Three years ago, I still believed that if I filmed what was in front of me—be it a short interview or unedited reality—its seriousness would carry it. I know better now. In the era of the algorithm, an eleven-minute testimony about dead students could simply vanish, while fifteen seconds of a man breaking down in front of the camera might travel far beyond the witnesses of such a tragedy. A still photo might hold a world and still be treated like nothing, while chaos is rewarded as truth. My journalist colleagues in Gaza capture our people’s suffering through captions crafted in English because clips in Arabic, as we’ve come to learn, don’t move through feeds unless they’re translated and packaged. Grief, apparently, has an optimal runtime.
The truth we’ve had to learn the hard way is that our stories—and therefore our realities—compete in a machine designed to keep people scrolling in an information ecosystem where not only do we see less of what we need to know in real time, but we also see only a specific fragment of what’s actually happening in the world. In turn, whatever images garner engagement coalesce and repeat until they become the template for what we think crisis-devastated places look like.
Gaza is a clear example. My people there have been accustomed to social media platforms controlling and deciding who counts as witnesses and what parts of their agony count as legitimate testimony.
As a reporter, it’s difficult for me to decide which stories get to be shared and which don’t. In Gaza, people would speak to me for hours, only for me to publish bits and pieces of our conversations. There were people who told me stories of strength as much as they spoke about the immense trauma they had to live through. They offered context and tried to explain the worlds they had lost. They spoke in complete sentences, though their lives had been cut short.
I remember a schoolboy who loved math and always sat in the front row before the war. He described a smell—chalk and oranges—because the kids ate breakfast in the classroom then. This child, who was almost ten, had to make sense of something that isn’t supposed to happen. This smell held within it a world of memories, and yet almost nobody will ever know this detail because it asks for time and attention, the two things the feed is designed to strip away. Public attention is funneled into a narrow channel. People don’t realize how much of what they call awareness is the product of filtration.
As journalists in Gaza, we worked together under the same pressures: post quickly, constantly, and in a way that survived network cuts, moderation systems, and the audience’s exhaustion; and above all else, as a team, try to avoid being targeted by Israel’s deadly attacks on journalists. Once, a colleague filmed a doctor in a hospital hallway sobbing into his hands. The clip spread across the internet within hours. Then she filmed him again—composed and precise—explaining the anesthesia shortage and which procedures were no longer safe. The second video, which had the information but not the collapse, didn’t travel. “The crying is what they want,” the journalist said to me. “The information is what they need.” The algorithm has a taste for tears but not for anesthesia protocols.
These people are not responsible for which footage gets the most clicks, but this is how the infrastructure of attention is built. Platforms can’t optimize for both understanding and reaction at once. Shock, sadness, anger, and pity keep people in the app. Context, on the other hand, slows them down; complexity makes people leave, and analysis competes with entertainment and loses.
*
Notably, the act of content suppression is often quiet. Content used to be removed to be neutralized, but now platforms have simply started to deprioritize certain topics, meaning posts can exist and still not reach anyone—visible if searched for but effectively invisible otherwise. There’s no obvious way to protest against this powerful form of control. We can’t rally against something we can’t prove happened.
I spoke with a researcher recently who studies digital rights in conflict zones. She said platforms choose the option that produces the least public friction. And because most viewers never know what didn’t reach them, the platform pays almost no reputational cost for deciding, repeatedly, that the safest choice is suppression.
Gradually, the digital narrative around Gaza has been shaped without an explicit policy of censorship. Platforms should not be seen as neutral pipelines when their editors have incentives that do not align with the truth. These systems would rather look for content that prioritizes immediacy over comprehension. Spectacle sells. Tech companies allow some images through—usually the ones that match what audiences already expect war to look like—then throttle the material that contextualizes that picture. They let the screaming through, but bury the testimony.
That doesn’t necessarily make all the platformed content false, only contingent, meaning we have to recognize that censorship now often looks like distribution failure. If a post is throttled rather than removed, the platform avoids a confrontation while still achieving the effect: fewer people see it, fewer people talk about it, and so fewer people build understanding from it.
After leaving Gaza, I started tracking this more deliberately. I couldn’t tolerate the feeling that something was happening in my feed that I couldn’t name. I noted what people posted and how far it went. There seemed to be an overlooked fact: analysis underperforms, and so do simple, quiet images. Anything that asks the viewer to think rather than react is disadvantaged from the start. Trauma—especially the kind that arrives in the right format at the right intensity—wins.
This does two things at once: It gives audiences the sensation of constant exposure (“I see Gaza every day”). At the same time, it shrinks the range of what Gaza is allowed to be in the public imagination.
