As you may have heard, although Twitter has over 300 million users, it’s struggling to make money. The company recently named one of its founders, Jack Dorsey, its newest CEO, in the hopes that this leadership tweak will revive “sagging user growth” and with it, profitability.
As someone relatively new to Twitter, I’ve noticed a peculiar phenomenon that may explain, in part, why Twitter struggles to make a profit. A recent study found that Twitter bots generate 24% of its content and that 5% of Twitter accounts are responsible for 75% of Tweets.
In effect, Twitter is a place where bots prevail. And where they don’t rule, people, acting like bots, rule. This uneasy person-bot rapprochement offers a fertile space for artistic exploration.
I recently chatted via email with app artist and Twitter bot maker Nora Reed. Her Twitter bot @thinkpiecebot elegantly captures the Twitter zeitgeist. What follows is a lightly-edited transcript of our conversation.
Your Twitter bot @thinkpiecebot was recently profiled by Jacob Brogan for Slate. The article mentions that @thinkpiecebot is one of several bots you’ve built for Twitter. Of all potential media, why did you target Twitter?
I’ve made generators in other mediums too—I have a swear generator, Orcwanker, and a generator of Dialectical Behavioral Therapy Strategies. Twitter bots are really easy to build, though, and making them means that I have something amusing popping up in my feed on a regular basis. I struggle with mental health issues and having bots showing up in a place I’m already checking and making jokes for me helps remind me I’m pretty funny and great even when I’m feeling low and not able to make new stuff myself. Always having a little army of Twitter bots making them helps with that too.
So, I guess it’s a combination of Twitter bots being pretty easy to build and implement and me already being on Twitter and being pretty familiar with the culture there. When you’re building a bot that’s on a social networking service where the stuff that it generates gets pushed into peoples’ feeds, you have to think about more than the technical—you have to think of how it fits into the culture that already exists on those sites. Twitter bots have been around for so long that how to do that is pretty well established—one of the best Twitter bot makers out there, Darius Kazemi, actually wrote a nice little article about it back in 2013, and it’s still relevant on how to build polite bots.
I had a look at Darius Kazemi’s Twitter bot etiquette piece. My impression is that, as a Twitter bot connoisseur, he’s encouraging bot makers to distinguish themselves from commercially oriented bots. They fall into varying degrees of niceness, from the Hootsuites and Buffers of the world, that manage social media for businesses, to the more black hat bots out there. For example, yesterday I got an email from Twitter telling me that Christian Bale had mentioned my feed @TheMonkMGLewis. Surprise, surprise—it turned out to be a follower phishing scam.
So how would you describe the Twitter “culture”?
This is one of those things that varies a lot depending on what circles you’re in, but in the corner of Twitter I hang out in, there are some soft rules for how to behave without being a jerk. Know who your audience is and how big it is and use it with care, especially if you’re engaging with someone with less power than you, avoid reinforcing social structures of oppression and marginalization, stuff like that. It’s not all necessarily applicable to bots—especially bots like @thinkpiecebot, which doesn’t actually engage with people on its own, but it is something I consider when choosing the phrases and formulas that go into it, and I have removed things from it when the headlines that it makes with them punch down more than up—that is, enforcing structures of marginalization instead of poking fun at the status quo.
That’s culture stuff from the corner of Twitter that I’m in, though, which spends a lot of time discussing feminism and other social justice topics. There’s definitely a more general Twitter culture that expects people to behave in certain ways and sees other ones as impolite, uncouth or making you look like, to put it bluntly, a n00b. That’s stuff like when to untag people who aren’t participating from an @-reply conversation, the technical details of @-replies (that is, replying to an actual tweet and not a user so that the conversation ends up being a coherent thread), stuff like that. These are the kind of social norms that develop on pretty much all platforms, and it’s important to build bots that follow them to avoid exploiting the platform and the community that uses it. Most of the bots that do that are in the “commercially oriented” category, yeah, but that doesn’t mean everything commercial is necessarily bad, it just means that people building spambots to direct traffic to their chum portals and ad-covered sites filled with stolen pornography don’t really have a lot of concerns for the basic social norms of the platform.
When it comes to the finer points of Twitter etiquette, I have to admit I’m a total n00b. I imagine you’re alluding to a tweet @thinkpiecebot cranked out recently that contained the keyword “AIDS”. You mention in a follow-up tweet that you’ve removed the offending term from the bot’s repertoire. It brings up an interesting issue—that no bot has real agency. Agency—and responsibility for the bot’s “speech acts”—are inevitably ascribed back to the creator. In a sense, a bot can never be more than a kind of virtual prosthetic. Or can it? What’s your fascination—if I can call it that—with bots and other automatic content generators?
