Meta is weighing whether ‘from the river to the sea’ is hate speech. But will it matter?
Social media companies have long struggled to effectively apply their moderation standards
The Meta logo; the company’s Oversight Board is reconsidering the moderation of the phrase “from the river to the sea.” Photo by Getty Images
By Mira Fox June 6, 2024
When Meta’s Oversight Board, which adjudicates controversial moderation decisions on the platform, asked the public to weigh in on whether “from the river to the sea” should be considered hate speech, they got thousands of responses — the second highest they’d ever received. (Trump’s temporary ban from the platform got first.)
An outsize proportion of those responses were from Jews, who see the phrase that has become emblematic of the pro-Palestinian movement, as an existential threat.
“I think they’re responding in such high numbers because of how important this issue is to them,” said Steven Terner, a geopolitical consultant who was asked to help Meta source experts on the matter.
Since Hamas attacked Israel on Oct. 7, there’s been a war on two fronts: one, of course, in Gaza, and the other online. And the debate over “from the river to the sea” has been centerstage.
Many pro-Israel activists hear the popular chant as a call to destroy the world’s only Jewish state. Meanwhile, pro-Palestinian activists see any attempt to police their advocacy as an attempt to dehumanize or censor Palestinians. And platforms such as Meta, the parent company of Facebook and Instagram, have found themselves mediating these debates.
Now, the Oversight Board, a team that reviews and refines Meta’s content moderation, is set to hear three cases about posts containing the phrase. These were all reported by various users as violating bans on hate speech, incitement to violence or supporting dangerous organizations. The posts cover a range of types — one is an open letter, one a comment and one a video — and one was seen by 8 million people, while another was only seen by around a thousand. In the initial moderation review, all were allowed to remain up.
According to a statement from the Oversight Board, Meta had reviewed its policies around the phrase after Oct. 7 but “determined that, without additional context, it cannot conclude that ‘from the river to the sea’ constitutes a call to violence or a call for exclusion of any particular group, nor that it is linked exclusively to support for Hamas.” Nevertheless, due to public outcry, the Oversight Board decided to do a more in-depth evaluation.
From which river to which sea?
In hearing the case, the Oversight Board allowed a period of public comment, during which everyday users and experts alike could weigh in; the response was overwhelming.
The Anti-Defamation League submitted a statement arguing the phrase constitutes hate speech, noting that the Hamas charter uses it, and urging the platform to censor posts using the phrase because it “has the effect of making members of the Jewish and pro-Israel community feel unsafe and ostracized.”
Meanwhile the Council on American-Islamic Relations (CAIR) submitted a statement saying that “no reasonable person” would call the phrase antisemitic, and arguing it constituted free speech and should be left up.
“Free speech principles should lead social media users to debate ideas with each other, not silence each other,” CAIR’s statement read, adding that the organization also does not support Meta policing Israelis advocating for a Jewish state from the river to the sea.
But not enough experts weighed in, so Meta hired a firm to find more, which is how Jonathan Sarna, a professor of Jewish history at Brandeis University in Boston, received the ask.
“Whenever you have expressions that have multiple potential meanings, then they all have this kind of controversy,” Sarna said he told Meta. “Given that we know that the Arabic had additional words” — Sarna here is referring to the Arabic version of the chant that translates to “from the water to the water, Palestine is Arab” — “and the truth is that even for a lot of people who use it in English, shall be free means shall be free of Jews.”
“Even though I know perfectly well that when some people use it that’s not what they mean, I think that since there’s plenty of other phrases in the world to express what you think that it’s quite reasonable that they put restrictions on it rather than find themselves complicit in hate speech,” he concluded.
The black box of moderation
Meta has historically struggled with moderation about the Israeli-Palestinian conflict, including removing posts referencing al-Aqsa mosque for supposedly supporting terrorism. Its ban on supporting “dangerous organizations and individuals” often leads it to remove educational or journalistic posts — such as a Washington Post article giving an overview of the Israeli-Palestinian conflict, that was removed for discussing Hamas as part of the area’s history — while leaving up others that are overtly hateful, such as a post calling all residents in Gaza a “savage horde.”
Both decisions were reversed after an investigation by the Oversight Board. But the Board does not see the vast majority of cases; in its first several years of existence, it heard 186 cases. To put that in context, in one quarter of 2022, users submitted 193,137 posts for review.
The examples above also seem relatively straightforward — it seems obvious that a piece of journalism explaining the history of Hamas is not the same as an endorsement. Nevertheless, the moderators’ initial decisions were incorrect. So how can Meta reliably moderate a phrase as murky as “from the river to the sea?”
