How this works

We collect official RSS feeds on a schedule and store them. For a few outlets that no longer publish a public feed, we use a Google News listing and keep only headlines that open on that outlet’s own website. Yahoo News has its own feed, which also reprints other papers; we only keep stories hosted on Yahoo. This website reads the latest saved edition. A separate collector updates that edition on a schedule. A page request does not refresh feeds. Source health shows unavailable and incomplete feeds; a missing feed is not evidence that an outlet ignored a story. We do not pull full articles onto our pages.

When several headlines are about the same event, we put them in one cluster. You see source excerpts about that event, then the original headlines grouped as left, center, and right. Each headline links out to the publisher. We do not reprint full articles.

Stories are listed as National or World. National is news about the United States. World is news about other countries, grouped by the country the events are about, not by where the publishing outlet is based. Those labels are not a bias score. If a country is not clear, we use World · International.

Each story page separates shared facts and source excerpts from where coverage differs. Matching headlines does not establish factual agreement. Each excerpt is labeled Source excerpt and shows its publisher, publication time, and original link. Examples from one outlet do not stand for every outlet on that side. Write-ups are extractive, not generated prose. They do not pick a winner or tell you who to believe.

We label the outlets in our feed as left, center, or right. Those labels describe the outlet, not every reporter who writes for it, and they are not a score for whether a story is true. We assign each label ourselves when we add an outlet to the feed. The labels live in our own list; we do not import a licensed bias scoreboard. We report headline counts and distinct-outlet counts separately. Several updates from one publisher count as one outlet. The left, center, and right outlet totals count distinct publishers on each side. Neither number rates the event itself.

We look at six desks: FactCheck.org, PolitiFact, AP Fact Check, Snopes, AFP Fact Check, and Reuters Fact Check. When one of them has checked a claim, we show their rating in their own words and link to their work. If no matching check is found in the six monitored desks, we say that. It is not a verdict and does not mean a claim was never checked anywhere. Attached checks name the specific claim and publication date; their wording is not a verdict on the whole cluster. Retrieval and matching can miss checks. We do not invent a verdict. We do not average the six desks. Snopes uses its official fact-check RSS feed, not its general news feed. We keep only Snopes-hosted fact-check links and read its published rating when the feed states one. If that rating cannot be read, we leave it unset and link to the source. AFP Fact Check first tries its official English RSS, with Google News as a fallback when that feed is unavailable. Reuters Fact Check uses Google News because no working official public RSS is available. Only English factcheck.afp.com articles and Reuters URLs under /fact-check are retained. These desks are separate from news outlets and do not count toward news coverage totals. Unavailable feeds, unresolved links, or unreadable claim metadata are disclosed on the Sources page; missing ratings are left unset.

Sometimes one side publishes a lot of headlines about a story and another side publishes almost none. We call that a coverage gap. In our feed, we flag a gap when a cluster has at least six headlines, one side has at least five, and another side has zero or one. That is a rule about this feed, not a claim about the media as a whole. The Coverage gaps page lists those stories and says which side is thin in our set. This rule still uses headlines, not distinct outlets. A separate Not found in our collection list shows monitored news outlets with no matched headline. Feed failures, limits, delays, and matching errors prevent any conclusion about those outlets’ actual coverage.

When public information exists about who owns an outlet, we say so and we link to the source. If we do not have a source, we leave it out. We do not guess.

Search and subject, country, and source-date filters cover the current collection only, up to 96 hours and at most 25 items per news feed. There is no historical edition archive without durable snapshot storage. Subject tags are conservative and may be unset. Most widely covered ranks the current collection by distinct outlets, not readership or importance. Publication times and collection times are separate, all shown in Mountain Time. A refreshed cluster is not necessarily new reporting, and we do not retain a first-collected history.

Documents behind the story appears only when an actual URL in a source item points to a recognized official document host. We do not guess document links. An official statement establishes what an institution said, not that every claim is true. Print views include excerpts, links, times, and correction notes, never full publisher articles.

Substantive corrections have a dated public entry, separate from routine feed refreshes. Report a problem prepares a copyable report; it does not send or permanently save it. The site does not collect contact information through that form.

Everyone can read every story, every coverage split, every coverage gap, and every fact-check. There is no subscription and nothing is locked behind a payment. We may show paid ads beside a story or at the bottom of the write-up. Ads never gate reading, and we do not take sponsors.

You can optionally get a free daily email of the clustered stories. The site itself stays free to read, and you do not have to join the list to browse, read stories, or use fact-checks. Donations are optional and separate.

If you want to help with the cost of running it, you can donate.

We do not reprint full articles. We do not use someone else’s licensed bias scoreboard. We do not add fact-check desks beyond FactCheck.org, PolitiFact, AP Fact Check, Snopes, AFP Fact Check, and Reuters Fact Check. We do not take sponsors, and we do not mark stories as brought to you by a sponsor. This version is a website, not a phone app.