Part Two — Licklider’s Dream

Members and Affiliates of the Intergalactic Computer Network
— J. C. R. Licklider, memorandum, 23 April 1963

Chapter Five: Rise of the Machines

In a few years, men will be able to communicate more effectively through a machine than face to face. That is a rather startling thing to say, but it is our conclusion.

— J. C. R. Licklider and Robert Taylor, ‘The Computer as a Communication Device’, 1968

He preferred to be called Lick.

The full name — Joseph Carl Robnett Licklider — went on the papers and the cheques. The people who worked with him remembered the warmest and least vain of the senior figures in their field, a man who gave ideas away without keeping count of them and who seemed genuinely surprised when anyone thanked him. In the room, he was Lick.

He was a psychologist.

This is the first thing that goes missing when the story is told backwards from the present, because we assume the medium was made by engineers, and it was — but not first. The man who imagined what the internet might be had a doctorate in psychoacoustics from Rochester, taken in 1942, and a training in the branch of empirical psychology that asks how a human being, dropped into a noisy and confusing world, manages to make sense of what reaches the senses.

He spent the war on that question in its most literal form. At Harvard’s Psycho-Acoustic Laboratory he worked on speech intelligibility in the cockpits of B-17s: how to build a radio, a microphone and an earpiece such that men flying through flak at fifteen thousand feet could understand one another over the engines.

It is a problem of acoustics, engineering, anatomy and attention all at once. And what a person learns from working on it is that the line between a human being and a machine is less solid than it looks. Get the design right and the two of them together carry a signal that neither could carry alone.

He thought about that for the next thirty years.

⁂

After the war he moved to MIT, and from MIT into the orbit of the strangest computing project of the early Cold War.

SAGE — the Semi-Automatic Ground Environment — was a continent-spanning air defence system built to spot Soviet bombers in real time and direct the response. It was a monstrosity. Most of the engineers who worked on it have said since that it would have been close to useless in an actual nuclear war, which by the time it was finished meant missiles rather than bombers.

But it produced something nobody had ordered.

The radar operators at SAGE consoles did not feed cards into a machine and wait for a printout. They sat in front of screens. They pointed at objects on those screens with a light gun. The machine answered at once. By the standards of every other computer then in existence this was almost unimaginably fast, and Licklider, consulting on the human-factors design, saw what it meant.

The computer used this way was not a calculator. It was a place to think.

The human at the console was not asking the machine for arithmetic and waiting. The two of them were working on the problem together, each doing what it was suited for, each handing back at the right moment — and the combination was performing acts of cognition that neither half could have performed alone.

This was not what computers were for. Computers in 1958 were enormous things that lived in cooled rooms, ate punched cards, and produced reams of paper that a human read afterwards at leisure. They were instruments of calculation, and the whole profession agreed about it.

Licklider had begun to suspect that the whole profession was suffering from a failure of imagination so complete that nobody in it had noticed.

⁂

In 1957 he left MIT for Bolt Beranek and Newman, an acoustics consultancy in Cambridge that was becoming, almost by accident, one of the most important computing laboratories in the country. And there, at the end of the fifties, he sat down and wrote the paper.

‘Man-Computer Symbiosis’ appeared in March 1960 in a technical journal with a small readership. It opens with a fig tree and a wasp.

The wasp lays its eggs only in the fig. The fig is pollinated only by the wasp. Taxonomically, the two are entirely separate organisms; functionally, they are one thing, and neither continues without the other. This, Licklider proposed, was what human beings and computers might become to each other. Not a master and an instrument. A partnership in which each did what the other could not, and the pair thought things that neither would have thought.

The grand metaphor rests on a small and slightly embarrassed piece of self-observation.

He had kept track, for several months, of how he actually spent his working hours. Not how he imagined he spent them. How he spent them. And he found that around eighty-five per cent went on activity that involved no thinking whatever: looking things up, plotting data, running the same equation again with a different parameter, drawing graphs, hunting references, filing notes, retyping drafts. The work he had been hired for — the work he had trained two decades to do — took up perhaps fifteen per cent of the time available to do it.

The rest was friction.

If a machine could be made to absorb the friction, he reasoned, a man might spend most of his life on the only part of the job that required him.

That is the whole of it. The computer takes the plumbing; the human does the thinking. The machine is scaffolding — holding what a mind cannot hold, finding what a mind has mislaid, running what a mind cannot run, and doing all of it fast enough that the thinking is never interrupted by the doing.

Notice what is absent from the design.

