Hypertext organizes by associative links, not a single hierarchy

What it is

A hypertext structure connects objects with “related to this” links into a graph. It does not require a shared parent tree first. Encyclopedia articles, citations, and “frequently bought with” are this organization: the next hop is decided by association, not by class membership. People can enter the graph at any node, not only at the root. It answers “what else is tied to this.” It does not answer “which class it belongs to” or “which step is mandatory.”

Pure hypertext without a backbone is still a structure type. Place is then maintained by the current node and the anchor text of outbound edges, not by a path.

Why it happens

Associative edges encode co-occurrence, cause, substitution, citation—relations that cross category boundaries. If the task already walks a relation (definition to example, product to accessory), a hierarchy forces a climb back to the parent and down again, and scent breaks on the return. Hypertext makes that edge a visible next step; the decision becomes “follow this relation or not,” not “guess the parent.”

A graph has no unique “up.” The deeper the walk, the harder it is to say where one is in the whole graph. Anchor text is the remaining scent; once it reads “click here,” the edge remains and the structural information is gone. Cycles and multiple entrances are normal for hypertext, not errors—unless the system still pretends there is one correct path.

Studying it

Evaluate hypertext with relational tasks, not classification tasks.

• Paradigms: start from one article and reach a relational target (accessory, counterexample, cited work); compare a pure link graph, a pure tree, and a tree plus links. Record lostness (repeat visits, inability to sketch the path taken).
• Independent variables: quality of anchor text, out-degree, whether a trail of visited nodes is offered instead of hierarchical breadcrumbs, presence of cycles.
• Dependent variables: success on relational tasks, failure of classification tasks on a pure graph, subjective sense of place, revisits to the same node.
• Methodological note: measuring hypertext with a find-the-class task predetermines “graphs lose to trees.” The task type must match the relation the structure type claims to serve. Report lostness with behavioural metrics, not only “did it feel messy.”

Where it stops holding

When objects have no stable, nameable relation, bolting on “related” becomes a recommendation black box, not hypertext structure. When law or a linear process demands a unique order, hypertext’s many entrances are a hole. On a tiny collection a pure graph collapses into a few mutual links, cheaper as a shallow tree. Search can jump into a node by keyword; that is not an associative edge and cannot stand in for the relations hypertext claims to organize.

Applying it

• Build edges only for relations that exist and can be written as short anchor text: accessory, counterexample, prior concept, another format of the same policy.
• Anchor text should preview the relation type. Do not use “more” or “click here.”
• Offer a local map of the current node or a trail of the walk. Do not fake a graph as a tree with parent–child breadcrumbs.
• Verify by starting from an article unrelated by class and reaching the target by association only. If the path requires going home and down the tree, hypertext is not yet a structure—only decorative links on a tree.

Related

• Within the group: G1.09.1 Hierarchy organizes content by parent–child links and is the usual default · G1.09.2 A matrix locates the same content by several attributes at once · G1.09.4 Sequential structure forces a fixed order, as in tutorials and wizards · G1.09.5 Most systems keep hierarchy as the backbone and add hypertext laterals · G1.09.6 Structure type and organization scheme are independent dimensions · G1.09.7 A matrix demands complete metadata; missing fields block location
• Adjacent: G2.03 Network and associative navigation · G1.01 Definition and scope of information architecture
• Search terms: hypertext structure · associative navigation · nonstrict hierarchy