Our understanding of complex systems — from physical cities to the digital web — depends not only on the informational content embedded within them, but also on the form through which the information is structured. These relational structures determine how places and media are organized and experienced; sensed and made sense of.

In urban contexts, spatial legibility and semiotics have been expressed through metaphors of reading and writing the city. Drawing upon the analogy of the text, hypertext presents a more suitable form for representing urban space: decentralized, nonlinear, and dynamic. Hypertext and urban theory converge among post-structuralist principles of distributed authorship and relational meaning. Spatially linked information networks afford association, provenance, and orientation — qualities which are destabilized in the contemporary algorithmic web.

The term hypertext was first introduced by computer scientist Ted Nelson in 1965 — over twenty years before the invention of the World Wide Web. Nelson used the prefix hyper- (beyond or in excess) to describe “text which is not constrained to be linear … which contains links to other texts" (Nelson). From its origin, hypertext produced a new navigable spatial logic for structuring knowledge. While today we may take hyperlinks for granted, they were instrumental to the decentralized nature of information shared on the Internet. Essential properties of the web — interconnectedness, non-linearity, distributed authorship — could not have been realized without Nelson's inventive media infrastructure.

In the 60s, urbanists began theorizing approaches to representing complex urban environments through non-hierarchical systems. In The Image of the City, Kevin Lynch wrote that people use mental maps to make sense of urban environments. He proposed a lexicon of elements — nodes, paths, edges, districts, and landmarks — to illustrate how cities are rendered legible when experienced by moving bodies (Lynch).

In A City is Not A Tree, Christopher Alexander borrowed vocabulary from the field of computer science — in which a tree is a common data structure made up of nodes and edges — to argue against the analogy of hierarchical trees in urban planning. Instead, Alexander proposed the semi-lattice, a more complex model of overlapping categories which is characteristic of natural, unplanned cities. One justification is that this bidirectional network logic is found recurrently in social structures: “the systems of friends and acquaintances form a semilattice, not a tree" (Alexander).

A number of proposals have imagined spatial visualizations of hypertext based on real-world metaphors. One notable case is Andreas Dieberger's 1998 dissertation A City Metaphor to Support Navigation in Complex Information Spaces, which puts forth the Information City as a “conceptual user interface metaphor based on the structure of a city.” (Dieberger). Picture a version of Wikipedia that's structured such that each wiki page is a room in a building, which is part of a neighborhood, which together construct a city. Dieberger based the Information City's topology on Lynch's lexical elements: container, landmark and path. The provocation bridged the gap between urban and information networks, highlighting how spatial arrangement can serve as both metaphor and method to express non-hierarchical relationships between discrete elements.

The convergence of networked thinking across disciplines stemmed from a broader intellectual shift away from unified narrative and stable representation. From literature to philosophy, postmodernist thought presented knowledge as relational and situated. The pioneering hypertext scholar George Landow applauded hypertext for instantiating Deleuze and Guattari's theory of the rhizome: a post-structuralist assemblage of multiplicity in opposition to singular authorship (Bishop). While arborescent models fix knowledge as hierarchical and stemming from a single origin, rhizomatic structures have no center, no beginning, and no end (Deleuze and Guattari) Decentralized spatial networks — from the web to the city — operationalize the rejection of metanarratives.

Co-constructed networks like the web actualized post-structuralist ideas by providing an infrastructure with infinite capacity for adaptation. Nelson's earliest ideation of hypertext stipulated that “such a system could grow indefinitely, gradually including more and more of the world's written knowledge” (Nelson). With no regulation over the creation of new links, half a century later the topology of the World Wide Web has been incrementally sculpted by millions of individual contributions.

Like the web, the unplanned city grows organically: its texture and form are constructed by many multi-temporal hands. Hyperlinks branch into offshoots; desire paths are grafted from paved streets. The word graft originates from the Greek noun stylus, or verb to write. “The city is grafted in this graphic sense,” writes media studies scholar Shannon Mattern in A City is Not a Computer. “It's a polyglot palimpsest of codes and scripts and plans" (Mattern).

If "the city and the book are opposed forms" (Raban) then perhaps hypertext offers a more appropriate form of representation for urban systems. The city and the web meander, evade linear narrative, are continually reinvented, and resist static representation. The city's “spread, contingency, and aimless motion” cannot be forced “into the tight progression of a narrative" (Raban). Likewise, hypertext is “interconnected in such a complex way that it could not conveniently be presented or represented on paper" (Nelson).

