The Death of the Author
As everybody knows, Nietzsche announced long ago that “God is dead.” Or more accurately, he wrote the parable of the madman, who was the first to realize that God was dead. As Nietzsche’s madman tried to spread his message, he was thrown out of countless churches and each time he responded stubbornly, “What are these churches now if they are not the tombs and sepulchres of God?”
Nearly a century later, Roland Barthes declared the death of the author. The connection to Nietzsche was clear implicitly, but it was also made explicit when Barthes spoke of the Author-God: the intelligent designer of the text who imbues it with order and meaning, and to whom all criticism and interpretation must ultimately be addressed. The Author-God is dead, and he’s been replaced:
We know now that a text is not a line of words releasing a single ‘theological’ meaning (the ‘message’ of the Author-God) but a multi-dimensional space in which a variety of writings, none of them original, blend and clash. The text is a tissue of quotations drawn from the innumerable centres of culture.
Roland Barthes, “The Death of the Author”
Talking about death is uncomfortable–and it’s hardly the easiest way to attract an audience. Unsurprisingly, theorists and philosophers, so often the harbingers of death, rarely gain mass appeal. Whether it’s deconstructionists like Roland Barthes and Jacques Derrida, or accelerationists like Nick Land, theorists get accused of encouraging apocalypse. But that misses the point: their ideas aren’t aspirational, but despite this and perhaps because of this, they’re often right. Camus put it well in his appraisal of Nietzsche. Nietzsche, he said, “recognized nihilism for what it was an examined it like a clinical fact.” Theorists aren’t in the business of architecting utopia: their job is to diagnose the present.
Returning to Barthes, his diagnosis was of course a prophesy of death: The Death of the Author. Writing, he said, has taken on a new meaning. It is now apparent that the author doesn’t build a world of his own ex nihilo, but rather channels innumerable cultural influences and presents not a linear monolith, but rather a galaxy of signifiers, open to interpretation. This gives way to the Barthesian circle of life: “the birth of the reader must be at the cost of the death of the Author.”
In the new world, prophesied by Barthes and others, and realized by technologists over the following decades, writers become mere relay nodes in a global network of symbols and ideas. It was already true, in 1977, that the genius of the author had been subsumed by semiotic bricolage. Accelerating technology rebuilt the literary environment to make this an unavoidable fact in the media we now inhabit.
Crying out hopelessly from the zeitgeist of the ‘70s and '80s, a technologist named Ted Nelson wanted the media architects of his day to know that a profound transition had already begun. The author is now a bricoleur, he declared: our new media systems must honor this. As computing technology advanced and our media systems evolved, Ted remained a persistent advocate for a comprehensive redesign that he called Xanadu. His ideas were buried by the explosive success of the Web, but then, decades later, many of his ideas were revived among a set of tech hipsters: roamers and digital gardeners and hypertext maximalists. I was among this set. I spent my senior year of college in close communication with Ted Nelson as I set out to write his story. The redesign of our media environment for its new life on computer screens felt like one of the most interesting chapters in the history of technology. And for me, as someone studying history, the fate of writing and research held a special significance.
What’s so appealing about the idea of hypertext authorship? I now believe it’s a sort of nostalgic, reactionary appeal. Hypertext authorship is the retrieval of the author at a higher level of abstraction. When authorial genius becomes inaccessible, bright minds can turn instead to systematic curation and organization of ideas. This is what Ted sought: the writer of the future need not labor to produce a glimmering monolith of text, but rather, writers can enter into free play in the realm of networked ideas.
Nonetheless, even this vision falls short. Hypertext authorship centers a new kind of authorship, in which the network of ideas in the author’s head can itself be shared, without being reduced to a line of words. The death of the author, however, will not give way to a new kind of author. Roam and other kinds of hypertext personal knowledge management systems are nothing more than the “tombs and sepulchres” of the author. They still revolve around the microcosm of the Author-God, in the form of collected notes and clippings. The Barthesian circle of life will not be realized in the proliferation of a million hypertext authors. Rather, the role of the author will continue to shrink until it is completely consumed by the act of reading.
NLP, LLMs, and GenAI were the missing piece. The text is where a “variety of writings, none of them original, blend and clash,” but until now these interactions couldn’t be modeled comprehensively. Now we have the model. A systematic way of reading will make the death of the author complete. At that point, when texts can be mathematically modeled as a galaxy of signifiers and the interactions between ideas can be calculated automatically, reading will finally outgrow its dependency on the author. Historically, it’s been the job of the author to elaborate new ideas as an outgrowth and recombination of the ideas that came before. But soon, all these latent ideas will be readily discoverable by the reader and, empowered by new kinds of tools, readers will be free to roam farther than any author could reach.
As for Ted Nelson’s ideas, I think they’ll still have their day. When readers explore the unbounded multi-dimensional space of ideas, it might after all look a lot like Xanadu.
P.S. The irony isn’t lost on me: this essay itself probably lives somewhere in the latent space of an existing or soon to exist Large Language Model.