Well, the Author is Dead. Now What?

In 1967, French literary critic Roland Barthes wrote an impossibly influential essay called, in English, “the Death of the Author”. The critical framework he lays out in this essay became the foundation for how people of my generation were taught to read literature, starting in junior high school, so I won’t go into his arguments in detail, but in summary, they are:

1. An artwork cannot be said to have one single, definitive meaning.
2. Previous ideas about an artwork’s single meaning are flawed because they rely on using information about the author’s intentions to determine that meaning, but this information is actually not relevant.
3. The author’s intentions are fundamentally unknowable in most cases anyway.

Point 1 seems defensible to me—artworks exist intersubjectively, like language, and a word does not have a definitive meaning. The meaning of words is shared between all members of a language community, and those meanings can shift over time for good reasons, bad reasons, or no reasons. So just as the letters b i r d refer to many different concepts, so do the words that make up a text. If someone says that such-and-such a word or a novel has some given meaning to them, I cannot say they are wrong about that meaning; I can merely choose to consider their meaning irrelevant to my meaning. My meaning-making project may look very different from theirs, which is their prerogative.

Points 2 and 3 never sat well with me, but for years and years I have been in a vocal minority. Merely to say “Well, I don’t think that’s what the author meant” invites the catchphrase “death of the author” like Barthes had presented a mathematical proof, an inviolable Law of Literature, that to disagree with was merely to be ignorant of, like trying to invent a perpetual motion device.

But of course that is not what Barthes did. Implicit in Point 1 is the idea that there is no objective point-of-reference in art criticism; it is impossible to “prove” that someone’s analysis is “wrong” because that is not what an analysis is. But if that is so, how can my analysis, into which I do care to take the author’s intention into account, be wrong?

This is ostensibly where Point 3 comes in: you can’t take the author’s intention into account, because you don’t know it. You probably can’t know it.

Okay? That’s actually true every day. Noël Carroll argues in On Criticism, which I can’t lay my hand to at the moment, that we resolve ambiguity in every day conversation by making inferences about other people’s intentions all the time, and that this is a perfectly valid and respectable thing to do when trying to communicate.

And we are trying to communicate, aren’t we?

Not so, says Barthes! This is where the essay loses more than my agreement, it loses my sympathy. He writes:

[L]inguistics has just furnished the destruction of the Author with a precious analytic instrument by showing that utterance in its entirety is a void process, which functions perfectly without requiring to be filled by the person of the interlocutors: linguistically, the author is never anything more than the man who writes, just as I is no more than the man who says I: language knows a “subject,” not a “person,” end this subject, void outside of the very utterance which defines it, suffices to make language “work,” that is, to exhaust it.

If I understand him correctly, he’s arguing here that linguistics has demonstrated that language is a fundamentally mathematical process rather than a mystical one; Chomsky’s Generative Grammar can make a sentence with no content, so why not a novel with no author? You don’t need a soul to write a book, so why treat a book like it has one?

When I was in high school, I struggled to explain my fierce objections to this philosophy to my English teacher. It occurred to me that these methods of analysis could be applied equally fruitfully to literature that had had no author at all, to artworks created without artists. But those were mere hypotheticals, thought experiments in which you encountered a beautiful painting on a lifeless alien world, a non-painting created by chance, by mere geology, or a novel created by pulling Scrabble letters out of a hat. I never brought these ideas up to my teacher because I knew he could dismiss them as so improbable as to be irrelevant.

But I have lived to see my hypotheticals come true.

The Author is Dead, Long Live ChatGPT

In the age of generative AI, we are surrounded by text with no author. We have met the void process whose subject creates empty utterances, the scriptor with “no being that precedes or transcends its writing.”

And we hate it. We recognize it as an inhuman, an anti-human force. We reject it.

This rejection is expressed most succinctly and most forcefully in the phrase, “I don’t see why I should bother to read something that nobody bothered to write.” I wish I could find out who coined the phrase and give them due credit. They should have a plaque erected in their honor, and under the plaque should be a note addressed to Roland Barthes, that here was a person, no mere subject, whose being precedes and transcends what they have written, and this is what gives their words meaning.

Not meaning in the semantic sense, that is to say what makes us understand those words; rather, it is what gives those words value, it is why we care about what is said. Roland Barthes’s authorless text embraces a diversity of meaning but ultimately no value of any kind.

Barthes wrote:

In this way is revealed the whole being of writing: a text consists of multiple writings, issuing from several cultures and entering into dialogue with each other, into parody, into contestation; but there is one place where this multiplicity is collected, united, and this place is not the author, as we have hitherto said it was, but the reader: the reader is the very space in which are inscribed, without any being lost, all the citations a writing consists of; the unity of a text is not in its origin, it is in its destination[.]

This attitude is what, in my cynical view, makes Death of the Author so appealing to some people: it reassures them that their perspective is correct, or least not subject to correction by others. It is impossible for a reader to be wrong, and that is very comforting to someone who instinctively recognizes how intimate the process of reading is and feels threatened by the prospect that those personal feelings are in some way deficient. It is tempting to say “I am the one reading this book, and I am going to read it however I like.” Well, sure, you can do that, and nobody can stop you.

But if, as I argued in the post I linked in the last paragraph, the value of reading is in the space it makes for other people, for empathy engendered by thinking other people’s thoughts in your own head, to close yourself off from this correction is to resist that very empathy. To disregard the author’s intention is to reject their otherness, to flatten them to an extension of yourself. This feels to me like the primary rhetorical move that teaching this methodology so widely has enabled: to turn reading selfish. No wonder it’s so popular.

But Large Language Models are exposing the fact that this methodology does not match all of our instincts, all of our intuitions. We have read writing by a dead author, by an author with no intentions. We don’t want it. We want books that are written by humans; not because humans are better at writing (though that is mostly still the case), not because humans use less water or electricity, or any other contingent fact about the current state of generative AI technology, things that may not be true tomorrow. We want books that are written by humans because we care what humans have to say. We care about what Adam Savage calls their point of view. Maybe someday a charming android with a positronic brain will have a point of view, and I will be delighted to read the novels it writes. But LLMs do not have a point of view. They do not have anything that they are trying to do. They have no intentions. They do not intend.

So I am going to continue caring about who an author is and what they think. I am going to continue to let that information color the sense I make of their writing. I am going to carry on treating the author like a living person. Barthes’s assertion that it does not matter what an author thought is an incomplete one. Mattering, like meaning, does not exist in the physical world, to be detected and demonstrated and proven. He means merely that it does not matter to him. It matters to me.