The World Wide Web and the Decision Not to Patent It

A documentation problem inside a physics lab

CERN, the European Organization for Nuclear Research based near Geneva, Switzerland, employed thousands of researchers who moved in and out of long-running experiments, each accumulating notes, results, and internal documents in formats that rarely spoke to one another. On March 12, 1989, Tim Berners-Lee, an English computer scientist working at the laboratory, proposed what he called an information management system: a way to link one document to another regardless of what machine or program had produced it. He built the idea himself over the following months. By mid-November 1989, he had implemented the first successful exchange between an HTTP client and an HTTP server, the two ends of the connection that still moves nearly everything on the modern Web.

The scale of what one person did in that period is easy to understate. Berners-Lee designed the Uniform Resource Locator, the addressing scheme that gives a page a unique location; HyperText Markup Language, the formatting language a browser reads to display it; and the HyperText Transfer Protocol, the exchange that carries it from server to browser. He then wrote the first web browser, which he called WorldWideWeb, and the first web server to answer it. Three technologies that still define how the Web is addressed, formatted, and transmitted trace to a single author working inside a physics laboratory, not a technology company. He later founded the World Wide Web Consortium to steer the standard's ongoing development, a step that moved control from one person toward a shared body rather than toward a single owner.

The project was not conceived, at first, as something the public would ever touch. It solved a problem local to CERN. Berners-Lee described his goal as a universal linked information system, and the Web that resulted was invented in 1989 and stayed largely inside the laboratory's own use for several years before it opened to the public in 1993. Its purpose in the interim was practical: keep a growing, high-turnover research organization from losing track of its own work.

What turned a lab tool into a global standard was not the code itself. It was a decision CERN's administration made about who would be allowed to own it.

The document CERN did not have to sign

On April 30, 1993, CERN's Director of Research, Walter Hoogland, and its Director of Administration, Helmut Weber, signed a short document that settled the question. It stated plainly that CERN relinquishes all intellectual property rights to the Web's code, in both source and binary form, and that permission is granted for anyone to use, duplicate, modify, and redistribute it. No royalty. No license fee. No requirement to ask first.

The alternative was available, and at the time, ordinary. Software protocols could be patented; software could be licensed; a laboratory that had funded years of a researcher's salary to build something this valuable had every institutional basis to charge for it, the way universities and corporations routinely did, and still do, with commercially useful inventions. CERN chose not to. The declaration put the Web's core code, not merely a demonstration of it, into the public domain: free for a competitor, a university, or a lone programmer to build on without CERN's permission and without CERN taking a share.

The consequence showed up almost immediately in what got built on top of it. A team at the National Center for Supercomputing Applications, at the University of Illinois, began work on its own browser in December 1992, months before CERN's formal declaration, already relying on the openness Berners-Lee had been extending informally. Released as NCSA Mosaic in January 1993, it was the first browser to display images inline with the text of a page rather than in a separate window, a small design choice that made the Web look, for the first time, like something a non-specialist would want to use.

What CERN's decision meant, in practical terms, is that nobody who wanted to build a browser, a server, or eventually a business on the Web needed to negotiate with anyone first. The standard was there to be used, not licensed. That single fact is what the next fifteen years of Web history sit on top of.

What refusing the toll bought

Every dominant information medium before the Web had an owner who controlled access to it, and that owner captured a share of whatever value moved through it. Telegraph and cable networks were run by companies that charged by the word. Broadcast spectrum was licensed by governments to operators who sold advertising against it. Even the commercial dial-up services that predated the open Web worked the same way: a company owned the network, and anyone who wanted to reach its subscribers paid that company first.

The Web broke the pattern at the level of the protocol itself. Because HTTP, HTML, and the URL were unpatented and unlicensed, nobody had to ask CERN, or pay CERN, to build a server, write a page, or launch a browser. A hobbyist and a well-funded company faced an identical entry cost for the standard itself: zero. What each still had to build, and what still cost money, was everything layered on top of it: the software, the content, the audience, the business model. But the floor everyone stood on was free and common to all of them.