Generally, Meta has difficulty consistently moderating, and the Oversight Board has noted recurring issues. Neutral educational or journalistic posts are regularly taken down for naming terrorist organizations, and in numerous decisions, the Board emphasizes the danger of removing journalism.
“This is a recurrent problem, which has been particularly frequent during the Israel-Hamas conflict,” the Board wrote in the decision reinstating a journalist’s interview with a Hamas leader. And, it added, these kinds of mistakes “significantly limit users’ free expression, the public’s access to information and impair public discourse.”
But policy recommendations made by the Oversight Board are non-binding; Meta does not have to change procedures flagged by the Board and, indeed, has often decided not to, citing the cost of implementation.
Meta’s moderation depends on a mix of algorithmic moderation and calls made quickly by teams of moderators who sort through thousands of posts; they’re often not employed directly by Meta but instead by contractors or subcontractors.
Beyond that, the process is a black box. When I asked to speak to someone about the review process for “from the river to the sea,” a Meta spokesperson sent me a brief statement that offered little elucidation: “While all of our policies are developed with safety in mind, we know they come with global challenges and we regularly seek input from experts outside Meta, including the Oversight Board.”
Missing context
While making the call to ban a video showing overt physical violence may seem simple enough, social media is also full of far more subtle decisions that require deep knowledge of the subject matter — such as the constantly evolving set of dog whistles and coded references that hate groups use. Yet the thousands of workers tasked with these decisions across the world often have to make their calls without seeing the relevant context; a moderator told The Markup earlier this year that they would often see only a single post in a set of posts.
And in contrast to Meta’s direct employees, the moderation teams are not well-paid, cushy tech jobs; reporting from The Verge in 2019 found that moderators in the U.S. made under $30,000 a year, had bathroom breaks monitored and timed, and were given little to no support with their job. In short, they are not experts prepared to make careful calls on complicated issues that require a deep understanding of Middle Eastern history or geopolitics, nor are they given time to carefully consider their choices.
Even when experts are engaged, they often operate with little information. Steven Terner, the geopolitical expert sourcing experts on “from the river to the sea,” was hired by a Meta subcontractor, not by Meta itself. And he knew little more about the project than anyone could read on the Oversight Board’s website.
“I’m a member of various expert networks for companies that take these kinds of projects on,” he told me over Zoom. “I don’t really know who the other experts are. I just know that I got an email asking if I wanted to participate in the project.”
Terner, who was asked to find 20 respondents, sent out an email to contacts he had. It quickly became clear that his email had taken on a life of its own, and spread far beyond his own network; he said he was receiving emails from people he’d never met, who had seen the call for experts in their alumni groups or through email listservs.
“After two days, I ended up with hundreds,” he said. But, he added, “they implied that the people responding weren’t diverse enough. Most of my respondents were coming from Jewish advocacy groups and clergy,” though he clarified that this wasn’t said directly — and was unintentional on his part.
This was not Terner’s first time doing a project like this; he said that since the war began, he has been doing media review and translation for social media companies — he’ll receive a set of thousands of videos that have gone viral and be asked to translate and explain them.
“It was cool, paid very well, but watching videos of — the content could be offensive, it could be nonsense, it could be violent, it could be upsetting,” Terner said. “I don’t miss it.”
Though Terner does not make decisions on which posts to leave up and which to remove, his information likely is definitive for moderators who otherwise would have no context for the videos. That’s helpful — it’s certainly better than not having any expert information — but Terner said he also feels unable to fully contextualize the videos.
“I think it’s kind of problematic to be this far cut off because sometimes you don’t really understand what you’re looking for,” he said.
Particularly with a much-debated phrase like “from the river to the sea,” context is often key. Did the post continue the phrase into “Palestine will be free,” as it usually goes? Did they pair it with a stated desire to destroy Israel, or was it a vague post about ending the war? Were they, in fact, Israeli and stating a desire for Israel to control the land between the Jordan River and the Mediterranean Sea?
Perhaps the Oversight Board will decide to set a precedent of banning the phrase, no matter the context. But that seems unlikely; nearly all of their moderation decisions seem related to context. Unfortunately, those making the actual decisions each day don’t get any.
A message from our CEO & publisher Rachel Fishman Feddersen
I hope you appreciated this article. Before you go, I’d like to ask you to please support the Forward’s award-winning, nonprofit journalism during this critical time.
We’ve set a goal to raise $260,000 by December 31. That’s an ambitious goal, but one that will give us the resources we need to invest in the high quality news, opinion, analysis and cultural coverage that isn’t available anywhere else.
If you feel inspired to make an impact, now is the time to give something back. Join us as a member at your most generous level.
— Rachel Fishman Feddersen, Publisher and CEO
