What is it like to be a bot? — Data & Society: Points — Medium

What is it like to be a bot?

suspending our suspicious stance towards bots, in favor of possibility

“Algorithms aren’t wizardry.”

It’s been a popular statement to make in academic circles. And it’s well intentioned, I know that. In saying that algorithms aren’t magical, scholars aim to demystify a concept that both intimidates and obscures. “Algorithms aren’t something we can’t understand” — they’re something real and rule based, they’re something we can learn.

It’s an effort to ground the seemingly daunting complexity of computational processes and empower everyday people (and the experts among them) to figure out how they work. Furthermore, this critique aims to hold algorithms accountable. In thinking of algorithms as magical, we miss the very human makers who encode them with human values and ultimately prejudices. It’s a reminder that the things algorithms do are the products of the people who make them.

The project of demystification is a familiar one to critical theorists, who see their role in the academy as shining a light on the systems of oppression that operate without our knowing. Demystification is pinned on the idea that, if we just knew how things happened, they would lose their power; we’d be liberated from their spell. And maybe they’re right. That’s the first credo of magicians after all: never tell your secrets. Because, when you know how it’s done, it isn’t magic anymore. Magic requires not knowing. Because the magic isn’t really real, it only seems that way. The Wizard of Oz is only a man behind a curtain.

“What demystification uncovers is always something human” (Jane Bennett, Vibrant Matter)…

Bots are a productive site to study our connection to emergent technology, our feeling for — and our feeling with — the almost human. For so many people, myself included, a bot might as well be magic. When bots use preexisting text as input, they place our everyday world into the magic box, and then they slice it apart. Bots like the NYT Minus Context manage to be poignant (“Trust me when I say: You will not care”), amusing (“Space is kind of where it’s at”) and confusing (“trying to kiss the behind of a hummingbird in midflight”). They leave me wondering about the conversation that could have produced something either so familiar or so strange to me. The first warm day of summer always knocks me dead in the same way, everything is so fragmented and rumbling — all the conversations flying out and through open windows.

Other bots work through combination: the same sleight of hand that begins with an everyday object and, poof! produces something else all together. Examples are bots like Two Headlines and Think Piece. They work simply enough, pulling from a corpus of news headlines or sidebar categories. And the objects are familiar: Apple, the Blackhawks, LinkedIn, Democrats, millennials, gender theory. They’re the tired but familiar topics of conversation: the timely talking points that can be rattled off to any acquaintance (“_________, am I right?” “how bout them _________?”) It reminds me just how bot-like humans can be; how formulaic I am, or the news cycle, even critique, is.

Demystification relies on an inert kind of materiality that makes it possible for a thing (a bot) to be real or fake, rather than treating our interactions with bots as an instance — as a meeting. What might research look like if bots were taken seriously as social actors in sociotechnical systems? What are the experiences, practices, meanings, and lived experiences in a world where we live with bots?

Following Jane Bennett’s Vibrant Matter, this type of analysis might follow two threads of attention: 1) to the humans who are enchanted and 2) to the non-humans that produce those effects. Thinking about meeting and magic in this way provides a space in which the suspicious stance we are so familiar with is suspended in favor of what Paul Willis calls “the ethnographic imagination” — the realization that “what’s going on” isn’t as simple as a mechanism; it’s bound up in how they go on.

1. Enchanted humans

On the critical algorithm studies reading list published by Nick Seaver and Tarleton Gillespie there were five publications, on a list of 150, that focus on the way users experience or make sense of algorithms. For all our thinking about bots, we do very little thinking with the people who encounter them. Why do thousands of people follow accounts that are openly automated? (As bot maker Allison Parish pointed out in our workshop: we don’t love a teddy bear because we think it’s an actual bear.) Bots maintain their magic, despite us knowing they’re just conditionals.

Even in the absence of understanding, people develop complex relations with technology — ways of explaining, ways of using, ways of working with (and around) it. By studying these spaces we can come to better understand the “power” of bots.

Put simply: What is it that the magic does?

2. Non-humans

“We humans embrace the world with the intention of covering it. He, on the other hand, embraces the world with the intention of incorporating it” (Vilém Flusser, Vampyroteuthis Infernalis).

As Gina Neff recently observed (on Twitter, fittingly) ethnography of bots would require that bots have a worldview. Like bats see with sonar, bots see with data. Is there a uniquely bot-like subjectivity, a way it is like to be a bot? I’m not quite sure what this subjectivity might look like, but in thinking here — thinking about bots on their own terms — we may develop an understanding that opens space for possibility, rather than reduction.

There is a whimsy and surprise in the bot worldview — in the way things become new, in the way they become different, when the actor doesn’t work with the same rules we do. They understand differently because of their reliance on combination.

Put simply: How does a bot make sense? What could bot thinking do?