The Dream Machine - M. Mitchell Waldrop

Partial Excerpts (antilibrary)

Prologue:       Tracy’s Dad

the very first thing he’d always do was turn to Tracy and his thirteen-year-old sister Lindy and ask, “What have you done today that was altruistic, creative, or educational?”

Once, at a convention, a guy had leaned over Dad’s shoulder and offered fifty dollars for a page of his doodles

And then there were the art galleries. On the weekends Mom liked to have a little time for herself, so Dad would take Tracy and Lindy out to look at paintings, usually at the National Gallery of Art down on the Mall. It was mostly the Impressionists that Dad liked

Dad and Mom managed to fill up that house with actors, directors, writers, artists, and all kinds of wonderfully oddball people—not to mention Dad’s students, who would drop by at all hours

His father was crazy about computers. Back home in Cambridge, at a company called Bolt Beranek and Newman, the members of Dad’s research group had had a computer that they’d modified themselves

light pen—

Tracy’s dad had come to believe that one day all computers would be like his machine up in Cambridge. They would be humane and intimate.
Ultimately, in fact, they would enter into a kind of symbiosis with humans

they could scatter dozens of remote terminals around the campus and in people’s homes. And then, through the technology of “time-sharing,” they could tell their big, central machine to dole out little slices of processing time very, very rapidly, so that each user would feel as if the thing were responding to him or her as an individual

Project MAC would not only introduce hundreds of people to the joys of interactive computing but also evolve into the world’s first on-line community, complete with on-line bulletin boards, E-mail, a “freeware” exchange—and hackers

Meanwhile, Tracy’s dad was also gambling on a soft-spoken, rather lonely guy who had approached him on practically his first day at the Pentagon, and whose ideas on “augmenting the human intellect” had proved to be identical to his own notion of human-computer symbiosis. Douglas Engelbart had been a voice in the wilderness until then; his own bosses at SRI International, in what would soon become Silicon Valley, thought he was an absolute flake

Engelbart, with his group, would go on to invent the mouse, on-screen windows, hypertext, full-screen word processing, and a host of other innovations

Engelbart’s December 1968 presentation at a computer meeting in San Francisco would blow about a thousand minds at once—and later be remembered as one of the turning points in computer history

and a talented portion of which would soon be gathering at PARC, the Xerox Corporation’s legendary Palo Alto Research Center. There they would put Dad’s “symbiosis” vision into the form we are still using more than three decades later

finally, there was networking. Now that he had come to work for the Pentagon, Tracy’s father was actually spending most of his time on airplanes, constantly seeking out isolated research groups that were already doing work consistent with his vision of human-computer symbiosis; his goal was to forge them into a nationwide community

On April 25, 1963, in a memo to “the members and affiliates of the Intergalactic Computer Network,” he would outline a key part of his strategy: to connect all their individual computers and time-sharing systems into a single computer network spanning the continent.

By the late 1960s such visions would inspire Dad’s hand-picked successors to implement his Intergalactic Network, now known as the Arpanet. By the 1970s, moreover, they would begin to expand the Arpanet even further, into the network of networks known today as the Internet

Advanced Research Projects Agency Command and Control Research
J. C. R. Licklider, Director

Chapter 1:      Missouri boys

Chapter 1

Missouri boys

Whenever I would finally come to Lick with the proof of some mathematical relation, I’d discover that he already knew it. He hadn’t worked it out in detail, he just ... knew it. He could somehow envision the way information flowed, and see relations that people who just manipulated the mathematical symbols could not see

Extremely intelligent, intensely creative, and hopelessly generous

born in St. Louis on March 11, 1915

the boys of St. Louis were model-airplane-mad. And nobody could build them better than Robnett Licklider

one of the kit-manufacturing companies paid his way to a national air show in Indianapolis so he could show the visiting fathers and sons how it was done

And then, with the approach of that all-important sixteenth birthday, his interest shifted to cars

Underlying his meticulous craftsmanship and insatiable curiosity was a complete lack of patience for sloppy work, easy solutions, or glib answers

He also possessed a streak of mischievous anarchy

he started introducing himself by his fraternity nickname: “Call me ‘Lick.’”

