Did My Brother Invent E-Mail With Tom Van Vleck? (Part Two)

This is part two of a five-part series. Parts 1 | 3

2.

THE INVENTION OF E-MAIL

I asked Ben Berman, a Harvard student who works in my office and is a computer whiz, to get on the phone with me and Van Vleck. I was afraid that there were too many arcane details, and I wanted to understand more fully what Tom and my brother had been working on. If you could dig into the CTSS code itself, could you determine who wrote what, when?

BEN BERMAN: I’ve been looking over the complete Multics source code that is dated as late as 1973. This is the batch that’s brought up on the Web site. And I’m finding a lot of comments and how people would sign off in the code. How did you indicate that the piece of code was written by you and how did Noel? How did members of the team do that?

TOM VAN VLECK: Noel’s code might be indicated just by his initials and a date, or initials, date and a comment. Before e-mail was invented, everybody logged into the time-sharing system or submitted batch jobs by indicating two items of accounting data used by MIT. One was the problem number, and that was a number like T0234 or M1416 or something like that. Then programmers were assigned a programmer number, which increased in sequence. The lowest numbers were the first users of the machine. Noel’s programmer number was 3187 and mine was 2962. I still remember those, and a lot of other people’s, too. It was just something you just knew. Before e-mail, if we wanted to leave messages for each other, we would just put a file somewhere that the other person would notice and read. And then when we invented mail, you still had to say to whom you were sending the mail by giving a problem number and a programmer number. At the time, you had to know the mail address of the person you were sending it to, just like you do now. You had to know some arbitrary string of letters that would get the mail to them.

ERROL MORRIS: What were your arbitrary string of letters? And Noel’s arbitrary string of letters?

TOM VAN VLECK: Well, let’s see. Noel’s string of letters would have been a problem number and a programmer number, because that’s how we started. So, I would send mail to T0234, that would be the problem number. 3187 was Noel’s programmer number. [In a Scientific American article on time-sharing, the authors provide a print-out of all the people on the system on May 27, 1966. The print-out reproduced in the article just happens to include my brother’s name. It is like a photograph, a snapshot of moment in time. It is a portrait of the community of users that were assembled around the mainframe.[14] At the bottom of the page, the article also includes an example of CTSS mail. [15] ]

You just kind of knew everybody’s programmer number that you needed to correspond with, or you could look it up. But for most people, you just knew it. You may want to at some point dig around and get a copy of the CTSS source code. Some of it has Noel’s name in it.

BEN BERMAN: Is there a particular strategy for figuring out what was written by whom? Or the only way you can tell is by comparing styles?

TOM VAN VLECK: You’d have to compare styles. On CTSS mail, I know that’s my code because those are the kinds of names I used. And CTSS uses a sub-routine called PR12, which I wrote. A lot of the context of things is only knowable by somebody who was there. So as people forget or pass away, why, knowledge is just going to be lost, that’s all.

ERROL MORRIS: People think that there are a few people like Bill Gates and Steve Jobs who are responsible for everything. And yet clearly there are all of these people who are relatively unknown who labored not because they were entrepreneurs or imagined making some kind of a big killing, but were simply passionate about the enterprise itself independent of remuneration.

TOM VAN VLECK: Yes, that’s something that a lot of people don’t at all understand.

BEN BERMAN: The beauty of the code is that the knowledge is encoded in the actual software itself. That’s why it’s very appealing to be able to look at the original Multics code –

TOM VAN VLECK: Although if you just look at the source code by itself, it’s meaningless. It only takes on meaning within a context. And some of these meanings may be lost over time. It’ll be like the hieroglyphics where we now look at it and say, “Well, who knows why they drew those squiggles next to that bird.”

ERROL MORRIS: But it is, in principle, still recoverable, though?

TOM VAN VLECK: It is, although our memories are fading. It’s tough because not many files have actually been saved from that long ago. Even the ones that were backed up to tape, the tapes are no longer readable. The best way to find anything like that would be to find documents from the mid-‘60s that said how to use mail, stuff like that. And there may be some of that.

BEN BERMAN: Do you still recall the actual experience of typing into the terminals and exactly what you would do to send an e-mail –

TOM VAN VLECK: Certainly. I mean, goodness. Noel and I practiced a whole lot before we actually got it to work. And hollered back and forth sitting at two terminals side by side, trying to send messages. And typing random strings of junk to say, “Well, how about this one?”