The Arcades Project, Walter Benjamin's massive collection of unfinished writing on 19th century urban life in Paris, is the first well-known attempt to capture this rhizomatic, fragmented form of writing on paper. “With its blind alleys and its maze-like tangle of passages, The Arcades Project resembles nothing so much as a city,” states a review of the work. “To read it is to take on the aimless motion of the flâneur, the idle urban stroller” (George). The text's modularity lends itself to hypermedia: in 2006 it was adapted into a hypertext project called Reading the Ruins, which orders and reorders passages for endlessly unique readings (Peaker). The Information City imagines text as city; Reading the Ruins imagines city as text.

If “metaphors of the book inform how we 'write' and 'read' the city” (Mattern), then metaphors of the web influence how we construct and navigate the city. Manuel Castells's 1996 book The Rise of the Network Society documented how information technologies like the Internet produce a new social structure (Castells). Shortly after, Albert-László Barabási discovered scale-free networks — whose properties recur in urban and social systems — upon mapping the topology of the web. The structural relationship between digital information and cities is not merely metaphorical: “in both computing and urban planning, this arboreal language is simultaneously formal, genealogical, operational, political, and epistemological" (Mattern).

A Latourian analysis grants agency to the World Wide Web itself. The digital infrastructure of information networks does not simply represent reality; it functions as an agent and mediator of social relations. Hypertext does not describe the city so much as it actively constructs it.

Our mental maps of urban spaces are constructed through relational association, connecting similar concepts and places. Similarly, hyperlinks connect related media on the web. Readers have agency to navigate freely when exploring forking content, constructing meaning as they forge their own paths. In this nature, hypertextual information systems foster associative thinking, sense-making, and provenance.

Post-structuralism posits that complex systems themselves do not contain stable or universal meaning. Rather, meaning is constructed by the reader, the user, or the flâneur. Only through navigation and sensing does the thread, path, or narrative emerge. The routes traced by users meandering across the web serve as through-lines that allow them to construct mental maps of digital space. Similarly, one's physical journey through a city functions as the backbone which upholds more affective impressions of place.

Today, much of the media consumed on the Internet is no longer discovered independently via hypertext navigation, but suggested by recommendation algorithms. Actively seeking out information is replaced with passively consuming streams of content. “As systems take on more decision-making, users have less control over what they see and how meaning or relationships between content are constructed" (Liu). When system logics are obscured, information is unsituated and provenance is lost.

Algorithmically recommended content on the web is not networked, but sequential and linear. Media surfaces from the algorithmic black box like mysterious omens; attempts to derive meaning are a losing game. There are no forking paths nor desire lines — only the freeway.

Our cities are familiar. We come to know our blocks and neighborhoods through embodied experience: the corner store, a neighbor's stoop, our daily walk to the metro. By repeated observation and interaction we make sense of our surroundings and weave detailed memory maps. These same perceptual behaviors carry over to digital space: a 1997 study reported that 58% of users' web pages accessed are revisited to previously seen pages (Tauscher). In the Information City the question is "what can I see from here?" (Bernstein), highlighting the fact that the user's perception of media is grounded in their situated position. Users exist inside the web; as they navigate hyperlinks they are “transported” from one media site to another. In both digital and physical space, sense-making emerges through repeated navigation, orientation, and embodied experience.

The post-hypertext web — characterized by siloed platforms and recommendation algorithms — imbues the user with a metaphorical birds-eye view: the "view from above, from nowhere" (Haraway). Algorithmic media prioritizes novelty over provenance. These logics extend beyond digital space to shape how cities are navigated offline, as urban experiences are increasingly mediated by algorithmic suggestions. In 2022, a senior vice president at Google shared that internal studies revealed nearly 40 percent of young people opt for TikTok or Instagram recommendations over Google's knowledge network when searching for a restaurant (Huang).

Predictive recommendation algorithms that direct direct users to similar places based on their profile activities are subject to criticism for their lack of serendipity. The absence of serendipity stems in part from digital spaces' low sensoriability — the capacity of an environment to be sensed — thereby producing a feedback loop of incomprehensibility and disorientation (Smets).

“In any organized object,” warned Christopher Alexander in 1965, “extreme compartmentalization and the dissociation of internal elements are the first signs of coming destruction" (Alexander). The algorithmic web contributes to the fragmentation of digital space by flattening relational structures and obscuring emergent associations. The digital world does not simply mirror the physical world, but constructs it: while hypermedia enriches our mental maps of urban space by linking representations of real-world entities, linear algorithms impose false metanarratives that sever the complex relationality of place. Preserving networked information systems and the open web is necessary to sustain infrastructures that afford participatory sense-making. In reading sensorial and sensible hypertext, we write richer, more complex, and navigable hypercities.