The scale of what grew from that floor is hard to state as a single number, and doing so would overstate the precision of any one figure. What can be said with confidence is that the companies whose businesses now sit on top of the open Web, browser makers, search engines, retailers, social platforms, and the infrastructure underneath all of them, are collectively worth an amount measured in the trillions of dollars by any reasonable reckoning of public market value. None of that value flowed back to CERN as a royalty, because CERN had left no royalty to collect.

This is one of the clearer cases in modern economic history of value created by a deliberate refusal to capture it. CERN did not fail to monetize the Web. It chose, in a signed and dated document, not to.

Invented in Geneva, capitalized in California

The Web's origin and its commercial capture happened in different places, under different flags. CERN is an intergovernmental organization funded by European member states, headquartered in Switzerland, and its research mission has nothing to do with building consumer software companies. The invention was European public-sector work, made openly available to the world at large. What happened next was almost entirely American, and almost entirely private-sector.

Mosaic, the browser that made the Web usable for people outside research institutions, had been built at a US federally funded supercomputing center, but its commercial future was decided in Silicon Valley. Mosaic's co-creator, Marc Andreessen, left the university project and, with the Silicon Valley entrepreneur Jim Clark, founded Mosaic Communications Corporation, soon renamed Netscape Communications. Netscape's browser, Navigator, overtook the original Mosaic in market share by late 1994, and Netscape spent the next several years as the company most associated with the Web in the public mind.

The debate: openness that liberates, and value that concentrates

The historical argument over what CERN's decision proves splits fairly evenly, and both halves are true at once. The case for liberation is straightforward: an unpatented, unlicensed Web could spread everywhere at once, adopted by browsers, servers, and eventually a majority of the world's population, without a single company or government able to charge a toll or restrict who joined. Set that against the earlier commercial online services, each built on a proprietary, closed network, which stayed fragmented into separate walled gardens precisely because each one was owned. By the mid-1990s the Web had become the dominant information systems platform of its era and the primary tool billions of people now use to reach the wider internet, a scale reached within roughly a decade of a decision that produced CERN no direct revenue at all.

The case for concentration is just as real, and it runs through the same history. Openness at the protocol layer did not produce openness in who captured the economic value that layer enabled. The absence of a toll on HTTP, HTML, and the URL did not stop toll booths from appearing one level up, in browsers, in search, in the platforms that eventually organized how nearly everyone finds anything on the Web. Those toll booths were overwhelmingly American, built with American venture capital, and the wealth they generated stayed concentrated among a small number of firms clustered around Silicon Valley rather than distributed globally or returned to the European institution that had made the whole layer possible.

What cannot be settled, because it never happened, is what the Web would look like today had CERN patented HTTP and licensed it the way a technology company ordinarily would. It is a reasonable argument, made by many historians of the period, that a licensed and enclosed Web would have fragmented into competing proprietary systems the way the earlier online services had fragmented, each optimized for its own paying customers rather than for universal reach. It is not a provable claim. Assigning the Web's global, non-fragmented reach to this one decision, rather than to it combined with Mosaic's usability and the Internet's own prior openness, belongs in the emerging or contested tier rather than the established one, even though it is the most common reading among people who have studied the period closely.

The pattern that repeats now

The specific mechanism that made the Web universal, an open, unowned protocol that any browser or server could speak, is a decision belonging to 1993, not to the present. But the underlying question it raises, who controls the layer that decides what gets found, is not a historical curiosity. It is the live question behind how businesses are discovered today; only the layer in question has moved.

A generation of businesses built pages assuming an open, crawlable Web would keep working the way it had since 1993: any browser could reach any page, and a search index built on top of that openness would surface it. What has changed is not the openness of HTTP itself, which still carries the same unlicensed protocol Berners-Lee designed. What has changed is that a growing share of the reading now happens once, inside an AI system, which then decides on the visitor's behalf which businesses to name in its answer and which to leave out. The protocol stayed open. The reading did not become any less consequential for being automated.

This is where the Web's founding choice and the present moment share a genuine connection, rather than a manufactured one. CERN's decision determined who could build on the Web at all, at the level of the standard. It said nothing, and could say nothing, about who would get named once billions of pages existed and something had to choose among them. That choosing now happens inside AI answer engines, and a business's position in that choice depends on whether its own pages are written in a way those systems can read, verify, and act on, not on whether the underlying Web remains free to build on. The openness Berners-Lee designed for is still there. What sits on top of it has changed completely.