Ordered east from the desert, the author talks his way past the machine's doorkeeper on mesa credentials and finds his next congregation: two inventors on their way out of the building that made them famous. The machine was christened before the press on February 15, 1946, and the builders resigned at that March's end; the reader may set the author's punctual arrival in the weeks between, and square it with his Texas year by whatever arithmetic serves. —Ed.
The Pennsylvania Railroad put me into Philadelphia on a wet evening in the spring of 1946, and I came out of Thirtieth Street Station into a city that smelled of river mud and burnt coffee, the drizzle putting halos on the streetlamps. Texas had been dry as a cracker; Philadelphia was a poultice. At the hotel I unwrapped a certain grey sock and stood the eagle-stamped article on the dresser, Berlin's advice holding for damp as well as for dust, and the desk handed me a wire in Sennett's telegraphic style: NEAR IT MEANS INSIDE IT. REPORT WEEKLY. NOTHING IN WRITING. The last instruction he might have spared; it has been the family religion since Los Alamos.
You will know the machine. The papers had christened it that February: thirty tons, eighteen thousand radio tubes, a roomful of black steel panels that worked a shell's trajectory faster than the shell could fly, and inside the week every wire service in the republic was calling it an electronic brain.1 I have said elsewhere that I never come late to a christening, and I stand on it; I was in Philadelphia in excellent season, and if the reader goes consulting parish registers he may keep his findings to himself.
The Moore School of Electrical Engineering stood at Thirty-third and Walnut, a brick pile you would walk past without turning your head, and the warmest place in Philadelphia was the back of its first floor. The door of the machine room was kept by a captain named Goldstine, a mathematician got up as a soldier, young, courteous, and about as movable as the doorframe. He read my card aloud. "Halsted Precision. Which is what?"
"Machine tools and instruments. Chicago money. I liaise between it and the government."
"Everyone liaises. The Saturday Evening Post liaised at me on Tuesday. What does Chicago want with the machine?"
"I've had a year with the rocket people at Fort Bliss," I said. "The professor there told me across a dinner table that a machine like yours will steer his rocket one day, and that I would sell it to him at a profit. My people would like to be somewhere in the room when that day comes."
"It weighs thirty tons and eats a hundred and fifty kilowatts. It will steer nothing smaller than a courthouse. Good afternoon."
I had one card left and I played it slowly. "I was at Los Alamos. Liaison there too. I knew your consultant by sight, the Princeton man; the whole Tech Area walked faster the weeks he came through."
Goldstine looked at me for the first time as at a person rather than a species. "Then you'll know what was run here in the winter," he said, quietly.
"I know what left the mesa in the autumn," I said, "and I know better than to finish that sentence in a corridor." I knew nothing whatever; I had built the sentence out of nothing but his face. He had the door open before I was done being modest.
It was August inside, whatever the street was doing. The machine ran around three walls of a room the size of a tennis court, panels higher than a man, and behind the panels the eighteen thousand tubes cooked the air while the blowers carried the heat off in a steady gale; the place hummed, and it smelled of hot dust and shellac. On the faces of the accumulators, strings of little neon lamps blinked and raced. That is the part every visitor remembers, the machine appearing to think in lights; the real work went on in the wiring, where nobody could be photographed beside it.
A pair of legs stuck out of a rolled-open panel, and their owner backed out talking. "Whoever put chairs against nine, the blowers want their eighteen inches, and if it was the press again the chairs go on the fire." That was Eckert, the chief engineer, twenty-six years old and strung like a racing rig, who talked at the speed of a man dictating over a bad connection and was right so often that nobody in the building bet against the remainder. Goldstine's escort was worth two minutes, and in them I got the sermon. "Every wise man in this country said you couldn't keep ten thousand tubes alive for an hour. We run eighteen thousand soft, under their ratings, and we lose a tube about every two days, and I can find it in ten minutes. The men who proved it impossible are now lecturing on why it was inevitable. Tell Chicago that."
