Escape the social media matrix

Plus: Israel’s AI-assisted bombing. Congress abetting ethnic cleansing? Why this automation wave is different. Zelensky on the defensive. And more!

Save the date: At 11 a.m. US Eastern Time on Thursday, December 21, I’ll do a livestream video with longtime collaborator Nikita Petrov. Paid newsletter subscribers can sit in on our Zoom session and submit questions, and everyone else can use YouTube’s comments function to kibitz. Details in next week’s Earthling.

One reason it’s in America’s interest for the fighting in Gaza to stop is that the fighting in Gaza is tearing America apart.

OK, that’s a slight overstatement. Did it strike you that way—as just a bit hyperbolic, lacking in nuance? Did you think, “Yes, the Gaza War has heightened tension between pro-Israel and pro-Palestine factions, and has even led to an increase in anti-Semitic and anti-Muslim or anti-Arab incidents—but ‘tearing America apart’ is a bit strong?”

I hope so—especially if you’re an American. Because I think America will be less likely to get torn apart—by the Gaza War, by the coming presidential election, by politics generally—if Americans are good at spotting overstatement and at resisting the temptation to engage in it.

Unfortunately, the media landscape—mainstream media, independent media, but especially social media—seems almost designed to make that hard. The resulting reality distortion field mediates much of our perception of the stresses the Gaza War is putting on American society. The consequent misperceptions then add to the stresses, and these new stresses are themselves subject to perceptual distortion, and so on. It’s a positive feedback cycle that could, given a few more months, or even weeks, render the first sentence of this piece non-hyperbolic.

So I’m going to spend the piece looking at how the reality distortion field works in hopes of helping people escape it.

Let’s start with a tweet from Jonathan Schanzer, a vice president of the intensely pro-Israel think tank Foundation for Defense of Democracies (and, disturbingly, a former analyst of terrorism finance at the US Treasury Department): “These protests are not simply in support of Palestinians. They are backing Hamas. Which means they are backing Hamas’s Iran-backed terrorism that is powered by annihilationist, jihadist, and anti-Semitic ideologies." 

The evidence Schanzer provided for this characterization of pro-Palestine protests broadly was a single photo of a single Hamas flag at a single pro-Palestine protest. Since I saw Schanzer’s tweet a month ago, I’ve scanned a number of videos and photos of pro-Palestine protests to see if I could find Hamas flags in the sea of Palestinian flags. No luck so far.

In a way, Schanzer is doing America a favor. By explicitly going so far beyond the data at hand—a single data point—he’s making his overgeneralization hard to miss. In the absence of such flashing red lights, it’s easier to fall for overgeneralizations. In fact, sometimes a single data point, presented with no commentary at all, can be a powerful reality distorter: We do the overgeneralizing ourselves, at an unconscious level. That’s especially true if we’re part of a group that naturally finds the single data point unsettling. Consider these two cases:

1) Suppose you’re an Arab or Muslim and you see the now-famous video of a New York street vendor of Arab descent being subjected to an anti-Islam tirade by a very vehement man (who, it turned out, had worked in the Obama State Department).

2) Suppose you’re Jewish and you see the now-famous video of protesters outside of Goldie, a Jewish-owned restaurant in Philadelphia, chanting “Goldie, Goldie, you can’t hide, we charge you with genocide.”

In both cases you’d probably feel a sense of personal threat—even though, if you sat down and did the math, you might conclude that this incident, by itself, doesn’t signify an appreciably raised risk level. This feeling of threat is a kind of unarticulated overgeneralization from a single data point.  

And it’s a natural reaction—in the literal sense of being grounded in human nature. In the environment in which humans evolved, the visual image of someone threatening the group you identify with would have signified the actual physical presence of that someone—a threat confronting you then and there, in real time. And an acquaintance of yours reporting that they’d seen such a threat would have meant that the threat was nearby.

Mass media changed this, bringing news and images from far away, and social media, of course, has changed it big time. As the Canadian academic Regina Rini recently put it, “Twitter is an engine for selectively highlighting the worst things said by the worst representatives of the political groups you hate.” Ethan Mollick, a professor at Penn’s Wharton School, puts it this way: “Your timeline is full of what enrages you, not what is representative.”

The outrageous behavior you see in your timeline can be unrepresentative not just in a quantitative sense (in the sense of not being typical of the ideological, national, or ethnic group the outrageously behaving people are part of) but in a qualitative sense; the version of the incident that circulates on social media may distort what happened in the first place.

For example: The protesters at Goldie were commonly depicted on social media as targeting the restaurant because it was Jewish-owned. The governor of Pennsylvania said they focused on Goldie “simply because it’s owned by a Jew and [they] hold that Jew responsible for Israeli policy. That’s the definition of antisemitism.”

The Philadelphia Inquirer’s account of the protest’s origin is too complicated to briefly summarize, but one takeaway is that Goldie wasn’t targeted just because its owner was Jewish. The protesters, rightly or wrongly, believed that the owner had fired workers for wearing pro-Palestine pins, had raised funds for an Israeli charity that supports the Israeli military, and had misled workers about what the charity did.

