Curated news vs algorithmic feeds

Summary: Curated news is picked by an editor against a written standard and is the same for everyone. An algorithmic feed is picked by a model to maximize engagement and differs for every viewer. Research shows engagement-ranked feeds leave you feeling informed while making you less accurate and more polarized. For news, curated wins. Feeds are fine for low-stakes discovery.

Curated news vs algorithmic feeds

A curated news service has a human editor picking what runs, ordering it by significance, and publishing the same finished product to every subscriber. An algorithmic feed has a recommendation system picking what each user sees, ordering it by predicted engagement, and serving a different product to every viewer. Curated news answers "what should a citizen know today?" An algorithmic feed answers "what will keep this person scrolling longest?"

For how editors make those calls, read How news gets chosen. For what happens when the picker is a language model, Can you trust AI news summaries? has the numbers.

What's the difference between curated news and an algorithmic feed?

The difference is the selector, the standard, and the audience.

• The selector. Curated news has a person who can name a reason for picking a story. An algorithmic feed has a model that can't, beyond noting that the story scored highly on an engagement target.

• The standard. Curated news judges stories against a written editorial standard: significance, magnitude, lives affected, public-interest weight. An algorithmic feed judges them against an engagement target: clicks, dwell time, replies, shares, watch completion. The system doesn't score whether the story informs anyone.

• The audience. A curated brief is the same for every subscriber. A reader in Phoenix and a reader in Boston get the same six stories in the same order. An algorithmic feed differs for every viewer because the model makes a per-user prediction.

A curated brief ends, and an editor is accountable for the picks. A feed doesn't end, and even its own engineers can't fully explain the model.

What does an algorithmic feed optimize for?

An algorithmic feed is a closed-loop optimization system. It observes micro-actions (clicks, scroll pauses, skips, scroll-backs), builds a predictive model of behavior, ranks content to maximize a target variable, serves it, and retrains on the response.

Across major platforms the target is some version of engagement. The platforms have published architectural details that confirm this: candidate generation from a large pool, a ranking step that scores each candidate against a neural network predicting likely actions (likes, replies, shares, dwell time, watch completion), and a heuristics layer applying safety and inventory rules before serving. The system scores the predicted action rather than the value of the content.

1. Engagement and accuracy diverge. A University of Copenhagen study published in April 2026 compared personalized engagement-based ranking against random, non-personalized engagement, bridging-based, and intelligence-based ranking. The personalized engagement feed produced more polarized and less accurate beliefs than every alternative, while receiving higher "insightful" ratings from its users. People felt informed and were less so.

2. Stated preferences and revealed preferences diverge. A 2026 CSCW study of young adult social media users found a systematic gap between what users say they want from a feed (accuracy, diversity, high-quality information) and what they engage with. Asked to hand-curate an ideal feed for a made-up persona, the same users produced feeds that scored higher on quality and satisfied them more than a simple engagement-ranked feed.

Optimizing for engagement produces a feed that wins on revenue and satisfaction ratings and loses on the outcomes (accurate beliefs, high-quality information) readers say they want.

What does a curated news service optimize for?

A curated news service optimizes for significance: a small set of stories the editor judges important enough to take up your attention today. The brief is finite because the editor decided what was worth your morning and stopped there.

The selection bar is explicit and external. Editorial standards usually name a few criteria: the size of the change a story represents, the number of lives affected, the public-interest weight of the underlying decision, the depth of supporting evidence. A celebrity feud doesn't get in on popularity alone. A regulatory filing makes the cut despite being unpopular.

The discipline costs something. A curated service pays editors, builds editorial process, and accepts the friction of human judgment with all its bias and inconsistency. An algorithmic feed scales to billions of users with little editorial labor.

What does the research say about each?

• Filter bubble effects exist, especially for moderates. Two experimental studies published in 2025 in Germany and the United States built news recommender systems biased toward each participant's political preferences. That bias increased ideological polarization among politically moderate users in both countries, with no effect on how much those users disliked the other side.

• Recommendation systems create information cocoons. A 2025 multidimensional assessment of seven news recommendation algorithms found that over repeated rounds, most of them narrowed the range of topics users saw and pushed users into more closed, polarized groups. The cocoon effect is built into the design.

• "News will find me" eroded source evaluation. An April 2026 Penn State study found that readers with a strong "news will find me" (NFM) mindset, who rely on algorithms and social networks for news (about one in three Americans in 2015 survey data), judged stories recommended by algorithms or shared by friends as just as credible as stories recommended by editors. The mid- and low-NFM groups still distinguished between source types.

• Alternative ranking algorithms can produce better outcomes. The same University of Copenhagen study tested simple alternatives against engagement ranking. Every one produced more accurate and less polarized beliefs than the engagement baseline. Feed outcomes come down to design choices rather than automation itself.

Engagement-maximizing algorithms produce outcomes that diverge from what readers say they want. Design alternatives exist, but the major platforms haven't deployed them at scale.

When is an algorithmic feed the right tool?

Algorithmic feeds are useful for discovery where the content pool is too large for any editor to filter and the goal is surprise. Music recommendation, niche video, and hobby communities all benefit from a model that sees patterns across millions of users and surfaces what a human curator would never get to.

They're also useful when a wrong recommendation is cheap. A song you skip after thirty seconds costs you little, and a polarizing story that skews your view of an election without your noticing costs a lot.

Algorithmic feeds fail where you're trying to keep an accurate model of the world. News is the canonical example. Engagement is structurally misaligned with a sound picture of what happened today, and the mismatch is a property of the design pattern rather than a bug in any one platform.

How do you tell which one is serving you?

A short test for any news source in your current rotation.

1. Can the source name its bar? If the answer is "the algorithm decides based on what I've engaged with," it's a feed. If the answer is a written editorial standard you can read, it's curated.

2. Does the source end? Curated products are finite and stop, and feeds don't.

3. Does everyone see the same thing? If a friend can read the same edition and talk with you about the same six stories, it's curated. Feeds differ per viewer.

4. Is the optimization target visible? Curated products optimize for something the source names openly (significance, accuracy, depth). Feeds optimize for engagement, and the source will rarely say so.

5. How does it feel after thirty days? A curated source should leave you better informed and not worse-feeling. A feed often does the opposite.

One week of each, then compare

Subscribe to a curated news service for seven days. The next week, go back to algorithmic feeds for news and track how often you open them. Then write down three things: which stories from each format you remember, how informed you feel about events you didn't personally select, and how much time each format took.

Our bet is that for most readers the curated week wins on all three, because feeds are built to be forgotten, tuned for engagement rather than coverage, and never end.

The informed.now angle

informed.now is a curated SMS news service. An editor selects what runs each weekday, the brief is the same for every subscriber, and it ends. There's no algorithm choosing per user. Our methodology has the full editorial standard.