“As We May Think” was published in The Atlantic Monthly in July 1945, across eight numbered sections over eight pages. There are no headings. Vannevar Bush was then the director of the Office of Scientific Research and Development, the wartime agency that had coordinated thousands of American scientists across radar, the proximity fuse, and the bomb. The war was ending; he had written, in a sense on behalf of all of them, an essay about what science should do next.
The essay’s reputation has by now nearly replaced the text. In the standard retelling, Bush “predicts” hypertext, or the internet, or the web, and the story moves on. Reading the text closely produces something stranger and more specific: a careful engineer in 1945 describing the best information machine 1945 physics could support, with the specific acknowledgment — in the final section — that the mechanisms he described were contingent on 1945 technology, not fundamental to the concept. That acknowledgment matters.
The text cited throughout is the Atlantic Monthly version. [1] The Life magazine republication (September 10, 1945) was more widely seen at the time and included photographs, but was not located in a readable text form and was not read directly; whether it differs from the Atlantic Monthly version in any material way is unknown.
”The summation of human experience is being expanded at a prodigious rate”
Bush opens not with a machine but with a problem. The scientific record is growing faster than any individual can navigate: “The summation of human experience is being expanded at a prodigious rate, and the means we use for threading through the consequent maze to the momentarily important item is the same as was used in the days of square-rigged ships.” [1] The indictment is precise — not that the record has become large, but that navigation hasn’t changed. Volume is not the problem. Access is.
He names the cost of inadequate access with Mendel: “Mendel’s concept of the laws of genetics was lost to the world for a generation because his publication did not reach the few who were capable of grasping and extending it.” [1] The diagnosis is the same across Sections 1 through 5: not the quality of the science but the failure of its dissemination.
Sections 2, 3, and 4 inventory the available technologies: improved microfilm and miniature cameras for recording; dry photography processes that write directly without wet chemical development; voice-controlled transcription assembled from existing components (the Voder, the Vocoder, the stenotype); computing machines for repetitive arithmetical and logical operations. These sections read as a brief to a capable engineer on the state of the toolbox — here is what exists, here is what is nearly available. The tools are not the problem.
Section 5 names the problem the machine will solve. Not storage: “we can enormously extend the record; yet even in its present bulk we can hardly consult it.” [1] Not computation — the arithmetical machines of Sections 3 and 4 handle repetitive manipulation. The bottleneck is retrieval. “The prime action of use is selection, and here we are halting indeed.” [1] A different way of finding what you already have was what the diagnosis required.
”It consists of a desk”
The memex, as described in Section 6, is a piece of furniture.
“It consists of a desk, and while it can presumably be operated from a distance, it is primarily the piece of furniture at which he works.” [1] On the desk’s top are “slanting translucent screens, on which material can be projected for convenient reading.” [1] The screens are not displays in any computational sense; they project microfilm. “The matter of bulk is well taken care of by improved microfilm. Only a small part of the interior of the memex is devoted to storage, the rest to mechanism.” [1]
Everything stored in the memex is on microfilm, including the user’s own additions. “On the top of the memex is a transparent platen. On this are placed longhand notes, photographs, memoranda, all sort of things. When one is in place, the depression of a lever causes it to be photographed onto the next blank space.” [1] The note is captured optically, using the dry photography processes described in Section 2, writing directly onto microfilm in place. There is no transcription step, no conversion to text, no digital encoding. The document becomes a microfilm frame.
Navigation through the stored material is also mechanical. “On deflecting one of these levers to the right he runs through the book before him, each page in turn being projected at a speed which just allows a recognizing glance at each.” [1] Faster: “If he deflects it further to the right, he steps through the book 10 pages at a time; still further at 100 pages at a time. Deflection to the left gives him the same control backwards.” [1] The entire navigation vocabulary is a lever’s degree of deflection. There is no shortcut, no pointer, no way to specify a destination independently of the lever.
