Divitis

The defining characteristic of Divitis is a document object model (DOM) composed almost entirely of <div> and <span> elements. These elements are inherently semantically neutral—they convey no programmatic meaning about the content they contain. An AI parser encountering a page built with Divitis cannot distinguish a navigation landmark from an article body or a footer, as every structural block is an identical, meaningless container. This forces extraction algorithms to rely on fragile, class-name-based heuristics rather than the robust, standardized accessibility tree.

To compensate for the lack of native semantics, developers with Divitis rely entirely on CSS class names (e.g., class="header", class="article") to convey meaning. This approach is fundamentally brittle because:

• Class names are not standardized and vary wildly between projects and teams.
• AI crawlers and search engines do not parse CSS class names as reliable semantic signals.
• The meaning is only apparent to human developers reading the source code, not to machines. This creates a semantic gap between developer intent and machine interpretation.

The browser constructs an accessibility tree from the DOM to expose content to assistive technologies and programmatic agents. Divitis starves this tree of meaningful roles. Without elements like <nav>, <main>, <article>, or <aside>, the accessibility tree becomes a monolithic block of generic containers. Screen readers lose the ability to offer landmark navigation, and AI agents cannot efficiently skip to relevant content regions. Remediation requires manually patching every <div> with ARIA roles (e.g., role="navigation"), which duplicates the function of native HTML elements.

Divitis often correlates with a broken or non-existent heading hierarchy. When all content is wrapped in <div> elements, developers may use visual styling (bold, larger font-size) to simulate headings rather than using <h1> through <h6> elements. This practice is catastrophic for AI content extraction because:

• Heading elements are the primary mechanism for AI models to understand document outline and content importance.
• Without a proper hierarchy, the document appears as an undifferentiated wall of text.
• Retrieval-augmented generation (RAG) systems rely on heading structure for accurate chunking and retrieval.

The cure for Divitis is achieving First Normal Form (1NF) for the DOM, a concept borrowed from database normalization. This involves replacing generic <div> elements with the correct semantic HTML5 equivalents:

• <div class="header"> → <header>
• <div class="nav"> → <nav>
• <div class="content"> → <main>
• <div class="post"> → <article>
• <div class="sidebar"> → <aside>
• <div class="footer"> → <footer> This transformation provides programmatic determinism, ensuring AI agents and accessibility tools can reliably interpret document structure.

Divitis directly undermines Generative Engine Optimization (GEO) efforts. AI models constructing answers from retrieved content prioritize documents with clear, machine-parseable structure. A page suffering from Divitis presents a low signal-to-noise ratio to AI crawlers, as every content block appears structurally identical. This reduces the likelihood of:

• Accurate entity extraction for knowledge graph injection.
• Successful semantic chunking for RAG system retrieval.
• High-confidence citation in AI-generated overviews. The absence of semantic landmarks makes it difficult for answer engines to isolate and cite specific facts from the page.

The direct antidote to Divitis. Semantic HTML uses elements according to their intrinsic, programmatically determined meaning rather than for visual presentation alone. Elements like <article>, <nav>, <aside>, and <section> explicitly communicate content roles to AI parsers, enabling accurate extraction of document structure. Unlike <div>, which conveys zero semantic value, semantic elements populate the accessibility tree with meaningful roles, states, and properties that both assistive technologies and generative engines rely on for content interpretation.

A logical, nested structure of <h1> through <h6> elements that defines the document outline. Proper heading hierarchy communicates the relative importance and parent-child relationships of content sections to search engine parsers and accessibility bots. Divitis often manifests as using styled <div> or <span> elements instead of actual heading tags, which renders the document outline invisible to AI crawlers. A well-structured heading tree is one of the strongest signals for information architecture that generative engines use to understand content scope.

A set of predefined roles—banner, navigation, main, complementary, contentinfo—added to HTML elements to programmatically identify distinct page regions. When developers suffer from Divitis, they wrap everything in generic <div> containers, forcing them to manually apply ARIA roles to restore lost semantics. Native semantic elements like <header>, <nav>, and <footer> have implicit ARIA landmark roles, eliminating the need for redundant role attributes and ensuring AI agents can efficiently navigate document structure without ambiguity.

A parallel structure generated by the browser from the DOM that exposes semantic information, properties, and relationships exclusively to assistive technologies and programmatic agents. Divitis produces a flat, undifferentiated accessibility tree where every node is a generic container with no meaningful role. This forces screen readers and AI parsers to guess at content relationships. Semantic HTML elements automatically populate the accessibility tree with correct roles—<main> becomes a landmark, <button> becomes an interactive control—creating a rich, navigable structure for machine interpretation.

The principle that the meaning, state, and value of a user interface component can be reliably interpreted by software, including AI agents, through standardized, machine-readable properties. Divitis violates this principle by replacing semantically deterministic elements with neutral containers. A <div onclick='submit()'> is not programmatically a button—it lacks the implicit role, keyboard interactivity, and accessible name computation that a <button> provides. For AI-driven search engines, programmatic determinism ensures content structure is unambiguous and extractable.

The formal groupings defined by the HTML specification—Flow, Phrasing, Sectioning, Heading, Embedded, Interactive, Metadata—that dictate where an element can be used and what its semantic purpose is. Divitis ignores these categories entirely, treating <div> as a universal container. Understanding content categories helps developers select the correct element: sectioning content (<article>, <aside>, <nav>, <section>) defines the document outline, while heading content defines section labels. AI parsers rely on these categorical distinctions to build accurate document models.