You scroll through your feed. News, memes, opinions, videos—a constant stream of content tailored just for you. But here's the unsettling question: what are you not seeing?

Every social media platform uses algorithms to decide what appears in your feed. These algorithms don't show you everything—they can't. Instead, they curate a personalized reality, filtering billions of posts down to the few hundred you'll see today.

This creates a modern version of Descartes' evil demon: an algorithmic intelligence that controls your perception of reality. But unlike deepfakes that create false content, filter bubbles do something more subtle and perhaps more dangerous—they hide real content, making your view of the world systematically incomplete.

The Algorithmic Curation of Reality

Social media platforms don't show you a random sample of content. They show you what their algorithms predict will keep you engaged.

News feed algorithms consider thousands of signals to decide what you see: who you interact with, what you click, how long you watch, what you share. The result is a feed optimized for your engagement, not for truth or completeness.

Recommendation algorithms suggest content based on your viewing history. Watch one video about a topic, and the algorithm will suggest more like it. This creates recommendation pathways where each piece of content leads deeper into a particular worldview.

Some platforms use machine learning to predict what will capture your attention. The algorithm learns your preferences so quickly that users often report feeling "seen" by the platform—as if it knows them better than they know themselves.

The result: each person experiences a different informational reality. Your feed is not my feed. Your recommendations are not my recommendations. We inhabit separate information bubbles, curated by algorithms optimizing for engagement.

The Invisible Censorship

The most troubling aspect of algorithmic curation isn't what you see—it's what you don't see.

If a platform's algorithm decides certain content won't engage you, that content simply doesn't appear. You can't know what you're missing because you never see it. The algorithm acts as an invisible filter, removing content before it reaches your awareness.

This is different from traditional censorship. When a government bans a book, you know the book exists and has been banned. But when an algorithm deprioritizes content, it vanishes silently. You don't know what you're not being shown.

You might think you're seeing a representative sample of what's happening in the world. But you're seeing a curated selection optimized for your engagement, not for accuracy or completeness.

Echo Chambers and Confirmation Bias

Algorithmic curation amplifies a natural human tendency: confirmation bias. We prefer information that confirms our existing beliefs and avoid information that challenges them.

Social media algorithms detect this preference and exploit it. If you engage more with content that aligns with your views, the algorithm shows you more of it. If you scroll past content that challenges your views, the algorithm shows you less of it.

The result is an echo chamber: a self-reinforcing information environment where your beliefs are constantly confirmed and rarely challenged. You see evidence for your worldview everywhere because the algorithm is showing you that evidence while hiding contradictory information.

This isn't a bug—it's a feature. Platforms optimize for engagement, and people engage more with content that confirms their beliefs. Challenging content might be valuable, but it's less engaging, so the algorithm deprioritizes it.

The Fragmentation of Shared Reality

When everyone sees a different curated feed, we lose shared reality.

In the past, people might have disagreed about how to interpret the news, but they at least saw the same news. Today, people see fundamentally different information. What's trending for you might not be trending for me. What you consider a major story might not appear in my feed at all.

This fragmentation makes productive disagreement nearly impossible. We're not just interpreting the same facts differently—we're working from entirely different sets of facts. We inhabit different informational realities, and we often don't realize it.

Political polarization is partly a result of this fragmentation. When conservatives and liberals see completely different news feeds, they develop incompatible understandings of reality. Each side thinks the other is ignoring obvious facts—but those "obvious facts" never appeared in the other side's feed.

Radicalization Through Recommendation

Some researchers have raised concerns that algorithmic recommendation may lead users down radicalization pathways.

The proposed pattern works like this: someone watches a mainstream political video, then the algorithm recommends something slightly more extreme. They watch that, and the algorithm recommends something even more extreme. Each step seems reasonable—just slightly more intense than the last—but the cumulative effect could be radicalization.

This happens because algorithms optimize for watch time, and extreme content tends to be more engaging. Moderate, nuanced content doesn't trigger the same emotional response as extreme, simplistic content. So the algorithm may learn to recommend increasingly extreme content to maximize engagement.

However, research on this phenomenon is mixed. Some studies suggest recommendation algorithms can promote extreme content, while others find little evidence of systematic radicalization. The debate continues, but the concern remains: if algorithms optimize for engagement rather than accuracy or balance, they may inadvertently amplify extreme viewpoints.

The Emotional Manipulation

Algorithms don't just curate what you see—they curate how you feel.

