Jay Akkad


The History of Email


I.                   Introduction

In 1994 while majority of the world was still learning how to spell “Internet”, there were a few technophiles like Bob Taylor, Ray Tomlinson, Abhay Bhusan, Jim McQuillian and several others who were already celebrating the first 25 years of the existence of the Internet that they had helped build.  This paper aims to provide a brief history of the invention of arguably the most useful by-product of the internet, E-mail.    

II.                BBN Technologies

This past summer I had the amazing opportunity to intern at BBN Technologies, the famous technology firm founded by two professors and a graduate student from MIT, Bolt, Berenak, and Newman.  In their book “Where Wizards Stay up Late, The Origins of the Internet”[1] Katie Hafner and Matthew Lyon refer to BBN as the “third university”.  Located in Cambridge MA, a stone’s throw from MIT and Harvard, BBN has housed several renowned scientists (wizards) and been responsible for a wide array of technological innovations.

One of their more famous “wizards”, Ray Tomlinson, credited for writing the first email program, and sending the first message between two computers can still be found in his office on the fourth floor.  When I met him, I asked him the same question that every mesmerized visitor asks him, “Why did you choose the @ sign in the email?”  While I was too awestruck in his presence to remember the answer, Ian Hardy in “The Evolution of the ARPANET email”[2] quotes him as saying “I got there first, so I got to choose any punctuation I wanted … I chose the @ sign because it was the least used symbol on my Model 33 Teletype.”

III.             Early Electronic Messages

Before Tomlinson wrote his celebrated email program in 1971, there were several mail programs in existence.  In the late 60s computers were very large, expensive, and thus extremely rare.  Researchers had instituted the concept of time-sharing where various users could log on to a computer from a terminal and share processing resources available on the computer.  Within the time-sharing system electronic messages were sent to the various users.  A mail program allowed a user to put a file in another users’ folder, but did not allow anyone except the owner of the folder to read or delete any files.  This ingenious system however, was limited to sending electronic messages to users who had access to the same computer.  

IV.            Invention of E-Mail

In early 1971 Tomlinson had been working on CYPNET, a file transfer protocol (FTP) for the ARPAnet.  CYPNET allowed a user to send a file from one computer to another using the ARPAnet backbone.  Tomlinson decided to write a simple hack, SNDMSG, where the file that was carried from one computer to another was in fact just a mail message.  While SNDMSG wasn’t too technologically challenging, it was “culturally it was revolutionary.”[1] When asked why he wrote the email program when no one had asked for one he is quoted on BBN’s website as saying “Mostly because it seemed like a neat idea.”[3]

In July of 1972, Abhay Bhusan, a professor at MIT was writing the final specs of the ARPAnet file-transfer protocol.  Upon the suggestion of his colleagues, he added Tomlinson’s email programs to the final product.  Tomlinson’s program was an immediate hit.  A report commissioned in 1973, only a year after Bhusan had included it in the FTP specs, showed three quarters of all traffic on the ARPAnet to be email!  The report was commissioned by then ARPA director, Stephen Lukasik, who himself was a big fan of email.  He was known to carry his 30 pound “portable” computer and check mail on a regular basis.  He was the first high ranking person within ARPA to use email for business.  He insisted that everyone communicate with him over email and thus incorporated e-mail as a means of communication within the ARPA culture.

By 1975 there were hundreds of researchers using email over the ARPAnet.  As the load on the network increased, the weaknesses of the system were highlighted.  The routing protocol which was written by Tomlinson’s colleague at BBN, William Crowther, was very simple and elegant, but too primitive for heavy traffic.   Jim McQuillian, another wizard at BBN realized that as the network grew, Crowther’s algorithm would be unable to provide upgrades to the routing algorithm fast enough and would cause network congestion. 

