Meet the Twitter Employee in Charge of Making Sense of Its Trending Topics

Meet the Twitter Employee in Charge of Making Sense of Its Trending Topics

Sleuthing the origins of those strange, mangled trends on our timelines is a full-time job.

A Twitter bird is seen flying near a hashtag symbol.
Illustration by Slate

One day in late November, the city of Cedar Rapids trended on Twitter. Nothing of obvious note was happening in the Iowa county seat—it wasn’t playing host to insurgent political instability, or college basketball heroics, or any of the other triumphs or tragedies that tend to cause a flare-up on social media. Instead, the popular Twitch personality Quackity had simply quoted one of Hillary Clinton’s old Snapchat stories live on stream. “I’m just chilling in Cedar Rapids,” said Clinton in the clip, overlooking the titular Cedar River, hot on the Democratic primary trail in 2015. Quackity wields enough influence to bless anything—literally anything—with absurd virality in a matter of seconds, and thousands of his fans soon blasted the quote across their own pages, causing Twitter’s trending section to take notice. Other Twitter users desperately sifted through their feeds to uncover whatever tragedy or chaos must have clearly been underway in eastern Iowa, while the company’s moderation team quickly mobilized to assuage our frayed nerves. No, there wasn’t anything to fear, save for some bespoke gamer ephemera.

ADVERTISEMENT

“While playing the horror game Slender: The Arrival, Quackity referenced Hillary Clinton’s viral 2015 Snapchat video, in which the former Democratic presidential candidate said she was ‘chilling in Cedar Rapids,’ ” read the caption that was quickly appended underneath the trend in Twitter’s sidebar. Crisis averted.

It feels like these oft-bizarre, oft-anxiety-inducing incidents happen every day on Twitter. A non sequitur pulses through the “What’s Happening” tab, thanks to the collective posting of a subculture lurking on the platform, and the company’s content guardians attempt to decode its origins as best they can in order to keep its user base apprised. In previous iterations of the service, Twitter let the trending page speak for itself; it simply collected the platform’s locally, nationally, and globally surging topics, while offering no further clarification or contextualization for what people were tweeting about other than the original texts. But beginning in 2020, Twitter started adding short, expository paragraphs to some of these daily topics, with the hope that the average user wouldn’t feel completely bamboozled when, say, a Twitch teen hijacks the algorithm to make it seem like some celebrity has died.

ADVERTISEMENT

This seems like a fairly straightforward concept, but Twitter has grown increasingly cavernous throughout its 14 years of existence. The memes and hashtags that occupy the trending tab today are almost ouroboroslike in nature—to get the joke, you often must rely on a tight cycle of references, insider knowledge, and necrotic, extremely online brain poisoning. The “What’s Happening” tab is less a big-tent reflection of the discourse and more a distillation of the vast, unknowable variations of the human experience; it is composed of all of these pocket dimensions, each equipped with their own insulated, myopic posting norms. Here’s an example: Last month, the word platinum popped up in the trending feed. Why? Twitter described the origins in its caption this way: “The Pokemon Company announced remakes of Pokemon Diamond and Pearl, but they did not mention the fan-favorite Pokemon Platinum, a later version of the game which addressed many of the issues associated with Diamond and Pearl, like the extremely slow pace.” What a densely specific explanation behind the uptick in the use of just one word.

A few weeks earlier, the algorithm picked up on the word gatekeeping—another fairly neutral term that could be applied to pretty much anything. Twitter’s exhaustive commentary: “People are discussing gatekeeping in pop culture after a user suggested if someone’s anime knowledge is solely based on popular titles like Demon Slayer, Naruto, Inuyasha, Bleach, One Piece … then you’re NOT an anime fan!”

ADVERTISEMENT

A perverse realization washed over me when I saw that last one. Clearly, Twitter employs actual people who spend their days attempting to parse the pandemonium of the platform. They’re required to walk with the weeaboos, the gamers, the K-pop stans, the Laker bros, and the incessant scolds on behalf of those of us who aren’t familiar with their syntax. They bravely serve as the interconnective tissue between every microscopic faction of the Terminally Logged On. It sounds like simultaneously the best and worst career in the world, and I needed to know how anyone could keep it up without losing their mind.

Thankfully, Twitter was kind enough to arrange an interview with Joanna Geary, the company’s senior director of curation, and a woman named Victoria, who makes a living by typing up trending captions; she requested we not use her last name for privacy. Initially, Geary said, Twitter considered offloading these blurb-writing duties to an A.I. program. If we could feed a computer enough tweets about a specific controversy, Twitter leadership thought, perhaps it could generate a precise summary of the action. But Twitter found that the machine learning was never quite good enough. “You need the nuance and context, and it needs to be coming from a person,” says Geary. “That’s how [we started].”

ADVERTISEMENT

Twitter announced its contextualization campaign on Sept. 1 and tasked a slew of writers all over the world to add a little bit of supplementary text to the day’s news. Geary says that the team tends to ignore the hourly headlines that are overwhelmingly obvious; there doesn’t need to be a short inscription on, say, “#MondayMotivation” or something similarly self-explanatory. Users can figure out that one for themselves. But whenever something more esoteric floats to the platform surface—be it a nuclear Marjorie Taylor Greene take or the whims of a mega YouTuber—this crack team of professional explainers gets right to work.

Victoria tells me that, because of this work, she’s become exponentially more familiar with Twitter’s gradient of fandoms. The way she puts it, her responsibility is to assimilate among those who keep pushing those strange, mangled trends to our timelines. In practice, that means Victoria had to recently embark on an all-encompassing Minecraft deep dive. The massively popular block-building game, first released in 2011, is in the midst of a minor renaissance, and the names and catchphrases of individual Minecraft streamers frequently orbit to the top of the trending apparatus. Victoria resolved to embed herself in the game’s lexicon, to truly know it and speak it so she could parse its frequently appearing trending terms. There’s no educational cheat sheet in a job like this.

