Personalized News: How to Get Only the Stories You Actually Care About

Personalized news has a reputation problem. For most people, the phrase conjures an image of algorithmic feeds that know you a little too well. platforms that track what you click, measure how long you hover, and serve up more of whatever keeps you engaged. The result is a news diet that's heavy on outrage, light on substance, and eerily tuned to your emotional vulnerabilities.

That's one kind of personalization. It's the dominant kind, and it's the kind most people have experienced. But it's not the only kind, and the distinction matters. because the difference between engagement-driven personalization and interest-based personalization is the difference between a news experience that respects you and one that exploits you.

How engagement-driven personalization works

The major news and social platforms personalize content using behavioral signals. They track what you click, what you scroll past, what you share, what you linger on, and what makes you come back. These signals feed into recommendation engines that optimize for a single metric: engagement. Time on platform. Click-through rate. Session frequency.

This sounds reasonable until you examine what "engagement" actually measures. Engagement is not interest. Engagement is not importance. Engagement is not value to the reader. Engagement is the likelihood that a piece of content will capture attention. and attention is most easily captured by content that provokes strong emotional reactions. Anger, fear, outrage, and tribal identification are the most reliable engagement drivers. Calm, nuanced, genuinely informative reporting is not.

The result is a personalization system that doesn't learn what you care about. It learns what makes you react. And over time, it gives you more of what makes you react and less of what would actually inform you. Your feed becomes a mirror of your impulses, not your interests.

This has consequences. Filter bubbles. where people only see information that confirms their existing views. are a well-documented outcome. But there's a subtler effect: the gradual degradation of the reader's own sense of what they want. When an algorithm decides what you see based on your past behavior, you lose the ability to discover something outside your established patterns. Serendipity. the encounter with something unexpected that expands your understanding. is engineered out of the experience.

The alternative: interest-based personalization

There's another way to personalize news, and it's considerably simpler. Instead of tracking behavior and inferring preferences, you ask the reader what they care about. Then you find the best reporting on those topics and present it.

This is how personalized news worked before algorithms: you chose a newspaper, you chose which sections to read, and you skipped the rest. The newspaper didn't reshape itself around your behavior. It presented an editorially curated selection and let you navigate it. Your personalization was your choice, made consciously, not inferred from surveillance.

Interest-based personalization preserves that reader agency while adding precision. Rather than choosing a whole newspaper and ignoring the sections you don't care about, you specify your actual interests. technology and climate science and foreign policy, say, but not celebrity culture or sports. The system finds the most important stories within your declared interests and assembles a newspaper from them.

The critical difference is the input signal. Engagement-driven personalization uses behavioral data: what you did. Interest-based personalization uses declared preferences: what you said you want. These sound similar but produce radically different outcomes.

Behavioral data captures impulses. You click on an outrage headline not because you want more outrage but because the headline was designed to be irresistible. The algorithm records that click as a preference and serves more outrage. Your feed drifts toward content that hijacks attention, not content that serves your actual interests.

Declared preferences capture intent. When you say "I care about technology and climate science," you're expressing a genuine interest that reflects your values and priorities. There's no ambiguity, no inference required, no risk of mistaking a compulsive click for an authentic preference.

How Edition's approach works

The personalization system in Edition is built on declared preferences rather than behavioral tracking. Here's what that looks like in practice.

When you set up your newspaper, you choose the categories that interest you. Technology, business, politics, science, culture, health, world affairs. you select the ones you want and ignore the ones you don't. You can also weight your choices: if you care about technology more than business, you tell the system, and your newspaper will reflect that weighting with more technology coverage and less business coverage.

You can also exclude specific topics. If you're interested in technology broadly but don't want cryptocurrency coverage, you can exclude it. If you follow politics but want to avoid horse-race election coverage, you can exclude that too. Exclusions are hard filters. excluded topics won't appear in your newspaper regardless of how important the system considers them.

With your preferences set, the system monitors over fifty sources for the most important reporting across all categories. Articles are classified by topic, clustered to eliminate duplicate coverage of the same story, and scored against your specific preference profile. The highest-scoring articles make it into your morning newspaper.

