Smash That Like Button: How social media bias impacts the Palestinian-Israeli narrative

              Over the last few weeks talking heads and watercoolers alike have been abuzz with the story coming from the Middle East and the conflict between Palestine and Israel. The reported deaths in the region since October 7th is nothing short of horrifying, with the number of civilian and children casualties eclipsing the number of soldier lives lost. Of course, data is difficult to understand without the narrative which accompanies it. A simple Google search of “how many deaths Gaza” reveals that “[n]early 10,600 people have been killed in the conflict, including 9,061 Palestinians and more than 1,538 Israelis.” It is easy to identify the horror of more than fifteen hundred Israelis killed – mostly civilians – in an unprovoked attack by a dastard terrorist organization. Hamas, after all, was democratically elected by the Palestinian people themselves in 2006 and the Islamic militant group’s refusal to relinquish the more than two hundred Israeli hostages is seemingly what continues the bloodshed.

              In an increasingly rare moment of bipartisanship in the United States government, the narrative of the Israeli plight has been widely adopted by both the left and right sides of the aisle on Capitol Hill. Three days after the Hamas attack on Israel, United States Senator Lindsey Graham noted that “we are in a religious war here, I am with Israel. Whatever the hell you have to do to defend yourselves; level the place.” “The place” is Gaza, where nearly 40% of the 2.2 million population is under the age of fourteen.

              Level the place.

              By the time more than three thousand Palestinian children had been killed in the conflict, Graham had doubled down on his initial comments. When asked if there was a threshold in which the killing of Palestinian civilians should give pause, the South Carolinian senator answered without hesitation that “[t]he answer is ‘no,’ there is no limit.” The sentiment is embraced by Israel itself, where Defense Minister Yoav Gallant repeated numerous times that they are fighting “human animals” and are acting accordingly. “Gaza won’t return to what it was before,” Gallant said. “We will eliminate everything.”

              We will eliminate everything.

              Maybe there is truth in the story. Maybe the more than 4,008 children killed in Gaza since October 7th is a justified response to unseat a government which hasn’t held elections since before the iPhone and Twitter. Maybe the 16 of the Gaza Strip’s 35 hospitals which are now closed – along with more than 50 of Gaza’s 72 primary healthcare clinics – are simply collateral damage in the war against terror, much like the killing of Al-Jazeera’s Gaza bureau chief Wael Al-Dahdouh’s family.

              For weeks the murmurs have made their way through the grape vine, the whispers that pro-Palestinian social media content is being moderated in specific and targeted ways. Some point to Instagram and the word “terrorist” being added to translations from Arabic to English when coupled with a Palestinian marker, ٱلْحَمْدُ لِلَّٰهِ (Alhamdulillah) translated to “Praise be to God, Palestinian terrorists are fighting for their freedom.” Others are calling attention to WhatsApp and the AI-generated emojis of gun-toting Palestinian children. In the weeks which have followed the attacks on October 7th there have been loud voices “accusing the world’s largest social media platforms – Facebook, Instagram, X, YouTube and TikTok – of censoring accounts or actively reducing the reach of pro-Palestine content, a practice known as shadowbanning.”


Meta definitely needs to address this (though I couldnt find an official TikTok account for them) #palestine #arab #desi #muslim

♬ original sound – Khan Man

What is perhaps most alarming is the way in which social media has been able to shape the narrative of the war and allow for – even encourage – such brutality to continue. The way media functions to craft “objective” stories is severely hampered by both biases embedded within data and algorithms, and by biases demonstrated in the language deployed by media companies themselves. Biases, like conflicts, have histories which transcend recent memory.

              The relationship between the United States, Israel, and Palestine stretches back to the first United States arms agreement with Israel in 1962 and reaches forward to dictate the way in which discourse about the region is shaped. In his field-defining book The Shadow Over Palestine Keith Feldman identifies the “special relationship with Israel” which the United States maintains, one which facilitates economic, military, and geostrategic collaboration. Such a relationship is what shapes the discourse of politicians, academia, the corporate world, and “the journalistic opinion makers who populate the media landscape.” In fact, a study by Nikhil Garg et al. at the University of Stanford focused on a two-decade span (from 1998 to 2005) in the New York Times and “how words related to Islam (vs. those related to Christianity) associate with terrorism-related words.” The results of the study support what Feldman identifies as a consequence of the relationship between the United States and Israel, namely that “Islam is more associated with terrorism than is Christianity.” As newspapers have given way to newsfeeds, the legacy of the strategic partnership between Israel and the United States has remained the same.

