"You Are an Interfacer of Black Boxes"
Computers surround us, about 50,000 of them now: just a decade ago there were 1700. Most of them are used for recordkeeping, or “sophisticated paper-pushing,” as the industry calls it. Ultimately they record almost every bit of money that changes hands, and they file various items of information: where you were born, where you want to fly, the numbers that you are known by, your vulnerability to direct-mail advertising.
This is one function, storage and retrieval, but computers are also capable of making certain kinds of decisions. Fed information, they may recommend action and thereby regulate refineries, diagnose illnesses, and fire retro-rockets. More important, they imitate reality. Probably the greatest use for computers beyond data processing is for the performance of simulations—war games, marketing strategies, stress analyses of aircraft—which allow the user to discover the consequences of an action without taking it.
No one is unaffected by all this, but for most of us the computers remain remote. They are mildly disquieting. It may be hard to take threats of the automated society of the future seriously, but computers say something unpleasant about our condition in the present. They are incarnate metaphors for the brain, and the brain often suffers in comparison. So it seems to “most of us.” But for an increasing number of people computers stand at the center of life.
Perhaps as many as 300,000 people—the programmers—tell computers what to do. lake computers themselves, programmers are mostly in their twenties. Their talents are indispensable, and they incite the economy to lust. However many there are, it is generally agreed that half again as many could be hired now, and twice as many as soon as the “fourth generation” machines appear. Intelligent voices seek programmers on the radio, and the want ads suggest rewards, “fringes,” and the mystique:
. . . Programmer, Has Your Ability Outgrown the Software? . . . Start $13.5—18,000++ . . . $2000 Guar. Increase Over Present Earnings for 3rd Generation People Interested in Real Time, Time-Sharing . . . Salary Open . . . Fly Free . . . Take a Trip on Us . . . Stock Options . . .
I have lately been talking to programmers in and around Cambridge, Massachusetts, where the buried circuitry of computers and computer people that binds any city is somewhat more manifest than elsewhere because of the universities and their spun-off, and symbiotic, research enterprises. One such place is the Smithsonian Astrophysical Observatory. Smithsonian (which employs about twenty-five programmers) tracks man-made satellites, studies the habits of meteors, and smiles on the whims of its employees; once one of them tried to program a theory of the creation of the universe, but “only got about ten micro-seconds into the job. ("Micro-second is a millionth of a second. As away of saying speed, it’s being replaced by “nano second, a billionth; “fempto-second,” a trillionth, is on the way.)
by Richard Todd
I went to Smithsonian to see Rudolf Loeser, a 1961 graduate of Harvard and Smithsonian’s senior programmer. Before I got to him, however, I saw his computer, on the first floor of Smithsonian’s offices in an industrial park. It was a stylish-looking machine of several great, rectangular components, faced in blue panels, and it occupied a glassenclosed space that resembled an automobile showroom. I had never seen a computer before. (Increasingly, this is an admission of provincialism, even cultural deprivation.) The visible signs of its work were whirling reels of magnetic tape on one component, and printed data—printout—emerging from another, in abrupt chunks. An adolescent occasionally dashed to the computer with a reel of tape, and replaced one already on the machine, trying, it seemed, not to waste the computer’s time. After watching him for several moments I started up to Loeser’s office.
The computer I’d seen was Control Data 6400, the largest computer (in terms of its abilities) in Cambridge. “It can perform 800,000 additions per second,” Rudy Loeser told me, “or 200,000 multiplications. I don’t know how to put that into easily comprehensible terms. There are what?—three billion people in the world. I suppose if you had data on the number of hairs on each of their heads, the computer could give you the total number in about an hour. Perhaps an hour and a half.”
Loeser, who has a broad, placid face and wavy hair, was born in Germany, and speaks with a slight accent. He speaks quietly, with a likable, if demanding, precision. He often pauses before a sentence, delivers it quickly, and is silent again. He is conscious of being considered unspontaneous. “I’ve changed a lot since I was married,”he remarked. “Before that people used to say that I was the most inhuman person they had ever met.”
Loeser was at work on a program that had occupied most of two years. It was called Pandora, and its function, he said, was to test certain hypotheses about the composition of the sun. In written form, Pandora filled two fat blue binders.
“We hope we will make some real statements about the sun,” Looser said. Talking with Loeser at first I had a feeling of great separation, as if we were two strangers in a waterfront bar, each with poignant but ineffable stories to tell. He was speaking of programming, he gestured at a program, but I had little concept of what one was.
