In contrast to SGML or XML, HTML is a specific markup language that contains a fixed set of elements and attributes. HTML has a limited repertoire of structural tags like headings, lists, and links, some tags for encoding formatting information like text attributes and layout, and very few tags for encoding types of information content. This design decision by Tim Berners-Lee, the inventor of the Web, was the right choice because it made HTML easy to understand and implement, leading to its rapid adoption. The idea of naming things in plain text with the content "between the tags" is highly intuitive, so even children can create Web pages.

However, as the Web evolved, HTML's initially simple tag set encouraged "hand-crafting" of Web pages, misuse of tags to achieve formatting effects, and proprietary additions to HTML by browser and application developers who needed richer markup to support additional functionality. Furthermore, while HTML can be described using a DTD, the vast majority of HTML on the Web is invalid. Taken together, HTML's fundamental limitations and typical usage without validation make it difficult for search engines and automated processes to exploit Web information because of the lack of reliable semantic encoding. XML can solve these problems with HTML and give the Web a much stronger capability for electronic commerce. XML makes it possible to encode information with meaningful structure and semantics in a very accessible notation that is both human-readable and readily processable by computers. While XML 1.0 adds no new modeling capabilities beyond those that have been available in SGML for over a decade, the simpler XML syntax makes it much easier for non-specialists to participate in the design of new markup languages..