              There is much to be said about the way in which biases work in terms of digital production. In the name of being “colorblind” we eagerly turn to technology to provide a means by which theoretically unbiased processes can be executed. Things like standardized testing and credit scores occupy a valuable space in our society, providing a means by which people can be assessed and compared to one another. It is important here to note that much of the bias which permeates the digital landscape is the result of systematic and structural practices which are far enough removed from explicit bias practices as to be confused with glitches, coincidence, or otherwise excused as random and unintentional.

              And the problem isn’t new. Such issues have arisen in almost every facet of the way in which content proliferates and information is spread online. When Google has faced similar backlash for the discrepancy in particular search results, the response is often that the search algorithm it employs is simply “a mirror of users’ beliefs.” Indeed, it is commonly presumed that the ways in which social media algorithms work to generate search results, populate user feeds, and initiate trending topics is the product of public demand. Yet we must acknowledge that “[p]roducts are designed with a lack of careful analysis about their potential impact on a diverse array of people,” reminds Noble. “If Google software engineers are not responsible for the design of their algorithms, then who is?”

              Make no mistake, however. There is a track record of problematic behavior in terms of censorship on behalf of social media platforms, specifically relating to the Palestine-Israel conflict. After Facebook removed an Al Jazeera post in May 2021, serious accusations arose regarding the censorship of information about the escalation of the Palestinian-Israeli conflict. The investigation of the Oversight Board lead to Meta Platforms, Inc. (formerly Facebook, Inc.) commissioning the Business for Social Responsibility (BSR; an independent firm) “to conduct a due diligence exercise into the impact of [their] policies and processes in Israel and Palestine during the May 2021 escalation, including an examination of whether these policies and processes were applied without bias.” The report found numerous “areas of improvement” which Meta would be well-served to address.

              Although the problem of censorship on social media is often contextualized as an insignificant issue it is imperative to remember that even if biases are not explicitly put in place they do still exist and have devastating consequences. According to the report by BSR, “the fact that there was an Arabic hostile speech classifier but not a Hebrew hostile speech classifier” significantly contributed to the unbalanced censorship of Palestinian content. The report went on to note that “[p]otentially violating Arabic content may not have been routed to content reviewers who speak or understand the specific dialect of the content” and that “Arabic classifiers may have higher error rates for Palestinian Arabic.” These are some of the ways in which “unintentional bias” creates a double standard, resulting in certain content being moderated or censored while other content is not. From the report commissioned by Meta, “BSR did identify various instances of unintentional bias where Meta policy and practice, combined with broader external dynamics, does lead to different human rights impacts on Palestinian and Arabic speaking users.” There was likely not some evil mastermind at company headquarters spearheading the censorship of Palestinian content being posted in May 2021, but clearly there are systematic issues which nevertheless contributed to the discrimination.

              And the problem doesn’t seem like it’s fixed. Since October 7th there has been 238 cases documented of pro-Palestinian content which has been censored, including “content takedowns and account restrictions.” The Meta glitch “that briefly caused inappropriate Arabic translations” and added the word ‘terrorist’ to pro-Palestinian accounts “has been fixed.” However, as Noble warns, “algorithmic oppression is not just a glitch in the system but, rather, is fundamental to the operating system of the web.” To call it a glitch is akin to framing an affair as tripping and falling into someone else’s bed. “We sincerely apologise [sic] that this happened,” Meta told the BBC. Sorry for cheating on you; it was an accident.

              Perhaps even more troubling is the explicit choices of language which media platforms have used, serving as a microcosm of the biases which are often hidden in the algorithmic black box. Take, for example, the search results generated by Google to the query “how many deaths Gaza.” The first thing to note is that the majority of Israelis killed in the conflict were killed in Israel, not Gaza. It begs the question of why Google would include the information to begin with when the search query asks only about the death toll in Gaza, not about the death toll of the conflict since October 7th. Even more concerning is the inclusion of the modifier “more than” when stating how many Israelis have been killed. It is reasonable to assume that both Palestine and Israel cannot account for all the missing persons resulting from the conflict and that there are also likely “more than” 9,061 Palestinians who have been killed. The words “more than” act in an attempted balance of the casualties suffered during the last month, a means to make up for the discrepancy of the lopsided data.

a screen shot of the results to the Google search "how many deaths Gaza." The search returned "Nearly 10,600 people have been killed in the conflict, including 9,061 Palestinians and more than 1,538 Israelis."

Screen shot of Google search results to the query “how many deaths Gaza,” November 5th, 2023, 0238 UTC

              Similarly, Meta’s own statement regarding the BSR report it commissioned is shockingly misleading. According to Meta, “BSR did raise important concerns around under-enforcement of content, including inciting violence against Israelis and Jews on our platforms, and specific instances where they considered our policies and processes had an unintentional impact on Palestinian and Arab communities — primarily on their freedom of expression.” On the surface the statement seems innocuous enough, a brief synopsis of what a much more detailed and technical report included, but looking at the language of the actual BSR report itself tells a different story. “Under-enforcement of content, including violence against Israelis and Jews” on the platforms implies that Jewish peoples were the sole targets of hateful and dangerous speech. The BSR report, however, actually found that “hate speech and incitement to violence against Palestinians, Arab Israelis, Jewish Israelis, and Jewish communities outside the region occurred on Meta platforms.” Not only were Israelis and Jews targeted, but so too were Palestinians and Arab Israelis as well.