“A set of instructions” is a frequently used shorthand way of explaining a program. I had read the phrase, but when I asked Loeser about it he seemed uncomfortable. He said that it was not only an oversimplification, but “rather anthropomorphic.” It appeared, he felt, to imply that the programmer had only to describe his desires and the understanding computer would enact them.
But if you look at a programmer “behavioristieally,” that is, in fact, surprisingly close to what happens. His work is done on paper. What he writes often appears as “instructions,” indeed as orders. Although Loeser’s own work, he said diffidently, was not very accessible to an outsider, everyday programs closely resemble the experience they describe. For example, a typical program for balancing a checking account begins “Get Check.”
PO. GET CHECK
MOVE ACCT-NO TO
ACCT-AO-CURRENT.
P1. GET MASTER.
IF ACCT-NO-MAST IS NOT EQUAL TO
ACCT-NO-CURRENT GO TO P1.
IF BALANCE-MAST IS LESS THAN
BALANCE-CHECK GO TO OVERDRAFT.
SUBTRACT BALANCE-CHECK FROM
BALANCE-MAST.
ADD ONE TO NO-OF-CHECKS.
GO TO PO.
Allowing for some abbreviations, this appears to be written in English. Actually, it is written in a language called COBOL, which stands for “common business oriented language,” the universal tongue used for computerized bill-collecting. COBOL contains a great many words of pure English and some coinages. When you watch a programmer jotting it down, you might think he could embellish his instructions at will, but not so. Every computer language has a strictly delimited vocabulary and a rigorous grammar.
COBOL is one of the easiest of literally hundreds of computer languages. Rudy Loeser works in one called FORTRAN (formula translator) , which is useful for expressing mathematical and scientific problems. Loeser can use several other languages; “learning a new one,” he said, “is almost a trivial matter.” (But a great many programmers make a living knowing, or caring, about only one. According to most computer people, the ability to pick up another language “just by reading the manual” is one mark of a gifted programmer.) Loeser mentioned some of the rarer languages, which have names as inviting as Saskatchewan towns; JOVIAL, JOSS, SNOBOL, PLEASE, and LISP. His own, FORTRAN, is more obscure-looking than COBOL, partly because it refers to less concrete phenomena. “I imagine someone processing Diners Club bills lias a firm sense ol reality,” he said.
But in my case the connection with reality is generally . . . not there.”
He showed me a small section called a “subroutine" of his vast program. Its function was simply to compare two numbers and mark them with an asterisk.
subroutine splice $ (I, ITV, MARK) c RUDOLF LOESER c SERVICE ROUTINE FOR HAWSER MARK = 1H IF (I-ITV) 101, 100, 101 100 MARK = 1H* 101 RETURN END
Computer talk rapidly escalates out of sight. Aside from the arcane computer languages themselves, ordinary conversation among programmers often occurs in a private tongue. It is nearly a subroutine ol English, and listening to it means listening to familiar syntax, to not wholly foreign words, but to meanings that remain entirely obscure. My eyes fell on one of Loeser’s interdepartmental memos:
To change fatal error 78, bad data, to non-fatal error user must initialize registers in the routine with a single call to BLISS prior to any read or decodes . . . SCROG is updated to agree with SCROGX.
But these are all symbols, and you are willing to believe they stand for something.
One thing that was bothering me, however, was the distance between the programmer and the computer. We were sitting a floor above the machine, and I found it surprising to realize how little a programmer, who is easily imagined in a tense posture next to a computer, need actually see of one. What connection did the language have with the computer? Loeser explained that that was the reason he had been reluctant to call the program a “set of instructions.”Its written form is only one of several states a program occupies in its transition from the imagination of the programmer to the memory of the computer. “It’s an interface between the programmer—the human being—and the computer,”Loeser said, using an omnipresent bit of computer lingo.
“Interface” refers to anything that mediates between disparate items: machinery, people, thought.
The equipment that makes the computer’s work visible to the user is often called an “interface,” and the word is used highly metaphorically, as in “the interface between man and the computer, between the scientist and society.”
Computers are troubling not just because we don’t understand them (most of us deal with many machines we don’t understand) but also because we suspect that we finally couldn’t understand them. And we are, most of us, right. Reassuringly, this is even true of many programmers. “It’s possible to be a nine-to-five programmer,” he said, “and never think about what happens within the computer. To some programmers it might as well be a hamster on a treadmill generating the output. The computer—for many purposes—may be thought ol as a black box.”