What he had stumbled across was “physiological” psychology, a line of research that was then in the midst of a remarkable period of ferment. Today it would be known as neuroscience

Lick was in heaven. Physiological psychology had everything he loved: mathematics, electronics, and the challenge of deciphering that ultimate gadget, the brain

In 1937, he finished his undergraduate career at Washington University with a triple degree in physics, mathematics, and psychology. He stayed on for another year to earn a master’s degree in psychology

Wars are noisy, to put it mildly. And by the eve of the United States’ entry into World War II, it was clear to the Army Air Corps that war had become intolerably noisy in military aircraft

So in the late summer of 1940, with the Battle of Britain already raging over London, the Air Corps decided that its increasingly urgent preparations for war should include a major research program in practical acoustics

leadership of the project fell to Leo Beranek

To head this Psycho-Acoustics Laboratory, as it came to be known, Beranek selected a thirty-three-year-old Harvard psychologist named Stanley Smith “Smitty” Stevens.2 Good choice

by the end of the war, his lab would be a beehive of almost fifty young researchers. Among the first, of course, was an affable twenty-seven-year-old named J. C. R. Licklider

It also has to be said that life under Stevens was not exactly low-key. “When [Stevens] was seriously interested in a problem, he could move forward only at full speed,”

And yet Miller, Lick, and many others somehow managed to thrive in that atmosphere, Miller noted

Lick, in particular, was in his element. He was just as committed to clarity and excellence as Stevens was. And besides, playing around in the lab until the wee hours was his idea of fun.

went out to the clothing stores and bought up all these snaps

He would stay in the lab until one or two in the morning putting together boards and wires with snaps on them. He called this setup the Snappiac. He knew that if you asked the university shop to build you a gizmo for your experiment, it would require two or three months. But with the Snappiac, he could rearrange components very easily, so he could put together any kind of amplifier or electronic components he wanted.”

in his soft-spoken way, Lick was proving to be a natural-born leader

Back in Cambridge, meanwhile, Lick had attracted the somewhat perplexed attention of Louise Carpenter Thomas, a twenty-three-year-old Kansan who worked as one of the only two secretaries (and only two females) in the lab. She was married to a graduate student over in the main psychology department

Everything he did—whether it was the puns, which were constant and horrible, or the stories he would tell—everything he did was multilayered

Louise’s divorce was finalized in late 1944. She and Lick were married in Harvard’s Appleton Chapel on January 20, 1945

when he went after the people at the top, the ones who could really teach him something, he was rarely finding them at Harvard. More and more, he was finding them two miles away, on the other side of Cambridge—at the Massachusetts Institute of Technology

since the beginning of the war, MIT had been the home of a top-secret crash program known as the Radiation Laboratory

in the 1930s, under the inspired presidency of the physicist Karl Compton, MIT had begun to transform itself into a world-class center for physics, chemistry, mathematics, architecture, and much else besides.

right there at the intellectual epicenter was a portly little man who could often be spotted wandering through the MIT campus in a natty suit and vest, his stubby fingers gripping an enormous cigar

How’s it going?” Norbert Wiener would invariably ask as he dropped into someone’s office unannounced

Wiener’s impromptu wanderings were known around campus as the Wienerweg, German for “Wiener’s path,” and might include conversations with anyone from janitors on up.

entry into nearby Tufts University at eleven. In 1913 he received his Ph.D. in mathematics from Harvard, at age eighteen.

studying with such figures as the philosopher-mathematician Bertrand Russell and the great German mathematician David Hilbert

The school wasn’t oriented toward research at all. However, no one seems to have informed Wiener of that fact, and his mathematical output soon became legendary

People, in short, were not Wiener’s forte. He was as famous for his feuds as he was for his mathematics.

amid all that explosive intellectual energy in wartime Cambridge, Wiener was doing as much as anyone to provide a focus for it, a forum where people could get together and talk until all hours about crazy, visionary ideas—especially his kind of ideas

The venue was Wiener’s monthly series of “supper seminars

Lick was usually there, of course, along with Stevens, Miller, and many others from the Psycho-Acoustics Lab

“The thought of every age is reflected in its technique,” Wiener asserted

Unlike clocks and steam engines, he argued, the emerging technologies of the modern age didn’t just operate blindly, forging ahead without any reference to the world around them. Rather, they operated responsively, taking in information from their surroundings to guide their future actions

They already exist as thermostats, automatic gyrocompass ship-steering systems

these so-called control mechanisms, or servomechanisms

the dominant sciences of the past had largely been physical sciences

But the central sciences of the modern era would increasingly deal with systems-level issues such as communication, control, organization, and information

the new sciences would point the way toward the fundamental nature of life and the mind, mysteries that the physical sciences had never been able to touch

Out of all the developments that had led him to this conviction, Wiener would explain as Lick and the others happily listened, perhaps the first and most important was the automatic computer. In fact, he had been following this technology ever since the mid-1920s, when his wanderings through the MIT campus had led him to the electrical engineering department and a professor there named Vannevar Bush.