The scoring considers multiple factors beyond just topic match. Editorial importance. is this a story everyone should know about?. carries significant weight. A major geopolitical event might appear in your newspaper even if foreign policy isn't your top category, because some stories are important enough that excluding them would leave you meaningfully less informed. Source diversity also matters: the system avoids pulling too many stories from the same publication, ensuring your newspaper reflects a range of editorial perspectives.

What the scoring does not consider: how you've interacted with past newspapers. There's no tracking of which stories you read, how long you spent on them, or whether you opened the newspaper at all. The system has no behavioral data because it doesn't collect any. Your newspaper tomorrow will be based on the same declared preferences as your newspaper today, unless you explicitly change them.

No filter bubbles, by design

One of the most damaging effects of engagement-driven personalization is the filter bubble. the tendency of algorithmically curated feeds to show you only what you already agree with, reinforcing existing beliefs and shielding you from contrary perspectives.

Filter bubbles are an intrinsic property of engagement optimization. People engage more with content that confirms their views than with content that challenges them. An engagement-maximizing algorithm will therefore systematically filter out challenging perspectives, not out of ideological intent but as a mathematical consequence of optimizing the metric it was told to optimize.

Interest-based personalization doesn't have this problem, because it doesn't optimize for engagement. It optimizes for topic relevance and editorial importance. If you say you care about climate science, the system finds the best climate science reporting. regardless of whether it confirms your existing views on climate policy. If you say you care about politics, you get the most important political stories. not the ones most likely to make you angry.

The system also includes an editorial importance signal that's independent of personal preferences. Some stories are significant enough that they should appear in every reader's newspaper. A major international crisis, a landmark legal decision, a scientific breakthrough of broad consequence. these are stories that any informed person needs to know about, regardless of their declared topic preferences. This editorial floor ensures that personalization narrows coverage to your interests without creating blind spots about major events.

The question of serendipity

A common concern about any personalization system. engagement-driven or interest-based. is the loss of serendipity. If the news is filtered to match your interests, how do you encounter the unexpected story that broadens your perspective?

This is a legitimate concern, and the answer involves understanding what serendipity requires. Serendipity isn't randomness. showing someone random articles doesn't create serendipity, it creates noise. Serendipity is encountering something you didn't know you'd find interesting, presented in a context where you're receptive to it.

A well-curated newspaper is actually good at producing serendipity, because the editorial importance signal surfaces stories that matter regardless of topic category. A reader who cares primarily about technology and business will still see the cultural story that everyone is talking about, the scientific finding with broad implications, the international development that affects everything else. These stories appear not because they match the reader's declared preferences but because they pass the editorial importance threshold.

The broadsheet format itself supports serendipity in a way that linear feeds don't. When you look at a newspaper page, you see everything at once. the story you sought and the story you didn't. Your eye lands on a headline you wouldn't have clicked on in a feed, and because it's right there on the page, you read it. The spatial layout of a newspaper creates encounters that a sequential feed, which shows you one thing at a time, cannot.

What personalization should and shouldn't do

Personalized news should make you more informed about the things you care about. It should save you time by filtering out coverage of topics you've explicitly said you don't want. It should preserve your ability to encounter the unexpected. And it should do all of this based on your conscious choices, not your unconscious behaviors.

Personalized news should not manipulate your emotional state to keep you engaged. It should not infer preferences from surveillance. It should not create filter bubbles that narrow your worldview. It should not optimize for the platform's revenue at the expense of the reader's experience.

The difference between these two approaches is not subtle, and it's not just philosophical. It's the difference between a news experience that leaves you informed, calm, and done. and one that leaves you anxious, outraged, and reaching for your phone again in twenty minutes.

The method of personalization matters enormously. How a system knows what to show you. whether through behavioral surveillance or through your own declared choices. determines whether it serves you or exploits you.

Your news should be personalized to your interests, not to your impulses. That distinction is everything.