A black box. Loeser explained, is anything considered in terms of input and output, without worrying about processes. It may refer to parts of a computer, or parts of a program. The idea is not so complicated: to simplify by abstracting an intricate system into manageable components. But it has a particular relevance among computers, where everything, looked at too closely, takes on a bellygripping complexity. “Black boxes” are as handy a concept as “interface”: “interfacer of black boxes" is not a bad definition of a programmer.
I asked Loeser if he could tell me something about what happened in the mind’s black boxes when he was working on a programming problem. He said that he usually let the problem rest in the back of his mind, turning to it for only a half hour at a time, over a week or so, counting on a solution to begin to generate itself, before he began to bear down hard. Even when the answer is at hand, he said, it is never entirely comprehensible:
“Often you’ve deluded yourself into thinking that you have a clear, simultaneous understanding of the problem, but you don’t. It s impossible to see the plan as a whole. I often see it as a ticker tape passing through my mind. I know I’m there if I can follow each step as it goes by.”
Rudy Loeser is spontaneously called “a true professional by his colleagues, and I wondered what that meant, how he felt about the programmer and his “interface” with society.
He said, “It’s the ‘in thing’ now to characterize programming as a profession. But I’m not sure it’s justified. It’s errant nonsense to call some novice just out of programming school a ‘professional.’ And most programmers become dominated by their salaries when they find they can extort almost anything they want. I’m afraid you don’t find many altruistic programmers.”
Loeser sees the computer as a “tool.” “Of course it is by far the most versatile tool we have, and it sets its users apart from other tool-users. The great source of fear about the machine is that people don’t understand them. People don’t understand us either, so they mistrust us. . . .
“I am concerned. But not about the computers —about the motives of the people who use them. The nightmare vision, of course, is hordes of militarists running rampant. You won’t see it happening if only the engineers become really concerned. Man could do the right thing, but will he? It depends on the extent to which we become involved. We must commit ourselves with deeds. And we don’t, not enough. How do I feel? I feel mostly apprehensive—apprehensive and guilty.
“Eventually we will build a machine we don’t understand. . . . At one point we will wonder if we should rely entirely on biology for the continuance of human culture. Machines may be made partly of living material, for humane and emotional functions. . . . And someday we will pass the torch on to the machines.”
One of the broad distinctions that programmers make among themselves is between “applications" and “systems programmers. Rudy Loeser is purely an “applications” programmer, which means that he uses the computer to solve problems from the world (or, in his case, the universe) of people and things. “Systems" programmers worry about the programs built into the machine, “the software,” which allows the computer to receive other programs, to switch from program to program, and in general to keep itself operating efficiently. Aaron Kronenberg is a young programmer with a special interest in systems work.
I met Kronenberg at Abt Associates, a Cambridge think-tank, specializing in “scientific solutions to social problems.”Abt was founded just four years ago, and it has had spectacular success: its billings have doubled each year and now approach $4 million. It is full of élan: extraordinarylooking girls in pantsuits and nano-skirts hurry through the white-carpeted hall, and at closing time bearded young men in lumberjack shirts throw Frisbees. It is as if everyone’s college-age children had come in to see where their fathers work.
Kronenberg, who is himself only twenty-three, has a reputation for brilliance and eccentricity. (“See Kronenberg,” one of his colleagues said, “his head is in some interesting places. He used to be a disc jockey you know.”) He weighs well over two hundred pounds, and his usual uniform is tight trousers and a white shirt, too small, straining open at the neck. He wears thick glasses, has troubled black hair, and the day I met him he wore a button that declared, “I am a rat fink!
“So you want to learn about computer,” said Kronenberg, who often speaks an article-less Indian patois. “I give short magical mystery tour.” We went to his office, where, on the blackboard, he gave me an overview of the computer’s functions —often addressing the imaginary machine as “Hey, baby,” or “This yo-yo here”—and of his own duties.
Most of his work, he said, involved consulting to companies whose computers have run amok, and whose regular programmers are no longer able to understand them. (When computers hit a snag they often print out dense reams of digits, which represent all the data that’s been given to them. This is called a “dump.” Searching out the error in a dump is referred to as “debugging,” and it occupies much of any programmer’s time.) Kronenberg’s consulting assignments tend to be tense, as the expensive machinery sits idle. “Essentially, I’m a rescue service,” Kronenberg says. “They fly you out to California and parachute you into Fresno and say you will return in seventy-two hours, you will bring program. It generate ulcer.”
