What Is Hate Speech? Elon Musk, Experts Have Different Ideas

Elon Musk has claimed that "hate speech" on Twitter has taken a huge dive since he bought the social media platform, experts told Newsweek that his data doesn't prove anything.

The Twitter owner shared a graph from an unspecified date in October to November which marked that "hate speech impressions" were lower in that timeframe.

According to the graphic, hate speech impressions spiked at over 10 million in the days after October 20, and dropped to their lowest at 2.5 million just before November 22. They rose a little again after that date with Musk claiming it was only 1,500 Twitter accounts responsible for the impressions.

"Hate speech impressions down by 1/3 from pre-spike levels. Congrats to Twitter team!" Musk tweeted alongside the graph. He added in his replies: "I have half a mind to wag my finger at the 1500 accounts that caused the spike, but I shall forebear [eye-rolling emoji]."

"Reducing the max allowed tweets/day to a number below what a speed typist on meth could do was helpful," Musk later tweeted.

Conservative pundit Jordan Peterson questioned the graph and Twitter's definition of free speech, tweeting, "How do you define hate speech @elonmusk a serious question asked in good faith."

Musk replied to him: "This is simply based on the same list of terms that Twitter has used for some time, so ceteris paribus," invoking the Latin phrase meaning, "all other things being equal."

Ellen Judson from the United Kingdom-based research organization Demos told Newsweek it's "not at all clear" what Musk's data is "meant to show."

She guessed it could refer to the number of impressions or content taken down under Twitter's Hateful Conduct Policy. Those impressions are defined as the "times a user is served a Tweet in timeline or search results."

"Although mentioning 'list of terms' suggests perhaps it's the number of impressions that content containing certain terms associated with hate speech have had. On either metric—we can't ascertain if it's accurate or not," Judson said.

But whether Musk's data was accurate was beside the point because it seemed to just be looking at numbers, "without any context, on impressions of one particular kind of (unclearly defined) content over a very short period of time," she added.

The alleged drop in hate speech could have also occurred after Musk laid-off thousands of Twitter staff, potentially resulting in those monitoring this kind of data no longer being around to do their jobs, said Triantafillos Loukarelis, the chair of The Council of Europe (COE) Steering Committee on Anti-Discrimination, Diversity and Inclusion.

"As we know people, human beings are important to the process of research," Loukarelis told Newsweek, adding that technology could be fallible and humans were better able to discern data, further adding to Judson's point that context was everything.

"It doesn't show us that Twitter is becoming a systematically safer place or a healthier place, or that users' experiences have positively changed in a meaningful way."

Many groups around the world, including lawmakers and those advising them have been closely monitoring hate speech on Twitter.

The COE put forward a series of recommendation reviews earlier this year for social media to respect criminal, civil and administrative law concerning hate speech and was adopted by all 46 member states. Major states like the U.K., Sweden, and Germany, among others, are watching very closely the outcome of the reviews to help develop local legislation around hate speech and social media.

Loukarelis explained the COE relied on data from Germany regarding the number of hate speech impressions on Twitter.

Germany is currently the only country that obliges Twitter to report its hate speech statistics, which it does in a bi-annual report to the government.

"The statistics show us in the first [report this year] that 130,000 complaints were put into consideration. And less than 120,000 cases have been faced by Twitter with a specific measure," Loukarelis told Newsweek.

The data helps the COE to formulate methods to "face this phenomenon," and while it was still too early for third parties outside of Twitter to measure if Musk's recent data was accurate, Loukarelis pointed to the fact online hate speech continues to occur, "and is, of course, a big issue."

Musk acquired Twitter on October 27 for $44 billion and promised a few days later that "Twitter's commitment to brand safety is unchanged."

Twitter's hateful conduct policy prohibits targeting "others with repeated slurs, tropes or other content that intends to dehumanize, degrade or reinforce negative or harmful stereotypes about a protected category."

It also bans wishing or calling for violence on a person or group of people or threatening anyone, especially on the basis of their "race, ethnicity, national origin, caste, sexual orientation, gender, gender identity, religious affiliation, age, disability, or serious disease."

While Twitter does not seem to have a specific definition of hate speech, its rules against hateful conduct are similar to the COE's definition.

Having a clear definition of hate speech was a "paramount part of the work on detecting hate speech," along with accurate data, Loukarelis said.

"This is important because if we have to find effective policies to fight this phenomenon, we have to have a precise data," Loukarelis explained.

This would help with the rapid processing of complaints, "removing of hate speech without delay," and the efficient reporting of criminal hate speech to authorities.

"It would be much easier for a national legislature and international legislature to be clear on the indication of hate speech and how to protect people," Loukarelis said.

Along with the swift detection of hate speech, a clear definition of it would help to ensure an ongoing commitment to the freedom of expression.

"We have to find a balance between the freedom of expression and the fight against hate speech," Loukarelis explained. "Legislation is crucial to [help] find ways to create this distinction."

The COE chair brought the issue back to the human side of hate speech, which he asserted was a matter of "human rights and democracy."

"The most serious cost concerning victims of hate speech, is they tend to isolate themselves...minorities are particularly affected by hate speech," he said. "We have to be responsible when we talk about hate speech because they are the lives of millions of people who are not so lucky as we are, so we need to be very careful in handling the cases for minorities and vulnerable people."

Contrary to Musk's claims, research has found the use of hate language, including racist, homophobic, transphobic and antisemitic slurs, jumped significantly in the first few days once Musk took the reins at Twitter, compared to the 2022 average, according to the U.K.-based Center for Countering Digital Hate (CCDH).

Twitter's former head of safety and integrity Yoel Roth wrote on October 29: "Over the last 48 hours, we've seen a small number of accounts post a ton of Tweets that include slurs and other derogatory terms. To give you a sense of scale: More than 50,000 Tweets repeatedly using a particular slur came from just 300 accounts."

He also tweeted that while Twitter has a comprehensive policy to try and protect its users from abuse, it does not have a list of outlawed words or phrases.

"This DOESN'T mean we have a list of words that are always banned. Context matters. For example, our policies are written to protect reclaimed speech," Roth said in the Twitter thread.

It is not the first time Musk has claimed "hate speech" has fallen under his watch, tweeting on November 4 that "Twitter's strong commitment to content moderation remains absolutely unchanged. In fact, we have actually seen hateful speech at times this week decline *below* our prior norms, contrary to what you may read in the press."

When Musk took over Twitter he promised the "return" of free speech, even allowing some accounts previously banned from the site to return, including Donald Trump and Kanye West.

"It is important to the future of civilization to have a common digital town square," he wrote to advertisers last month. "That said, Twitter obviously cannot become a free-for-all hellscape, where anything can be said with no consequences! In addition to adhering to the laws of the land, our platform must be warm and welcoming to all."

Newsweek reached out to Twitter for comment.

Correction 11/28/22, 3:40 a.m. ET: A quote from Triantafillos Loukarelis has been changed to reflect the fact that a report on hate speech showed that 130,000 complaints were put into consideration, not 830,000.