HTML is markup, not a programming language — use it like one
Treat HTML as structure for navigation, SEO, accessibility, and render performance, and replace div soup with semantic elements the browser and crawlers can understand.
This article is published in English.
Sources: HTML Is Not What You Think
The Hook
HTML is still treated as the easy starter skill—tags, forms, a few layouts—while real pages have grown enormous. Typical documents now ship thousands of DOM nodes, yet many sites are still authored with habits from an earlier, smaller web. That gap shows up in four places: how people navigate, how search engines interpret structure, how efficiently browsers paint, and how crawlers and AI systems extract meaning from markup.
HTML Is Not a Programming Language
Markup structures content; programming languages express behavior. The contrast is blunt:
- Logic: Languages branch with conditionals. HTML does not decide.
- State: Languages keep variables. HTML has none.
- Output: Languages compute values. HTML describes what should appear.
Calling HTML a programming language confuses categories. It is a markup language for structuring and describing content the browser will present.
The Internet Has a DIV Problem
Nested anonymous boxes remain common:
<div><div><div>
<div>Text</div>
</div></div></div>
A semantic alternative names roles instead of stacking wrappers:
<article>
<header><h1>Title</h1></header>
<main><p>Text</p></main>
</article>
Heavy div nesting hurts accessibility trees, slows maintainability, and gives the browser more meaningless structure to walk. Prefer elements that declare purpose.
HTML Has an Identity Crisis
HTML began as a document format for publishing. It now underpins software-like experiences: dashboards, PWAs, games. A short timeline:
- 1991: Academic documents
- 1996: Table-driven layout
- 2004: AJAX-era web apps
- 2024: PWAs, dashboards, games
The language was never designed as a full application runtime, yet the modern DOM exposes a large API surface that application frameworks sit on top of.
What the Browser Really Sees
Rendering is a pipeline, not a single pass:
- DOM tree — HTML parses into elements and structure
- CSSOM — styles attach
- Render tree — DOM and CSS combine
- Layout — geometry is computed
- Paint — pixels are drawn
Poor markup can hurt performance before CSS or JavaScript become the bottleneck. Deep, anonymous trees cost layout and accessibility work even when styles look fine.
Semantic HTML Saves the Web
Semantic tags give browsers, search engines, and assistive tech clearer structure:
<article>marks independent content for readers and indexers<nav>identifies navigation landmarks<main>highlights the primary content region crawlers prioritize<figure>pairs media with captions in a way parsers can recognize
The Hidden Cost of Bad HTML
- Performance: Oversized DOMs slow rendering on modest devices
- Accessibility: Weak structure frustrates screen readers and keyboard users
- SEO: Search systems still lean on HTML structure before most client JavaScript runs
HTML5 Changed Everything
Before HTML5, rich media often depended on Flash or Java applets. Native elements and APIs replaced much of that stack:
- Multimedia:
<video>and<audio>displaced plugin players - Graphics:
<canvas>and<svg>covered drawing without applets - Storage:
localStoragereduced cookie-only client persistence
HTML today is a Living Standard stewarded by WHATWG and still evolves in place rather than freezing as a numbered “HTML6” snapshot.
The Future of HTML
Frameworks change authoring workflows; browsers still consume HTML at the end of the pipe. Trends that keep markup central:
- Native web components for reusable pieces without a framework mandate
- Accessibility-first design as a baseline expectation
- AI-readable structure where semantics improve extraction and SEO
- Server-rendered HTML that stays fast and usable before hydration
Structure, accessibility, and semantic clarity will matter more, not less, as machines join humans as primary consumers of pages.
Consequences — The Challenge
HTML has outlasted every fashion of the web—from brochure sites to AI-assisted apps—as the quiet substrate underneath. Learning tags is no longer the hard part. The hard part is living with the long-term cost of using them carelessly: inaccessible div soups, sluggish DOMs, and markup that neither people nor crawlers can trust.
Primary references include the WHATWG HTML Standard at http://html.spec.whatwg.org/, MDN at http://developer.mozilla.org/, practical antipatterns at https://htmhell.dev/, and Web Almanac measurements at http://almanac.httparchive.org/.