Kronenberg’s consulting fee—Abt’s charge for his service—climbed to about $300 a day as he became better known in the business. His own base salary, he said, was $14,400, with bonuses and stock options that brought his income to about $17,000 a year. Programmers may, if they choose, move around a great deal, and many of them do. Kronenberg was weighing two offers: “I hope to push my salary up a couple of figures anyway.”He said that he’d been talking about it to his boss, Clark Abt, whom he calls Leader, “With offers I have, Leader listen.”
Kronenberg’s self-confidence sometimes irritates his colleagues. (“I could strangle him,” one girl said. “Obviously he was hired for his ability, not his personality.”) But Clark Abt said, “Ron has a few rough edges, but they’ll smooth off. He has great growth potential.”
Kronenberg, who says of himself, “I’m probably the weirdest guy in this business,” and whose computer recognizes him by the password MADMAN, is atypical in many ways, especially in his education. Most programmers have at least a bachelor’s degree, but Kronenberg is a college dropout. He entered Rensselaer Polytechnic Institute, intending to become a physicist, but left in boredom after two years. (Not before he had taken some programming courses, though, and founded a campus “computer society.”) After RPI, he became a disc jockey. As Art Matthews, he was morning man for WXKW, Sarasota, and he can give a convincing reproduction of his radio self, as he did one afternoon at Abt, lowering his voice and dropping his head to his chest: “Movin’ and groovin’ with Big Daddy Madman Matthews on soooooo-oulful 1600 W X K W 76 degrees in the big bag outside and time for: Muuuuu-sic!”
Meanwhile, he was programming for the city of Sarasota, and he was becoming more involved in computers, and he realized that he had highly marketable skills. “Next step. Whirlwind coast-tocoast tour.” He ended up working at the Yale University computer center, where he helped set up a “time-sharing system,” and then he came to Abt.
“Time-sharing” is Kronenberg’s particular interest. It is a way of accommodating several users, with the appearance of simultaneity, in a single computer. The users may be spread throughout a city, state, or the world. They communicate with the computer through “terminals, which are connected to the machine by telephone lines. The terminals resemble IBM electric typewriters; in fact, most of them are IBM electric typewriters with some serious modifications. The effective difference is that the machine types back. At Kronenberg’s terminal, when he typed in MADMAN, it typed back READY. A similar interface is the graphic display terminal, like a TV screen, which also interacts with the user, who may enter data by means of a “light pen,”like a small flashlight. The distinctive thing about time-sharing, from the programmer’s viewpoint, is that his transactions with the machine take place in “real time"; that is, the computer interacts with him step by step. If he makes an error, he doesn’t have to wait for his whole deck of cards to be read by the machine to find out; the computer calls attention to the mistake in its next breath.
Time-sharing represents only a small portion of the computer industry now, but it is growing. The terminals may be rented for very little, though the charges for an hour of the computer’s time run to several hundred dollars. Time-sharing makes the machines more accessible to human beings—“conversational" is the industry’s word for the relationship—and it promises new uses for computers. Already, time-sharing systems enable airlines to confirm reservations immediately. When the day comes (it’s often predicted) that everyone has a “computer in his kitchen,” it will really be a timesharing terminal, able to retrieve material from the Library of Congress or the New York Times archives, to predict one’s future tax problems, or to enact marital game theory.
Time-sharing systems are, like any computer operation, only as good as the programmer who readies them for nonprogrammers. Kronenberg is working on methods of increasing the machines’ ability to respond to everyday language. “I’m interested in the machine-user interface,” he says. “I’m trying to work out ways to program the machine so a sociologist can sit down and talk sociology talk to it and the machine will answer back in sociology talk. There’s a computer mystique—people are afraid of computers. But they shouldn’t be. Computers are good guys.”
Kronenberg is totally absorbed in computers. His habit, he said, was dinner at a delicatessen near Abt and work long into the night, and work at home as well. Some everyday concerns find him oblivious: he doesn’t drive a car, he walks in rain or freezing temperatures without a coat, his dates sometimes begin at midnight.
