News, in the order it actually happens.

If you've spent five minutes on a news app in the last decade, you've been ranked. The feed isn't reverse-chronological; it's a model output. The model is trained on what made other people scroll, click, and rage-share. The natural endpoint of that loop is a feed that's optimized to make you keep using the app, not to inform you.

WeSearch goes the other way. Our home feed is deduplicated, source-diverse, and time-ordered. We pull from 700+ editorial RSS feeds — left, right, center, international, science, finance, tech — and merge them by recency, with simple deduplication when the same headline shows up across sources. There is no ML model picking which headline gets shown to you. There is no "Suggested for you." There is no "You might also like."

How it actually works

• Pull, don't predict. Every 5 minutes we pull a fresh batch from each subscribed feed.
• Merge by time. Stories sort by publish time, newest first. That's the whole sort.
• Dedupe by URL. Same headline from three wires? You see it once.
• Categories are honest. Each source's category comes from a static directory we maintain, not from semantic clustering. Tech is tech, World is world.
• Trending is opt-in. The 24-hour trending row counts community reactions, not engagement velocity. If you ignore it, your feed stays untouched.

What you give up

An algorithmic feed is, in some narrow sense, more efficient — it learns what you click and shows you more of that. The trade-off is that it builds a bubble around you, slowly, without telling you. WeSearch refuses that trade. You'll see headlines you wouldn't have clicked on. That's the point.

What you get

The same news a working journalist or analyst sees in their RSS reader, with deduplication done for you, comments and reactions added, anonymous identity included, and zero data harvested off the back. It's a community hub, not a personalization engine.

Browse by category, not algorithm

If you want to filter, do it manually. Each category page lists the most recent stories from every source in that bucket: world news, US news, politics, technology, science, business, markets, health, climate, AI, culture, media.

The longer argument

For a more complete argument about why we made this design choice, see why no algorithm. The short version is: engagement and informedness are not the same objective, and any feed optimized for the first will, at scale, undermine the second. We chose the second. Chronology is dumb on purpose.

Why algorithmic feeds quietly drift

The structural problem with engagement-optimization isn't that any single ranking decision is wrong; it's that the feedback loop drifts the feed over time toward what produces clicks rather than what produces understanding. The drift compounds: a feed that learned you click on outrage will show you more outrage, which trains your taps further toward outrage, which the feed reinforces. After several years, the reader notices their news feed is angrier and dumber than they remember it being and assumes the world has changed; usually the world hasn't changed nearly as much as the feed has. Restoring chronology breaks that loop.

None of this is novel — Yochai Benkler, Jaron Lanier, Tristan Harris, and many others have made variations of the argument over a decade. The novel thing is that very few news products actually act on it. Defaulting to chronological is unfashionable but cheap and structurally sound.

Trade-offs we accept

• Slower velocity for "what's important right now." Algorithmic feeds optimize for surfacing the breaking story instantly. Chronological feeds rely on you scrolling. We mitigate with the 24-hour Trending row (count-based, not model-based) for high-engagement stories.
• No personalization. Same feed for every reader. Some readers love this; others find it less efficient. We trade efficiency for accountability.
• Information overload. 700+ sources × 24 hours = a lot of headlines. We expect readers to scope by category or use keyword push for specific interests.
• No "you might also like" surface. Discovery happens through category browsing or following specific publishers, not algorithmic recommendation.

Bottom line

• WeSearch's home feed is chronological and identical for every reader. No algorithm is making per-reader ranking decisions.
• Categorization, source filtering, and keyword push give you control over what reaches you, but the sort is always time.
• The trending row is a count, not a model — it surfaces what readers are reacting to without applying engagement-prediction.
• If you want algorithmic personalization, this isn't the right tool. If you want a structurally accountable feed, it is.

Frequently asked

Frequently asked