Screenwriting
Narrative Design
Live-responsive writing
In 2019, I joined the team making Artificial, a groundbreaking streaming series produced by Twitch that was a lot of things: it was half performed live (like theater), half prerecorded (like traditional TV), and fully interactive with the audience. During my time there, the show won an Emmy for Outstanding Innovation in Interactive Media, a Peabody award, and a Webby award.
This project planted the seeds in my life that would blossom into a love for all things interactive and a passion for game writing and narrative design.
The conceit of the show is that a brilliant scientist has created an artificial intelligence robot in the image of his deceased daughter. In order to help her grow into a “human,” he set up this Twitch stream so she could interact with thousands of real humans and better understand what it means to be human.
Before each season, we would break the overall arc for the show, but we’d leave lots of room for audience choices to dictate where the story went and how it resolved. In a normal show, you’re at least a few months ahead of production (network) or you’ve written all the episodes before filming starts. Because we relied so much on audience choices, we never had a finished script until the week before the show aired. It was scary and thrilling to not have that safety net of episodes built up; by necessity, we got good at writing and rewriting at blazing speeds in order to get the scripts done in time.
For big choices, we’d build in branching nodes that would play or be performed based on whatever the majority of the audience voted for. These could be as small as how will one character greet another, or as big as whether or not a character lives or dies.
The scene below shows Lilith, the AI robot, interacting with Asher for the first time. When the episode aired, the audience voted for Lilith to greet him with a joke, so the other two options were never performed.
A popular feature of the show was how we let the live Twitch audience interact directly with the characters. The actors weren’t reading Twitch chat during the show for obvious reasons. But we, the writing team, were constantly reading the chat and pulling anything we found interesting. We’d take questions and comments (even from trolls) and write dialogue in the characters’ voices, keeping all the lore in mind, and put them in a Google Doc the actors had access to. Then, the actors would perform those lines live essentially on a cold read.
It was exhilarating to pull an audience question, craft an in-character joke on the fly, and have it performed all within two minutes… and get a good audience reaction!
Season 3 of Artificial was produced during 2020, right after the beginning of the pandemic. With that context, we decided to do the entire season remotely which gave us a huge writing challenge… how do you craft dramatic, heartfelt, or funny scenes when the actors can’t physically interact with each other?
Luckily an audience choice gave us the perfect lead-in without even knowing it: at the end of Season 2 the Twitch audience chose to deactivate the AI being, Sophie, because she had killed a person. That meant we had a good reason to keep the “rebooted” AI, Lilith (named by the audience), separate from everyone else.
We also introduced two forms of interactivity that made it even harder on ourselves: audience created scenes and audience created characters.
The first character the audience created for us ended up shaping the entire arc of Season 3 when they chose to make her a private investigator. We didn’t want to just pay lip service to their choices, so we ended up structuring the entire dramatic arc of the season around her investigation of the robot and a trial for its creator.
Writing for Artificial was such a special experience that I’ll always cherish. We made something groundbreaking that was recognized by several prestigious industry awards and found a passionate audience that constantly theorized, made fan art, and talked about their love of the show in our discord. As a writer, it constantly challenged me and made my work better. From poll-based interactivity to scrambling to finish a sentence ten seconds before it was read live to an audience of 30,000, it gave me a trial by fire in writing interactive narratives and I emerged (mostly) unscathed.
If you want to see what it was like watching this show live, check out the VODs on Twitch where you can see how the audience reacted live to each moment.