It does not capture attention. It does not measure engagement. It does not infer desire from clicks, or maximise time on device, or carry advertising, or have a feed. It is a quiet, fast, patient assistant, and its entire purpose is to make the person at the keyboard a better thinker than they were. It is a tool in the oldest sense — an extension of a human capacity, owned by the human, serving the human’s ends.

Like the fig and the wasp, it only works if both of them live.

⁂

Two years later he was given money.

In October 1962 Licklider was appointed to run the Information Processing Techniques Office at ARPA — the Pentagon’s blue-sky research agency, founded in the panic after Sputnik to ensure the United States was never again technologically surprised. His budget, in present money, ran to hundreds of millions. His hand was remarkably free.

The Pentagon wanted command-and-control systems that could survive a nuclear exchange. What Licklider funded was an entire research field devoted to human beings and computers thinking together.

He funded Douglas Engelbart at the Stanford Research Institute, whose Augmentation Research Center would produce the mouse, hypertext, real-time collaborative editing, the cursor and the windowed display, years before these ideas reached ordinary consumer computing. He funded Project MAC at MIT, and with it the time-sharing systems that let many people use one machine at once — the technical foundation of everything later called online community. He funded Ivan Sutherland, and with him computer graphics. He funded, through the men who followed him at IPTO, the work that became the ARPANET.

It should be said plainly that this money came from the Department of Defense, at the height of the Cold War, in the years America was escalating in Vietnam.

Licklider knew that. He was not a naïf who wandered into the Pentagon and was handed a chequebook by mistake. He was a man who had understood the institution well enough to describe humane research in language a defence agency would pay for, and who then spent the money on what he had wanted to spend it on all along. Whether that makes him a subversive working the system from inside, or a man who told himself a comfortable story about the source of his budget, is a question I cannot settle. The answer probably contains some of both.

What is not in question is what he did with it.

⁂

In April 1963, six months into the job, he wrote a memorandum. He addressed it to the ‘Members and Affiliates of the Intergalactic Computer Network’.

The title was a joke. The content was not.

The memo proposed, for the first time in the institutional record, that the various projects ARPA was paying for should be built not as separate machines but as a network — common standards, shared resources, and the ability for any researcher at any site to draw on the computing power and the data and the colleagues available at every other site.

The absurd name was doing work. Licklider was describing something so much larger than anything that existed that a deliberately ridiculous label was the only way to say it out loud without being dismissed. Give a thing a silly enough name and you can propose it seriously.

What he had described was the internet. Not the internet that came to exist — the internet as he meant it: a distributed community of minds drawing on distributed resources in the service of distributed thought.

The computer was the medium, not the destination. What travelled across it were ideas, questions, half-formed thoughts belonging to the people connected to it.

⁂

A problem can be larger than any single mind available to it. What Licklider was describing was an arrangement in which minds might hold one problem or cultivate one idea between them — each carrying a part, none carrying the whole, and the whole nevertheless becoming visible to all of them.

That is not a claim about efficiency. It is a claim about apprehension. He thought a network of linked minds might see something no individual mind could see.

Which is the same problem Sagan would spend twenty years solving with a camera.

And then there is a fortnight in December 1968 that nobody has ever set side by side.

On the ninth, in a hall in San Francisco, Douglas Engelbart sat at a console in front of a thousand computer professionals and showed them, in ninety minutes, the mouse, the cursor, hypertext, windowed displays, shared editing, and a live video link to a colleague thirty miles away. It is called the Mother of All Demos. The laboratory was Licklider’s doing; he had put ARPA money into Engelbart five years earlier, when almost nobody else would.

What Engelbart demonstrated that afternoon was that a mind could be handed a whole body of work and go straight to any part of it — as fast as it could decide where to look.

Fifteen days later, William Anders looked out of a window above the Moon and photographed the Earth.

From here, they look like the same operation. Both were attempts to make a whole available to a mind that could not otherwise hold it. One did it by leaving the planet. One did it by linking the minds. Neither man knew what the other was doing, and nobody at the time remarked that the two demonstrations had fallen in the same fortnight of the same year.

Nobody had any reason to. The connection is only visible from here.

⁂

Five years after the memo, back at MIT, Licklider set out his most complete public statement of the case with his ARPA successor Robert Taylor. ‘The Computer as a Communication Device’ appeared in April 1968, and opens with the most famous prediction in the history of computing: that within a few years, human beings would communicate more effectively through machines than face to face.

The ARPANET had not yet sent a packet.

