For Twitter bans, a 'lack of transparency' is a bigger issue than political bias, experts say

Twitter's decision to ban former Ontario teaching assistant Lindsay Shepherd from its platform earlier this week has once again sparked accusations that the company unfairly targets people with certain viewpoints. 

But the bigger issue, say observers, is not whether Twitter holds a political bias, but the confusion and secrecy surrounding its decisions to punish some users.

"It's not so much whether or not there's bias, it's that it's easy to perceive bias because of the lack of transparency and accountability on Twitter," said Fuyuki Kurasawa, an associate professor of sociology at York University.

"The broader problem is that Twitter is inconsistent and is not transparent in its decision-making processes when it comes to whom it bans," said Kurasawa, who is also director of the university's global digital citizenship lab.

Nikki Usher, an associate professor at the University of Illinois College of Media, agreed it's difficult to hold Twitter to account.

"We don't really know what they're doing because all we can see is what they say their established standards are. But we don't have any transparency about how they're actually going through and moderating these comments."

Shepherd, who made headlines after she was disciplined by Wilfrid Laurier University for showing a clip to her students of controversial professor Jordan Peterson discussing gender-neutral pronouns, said she received a notice from Twitter that her account had been suspended for violating rules against hateful conduct.

She immediately appealed the suspension, and shortly after received a notice from Twitter that her account "will not be restored," effectively banning her for life.

Although Twitter wouldn't tell Shepherd how she violated their hateful conduct rules, the ban came soon after a tweet spat she had with Jessica Yaniv, a transgender activist.

Yaniv had tweeted some ugly remarks about Shepherd's genitalia, Shepherd told CBC News. At one point, Yaniv tweeted: "I heard @realDonaldTrump is building a wall inside of your uterus aka your "reproductive abnormality" hopefully the wall works as intended."

Shepherd believes that was a reference to her septate uterus, an anomaly in the body that can lead to an increased rate of pregnancy loss. 

Shepherd tweeted that "at least I have a uterus, you ugly fat man" and also tweeted: "This is how men who don't have functional romantic relationships speak. But I guess that's kinda what you are."

Who reviews the guidelines? 

Shepherd said the fact that Yaniv still has an account is a "textbook case" of a double-standard employed by Twitter, that those within the transgender community are part of a "protected class" and can remain on the platform regardless of what they say.

"Yes I misgendered which is against the Twitter Rules. OK, fine," Shepherd said. "But the thing is, what's also against the Twitter Rules is sexist insults. So technically [Yaniv] should be gone too."

Kurasawa suggested this case exemplifies the confusion around Twitter's rules. While Yaniv's tweets about Shepherd were "vile," he said personal insults, even about someone's medical issues, are permissible, according to the standards set by the platform.

The Twitter Rules, published on the company's website, outline the reasons an account may be suspended. These include threats of violence, targeted harassment, hateful conduct and posting graphic and sexual violence.

"The issue is not so much the guidelines," said Kurasawa. "The issue is who are the people at Twitter who are making these decisions, interpreting those guidelines."

When a person is suspended or banned, little explanation is given, he said, meaning people don't understand the basis for Twitter's decisions and how the company has interpreted its own rules. 

At the TED2019 conference in Vancouver, Twitter CEO Jack Dorsey would only say that when it comes to bans, Twitter focusses on conduct versus content, CBC's Alex Migdal reported.

Twitter Canada wouldn't comment on individual cases like Shepherd's and referred CBC News to its Twitter Rules guidelines. 

Accusation of bias

Shepherd's complaint about a double standard is often repeated by commentators on the political right, who have accused Twitter of bias against conservatives.

Earlier this year, Republican lawmaker Devin Nunes sued three Twitter users and Twitter itself for defamation. In his lawsuit, Nunes claimed that Twitter "actively censors" conservatives.

Richard Hanania, a social scientist and research fellow at Columbia University's Saltzman Institute of War & Peace Studies, analyzed these claims.

Hanania compiled a list of prominent people who had been banned from Twitter. He found that of the 22 people banned, 21 had supported Donald Trump. 

Acknowledging the small sample size, and that some of those who were disciplined were not exactly "innocent angels," Hanania still thought the results were clear: Twitter is biased against conservatives.

"Even if you assume conservatives are three times more likely to violate Twitter's terms of service ... you're still very, very unlikely to get a result that skewed even with the small sample size," he told CBC News.

Twitter's CEO Dorsey has rejected such accusations of bias. "Twitter does not use political ideology to make any decisions, whether related to ranking content on our service or how we enforce our rules," he testified before the U.S. Congress last year.

"We believe strongly in being impartial, and we strive to enforce our rules impartially."

Usher said not just conservatives, but left-wing activists and marginalized groups, will complain that their speech is being compromised or banned by Twitter. She said if people could see the entire list of banned accounts, then possibly they could determine if there are some partisan trends.

Cherry pick

"We don't have a really good sense of who's getting banned and who's not getting banned because all we have is the data we can cherry pick," Usher said.

That's why she believes Hanania's study has some significant methodological problems, including an "extraordinarily cherry picked sample."

"We don't know the context for these bans. It's a really limited selection of people," she said. 

"Just looking to see if there is equality in who's getting banned doesn't tell you much about the content of the speech itself."

Meanwhile, Kurasawa said that at the very least, in high profile cases, Twitter should publicly explain their decision making processes. Yet he understands why they'd be unwilling to do so.

"The inconsistencies with which they enforce their own regulations ... would probably be exposed," he said. "It wouldn't be favourable to Twitter to do that and therefore it would be probably undermining of corporate image."