In the blazing heat of the summer of 1947, virtually no houses in Washington were air-conditioned. We had a small house, belonging to an NIL person on leave, in the middle of a housing development on the outskirts of Bethesda, Maryland. Surprisingly, since the owner was a chemist with a Ph. D., the suburb was very much a working class one. No trees had been left standing, and the little one-storey houses had no attics to provide a layer of insulation between the sun and the inhabitants. Most people had whole batteries of fans forcing air through their little boxes of houses. The artificial blizzard was noisy with the whirring of so many fan blades, and the wind felt as if it came from the Sahara desert. But it eventually became clear to everyone that it was better than sitting dripping in the horrid hot humidity.

At work, the climate was quite different. The computer, occupying most of the floor space of what had been a warehouse, had to be air-conditioned. Since the offices of our group, arranged around the periphery of the building, opened into the computer space, one had only to leave one's door open to be cool.

While Maria and I were both assigned to Complab, the group most closely connected with the computer, we had quite different missions. I was the technical writer for the summer. My function was to re-write reports and proposals, making them as clear and persuasive as possible. Maria's job was nothing less than to learn to be a computer scientist, partly in apprenticeship to the other seven members of the group, partly by reading what existed in the way of manuals and literature, and partly by trial and error on the machine itself. As Sam White, the leader of the group, said to her,

"Go get confused, and then come and ask us questions."

Complab was almost entirely civilian, with only one retired colonel. White himself was an engineer who nevertheless had the manner of a commanding officer. He was impressive and a little frightening, but within an informal context. Everyone but Maria and Leslie, a young woman in her mid-twenties, called him by his given name, but no one heard that Sam wanted to see them without something of the feelings of a student called to see the headmaster.

The second in command, Maylon Harris, had a Ph. D. in philosophy, and the look and mien of a philosopher, but could scribble out computer programs so complex that no one could follow them. Amazingly enough, some of them ran the first time.

There were two things going on at the time. One was the progressive discovery of the uses to which the big new computers could be put. Back at Hiram Mason, Lefty was telling people that the computer was more important than the atomic bomb, and, despite the vast gulf between their knowledge and Lefty's, the people in our group would have agreed with him.

The other thing, exciting in a different way, was the increasing impossibility of denying that the Soviet Union was becoming our enemy. Stalin was no longer referred to as 'Uncle Joe.' Virtually everyone realized that he was an implacable enemy, now armed with atomic weapons, who would destroy the United States if he could.

What seemed to deter him from doing that was the newly- formed Strategic Air Command. With its fleet of B-29s carrying atomic bombs, it could probably get enough bombers past the defending fighters to destroy much of Russia and kill many millions of its citizens.

But it was a close thing. The B-29 was ageing, and the massive B-32 wasn't yet ready. The B-29s didn't have the range to hit the Soviet Union and return, but, with aerial refuelling on the way out, they could hopefully make it to a friendly or neutral country. The other iffy thing was that the Russians were used to millions of casualties, and Stalin might not be deterred. To make things still more interesting, the Soviets were developing their own version of SAC, and might surpass it. Suffice it to say, the long period of what the intellectuals called nuclear angst was getting well under way at that time.

The object at NIL was to draw these two developments together and make predictions, based on computer analyses, concerning a nuclear exchange. These would then be forwarded directly to the gentlemen who comprised the Joint Chiefs of Staff.

Complab was part of a larger group, Simforce, which was responsible for the war gaming. Military and naval staffs had for many years conducted war games. These were sometimes actual exercises in the field or at sea, but, as often, they were board games of one sort or another. Games of both sorts were used, not only to predict the outcome of battle under various circumstances, but to train officers in tactics and strategy.

At the time that Maria and I came along, war gaming was about to be made made scientific. On one hand, a great effort was made to state precisely the assumptions on which the rules of the game were based, and, hence, to refine the predictive value of the game. At the same time, the whole simulation became much more complete. There were games within games within games, and one could zero in on any aspect of warfare in as much detail as one wished.

The simulations of such things as nuclear exchanges were so complex that it was becoming almost impossible to play them by hand, and the other great development was their computerization. This, at least in theory, would allow the same game to be played enough times to allow the results to be statistically significant. Of course, the simulation of World War III was a gigantic project. It was, in fact, beyond the means then available, but that didn't prevent the operations researchers of the time from having a stab at it.

