Vannevar Bush

Vannevar Bush wrote “[[bush-as-we-may-think-1945|As We May Think]]” in The Atlantic Monthly in July 1945, in the closing weeks of World War II [1]. He was, at that moment, the head of the U.S. Office of Scientific Research and Development — the wartime science administrator who had directed the [[manhattan|Manhattan]] Project, the proximity fuze, radar, and the rest of the U.S. wartime science apparatus. The essay was his answer to the question of what scientists should do next, now that the war was nearly over.

His answer was that the next great scientific project should be aimed at the problem of consultation — the gap between how much humans had managed to record and how little of it any one human could actually consult and use. He named the problem with unusual precision. He proposed a solution. And he predicted, with eerie accuracy, what the device that solved it would look like: a desk that stored all your books, that built associative trails between items, that let you and your colleagues share trails [[daoism|the way]] you might share annotations.

The device he described did not exist in 1945. The microfilm-based version he sketched as the most plausible engineering approach turned out not to be the right substrate; the digital substrate that did the job came thirty years later. But the framing — the problem of consultation, solved through associative trails maintained by a personal device — has shaped every layer of computing that has tried to address knowledge work since.

What “As We May Think” said

The 1945 essay is short — about 9,000 words — and worth reading in full [1]. Its key moves:

• Names the problem. The rate at which human knowledge is being recorded vastly outstrips the rate at which any human can consult it. Indexing systems based on hierarchy and category — the standard library science of 1945 — fight against how the human mind actually retrieves information.
• Names the alternative. The mind retrieves by association, not by hierarchy. The right tool for human knowledge work would store information [[daoism|the way]] the mind uses it, with associative trails between items rather than fixed taxonomies.
• Proposes a device. The Memex is described as a desk-sized machine containing all of a person’s books, periodicals, and notes on microfilm. The user can pull up any document, project it on a screen, and — critically — link it to other documents by building associative trails. The trails are saved, named, and can be shared with colleagues.
• Predicts a profession. Bush coins the term trail blazers for the new class of people whose job will be to build and maintain associative trails through the world’s literature on behalf of others. He treats this as obviously valuable, and predicts it will become a recognized profession.
• Identifies what is missing. The essay does not solve the maintenance problem. Bush imagines individuals building their own trails and sharing them, but does not have an answer for the upkeep cost of large trail-systems at population scale.

The essay’s tone is unusual for a science-administration document. It is genuinely speculative — Bush is willing to extrapolate from technical possibility to social consequence — but it is also concrete. The Memex’s mechanical operation is described in enough detail that engineers tried for the next thirty years to build something matching the description.

What the lineage built

Every figure in the [[Memex Lineage]] inherits something from Bush, and each leaves something Bush named unaddressed [1, 2]:

• [[doug-engelbart|Engelbart]] (1962) — extends Bush from passive retrieval to interactive augmentation. Adds the tight human-machine loop. Builds the first working hypertext system. Does not solve the maintenance problem at scale.
• [[ted-nelson|Nelson]] (1965 onward) — extends Bush with a fully-elaborated document substrate (transclusion, bidirectional links, micropayments, version preservation). Coins the vocabulary the rest of the lineage uses. Does not ship.
• [[tim-berners-lee|Berners-Lee]] (1989–91) — ships a deliberately lossy form of the vision. The web has hypertext but loses the associative-trail-as-first-class-object idea. Maintenance is offloaded entirely to the user, with no native support.
• Wikipedia (2001 onward) — proves that human-maintained associative knowledge bases at scale are possible, but only with thousands of volunteers and strict editorial governance. Solves Bush’s maintenance problem at the cost of editorial flattening.
• Roam, Obsidian, Logseq (2013 onward) — give individuals Bush-style associative trails on their own documents. Maintenance is back on the user. Most personal knowledge bases die at the maintenance step.
• [[karpathy-llm-wiki|LLM Wiki]] (2024 onward) — the LLM is cheap enough to be the trail blazer Bush predicted. Maintenance happens at near-zero marginal cost. The vision Bush sketched in 1945 finally has a working implementation, eighty years late.

This is what the [[Memex Lineage]] playbook documents. Bush is the figure at the head of the line.

What he got right and what he missed

Bush got an unusual amount of the future right. Eighty years later, the Memex’s design — personal knowledge appliance, associative-trail-based, share-able to colleagues, with a trail-blazer profession around it — is recognizably the system humanity is now actually building.