V.               Standardization

With the growing popularity of email, it was obvious that email couldn’t just be a program that piggybacked on FTP. A separate mail-transfer protocol was needed.  In the early 1970s, the ARPAnet was a network for researchers to experiment and write new protocols.  But for a program like email to work, a universally accepted protocol was required.  Bhusan, in 1973, led an ad hoc committee to bring some order in the anarchic world of ARPAnet research.

All data packets in the ARPAnet needed headers that contained some crucial information for routing.  Email required additional information.  Machines on the network ran into problems when parsing the headers due to the lack to standardized headers.  This problem grew with increase in the number of different mail programs and the number of connected computers.  There was a lot of debate over what information was to be stored in the headers.  Some users advocated for very little information like sender, recipient, date, and time.  There were others who often included a great deal of other information like the job number from the processor, the security level, the number of spelling errors etc…  There was even debate over how the email address should be structured.  

For example, in Multics, an operating system used by some in the 1970s, the ‘@’ symbol indicated a kill line symbol.  This meant that any set of characters followed by the '@' sign were ignored by the Operating System.  Tomlinson however used the Tenex operating system built at BBN where the ‘@’ sign wasn’t used for anything and thus had decided to use it in the email address name to concatenate the user with the host server.  There were various suggestions on standardization of the formats and protocols for email.  RFC 561, 680, 724 and 733 are just a few of these examples. 

VI.            External Threat

In 1976 there were 98 nodes on the ARPAnet.  On average, each node sent out ten thousand email messages that year.  While this didn’t come close to the fifty billion pieces of first class mail handled by the United States Postal Service (USPS) that year, e-mail’s steep growth curve didn’t go unnoticed.   IBM came out with MESSANGER and the Computer Corporation of America came out COMET, both products to provide email services for the private sector.  In 1976, the assistant Post Master General was quoted as saying “We are being bypassed technologically.”[1]  In 1979 President Carter was supporting a USPS proposal to offer limited electronic message service where the messages would be transferred from one post office to another electronically, and then would be brought to the consumer’s address by the mailman.  This threat brought the research community together and they along with the US Justice Department and the FCC opposed any plans of government intervention in e-mail.  They all successfully lobbied for E-mail to be left up to the free market. 

VII.         SMTP

With the continuation of the growth of e-mail, in the early 1980s new research produced the simple mail transfer protocol (SMTP).  Its objective as stated by Jonathan Postel, in the RFC 821 was to “transfer mail reliably and efficiently.”[4] The most important feature of SMTP was its ability to relay mail across a variety of transport networks.  

The simplicity of SMTP has greatly contributed to its ubiquity.  It follows a client server model.  The objective of the mail client is to establish a two way communication channel with one or more servers when an email is to be transmitted.  The server could be a destination or a relay node.  If the server is a relay node, it would check its routing tables and send the message to the next server.  SMTP includes a formal hand off procedure where the server would respond to each command with a reply.  The relay servers, using any given routing algorithm would eventually pass the message to the destination server where the message would await being picked up from the mail client of the recipient.  Today, more than forty million email messages are sent out each day using SMTP, a protocol developed in the 80s as the RFC 821 and further revised under RFC 2821 in April 2001.   

VIII.      Conclusion

The history of E-mail is mainly one where researchers and government agencies such as ARPA worked together with a common objective: to better communication.  Email and the Internet are not a product of one person, group, or organization but rather the fruit of many brilliant bureaucrats, scientists and engineers working in unison to revolutionize communication technology.

IX.            Works Cited

[1] Hafner, K., & Lyon, M. (1996). Where wizards stay up late: the origins of the internet. New York: Touchstone.

[2] Hardy, I. (Spring 1996).  The Evolution of ARPANET email.

[3] The First Network Email.  Retrieved October 2, 2004 from the World Wide Web: http://www.bbn.com/email

[4] Klensin, J. (April 2001) Simple Mail Transfer Protocol.  Retrieved October 3, 2004 from the World Wide Web: http://rfc.sunsite.dk/rfc/rfc2821.html