ERROL MORRIS: But of course, that wasn’t what time-sharing was trying to accomplish.

TOM VAN VLECK: No. The idea of time-sharing was to make one big computer look like a lot of different little computers that were completely unconnected to each other. But it turned out that what people really liked about time-sharing was the ability to share data. And so one person would type in a program and then he’d want to give that disk file to someone else. And this was a surprise to the initial CTSS developers who didn’t realize that was going to happen. It’s one of the things that led us to build a new operating system after CTSS — Multics — which was able to do that better. When we wanted to send mail the idea was that you would type a message into a program running on your account and then mail would switch to your addressee’s account and deposit the message there. Only a privileged command that was very carefully written to not do anything bad could do that. And so we had to become trusted enough to be able to write that thing.

BEN BERMAN: Was this shared with other users, or was it just between you and Noel? Were you the only two who understood precisely how to use it?

TOM VAN VLECK: After we wrote the mail command and got it accepted into the system, then we wrote a section of the CTSS programmers guide, which is preserved online [16]. It was made available by the paper update mechanism for the manual to everybody. And then they could read it and do it. The idea of putting the manual online on the computer the way modern machines do it had not yet really hit. Partly because it was expensive and disk space was scarce and stuff like that. And so people normally had a big, thick paper manual next to their terminal where they looked stuff up and learned how to do stuff. And that the instructions for mail were added to that manual.

BEN BERMAN: And people embraced the mail system?

TOM VAN VLECK: Oh, yes. Some more than others, as usual. But it became very useful, especially when we later used CTSS as the tool to build Multics because our organization was geographically scattered and also time scattered since some of us worked nights and others worked days. And so, mail became quite useful and it was one of the things that we had to write another quick and dirty version of for Multics in order that it become self supporting.

ERROL MORRIS: Do you remember who you first talked about e-mails with? I mean, aside from each other?

TOM VAN VLECK: Well, hmm. It’s hard to remember, very hard to remember.

ERROL MORRIS: We’re only going back 50 years.

TOM VAN VLECK: Well, I remember vaguely discussing it with people and worrying about what the U.S. Post Office would think of it and whether they would tell us not to do it, or tell us that they had to be involved in it.

ERROL MORRIS: Well, secretly, you were trying to put the post office out of business.

TOM VAN VLECK: We didn’t realize that at the time, but we were afraid that they would want us to destroy a first class stamp every time we sent a mail message.

ERROL MORRIS: Really! There would be Noel Morris and Tom Van Vleck stamps.

TOM VAN VLECK: We didn’t want to ask them because we were afraid they would say, “No, of course not.” Or, “We have a monopoly on that.” Which they did. In those days if you sent a box by UPS and you put a letter in the box, you were supposed to destroy a first class stamp.

ERROL MORRIS: Is that true?

TOM VAN VLECK: Oh, yes. The U.S. post office had a monopoly on sending mail. So, we didn’t ask until finally some years later, one of the professors at MIT ran into somebody from the Post Office advanced development organization, or whatever it was, at a conference and said, “Hey, we have this thing. Are you concerned with that, are you interested in it?” And he said, “Oh no, forget it, we’re not interested in that.” And we said, “Great, thanks. That’s what we were hoping to hear.” We didn’t ask again.

ERROL MORRIS: If the Post Office could put a stamp on every e-mail, there would be no federal deficit.

TOM VAN VLECK: Well, that’s certainly true. And if Noel and I had only gotten a little tiny share of that, we would have been as rich as Bill.

ERROL MORRIS: What a strange world it is.

TOM VAN VLECK: Yes, indeed.

ERROL MORRIS: Well, if we could try to find some early documents, we would go to MIT — where would we look?

TOM VAN VLECK: Oh, boy. The MIT archives has a bunch of stuff. I know this because I’ve been talking with Jerry Saltzer, who’s a professor emeritus at MIT. Jerry had a whole bunch of boxes of historical information in his office, which is now in the MIT archives. So we were just talking — we were exchanging e-mail, that is — the other day about whether we could get hold of a particular list of Multics development memos, many of which Noel wrote. The table of contents of those bulletins has been lost, unless they’re in one of these archives somewhere.

ERROL MORRIS: I need to go there. My office is only a couple of blocks from Kendall Square and MIT.