He missed:

• The substrate. Microfilm was the wrong material; he could not have known. Digital storage on commodity hardware turned out to be the right one, available thirty years later.
• The maintenance cost. Bush imagined trails being built and maintained by individuals. He did not anticipate that the maintenance burden, at any serious scale, would exceed what individuals or even small teams could sustain. The maintenance problem ate every Memex-descendant system between 1945 and 2024 — including, in different ways, Xanadu’s failure to ship and the personal-knowledge-base graveyard of the 2010s.
• The economic substrate. Bush wrote in an era of public-[[mission-district-sf|mission]] research funding. He did not anticipate that the natural deployment vehicle for his vision would be commercial platforms whose business models would be fundamentally misaligned with the Memex’s user-serving framing. Nelson’s micropayment idea was an attempt to address this; the web shipped without it; the resulting attention economy is the visible cost.
• The institutional dimension. Bush imagined the Memex as personal infrastructure. Wikipedia revealed that the Memex’s most successful real-world descendant would be institutional — a public-benefit nonprofit with editorial governance — not personal at all. The personal-knowledge-base market took until the 2020s to find a path past the maintenance wall, and the path it found (the LLM as trail blazer) was not one Bush imagined.

Why this matters for 0mn1.one

[[0mn1one|The platform]] is, in its wiki form, an attempt to ship Bush’s vision in its faithful form for the first time at the personal-and-small-organization scale. Specifically:

• Associative trails as first-class objects. The wiki has a trails surface (src/content/trails/) that explicitly treats Bush’s “associative trail” as a primary unit of content, not a derived structure. Each trail is a curated walk through wiki entries, with narrative connecting each step to the next. Bush’s predicted trail-blazer work is performed on this platform.
• Maintenance solved by inference. The vault’s maintenance cost has dropped from human-only to inference-shared, which is the move Karpathy named (in the [[Memex Lineage|Memex Lineage]] playbook’s modern entry) and which [[0mn1one|the platform]] exists in part to demonstrate.
• Author [[eating-the-landscape|identity]] preserved. Bush did not address authorship persistence; [[0mn1one|the platform]] does, for its human contributors.
• [[mission-district-sf|The mission]] framing. Bush wrote in a 1945 spirit of public-benefit science. 0mn1.one’s framing — [[values-aligned-commerce|values-aligned commerce]], revenue serving infrastructure for life — is in conscious continuity with that spirit. The platform is one attempt to put the Memex back in the kind of moral framing Bush originally proposed it within.

He died in 1974, twenty years before the web shipped. The trail-blazer profession he predicted is now real. [[0mn1one|The platform]] is one place where it is being practiced.

Cultural and historical

Vannevar Bush was born in Everett, [[barre-ma|Massachusetts]] in 1890 [1]. PhD in electrical engineering jointly from MIT and Harvard in 1916. Co-founded Raytheon in 1922. Served as MIT vice president and dean of engineering in the 1930s. As director of the Office of Scientific Research and Development (1941–1947), he ran the largest concentrated science-administration effort in U.S. history; the [[manhattan|Manhattan]] Project reported through his office. After the war, he wrote the foundational report Science, the Endless Frontier (1945), which argued for sustained federal investment in basic research and led directly to the founding of the National Science Foundation.

He was, by every standard available, one of the most consequential American scientific figures of the twentieth century. “[[bush-as-we-may-think-1945|As We May Think]]” is somewhat overshadowed in his official biography by his wartime administrative work, but it is the part of his legacy that the most people now read.

He died in 1974 in Belmont, [[barre-ma|Massachusetts]], age 84.

Lenses still to grow

• “[[bush-as-we-may-think-1945|As We May Think]]” in primary form — direct quotation from the 1945 Atlantic essay, ingested in full. Available freely online.
• The wartime science-administration role in detail — the [[manhattan|Manhattan]] Project administration, the OSRD, the relationship with FDR
• Science, the Endless Frontier as a separate entry — the report that led to the NSF
• The Differential Analyzer — Bush’s 1930s analog computer, an important early computing milestone
• The relationship to the Roosevelt administration and the post-war science establishment — historical context for who Bush was politically and what he was trying to achieve
• Comparison with H. G. Wells’s World Brain (1937) — the speculative-knowledge-organization tradition Bush was writing within

See also

Auto-generated from this entry’s typed relations: frontmatter, grouped by relation type so the editorial signal isn’t flattened.

• Extends: [[memex-lineage]]
• Enables: [[doug-engelbart]] · [[ted-nelson]] · [[tim-berners-lee]]
• Member of: [[person]]

Sources

1. Bush, V. “[[bush-as-we-may-think-1945|As We May Think]].” The Atlantic Monthly, July 1945. The originating text. Ingested in source-note form at _knowledge/sources/bush-as-we-may-think-1945.md.

2. Wikipedia: Vannevar Bush. Biographical timeline, the OSRD years, Science, the Endless Frontier, “As We May Think” historical reception. https://en.wikipedia.org/wiki/Vannevar_Bush

The 1945 essay is short and freely available; future revisions of this entry should quote it directly rather than paraphrase. The Bush biography by G. Pascal Zachary (Endless Frontier, 1997) is the standard secondary source and worth ingesting in a future pass.

Page filed 2026-05-03. Completes the modern [[memex-lineage|Memex Lineage]] core (Bush, Engelbart, Nelson, Berners-Lee all now have wiki entries).