Is Twitter Really Censoring Free Speech?

I write about the broad intersection of data and society.

The Twitter logo. (Leon Neal/Getty Images)

The past month has seen a flurry of high profile announcements chronicling just how all-powerful social media companies have become in their control over what we see online. From Twitter’s nonchalant reminder of its ability to ban world leaders and their posts, to Facebook’s actual deletion of a head of state, Silicon Valley has been on the move to remind the world that it and it alone decides what we are permitted to see in its walled gardens that define our modern web. As we take stock of a new year, what does 2017 teach us about what to expect in the coming year?

The power of social media companies to determine what crosses their digital borders has been in the headlines this week with two major stories: Facebook’s redesigned News Feed that deemphasizes commercial and news content in favor of friends and family and a set of undercover videos by Project Veritas that claim to show several Twitter employees openly discussing how the platform limits or deletes posts or entire user accounts.

Project Veritas is known for selectively editing its videos and the manner in which the most recent videos were filmed and edited makes it difficult to fully assess their contents and the veracity of the claims they appear to make. The broader question, however, is why such films received the attention they did in 2018. The short answer is that they address a topic that the social media platforms themselves have been immensely reluctant to discuss publicly: how they make the myriad decisions each day of who and what to delete or restrict on their platforms.

When asked for comment on several specific claims made in the videos, a Twitter spokesperson issued a statement saying “The individuals depicted in this video were speaking in a personal capacity and do not represent or speak for Twitter” and that “Twitter is committed to enforcing our rules without bias and empowering every voice on our platform, in accordance with the Twitter Rules.” However, when asked about the specific claims made in the video regarding content moderation, Twitter would provide comment only regarding “shadowbanning,” saying “Twitter does not shadowban accounts. We do take actions to downrank accounts that are abusive, and mark them accordingly so people can still to click through and see these Tweets if they so choose” and referred to its Help Center regarding “Limiting Tweet visibility."

When asked whether the company unilaterally denied the allegations of “unwritten rules” and political bias in its content review teams that determine what content is considered a violation of its rules, the company responded to several other questions, but did not issue a denial or any other comment regarding the bias statements beyond denying the existence of “shadowbanning,” nor provide further comment regarding the question of bias or reviewer composition. Unfortunately, this appears to be the standard practice today of Silicon Valley companies when confronted with questions of how they decide what is permissible speech online: simply remain silent and wait for the story to pass, rather than take the opportunity to provide their users with more detail about how the online world they call home works.

In particular, when confronted with questions of bias, whether relating to political affiliation in the US or government pressure internationally, companies have remained largely silent, refusing to provide any significant detail as to their moderation policies. Why is it that in 2018 the platforms we use to communicate with each other operate as opaque black boxes into which we have absolutely no insight or voice and simply accept that a handful of people in Silicon Valley will decide what a third of the earth’s population have the right to talk about?

When asked whether Twitter would consider releasing its full set of guides, manuals, documentation, tutorials, training materials and all other materials given to its reviewers or a justification for why it believes this material cannot be released, the company did not respond. In the past companies have claimed that releasing such material would enable bad actors to know just what they can get away with, but such arguments hold little merit in that every day bad actors post material looking to see just how close to the line they can tread without consequence.

Setting aside the specifics of its training manuals, it is also noteworthy that the companies have similarly steadfastly refused to provide aggregate demographics regarding their content review staff. Releasing basic top-level statistics as to gender, race, languages spoken, countries they hail from, self-identified political, social, religious and other affiliations and other demographics would go a long way towards refuting the bias claims that regularly surface regarding the companies’ review staff.

Understanding the composition of the review staff used by major social media platforms would help shed light on the languages and cultures that might be underrepresented and the kinds of hidden biases that can lurk unnoticed. After all, as I showed in 2016, Facebook’s News Feed was indeed extraordinarily biased, but in a way that others weren’t talking about: geographically.

Twitter also declined to respond when asked whether the company would be open to convening an external panel of academics and other experts from outside the company, providing them a large dataset of tweets and accounts it has limited, deleted or otherwise taken action on, and allowing them to produce a summary report for public distribution that would assess Twitter's accuracy and biases. In the end, even if the company felt that releasing any information regarding the aggregate demographics of its reviewers or any detail of its review process would harm its operational security, it is unclear why the company will not commit to allowing an independent external panel to assess its work. After all, if a blue ribbon panel of top scholars and data scientists from across the world were granted unrestricted access to its review materials and the actual records of what it has and has not taken action on to analyze them, it would go a long ways towards either confirming or finally refuting once and for all the myriad questions of bias that naturally occur when companies operate in strict secrecy.

Even simple questions like the percent of Twitter's accounts that are bots and how much of its content is automatically generated are complete unknowns. External groups routinely provide their own assessments, but beyond vague statements, the company has to date declined to provide firm hard numbers on just how much of its content and viewership is made of carbon rather than silicon. When asked whether the company would permit external auditing of its numbers, a spokesperson said they had nothing to add beyond the company’s previous statements.

Twitter’s opaque moderation policies also make it more difficult for the company to fight “fake news” and misinformation campaigns that leverage the anonymity of their platform. Parody accounts look and act very similar to the real accounts they satirize, but if they look too close to the real thing and aren’t clear enough about their satirical role, it can be difficult for users to tell the difference. For example, in the leadup to last year’s election, the Russian government is believed to have operated a troll account designed to look like the Tennessee State GOP. The Twitter account used the State Seal as its logo, “@TEN_GOP” as its handle and “Tennessee GOP” as its title, with its bio saying “I love God, I Love my Country.” Only later did it change its bio to “Unofficial Twitter of Tennessee Republications,” which could still easily leave unsuspecting users thinking it was operated in some fashion by or with the knowledge of the state GOP.