What the memex is solving, at the hardware level, is a storage-density problem: how to fit a large library into a desktop device and retrieve from it quickly. Microfilm was the right answer in 1945. The reduction ratios Bush mentions in Section 2 — 100:1 being achievable — are enough, Bush calculates, to put the Encyclopaedia Britannica in the volume of a matchbox. [1] The projecting screens solve the readability problem microfilm creates. The levers solve navigation within the mechanical range the film transport can support. Each mechanism is the 1945 solution to a 1945 constraint. The device looks exactly like what it is.
”Photographs the whole trail out, and passes it to his friend”
Section 6 also introduces the memex’s organizing principle: associative indexing. Rather than requiring a researcher to know in advance the hierarchical classification under which something was filed, the memex lets users build personal trails through linked items. Section 7 shows this in use.
Bush’s example is a researcher studying why the short Turkish bow was apparently superior to the English longbow in the skirmishes of the Crusades. “He has dozens of possibly pertinent books and articles in his memex. First he runs through an encyclopedia, finds an interesting but sketchy article, leaves it projected. Next, in a history, he finds another pertinent item, and ties the two together. Thus he goes, building a trail of many items.” [1] The trail is named and stored: “When the user is building a trail, he names it, inserts the name in his code book, and taps it out on his keyboard.” [1] The code book is physical; the trail is a sequence of linked microfilm frames.
Sharing a trail means transferring a physical object. “So he sets a reproducer in action, photographs the whole trail out, and passes it to his friend for insertion in his own memex, there to be linked into the more general trail.” [1] The transfer mechanism is a photograph — a tangible copy of a microfilm trail, handed from one person to another, physically inserted into a separate device. There is no shared infrastructure. There is no remote access. The two memex devices are independent; the connection between them, when it exists, is an artifact passed by hand.
In 1945, this was the only architecture available for sharing indexed material. A document could be faxed electrically, as Bush notes in Section 2. But transmitting a navigation structure — a named sequence of associations across a personal library — would have required a network that did not exist and a protocol that had not been conceived. The physical handoff is not a failure of imagination. It is the right answer to what the technology could do.
”Must we always transform to mechanical movements”
Section 8 is where Bush says what kind of document the essay is.
Having described the memex’s microfilm, its levers, its translucent screens, its physically-transferred trails, Bush pauses in the final section and asks whether the mechanical intermediaries are necessary at all: “Must we always transform to mechanical movements in order to proceed from one electrical phenomenon to another?” [1] The example he reaches for is the human nervous system: “when the eye sees, all the consequent information is transmitted to the brain by means of electrical vibrations in the channel of the optic nerve.” [1] The body transmits information electrically without mechanical conversion. “The impulses which flow in the arm nerves of a typist convey to her fingers the translated information which reaches her eye or ear.” [1] Is there a way to do the same in a machine? “Is it not possible that some day the path may be established more directly?” [1]
He pulls back. The possibility “hardly warrants prediction without losing touch with reality and immediateness.” [1] Bush has been calibrated throughout about what is available versus what is speculative; here he applies the same standard to his own conjecture.
This passage establishes that every mechanical constraint in the memex — the levers, the microfilm, the physical trail photographs — is a contingent answer to the problem, not the problem itself. The problem is selection, fast associative access, and the ability to share navigation through a large record. The memex solves those problems with the best available mechanisms. Bush is clear that better mechanisms would solve the same problems more directly if the electrical-to-electrical path could be found.
The essay has been read as prophecy because the problems it identified were real and the solutions that eventually arrived — hypertext, networked file systems, the web — do in fact address them. But the solutions arrived with none of the mechanisms Bush described. Hypertext is not microfilm. A search index is not a lever. Sharing a URL is not handing someone a photograph of your navigation history. The essay was right about the problem and right that the mechanical constraints were not fundamental to it. It did not claim to know what the solution would look like.
A careful engineer in 1945, working from what 1945 physics offered, described the best machine 1945 physics could support. In the final section he acknowledged that better machines would come when better physics allowed it. Reading it as prophecy is a way of not reading it.