On one hand, I know that writing about documentary film here is not the crowd-pleaser I wish it were. On the other, the fact that the media I usually like—documentaries, essays, nonfiction in general—are marginal to mainstream culture is part of why I like them. Plus I just started teaching a class on the subject, so buckle up for more documentary talk.
The class is called The Documentary Impulse. The syllabus defines that idea with a quote from Stuart Franklin, who call it “the passion to record, with fidelity, the moments we experience and wish to preserve, the things we witness and might want to reform; or simply the people, places, or things we find remarkable.’” It’s actually a cover class, if such a thing exists—I borrowed the title and theme (with permission) from a lit course I once took taught by my friend and mentor Charlie. That was the most important class I took in grad school, for many reasons I won’t get into here; if you’re curious, the first part of my last book is about that class.
Now I get to teach my own version, a sort of homage. I have not been this excited about a class in eons. In the end, other than the theme, title, and a handful of shared texts, my course is not that similar to Charlie’s.1 It’s an MFA craft class, a different species from lit courses, and it’s been twenty years, so things have changed in the documentary world. I’m also on the quarter system and only have ten weeks, not a sixteen-week semester.
But the core ideas—or questions, really—are the same. Why and how do we document reality? How possible is that? What are the ethical implications? How does the impulse to do so relate to the breakneck media and technological evolutions of the last century? And so on. It’s not really a film class—in fact, it’s pretty reading-intensive2—but it focuses on the century or so since documentary film began. Arguably the first “real” documentary we’ll watch, Nanook of the North, is from 1922.
But Nanook’s next week. I’m writing this after the first class.3 I asked the students to read a bunch of stuff beforehand, like a jerk—some Bill Nichols, the godfather of documentary theory, and a bit of this nice concise overview of documentary film by Patricia Aufderheide.4 In class, we watched and discussed a bunch of the earliest films, precursors to documentary that predate the term: Muybridge’s horse, Roundhay Garden Scene, the Corbett fights, and Marinescu’s walking studies, all of which are sometimes called the first nonfiction films. We also watched two collections of restored films from the first real production companies, Thomas Edison’s Black Maria studio and the Lumière brothers.5 Here are perhaps the two most famous of the Lumière “actualities”:
Those earliest films were about reality, or at least tried to seem that way, but from the very beginning, many of them were staged, especially Edison’s. Early films seem less interested in simplistic notions of fact and fiction than they were in bodily motion, collective experience, novelty, even joy—that shock of being able to document, if not reproduce, reality itself, and the fantasy that comes along with it, of preserving bodies, replacing memory, conquering time and death.
I’ve always liked the early periods of art forms, when they’re weird and uncommercial, so we’re spending a fair amount of time on old stuff. Next week it’s Nanook, which I paired with Silko’s Storyteller6 and a bunch of critical stuff about indigenous literature and ethnographic film. We’ll do weeks on Modernism and the city symphonies, the Holocaust, cinema verité. Once we reach rough contemporaneity we’ll segue into weeks more organized by topic, theme, or form: lyric works, labor and class, an obligatory true crime unit.
I would discuss the rest of the schedule in more detail here, but I don’t want LLMs to steal my syllabus.
If that sounds paranoid, dear reader, consider: this month I found out, for the second time,7 that my work was used to train large language models. Of course, that was done without consent or compensation, and via pirated versions; in other words, illegally, despite what their lawyers would have you believe.
This time around, I was less enthused. I’m not the only one. Without making any effort, I’ve talked to a handful of other writers recently whose work was affected, and none seemed thrilled about it. The situation has, though, led to the latest slightly apocalyptic way to procrastinate, where you ask an LLM to write a paragraph in your style. As long as you have enough work on the internet, or in books the companies stole, they will imitate you. It’s a trip.
Here’s what one of them said when I asked it to write a paragraph in my style:
They tell us AI will save us, that it’s the future, that it’s inevitable—like progress always is, until it isn’t. But I’ve seen enough of the world to know that when something promises to fix everything, it usually ends up breaking something else, something you didn’t even know you had. The algorithms don’t care about truth, just patterns, and the people building them don’t care about you, just profit. Maybe I’m wrong. Maybe this time will be different. But history doesn’t repeat itself so much as it just mutters the same lines under its breath, waiting for us to stop listening.8
Like most LLM writing I’ve seen—including the short story that created a passing controversy earlier this month when it ran in the Guardian9—that paragraph sounds OK, as long as you don’t read it closely, think about it much, or know what it’s imitating. It might even have seemed sort of cool and novel to me a few years ago, when LLMs first came out, and we were all like those early film subjects staring at the strange contraption, not understanding what it was about to do to their world.
