The secret list of Twitter VIPs getting boosted over everyone else
Congratulations to Ben Shapiro, AOC, and ... LeBron?
Last week, Twitter began notifying users who were verified under the company’s previous regime that their blue check marks will be taken away unless they become paid subscribers for $8 a month. It’s about “treating everyone equally,” CEO Elon Musk tweeted in response to a critical William Shatner. “There shouldn’t be a different standard for celebrities imo.”
But Twitter does have a different standard for celebrities – including Musk himself. For months, the platform has maintained a list of around 35 VIP users whose accounts it monitors and offers increased visibility alongside Elon Musk, according to documents obtained by Platformer. The list, which spans the political gamut and also includes several journalists and celebrities, includes:
NBA All-Star LeBron James
Daily Wire founder and conservative commentator Ben Shapiro
Pseudonymous conservative commentator @catturd2
Rep. Alexandria Ocasio-Cortez, D-NY
President Joe Biden
YouTube star MrBeast
Venture capitalist and Twitter investor Marc Andreessen
Weird Twitter pioneer @dril
Comedian Jaboukie Young-White
Tesla community account @teslaownerssv
Journalists Matt Yglesias, Glenn Greenwald, Noah Smith, and Adrian Wojnarowski
(Platformer is not publishing the full list, whose makeup has changed slightly over time, to protect our sources’ identities. All the names above are still on the list.)
Menswear writer Derek Guy, aka @dieworkwear, is also on the list, which could explain why many of his posts went inexplicably viral earlier this year.
The list was originally created to monitor the engagement received by Twitter power users — to understand how changes to recommendation algorithms and other systems might affect their visibility on the platform. Conservative users have long complained that Twitter reduced their prominence in search results, which they often likened to “shadow banning,” and that their tweets get fewer impressions than their follower counts suggest they should.
In December, Musk noticed that his own engagement on the platform was dropping, and worried Twitter’s ranking systems could be suppressing his tweets. In response to his questions, an engineer created a list of a few dozen VIPs, with the goal of monitoring how changes to Twitter’s ranking systems affected their accounts. If the engineers identified segments of the ranking system where tweets from these accounts were “dropping,” they could tweak the code to ensure that tweets from those accounts were always shown. (The person responsible for creating the list declined to comment, but Platformer spoke to multiple employees with direct knowledge of the project).
Results from the project did not satisfy Musk. In February, the engineering team rolled out a change that boosted the visibility of Musk’s tweets over everyone else’s. But the team also allowed the list of VIPs to bypass heuristics that might otherwise limit how often their tweets are shown on the For You tab.
This is not the same as the boost that Musk’s account received. But it has the practical effect of ensuring that tweets from these accounts are more widely seen than other users’. For example, the core Twitter ranking algorithm is designed not to recommend too many tweets from one account. Being on the VIP list exempts those accounts from that rule, ensuring that more people see what they have posted.
These changes would seem to fly in the face of Musk’s stated goal to eliminate Twitter’s “lords and peasants” system, which elevated some users over others in a different way — manually verifying some notable accounts to reduce the risk of impersonation. Twitter engineers say the chief executive repeatedly flags the complaints of high profile users over those of average ones, effectively forcing them to act as a highly paid customer support team for Musk and his most favored users. (Not that every user on the list is a friend of Musk’s — he has criticized AOC in the past, for example.)
II.
On Friday, in a bid to make the platform more transparent, Musk has said he will publish Twitter’s recommendation algorithm. Whether that happens is a big “if” – Musk previously promised to open source the algorithm in February.
If he follows through, though, the code that’s released is unlikely to provide a complete picture of why people see certain tweets, company sources say. “If and when he shares the code, he’s going to choose a very specific part of it,” says one person staffed on the project.
In part, that’s because Twitter’s recommendation algorithm is constantly changing. When you log onto the platform, the algorithm looks at a variety of signals, including your previous likes and retweets and your follow graph (who you follow, and who follows you).
From there, the system chooses which tweets to show you. Typically, this includes a mix of tweets from accounts you follow, tweets from other accounts the system predicts that you will find engaging, and ads. Twitter calls the process of sourcing tweets this way “candidate generation.”
Then the recommendation system applies a series of filters. Among other things, the filters are meant to ensure that the same tweet isn’t shown twice, that it isn’t from someone you have blocked, and that it doesn’t violate the company’s community standards.
Initially, Twitter might be considering thousands of possible tweet “candidates” to show you. The next step is to narrow down the candidates into a list of about 50 that will show up in a user’s feed. Twitter does this by applying a score to each tweet. That score dictates if a tweet is shown, and if so where it shows up in the list.
This step is critically important in determining what you see – and it’s part of the reason employees worry the code that’s released won’t provide true transparency. As CEO, Elon Musk has control over the final score that tweets receive. He has recently manipulated these scores to ensure his tweets are boosted above everyone else's. More recently, his engineers have applied similar filtering to the list of VIP users.
Given that reality, sources said, it might not matter that much how Twitter’s recommendation code is written. If a secret list of users are given preferential treatment, what’s the point of releasing the ranking algorithm at all?
III.
Over the weekend, the New York Times broke the story that parts of Twitter’s source code were leaked online; the platform is subpoenaing GitHub to get information on the person who leaked it. Twitter suspects that the perpetrator was a former employee who was laid off in early November, multiple sources told Platformer.
While that code is no longer available online, an archived version of the GitHub page shows the folders that were released in the leak. They point to someone who had access to Twitter’s backend infrastructure.
While some people have pointed out – perhaps in jest – that Musk was already going to open source the algorithm, the code that was released does not appear to be related to how Twitter generates tweet candidates. It does contain information that could allow hackers to trick Twitter into believing the hacker was a different user, or deanonymize accounts – both serious security risks, assuming that the code falls into the wrong hands.