The day I saw it, his big apartment in the South End, a hiply fashionable neighborhood of old brick houses with bow fronts, had a scarcely inhabited look. But in his bedroom, an anarchy of bedsheets and books, Kronenberg kept a timesharing terminal so that he might jump from bed and type out the solution to a problem. The terminal rents for $115 (by special arrangement, the time comes free), and he spends $150 a month for books, “mostly in the field, to keep up.”
Not alone among programmers, Kronenberg is worried about maintaining his own parity with the state of the art,”not about the purported evils of automation. “I don’t think the computers will put me out of business for a few years,” he says, “I don’t think they’ll appreciate beauty in quite the way I do, but if so, a certain joy will be gone out of life. But it’s a long way off. And anyway I doubt that the machine ever will be a good disc jockey.”
I had talked at length to two programmers. They didn’t seem much alike, and I spent a couple of days in assorted computer installations, looking for a commonality. It was a futile search, of course, and actually Rudy Loeser and Ron Kronenberg, despite their styles, did share something: they were near the top of a profession that contains great variations in ability. If there is anything that computer people seem to agree upon, it is that the difference between an ordinary programmer and a good one is the difference between the runway and the air. “A good programmer isn’t worth two median ones,” one computer executive said, “he’s worth ten.”
At one level, it is possible for many people to become programmers. I talked to a programmer who did payrolls for a hospital, and asked him why he got into the field. “Because I failed in the restaurant business,” he said. And in fact, we will probably all be programmers of a sort in a few years, with the proliferation of computer terminals in the home.
Programmers in more recherché spots look down at the data-processors in business. A university programmer said that “most people in banks are doing things that probably just don’t need to be done,”referring to the possibility of standardized software to perform routine work. Nevertheless, the banks represent in miniature a world in which computers and their human aides sit uneasily in a traditional setting.
I talked to George McQuilken, twenty-five, an IBM systems engineer at the State Street Bank. (IBM sells the full-time service of systems engineers along with the computer; the SE’s live at the installation, consultants to the regular programmers.) McQuilken said, “Ask a banker what his biggest problem is today and he’ll say computers. Ten years ago he would have said paper work. Computers cost too much, and to run them he has to hire a bunch of kids, at ten and twelve a year, and they have beards and rotten personalities.”
The programmers at the bank, in fact, were conventional, if not bankerly, in appearance, but Bradford Tripp, vice president for computer operations, warned that “one kook can ruin it for everyone,” as a board member comes through and says, “So, you’ve got them here too.”It is true anyway that the bank’s programmers can sometimes achieve an independent style. I was introduced to one programmer working on a problem more complex than check processing, a financial simulation. I asked him if he could tell me something about it. He looked saddened; and after a silent moment he turned and walked away.
Somewhat surprisingly, even in their own milieu, programmers are found disquieting. They’re a little like low-level diplomats,” said Professor Joseph Weizenbaum, a computer scientist at MIT. “They don’t have any real decision-making power, but it ends up that they’re making all the decisions, simply by telling you what you can and can’t do. And they feel misunderstood, by people ‘up there,’ which gives them a cliquish sense of professional pride.”
Weizenbaum feels that programmers are largely of a distinct personality type, which tends to be a distortion of the qualities that make good scientists in general.
“I’m talking about a universally recognized phenomenon. People get hooked. They begin to behave in a way that resembles addiction. They refuse food. They refuse their girlfriends. I’m quite serious. The word ‘compulsion isn’t far removed; in fact, it’s correct.
“When a programmer finally gets his program perfected, what does he do? He sabotages it. I don’t mean he literally, consciously wrecks it. But he goes in and says: ‘This can be done better!’ So he destroys what he’s done and gets into a terrible panic, and then he’s happy again.
“He’s just like a compulsive gambler. He’s not interested in winning but in keeping the game going. Why? He’s emotionally involved in a struggle for control. These people have suffered a major defeat sometime in their struggle for control. They feel they can’t achieve the kind of power they need. They have given up the real world and begin to operate in a magic world in which they believe they can be omnipotent.”
Weizenbaum is a member of Project MAC, MIT’s huge computer sciences program. (MAC is a double acronym: “multiple access computer”; “machine-aided cognition.”) The project is credited, among other things, with most of the developmental work in time-sharing. MIT, probably the most computerized university in the world, has forty-odd computers of various sizes and hundreds of time-sharing terminals. Some knowledge of programming belongs in almost every undergraduate’s repertory; the elementary course in programming, “Introduction to Automatic Computation,” attracts more students than any other at the school. This year under the direction of Robert Fenichel, a young professor of electrical engineering, the course has begun to be taught by machine. Fenichel calls his automated course TEACH. At the beginning of the term students meet and are given a mimeographed “handout” that begins “You should expect to have little formal contact with the instructors . . .”
