Total authorship and generative AI

Prompt: “poetess in a field of paper, watercolor” in Midjourney for the central image, then outpainted with DALL·E 2 using the prompt “a field of paper and poppies, watercolor.”

As I keep thinking about tools for imagination and what it means to go with the grain of generative AI, one term keeps resurfacing like a chorus: total authorship. To my mind, total authorship means end-to-end responsibility for a creative work, with a single human imagination making every meaningful decision along the way. My hope with this post is to recount the brief history of the term “total authorship” as far as I can trace it, as a point of reference for future conversations about generative AI and creativity more generally.

Total authorship is relevant for generative AI because some of what catches us off-balance about these new tools is that they explode the scope of what a single person can create “on their own,” in collaboration with AI. Generative AI multiplies what an individual can do, and a common intuition is that this will drastically reshape the landscape of what gets expressed, and by whom.

I first heard the term total authorship from Robin Sloan, a friend and for the past decade and a half one of my favorite thinkers. Robin is now a traditionally-published novelist, but one of his early creative writing efforts unfolded in the early days of Kickstarter. In the summer between my two years at business school, I interned at Kickstarter and asked Robin if I could write a retrospective of his project; he graciously agreed to participate. The resulting case study for Kickstarter’s blog captures the concept of “total authorship” as I first encountered it:

Robin’s vision, from the beginning, was to write, design, illustrate, and ship the book himself. Reflecting on that decision, Robin would later say, “In retrospect, I would perhaps not have done everything myself. At the time, I was really intent on this idea of ‘total authorship.’ I was just very compelled by—writing the words, drawing the pictures, creating those illustrations, designing the whole book, picking the font—the whole thing.”

I was so taken with Robin’s take on the term that when I was invited to be a fellow at the Berkman Klein Center for Internet and Society at Harvard starting in the fall of 2012, I included “total authorship” in my statement of interests:

Diana is an MBA candidate at Harvard Business School. As a co-creator of ROFLCon, her interest in internet culture runs deep. Most recently, this interest has expressed itself in learning Ruby (a programming language) and applying an open-source ethos to mentoring. She's also exploring the concept of total authorship as it relates to art.

Burrowing in to Google find other references to “total authorship,” there aren’t many, but one stands out as sharing the same meaning: a 2014 interview with comic artist Danijel Žeželj. In response to the interviewer asking “You’ve worked for big American publishers (DC/Vertigo and Marvel) and for smaller independent ones like we are, as well. It seems that this distinction in your case is inexistent or that you at least don’t care about it. Is there, after all, a difference in working for one or the other – and if so, how does that influence your creative process?,” Žeželj replied:

There are similarities and differences, but often these kinds of work complement each other. The projects for the big publishers are mostly done for scripts I haven’t written myself. I was lucky to have cooperated mainly with great script-writers, so the process was a positive experience and a challenge and I wasn’t forced to make compromises regarding the quality of work. Work for corporate publishers has its own limits, but inside these limits there is always room for creativity. On the other hand, working for oneself has its limitations as well, so these walls and barriers are quite relative. The main difference is that total authorship requires total responsibility.

Total authorship requires total responsibility. There are limitations to every creative configuration, but end-to-end ownership leaves less to accident; one person has to care about the whole and create anything that’s missing for it to be complete.

Deeper on Google Books, searching for “total authorship” turns up a book called Auteurs and Authorship: A Film Reader, and the following passage on the role of the director in filmmaking:

There can be no clear distinction between supervising the cameraman and creating the images, or between advising the actors and moulding the performances. But unless we consider acting and photography to be the whole process of film-making, the director is still a long way from the total authorship that is often claimed for him.

This passage gets at a tension at the heart of total authorship. Taken literally, total authorship is an unreachable extreme—we always stand on the shoulders of giants. If our medium is words, those words are legible because they’re held in common; we didn’t invent each component part. If our medium is paint, the pigment comes from somewhere. But as a concept, total authorship helps capture an important insight: when all iteration and cohesion of a creative work can happen inside a single person’s head, there can be a different quality to the output. High-integrity idiosyncrasy becomes possible to a degree that would have been difficult any other way.

There’s a reason that novels are the ground truth of so many fictional worlds that have since spun into movies, theme parks, games, and more. Words are low-overhead, flexible, fast to get down. One word “costs” the same as any other; we can litter paragraphs with glimmering, unforgettable details at no extra expense. With the text-to-everything dynamic currently dominating the world of generative models, the person who would have been a novelist can become, overnight, a graphic novelist—and soon, perhaps, a film director.

Total authorship is not the end of the story for generative AI. As excited as I am by it, I’m even more interested in how generative AI will sway the nature of interpersonal collaboration. But since total authorship captures an eddy of awe swirling around what generative AI unlocks, I wanted to pull together this exploration to anchor that perspective as a clean source of contrast for alternative visions to come.

Speaking of authorship, Robin Sloan has a new novel on the way!