The tweet in question was “How AIDS Will Save Transgender Ideology”, which is the kind of thing that is totally plausible as an actual thinkpiece by the kind of bigot that the bot is making fun of, but it really failed to hit the target of making fun of the bigots themselves and moved into the “ironic transphobia” category. It felt, to be honest, like the kind of thing that comes up in a game of Cards Against Humanity played by a bunch of jerks, and I want it to be better than that. I honestly should’ve known better than to include AIDS in there at all. I’ve had similar issues with @man_products, which, because of the nature of its jokes, has a vocabulary with a lot of violent keywords in it, and I wanted to make sure that it stays absurd instead of in the “too real” category. With that one, I ended up removing some of the subject lines from the second half of the generated content so that what it came up with didn’t end up sounding like misogynist violence.
I think of my bots as more similar to fictional characters than anything else and put myself into the same category of responsibility over them as an author has over her characters. I have just as much of a duty as a bot-creator to try to avoid harm with my automatic generators as, say, a novelist does to avoid causing harm with her books.
My fascination with bots and generators is pretty simple, though: they make me laugh. I can make a bot that tickles my fancy, press a button, and have pretty good jokes come out the other side. There are other reasons, too—I’ve always loved playing with language and things like headlines and advertisements can use such repetitive phrasings that it’s really easy to imitate them or introduce absurdity or the surreal. There’s also a fair amount of frustration in the bots too, though—I’m obviously pretty annoyed with thinkpieces in general and making fun of them makes it easier to deal with that.
It’s mostly the joke thing, though.
That’s what has always drawn me to satire—that it’s both funny and subversive. You mentioned earlier that, for you, bot-making is therapeutic. In addition to the laughter your bots provoke, how so? Is it the act of creating or communicating or something else that’s therapeutic?
It’s having something creating content and constantly pushing it out when I’m not able to do so myself. I get little reminders that I can make cool stuff in my timeline in the form of bot-tweets when I’m too fatigued/tired/depressed/anxious/in pain/etc. to be making new stuff right then.
Also, they remind me to stop tweeting and drink water, as well as to take my meds.
For the technically inclined, tell us about how you make your bots. What coding tools do you use? What’s your process?
My bots are coded in Tracery and use Cheap Bots Done Quick. I use the Tracery Visual Editor to code them. Basically, they all have a list of formulas, and then they grab from word/phrase lists to fill them in.
For example, one of @thinkpiecebot’s formulas is “Could #reactthing# Have Caused #disaster#?” That one grabs from a list of things reactionaries tend to get up in arms about and a list of disasters and spits out phrases like “Could Nicki Minaj Have Caused El Niño”. @thinkpiecebot is one of my more complex bots. It has 26 formulas and over 600 different words/phrases that can be put into them, and I keep finding new stuff to add to it. Most of them are a lot simpler. My newest one, @DOTHINGSBOT, only has about 25 different phrases that it puts in different combinations.
My process mainly involves me sitting down over a day or two and working on the bot while tweeting out examples of the kind of stuff it’s coming up with—including funny ways that it breaks—and getting feedback and ideas from my followers. A friend of mine actually gave me the @DOTHINGSBOT idea and walked me through a lot of making it, but all of my bots end up with a lot of community feedback in their making, and my followers are great at giving me new words/phrases for the ones like @thinkpiecebot, which I update regularly.
Here’s a perhaps unfairly loaded question for you: what is art?
My dad’s a journalist, and he has a great little spiel he uses to talk about this where he talks about art versus craft. The metaphor he uses is coffee mugs. A coffee mug can be beautiful, but it fails to be a coffee mug if it fails to carry the coffee. He uses this to talk about journalism. An article that fails to inform its audience is, in his words, failing to “carry the coffee”, and as a craftsman, it’s his primary job to make sure that his stories do that. They might also be beautiful or profound or have additional “artistic” values, but if they fail to carry the coffee, he’s failing as a journalist. (He wrote more about this metaphor over on his blog.)
Anyway, pointing out the difference between art and craft has been the only useful way to define the word “art” for me. I think leaving art itself as a vague term ends up being useful because artists get to poke at the edges of what would be classified as “art” by most people and that makes it challenging and exciting. It also tends to make it obvious who the people being gatekeeping jerks are because they’re the ones pointing at something and claiming it “does not count as art” for one reason or another.
(I grew up in a family of artists and craftsman of various sorts, and both my parents are super into Marcel Duchamp. It probably shows.)
What’s the difference, if any, between an artist and a self-avowedly queer artist?
I think the important thing about this question is less “queer” and more “self-avowedly”. I choose to identify my queerness as influential on my work, and I want to put big signs on both myself and my work that say NOT HETEROSEXUAL and then hang out with other people who put themselves in similar categories.