The Federal Trade Commission has already opened at least data privacy investigation into the company; depending on what happens with the code leak, it may soon have something else to look into.
Other news and notes from around Twitter:
In an effort to save money, Twitter is scaling back its content moderation team even further, and relying more heavily on automated systems to police content than ever before. Boring Company CEO, who has been on loan to Twitter, Steve Davis is heading up the project.
Twitter’s payroll team sent some former employees an email noting that the W2s they received may have been incorrect. “During a recent audit, we discovered that the ESPP (Employee Stock Purchase Plan) disposition that occurred in October 2022 was not recorded,” the company said. “As a result, employees who purchased shares through our ESPP program and sold those shares in October 2022 may have received incorrect W-2 forms for the 2022 tax year.” A former employee tells Platformer: “This is what happens when you fire most of your payroll staff…”
We broke the news over the weekend that Musk made good on his promise to give Twitter employees new compensation packages, including stock grants. Those grants are based on a $20 billion valuation, Musk said. That’s roughly half of what Twitter was valued at when Musk bought the company. The CEO says he sees a “clear but difficult path” to getting Twitter valued at $250 billion.
Talk about this edition with us in Discord: This link will get you in for the next week.
Governing
President Biden signed an executive order restricting the use of commercial hacking tools by federal agencies, as officials say at least 50 US personnel were compromised by commercial spyware overseas. (Dustin Volz / Wall Street Journal)
A TikTok ban would have to satisfy First Amendment scrutiny to survive a constitutional challenge, this piece argues. (Jameel Jaffer / New York Times)
A TikTok ban would be a blow to Hollywood, forcing movie studios, record labels, casting directors, agents, and actors to radically shift the way they do business. (Taylor Lorenz / Washington Post)
TikTok temporarily banned Enes Kanter Freedom, the former NBA player known for his outspoken political activism against China. (Drew Harwell / Washington Post)
France banned TikTok and “all recreational apps” from government-owned devices, citing security concerns. (Simon Sharwood / The Register)
And here’s a fun chaser: France is moving forward with its plan to roll out an AI-powered surveillance system ahead of the 2024 Paris Summer Olympics. (Laura Kayali / Politico)
Millions of Americans are downloading Chinese-designed VPNs that pose a greater privacy risk than TikTok, with no accompanying outcry from lawmakers. (Joseph Menn / Washington Post)
Utah’s sweeping new social media law could limit the use of TikTok and Instagram by kids. Civil liberties experts say it comes with worrying privacy implications. (Natasha Singer / New York Times)
The UK’s antitrust watchdog narrowed its probe of Microsoft acquisition of Activision Blizzard, saying new evidence “provisionally alleviates” concerns in relation to the supply of video game consoles in the UK. (Natasha Lomas / TechCrunch)
Twitter blocked 122 accounts belonging to journalists, authors, and politicians in India in response to requests from the Indian government. Twitter used to fight requests like this. (Samriddhi Sakunia / Rest of World)
In Germany, authorities are using an AI system to crack down on porn on Twitter, TikTok, and YouTube, and creators can face criminal proceedings. (Matt Burgess / Wired)
Researchers identified potential malware in some versions of the Chinese shopping app Pinduoduo, days after Google suspended it from its app store. Bad!! (Sarah Zheng / Bloomberg)
Industry
Elon Musk tried to take over OpenAI in 2018 but was rebuffed by co-founder Sam Altman. (Reed Albergotti / Semafor)
Microsoft threatened to cut off rival search engines from licensing its search data if they don’t stop using it as the basis for their own AI chat bots. (Leah Nylen and Dina Bass / Bloomberg)
Venture capitalists are racing to land the next big AI deal and take on Microsoft and Google. (By Krystal Hu and Jeffrey Dastin / Reuters)
Apple employees have reportedly grown skeptical of the company’s augmented reality headset which is set to debut in the next few months. (Tripp Mickle and Brian X. Chen / New York Times)
Tim Cook praised Apple’s symbiotic relationship with China in his first visit to the country since the pandemic. (Joe Leahy / Financial Times)
Twitter will start taking away peoples’ verification badges unless they pay for Twitter Blue starting April 1. (Todd Spangler / Variety)
Twitter Blue has only made $11 million in mobile subscriptions since it relaunched three months ago. A truly pathetic amount. (Jagmeet Singh and Ingrid Lunden / TechCrunch)
Twitter may be working on a feature that lets users hide the blue checkmark they got by paying for Blue. Guys just try Mastodon, we’re begging you.(Mitchell Clark / The Verge)
Twitch is alienating creators, fans, and employees as it drives toward profitability. (Nathan Grayson / Washington Post)
One of Europe’s largest ammunition manufacturers is facing a roadblock to the planned expansion of its largest factory because a new data center for TikTok is using up all the spare electricity in the area. OK so TikTok is good now?(Richard Milne / Financial Times)
ProPublica says it found OBN, the mysterious scammer who gets Instagram influencers banned and extorts them for money. (Craig Silverman and Bianca Fortis / ProPublica)
New AI tools can mute and ban video game players who violate conduct rules on voice chat. (Sarah E. Needleman / Wall Street Journal)
The deepfake porn industry is already flourishing and some creators accept Visa and Mastercard. (Kat Tenbarge / NBC)
Those good tweets
For more good tweets every day, follow Casey’s Instagram stories.
Talk to us
Send us tips, comments, questions, and suggested Twitter VIPs: casey@platformer.news and zoe@platformer.news.
The secret list of Twitter VIPs getting boosted over everyone else
That ProPublica article on exploiting Meta's poor technology and mismanaged customer service is crazy AF.
11 million for Twitter Blue in three months is truly a pathetic number