Algorithms that customize marketing to your phone could also influence your views on warfare

6 min read

You clicked on running shoes last week. Now every app you open shows you running shoes. Different brands, different styles, but always running shoes. The algorithm learned what you want.

Except it didn’t learn what you want. It learned what you clicked. And now it’s showing you more of that, because more of that keeps you scrolling, and scrolling is the product.

That same system, the one that tracks your shopping habits and serves you ads, is now tracking your political attention. It’s learning which headlines make you stop. Which outrage makes you share. Which framing of a war, a protest, or a policy debate keeps you engaged.

And just like with the shoes, it’s not showing you what’s true. It’s showing you what works.

From Products to Perspectives

Recommendation algorithms were built to solve a commerce problem. If someone buys a book about gardening, show them more books about gardening. Simple. Effective. Profitable.

But the infrastructure didn’t stay in the shopping cart. It moved to news feeds, video queues, and search results. The same system that recommended products started recommending perspectives.

When you watch a video about a geopolitical conflict, the algorithm doesn’t ask whether the next video is accurate. It asks whether you’ll watch it. If you watched the first video to the end, the algorithm assumes you want more like it. More of that tone. More of that framing. More of that emotional register.

If the first video presented the conflict as a clear moral binary, the next five will too. If it focused on outrage, you’ll get outrage. If it vilified one side, the recommendations will reinforce that vilification.

The algorithm doesn’t care about truth. It cares about retention.

How Personalization Became Persuasion

You might think you’re immune. You fact-check. You read multiple sources. You’re media literate.

But personalization doesn’t announce itself. It doesn’t feel like manipulation. It feels like the internet finally understanding you.

You search for information about a protest movement. The algorithm notes which results you click, how long you stay on each page, which videos you watch all the way through. It builds a profile. Not of who you are, but of what keeps you engaged.

Then it shows you more of that. Not because it’s representative. Not because it’s accurate. Because it works.

A teacher in Ohio searches for information about a foreign conflict. She clicks on an explainer video that presents the conflict through a humanitarian lens. The algorithm flags that. The next search shows her three more videos with the same framing. The sidebar recommends opinion pieces using similar language. Her YouTube homepage fills with content that echoes the same perspective.

She didn’t choose an echo chamber. The algorithm built one around her.

The Infrastructure of Consensus

This isn’t just about individuals forming biased views. It’s about large populations being shown fundamentally different realities.

Two people search the same phrase about the same war. One sees videos emphasizing civilian casualties and humanitarian crisis. The other sees videos emphasizing strategic necessity and national security. Both believe they’re getting a neutral view of the conflict. Both are being fed a curated reality based on their prior engagement patterns.

The algorithm isn’t neutral. It’s not trying to inform. It’s trying to maximize watch time, clicks, and ad revenue. And it’s learned that strong emotion, moral clarity, and us versus them framing are incredibly effective at keeping people engaged.

So that’s what it serves. Not the most accurate information. The most engaging version of information.

This matters in classrooms because your students are living inside these systems. They’re not encountering ideas and forming opinions. They’re being served opinions and encountering the ideas that support them.

Marketing algorithms work by segmentation. They divide audiences into groups based on behavior, then serve each group the message most likely to convert them.

That logic has migrated wholesale into political communication. Campaigns, advocacy groups, and media outlets use the same targeting tools that sell products to sell perspectives on policy, protest, and war.

A student who watches gaming content gets ads for a political candidate framed through gaming culture references. A student who follows environmental accounts gets the same candidate’s message reframed around climate policy. A student who engages with military content sees that candidate’s foreign policy stance emphasized.

Same candidate. Different message. Personalized for maximum resonance.

This isn’t new in politics. Campaigns have always tailored messages to different audiences. But the scale, precision, and automation have changed. And the line between informing and manipulating has dissolved.

Your students aren’t just seeing different ads. They’re seeing different facts, different framings of conflicts, and different versions of what’s happening in the world. All optimized for engagement, not accuracy.

You can’t opt students out of this system. They live on these platforms. But you can teach them to see it.

Start with a simple exercise. Have students search the same term on their phones. Compare results. Not just the order, but the actual content being surfaced. What’s emphasized? What’s missing? What assumptions are baked into the framing?

Then go deeper. Ask them to notice when a platform starts serving them content they didn’t search for. What did they click yesterday that led to this recommendation today? Can they trace the logic?

Teach them that personalization isn’t the same as accuracy. The algorithm doesn’t know what’s true. It knows what keeps you watching.

Have them practice switching contexts. If they usually watch political content from one perspective, what happens when they intentionally engage with another view for a few days? How does their feed shift? What gets amplified? What disappears?

This isn’t about forcing ideological balance. It’s about making visible the systems that shape what students see, and therefore what they think is normal, true, or worth caring about.

Marshall McLuhan said the medium is the message. The structure of a communication technology shapes what can be said, who gets heard, and what’s even thinkable.

Personalized recommendation algorithms are the medium now. And the message they’re sending is this: your reality should be comfortable. It should confirm what you already believe. It should feel engaging, not challenging.

That’s terrible for selling shoes. It’s catastrophic for understanding war.

When students encounter information about conflict, protest, or policy, they’re not just consuming content. They’re being shaped by an infrastructure designed to maximize their engagement, not their understanding.

And that infrastructure is learning. It’s getting better at predicting what will keep each person scrolling. Better at serving the version of reality most likely to hold attention. Better at making the filter bubble invisible.

You can’t dismantle that system. But you can teach students to see it. To question why they’re being shown what they’re being shown. To notice when their feed stops challenging them. To recognize that the algorithm’s job isn’t truth. It’s retention.

And in a world where the same system selling products is shaping perspectives on war, that recognition isn’t just media literacy. It’s survival.

Related Reading:

By Digital Alma

About the Author: writes Digital Alma, a newsletter about cyberpsychology and what it means to become yourself in a world that archives everything. For reflections that don’t make it to the essays, subscribe at .