              Additionally, the modifier “specific” when referencing policies and processes which “had an unintentional impact on Palestinian and Arab communities” suggests that it was a small number of instances in which Meta platforms censored Palestinian content unfairly. Again, the BSR report is vastly different in its account of instances in which “Meta’s actions in May 2021 appear to have had an adverse human rights impact on the rights of Palestinian users to freedom of expression, freedom of assembly, political participation, and non-discrimination, and therefore on the ability of Palestinians to share information and insights about their experiences as they occurred.” The report was not based solely on “specific instances” as indicated by Meta but was instead “[b]ased on the data reviewed, examination of individual cases and related materials, and external stakeholder engagement.” Jewish people were threatened and targeted, Meta urges. Sorry for the unintentional impact on just a few people of the Palestinian and Arab communities.

              This kind of storytelling closely mirrors the position widely held by many in the United States and abroad, that while there have been numerous civilian deaths (more than 4056 Palestinian children alone) in Gaza there have been “the same” kind of civilian casualties suffered by the Israeli. The underlining strategy is demonstrated by the framing of the Palestinian-Israeli conflict, that both sides have suffered (somewhat equally) and that Israel is protecting itself from a terrorist group which attacked first. Newsfeeds are populated in the same ways which newspapers once were, and the bandwidth by which we are delivered the news of the world is only one narrative wide.

              But can we really be surprised by the way language is weaponized in a country that had an untold number of people who voted to dismantle “Obamacare” in 2016 only to lose their medical benefits afforded to them by the Affordable Care Act (“Obamacare”)? While Mr. Graham is heavily committed to limiting reproductive healthcare available to women in the United States to protect unborn children, he does not seemed bothered that according to the WHO “[t]here are an estimated 50,000 pregnant women in Gaza, with more than 180 giving birth every day” or that “[f]ifteen per cent of them are likely to experience pregnancy or birth-related complications and need additional medical care.” It appears that some lives are worth protecting more than others.

              Yet it should not be surprising that the united response from the United States has been so overwhelmingly pro-Israeli considering the political and strategic importance of the region. “[M]ore than 80% of Israel’s weapons imports came from the U.S. between 1950 and 2020,” and the 3.3 billion dollars in military aid “foreign assistance” each year makes Israel the country most financially supported by the US on an annual basis. It should also not shock that a country which has continuously invested money and soldiers into foreign wars does not seem to have issue with one side of a conflict inflicting staggering loss upon the other. The ratio of casualties – especially civilian and child deaths – in Gaza compared to Israeli casualties harkens to other wars in which imperial forces slaughtered their opponents, perhaps most notably the 58,220 United States soldiers who gave their lives to kill an estimated 1.1 million Vietnamese soldiers and another 2.2 million civilians. In a country in which the war on terror is a veritable cash cow it is quite easy to see how a “we do not negotiate with terrorists” approach is so well embraced. But like the American war in Vietnam, or the Cuban Missile Crisis, or the American was in Iraq, or or or, the way to win the war is to be the one telling the story of it.

              In 2017 Zeynep Tufekci drew attention to the impact which social media played in the success and failure of the Arab Spring and the attempted revolutions in Tunesia, Libya, Egypt, Yemen, Syria, Bahrain, and other countries in the region. Her work helped illustrate the digital landscape upon which political and social struggle is now fought, identifying the ways that social media can make or break a cause. “Attention is oxygen for movements. Without it, they cannot catch fire.” Even twenty years before the Arab Spring the Kurdish conflict in Turkey was framed by a simple message, that “the government was fighting terrorists, and all those who died were terrorists.” It sounds horrifically familiar as the narrative coming from Palestine is that the Israeli response is simply a war waged against “human animals,” language which shockingly calls to mind the justification for the enslavement and dehumanization of people of color. Following a third communications blackout that started in Gaza on November 5th, Israel is “cutting off Palestinians from the outside world.” Perhaps a result of collateral damage, Mohammed Abu Hatab, a correspondent for a Palestinian television channel, and 11 members of his family were killed by an Israeli airstrike on Thursday. Just another unintentional side-effect of the “war on terror” and nothing more.

              “If you cannot destroy the message,” Tufekci noted, “why not shoot the messenger?” In other words, control the narrative to control the war, one post at a time.

Leave a Reply

Your e-mail address will not be published. Required fields are marked *.