by Alec Nevala-Lee
“Retention,” my latest short story for Analog (July/August 2020) [on sale now], was originally written as an audio play, and before you read any further, I’d recommend that you give it a listen. Amazingly enough, it exists in two different versions, both performed by wonderful actors. One is an episode of the science fiction audio podcast The Outer Reach, which is available for free online, featuring Aparna Nancherla (Inside Amy Schumer, Master of None) and Echo Kellum (Arrow). The second is the concluding chapter in Syndromes, a new audio collection of my short fiction from Analog, which consists of thirteen stories narrated by Jonathan Todd Ross and Catherine Ho. For most of the audiobook, Ross and Ho trade off on reading duties, but they team up at the end for a new version of “Retention,” which is actually an audio performance of the print adaptation of my original radio script.
As a result, there are three subtly different variations of “Retention”—two for audio, one for the printed page—and no “definitive” version, which reflects the unusual circumstances behind its creation. I never thought about writing for audio until a few years ago, when I was approached by Nick White, a radio producer in Los Angeles who had gone to high school with my brother. Nick was developing an audio science fiction series, purely as a labor of love, and since he knew that I had published some stories in Analog, he wanted to know if I’d consider adapting one for the show. I was more than willing, but after taking a hard look at various candidates, I decided that none was particularly suited for the format. In most cases, there were just too many characters, and I couldn’t think of a plausible way to adapt them using audio alone, so I decided to write something from scratch.
My only rule was that it had to be a story that could be told entirely through dialogue and sound. When you listen to classic science fiction radio shows like Dimension X and X Minus One, you find that they rely on a few basic principles to dramatize the situation. For instance, we rarely hear more than two voices at once. Even in print, it can be hard for the reader to keep track of more than two new characters at a time, and when you don’t have any visual cues, it’s best to restrict the speakers to a number that the listener can easily follow. A scene between two characters, especially if they can be readily distinguished from each other, is immediately more engaging than one in which we have to keep track of three similar voices. Years ago, in fact, I concluded that if I were trying to adapt a story for radio, I’d start by asking if it could be structured as five two-person dialogue scenes, ideally between a man and a woman.
This is the same structure that I ended up using for “Retention,” largely as a safety net to make up for my lack of experience. Since I’m lucky enough to be married to a professional podcaster, most of what I understand about audio storytelling comes from radio journalism, in which clarity is crucial. You can’t conveniently rewind to listen to a section that seems unclear, and if you stop to figure out what you’ve just been told, you’ll miss what comes next. That’s why radio shows like This American Life are constantly telling us what to think about what we’re hearing. As Ira Glass puts it in Radio: An Illustrated Guide, ”This is the structure of every story on our program—there’s an anecdote, that is, a sequence of actions where someone says ‘this happened then this happened then this happened’—and then there’s a moment of reflection about what that sequence means, and then on to the next sequence of actions.”
Glass sums up this structure as “anecdote then reflection, over and over,” and for the sake of narrative clarity, I decided to follow an analogous set of rules for my script. It also struck me that the best way to orient the listener was to start with a readily identifiable genre of “found” audio and see what kind of story it suggested. In my earliest emails to Nick, I pitched building an episode around an emergency hotline call—an idea to which I still might return one day—or a series of diary entries from a spacecraft, which Christopher Nolan had already used as a plot point in Interstellar. In the end, I was inspired by a viral phone recording of the journalist Ryan Block trying to persuade a representative from Comcast to cancel his account, which I liked because it would establish the premise at once. We’ve all been in a similar situation, and after less than thirty seconds, we know what the story is about.
“Even in print, it can be hard for the reader to keep track of more than two new characters at a time, and when you don’t have any visual cues, it’s best to restrict the speakers to a number that the listener can easily follow. A scene between two characters, especially if they can be readily distinguished from each other, is immediately more engaging than one in which we have to keep track of three similar voices.”
The difference between “Retention” and the equally infuriating phone calls that happen every day is that this story takes place at an undefined point in the future. It wasn’t hard to imagine a scenario in which companies like Comcast have taken over every aspect of our lives, with algorithms delivering content and services based on our user profiles, until we’re both totally dependent on the system and increasingly isolated from others. I also decided that the service representative would actually be a chatbot, or an artificial intelligence designed to mimic a real person, which is one of the first ideas that would have occurred to anybody. But I wondered if it might be possible to have the customer be an artificial intelligence, too—an emergent property of the system that accidentally arose from the client’s preferences and search history. The original customer, I realized, might not even exist.
At that point, I had the germ of a story. I’d been interested in the Turing Test and conversations with artificial intelligences ever since middle school, when I read a hilarious dialogue in the book Metamagical Themas between the cognitive scientist Douglas R. Hofstadter and an entity that claimed to be a computer program named Nicolai. Two of the earliest chatbots were ELIZA, which was supposed to imitate a psychiatrist, and PARRY, a simulation of a patient with schizophrenia, which allowed for a stereotyped way of speaking that would be relatively straightforward for a program to emulate. They complemented each other in obvious ways, so it’s no surprise that someone decided to see what happened when they were matched up with each other. Here’s a short excerpt from the full conversation, which took place in 1972:
ELIZA: What else would you like to discuss?
PARRY: I am getting tired of answering the same questions over and over.
ELIZA: What resemblance do you see?
PARRY: Why do you need to know that?
ELIZA: What do you think?
PARRY: I don’t understand your motives.
ELIZA: Why don’t you understand my motives?
Under the circumstances, it isn’t surprising that neither program was at its best. Each one depended on canned phrases, repetition, and leading questions to keep the conversation going, and they both suffered from the absence of a human partner. But I’ve never forgotten the idea of two chatbots talking to each other. The final transcript of the conversation between ELIZA and PARRY runs about six pages, but in theory, it could have continued forever. (In 2017, a similarly interminable conversation between two Google Homes named Vladimir and Estragon generated a brief flurry of interest online.) It seemed to me that I could write something poignant and creepy about a customer service call that never ended because it paired off two artificial intelligences that refused to give up. In the finished story, my two characters, whom I named Lisa and Perry, end up talking for over eight hundred years.
At this point, I had a decent framework for the plot, and the rest amounted to a series of technical challenges. Fortunately, I had a lot of material. I listened to Ryan Block’s conversation with Comcast and noted down scraps of potential dialogue. Even more useful was a copy of the standard Comcast script for customer retention, complete with canned phrases—“I understand that your needs have changed and you want your services to reflect that”—that wouldn’t be out of place in ELIZA. The rest practically wrote itself, and I delivered a draft to Nick on January 8, 2016. It took another year to be cast and recorded, and I’m relieved to say that I’m pleased with how it turned out in The Outer Reach. Casting Aparna Nancherla as Lisa was an inspired touch, and I’m not surprised that she’s gone on to much greater things since the episode was released.
“Retention” gave me a taste, in a modest way, of how it feels to hand an idea over to somebody else for completion. (It was also fascinating to hear the more recent audiobook version, which took essentially the same script and wound up with a strikingly different interpretation.) I’ve spent most of my writing career thinking about issues of clarity. For a script, this isn’t an abstract goal, but a strategic element that can determine how faithfully the story is translated. Any missteps will only be magnified in the subsequent stages, so there’s a huge incentive for the writer to make the narrative as logical as possible. This is especially true when you’re providing a template for someone else to finish, but it also applies when you’re writing for ordinary readers, who are turning the story into a movie in their heads. Much to my delight, when I decided to adapt it for Analog, I found that I didn’t need to change much of anything. Different formats have different rules, but as Stephen Sondheim once wrote, it’s all in the service of clarity. And that might have been the most useful lesson of all.