Bots can be the absolute worst part of Twitter. They can also be the best. In the last few years, artists and programmers have turned Twitter bots into an internet-native art form, producing bots that are often hilarious, usually weird, and sometimes unexpectedly poetic. Some recontextualize words (@everyword) or images (@museumbot), while others leverage Twitter’s flexibility to produce commentary (@thinkpiecebot), journalism (@congressedits), or generative art (@greatartbot), sparking amusing or illuminating juxtapositions in your timeline.
Why Twitter? Its 140-character constraint is perfect for bot-making, says Darius Kazemi, a prolific bot-maker (he’s made something like 50 Twitter bots, and many more on other social-media platforms) and organizer of Bot Summit. “It’s easier to fill 140 characters with something interesting than, like, a Tumblr post, which could be thousands and thousands of words,” he explains. “There’s not a hard expectation on Tumblr of what a Tumblr post should look like and how it should be composed, whereas it’s pretty codified about how to write a tweet. Retweeting, faving, lists — all the basic functionalities that we take for granted are really great for bots.”
That combination of formal restriction and frictionlessness has led to the rise of @oliviataters, a bot that mashes up results for Twitter searches for adverbs like literally and truly to create tweets so convincing — and sometimes profound — that Twitter users interact with her as if she’s a real teenage girl, and @everyword, a linguistic experiment (2007-2014) composed of individual tweets of every single word in the English language. But there are so many other creative and useful bots out there that don’t get the attention they deserve. So we spoke to some of our favorite bot-makers about their favorite Twitter bots — their own, and made by others in the #botALLY community — to surface some lesser-known projects worth following.
@Thricedotted (@420worldclock, @wikisext): “@mothgenerator is astounding to me. Every single moth it generates is utterly beautiful and completely unique … its entire aesthetic feels so perfect to me. And the moths themselves are beautiful on their own. But to know that they are being posted by a bot with no human intervention — I find it amazing that the creators found a way to capture this aesthetic so precisely with an algorithm. It just impresses me so much.”
Created by: John Emerson
John Emerson (@FISAbot, @NYTanon): “The bot tweets images of random objects from the Museum of Modern Art collection four times a day. The bot was inspired by Darius Kazemi’s @MuseumBot, which does the same for objects in the Metropolitan Museum of Art. The range of the MoMA collection covers a wide swath of the 19th and 20th century, so it makes for a nice surprise. I always love the Russian avant garde typography. The bot’s followers really dig the mid-century modern product designs whenever those pop up.”
@Twoheadlines draws from Google News, automatically combining the subject of one headline with the action of another.
Created by: Darius Kazemi
Darius Kazemi (@sortingbot, @wowsoportland): “Part of the reason it’s funny is it’s timely — it’s always talking about what’s in the news right now because it’s pulling from Google News. The other advantage is that, much like Twitter, news headlines have a very specific way they’re written, both within publications and across publications. Like all entertainment headlines, all politics headlines, kind of all look the same. So seeing a headline about Beyoncé written the way you’d see a finance headline is very funny. Similarly, seeing a headline about Barack Obama in a way that you’d normally see a B-list celebrity written about is funny. It plays with the convention of headline-writing itself and subverts those expectations. Its hit rate is very high. Probably four or five tweets a day are very funny, which is a pretty high hit rate for a bot.”
@Nice_tips_bot shares life advice from Wikihow to brighten your day.
Created by: @Thricedotted
Dubbin: “I love it because it takes something that could be dry — scraped text from WikiHow articles — and imbues it with personality. This bot feels like a well-meaning fairy that pops into my timeline to offer earnest, sometimes urgent advice, and that tone stays consistent even if the advice is terrible or self-destructive out of its original context. I’m a huge fan of all of @thricedotted’s work, they really go that extra mile to give their bots character.”
Thricedotted: “This bot was sort of a ‘spinoff’ of @wikisext, which kicked off my interest in using content from WikiHow in a few projects. Some articles on WikiHow include “tips” at the end, and I imagined @nice_tips_bot as a well-intentioned friend who just wants to help folks with the occasional out-of-context reminder. Compared to some other bots I’ve made, it’s very simple from a technical perspective — it ultimately depends most on the content it pulls from WikiHow to remain interesting. But it still has a distinctive voice in its framing of this content, and that’s something that’s very important to me and the way that I imagine people interacting with my bots. They are not fully fleshed personalities within themselves, but if a person is willing to bridge a narrative together from the bot’s tweets, some really neat interactions can arise.”
@NYPDedits tweets anonymous Wikipedia edits made from IP addresses in the NYPD.
