Dave Crocker is a principal with Brandenburg InternetWorking, a technology consultancy. A member of the ARPANET research community starting in 1972, Crocker’s work experience includes stints at the RAND Corporation, MCI Communications Corp, and Digital Equipment Corp. He is a 2004 recipient of the IEEE Internet Award for his work on e-mail. As a result of developing this piece, he has started the Web site emailhistory.org, to develop consensus views about milestones in the evolution of electronic mail.
Mail is a simple concept. An author writes a message, affixes a recipient’s address and gives this to a delivery service, which moves the message to the recipient who reads, replies, files or deletes the message.
In 1975, before the days of personal computers, a leading industry trade magazine, Datamation , envisioned the future of electronic mail with an editorial cartoon showing a man at a desk pointing to his “terminal.” The caption read, “This? Oh, this is the display for my electronic junk mail.”
Two years later, in 1977, the maturing e-mail industry got its first newsletter, “EMMS: Electronic Mail and Message Systems.” Then, in 1978, three years after Datamation published its editorial cartoon, I received what is widely acknowledged to have been the first spam message, sent over the ARPANET by Gary Thuerk, a marketer for my eventual employer Digital Equipment Corp. The message went to roughly 400 users, and the community was swift in denouncing this “flagrant violation of the use of ARPANET,” according to a blog post by ClariNet Communications Corp. founder and publisher Brad Templeton.
The evolution of electronic mail development is a story of collaboration filled with personalities, software systems and acronyms. My own involvement in the development of electronic mail began in 1972, shortly after Ray Tomlinson chose the “@” symbol to trigger his new mechanism for transmitting messages from one computer to another. E-mail had existed on individual machines at least from 1965, according to technologist Tom Van Vleck, who was working for MIT at the time. Tomlinson, however, was the first to connect machines. I was an undergraduate studying psychology at the University of California Los Angeles (UCLA) when the ARPANET research group in engineering hired me to do technical editing and user support.
While working on the ARPANET, I discovered the collaborative culture that became the foundation of today’s “open systems” practices. Ideas, documents and software were freely shared within this community, and anyone who wanted to learn and contribute could participate, including me — a psychology student who had never taken a programming course.
It helped that my brother, Steve, taught me computer programming when I was twelve. At the time, the conventional wisdom dictated that, in order to learn computer programming, one needed an advanced mathematics degree. Today, Steve is the board chair for the Internet Corporation for Assigned Names and Numbers (ICANN), which coordinates the unique identifiers across the Internet. More significantly for the current narrative about the early days of the ARPANET, he established the collaborative tone and open availability of the Internet’s Request for Comments (RFC) technical document series, which remains, for many, a cornerstone of the Internet’s open culture.
I found e-mail to be a compelling technical sandbox. It provides opportunities for assisting group collaboration while presenting significant technical challenges. Designing a good user interface and crafting reliable, fast mail delivery require creative research and careful engineering. My first technical suggestions for e-mail were published as an RFC in 1973, a year after I started work on the ARPANET, and I was delighted to see my thoughts taken seriously by the community.
This led to my being asked to help with a two-year effort focused on resolving the confusion created by emerging, independent e-mail formats. The four-person project team, which included Ken Pogran, John Vittal and Austin Henderson, was tasked with developing a single standard. We were spread across the United States, each working at different companies, and we used e-mail itself as our discussion forum, often consulting others from the wider community.
We published the resulting technical specification in a 1977 RFC titled “Standard for the format of ARPA network text messages.” The document established what I believe was the first, formal standard for an e-mail object’s format — the message content itself. Although primarily codifying existing practice, we also seasoned the specification with a few extra features, such as distinguishing the author from the agent who posted the message and allowing a separate address for replies.
Between 1978 and 1982, I was a graduate student studying computer science at the University of Delaware. During that time, I developed the Multichannel Memorandum Distribution Facility (MMDF), a program for relaying e-mail among different kinds of networks, including via telephone calls. The program was eventually funded by the National Science Foundation (NSF), as part of its CSNet initiative. CSNet extended the reach of the ARPANET/Internet to more computer science research and academic sites, and served as the forerunner to a larger NSF Internet project, NSFNet, which helped to formulate the distributed and independent ownership and administration of today’s Internet.