Despite 11 months of the real Tennessee GOP formally complaining to Twitter about the impersonating account, the company took no action, removing the account only after the company faced substantial scrutiny for the role its platform played in Russian influence campaigns. When asked for comment a spokesperson referred to its impersonation policy, which requires that such accounts “clearly stat[e] it is not affiliated with or connected to any similarly-named individuals.” In this case, the @TEN_GOP account did not include such language and in fact could easily have been confused for the real account, but Twitter did not respond to a further request for comment, including the use of the Tennessee State Seal as the account’s avatar.

This of course played out again last year with the rise of the “rogue” US Government agency accounts, from NASA to the National Park Service. As the accounts began attacking each other and even launching fundraisers, there was little for the average person to be able to know just which accounts, if any, were actually run by current or former US Government employees from the agencies they claimed to support.

Moreover, this lack of transparency can have very real consequences. When the official Twitter account of the President of the United States was briefly deactivated last fall, Twitter released precious little detail about how a single contractor could allegedly shut down the Twitter account of a head of state. In response, a Twitter spokesperson offered by email only “We won't have a comment on a former employee. We have taken a number of steps to keep an incident like this from happening again. In order to protect our internal security measures we don't have further details to share at this time.” Yet, once again we are left in the dark as to how much has really changed.

It is a remarkable turn of events that the company that once congratulated itself as “the free speech wing of the free speech party and famously informed Congress it would not stop alleged terrorists from leveraging its services has evolved to slowly and steadily distance itself from its free speech ethos. With each update of its terms of service, the company has moved a bit further towards prioritizing commercial reality over the anything-goes mentality upon which it was founded.

However, the ideal of unfettered free speech still factors prominently into the company’s public ethos. In an interview last year, the company’s Head of Global Public Policy Communications offered “I am passionate about protecting and empowering all people to freely express themselves, even when that can lead to challenging conversations, because ultimately it is only through connecting with others that we can learn, grow, and evolve. Twitter has revolutionized the way people communicate and learn from one another, and I am proud to represent a company that both has such a strong commitment to free speech and approaches this issue with such care and thoughtfulness.”

Putting this all together, is Twitter actively censoring certain political views or are Project Veritas’ videos the result of selective editing and employee bravado? We will simply never know. Yet, the fact that we will never know is the problem. Whether it is claims of conservatives being censored on Twitter, allegations of government influence on Facebook, or very real geographic biases, the platforms we communicate through are no longer neutral. Here in the United States, your cellular provider doesn’t actively monitor your phone calls and mute out the topics they don’t like. The USPS doesn’t inspect every letter for adverse political views. Your ISP doesn’t block access to sensitive topics like Tiananmen Square. In other countries such censorship is a routine and accepted part of daily life, but here in the US we have grown accustomed to neutral communications mediums that connect us to others without actively moderating what we are permitted to say. In their place, the online world that was supposed to bring us together and tear down the last bastions of censorship has instead created the greatest censorship and surveillance infrastructure the world could ever imagine.

President Donald Trump concocted a blatant lie in a tweet Saturday morning, accusing Democrats of “defending the life of Qassem Soleimani” who he called “one of the worst terrorists in history.”

There is no evidence that any Democrats have made such a defense. After Trump ally and suspected Republican pedophile Rep. Doug Collins (R-GA) made a similar assertion earlier this week, he apologized soon thereafter.

Russians Think Soleimani Was Great, and Trump’s a Big Loser

Writing on Twitter, the president asked, “Where have the Radical Left, Do Nothing Democrats gone when they have spent the last 3 days defending the life of Qassem Soleimani,” adding, “He was also looking to do big future damage! ‘Dems are “unhinged.’”

Earlier in the week, Nikki Haley, Trump’s former Ambassador to the United Nations, claimed on Fox News that Democrats were “mourning” the loss of the Iranian general. When pressed on what she meant, she doubled down.

“Mourning comes in different forms,” Haley’s office said in a statement. “It doesn’t have to be literally crying over the casket like Ayatollah [Ali] Khamenei. Leading Democrats are aggressively arguing that we would be better off if Qassem Suleimani was still alive today. That is effectively mourning his death.”

Democratic presidential candidates have been explicit about Soleimani’s legacy. “No Americans will mourn Qassem Soleimani’s passing,” Democratic presidential hopeful and former Vice President Joe Biden said. “He deserved to be brought to justice.”

Sen. Elizabeth Warren (D-Mass.) also called the Iranian general “a murderer, responsible for the deaths of thousands, including hundreds of Americans” even as she questioned the strategic reasoning behind the timing of the strike that took him out.

The mistruth aka filthy fucking lie follows a barrage of other eyebrow-raising claims by the president. On Friday, he claimed that four embassies were under threat from Iran. But, as the Washington Post reported, “that was at odds with intelligence assessments from senior officials.”

The Trump administration has been under increasing pressure to supply concrete evidence about the “imminent threat” that led them to take out Soleimani, the head of Iran’s elite Quds Force.

Bigger Fatter Politics

Sunday, January 12, 2020

Hezbollah Warns Trump And His MAGAts

Twitter Refuses to Be Transparent

Is Twitter Really Censoring Free Speech?

Treasonous President Donald Trump Concocted a Blatant and Dangerous Lie

Sites Trump Supporters Hate