Twitter CEO Jack Dorsey acknowledged that his social-media platform recently switched off an algorithm that has been connected with the practice of shadow banning—the unannounced suppression of a social-media users’ content. The practice appeared to be directed mainly toward conservatives, and in particular, supporters of President Donald Trump.
Dorsey also said the company has a long way to go to ensure its algorithms aren’t biased, according to his written testimony released on Sept. 5 before he answered questions of the House Committee on Energy and Commerce.
While he didn’t say the nixed algorithm targeted conservatives, he said it was one of the algorithms used to filter out from search results the accounts “that had a higher likelihood of being abusive.”
The so-called “quality filter” removes affected accounts from the “latest” category of search results, unless the user manually switches the filter off. The filter then snaps back on after each search.
The Epoch Times previously reviewed dozens of Twitter accounts of Trump supporters and opponents that appeared to exhibit similar patterns of behavior. Only the Trump supporters were targeted by the filter.
“Twitter decided that a higher level of precision is needed when filtering to ensure these accounts are included in ‘latest’ by default,” Dorsey said. “Twitter, therefore, turned off the algorithm.”
A Twitter spokesperson clarified that by “these accounts,” Dorsey meant “abusive accounts.”
The Epoch Times reported on Sept. 2 that the previously tested pro-Trump accounts no longer appeared to be affected by the filter starting around Aug. 31, when Dorsey announced he’d testify to the House committee as well as the Senate Intelligence Committee.
Biased Machines
Dorsey rejected the idea that Twitter filters content based on “any affiliation, philosophy, or viewpoint.” But he acknowledged that algorithms can end up being unintentionally biased.
“We recognize that even a model created without deliberate bias may nevertheless result in biased outcomes,” he said. “Bias can happen inadvertently due to many factors, such as the quality of the data used to train our models.”
He said his company is “very, very early” in its work on addressing this issue.
It will be the “engineering rigor” that will prevent Twitter employees from injecting their biases into their work, he said.
“They’re looking for fairness, they’re looking for impartiality,” he said.
However, that doesn’t appear to have previously been the case.
A Culture of Bias
A Twitter engineer, Pranay Singh, previously told undercover reporters from the Project Veritas that most algorithms designed to identify automated “bot” accounts were actually centered around conservative topics.
“You look for Trump, or America, or any of, like, five thousand keywords to describe a redneck,” he said.
Singh, however, didn’t appear to talk about an organized effort to oppose Trump, but more a disbelief that die-hard Trump supporters on Twitter could be genuine.
“Just go to a random [Trump] tweet and just look at the followers. They’ll all be like, guns, God, ‘Merica, like, and with the American flag and, like, the cross,” he told the undercover reporter. “Like, who says that? Who talks like that? It’s for sure a bot.”
Anti-Trump bias at Twitter wasn’t so much an official policy as a company culture, according to Mo Norai, a former Twitter content review agent, who spoke to an undercover reporter on May 16, 2017.
“As a company, you can’t really say it because it would make you look bad, but behind closed doors are lots of rules,” he said. “Like, ‘Hey, you gotta do this this way.’ Or something like that. It was never written, it was more said.”
Twitter employees were “probably about 90 percent anti-Trump, maybe 99 percent anti-Trump,” at the time when he worked there, Norai said.
Trump, Sessions Weigh In
After the Senate Intelligence Committee hearing, which continued on Sept. 5 with Facebook and Twitter testifying, Attorney General Jeff Sessions announced in a statement that he’ll meet with “a number of state attorneys general this month to discuss a growing concern that these [social-media] companies may be hurting competition and intentionally stifling the free exchange of ideas on their platforms.”
Trump criticized social media at a rally in Indiana on Aug. 30.
“You look at Google, Facebook, Twitter, and other social media giants and I made it clear that we as a country cannot tolerate political censorship, blacklisting, and rigged search results,” Trump said. “We will not let large corporations silence conservative voices.”
In a Sept. 4 interview with The Daily Caller, Trump said social-media giants have interfered in elections.
“I mean, the true interference in the last election was that—if you look at all, virtually all of those companies are super-liberal companies. They were in favor of Hillary Clinton, he said. “Now maybe I did a better job because I’m good with the Twitter and I’m good at social media, but the truth is they were all on Hillary Clinton side, and if you look at what was going on with Facebook and with Google and all of it, they were very much on her side.”
Ajit Pai, chairman of the Federal Communications Commission, said in a Sept. 4 blog post that tech giants need more transparency. He, however, opposed stringent regulation of online services similar to the way utilities are managed. Such a level of regulation was previously advocated for by figures like progressive billionaire George Soros.
Reuters contributed to this report.
