Discover more from Gideon's Substack
An obsolete streaming contract and the threat from A.I. pose very different problems
A striking sign in Los Angeles. Sadly, I do not know.
I’ve never worked on a big-budget film or a series. Everything I’ve been involved with has been either an indie film or something even smaller (like an indie short). So take my thoughts below with a big grain of salt. But I’ve been thinking a bunch about the issues at stake in the WGA and SAG-AFTRA strikes, and the more I think about them the more I think that one of the conflicts is not like the others. That conflict revolves around artificial intelligence.
Gideon's Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
All the other issues are classic labor-management contract disputes. Residuals are the easiest to explain. Streaming services generally operate under a new media contract that was written during an era when it wasn’t clear whether there would even be material revenue from putting video on the Internet, and that contract was deliberately written in such a way as to not hamper the development of a novel and uncertain channel. Now that streaming video on the Internet is the dominant way that people watch movies and television, that contract has been revealed as wildly unfair, offering a far lower share of the rewards from success than is the case for contracts for broadcast television or film. This piece from The New Yorker does a good job of describing that unfairness from the perspective of union actors; the perspective of union writers is broadly similar. And the solution to this problem is very straightforward: demand a fairer distribution of profits and strike to get it if it isn’t offered. And that’s precisely what is happening.
The folks on the business side will push back broadly with two points. First, they will argue that the data to calculate residuals in a manner roughly comparable to the metrics used for other distribution platforms is a valuable and proprietary trade secret in the streaming world, and therefore they can’t use it as the basis of a contractual obligation without damaging their business. Second, they will argue that the streaming space is overcrowded and that apparently robust current profits are actually under considerable threat going forward, so making concessions to labor based on recent numbers would lock them into terms that can’t be sustained going forward. These objections are very readily dispensed with. To take the second point first, the answer to an overcrowded space is consolidation. Yes, consolidation will mean some reduction in the number of projects, and hence some reduction in the number of jobs, but that’s not a terrible trade if the jobs are more reasonably structured. And if a consolidation needs to happen, far better it happen under a fairer contract than one that makes it easier for management to screw labor.
As for the first point, I actually think it’s true, but it’s nothing labor should care about. Everyone in the industry would benefit from more transparency except for the streamers themselves—and that very much includes the legacy competitors of the streamers who have been moving into the streaming space for defensive reasons and, generally, losing a pile of money doing so. That disjunction of interests between the streamers and the legacy companies is actually the best hope for labor to prevail on this point in negotiations. I understand why Netflix would fight tooth and nail to keep their data to themselves, but I’m less clear why NBCUniversal would fight tooth and nail for Netflix’s interests; Peacock is just never going to swing a comparable stick, and I’m sure they know it. More generally, though, while labor has every reason to care about the viability and growth prospects of the industry, because if they hobble it their members lose jobs, they have no reason at all to care about the share prices of the companies they are bargaining with, and that’s ultimately what’s at stake in the fight over residuals.
A lot of the more arcane issues that the WGA is fighting over and that don’t have real parallels for SAG—the proliferation of mini-rooms, the way shorter seasons have dramatically cut effective annual pay, etc.—fall broadly into the same bucket as the far more central important residuals question. These are all classic labor-management disputes, and while I can’t say that labor necessarily has the best argument or the leverage to win in every case, it really is enough to know that it’s management vs. labor to be able to decide broadly whose side you want to be on.
But there’s one question that I don’t think is really like that: A.I. Let me explain why.
The disputes about A.I. are still partly theoretical at this point, but only barely. Digital video and then CGI have already radically transformed filmmaking twice over, and it is very clear that the studios see even huger efficiencies to be achieved through the deployment of A.I. They may very well be right about that—and if they are, they are not going to give ground.
Let’s take the writers first. This Tim Lee piece investigates how existing A.I. has already transformed the writer-centric field of translation services. Automated translation works well but far from perfectly, so if you want to translate something the right way to approach it is to hire a translator and have the translator use A.I. as a tool to speed up the process of translation. The result could have been a dramatic reduction in the number of people working in translation, but it hasn't been because demand for translation has risen as the cost of translation has fallen. Lee’s bottom line conclusion is that while A.I. has depressed wages somewhat, it hasn’t resulted in less employment or worse employment conditions; instead, by driving prices down, it has expanded the market.
Is that what will happen with screenwriting? I don’t think so. The reason is the attention economy. There are only so many hours any of us are awake in a given day, and I have a feeling we’re already approaching the maximum number of waking hours that we can spend staring at a screen. The streaming space is overcrowded because every streamer is shoving way too much “content” through their pipes in a bid to make themselves an essential subscription. Meanwhile, once you subscribe, it doesn’t cost a consumer anything to watch a given show—no money, anyway, just time. The result is that most of that “content” never had a chance to be financially viable, and lots of it is being canceled or even disappeared.
Making the process of writing a screenplay faster or cheaper doesn’t solve the essential problem here, which is a shortage of time and a ruthless competition for available eyeballs. It just makes it easier to keep the brutal arms race going a bit longer before the moment of consolidation. And when that moment of consolidation arrives, the number of screenwriting jobs should shrink, because A.I. will make it possible to make a show with far fewer writers, just as robots make it possible to make cars with far fewer workers on the assembly line.
