APL and e-mail

[Roger|Dyalog]

[Hui and Kromberg 2020] says:

An important factor in these personal relationships was the worldwide e-mail system Mailbox (666 box) extant at IPSA since the early 1970s [Goldsmith 1980; Goldsmith 2010]. Mailbox provided for discussion groups on language design, programming questions, newbie help, etc. as well as person-to-person exchanges. ... IBM developed its own e-mail system after a key developer (Jon McGrew) used Mailbox and consulted with its implementer [McGrew 2016].

The cited references are interesting and deserve to be better known. [Goldsmith 1980] is the paper by Leslie H. Goldsmith, Corporate Communications Using the SHARP APL Mailbox, 1980 APL Users Meeting Proceedings, 1980-10. [Goldsmith 2010] is from APL Quotations and Anecdotes:

Lower Canada College in Montreal was fortunate enough to have been introduced to APL when John Brown, then head of the Mathematics Department of the school, met Ken Iverson at an IBM conference in the late 60’s. John excitedly brought IBM’s APL to the School, and this was later replaced by a connection to SHARP APL (then APL Plus). Everyone used the same account number, and so had access to everyone else’s information. Some people, including myself, started to reason about how one could protect workspaces and files (when these were released in 1970) from prying eyes. We started to think beyond (workspace) locks and (file) magic numbers, and this led to techniques for monitoring application security, trapping snoopers, encrypting data, hiding programs, etc.

The first Mailbox (the term “e-mail” hadn’t yet been coined) was developed by Larry Breed at STSC during a period in time when IPSA and STSC collaborated on many language and environmental features. Larry’s Mailbox was actually the world’s second electronic mail application that I know of, the first having been written in 1971 by Frank Bates III of Mobility Systems. Frank’s application demonstrated the efficacy of communication through this medium, but was not a robust, commercial application and was not secure. In early 1972, Larry set out to build something that was.

There was something appealing about trying to break into a messaging system. For one, you can learn a lot trying, and even more when you’re successful. Additionally, Larry had introduced some novel changes into the APL interpreter to allow name masking, and this made the allure still greater—even mysterious. After having penetrated numerous other applications running on the mainframe, it wasn’t hard to convince myself to try to break into the Mailbox. I did this successfully on many occasions, using a variety of techniques that I developed. Each time, I would report what I had done to David Keith, branch manager of the IPSA Montreal office, and eventually that hole would be plugged and I’d move on to find a new one. Years later, Larry told me he was late for the annual STSC Christmas party in 1972 because he was busy fixing my latest reported vulnerability.

One Saturday in 1973, I recall being at the School browsing through everyone’s messages when I came across one written by David Keith about a confidential sales prospect. The message was addressed to Ian Sharp and a few senior managers in the company. It went on for some time about the nature of the opportunity and the likely next steps in securing it. In conclusion, David reiterated the confidentiality of the matter and, almost prophetically, added “that goes for you snoopers at LCC, too!” It was a mirthful moment.

A couple of months later, David Bonyun of I.P. Sharp, then System Librarian and in charge of all public libraries, contacted me and asked if I would write a new Mailbox that was functionally similar to the existing one but “Leslie Goldsmith proof”. Doing this would certainly bring to bear all that I had learned about trying to protect data at LCC.

My first version of 666 BOX was written as a part-time activity in the late summer of 1973. IPSA did not have access to the source for the version they had been running, so it was written from scratch. Although it underwent many significant changes over the coming years, dramatically increasing its capacity and later introducing cross-domain message transfer, I’m not aware of any successful attempts
to penetrate it.

—— Leslie Goldsmith

The immediately following anecdote is related:

It came time to negotiate the remuneration for the mailbox work. Ian Sharp offered a fixed sum; Leslie Goldsmith countered with a royalty approach of one penny per message. The negotiations demonstrated the financial acumen of both parties: if Leslie had prevailed he would eventually have received several orders of magnitude more than the fixed sum.

—— Roger Hui

[McGrew 2016] is the paper by Jon McGrew, Forgotten APL Influences, APL-Journal, volume 35, numbers 1 and 2, 2016. The following is the section on e-mail from this paper:

Although there are some differing opinions as to when the first email system appeared, credit has at times been given to an MIT inhouse system, which was developed in 1965.[32] This was a very limited system, intended for use just within their own campus. The other contender is a system created by Frank Bates III of Mobility Systems, written in 1971.[34] These were both very limited systems, with neither one handling any commercial usage.

A discussion of “Internet in Its Infancy” [32] tells us that “By the end of 1971 there were 23 computers at 15 different locations connected to the ARPANET … The following year [1972] the first true email software was written, and email rapidly became the most popular application on the network.”

