In plain terms
Before the Web, reading a document stored on another machine meant knowing that machine's own access system: FTP, Gopher, WAIS and Telnet each had their own addressing, incompatible with the others. Berners-Lee proposes three building blocks that hold together: a universal address (URL), a single way to request a document (HTTP), a common format in which to write it (HTML). The equivalent of a postal address valid in every country, paired with a language every post office can read. The decisive shift is not technical but legal: by placing the code in the public domain on April 30, 1993, CERN gives up all royalties, and publishing no longer requires anyone's permission. The adoption figures cited below remain orders of magnitude, dependent on measurement methods.
Discovery
| Parameter | Value |
|---|---|
| Proposal date | March 1989 (memo "Information Management: A Proposal") |
| First site online | December 20, 1990, info.cern.ch |
| Public announcement | August 6, 1991 (post on alt.hypertext) |
| Inventor | Sir Timothy John Berners-Lee (born June 8, 1955, London) |
| Place | CERN, Geneva, Switzerland |
| Server machine | NeXT Computer (Motorola 68040 CPU, 25 MHz, 16 MB RAM) |
| Technologies created | URL (addressing), HTTP/0.9 (transfer), HTML (markup) |
| License | Public domain (April 30, 1993, CERN decision) |
Technical explanation
1. HTML (HyperText Markup Language) — Berners-Lee defines a markup language derived from SGML (Standard Generalized Markup Language, ISO 8879:1986). The first version has only 18 tags (<a>, <p>, <h1>–<h6>, <ul>, <ol>, <li>, <address>, etc.). The key concept is the hypertext anchor <a href="...">: a clickable link to any document on any server.
2. HTTP (HyperText Transfer Protocol) — A request-response protocol in plain text. HTTP/0.9 supports only one method: GET. The client sends GET /path\r\n, the server returns the raw HTML document and closes the TCP connection. No headers, no status codes, no MIME types. The simplicity is deliberate: any developer can implement a client in a few hours.
3. URL (Uniform Resource Locator) — Berners-Lee unifies the addressing of network resources into a single syntax: protocol://host:port/path. Before URLs, each system (FTP, Gopher, WAIS, Telnet) had its own incompatible addressing scheme. The URL federates all these protocols under a universal convention.
4. The first browser-editor — The "WorldWideWeb" software (later renamed Nexus) on NeXT is both a browser and a WYSIWYG editor: Berners-Lee designs the Web as a read-write space, not read-only. This vision would be lost for 15 years, until wikis and Web 2.0.
Why it worked
The Web is neither the first network nor the first hypertext (Apple's HyperCard dates from 1987, Ted Nelson's Xanadu project from 1960). Its victory rests on three architectural choices:
| Architectural choice | What it implies |
|---|---|
| Decentralization | No central server: anyone can publish |
| Fault tolerance | A broken link degrades gracefully, without systemic crash |
| Simplicity | HTML learned in an hour, server deployable in a day |
CERN's April 30, 1993 decision to place the code in the public domain — no royalties, no license — is the founding act. Without it, the Web could have remained proprietary, like CompuServe or AOL.
Causal chain
Vannevar Bush imagines the Memex (1945) → Ted Nelson defines hypertext (1965) → ARPANET creates packet switching (1969) → TCP/IP unifies protocols (1983) → Berners-Lee proposes the Web (1989) → CERN releases the code (1993) → Mosaic/Netscape democratize browsing (1993–1994) → Google organizes information (1998) → Web 2.0 turns the user into a producer (2004) → mobile-first overtakes desktop (2016)
Anecdote
The red sticker Berners-Lee stuck on the case of his NeXT read: "This machine is a server. DO NOT POWER IT DOWN!!". Had anyone unplugged it, the only web server in the world would have gone offline. That NeXT is today on display at the CERN museum in Geneva.
Limits and controversies
Market-share and adoption statistics vary according to measurement methodologies (surveys, traffic analyses, publisher declarations). The figures presented are orders of magnitude drawn from specialized sources. The history of computing is subject to priority controversies not exhaustively covered here.
Sources
References verified during the August 2026 fact-checking audit: these are the pages
against which this bulletin's claims were checked.