The rest of the paper is a manifesto for what would later be called online community. Licklider and Taylor set out how networked machines would let people separated by geography form working groups around shared interests. They anticipated electronic mail. They anticipated message boards. They described something close to Wikipedia — a body of knowledge maintained collaboratively by distributed contributors. They described a person in a small town taking full intellectual part in a community whose members lived on five continents and had nothing else in common but the same problem.

And they worried about who would be let in.

They understood that a medium of this power, unequally distributed, would produce an inequality of a kind the world had not previously had to think about, and they wrote that the question of access was political rather than technical.

They were right about an astonishing amount of it, which is what makes the paper difficult to read now. Every part of the machinery arrived. By 1972 the network was working. The people who built what came next had those same papers available to them and arrived at something close to the opposite.

⁂

It is worth stopping for a moment on what that was actually like.

A graduate student sits in a terminal room somewhere in the early seventies. It is late. The room smells of hot electronics, and the machine makes a noise like a fan heater.

She has a problem she cannot solve, and she has just found out that a man at another university, two thousand miles away, has been working on the same thing for a year. She has never met him. She types a message and sends it, and there is nothing to do but wait, because he is asleep in a different time zone.

In the morning there is an answer. It is long. He has thought about it overnight and worked something out that she had not, and at the end of it he asks her a question she cannot answer either.

Nobody is selling either of them anything. Nobody is counting how long they spend. There is no feed and there is nothing else on the screen. Two people who will probably never meet are thinking about one problem, in turn, across two thousand miles, because a machine has made it possible for them to hand it back and forth.

That was the thing. That is what it was for.

It still works, incidentally. You can still do exactly that today. It is just that not so many people do, and the machine on which you would do it is now doing several other things at the same time, all of them louder.

⁂

Licklider died in June 1990.

He had seen the ARPANET become more than a research network. He had seen electronic mail become, among researchers at least, the ordinary way of doing business. He had not seen the network opened to commercial traffic, which began in earnest in the year after his death, and he had not seen what commerce would do with it.

He died with no strong reason to believe that the thing he had spent his life imagining was not, at last, arriving.

He was wrong about that. He was not wrong about what the medium could do.

The architecture he described is still what the underlying network actually predicts. Everything laid over it since — the engagement metric, the recommender, the fifteen-second increment, the profile — is a second structure, built on top by people who looked at what he had made and saw a supply of something they could take.

The first structure is still there. One can still send an email. One can still keep a small site. One can still join a forum organised around a question rather than around a person. It is legible in the architecture to anyone who knows where to look — and almost nobody who uses the medium today would describe its purpose in the terms its designer used.

⁂

Dostoevsky drew a version of this as a man in 1868, in a novel that was not about technology and could not have been.

The parallel is not biographical and does not need to be. What it offers is a way of seeing what happened to Lick that the history of the technology, by itself, cannot supply.

Chapter Six: The Prince Who Saw

The novel begins on a train.

A young man in a thin Swiss cloak, badly under-dressed for the November cold of central Russia, sits in a third-class carriage of the Warsaw–Petersburg line as it pulls into the capital at nine in the morning. He has been away for four years, in a sanatorium in the mountains, treated for an illness from which he has now mostly recovered. He has almost no money, no family waiting for him, and a single small bundle. The autobiography he can offer his fellow passengers amounts to a few fragments of a childhood he barely remembers.

His name is Lev Nikolayevich Myshkin. He is the last of his line, and the title of prince, in his case, is mostly an embarrassment — attached to a man who owns nothing and knows almost nobody. He is travelling, this November morning, into a society that will spend seven hundred pages discovering what use can be made of a man whose goodness is open and unguarded.

⁂

This was the problem Dostoevsky had set himself, and he knew even before he began that it might not be solvable.

In a letter to his niece Sofya, written from Geneva in January 1868 while the novel was still in its opening chapters, he described the difficulty. The idea was to depict a positively beautiful man, and nothing in literature, he believed, was harder. Moral beauty — the unmediated goodness of a soul that has not been bent — is almost impossible to render, because the writer’s own corruption keeps leaking into the writing. Christ was the only complete instance. Don Quixote was the closest secular attempt, and beautiful chiefly because he is ridiculous.

Dostoevsky would try, at full scale, to make a beautiful man who was not ridiculous.

What the world ends up calling that man is the novel’s title. The frame the author worked under was different. He was writing about goodness — direct, unprotected, incapable of the small flexings of self-interest that ordinary people perform without noticing — and his question was what such a goodness would look like once it was set down in a society organised, like every society, around the management of self-interest.

⁂

The illness supplies the novel’s central image, and Dostoevsky wrote it from the inside; he had the disease himself.