Chapter 2:      The last transition

Chapter 2

The last transition

Best of all, for Wiener, was that Bush shared his fascination with computing devices

Wiener’s lifelong interest with the workings of the brain and with the fundamental underpinnings of intelligence (in 1909 he had actually started his graduate studies at Harvard in the zoology department

until his hopeless clumsiness at the lab bench sent him back to mathematics, he had hoped to make brain science his career

For Bush, the inspiration had come early in the 1920s, when he tackled one of the most vexing technical problems of the day, the instability of electric power networks

Differential Analyzer, an elaborate system of gears and pulleys and parallel rods that took up most of a large room

setting up the analyzer for something as complicated as the network could take as long as two days. But no matter: once all the connections were complete, solving the equation was as simple as starting the analyzer’s motor

By the mid-1930s, researchers from all over the world were flocking to the analyzer, using it not just for electrical engineering but for atomic physics, seismology, and astrophysics

And the original analyzer had undergone numerous upgrades—including several suggested by Norbert Wiener

Bush himself, meanwhile, had even bigger plans. In the spring of 1935, once the Depression had eased a bit, he got the Rockefeller Foundation to contribute eighty-five thousand dollars for the development of a second-generation Differential Analyzer based entirely on high-speed electronic circuits

after the “Rockefeller Analyzer” was finally dedicated, in 1942, following numerous delays due to its complexity, it would go on to do yeoman computational work during World War II.

Bush had become increasingly distracted from his computer work ever since 1932, when MIT president Karl Compton asked him to serve as the first dean of MIT’s newly formed engineering school

even greater anxiety over the looming threat of a new world war, and a conviction that the United States would almost inevitably be drawn into that conflict

in late 1938 he accepted an offer to become director of the Carnegie Institution of Washington, a major private research foundation that just happened to be located at the seat of power, where he hoped he could have some influence

1933. Toward its end, almost in passing, he fantasized about being able to fit an “unabridged dictionary on a square foot of film,” and about “the contents of a thousand volumes [being] located in a couple of cubic feet in a desk

Bush was talking about microfilm, of course

idea to reach a wider audience. So on December 7, 1939, having prepared an article that he hoped would be accessible to lay readers, he submitted the forty-five-page typescript of “Mechanization and the Record” to Eric Hodgins, the publisher of Fortune magazine

Bush was writing here for a general audience, so perhaps his emphasis on nonprofessionals shouldn’t be a complete surprise. Nonetheless, his vision of high technology’s enhancing and empowering the individual, as opposed to serving some large institution, was quite radical for 1939

Even more radical, however, was Bush’s second new idea, which was the mechanism for implementing his vision. Instead of storing those countless microfilmed pages alphabetically

arbitrary—Bush proposed a system based on the structure of thought itself. “The human mind ... operates by association

Such a device needed a name, added Bush, and the analogy to human memory suggested one: “Memex.”

once a Memex user had created an associative trail, he or she could copy it and exchange it with others. This meant that the construction of trails would quickly become a community endeavor

Today we know it as hypertext

The Memex article we know today is a revised version that appeared in the July 1945 issue of the Atlantic Monthly under the title “As We May Think.”

On September 1, 1939

Hitler’s invasion of Poland

Six months afterward, following an intensive backstage lobbying effort, Bush walked into the White House bearing a one-page proposal for a National Defense Research Council (NDRC), a high-level committee that would coordinate all the country’s war-related technology development—with Bush, naturally, as its chairman. The lobbying paid off.

Norbert Wiener wasted no time. Indeed, he’d felt the coming crisis even more keenly than Vannevar Bush had.

So in the late summer of 1940, when Bush asked researchers all over the country to propose R&D projects that would help prepare the nation for its likely entry into the war, Wiener answered quickly. Build computers, he advised his friend

Wiener mailed off his twelve-page memorandum on September 21, 1940, fully expecting a yes from Washington. In due course, however, the answer came back: no. Bush gently suggested that his friend turn his talents toward something more, um, practical

But in the end, with research dollars scarce, Bush clearly felt he had to make choices

his most urgent priorities had to be the development of radar, antiaircraft fire control, and antisubmarine warfare

Actually, all Bush had been looking for was a bright young electrical engineering graduate who would be willing to work his way toward an MIT master’s degree.