This first “true” email system was created by Larry Breed, one of the founders of Scientific Time Sharing Corporation (STSC) in Bethesda, Maryland. He created it in 1972, and it was of course implemented in APL. This very successful email system was described as the “APL*PLUS Message Processing System Mailbox,” or known to their customers simply as “the Mailbox.”[33]

In that same Internet history article, the following statements were made about the STSC mailbox:

“In 1976, email was first used to gain political power. Jimmy Carter and Walter Mondale used email during their U.S. presidential campaign to co-ordinate their schedules. They won the election and Carter be came a strong supporter of the internet. Each of their emails cost about US$4 to send.”[32]

White House Press Secretary Jody Powell talked about how they used the STSC Mailbox to help them with Jimmy Carter’s successful 1976 U. S. Presidential campaign, saying:

“One of the constant problems in a political campaign is caused by the fact that things happen in rapid succession. The candidate tends to be one place, most information —the ‘good’ information—tends to be another. We of course used the Mailbox system to get that information and have it available when the candidate and travel party needed it. It saved us some time, it saved us some money … I think you’d have to say, all in all, that it worked out pretty well.”[34]

This same email system was also used by I. P. Sharp Associates (IPSA) in Toronto; these two “friendly-rival” APL timesharing companies shared the same code for their email. Leslie Goldsmith at Sharp rewrote it to incorporate greater security. That new facility was then known to the Sharp APL customers as “666 BOX”.[34,35]

So, with just one or two very limited, inhouse systems preceding these early APL mail systems, I can’t quite claim that APL actually “invented” email, but in a practical sense, maybe that is the case: APL systems offered their users the first “real,” widespread, global email service. And all of this was available to APL users before the term “e-mail” had even been coined.[36]

I had purchased a couple of APL terminals to use at home (…yes, terminals, not PCs). Starting in the mid-1970s, I was a guest user of the Sharp APL system, and had an email account there, and I thought, this is terrific—we should have something like this. So although I never saw any of their code, I created an email system[37] at IBM based on the ideas that I got from the I.P. Sharp email system.

Email was still quite rare in these days, so following the Sharp system, mine ended up also being another one of the first email systems (but obviously not the first)— and of course, all written APL.

Now, since I am not an authority on the STSC or IPSA mail systems, let me tell you about the follow-on system that I created. In support of our timesharing system, I had a great manager at IBM, Bill Davis, who gave me the freedom to work on the projects that I felt needed to be done, so having seen the IPSA mail system, I decided email would be a good thing to have.

It started as a personal project in 1979, and I released it to the world in the summer of 1981… and by “the world” I mean that on Day One, our users in seventeen countries had email. They were all running on our mainframe computers there in Kingston, New York, so we had people in Japan, Australia, Switzerland, Germany, France, England, Argentina, Brazil —all around the world— logging on to the Kingston system to do their daily work, and now for the first time, also to get their email.

Before I put my email system together, I went to Toronto and met with Leslie Goldsmith to ask him what I should be really focusing on if I create my own email system. He said the Number One Priority needs to be security—he said that they have competing companies on their system, all using email, and they have to know that it is secure, that nobody else can see their email. So that was one thing that I made a big point of with my system, to ensure that every thing was very secure and that nothing could get misdelivered. There were even some new features added to the APL interpreter itself to ensure the security of the messages.

Now, flash forward to today and look at our current email providers: How secure is email these days? If you were signing up for this conference, you might go on to a website with your credit card number to reserve a hotel— that may be fine. But would you send it by email? I would suggest not.

A lot of different groups within IBM used our email system, but I was particularly interested in hearing that IBM Headquarters in Armonk became interested in this system and adopted it as their method of sending out all of the IBM Corporate Communications announcements around the world. So all of the new product announcements and executive promotions and IBM earnings statements, and so forth—everything that went up on the (physical) IBM bulletin boards worldwide—came through my email system.

All of these various APL-based email systems predated CompuServe, just in case anyone still remembers that (limited email in 1989 and full support in 1992), AOL email (1991), and Yahoo (1997). It greatly predated Google (1998) and Google’s Gmail (2004), and of course it predated personal computers.

It also predated the commercial Internet (early 1990s). Sharp had their own IPSANET; so how did I send email around the world at IBM? Well, IBM had its own global network, called VNET [38] (mid-1970s), some portions of which still exist, but of course the Internet has taken over for a lot of that now. An attempt to standardize email formats came in September of 1973 from the Internet Engineering Task Force (IETF),[39] but it was then nearly a decade until SMTP (Simple Mail Transfer Protocol) was introduced as the standardized format for email messages, using the now-familiar “@”-symbol in email addresses (August 1982).[40]

All of this came a decade after many APL users had already been using global email.

I’ll emphasize again that of course I realize that I was certainly not the first one to create an email system; I don’t claim to be, and I don’t even claim to be the first one to create it in APL code. I commend Larry Breed and the others who got the APL world online with email.

The point of this discussion is that APL was very early to the game with email. In fact, it was early enough that with the first release of my system I had to explain to people what “email” was. Some people said to me, “Of course I know what ‘mail’ is, but what is the ‘e’ part? …Does that somehow make it different?”

Yes… yes, it does….

A final anecdote, from years later:

At the banquet of the 2014 Jsoftware Conference I sat at the same table with Leslie Goldsmith and a young colleague (and others). Well into the evening the young colleague asked Leslie, “Do you program?” (I suppose there was some doubt because Leslie was now a senior executive of a company.) I quietly said that Leslie programmed in APL one of the world’s first e-mail systems. Leslie added that it was the world’s first commercial e-mail system, and that he did it while a high school student. The young colleague exclaimed, “Now you are starting to impress me!”

—— Roger Hui