Myshkin is epileptic. The seizures are preceded by an aura lasting a few seconds, in which the boundaries of ordinary perception dissolve. What arrives in those seconds is not a feeling but an apprehension: that everything is in its proper place, that the disharmony has been resolved, that one is for a moment in direct contact with something one has no other means of reaching. It cannot be held, and it cannot be brought back. Myshkin says he would trade an ordinary lifetime for it.

The seizure that follows undoes him. But while he was inside it, he had seen, and the seeing leaves a residue.

He returns from Switzerland with a directness of moral perception that the society he is walking into has no equipment to receive.

And none to refrain from using.

⁂

The novel returns again and again to a single fact about him: he cannot lie.

Not that he refuses on principle to tell untruths. The protective coating that ordinary social life requires — the small adjustments of self-presentation by which any of us manages to seem slightly more than we are and slightly less than we feel — is simply absent in him. He says what he sees. He loves what he loves, and he loves people most when they are damaged, and he says so.

When Nastasya Filippovna, the most ravishing and most damaged woman in Petersburg, sweeps into a drawing room and the assembled men assess her in the calibrated language of their desire, Myshkin tells her, in front of all of them, that she has suffered too much, and that he does not believe she is the woman she has made herself appear to be.

He says it without calculation, because in him the perceiving and the saying are one act.

⁂

Nastasya, recognising immediately what he is, begins to use him — as the witness to her own degradation, as the saviour she will not permit to save her, as the man whose love will give her death the shape of a sacrifice rather than a squalor. Rogozhin uses him as the rival whose existence legitimises an obsession. Aglaya uses him as the man who might rescue her from the role her family has assigned her. Ganya tries to use him for money. General Ivolgin uses him as a confessor for his lies.

Every significant character in the novel finds a use for Myshkin. His goodness is the resource on which the entire social machinery runs. And they use him not because he is foolish, and not because he fails to notice. They use him because he is what he is — open, unprotected, structurally unable to withhold what he gives. In a society organised around self-interest, that openness is an extraordinary natural resource, and almost anybody can find a use for it.

He is destroyed not because he is weak but because he is good and he is open.

I should say that this is a simplification, and that Dostoevsky’s Myshkin is a great deal stranger than the figure I have been describing.

He is not innocence. He is erotically entangled with two women at once and handles it disastrously. He is fascinated by suffering in a way that is not entirely healthy — he cannot stop looking at it, and the novel makes something uncomfortable of that. He causes real damage, repeatedly, to people he is trying to help, and Dostoevsky is careful to show the damage. The book is not sure he is admirable and does not ask the reader to be sure either.

What survives all of that complication is the structural point, and it survives precisely because it does not depend on him being good.

An unguarded openness, set down without anything between itself and the appetites that will find it, is vulnerable in a way that depends on the openness rather than on the virtue. Remove the openness and there is nothing to take. The quality and the exposure are the same quality.

⁂

The end of the novel is among the most devastating passages in nineteenth-century literature, and it is devastating precisely because nothing in it is violent.

Rogozhin has murdered Nastasya. Myshkin spends the night beside her body, in the dark of Rogozhin’s apartment, stroking his rival’s hair while Rogozhin drifts in and out of fever and weeps.

By morning Myshkin is gone. Not dead. Gone in a deeper sense. The seizures have taken him back, the Swiss doctor’s four years of work are undone, and he is shipped to the mountains to live out his life in the silent, uncomprehending state from which he was briefly woken.

Nothing in the city killed him. The city merely spent him. The goodness that arrived four years earlier in a third-class carriage was drawn down by the ordinary requirements of the people around him until there was nothing left to draw.

⁂

A man whose goodness was used by everybody who came near it. That is the figure, and it is the one that names what happened to Licklider.

Not the biography. Lick was received perfectly well. He was admired in his field, promoted, funded, and remembered warmly by everyone who worked for him; he died in his bed with his reputation intact. He was not consumed the way Myshkin was consumed. What was taken from him was taken from the design.

⁂

The standard story is technological. The internet was opened to commercial traffic in the early nineties. The dot-com boom created enormous wealth. Advertising emerged as the dominant business model, the engagement metric became central, and the system evolved, by a kind of impersonal market logic, into the attention economy we now inhabit.

Nothing in that story is factually wrong. It is wrong only as an account of agency. Systems of this kind do not evolve the way a coral reef evolves. This one was built, by particular people, making particular decisions for the ordinary reasons — advancement, the pleasure of building something one’s peers would call important, the wish to be on the side of a thing that was winning — inside structures of capital that rewarded one kind of building and not another. The polite version flatters the builders by describing their choices as weather. They were choices, made inside a system designed to make them feel inevitable.