His name was Claude Shannon

now here was the Differential Analyzer. “The main machine was mechanical with spinning discs and integrators,” he told Omni, “and there was a complicated control circuit with relays. I had to understand both of these. The relay part got me interested.

as he traced his way through the control circuits, he was also struck by how closely the relays’ mechanical operation resembled the workings of symbolic logic

Indeed, as Shannon told Omni, the analogy worked perfectly—and no one had ever noticed it before. So he made it the topic of his master’s thesis and spent most of 1937 working out the implications

Any desktop adding machine could do simple addition and subtraction; a relay circuit could decide

hands-on tinkerers were already plunging ahead with their own ideas for computing, and in the process achieving most of the other transitions on Wiener’s list

From decimal math to binary math

binary math not only had deep ties to logic, as Shannon had pointed out, but also greatly simplified the design of switching circuits—which is probably why it was independently rediscovered on several different occasions

In late November 1937, for instance, a young mathematical engineer named George Stibitz

being a mathematician, he was immediately struck by the same on-off, true-false, 1-0 equivalence that Shannon had perceived

Before long, Stibitz’s original task was forgotten, and it had become clear to everyone at Bell Labs that his relay circuits were much more than toys. In the early summer of 1938, his boss asked him if he could build a calculator that would do arithmetic with complex numbers—the kind involving “imaginary” quantities based on the square root of −1. Those quantities had turned out to have some very real applications in the design of AT&T’s new coast-to-coast system of long-distance lines

The result, completed in October 1939 at a then-breathtaking cost of twenty thousand dollars, was the Complex Computer. Equipped with some 450 relays and three modified Teletype machines for entering problems and printing out the answers

It’s hard to say which impressed the attending mathematicians more: the Complex Computer itself or the fact that they were communicating with it through the world’s first computer network. “Williams designed an interface so that the number signals could be serialized and transmitted over a standard telegraph line,” explained Stibitz

From mechanical to electromechanical to fully electronic

The first person to successfully take on the challenge of creating a fully electronic computer was a physics professor named John V. Atanasoff, of Iowa State University

1937, he just couldn’t stand it anymore

By 1939 he had his designs worked out and was seeking funding

Chapter 7:      The intergalactic network

Licklider would carry around one of these big thick briefcases crammed full of papers, and letters, and correspondence, and all kinds of stuff that he had to make decisions on. And he would occasionally ‘lose’ the briefcase

Not surprisingly, this sort of behavior infuriated Lick’s colleagues, who found that they had to badger their new director for days, weeks, or months to get anything done.

The near-universal enthusiasm that had greeted Lick’s accession the previous fall had become near-universal exasperation. At ARPA, his colleagues now realized, Lick had succeeded because he never had to be a manager; the job had left him free to do what he did best, which was be a visionary. But at Project MAC, where he did have to be a manager, the top administrators had long since quit in frustration

there’s no sugarcoating this fact, either: J. C. R. Licklider again and again took jobs that required him to be an administrator—and then was almost contemptuous of the skills it took to do those jobs right. It was perhaps his greatest failing

And sadly, it was a failing that was not confined to paperwork. For someone who was so warm and empathetic face-to-face, Lick could be remarkably maladroit in dealing with Tech Square’s delicate political sensitivities. Take his steadily deteriorating relations with the AI Lab, for example. It’s true that Minsky and his crew were infuriating at the best of times.

What characterized the “winners,” of course, was their fervent, almost religious conviction that they were on the verge of creating truly intelligent machines—not just understanding intelligence, or modeling intelligence, but creating it

It was a belief that flowed from Minsky himself, says another graduate student of that era, Patrick Winston, who would later succeed Minsky as the director of the lab. “I think he wanted to make a contribution as fundamental as Darwin’s and Freud’s

They never doubted for a minute that they could do it, agrees Winograd, who has since come to entertain a great many doubts

why everyone who wasn’t a winner was a loser: to Minsky and his followers, the “computer utility” vision that animated the rest of Project MAC was boring

At best it was a means to an end, a way to get the computer power they needed for their real work. Likewise this stuff about on-line communities: if the outside world was full of losers, why bother talking to it? And as for Lick’s vision of human-computer symbiosis, what was the point? Why waste your time augmenting human intelligence, when humans were virtually obsolete?