I had asked Fenichel too about the programming mind, and he had said, “You’re going to have a hard time finding out. The ur-programmer relates to his computer the way an Iowa farm boy relates to his old Ford, and neither one of them can tell you much about it. Some of them are virtually idiot-savants. Some are duo-maniacs: you’ll find a brilliant programmer who also knows everything about ballet—and nothing else.” Fenichel offered, however, to let me get some “hands-on” experience by participating in TEACH.
The machine turned out to be not nearly so chilly as might be expected. It took me a couple of tries just to “login,” but, successful, I was rewarded with HELLO, and my name! At moments TEACH is capable of a truly jarring testiness: GARBAGE FOLLOWS COMMA ON MAIN LIST . . . NOT AN EXECUTABLE STATEMENT. But it also pleases the student, after his first easy problem, with a mildly ironic CONGRATULATIONS, and after each session it types GOODBYE.
The transfixing ball ot characters that flies across the page creating printout at the rate of 15 characters a second is enough to fascinate the human participant. A short way into the course you begin to get a whiff as well of the mixed sense of power and subservience the programmer must feel, as the machine performs dazzling errands on a correct command, or traits silently (LISTENING AT LEVEL 1) for you to say a sensible thing. I had intended to spend 15 minutes with TEACH the first day and I spent three hours. But I did not fully realize its effect until I was in my car, where I sat for a moment, expecting that machine, too, to tell me what to do.
MIT’s programmers roam through the halls of the elegant building at Technology Square. Most of them are unstructured in appearance; by looks they might as easily be members of the Electric Cabbage. Artificial intelligence is on their minds too.
I talked to Jed Harris, a twenty-one-year-old student at MIT on leave from Beloit College to take graduate courses in computer sciences. He has glossy black hair at shoulder-length and a full beard, and describes his programming work as “like living inside a Bach fugue.” “In a few years,” Harris remarked, “it will begin to feel immoral to tell your computer what to do, just like slavery. Watch: as soon as the machines begin to simulate consciousness on the level of a dog, you’ll see protective societies forming, like the ASPCA.”
Jerry Yochelson, a 1967 graduate of MIT, and a full-time programmer at MAC, described the Project’s work with chess. He acknowledged that he was a devoted player, but that he had never beaten the chess program in one of MAG’s computers, which is now playing a low-level-B game and “has beaten many men and some computers.” Computer scientists like to engage their computers in chess, partly because the men themselves like the game, and partly because it is so complicated that the machine is not simply calculating multiple possibilities, it is enacting strategy. Some of its programs enable the computer to review previous games and analyze errors so that it improves upon itself. The game is represented on a screen, like a television screen; this is a form of graphic printout used for a variety of purposes. The screen is sensitive to light, and to move, the human player need only indicate the piece and the square with a light-pen. “It’s very disappointing to lose,” Yochelson said. “You ponder over your move, and make it, and then you just sit there and watch a piece disappear.”
Yochelson foresees without question a computerized world in which work will be optional, or perhaps luxurious. From the window of his eighthfloor office one sees the old red brick factories of East Cambridge, all quite susceptible to computer management. “Almost no one will have to work in twenty years. The twenty years may be off, but well have the necessary technology in that time, so it becomes a question of what most people want. Obviously, it will disturb many people. Computers pose a threat to people who don’t want to change.”
In the imagined workless society, programmers would, of course, not be the first to be displaced; but neither would they be the last. Almost at once, people saw in their work the implication that programmers would “program themselves out of a job. It’s one of the paradoxes of the occupation that twenty years ago it scarcely existed, and today it contemplates its own obsolescence.
In the present, programmers occupy an ambiguous social role, despite the fact that they perform some of the most sophisticated chores being done in the world and that, without them, organized life in the United States would choke. They seem to be thought of as something between professionals and occult tradesmen. If programmers’ meritocratic life-style nettles those around them, it is because of a sense that their ascendance is unearned. Their work is as crucial and as inaccessible as that of many scientists, and yet they stand apart from the educational hierarchy by which society keeps most of its brilliant members in bounds. “People have to listen to you in this job,” Jerry Yochelson said, “because they don’t know what you’re doing. I like that.”