The man in charge of Simforce was General Denison, lately in command of an army in Germany. His organization contained many other recently retired generals and admirals, and some officers on active service. They were then mixed with high-powered civilians, ranging from physicists to psychologists, with the idea of blending theory and military practice. Indeed, this combination worked rather better than one might have expected, and simulations of all sorts of military activity were produced at a great rate. The idea then was that the members of Simforce, having created the games, would work with members of Complab to put them on the computer. In practice, some members of Simforce learned to program, and almost everyone in Complab became adept at devising games.

The first job for anyone entering Complab, even one such as myself, was to learn to program. In those days before FORTRAN (or any other programming language), programming was a rather different affair. Our machine had registers of 36 bits, each three bits being represented by a single octal numeral. I had never heard of the octal number system, but I adapted. There were no eights or nines; 7 in octal represented 111 in binary, 5 represented 101, and so on, down to zero. The console of the machine, rather like the bridge of a ship, had a number of registers displayed on it. Each consisted of a row of 36 lights with little switches under them. A binary number such as,

101000011010111001011011010100100101

(503271332445 in octal)

would be represented by a lit bulb for each 1 and an unlit one for each 0. Since there was no screen attached to the machine (that invention came later), the quickest way of discovering what was stored in any given register was to ask the operator to connect that register with one on the console, and thus display it. Conversely, if one wanted to enter information, one set the switches on a console register to the desired number, and then had the operator transmit it to the desired location. Needless to say, the operators had fingers which could fly over those minute switches so fast that one could hardly follow them with one's eyes.

The social situation at the console was a rather tense one. Machine time was measured in the tens of thousands of dollars. The head operator, Mack Kowalski, had been a paratrooper in the war. He was, in theory, subordinate to the programmer, but, unless he happened to respect the programmer, he was very much the senior NCO faced with a green young officer. I found him quite frightening, particularly when he looked at me, his fingers itching for the switches, and said,

"Come on, come on. You've wasted another thousand bucks."

The experienced programmers helped him by throwing switches themselves, but, when I attempted as much, I got only,

"Keep your hands off the fucking console."

I soon learned to approach the machine only when it was operated by Anna Louise, a lovely southern woman whose fingers were equally fast, but who spoke slowly and sweetly.

The basics of programming weren't particularly difficult. Even a beginner such as myself could write a program which added a string of numbers together. However, what Mack considered a "real" program had many loops within loops, and invariably exhausted the entire capacity of the machine. Saving space was an obsession, and the most arcane manoevers were used to replace three instructions with two, or to pack information ever more tightly within the memory.

In practice, almost every real program modified itself as it went on, leaving no trace of the original. Maylon had written one program in which an instruction called its own number, thus modifying itself.

Once the first draft of a program was finished, it would be punched on to paper tape and loaded into the machine for de-bugging. Except in Maylon's case, no programs of any size ever ran properly the first time. It might be weeks, or even months, before they did.

It was quite easy to waste a de-bugging session and not learn anything at all. That happened to me more than once, even with my modest little programs. If the operator was the relatively inexperienced Ellen, she would quickly get as confused as I was. If Anna Louise was in what would have been the co-pilot's seat in an airplane, she would figure out what I was trying to do and correct at least one of my mistakes. If it was Mack, I could count on exactly three minutes before he flushed my program out of the machine and beckoned for the next programmer to slide into my seat.

One day, the next programmer was Maria. I stood back and watched as she gave Mack a set of instructions she had carefully written out in advance. His fingers flew, and he simultaneously jerked his head at her and said something. She, too, started setting switches on her side of the console. Mack hit the start button, and lights flashed all over the console, too fast for anyone to read them. Then the electric typewriter at my side began working. Maria turned and pounced on it. She shook her head, but seemed to know what was wrong. She and Mack set to work changing the program. After twenty minutes of frantic activity, Maria got the results she wanted and gave a little cry. Mack smiled and whacked her on the rear as Anna Louise, calling her congratulations to Maria, relieved him at the machine.

Maria, predictably, drifted toward Maylon. He was the best programmer and, in many ways, the nicest person in the group. He was also modest in a way that reminded me of Ralph. It wasn't uncommon to see Maylon in the office that Maria and I shared, perched improbably on some piece of furniture, talking about a program. Sometimes they didn't even have the program in front of them, and their activity reminded me of chess masters who play without a board.

It was just then that Maria came in contact with a gadget that NIL had bought at the behest of the generals. It cost fifty thousand dollars and looked like one of the television sets that were coming on to the market. However, it was designed to furnish a specialized form of entertainment.

Under the control of the computer, it would display on the screen a single character, which could be either a letter or a number or a military symbol. Inside the device there was a little matrix in a cathode ray tube, and the computer "bent" the ray to pick the desired character off the matrix, putting it on the screen. Since each character vanished after a short time to be replaced by the next one, it was an inefficient and cumbersome output device which left no permanent record.