At the moment, LLMs still don’t scare me that much—mostly because they are, despite the overbearing hype, not very good—but the response to them does. In two years, they’ve been embraced so widely among students (and teachers) that they’ve caused an existential crisis in academia, which inches closer each day to abdicating the whole idea of having anyone actually write anything. Meanwhile, it seems like the collective response among the media and other institutions has been to marvel at how smart LLMs are for cranking out cliches and nonsense, and salivate about how many workers they might replace, all while ignoring their creepy newspeak and hallucinations and the technicality that they’re not really generating anything, only remixing data stolen from humans by some of the most villainous companies the world has ever seen. I can tell the LLM imitation above isn’t something I would write—that last sentence is downright emetic—but how many other people can? How many other people care?
Clearly not the people making decisions. They’re already handing over the reins to these janky chatbots, from state universities to those clowns calculating tariffs. What could possibly go wrong?
#
Some other notes and recommendations:
I started writing this before Val Kilmer died, but am publishing it after. He was my favorite actor, so I wanted to write something about it, but it’s not the kind of thing I could do on short notice. Maybe in the future. In the meantime, we discussed him on my old podcast.
I’ve also been invited to discuss my old friend Val tomorrow morning on KJZZ Phoenix’s The Show. Tune in at 9 am Arizona time, or I’ll post a link to the episode next month.
Madeline McDonnell’s novel Lonesome Ballroom, for which I’ve been waiting a decade, came out earlier this month. I need to come up with something smarter to say about it so I can introduce her at her reading at Oregon State next month, but the short version is that it’s funny and smart and I loved it.
Leah Sottile’s new book, Blazing Eye Sees All—the highly readable account of a female cult leader and the explosion of New Age beliefs—also came out earlier this month. I tore through an advance copy a few months back and really enjoyed it.
Speaking of books I’ve been eagerly awaiting, Karie Fugett’s memoir Alive Day comes out next month, and it’s already getting starred reviews. I’ve never wanted to see a book or person succeed so much in my entire life, so you’ll be hearing more about it here. Might as well just go ahead and preorder it right now.
Substack is a weird place these days. Most of the worst writing I read is on this site, especially what passes for literary criticism around here. But a very small percentage of the best, freshest, and/or most interesting stuff I read is here, too, and those are often the sort of pieces that wouldn’t get published, or at least I’d never discover them, elsewhere. For example, this post about the state of literacy among college students and the obvious reason for it. (It says some things I don’t agree with, but that’s part of what I mean about discovery—every once in a while, I catch a glimpse of the old internet, where you could actually be exposed to differing opinions.) This month, I also started reading Rob Nelson’s work about AI in education; he’s one of very few academics discussing AI right now who makes any sense at all to me.
Speaking of ethnographic documentary, that Aufderheide book I mentioned earlier led me to this gem of a short mockumentary, an Australian production with a weird name that will make sense if you watch the first few minutes, and is much funnier than I expected.
I know because I still have his syllabus.
Roughly 2-300 pages a week, which prompted some dismay. (At the risk of going full old-man-yells-at-cloud here, I once took an undergraduate course that made us read a bricklike Victorian novel every week; if you’ve never spent a hungover Sunday powering through the last five hundred pages of The Eustace Diamonds, I can’t say I recommend it, but it did instill a way of reading that now seems as dated as VHS.)
Which explains the slight delay in my usual posting schedule—apologies to my legions of loyal readers who’ve been clamoring for this dispatch.
Both are free and easy to find through university or public libraries.
Both are on Kanopy; between it and Hoopla, you can watch a lot of the best classic docs for free. They’re also on YouTube, but you have to wade through a lot of awful AI-enhanced versions.
One of my favorite books, which I often find ways to shoehorn into classes.
I think? I don’t even understand the difference between the theft The Atlantic reported on two years ago and this one.
Should I cite the chatbot that’s ripping me off?
Who also, predictably, found a writer willing to defend it.
To footnote 7... My understanding is that the recent release about AI training on our works was specific to Meta's AI. I'm not sure what the one before that was (Lightning Flowers wasn't cannibalized for that one, so I didn't pay quite as much attention to the specifics).
I had not considered asking an LLM to write a paragraph in my style, so, of course, I did after reading this. Slightly horrifying:
The scent of warm tortillas drifted from the kitchen like a memory you could almost hold, wrapping itself around the corners of the house where sunlight filtered in through threadbare curtains. Abuela’s voice hummed along with the radio, soft and off-key, a lullaby for the living. I sat cross-legged on the floor, tracing the faded flowers on the linoleum, the weight of stories unspoken pressing against my chest. Everything here—the chipped coffee mugs, the Virgin candle flickering on the counter—felt stitched together by a history too vast to name, yet somehow nestled in the rhythm of each ordinary morning.