And what those people saw, looking at what Lick and his successors had built, was what Rogozhin and Ganya and Nastasya saw looking at Myshkin.

They saw a resource.

⁂

It is worth being precise about the mechanism, because the loose version of this argument is easy to refute and the exact version is not.

The early network was not built on trust in the sentimental sense. It was built for interoperability — for the specific purpose of letting machines at different institutions, running different hardware, under different administrations, exchange data without any of them having to ask permission of the others. That is a design goal, not a moral posture. But it carries a consequence. A system in which any node may address any other, and in which the network itself does not inspect what it carries, is a system with very few places to put a gate.

Consider what the designers left out. Mail was specified without any means of verifying that a sender is who the message claims — not from carelessness, but because in a network of a few dozen research institutions the question did not arise, and because building verification into the protocol would have required an authority to verify against, which was precisely what the design was avoiding. The same reasoning runs through the architecture. A link is an offer, freely made, and no permission is sought from the page being linked to. A page may summon resources from any other server in the world, and the reader is not asked.

None of this was surveillance. None of it was designed to permit surveillance. It was designed to let a small number of people who already trusted each other work together without asking anyone’s leave.

What it produced was a substrate.

The extraction came later, and it took real engineering. Nobody exploited a protocol. What happened is that a set of ordinary, individually reasonable capabilities — a mechanism for a site to remember a returning visitor, a mechanism for a page to load an image or a script from a third party, a mechanism for that third party to be the same third party across thousands of unrelated sites — turned out, when combined and commercialised, to constitute a surveillance apparatus that no single component had been designed to build. Cookies were a session-management convenience. The third-party request was a way to share a logo. Put them together, add an exchange that sells the resulting attention in the fraction of a second before a page renders, and you have a dossier assembled on a reader who was never asked, by companies whose names he has never heard, on sites that in many cases did not know the full extent of what they had let in.

The openness that made the early network useful was the same openness that made all of this cheap to build. Not because the architecture permitted surveillance, but because it declined to require permission for anything, and so there was no point at which anybody had to be asked.

A more guarded design would have resisted this. It would also have been slower, uglier, less interoperable, and it would not have done the thing Licklider wanted.

The qualities that made the design worth having were the qualities that made it worth taking.

And it is worth saying which quality, because goodness on its own is too vague. What was structural here was openness — and in both men, openness is what the goodness consisted of. Myshkin is not good and also open. He is good in the form of being unable to withhold.

In Lick’s case it had a technical vocabulary. Interoperability. Decentralisation. The absence of anybody whose permission was required. That is what his goodness looked like once it had been built into a protocol, and it is why he thought the thing would bring people together.

It was also, in the structural sense, the substrate of the betrayal.

Which is also why Myshkin belongs in this chapter at all. He is not a figure for the network because he is good. He is a figure for it because he is open — structurally unable to withhold what he has. That is the property in both cases, and it is the property that gets taken.

This is not a mistake he made. It is not a naivety he could have corrected with better foresight. There is no version of the design that delivers what he was after and cannot be harvested — the openness is not a flaw in the mutualism; it is the mutualism. He could have built something defended, and it would not have been the thing he meant.

Myshkin could have learned to lie. He would then have been somebody else.

⁂

Nobody had to reject what Licklider wanted. There was no room in which anybody decided against him, and I have no evidence that a single person who built the second structure ever gave his name a thought.

They only had to optimise for something else.

Licklider’s objective was the cognitive capacity of the person at the keyboard. The objective of what came after was measurable engagement. Same wires. Same protocols. Same architecture, unchanged and still doing exactly what it was designed to do.

A different number at the top of the page, and everything downstream of it rearranges.

That is worse than a betrayal, because a betrayal can at least be argued with. This required no argument, no decision and no ill will. It required a metric.

And the papers were never hidden. They sat in the libraries and the archives the whole time, perfectly legible, in a system that had assembled itself around a different question.

⁂

Both of them are still there.

Myshkin does not die at the end of the novel. He is returned to the mountains, alive, breathing, unreachable — present in the world and of no further use to it. Dostoevsky could have killed him and chose not to, which is a colder ending and a truer one.

And the medium Licklider imagined has not died either. It is all there, functioning, legible to anyone who knows where to look, underneath the second structure that was built on top of it.

Neither was destroyed. Both were spent, set aside, and left alive.

⁂

The message, profound as it was, did not serve the motivations of the people who received it. It was useless to them.

It was not that this was a message they did not care to hear. They simply had no use for it.

Darren Gall