still only one-way). As the bugs were ironed out and the traffic increased, so the expansion of Usenet began to accelerate.
Figure Two: Usenet Map June 1981
By the end of 1981, 150 computers were linked to the network and within a year the number had swollen to 400. Whereas initially Truscott and Ellis had anticipated two or three messages being transmitted daily, the number had increased to fifty a day by the end of 1982. By that time there were more than fifty newsgroups in existence. The contrast in content with Arpanet was noticeable. Whereas 67 per cent of Arpanet's groups were devoted to computer/technical topics, this accounted for only 25 per cent of Usenet's groups. 21 per cent of Arpanet's groups dealt with 'serious' discussion topics (most of which were fed through to Usenet as well) leaving 10 per cent as 'fun' groups (including chess and science fiction). By contrast, fully half of the Usenet groups could be described as 'recreational' and some (net.jokes) downright frivolous.
Usenet now began an explosive growth, as shown in the following data:
Usenet operated as a series of (big) linked computers whose "wizards" (owners is wrong word since these were still universities or research institutes) agreed to pass on all 'post' sent to them to the next recipient (immediately) and so on through the 'network'. Obviously, they also agreed to make them accessible to its users. The Usenet network was subsidised (often unknowingly) by these larger machine owners, who picked up the phone-bill. Around 1983, within the network, there developed a 'backbone' of reliable, larger computers through which most of the traffic flowed (if only because it was likely to arrive earlier than through other routes). The wizards of the backbone sites - a group of people using the lists and developing software - were in regular contact and became known as the "Usenet Cabal". It was the cabal that decided what would be carried, and how it would be catalogued.... ie. what groups should be created. BUT for the first years, with a limited community, there was little need for censorship.
In the history of newsgroups, the great renaming which took place in 1986/87 is the next main step. By now there were 241 news groups divided into three categories:
There were two
Second, it cost increasing amounts of money to route these all through the Net. This applied especially to the overseas links from North America which were established in 1982/83. The link to the UK was carried by through Centre for Seismic Studies, North Virginia (vax135); that with mainland Europe through a Philips Digital Equipment Corporation (decvax) in the USA to the Mathematics Centre (mcvax) at the University of Amsterdam (at $6 per minute) and that to Australia also from DEC to University of Melbourne (munnari).
The crux of the matter was one of cost. As one contributor expressed it already in 1982:
Thus, the decision was taken to rationalise the information flows (although this did not always pass without resentment). New categories added between July 1986 and March 1987 were:
This enabled some universities to drop the 'talk' and 'social' groups and for several years these groups and 'alt' (see below) were not transmitted to Europe.
The renaming exercise had seen the gods mingle with the mortals. Accusations of censorship arose... and not only among the netusers. Some of the wizards also thought things had gotten a little out of hand and sought to reconstitute a 'Usenet' outside the 'backbone' for this they created the stem:
The word was intended to represent 'outside the system'. The first group was alt.gourmand, introduced because the creator didn't like mod.gourmand or rec.recipes. Alt.sex and alt.rock-n-roll soon followed.
The Great Renaming represented the height of the Cabal's power, and also marked the beginning of its decline. The cabal was forced by the 'great renaming' controversies to democratise their naming procedures, which loosened its control. Moreover, the availability of new routes (particularly NSFNET - named after US National Science Foundation - backed up by five super-computers) swiftly removed the bottleneck in newsgroup traffic. And finally, in 1987, UUnet was set up - a non-profit-making news carrier, which had every incentive to carry and distribute the alt. group and others.
The Great renaming established the format of Usenet newsgroups that we know today. It was initially defined by the programming needed to send and receive it, but by the early 1990s this 'functional' definition ceased to be applicable and much of the news was transferred over the Web. It was carried to an ever wider public - first as internet access spread over wider sectors of the academic community and later as the Web opened up. By 1990 the number of newsgroups exceeded 1000 for the first time. It was also the year in which alt.sex became the most widely read group on the net. Four years later, the usenet exceeded the 10,000 mark. There was now a further fillip to the development. As the WWW became more user friendly, so commercial providers began to enter the scene in a big way. By 1993, the traffic on usenet spanned the globe (though heavily concentrated in the US and Europe).
Figure Three: Usenet Traffic Flows 1993
In 1995 Internet access providers as Compuserve, Aol and Prodigy all started business. The result was an explosion of news groups, not only into the absolute minutiae of utter trivia (always someone else's hobby) and, recently, by the proliferation of foreign language groups. Whilst the Web is supposed always to have been international, the fact is that for much of its history the lingua france has been English. This now changed and virtually every language group has its own collection of newsgroups:
What a great aid to language learning! By now, the number of groups must be close to 100,000. And, on a personal note, on January 22nd 1998, much to my delight, some enterprising fan created the newsgroup it.sport.calcio.fiorentina.