None of this (even if true) makes the protest a good idea. If I were moonlighting as an adviser to pro-Palestine protesters, one of my guidelines would be, “Avoid doing things that, as they circulate on social media, will lead some Jews to think about Kristallnacht.” But my point for now is that when you see an outrageous incident on social media, don’t assume you’ve seen all relevant context, or that the commentary accompanying it is accurate.

So too with the guy who harassed the Egyptian street vendor. The harasser claims the vendor had said he supported Hamas. This may not be true, and even if true it doesn’t justify his behavior, but the claim does increase the chances that this guy wasn’t just randomly harassing people of Arab ancestry. (Also, he turns out to be a repeat offender; he crudely insulted a Russian diplomat after the invasion of Ukraine, and he harassed the street vendor more than once. If you watch the various videos of him—including the one where he says that, if Israel has killed 4,000 Palestinian kids, that “wasn’t enough”—he starts to look mentally ill, which is more reason to think of him as not representative of his ideological tribe.)

I find that investigating the context of outrageous social media material—which often begins with seeing how it’s being processed outside of the tribe that it most outrages—usually alters my judgment of the affair. 

Sometimes, to be sure, the change is small. When I saw this week’s video of Rep. Elise Stefanik grilling three college presidents about how they’d handle students who call for Jewish genocide, and all three of them answering in wishy-washy fashion, I thought, “Wow, they really bungled this.” And, actually, I still think that. But I feel about three percent less harshly judgmental now that I’ve read Michelle Goldberg’s New York Times column on the incident.

Goldberg notes that if you watch the whole interrogation you realize that Stefanik set up the climactic parts of the interrogations—the parts that made it onto social media—by first equating, for example, the phrase “From the river to the sea” with a call for Jewish genocide. As I’ve noted before, genocide isn’t what most Americans who use that phrase have in mind. So Stefanik had conflated legitimate political speech and horrible murderous speech, a conflation that complicated things for the three presidents and is presumably one reason they failed to issue the kind of unequivocal condemnation Stefanik sought. (Though, again, it didn’t complicate things enough for me to quit marveling at their lack of dexterity.)

So is there a way to escape the reality distortion field? Or at least significantly dilute its impact? Two basic approaches come to mind:

1. Technical insulation. Take time off, maybe a lot of time off, from social media. Or use the tools social media gives you to dull the impact of its insidious algorithms. Block or mute the people who bring out the worst in you. And on some social media sites, including Twitter, you can create a kind discretionary insulation via the “lists” function: Assemble lists of different kinds of people (they don’t have to be people you follow). My own lists are topic-specific: people who tweet about Ukraine, people who tweet about AI, etc. But, however you organize your lists, the point is that you can decide which list you’re in the mood for, and which lists you don’t feel would be healthy for you at the moment. And on all lists, by the way, you’ll be free from the dreaded Twitter algorithm; all you’ll see is tweets and retweets from the people on a given list, in chronological order.

2. Psychological insulation. Expose yourself to a diversity of views: Follow people from your tribe and from other tribes—including even people in your own tribe who tend to share content that outrages you and even people in other tribes who are content that outrages you. Then try hard not to be outraged.

The second approach is hard, but the good news is that if you succeed you’ll be maturing psychologically and, in a sense, spiritually. You’ll be learning to maintain equanimity in a disturbing world, and your capacity for charitable judgment will grow. (There are disciplines that can help you nurture equanimity, including mindfulness meditation, so if you want to buy a book about that, feel free.)

You can also combine the two approaches: Use technical tools to weed out the biggest challenges to your mental health and then work at preserving your peace of mind amid the rest. Chances are you’ll wind up contributing to peace more broadly.

The Israeli magazine +972 reports on how AI is guiding the military’s selection of targets in Gaza—a system that one former intelligence official calls a “mass assassination factory.”

The AI-driven system (known as Habsora, or “the Gospel”) uses voluminous data streams to generate targeting recommendations—including the homes of suspected Hamas operatives—“at a rate that far exceeds what was previously possible.” Sources who have worked with the system say that the “emphasis is on quantity and not on quality” and that, although the automated selections are evaluated by humans, “there is little time to delve deep into the target[s].”

The Israeli military says the platform helps increase precision while minimizing harm to non-combatants, thanks to algorithms that assess likely civilian presence before a strike. However, +972 reports that the threshold for acceptable civilians deaths has been set at a higher level than in previous Israel-Hamas conflicts: 

In one case discussed by the sources, the Israeli military command knowingly approved the killing of hundreds of Palestinian civilians in an attempt to assassinate a single top Hamas military commander. “The numbers increased from dozens of civilian deaths [permitted] as collateral damage as part of an attack on a senior official in previous operations, to hundreds of civilian deaths as collateral damage,” said one source.

Three Russia-Ukraine updates:

1. With Ukraine’s summer offensive having failed, President Zelensky signaled a shift to a more defensive posture. “In all major sectors where reinforcement is needed, [we must] speed up building of structures,” he said in his nightly address. On Twitter, journalist Leonid Ragozin (a former Nonzero podcast guest) pointed out that Russia had six months to fortify its defenses before Ukraine’s summer offensive. “Now they [the Russians] are on the offensive and Ukraine doesn’t have any of that time.”