Any demand to limit the use of A.I., then, isn’t just going to redistribute revenue between shareholders and workers. It’s going to make the process unnecessarily inefficient. There’s a big difference between saying “pay workers more for their labor” and “pay for workers you don’t believe you need,” and I think management is simply going to refuse to accede to the latter.
The situation with A.I. is similar but, if anything, even worse for actors. I know a number of actors who have already been “scanned” by productions—usually for video games, but also for television series—because this makes it possible to insert them into scenes using CGI, which is a lot more efficient than filming them again. A.I. is going to vastly expand the scope of what can be done this way. I suspect we already live in a world where crowd scenes are more likely than not to be digitally-generated. I don’t think it’s an exaggeration to say that we’re very close to a world where hiring background actors generally is just not economical. It’ll just be so much cheaper to create them using A.I.
This, understandably, has the actors up in arms—particularly ordinary working actors who have few lines or who work solely in background with no lines at all. If your social media feed looks like mine, you’ve probably seen a steady stream of angry posts about a proposal that studios put on the table in their negotiations with SAG that background actors be compensated for one day of shooting to be scanned, after which the studio who did the scan could use that image as often as they liked without compensation. But it’s important to recognize that what’s apparently outrageous about this proposal isn’t that it doesn’t provide adequate compensation for labor; it’s that it doesn’t provide adequate compensation for the use of (purported) intellectual property. After all, even if the scanned actor’s image appears in thousands of subsequent shows, they didn’t do any work on those shows. They only did one day of work, being scanned. If they own their image in some inalienable way, it’s that property that they want to be compensated more fairly for the use of.
But are they going to be? I doubt it, because I don’t think that property is actually worth very much. The same A.I. that will make it easy to manipulate their scanned image will make it nearly as easy to create an image of an entirely fictional person. If that’s the case, then how much value can the image of an ordinary background actor possibly have? The entire profession of background acting—and the jobs of all the workers who support that profession, from hair-and-makeup people to wardrobe to second assistant directors—are at serious risk of being rendered obsolete. Moreover, a strong stance on property rights in one’s personal image will accelerate that obsolescence, not slow it down, by making it more clear that a fake person is cheaper than a real one.
Where, though, did that fake person come from? A.I. images are the product of training data, and so this question of property rights also gets brought up with respect to the training data set for the A.I. If you train it on a bunch of text or images that you don’t have rights to, haven’t you violated the property rights of the owners of that text or those images? Without wading too far into that question, which I’ve touched on before, it is again worth noting that a strong stance on property rights here benefits those who already own a lot of intellectual property, not those who would create intellectual property (or, you know, art) in the future. I’m quite certain that Warner Bros. Discovery has enough intellectual property to train an A.I. on. A starting writer or filmmaker? Not so much. So if it becomes cost-prohibitive to make films without using A.I. background actors, and you have to pay for the right to use copyrighted training data, who will be priced out of competition? Not Warner Bros. Discovery.
The bottom line, for both writers and actors, is that A.I. is like any other form of automation in terms of the threat it poses to their jobs. If they can generate more output, and the market can absorb that greater output, then consumers will benefit and the workers will still have jobs, albeit somewhat different ones, perhaps paying more, perhaps less. If they can generate more output, but the market can’t absorb greater output, then the cost of production will go down, and consumers will benefit, but the number of jobs will shrink dramatically, though the ones that remain may still pay at least as well. I think that’s what’ll happen to writers. If they can’t generate more output, though, then their jobs will simply go away. I think that’s a very real risk for a lot of actors. And because it’s a very real risk, with a very real benefit to the companies that currently employ them, I don’t think those companies are going to give ground in any important way in their negotiations.
I hope I’m wrong on all counts, by the way. Partly that’s because I am a writer. Partly it’s because I love actors, and I want to see them continue to get jobs. More so, I love actors as actors, and I dread the prospect of a future where their deeply human activity is replaced by a machine that feels nothing, when feeling is so essential to what it is an actor does. I had a marvelous time working with all my actors on my recent film, very much including the background actors (of which I had quite a few). Those background actors were a non-trivial part of my budget, and I believe they were worth every penny because they brought themselves to their tiny roles, and those selves mattered, and mattered in ways I couldn’t have anticipated without their presence in person, on set. In their absence, we’re left with just the director’s solitary self fiddling with knobs on a machine, doing precisely what he thinks he wants, and never learning that something else was possible. The essentially collaborative and hence surprising aspect of filmmaking will, I suspect, progressively be drained away in the brave new world aborning, and we’re going to feel that loss in ways that we can’t yet fully comprehend.
But feel it we will. I wish I saw a good way to prevent it, but I don’t. I’ve loved the theater for too long, and seen what has happened in that medium. Baumol’s Cost Disease is a brutal taskmaster, and as the rest of filmmaking gets more and more efficient the cost of humans goes up and up and up, until it starts to look like a luxury to use them, in the writer’s room or in front of the camera. And when it starts to look like a luxury, we’ll cut them, because the alternative is rising prices and a shrinking market. You and I will make that happen by our choices in the market, and then we’ll sit and amidst the ruins we have made and cry over what we have wrought.
Gideon's Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.