E-mail and Usenet have sparked immense changes in the social space of those who have experienced it...and will continue to do so. Unfortunately it has little to offer the professional historian. There are some newsgroups, but the discussion is not always at a high level (well, they are open access groups!) and, if you are thinking of using any material you might locate, don't forget that any message is only posted for a limited duration. The following groups, however, might be worth a visit:
Most of these are 'fun' sites. For the serious stuff (perhaps too serious) you have to go to the relevant lists.
As mentioned earlier, a newsgroup places all the news on a serve and the reader selects what he or she wants to read. In a list, specific items of news are sent to all readers who have indicated previously that they want to receive it. From the moment that e-mail facilities allowed users to store addresses and to forward messages to everyone on a specific address list, it was possible to construct lists. Arpanet's 'newsgroups' were really lists and many other networks that emerged in the 1980s also had people that ran lists devoted to specific subjects. The drawbacks lay in the fact that all these operations had to be performed manually, making the success of a list contingent upon the diligence of the list's moderator. Added to the human factor, lists were prone to all the normal frustrations associated with slow/failed connections, especially on the oversaturated transatlantic cables. Improved infrastructure would remove the second of these problems, human ingenuity the first.
In 1985 Eric Thomas (an American computer student working in Paris) designed the first automatic readdressing system but it was slow, primitive and allowed no interaction. Moreover, the initial contacts were all through human agencies. Within a year, however, he had added a user-interface system (including passwords and security checks) and could process all the requests that the previous machines were performing manually. Moreover, the interfaces were duplicated (so, with a machine in Europe, messages only had to cross the Atlantic once, to an American mirror). The system became operational in June 1986 and by the end of the year it was serving 41 lists. Eighteen months later there were 1000 lists on the system.
The system was continuously modified as demand increased. One major blip worth mentioning as the 'IBM crisis' around early 1990 as universities and other research organisations began getting rid of their (large, unwieldy and increasingly out-of-date mainframes) and thus cutting off their list managers in the process. THUS decided to go commercial and develop new software that was not dependent on the IBM operating system. In June 1994 this was ready.... new version made available in May 1995 linked software to Windows environment. The result is that Listserv today supports 23,000 public lists and almost 100,000 public lists delivering in a day 22 million messages. There is a useful entrance to most of these through: http://liszt.com/
For us, the most interesting development is the decision in 1995 of the National Endowment for the Humanities to fund a listserv program (including free software) for the Humanities in the form of H-Net which now hosts over 100 history discussion groups with a total of 60,000 members (0,1 per cent of all list-serv members)! These are accessible through: H-Net
Note, by the way, that economic history lists are not taken up in alphabetical order but appear at the end of the list! Each list will give you instructions how to subscribe, and many allow you to do so from the H-Net site. You will often be sent an on-line form to fill in and it might take from a day to a week to join. And follow the instructions (including full stops and spacing) meticulously... the message is being read by a computer, not a human!
These lists cover a wide range of historical subjects, but they vary in their usefulness and relevance. What most of them contain are:
It is the last two that makes them most interesting for history students. The reviews are far more descriptive and detailed than those found in academic journals and since debates are conducted among academics many are held at quite a high intellectual level, although often on esoteric subjects. But you don't have to read them all. Most discussion chains carry the same subject descriptor and so you can always delete the less interesting ones. As in most lists (academic or otherwise), the best advice is to 'lurk' for a while before participating actively. You can also get an indication of the type of discussions by visiting the group's archives. Get a feel for the 'style' of the group and the expected level of response... and don't get upset if some of the replies are not exactly sympathetic. If you decide to leave a group, don't forget to 'unsubscribe', or else your mail-box will soon overflow. And don't forget your netiquette.
Netiquette is nothing more than etiquette for the net. It and its first appearance in 1986 and attempted to set out the rules of acceptable behaviour. There are plenty of netiquette sites to be found via Yahoo but they are all typical to the one linked above... and they all read a bit like advice columns in the pages of the more staid women's' magazines. Still, some of the tips are not immediately obvious:
To flame is to express an immoderate opinion and it can easily lead to flame-wars... which degenerate into a slanging-match, much to the disgruntlement of other readers of the group. And they have ways of getting even. You don't want to get bombed (have your private e-mail address deluged by repeated messages until your box is full).
So it might makes sense to read one of those on-line manuals after all.
Back to homepage
Last update: 11 October 2002