Minsky and his crew were very much children of the sixties: brilliant, touchy, self-congratulatory, and deeply suspicious of the establishment—meaning the likes of Fano, Corbató, and now Lick. Handling them required exquisite sensitivity and tact on the part of the director. Bob Fano had managed the task fairly successfully. Lick, sadly, did not.

henceforth, Project MAC and the AI Lab would go their separate ways, as completely separate organizations.

By that point, the separation was probably as much a relief to Lick as it was to Minsky, but it was deeply embarrassing nonetheless. And in the meantime, ironically, Lick hadn’t been doing any better with the Multics crowd.

Originally planned for completion in 1966, the second-generation time-sharing system was already more than two years late

Fundamentally, says Corbató, Multics was just too complicated

three-way agreement among MIT, General Electric, and Bell Labs, no one was really in charge, which made for confusion and slow decisions

Second, the hardware needed a massive upgrade

meant that none of the software from the original machine was usable

All of which led to the final problem: PL/1, the new and supposedly all-purpose programming language that IBM had introduced along with the System/360. “In retrospect, PL/1 was a monstrosity to be trying to use to design a system,” says Corbató

what we all failed to foresee was that the language had gotten terribly complicated in its specification. And when the company failed to deliver, having strung us along for twelve or eighteen months, we were suddenly in deep, deep trouble, because we were totally dependent on this thing

IBM was running into equally severe problems in its crash program to develop a time-sharing computer compatible with the System/360 line

IBM eventually went through eight revisions of TSS, most involving the complete redesign of one or more major components. And yet, as Alan Perlis later groused, “it never succeeded in doing what it was supposed to do.”

Shared naïveté or no, however, Lick had begun to see Multics as a horrendous opportunity cost for Project MAC, a diversion of talent and effort that had left the actual users in limbo. He also knew that ARPA’s patience wasn’t going to last forever, given the new emphasis on “relevance.”

“Many are waiting to see how the second generation of general-purpose multiaccess systems [that is, Multics and TSS] turns out. [If they] turn out to be too complex to implement, or to operate efficiently, there may be a general retrenchment to noninteractive applications. ... That would be a great loss, I think, to the universities and, indeed, to all who want to use computers as aids or partners in learning, experimenting, modeling, solving problems, making decisions, and other intellectual activities.”

it wasn’t the only game in town anymore. The time-sharing system developed at Project Genie might have been somewhat less ambitious, but it was very capable nonetheless: just look at what Engelbart and his crew had done with it. Then there was DEC’s big PDP-10

Lick thought it was high time for Project MAC to get some answers. But that, alas, was where his maladroitness came into play. His way of dealing with the problem was to have ARPA call for a full-scale, outside review of Multics to determine if the time had come to terminate it.

insensitive, and autocratic. Worse, they felt betrayed. Lick, as the director of Project MAC, was supposed to be on their side,

It didn’t help that Multics itself in the end proved to be something of a disappointment

GE had long since lost interest. In 1970, in fact, the company sold its entire computer division to Honeywell

most people in the ARPA community ended up turning to the PDP-10 and Tenex, a new time-sharing system created for that machine by BBN. And then, of course, came the microcomputer revolution of the 1980s, which made Multics look like a complete dinosaur

If Multics became a dinosaur, however, it was a remarkably influential one.

Multics was a prototype for virtually all the operating systems to follow. And along the way, not incidentally, the project became a model of planning and documentation

And then, of course, there were two young Multicians, named Dan Bricklin and Bob Frankston, who would go on to galvanize the emerging microcomputer industry with a little program called VisiCalc, the first electronic spreadsheet

that’s not even counting its biggest, if least direct, legacy, which came in the form of a backhanded compliment

Bell Labs announced that it was dropping out of the Multics partnership

put the laboratory on track to buy a conventional GE 635 computer that would operate in a conventional batch-processing mode. But then a funny thing happened: two of Bell’s former Multicians, Ken Thompson and Dennis Ritchie, began suffering from withdrawal pangs

Thompson and Ritchie found themselves craving that interactivity. So, good hackers that they were, they decided to write a little time-sharing system of their own. They had to sneak it in on the side, since their superiors weren’t about to OK another time-sharing project

By mid-1970, they had a preliminary version up and running. Somewhere along the way, moreover, their homebrew operating system had acquired a name.

Unix

In the fall of 1969 he found himself embroiled in a controversy over the Cambridge Project, a separate ARPA-funded effort that he had helped to organize in collaboration with Bob Taylor, Jerry Wiesner, and a number of psychologists