The programmers had no use for it, but the generals wanted it as a vivid illustrative device for war games. Sam White decided that it would be a good project for Maria to write the programs to provide the device, called a Symbolotron, with the input that it needed, and to fool with it generally.

Maria soon found out what Sam had not told her, that the Symbolotron had been kicking around for six months, and that another programmer, a man named Landrum, had had no luck with it at all. Landrum had since departed, and the rather awkward name of the device had been replaced with the designation, "Landrum's Folly."

Maria quickly wrote a program to provide Landrum's Folly with the inputs the manufacturer's manuals said it needed. The device made a humming noise, but the screen remained blank. Maylon told her to call the manufacturer in Seattle.

The manufacturer was a small company that made specialized radar, and which employed a receptionist, apparently Japanese-American, who hardly spoke English. After several calls, Maria found someone who was familiar with the device with whom she could also communicate. He thought that the characters were appearing on the screen, but for too short a time to be perceived. What the manual didn't say was that a given input had to be repeated, with appropriate timing, for as long as the display was desired. This, apparently, was what Landrum had never discovered.

Maria then wrote the new program. It was more mysterious than most to anyone looking at it because one of the loops contained calculations which were designed solely to take up time. The cycle could be speeded up with optional jumps linked to console switches which would by-pass or include varying amounts of calculation.

The first trial brought results of a kind. Shadows appeared on the screen. Maria pushed a button on the console, and there appeared some lines. On inspection, they were revealed as the top fragment of a capital 'A', the bottom of a lower case 'w', and part of a stylized drawing of a tank. This, said Mack, was a hardware problem.

More calls to Seattle followed, including one that presaged the future, the kind where one holds the receiver in one hand and a screwdriver in the other. Mack was on hand to help, and, once Landrum's Folly was disassembled, he took over the phone. Suffice it to say, the Symbolotron was adjusted so as to display single complete symbols, even though they were seldom centered on the screen. Mack advised Maria to leave well enough alone, and not try to center the characters. He then added,

"That guy could hardly speak English, but I think he was trying to tell me something about overheating."

He then shrugged and waved to the next programmer.

On Fridays Complab had its meetings. One of its members would give an informal account of his or her recent work, occasionally with demonstrations on the machine. This seldom involved the war gaming done in cooperation with Simforce. Rather, it generally consisted of programs designed to make it more efficient to use the machine. In fact, most of those programs were fragments of what would later be operating systems and programming languages. Toward the end of the summer, it was Maria's turn to give a talk on Landrum's Folly.

The audience was maximally supportive, and Maria spoke well, hardly seeming shy at all. She ended, humorously, by saying that Landrum's Folly would now display sequentially the letters of the sentence,

"The cat is on the mat."

There was applause, and we all went out to the machine. The Symbolotron was perched beside the input devices with wires running across the floor. Mack relieved Ellen at the console, and loaded the paper tape on which Maria's program was punched. Soon, a large 'T' came on the screen, slowly followed by an 'h', somewhat off-center, an 'e', a symbol for a space, and so on. Sam then wanted it to congratulate Anna Louise on her birthday. Maria had on a piece of paper a code of the numbers which represented letters, and she called the numbers to Mack, who entered them. Someone called for greater speed, and a parameter was changed.

Landrum's Folly got half way through the birthday greeting when it halted, made a peculiar noise, and began to give off blue smoke. Before Mack could disconnect it, there was a loud pop, and the screen turned a peculiar color. As more smoke billowed out, Mack ripped out the wires which connected it to the machine, picked the Symbolotron up by its power supply cord, and proceeded with it to the men's room. The others were delighted by the episode, and there were many interpretations of the birthday greeting to Anna Louise. The latter was comforting Maria, who was near tears. When Sam noticed, he assured her that the demise of the device was in no way her fault. He added,

"We'll make them give us another one, and it had better be an improved model. I'll tell them they almost set fire to our installation."

She perked up and replied,

"I really did get to like it. If they could get it to display more than one symbol at a time, they might really have something."

These words were, of course, prophetic.

There was now only a week to go in the summer. I re- wrote a report and Maria helped Maylon de-bug a big program. There was a send-off lunch for us on the last working day, and we promised to return the next summer.

On the long way back on the train, Maria continued to talk about Landrum's Folly. Despite its demise, it had been the instrument through which she, a mere schoolgirl, had achieved a significant success in a decidedly adult world. I spoke kindly of it as we roared through small towns and headed for the mountain fastnesses of West Virginia in the gathering night.