How Since Then groups coverage and rates outlets

Since Then rates outlets, not individual articles. A source rating reflects an outlet's editorial tendency over time. Readers use the rating to know what leaning they are reading from before they begin, not to judge any single story in advance.

Since Then's editorial position is that the framing of news should be visible. Coverage is shaped by which stories an outlet chooses to run, where they are placed, and the language used to describe them. Naming that shape is part of the reading.

Since Then groups reports of the same event into one update. An update's page lists every report in it, oldest first, with each outlet's own headline, the short description from its feed when it has one, and a link to the publisher. We collect headlines and those descriptions, not full articles.

Where an update is shown as a single line (on the homepage, under a story's new coverage, and as the title of its own page), Since Then shows one of those original headlines, word for word, with its outlet named. A fixed rule picks it: a plain news headline before an explainer or analysis title, and both before live-blog labels, shouted words, exclamations, closing questions, very short headlines and loaded words; then outlets nearer the middle of the ratings; then a headline several outlets carried word for word; then the earliest. The rule never rewrites a headline and never hides an update, and every headline stays on the update's page. A story's full record lists each update under the headline our grouping first gave it.

Nothing on an update's page is written by Since Then. A story's summary or overview, where it has one, is reviewed by an editor before it appears, cites a report for what it says, and shows the date it was reviewed.

Since Then groups reports of one event into an update. When several updates turn out to be the same continuing matter, they are grouped again into a story, and the story is what the site follows over time.

Updates join a story on the language and entities they share with the updates already in it. Each story carries a name and a short description of what it covers, written from that coverage.

Every count of coverage on Since Then, on an update, a story, a chart or a People & topics page, counts sources: the news outlets that carried the report. An outlet counts once however many pieces it ran, and an outlet that ran a wire story, such as one from the Associated Press, counts as carrying it, because its readers got the story from it. Outlets in one ownership or syndication group count as one source on each side of the spectrum they appear on, so a chain whose papers carry two different ratings counts once on each of those sides. Only outlets the graphs count are placed on a side; section 12 explains which those are. An update's page lists every report, whatever its outlet's rating.

A source count says who carried a report, not who reported it themselves. Identical headlines from several outlets usually mean shared wire copy, and an update's page groups them and says they are not independent confirmation.

Each bar in a story's coverage chart is one week, and its height is a share: the sources on the left-right scale that carried the story that week, out of the sources on that scale that published anything we hold that week. A share keeps weeks comparable when the number of outlets we hold changes, which it does (see section 07). The bar is split into left, center and right, folding lean-left into left and lean-right into right, and the three parts add up to its height. Weeks run Monday to Sunday in UTC, so a report published on a Sunday evening in US Eastern time counts in the next week. Under the chart on each story page, every week lists its counts, the sources behind them and their reports. A story's Coverage page also charts collected articles per week, a different measure, named as such.

A week with no bar means no outlet on our left-right scale published a report that belongs to this story that week.

It does not mean nothing happened, and it does not mean nobody reported it. An outlet outside our roster may have covered it, and a report may exist that our grouping did not place in this story. A gap is a fact about our holdings, stated plainly so it can be checked, and it is never filled in.

A story is ongoing while at least one of its updates is dated within the last three weeks. After that the story page says it has gone quiet, and names the day of its most recent update.

Going quiet is a label, not a removal. A quiet story keeps its page, keeps its place in the lists that hold it, and becomes ongoing again the moment another update joins it.

Since Then has crawled the outlets we track continuously from July 2026. Before that the record is reconstructed from archives, and the site is still working backwards through them. Today coverage on this site begins in October 2021. A reconstructed month holds between about 800 and 2,300 updates, against about 6,700 to 12,800 in the months collected live. The site holds nothing earlier than October 2021.

This matters when reading any record that crosses July 2026. The live crawl holds about two and a half times as many outlets a week as the archive does, so the charts draw each week as a share of the sources publishing that week rather than as a count. That puts the two periods on one scale. It cannot add back reports the archive never reached, so counts and update lists from before July 2026 still understate what the press published. A story whose first updates sit outside what the archive has reached will open later on the site than it truly did, and the site names the earliest update we hold rather than the day the matter itself first drew coverage, because only the first of those is a fact. Archive coverage varies by month and by outlet.

