L ooking Back

The TX-2 Computer in mind. He puffed a few puffs on his pipe, but failed to give me an answer, and so I asked if he needed more time to Early computer technology at Lincoln Laboratory led to think about his decision. After a few computer and the functionality of your touchpad. more puffs, he said, “No.” My crest- fallen look must have caused him Some recollections by Ivan Suther- well understood. I recall wrapping a quickly to add that the “no” was no land of MIT and Lincoln Laboratory, dozen turns of wire around a pencil more time. His answer to my plan 1960 to 1963 to make an inductor, which, when was “Yes.” I was on my way to my the- connected in series with the tunnel sis. It was only much later that Wes I first joined Lincoln Labora- diode and a single dry cell, made and I became good friends rather tory as an intern for the summer a fine relaxation oscillator. More- than mentor and pupil, though some of 1960. Having decided to move over, I recall that it produced about of his mentor role remains with me from Caltech where I got a master’s 50 MHz, which at the time seemed to this day. Wes once told me that he degree to MIT for my PhD program, to me a very high frequency indeed. I designed TX-2 for Sketchpad long I had applied to Lincoln in the early also recall seeing the sawtooth wave- before I ever came to Lincoln. I guess spring. It got to be late spring and form on an oscilloscope. That was the he was confident that “if you build it, I had heard nothing, a silence that day I lost interest in ham radio, trad- they will come.” caused grave concern because my ing my interest in linear circuits for I found a bug in the logic of the wife, Marcia, was expecting our first the appeal of nonlinear devices. “Lincoln writer.” The Lincoln writer in August. I wanted to use the TX-2 com- was a Laboratory-built terminal One of my early assignments at puter for my thesis work—what later device consisting of an electric type- Lincoln was to characterize a tunnel became Sketchpad. I went to Wes writer with lots of special characters diode, newly invented and not yet Clark and described the project I had and two keyboards, one for upper

TX-2 was an experimental digital computer created at MIT in 1958. It was one TX-2 computer of a few first-generation large electronic digital computers in which transistors largely supplanted vacuum tubes. It was designed to facilitate and enhance real-time human-computer interaction. When first implemented, TX-2 had inherited the ferrite-core memory from its predecessor TX-0 (there was no TX-1). It also had two other random-access memory modules that could work concurrently to provide increased computing speeds. TX-2 was an experimen- tal tool to test many techniques and devices, among which were a magnetic- core memory unit and the first thin-magnetic-film memory unit.

William Kantrowitz, a systems programmer on the TX-2 computer, provides the following list of some of the highlights of the TX-2 computer. • Much of [Sketchpad] began on TX-2. • Early pioneering speech research was carried out on TX-2. • TX-2 had one of the first, if not the first, two-level memory paging systems. • Pioneering work in large memories was done with TX-2. • The Advanced Research Projects Agency network (ARPANet) derived from experiments on TX-2 with a prototype net between TX-2 and a computer at System Development Corporation in California. Jay Forrester is holding a magnetic- • The feasibility of using the ARPANet for packet speech transmission core memory . was first demonstrated on TX-2.

82 LINCOLN LABORATORY JOURNAL n VOLUME 19, NUMBER 1, 2012 L ooking Back

case and one for lower case. There I think the bug was that the reset a master’s degree from Caltech, my were about a dozen Lincoln writers worked from the keyboard but not engineering skill had never before in the Laboratory, each built into a when one typed from paper tape. been taken seriously; the memory small table with casters. The bug, I went to my boss of the time, Jack remains vivid. Part of is as I recall it, had to do with a par- Mitchell, and reported the bug. I learning to take one’s place in pro- ticular problem with changing case. was surprised to hear him ask me to fessional life. On the tape a “change to uppercase” propose a fix. I was even more sur- Although enormous both in mark served to herald each string of prised when my proposed fix, which physical size and in capacity for its uppercase characters, and a “change involved half a dozen additional day, TX-2 operated like a personal to lowercase” mark served to herald relays, was adopted and built into computer. Users sat at the console each string of lowercase characters. all Lincoln writers. Although I had and debugged code or did experi-

Sketchpad

The Sketchpad system was the first graphical computer interface. It made it possible for a man and a computer to converse rapidly through the medium of line drawings. Heretofore, most interaction between man and comput- ers had been slowed down by the need to reduce all communication to written statements that could be typed; in the past, we had been writing letters to, rather than conferring with, our com- puters. For many types of communica- tion, such as describing the shape of a mechanical part or the connections of an electrical circuit, typed statements can prove cumbersome. The Sketch- pad system, by eliminating typed statements in favor of line drawings, opened up a new area of man-machine communication. The currently used , or GUI, was is shown working with Sketchpad. The lightpen, based on Sketchpad. shown below, and the keyboard function control are the precur- sors for the mouse and right and left mouse-clicks prevalent today.