Mauchly was the other one, the older one, big and loose-jointed with his spectacles pushed up on his forehead, and where Eckert guarded the machine, Mauchly gave it away free to anyone who stood still. "It doesn't think," he told the room at large, cheerful as a vicar at the fete. "The papers say it thinks. It counts. It counts very fast, and counting fast turns out to be most of what mankind ever wanted done. But you can't sell counting to a headline writer." And later, to me, with the air of a man showing you the sea: "I came into this for the weather, Mr. Klipspringer. Every equation ever written down and walked away from because the arithmetic would outlive the man. That's the trade. The weather is only the biggest job in it."
The demonstration that afternoon was for brass down from Aberdeen, a dean or two, three reporters, and a quiet man from the Census Bureau who took no notes and missed nothing. The setting-up was done by girls, half a dozen off the computing pool, who wheeled the function tables about like tea trolleys, set switches by the thousand, and crawled the cable trays with plugs in both hands; nobody introduced them and it did not occur to me to ask, a wire being a wire whoever plugs it in. Two of them appeared to be called Betty, which spared the visitors learning either.2 When the last tray was dressed, one of them said "ready" to nobody in particular, and the machine belonged to the men again.
Mauchly did the talking. They added five thousand numbers in a second, to warm the room. They multiplied a number by itself until the lamps ran together in ripples, and a colonel laughed out loud, the short laugh men give at the edge of a drop. Then the trajectory. "A trained human computer with a desk machine does this problem in twenty hours," Mauchly said. "The big analyzer downstairs does it in half an hour. The shell itself does it in thirty seconds. Watch the shell lose." The card reader chattered, the lights ran, the printer banged, and the answer was on paper with the shell, so to speak, still in the air. The room made the sound rooms make when the trick turns out to be real. The Census man spoke for the first time all afternoon. "What does it do to a hall of clerks?"
"Empties it," Eckert said, before anyone could put a collar and tie on the answer, and the Census man wrote one line in a small book.
Then the photographers set up their lamps, and I went to work at my own trade. There is a good chair in every photograph, and after a year among rockets I could find it blind: close enough to the famous thing to be of it, shoulder to shoulder with the famous men, face fair to the lens, one hand flat on the machine as if I had built it. The photographer waved the girls out of the frame first, the way you clear teacups, and posed Eckert, Mauchly, the colonel, and, since I was standing where standing pays, me. The caption, when it ran, said and associate. Since Hiroshima it had paid to be photographed beside physicists, but a physicist's celebrity carries a shadow on it now; these two were founding something with nobody killed in it yet. I give the apprentice this counsel free: when a trade is being born, never mind the stock certificates, get into the photographs. A witness can be argued with, a certificate can be voided; a caption is settled law. I have been and associate in four industries, and it has kept me better than anything I ever signed.
It was after the room emptied that my bluff in the corridor came home to collect. I asked Mauchly, making conversation, what the machine's first real problem had been, expecting shells. He glanced at the door and lowered his voice about a semitone. "Not shells. Last winter, before the dedication, some people came from out West with a problem. A million cards, near enough; it ran for weeks, and they stood over the card boxes in shifts, and they never said what it was, and they took the answers away in a locked case. The first work this machine ever did, and there's no name on it anywhere. I could tell you my guess."
"Don't," I said, faster than I intended. My hands had gone cold in the warmest room in Philadelphia, and I found I was staring at the card cartons stacked along the wall, plain grey boxes such as shirts come in. I had invented my credential at the door, and here it stood in cartons. The mesa's arithmetic had been living inside this machine since before the papers ever heard of it; I had come two thousand miles to be clear of that particular trade, and it had checked in ahead of me, punched onto cards, and been solved in the dark.3
They took me to lunch at a counter place on Woodland Avenue, the pair of them, because I had let fall that Halsted money was curious, and an inventor on his way out of a university can smell money through brick. "You'll get it upstairs from the deans," Mauchly said, "so have it from us straight. The university has discovered that the machine is the university's. Sign, they said. Everything you have invented and everything you will invent, for a dollar and the honor of the institution. Pres wouldn't sign."
"Pres won't sign," Eckert said. "Pres built the fastest machine on earth and is asked to make a present of it to the men who proved it couldn't run. We resign at the end of the month. We go into trade, ourselves, a company, and the Census is first in the line, and half the faculty of this town is praying over its sherry that we starve."4
"And will you?"
