The Forms and Limits of Hypertext
I’ll confess that I’m a pretty linear thinker. I come by it honestly, having spent a bit over a decade studying and then practicing a pretty hardcore version of analytic philosophy.
“What’s analytic philosophy,” you ask?
A rigorous answer to that question would require more words than any sane person is willing to read, but the nutshell version is that analytic philosophers break claims down into their smallest constituent parts (be it terms, concepts, or propositions) and then analyze each part through the lens of formal logic.
As a good analytic philosopher, I’m on the record arguing that postmodernism as Theory is incoherent, at least when applied to moral or political questions.
And yet I’m fascinated by the very postmodern idea that links between texts are more important than the arguments within said texts.
For the last couple of years, I’ve been pretty obsessed with the idea of the nonlinear text. The text that doesn’t simply move the reader from A to B to C through a series of logical steps. The text that the reader creates by choosing which steps to follow.
So at the risk of having the Analytic Philosophy Police take away my diploma, I’m going to talk about hypertext theory, which is a pretty postmodern (or, more specifically, post-structrualist) concept. Fair warning, though: I’m still an analytic philosopher trying to explain postmodernism, so this is going to be pretty handwavy. (Insert your own joke about postmodernism and hand-waving.)
What Is Hypertext?
In its simplest form, hypertext is simply a set of nodes (call them cards or documents or pages or whatever) and a set of links between those nodes. The links can take two forms:
- Unidirectional links allow one-way movement between nodes.
- Bidirectional links allow back-and-forth movement between nodes.
I’m going to talk only about unidirectional links here. Bidirectional links are cool and all, but the HTML hyperlink is unidirectional, and there’s really what I’m interested in.
Hypertext can be organized in several different ways. Here are a few that seem particularly relevant to the web.
Axial Structure
Axial is a really pretentious label for something fairly simple — movement along a line. Axial hypertext isn’t strictly linear, though. Rather, readers are invited to take explore regular digressions from, then return to the main text.
Basically, this is how we write every piece of nonfiction on the web.
- A feature story with some included image galleries? Axial hypertext.
- A sidebar that expands over the main story, then shrinks back to its initial location when you’re finished? Axial hypertext.
- A video embed that offers some additional context for the narrative? Axial hypertext.
- A card stack at the end of a Vox article? Axial hypertext.
I could go on, but I think you get the point.
Arborescent Structure
Another terrible term. Arborescent just means treelike, but no self-respecting postmodernist uses a simple word when an esoteric one will do. Anyway, an arborescent structure has a defined beginning point, and many non-overlapping ending points (see figure 1 below).
If you rotate this structure 90° clockwise, you’ve pretty much got a sitemap. Indeed, an information architect might call this kind of structure hierarchical — each node is a parent of its connected nodes on the right, and a child of its connected node on the left. Hierarchical language makes a bit less sense when applied to narrative text, so we’ll stick with arborescent.
The most obvious example of an arborescent text is the Choose Your Own Adventure story. The CYOA stories provide lots of paths through the text, with each path leading to a pre-determined ending. The structure requires presenting readers with clear binary choices — a requirement that is likely to feel pretty artificial outside of fiction.
In the digital realm, a lot of video games utilize the arborescent model, but I’m not aware of much online nonfiction employing this type of structure. If you’re looking for a real-world example without having to go dig up a Choose Your Own Adventure book from your parents’ attic, go spend a few minutes with You Are a Bear.
Layered Structure
As the name might imply, layered structures involve two different layers of linked nodes. Each node in the top layer is linked sequentially to the next node in the top layer, as well as to its corresponding node in the bottom layer (see figure 2 below).
Typically, the top layer would contain text and the bottom layer would contain photos, sounds, and video.
There are a lot of cool possibilities for layered hypertext. The text of the Pitchfork essay, “Glitter in the Dark,” is a standard entry in the Musician with a New Album genre. What’s different is the layer of stop-motion images that accompany the text. While they aren’t strictly part of the main narrative, they do evoke a sense of personality that deepens the text.
A slightly different take on the layered structure, the publishing platform FOLD, lets users add a layer of multimedia to accompany a text-narrative. The platform, created by the Center for Civic Media at MIT, offers a slightly different take on the same idea. It utilizes a set of cards in the main narrative. Each card is then linked to a second layer that contains background information — videos, text, maps, etc.
Networked Structure
I’ll admit that this is the one that really fascinates me. In a network structure nodes are connected by multiple links, potentially overlapping links (see figure 3). There’s no particular starting point, and no ending point. It’s possible to hit every node, though there are multiple paths through the structure, and many of those paths would hit the same node multiple times.
You might recognize the network structure as the basic form of the web itself, in which any given page may link to a multitude of other pages — some or all of which may link back. We’ve gotten pretty good at building these kinds of links between articles.
But I think we’ve not yet really thought about what it might mean to create a single publication that comprises a network of linked nodes.
Indeed, I can’t say that I’ve really seen this tried anywhere, though in one of those bits of self-referential meta-ness that good analytic philosophers like me are supposed to disdain, this very thing that you’re reading is one part of an attempt at doing just that.
Related