While users and developers of the original ARPANET were readying for the transition to the Internet’s technical infrastructure of today, I collaborated with the wider community again, to revise the format specification. This produced what I believe was the first “Internet” standard for electronic mail in 1982. The standard was developed in parallel with the Simple Mail Transfer Protocol (SMTP), which is responsible for moving these messages across the Internet. The standard has been revised twice since then in order to refine language, remove obsolete sections and conservatively add improvements. Over the years, individuals and committees have added mechanisms for remote access and group discussion services (mailing lists), as well as improving capabilities for managing e-mail services and more. This tradition of selective increments, which preserves a stable operational base and therefore the user environment, has been a hallmark of Internet technical development.
By 1989, the Internet was expanding into the commercial sector, including the addition of connections to commercial electronic mail services, such as MCI Mail, OnTyme, Telemail and Compuserve. The connection to MCI, whose e-mail service I helped create, used an enhanced version of my “MMDF” system.
In spite of the changes over the years, an e-mail from the early 1970s is very similar to what we see in messages sent today. Those messages also contained “from,” “to,” “cc,” “bcc,” “subject” and “date.” However, in the 1970’s, messages were strictly text. Today, they can include pictures, Web pages, music, video and more. While there were ad hoc methods for including these, a standard for it was not developed until the early 1990s — around the time the Internet went global.
Beyond the similarity of message appearance, even more remarkable is that the e-mail service we rely on today has been in continuous operation since those early days. There have been many incremental improvements, of course, but the basic service, allowing users to exchange mail across the extent of the ARPANET/Internet, has not been replaced or interrupted in 40 years. It simply grew from a couple hundred users to a couple billion.
Besides more users, there is also more mail per user. In the 1970s, it was extraordinary for a person to receive as many as 50 messages a day. Today, it’s not unheard of for an individual to receive 1,000 messages a day. Although basic e-mail requires only simple technology, it is challenging to scale e-mail and make it an essential service for a large pool of users. Doing this requires adding functionality, designing a good user interface (usability), and providing non-stop reliability while delivering messages quickly.
Today, the e-mail equivalent of the U.S. Postal Service is provided by independent, interconnected private providers, such as your Internet access service and your employer. The typical machines that are used often process many millions of messages per day, delivering these messages within minutes or seconds and suffering only a tiny fraction of failed deliveries.
In terms of core e-mail service, I find that the most difficult design work is in making user interfaces that are intuitive for long-term use with a large archive of messages, although spam and phishing have created their own, daunting engineering challenges.
Larry Roberts received many more messages than most during the early days of ARPANET. As head of the ARPA office overseeing computer research, including creation of the ARPANET, Roberts wrote a basic program called “RD,” which permitted the selective handling of individual messages. However the user’s burden of creating replies remained the same as the one needed to create an initial message.
John Vittal overcame this burden with a re-coded version of Robert’s program, which he called “MSG.” It included a command for replying to a message, similar to what is used today. My impression at the time was that, as soon as this command became widely available, e-mail use exploded. As one of MSG’s early users, I found that it permitted quick, convenient, and sustained conversations similar to the conversations we have come to expect over e-mail today.
By the mid-1970s, other user-oriented e-mail programs arrived on the scene. Two of the more popular examples were “Hermes,” at Bolt, Beranek, and Newman, now BBN – a wholly owned subsidiary of Raytheon — and “Laurel,” which was in use at Xerox PARC. “Laurel” is widely acknowledged within the tech community as the first graphically-oriented e-mail user interface. The program also had the split client/server mode in use today. During this time, Albert Vezza, formerly the associate director of MIT’s Laboratory for Computer Science and chairman of the World Wide Web Consortium, led his team at MIT in building a system that ran on top of a database engine. My own “MS” system was developed while I was at The RAND Corporation. The program was later, re-cast to “MH” by RAND’s Bruce Borden and further developed by Marshall Rose at the University of California, Irvine. “MH” became the most commonly used e-mail client software in the technical community and enhanced versions are still in use. While the role of these systems was to interact with users, “sendmail” was developed by Eric Allman at University of California-Berkeley, for providing transfer and delivery services similar to MMDF.
E-mail is the product of collaboration among many innovators, each incrementally advancing the technology. In my experience, I have seen no controversy about the nature or origins of e-mail prior to recent articles in Time and The Washington Post, and the Smithsonian National Museum of American History’s decision this past February to acquire documents relating to the creation of a more localized program called “EMAIL.”
The reports incorrectly credited its author, a 14-year old in the late 1970s, as the “inventor” of e-mail, long after it had become an established service on the ARPANET. He called his system “EMAIL” and later copyrighted the software. Unfortunately, the articles neglected the contributions of individuals widely regarded as the true e-mail inventors. Many are still active, some — too many to name — even took the time to review this piece. The Post has already acknowledged its error and I appreciate the publication’s affording me this opportunity to provide a personal account of that time and that community.