As We May Think — Vannevar Bush and the Memex

In July 1945, The Atlantic Monthly published an essay by Vannevar Bush, who at that moment was about as powerful as a scientist can be. As director of the U.S. Office of Scientific Research and Development, he had coordinated the country's wartime science, including the work that led to the atomic bomb. The essay, titled As We May Think, asked what all that scientific energy should turn to once the war was over.

His answer was not a weapon or a cure. It was a way of thinking — or rather, a machine to help us do it. The problem Bush identified was one we'd recognize instantly today, decades before it became unavoidable: there was already too much to know, the record of human knowledge was growing faster than anyone could navigate it, and our tools for finding our way through it were hopeless.

The trouble with indexes#

Bush put his finger on something precise. The problem wasn't storage; we were already good at storing records. The problem was retrieval — and specifically, the artificiality of how we organized knowledge for retrieval.

Libraries and filing systems work by classification: a thing goes in one place, under one heading, found by tracing a rigid hierarchy of categories. But, Bush observed, that is not how the mind works at all.

The human mind… operates by association. With one item in its grasp, it snaps instantly to the next that is suggested by the association of thoughts, in accordance with some intricate web of trails carried by the cells of the brain.

We don't think in a hierarchy of folders. We think by jumping from one thing to a related thing, building idiosyncratic webs of connection. Bush's radical proposal was that our information machines should match that — that we should be able to link any record to any other, directly, by association, and follow those links the way the mind follows its own.

The memex#

So he imagined a device, and gave it a name: the memex, a contraction of "memory extender." Picture a desk. Inside it, on microfilm, is the user's entire library — books, papers, letters, notes, photographs. On its slanted glass surfaces, screens display pages on demand. There are controls for browsing, levers for paging quickly or slowly, a way to photograph new material straight onto the film.

That much is just a very good filing cabinet. The heart of the memex is what Bush called associative trails. The user can take any two items and tie them together — make a link — so that touching one immediately brings up the other. More than that: they can build a whole trail of linked items, a named path threading through dozens of documents, representing a line of thought or research. And they can save those trails, recall them years later, and — this is the part that still feels ahead of us — share them. A scholar could hand a colleague not just a conclusion but the entire path of association by which they reached it.

Bush imagined "a new profession of trail blazers" who would do nothing but build valuable trails through the public record for others to follow. It's hard not to read that as a description of something we've half-built and half-botched many times since.

The ideas that came true, and didn't#

Bush had no computer to build the memex with — it was a fantasy of microfilm and optics and mechanical levers, and it could never have worked as described. But the ideas were so clear and so right that they became a kind of program for the people who came after.

Two of them in particular carried the torch and said so. Douglas Engelbart read As We May Think as a young Navy radar technician in the Philippines, and later named it as a direct inspiration for his life's work on augmenting human intellect — the associative trails of the memex are recognizably the ancestor of the links in his NLS. Ted Nelson, who coined the word hypertext two decades later, was working in the same lineage; his trails became links, his memex became Xanadu.

And then, in 1989–91, Tim Berners-Lee built the World Wide Web, which made associative linking the everyday fabric of the world. The link Bush dreamed of is now the single most-used idea in the history of information.

But notice what got lost in transmission. The Web's links are one-way — a page can point at another, but the target doesn't know, and the connection isn't a shared, persistent, navigable trail in Bush's sense. We can link, but we mostly can't build and save and share the trails of thought that Bush thought were the whole point. We got the atom of his idea and largely dropped the molecule. (Ted Nelson has spent decades, with some bitterness, pointing this out.)

Why it still reads like a prophecy#

Read As We May Think today and the uncanny thing is not how dated it is but how current. The anxiety — drowning in records, unable to find or connect what matters — is the defining condition of the information age, and Bush named it in 1945. The proposed cure — tools shaped to the associative way the mind actually works, that let us build and share trails through knowledge — is still, in important ways, unbuilt.

That's the pattern you meet again and again in the tools-for-thought tradition, and Bush is where it starts: a vision stated early and clearly, partially realized, and still pulling us forward. He ended the essay worrying that humanity might apply its growing power foolishly. The memex was his bet on the alternative — that we might instead build ourselves better minds.