ADVERTISEMENT

“It was really confusing for us in the beginning,” says Victoria. “We weren’t coming from a place of, ‘I know all about these Minecraft content creators, I watch all these streams.’ ”

Victoria created a comprehensive Minecraft curriculum for herself. She started watching Twitch streams, she paged through zillions of tweets, and she pored over the amateur wikis that serve as the knowledge base for countless different fandoms. Only then, after gaining fluency, did she feel fully confident lending her descriptive talents to the furtive Twitter dramas laced between Minecraft personalities. “It wasn’t enough to know what they were talking about one day,” continued Victoria. “You needed to know all the words they were using. You needed that background knowledge.”

But that’s only half the battle. Sure, you’ve consumed a week’s worth of Minecraft videos and accrued a competent understanding of redstone, but how do you translate all of that abstruse information to a cadre of boomers who, again, just want to know what the hell is going on in Cedar Rapids? Victoria says that this challenge has completely rewired her brain.

2 Comments

  1. core i5 4690
    +3
    91
    91
    0
    < 0.01
    140
    28
    0
    www.technilesh.com/2021/07/the-ultimate-guide-to-tensorflow-learn.
    html
    Jul 01

    b4x com
    +2
    39
    39
    0
    < 0.01
    50
    36
    0
    www.technilesh.com/2021/06/which-best-platform-for-andriod.
    html
    Jun 12

    standardscaler python
    92
    92
    0
    < 0.01
    210
    41
    0
    www.technilesh.com/2021/06/machine-learning-model-2-project-for.
    html
    Jun 30

    standard scaler python
    95
    95
    0
    < 0.01
    210
    43
    0
    www.technilesh.com/2021/06/machine-learning-model-2-project-for.
    html
    Jun 30

    sklearn model_selection train_test_split
    83
    83
    0
    < 0.01
    210
    34
    0
    www.technilesh.com/2021/06/machine-learning-model-2-project-for.
    html
    Jun 30

    supervised and unsupervised learning tutorial point
    +2
    95
    95
    0
    < 0.01
    30
    34
    0
    www.technilesh.com/2021/06/the-best-machine-learning-projects-you.
    html
    Jun 28

    sklearn standard scalar
    85
    85
    0
    < 0.01
    1,000
    47
    0
    www.technilesh.com/2021/06/machine-learning-model-2-project-for.
    html
    Jun 30

    spark mllib tutorial java
    +2
    94
    94
    0
    < 0.01
    30
    26
    0
    www.technilesh.com/2021/06/the-best-machine-learning-projects-you.
    html
    Jun 28

    cisco internship

    ReplyDelete
  2. https://www.technilesh.com/2021/07/top-10-tech-events-you-cant-miss-tech.html
    https://www.technilesh.com/2021/07/the-top-10-most-innovative-tech.html
    https://www.technilesh.com/2021/07/best-ways-to-make-research-and.html
    https://www.technilesh.com/2021/07/the-top-10-internship-companies-for.html
    https://www.technilesh.com/2021/07/the-ultimate-guide-to-tensorflow-learn.html
    https://www.technilesh.com/2021/06/machine-learning-model-2-project-for.html
    https://www.technilesh.com/2021/06/10-projects-for-ai-and-machine-learning.html
    https://www.technilesh.com/2021/06/how-install-all-libraries-of-python-and.html
    https://www.technilesh.com/2021/06/the-best-machine-learning-projects-you.html
    https://www.technilesh.com/2021/06/the-future-of-artificial-intelligence.html
    https://www.technilesh.com/2021/06/realme-c-11-specification-microsoft.html
    https://www.technilesh.com/2021/06/programming-languages-which-should-you.html
    https://www.technilesh.com/2021/06/the-complete-guide-opinion-to-data.html
    https://www.technilesh.com/2021/06/today-jio-reliance-acm-conference-event.html
    https://www.technilesh.com/2021/06/machine-learning-basic-code-which-used.html
    https://www.technilesh.com/2021/06/how-find-best-master-degree-university.html
    https://www.technilesh.com/2021/06/the-future-of-automation-technology.html
    https://www.technilesh.com/2021/06/the-future-of-augmented-reality-what.html
    https://www.technilesh.com/2021/06/the-top-cloud-computing-mistakes-youre.html
    https://www.technilesh.com/2021/06/how-to-understand-object-oriented.html
    https://www.technilesh.com/2021/06/how-virtual-reality-is-changing-game-7.html
    https://www.technilesh.com/2021/06/how-to-build-iot-device-x-steps-to-get.html
    https://www.technilesh.com/2021/06/the-mobile-can-predict-search-result.html
    https://www.technilesh.com/2021/06/how-to-learn-any-skill-you-want.html
    https://www.technilesh.com/2021/06/how-computers-are-reading-your-brain.html
    https://www.technilesh.com/2021/06/choosing-platform-technology-for-your.html
    https://www.technilesh.com/2021/06/a-beginners-guide-to-coding-tips-for.html
    https://www.technilesh.com/2021/06/artificial-intelligence-what-it-is-and.html
    https://www.technilesh.com/2021/06/the-complete-guide-to-what-is.html
    https://www.technilesh.com/2021/05/What-is-Cloud-Gaming-in-2021-TechNilesh-gaming-andriod-gaming-mobile-to-how-cloud-work-tech-gadgets-now.html

    Reply

    ReplyDelete
Post a Comment
Previous Post Next Post