In 2014, Facebook conducted an experiment on 689,000 users without their knowledge.^[1] They manipulated users' News Feeds to show more positive or negative content, then measured whether this affected users' own posts. It did. Users shown more negative content posted more negative content themselves. Users shown more positive content posted more positive content.

This revealed something disturbing: platforms can manipulate your emotional state by controlling what you see. The algorithm isn't just a neutral filter—it's an active influence on your mood, opinions, and behavior.

Every platform does this, whether intentionally or not. The algorithm learns what emotional triggers keep you engaged, then shows you content that triggers those emotions. If outrage keeps you scrolling, you'll see more outrage-inducing content. If anxiety keeps you checking, you'll see more anxiety-producing content.

The Impossibility of Objective Reality

Here's the philosophical problem: if all your information comes through algorithmic filters, can you ever access objective reality?

You can't see what the algorithm doesn't show you. You can't know what you're missing. You can't verify that your view of the world is complete or accurate because you have no access to the unfiltered information stream.

This is the algorithmic evil demon: an intelligence that controls your perception of reality, and you can't escape it or verify its accuracy. You're trapped in a curated information bubble, and you can't see the walls.

Unlike Descartes' evil demon, which was a thought experiment, the algorithmic evil demon is real. It's operating right now, deciding what you see and what you don't see, shaping your understanding of reality in ways you can't detect or control.

The Attention Economy

Why do platforms do this? Because their business model depends on it.

Social media platforms make money by selling advertising. The more time you spend on the platform, the more ads you see, the more money they make. So they optimize for engagement—keeping you scrolling, watching, clicking.

Truth, accuracy, and completeness don't directly contribute to engagement. In fact, they might reduce it. Nuanced, complex information is less engaging than simple, emotional content. Challenging information that contradicts your beliefs is less engaging than confirming information that reinforces them.

So the algorithm learns to show you simple, emotional, confirming content—not because it's true or important, but because it keeps you engaged. Your perception of reality is shaped by what maximizes advertising revenue.

Can You Escape the Bubble?

The troubling answer is: probably not completely.

You can try to diversify your information sources, follow people with different viewpoints, and actively seek out contradictory information. But the algorithm still controls what you see. Even if you follow diverse sources, the algorithm decides which of their posts appear in your feed.

You can try to use chronological feeds instead of algorithmic ones, but most platforms have removed or hidden this option. And even chronological feeds are filtered—you only see posts from accounts you follow, which is itself a form of curation.

You can try to leave social media entirely, but then you lose access to information that only exists on those platforms. And other information sources—news websites, search engines, streaming services—use similar algorithmic curation.

The algorithmic filter bubble isn't something you can simply opt out of. It's the infrastructure of modern information access.

Living in the Algorithmic Vat

So what do you do?

Awareness is the first step. Recognize that your feed is curated, not comprehensive. What you see is optimized for engagement, not truth. You're living in an information bubble, and so is everyone else.

Seek out diverse sources. Actively look for information that challenges your views. Follow people you disagree with. Read publications with different editorial perspectives. Don't rely on a single platform or algorithm.

Question your reactions. When content makes you feel strong emotions—outrage, fear, excitement—ask why. Is the algorithm showing you this because it's important, or because it triggers engagement?

Verify before sharing. Don't assume something is true just because it appeared in your feed. Check multiple sources. Look for primary sources. Be skeptical of claims that seem designed to provoke emotional reactions.

Understand the incentives. Platforms optimize for engagement, not truth. Advertisers want your attention, not your enlightenment. Recognize that your information environment is shaped by economic incentives that don't align with your epistemic interests.

Tomorrow's Question

Tomorrow, we'll explore digital twins and AI models of individuals. If an AI can replicate your personality, preferences, and behavior, which version is the real you? The biological original or the digital copy?

Unlike filter bubbles, which shape your perception of external reality, digital twins raise questions about internal reality—your identity, consciousness, and self. If a perfect digital copy of you exists, are you still uniquely you?

The algorithmic evil demon controls what you see. But what if technology could replicate who you are? That's tomorrow's question.

References

[1] Adam D. I. Kramer, Jamie E. Guillory, and Jeffrey T. Hancock, "Experimental evidence of massive-scale emotional contagion through social networks," Proceedings of the National Academy of Sciences, Vol. 111, No. 24, June 2014. https://www.pnas.org/doi/10.1073/pnas.1320040111