I spend an extraordinary amount of time on Twitter. It's where I get my news, where I commiserate with the rest of the sane world while watching political debates on television, where at least once a week, I fangirl out about something great Seth Meyers did on his show. I have made so many friends on Twitter that it's basically like a Cheers that exists only in the digital realm (and has no booze). Twitter is its own universe, full of troll accounts, celebrity accounts, and bots, which are accounts programmed to do their own thing without a human being, which is basically what they told us the future would be like. Humans hanging out with bots, living in perfect harmony.
Some bots are funny, like the "Florida Man" bot that tweets out all the bizarre news that comes from America's wang; some are weird, like the bot that tweets out every color (with a swatch); and some are unintentionally existential, like the beloved Horse_ebooks.
A few weeks ago, though, I discovered a bot called the Ephemerides, which tweets out raw images from outer space taken by probes like the Voyager, accompanied by a computer-generated poem. The text of the poem is culled randomly from books on astrology and the deep ocean. The resulting tweet is almost always a beautiful and real depiction of what it's like to be a human being living in this vast unknown universe. Like, for example:
I soon found the woman responsible for my latest cyber obsession was Allison Parrish, a teacher in the Interactive Telecommunications Program at NYU Tisch (the program's tagline is "A Center for the Recently Possible," which sounds like something out of a Charlie Kaufman movie). The Ephemerides is one of several bots she's made — other favorites include Modernart_exe, which tweets the names of made-up works of art, and BrainTendencies, which tweets out "common and pernicious randomly generated cognitive biases that prevent YOU from making rational decisions." She also created a cool board game called Rewordable, where you make words out of syllables instead of letters, for when you're ready to take your Scrabble obsession to the next level. Talking on the phone with Allison opened my mind to a whole other world of art and a new mode of thinking about language and technology and the future.
Laia Garcia: What came first for you, an interest in technology or an interest in language?
Allison Parrish: I think they both happened so early that I can't really distinguish between them. I've been doing computer programming since I was a little kid. I got a computer for Christmas when I was five, and I started doing computer programming then. I think I knew I was in love with language when I read The Hobbit, in like fourth grade, because Tolkien does all this wonderful weird stuff with language, and that's sort of the thing that set me off down that road.
LG: What was the first Twitter bot you built, and how did that idea come about for you?
AP: The first one that I built was Everyword [a bot that tweeted every word in the English language], back in 2007. It was probably one of the first Twitter bots ever. People were still investigating what we could do with this API (or application programming interface) that Twitter bots have.
At the time, I was a graduate student where I teach now, at ITP and I was taking a class on how to make computers do weird stuff with text. They were talking about a project called Every Icon, which is this project that's very quickly going through every possible [design that can be made on a square grid that is] 32 by 32 pixels, [using just] black or white pixels. I had this cheeky idea: Well, if someone can make every "icon," I'm going to say every word on Twitter, and it went on for seven years. It's really more of this cheeky grad-school project. What is the simplest thing I can do to get a laugh out of somebody?
LG: You've since made many other bots, but the one that made me reach out to you is the Ephemerides. It's poetic, which is obviously part of your intent, but whereas other bots are usually funny, this one is very sad and poignant. How much of that attitude or feeling is part of the code that you write, or is it just a result of the randomness of it?
AP: Yeah, that's a good question and I don't know. Satire is one of the things that computer-generated text and other computer-generated stuff does well, because satire at some point is just mimicry. Like, the joke is that that genre of art or that social practice is just so predictable that I can make the computer program do it.
There's a resistance to the lyrical when it comes to highly formal work — in any genre, but especially in poetry. When you think of Kenneth Goldsmith, for example, or Dada poetry, where it's really about process rather than the output, you usually don't associate that with something that can feel lyrical, that can feel like it's personal, that feels like it's emotional. You're supposed to just appreciate that process just because it's clever and not because it's beautiful.
I don't think that's the limit on the emotional palette of computer-generated stuff, and that's what my art practice and my research has been about for the past couple of years, thinking about the other things that computer-generated text, specifically, is good at. Is it possible for computer-generated stuff to evoke these other emotions, to get away from cleverness and satire? Not because I don't like cleverness and satire, but because I'm interested in making those things available to people who work in the genre to see what can be done with those things. I think one of the things that computer-generated artwork is really good at is vertigo in the face of the infinite.
LG: "Vertigo in the face of the infinite." That's amazing.
AP: You can see that a lot of computer-generated art has this as a theme. I don't know if you're familiar with No Man's Sky, for example. It's a video game [that] basically generates an entire galaxy for you. You can land on every planet, there's alien species of animals on every planet, and the idea is that it just feels vast. That's kind of what I was trying to do with the Ephemerides.
[The emotional aspect] was an intentional part of what I was trying to do. But of course a lot of that comes from the source text that I used. Obviously if I did the same thing with Charles Dickens and Dr. Seuss or something, the poems would look very different.
LG: Was there a lot of fine-tuning once you built it and had it create the first poem?
AP: There's always tweaking in that process. I think with this particular project a lot of the tweaking was with the way it was treating language. The bot is not always perfectly grammatical, but that's OK. I was trying to minimize that too, to make it feel a bit more seamless. But ultimately I'm OK with it feeling glitchy. It brings out the texture of the poems.
This interview has been condensed and edited.
Laia Garcia is the deputy editor at Lenny.