Created by: John Emerson
Emerson: “I’d say my most important bot is NYPDedits, though it has not tweeted since April 2015. The bot monitors Wikipedia in real time for anonymous edits from IP addresses registered to the New York City Police Department. The bot was inspired by an article by Kelly Weill in Capital New York exposing anonymous edits from NYPD IP addresses to Wikipedia pages containing details of alleged police brutality, NYPD scandals, and well-known victims of police altercations, including entries for Eric Garner, Sean Bell, and Amadou Diallo. Since I launched the bot, it caught a few edits before going silent — including Walking Dead spoilers in an article about actor Tyler James Williams. (His father is an NYPD sergeant.) Since April no new anonymous edits have appeared — though the bot is still watching. The bot has a wide range of followers, activists, journalists, and concerned citizens interested in holding police to account. It was mentioned in a couple of follow-ups to Kelly’s piece, and seems to have had an impact on police editing.”
Created by: Liam Cooke
Allison Parrish (@eventuallybot, @deepquestionbot): “It’s minimalist, almost peaceful, sometimes hilarious, sometimes sad. I think the success of bots like this one and the Ephemerides lies in the fact that people do actually love poetry (all appearances to the contrary) — we are all so hungry for linguistic serendipity and new meaning and words that behave in unexpected ways. Poetry like this activates parts of the brain often left dormant. That’s part of the power of computer-generated language, I think: It’s a bit divorced from human intention, and all the presuppositions and habits and patterns that go along with that. A good Twitter bot gives us a glimpse of language that comes from a place utterly foreign to our own conventional understanding.”
@soft_focuses is another interesting poetic experiment.
Created by: @thricedotted
Pipkin: “This is a very quiet and mysterious bot from @thricedotted. Thrice is (in my opinion) making some of the most interesting bots on Twitter, and they’ve been a huge influence on me and how people deal with language and corpora (or source material) in that space generally. @soft_focuses is a moderate outlier to the bulk of Thrice’s cannon, which might generally trend a little funnier, or louder. But reading through its output, one is struck by how non-formulaic the tweets are; no simple Ad Libs construction here. It is a technical marvel and also one of the sweetest and kindest things on Twitter. Every time it shows up in my feed, I remember to look out the window.”
@desires_exe is definitively the best 50 Shades of Grey bot.
Created by: Casey Kolderup
Kazemi: “Unconventional Desire is not very popular, but I just love it. There was this meme that was going around when the 50 Shades movie came out. It was this scene where the guy says, ‘My desires are unconventional.’ And she goes, ‘Show me.’ So the reaction shot is just something absurd with the meme. This just grabs random pictures and puts it in that reaction shot. So like, his unconventional desire was a cat in a bow tie. And it’s this steamy sexy scene and all the sudden he’s like, ‘My desires are unconventional …’ with steak and potatoes. I would not call it the best bot out there by far. I could recommend you many better, more artful, possibly even more objectively funny bots, but this is the one that makes me laugh out loud literally every day.”
@RestroomGender is a random restroom-sign generator.
Created by: Tyler Callich
Pipkin: “Perhaps this is my own gender showing, but that bot is really doing the lord’s work; every four hours or so, it produces a new abstract gender (recently trash goshs, techno acids, monohumans), pairs it with a unicode symbol and braille translation, and tweets an image of a restroom sign. The resulting timeline is pretty cute and funny in and of itself (monohumans! trash goths!) but also points at real issues with public facilities access, transness, and contemporary relationships with bodies that don’t always feel like home.”
Created by: Allison Parrish
Parrish: “One of the affordances of generative text in general is satire and humor, and I have made my share of satirical and humorous (and borderline mean-spirited) bots. In making the Ephemerides, I was trying to expand my aesthetic range a little bit, to make something a bit more lyrical and evocative. The poetry is made by remixing two 19th-century texts, one on astrology and the other on oceanography, which combined have the feeling (or are intended to have the feeling) of otherworldly poems about the exotic icy landscapes of the solar system.”
Created by: Jia Zhang
Emerson: “I think CensusAmericans is my favorite bot right now. I find it so poetic. The bot reverse-engineers census data to create tiny snapshots of people’s lives. You catch a glimpse of all this hardship and triumph, joy and sadness, life’s milestones in a flash of 140 characters. It’s a reminder how much work there is to do to repair this country’s tattered social safety net, but also of the time we have with each other, and that every day is a gift.”
@phasechase is a linguistic game taken out of the classroom and shared on Twitter, a project by Rob Dubbin.
Created by: Rob Dubbin
Dubbin: “My favorite bot of mine is probably @oliviataters, but she gets enough attention so I’m going to talk about one that I love but the world at large doesn’t care about as much, which is @phasechase. @phasechase was inspired by a game that one of my smartest elementary school teachers, Mr. Streit, used to play, where you’d start with any word and change one letter to turn it into a different word, then another letter to turn THAT into a different word, and so on. You’d always end up so far away from the word you started with! I loved that little exercise and wanted to fit it into 140 characters, and so @phasechase was born. It often reveals poetic connections between lexical neighbors, and that kind of serendipity is what Twitter bots are all about.”