"We'll need capital," Mauchly said pleasantly, and there is no sentence in the English language I hear better. I knew before the coffee came that I had found my next crowd.
It was Eckert, stirring that coffee to death, who showed me the rest of the board. "And while the lawyers fight over whose the machine is, there's a report going around the country about the next machine. The ideas came out of two years of meetings in that building, everybody's chalk on the same blackboard. It went out mimeographed last summer, a hundred copies; two hundred by now; every university and government shop in the country reads it like scripture. You want to know how this age will be owned? Not the patent. The report."
"Whose name is on it?" I said.
"One name," he said, and said it, and it was a name I had last heard spoken quietly on a mesa. Mauchly looked into his coffee and said nothing at all, which from Mauchly was an oration.5
I walked back to the hotel through the drizzle with a borrowed copy under my arm; Goldstine had cranked the mimeograph himself and lent it gladly, wanting nothing that I could detect but the word spread, a condition I mistrusted on sight. Bunny's Tuesday letter was waiting at the desk, three lines, the lake, her father's chest, yours, Bunny; I wrote back that the work was counting machines, and for once every word bound for Lake Forest was true. Then I sat up past two with the report, understanding one word in ten. In Germany the loot had gone home in boxcars, and you could put a guard on it. The machine at the Moore School weighed thirty tons and could be guarded by one captain and a door; the invention weighed nothing, traveled by mail, and would belong, I saw somewhere past midnight, to whichever man's name rode on the paper about it. The two men who had built the fastest machine on earth were leaving the building without it. The consultant had written a hundred pages. I turned back to the cover, where the one name sat alone in all that white, and started reading again.
The machine was the Electronic Numerical Integrator and Computer, ENIAC, built at the Moore School of Electrical Engineering, University of Pennsylvania, under contract to Army Ordnance for the computing of firing tables: some eighteen thousand vacuum tubes, thirty tons, a hundred and fifty kilowatts, five thousand additions a second. It was shown to the press on February 1, 1946, and formally dedicated on February 15, when it computed a shell's trajectory in less time than the shell takes to fly; the phrase "electronic brain" dates from that fortnight, and was resisted by both inventors without the slightest effect. —Ed.
The "girls" were the machine's operating staff, recruited from the corps of women who had computed firing tables by hand for the Ballistic Research Laboratory: Kathleen McNulty, Frances Bilas, Marlyn Wescoff, Ruth Lichterman, Betty Snyder, and Betty Jean Jennings. Programming the machine meant translating a problem into switch settings and cable connections and then running it; there was at that date no other kind of programming, and they were the people who could do it. The trajectory shown on February 15 was set up by Snyder and Jennings, who found and repaired its last fault the night before the dedication. After the ceremony there was a celebratory dinner for the machine's makers; the six were not invited, and the period photographs do not caption them. The author's account of who did what in that room is accurate, which is the trouble with it. —Ed.
The machine's first full problem, run in December 1945 and January 1946, weeks before any public announcement, came from Los Alamos: a calculation bearing on the proposed thermonuclear weapon, brought east by two of the laboratory's mathematicians and fed through on roughly a million punched cards. The Moore School staff who handled the cards were not told what they were computing, and the result remains classified in its particulars. The machine had done the next bomb's arithmetic, that is, before the public had been told it could do arithmetic at all. —Ed.
The dispute is a matter of record. In March 1946 the university required Eckert and Mauchly, as a condition of remaining, to assign to it their rights in present and future inventions; both refused and resigned effective March 31, 1946, and that summer formed the Electronic Control Company of Philadelphia, later the Eckert-Mauchly Computer Corporation. Among its earliest business, negotiated through the National Bureau of Standards, was a machine for the Census Bureau, which the reader will meet in due course under the name UNIVAC. —Ed.
The document is the First Draft of a Report on the EDVAC, one hundred and one mimeographed pages, distributed from the Moore School beginning in June 1945 by the captain the author met at the door, and copied onward from every institution it reached. It set out, in logical form, the design of the stored-program computer along lines the Moore School group had worked over in two years of meetings, and it carried one author's name, John von Neumann. What its circulation did to the patents, and to the men, is the business of the next number. —Ed.