I think about this every time people tell me they’re “staying informed” through social media. They don’t know that they are being informed only about the slice of reality that survives engagement metrics and moderation filters. They know about the looped clips that performed well enough to be repeated until they became the story. People believe they’re seeing “so much,” not understanding that repetition can masquerade as coverage. Seeing the same kind of clip over and over creates a false sense of comprehensiveness. The feed feels saturated, so people assume reality is fully represented. It isn’t. A viral video could be true and still be misleading. It could be a factual substitute for the larger reality. It becomes the representative moment, the one everyone shares, until the brutal reality itself is reduced to a handful of images. That reduction changes how people interpret what they’re seeing and turns an ongoing political catastrophe into a set of emotionally powerful fragments.
Over the past two years, many photos and videos have circulated showing Palestinian journalists in Gaza being bombed live on air as they did their job. The clips traveled because they were perfectly compressible—scenes that could be shared without context, risk, or thought, not even captions. This footage made viewers feel something strong and immediate. It also made it easy to stop there and to treat feeling as participation. People feel and move on.
If the dominant frame is crying children and chaotic crowds amid the rubble, then the “appropriate” response becomes charity and grief. Though that is largely what defines life now, it misses the need for political pressure rooted in understanding. What becomes possible then is shaped by what people see and what they are able to demand. Once the public story is built out of that narrow band, it shapes everything downstream: what counts as credible, as urgent, what kind of policy response feels natural or extreme.
Meanwhile, child amputees from every half-destroyed medical facility across the Strip, speaking in Arabic, with no translation and no hook, likely remain unseen. Their tragedies are no less true, but it’s harder for the machine to monetize this kind of footage and harder for viewers to consume something that requires slowing down to understand, that requires contending with the actual needs of those being watched and identifying who has the power to stop the atrocities. If the dominant imagery is grief, the dominant response becomes sympathy. Sympathy is not what’s asked for. Sympathy without understanding what—and who—is responsible for the grief is politically weak. It produces mourning and charity more readily than it helps pave the way for accountability.
Undoubtedly, we have to separate the act of bearing witness from consumption. Viewing and sharing should not be confused with action. And feeling will never be the same as solidarity. The feed encourages that confusion because it benefits from it, trapping viewers in a cycle of watching to absolve themselves of the guilt of inaction. It offers people emotion as a substitute for agency, for confronting the witness’s responsibility toward the witnessed.
*
The same machine that curates audiences and what they are exposed to also reshapes those inside the catastrophe, posting through the trauma. It teaches them what their lives are worth online, and it tells journalists which parts of reporting will be rewarded and which will be ignored. It also shows civilians that if they want to be seen, they have to translate themselves, condense their pain, simplify their reality, and—often—injure themselves into legibility. They learn what kind of pain is “worth it.” They start to learn how to speak to the machine and master adaptation under pressure. I hate admitting how often I’ve participated in that training.
I’ve asked photojournalists and videographers for “better” footage when what they provided already represented the truth. I’ve encouraged subtitles, framing, hooks, and more explanatory captions that turn life into a product. I’ve thought in angles and time limits while standing in front of someone’s ruin. Even devastation has to respect the aspect ratio. This was itself a moral dilemma. I know the brutal arithmetic: if it doesn’t travel, it might as well not exist for most of the world.
At some point, the decision-making process behind what to report becomes more complicated. The work becomes less about documenting reality and more about producing a version of reality that survives distribution. We as journalists and writers start making choices that are not artistic in the classical sense, because we need to be strategic: Which moment will break through, which phrase will be quotable, which face will move people, which image is “graphic” enough to cause a reaction but not so graphic that it gets buried We are negotiating with the rules of a system that is indifferent to the facts. We stop being able to tell the difference between documenting and performing. The algorithm trains it out of us.
The “good victim” isn’t new, but platforms have given it a sharpened face. The good victim is digestible. He must be eloquent in English and must offer innocence and heartbreak without rage. The good victim can’t speak with complexity or political clarity that might discomfort or demand something of the viewer. The good victim doesn’t take too long. He just fits. He has to. He is algorithmically fluent. Anything outside that frame is punished with silence.
*
My family back home calls me sometimes and we talk about nothing important—schoolwork, the weather, food. Our calls don’t usually inspire posts. I don’t turn them into proof of anything. They are the private, unoptimized moments of my life. Part of me thinks that the only sane refusal left is to keep some humanity out of circulation and to insist that not everything must be converted into a unit of engagement to “count.” But refusal comes with a cost. Invisibility is part of the platform’s leverage: perform or disappear. Not only does the algorithm prefer beauty and simplicity, it convinces everyone—viewers, journalists, families under siege—that what it selects is what is real.
The algorithm doesn’t have to want anyone dead to produce a world where some deaths are seen and others aren’t, or to make certain stories circulate and others vanish. The public’s understanding is largely built from what performs rather than what explains. In that world, even when we think we’re paying attention, the line blurs between being informed and being attention-farmed.