Cathode ray tube screen Power dc coupled/ Display point signal ac coupled wire (bright dot) Shutter Shutter button Connector Coaxial cable

Split shell for Lens Photodiode removal of pen cap Transistor Rotate barrel for smaller Field of view preamplifier field of view to move diode back from lens

VOLUME 19, NUMBER 1, 2012 n LINCOLN LABORATORY JOURNAL 83 L ooking Back

ments online. Computer time seemed hang around to watch. Fred Frick, easy task. Seeking to duplicate their most available in the wee hours of who was then the director of the lab, work, Larry wrote a much more diffi- the morning, so I was up at least two looked at Tom and said, “Well, Tom, cult program to make TX-2 produce a mornings a week early enough to get I see you’re wearing shoes today, so continuously rising tone, a siren with to Lincoln for my 3:00 to 6:00 a.m. I guess it’s OK for you to stay.” The ever-rising pitch. Larry’s siren went computer time slot. Marcia, I recall, demo went well, as I recall. up and up in pitch, octave after octave, regularly got up long enough to see Every Wednesday, TX-2 changed until it should have been audible only me off, and later woke our two young a little. The hardware folk would add to dogs and not to people. What a children. It may be that my pleasant new features, improve performance, noise. Larry reported to me that the early morning experiences with TX-2 or do repair. We software types got guards rushed into the computer conditioned me to being an early to use the sometimes problematic room when his program first ran suc- riser, a property I retain to this day. result. Can you imagine the concern cessfully. No such siren had ever been The cafeteria served a nice breakfast that would spread in the software heard in the Laboratory, or anywhere starting about when I finished my group when a rumor went around else, and it could only indicate a disas- morning debug session, and I came that, today, the multiply operation ter requiring the attention of Lincoln’s to love sunrises. might not work properly? alert guard force! There was a Xerographic printer Hardware bugs also happened, I left the Laboratory in 1963, on TX-2. It printed on a roll of paper through component or engineering my fresh PhD in hand. TX-2 lived about six inches wide using a “charac- failure; TX-2 was, after all, an exper- on. Over the years, it grew ever more tron” cathode ray tube with a stencil imental computer. The quickest capable each Wednesday. It got more inside to shape its electron beam into way to get a bug fixed was to write and better memory. It got more input the many characters we used. The a very simple program that failed and output devices. It got a time- print wasn’t great, but it could print in spite of obvious correctness, and sharing system to participate in what all our special characters at reason- the easiest way to “explain” this bug was then another new field of com- ably good speed. This was, of course, to the hardware engineers was to puting: serving multiple users. I think long before xerographic computer show them a Polaroid photograph it is very significant that TX-2 had a printers were common. The group at of the panel of toggle switches that useful life of over 20 years. Few, if any, Lincoln had convinced Xerox to put “stored” 24 words of 38 bits each and of today’s computers can hope to be this printer together as a one-off spe- the console lights. Leaving a pro- useful that long, and, indeed, only a cial. The result was itself a bit experi- gram in toggle switch memory that very few whole families of computers mental, and though it printed OK, should obviously work, but failed, have had a similarly long useful life. xerography requires heat to fuse the was a sure way to get the hardware —Ivan Sutherland powder deposited electrostatically on people to fix a bug. Sutherland started his career at the paper. Sadly, if the paper jammed, Many of my memories are about Lincoln Laboratory as an MIT grad- it would sometimes catch fire, and if great people. Larry Roberts, a fellow uate student. After receiving his PhD, it didn’t jam, it often emerged highly graduate student at MIT and at Lin- he headed the Advanced Research charged with static electricity. The coln, was using TX-2 for early work on Project Agency’s Information Process- paper dropped from the printer into image processing. I once mentioned ing Techniques Office. Subsequent to a large steel bin with a sturdy steel lid to Larry a remarkable set of musical that, he worked for Harvard Univer- to use in case of fire. tones developed at Bell Laboratories. sity, the , Califor- On the day I was to demonstrate Each tone seemed, subjectively, to be nia Institute of Technology, and Sun Sketchpad for the first time to a group higher in pitch than the one before, Microsystems. He is currently a vis- of visiting dignitaries, Tom Stocke- yet, like Escher’s staircase, the set of iting scientist at Portland State Uni- brand, a fellow Caltech graduate who tones closed on itself. I failed to men- versity where, with his wife Marly was about as far from formal in dress tion to Larry that the Bell people had Roncken, he is leading the research as a person can be, asked if he could made a series of discrete tones, a fairly in asynchronous systems.

84 LINCOLN LABORATORY JOURNAL n VOLUME 19, NUMBER 1, 2012