Since Then does not produce original bias ratings. Ratings describe outlets, not individual articles or reporters, and draw on independent rater organizations whose methodologies are public. Most outlets currently carry a single published rating; where more than one rater covers an outlet, their scores are aggregated:

• AllSidesMethodology ↗

  Blind-survey methodology with editorial review across multi-partisan teams.

• Ad Fontes MediaMethodology ↗

  The Media Bias Chart: content analysis scoring outlets on bias and reliability axes.

• Media Bias / Fact CheckMethodology ↗

  Editorial assessment combining wording analysis, sourcing review, and factual-reporting history.

Where these raters distinguish between an outlet's news desk and opinion desk,Since Then preserves the distinction on the source page. A reader can see, for example, that an outlet's news desk is rated centrist while its opinion desk is rated Lean Right, a distinction that a single-number summary would lose.

Raters do not always agree. Since Then reports the disagreement rather than hiding it, and assigns a confidence tier based on how the panel converges:

When raters disagree by more than one unit, the consensus pulls toward the more centrist label. The default is conservative in the editorial sense: Since Then would rather under-state a tilt than mis-state one.

Every rated outlet sits somewhere on this scale. The colors carry across the product (feed pills, spectrum bars, cluster tilt chips), so the scale stays visually consistent wherever orientation is reported.

Not every source maps cleanly onto a left–right spectrum. Wire services, fact checkers, foreign outlets, and aggregators carry an axis tag that flags their editorial function in addition to (or in place of) a political rating.

An outlet's rating comes from the raters, and Since Then shows it exactly as they give it. Whether that rating puts the outlet's coverage on the left, center or right of a coverage graph is a separate question. The graphs are meant to show how partisan American news covered a story.

Most outlets count at their rating. Two kinds count differently:

• Local TV newsrooms count as center. Their local reporting is not partisan coverage, even when the station carries a left or right rating.
• International outlets, and outlets that mainly cover entertainment, sports or technology, are not counted as left, center or right at all. Their reports still appear on every update's page, and a story's Coverage page counts them in their own group.

These are editorial decisions by Since Then, and each one is recorded with its reason in the outlet's history. They never change an outlet's rating. Every outlet's source page states how it counts on the graphs and why.

Why the rule exists: when we measured it, 58% of the partisan coverage among the outlets we track came from left-rated outlets. Nearly all of that lean comes from foreign outlets, local TV and entertainment outlets, which carry left ratings with almost no counterpart on the right. Among the other American outlets we track, left and right are close to even. Without the rule, most stories read left for reasons that have nothing to do with how partisan outlets cover them.

Any reader may dispute a rating from an outlet's source page. Disputes enter a public audit trail and are reviewed against current rater evidence. A dispute can result in (a) no change, (b) a re-fetch of rater scores, or (c) an updated label, in which case the change is recorded in the outlet's orientation history.

The strongest disputes cite specific coverage, raters, or evidence, but every submission is read.

• A five-point political spectrum is a simplification. Real editorial positions are multi-dimensional; Since Then reports one axis well rather than several poorly.
• A source rating is not an article rating. Some left-rated outlets publish centrist coverage; some centrist outlets publish polemics.
• Raters disagree. Since Then shows disagreement rather than collapsing it into false certainty.
• Outlets drift. Ownership changes, editor changes, and desk changes affect orientation. Ratings are periodically re-reviewed.
• No rating system is fully objective. Every rating here shows the raters and dates behind it, so readers can check the evidence themselves.

Since Then aggregates content from RSS and Atom feeds across hundreds of sources. Only headlines, summaries, thumbnails, and metadata are stored. Clicking any article takes the reader directly to the publisher.

The homepage and its recent developments are the same for every reader: nothing is personalized, and nothing is ranked for engagement. Following a story lists what is new in it on your Following page; following an outlet keeps it there too.

Any legitimate news organization may request inclusion by writing to support@sincethen.news. Sources across the full spectrum are included, among them outlets whose editorial positions Since Then's operators find objectionable, because completeness is part of the product. Sources may be removed for consistently publishing fabricated reporting or for operation by state intelligence services.