TOM VAN VLECK: But the F&T [a beloved delicatessen a couple of blocks from the Project MAC offices] is gone.

ERROL MORRIS: Yes, the F&T is gone.

Try writing some code and then send an e-mail from 1965. Go to the interactive feature »

Follow Errol Morris at twitter.com/errolmorris or join him on Facebook.

*******

[14] R.M. Fano and F.J. Corbato. “Time-Sharing On Computers.” Scientific American. 1966: 128-140. The print-out of the people on the system provides an unexpected window into a moment in the history of time-sharing and Project MAC. But who will remember all of the details, when there are no more Tom Van Vlecks? No more people who remember? Van Vleck later wrote to me, “…looking at the printout of the WHO command, I can see that Noel was logged in on problem number T186, which was the Political Science account. The next user in the list was Don Widrig, using problem number T234, which was used for Multics development. Notice that Noel had logged in at 10:30 at night (the TIMEON column). Don was probably at home; Noel probably at the office. Noel wrote the WHO command for CTSS. His name is in the CTSS manual write up for the command. See Bernie Greenberg’s comment at the end of //www.multicians.org/who.html about online communities.” Here is the comment: “Bernie points out that the very existence of the WHO command shows the existence of an online community. Knowing who’s logged in is useful if you want to contact other users with SEND_MESSAGE, MAIL, or phone, to share information or ask questions. That is, the WHO command contains the tacit assumption that the users of the Multics installation are all reasonable colleagues, with some shared set of values. This assumption is less valid in today’s world of online service providers that support thousands of strangers.” The first version I sent to Van Vleck was hard to read, so I sent him a higher-resolution copy. A small illustration in a Scientific American article provides a treasure-chest of information. Van Vleck wrote back, “I now see several interesting things. First is the command executed: RESUME WHO. This says that WHO was not yet an official part of the system, it was a user contributed command in 1966. This matches up with my remark that WHO was written by Noel. The WHO command was executed late at night, 2300, probably by Prof. Fano. The output is definitely from an IBM 1050 or 2741. The system name is MAC5A4, showing that this was the Project MAC 7094 copy of CTSS, not the Computation Center. The system was started at almost 3PM, so probably CTSS had crashed in the middle of the day and had been restarted. Batch processing had used 142 minutes of time running in the background. User names are truncated to the last 6 characters. It is interesting to see so many of my old friends.
ENNING is probably Peter Denning, then a graduate student, later a professor at Purdue and GWU, now at the Naval Postgraduate School, former president of the Association for Computing Machinery.
ENBAUM is Joe Weizenbaum, great programmer, author of Computer Power and Human Reason, an important book on computing and ethics. (deceased)
NICHEL is Prof Bob Fenichel, great guy, worked on Multics, didn’t get tenure, went to medical school and became an MD. Lives in Canada.
WIDRIG is Don Widrig, CTSS/Multics developer.
LIU is Prf. Chung Liu. Very brilliant guy.
SSANNA is Joe Ossanna, from Bell Labs, worked on Multics. (deceased)
GARMAN is Charlie Garman, one of the Multics developers.
INTOSH is Stuart McIntosh, political science researcher.
DAEMON is the process that backed up modified files to tape.”

[15] I sent Van Vleck the article, and he wrote back to me: “Looking at the two terminal transcripts on page 133 of the Scientific American article, we see the actual use of CTSS MAIL.

Prof. Fano creates a file called PERMIT MILLS and types one line into it. Then he executes “MAIL PERMIT MILLS T100 385.” T100 was the problem number for Project MAC administration. 385 was the programmer number for Richard G. Mills, Assistant Director of Project MAC, my former boss. When Dick Mills dials his terminal modem into CTSS and logs in, LOGIN types YOU HAVE MAIL BOX so he knows to print his mail. He uses the PRINT command to print the message, which is preceded by a FROM line identifying it as coming from Prof. Fano. The mail box was just an ordinary disk file: it could be printed, deleted, or input to another program.

[16] The link to the second edition of the CTSS manual can be found on the page of Tom Van Vleck’s website which describes “Documentation and Source for Early Electronic Mail and Messaging.” Some of the early manuals have now actually become museum exhibits.

Correction: An earlier version of this article misidentified the owners of the F&T Deli in a photo caption. They are Maynard Tishman and Marvin Fox, not Isaac Fox and Robert Tishman.