HTML

HTML, the HyperText Markup Language, is the language web pages are written in. A document is text marked up with tags that say what each part is: a heading, a paragraph, a list, a link, an image. The browser reads the tags and builds a tree of elements, which scripts can then change; see Document Object Model.

Tim Berners-Lee designed it at CERN around 1990 as part of the World Wide Web. He borrowed the angle-bracket tag style from SGML, a document standard already used at CERN, and added the one feature that mattered most, the link. A 1991 description listed only about eighteen elements. Images came in 1993, when Marc Andreessen added an image tag to the Mosaic browser without waiting for agreement, which set a pattern.

Through the 1990s, browser makers added tags to win users, from blinking text to frames, and pages were written for one browser or the other. Standards bodies tried to catch up. The W3C published HTML 4 in 1997, then pushed toward XHTML, a strict version in which a single unclosed tag could stop a page from displaying at all.

Authors didn't follow. Most real pages were tag soup, full of unclosed and misnested tags, and browsers had to display them anyway; see Postel's Law. Each browser guessed differently, which made pages behave inconsistently. In 2004, people from Apple, Mozilla, and Opera formed a separate group to extend HTML instead. Their HTML5 did something no earlier version had: it specified exactly how to parse broken markup, so every browser now builds the same tree from the same mess. It also added video, audio, a drawing surface, and form controls that had needed plugins. Since 2019 a single continuously updated document has been the standard, with no further version numbers, though "HTML5" survives as a loose label for the modern web platform.

Old pages still work. A browser can still render a page from 1995, and the doctype line at the top of a file decides whether it uses old or modern layout rules.

Elements carry meaning as well as appearance. A nav element marks navigation, and a button announces itself as a button to a Screen Reader, which a styled div does not; see Semantic HTML. Appearance belongs to stylesheets. Pages built from meaningful markup work before scripts load and when they fail; see Progressive Enhancement.

Plain text formats such as Markdown exist mainly so writers don't have to type tags, and they're converted to HTML in the end. Developers can now define their own elements as well.