News for people, not algorithms.

If you've used a news app in the last decade, you've used a feed that an engagement algorithm reordered for you. The algorithm picks which headlines reach the top of the page based on what kept other readers in the app. The reader is the input; the engagement metric is the output; the news is the byproduct. This page is the opposite proposal.

The premise

News is most useful to a reader when it is (a) accurate, (b) recent, and (c) ordered consistently with other readers' views of the same period. The algorithmic feed undermines all three: it injects optimization-target stories that test well in the model rather than informational ones, it deprioritizes recency in favor of "evergreen" engagement, and it splinters the shared view of the period because every reader sees a different ordering. WeSearch is news ordered the old way: by publish time, deduplicated, identical for everyone.

What "for people" means

• The same news for every reader. One chronological feed, one shared view. You and another reader two timezones over see the same headlines in the same order at the same hour. More.
• Anonymous discussion. The conversation under each story is not gated by follower counts, real names, or persistent public identity. Most threads stabilize on substance because there is no follower-count incentive for outrage. More.
• Sources you can audit. The source list is public. You can see exactly which 700+ feeds the home is built from and decide for yourself whether the mix is balanced. Browse.
• Standards you can hold us to. Editorial standards are written down. Read.
• Privacy you can verify. No third-party trackers in the request layer; verifiable in DevTools. More.
• Funding without strings. Donations from readers, no advertisers, no investors, no premium tier. More.

What "not algorithms" doesn't mean

We're not Luddites about machine learning. We use AI for two narrow purposes: a clearly-labeled 3–5 sentence TL;DR on each story page, and a clearly-labeled daily editorial note. Neither affects feed ordering. Neither personalizes. Both can be flagged and corrected. Used surgically, AI is a useful tool. Used to rerank the news, it is the problem.

The honest trade

You will, on a chronological feed, sometimes scroll past stories you don't care about. You will, occasionally, see headlines that other readers in your political tribe have decided to ignore. Your feed will not be optimized to keep you in the app. Those are the costs.

The benefit is that what you read is what is happening, in the order it happened, across many sources at once, without a third party deciding which version of "the news" you get. We think that trade made sense for fifty years of newspapers and we think it still does.

Who this is for

Readers who already have an opinion about the algorithmic feed and want to read without it. Readers who pay for several news subscriptions and are tired of the fragmentation. Readers who lurk on Reddit's news subreddits but feel the platform politics weighing on them. Readers who used Google Reader, miss it, and never quite trusted Feedly's premium drift. Readers who care enough about news to scroll past a few headlines they didn't care about in exchange for the assurance that nothing was hidden from them.

The deeper claim

The argument here isn't aesthetic. It's that algorithmic-feed news products at scale produce systematically worse-informed readers than chronological ones do — that the engagement-prediction layer, accumulated across millions of readers and many years of training data, slowly bends the news ecosystem toward the headlines that test well in the model rather than the ones that are most informative. The bend is gradual; individual readers don't notice it at any single moment; but the population-level effect on public knowledge is measurable in the gap between what a careful chronological reader knows and what a typical algorithmic-feed reader knows about the same week's events.

WeSearch can't reverse that drift at the population level. It can offer individual readers a feed without the bend, and let them notice for themselves how different a chronological view of the same period feels.

Common pushback, addressed

• "Chronological feeds are inefficient." True in the sense that you'll see headlines you wouldn't have clicked on. False in the sense that engagement is the right efficiency metric for news.
• "Algorithms can be designed for informedness, not engagement." In principle yes; in practice, the platforms that have actually shipped have all converged on engagement metrics. We're skeptical of the principle until we see a working example at scale.
• "You're using AI for TL;DRs, isn't that hypocritical?" No. AI used to summarize a single story is a tool; AI used to rerank what reaches readers is the mechanism we're objecting to. The difference matters.
• "Won't readers just go back to algorithmic apps?" Some will. Some won't. The point isn't to convert everyone; it's to make the chronological alternative available for the readers who want it.

Bottom line

• WeSearch is structured around the premise that news belongs to readers, not engagement metrics.
• Every architectural choice (no algorithm, no tracker, no paywall, no follower count, no real-name requirement) flows from that premise.
• The trade-offs (information overload, no personalization, smaller audience) are the costs of that commitment.
• The alternative — algorithmic news that promises efficiency and delivers a slow drift in what news is — is the system we're refusing to build.

Frequently asked