Why I Stopped Tracking Five News Sites' Most Read Lists
By Rene Roost | Published: October 7, 2026
This site ranks news stories on one kind of evidence: each publisher's own Most Read list. The longer and higher a story sits on its list, the higher it ranks here. In August and September 2026 I stopped tracking five of those lists — The Dispatch, France 24, AP News, TMZ and Page Six — moved The Sun to a different desk, and closed the Celebrity desk. On their own sites the five lists were headed "Most Read", "Most Popular", "Trending Now" and "Popular With Members"; for short, I call them all Most Read lists.
None of these decisions is a verdict on anyone's journalism. Four of the five lists went because of one discovery: every publisher builds its Most Read list its own way. One publisher's list is a morning digest of yesterday, another's ranks a window that trails publication by a day, a third keeps the same few stories for days. They look alike on the page, but recorded around the clock they behave very differently, and only some of them keep pace with the day's news. Those are the lists this site's feeds need, and the ones I keep. Page Six went for a different reason: on its own, it couldn't carry the Celebrity desk. This report sets out the data behind each decision; the Q3 overview compares the lists that stayed. How the recording and ranking work is on the methodology page.
At a glance
| Source | Desk at the time | Tracked | What the data showed |
|---|---|---|---|
| The Dispatch | Politics & Policy (now US Politics) | Jul – Aug 2026 | A slow shelf: stories stayed about three days, most arriving a day or more after publication |
| France 24 | World News | Aug 2026 | A morning digest of yesterday: every story reached the list more than a day after publication |
| AP News | World News | Jan – Sep 2026 | A list a day behind: stories arrived about 20 hours after publication, in every month |
| TMZ | Celebrity | Aug – Sep 2026 | A list a day behind: 26 hours late, then up to two days at #1 |
| Page Six | Celebrity | Aug – Sep 2026 | A healthy list with no desk it fits |
| Celebrity desk | — | closed Sep 2026 | Left with one list that matched its name |
How the site uses a list
Each desk's page opens with its main feed, headed Most Read Right Now (the right-now feed, for short): stories published in the past 24 hours (48 on the Tech, Science and Asia desks), ranked by their lists. It's the part of the page readers see first; earlier days follow below it, one day at a time. The same rankings also accumulate into category archives — This Week, This Month, each quarter — where every source competes inside the same topic list, whatever its desk.
A list serves both only if its stories arrive within hours of publication and move on when readers do, and each retirement fell short in its own way. AP News, TMZ, France 24 and The Dispatch were late: their stories reached the list after they had aged out of the right-now feed, and the slowest then held top places in the archives far beyond their weight. Page Six was a sound list that lost its desk when the Celebrity desk closed.
Four kinds of Most Read list
| Kind | How it behaves | Examples | Can it drive a right-now feed? |
|---|---|---|---|
| Rolling window | Ranks recent reading continuously; stories arrive within hours and leave within a day or two | Most lists on the site — the BBC, Al Jazeera, The Hill | Yes |
| Morning digest | Rebuilt about once a day from yesterday's reading | France 24 (06:00–08:00 Paris), The Dispatch (00:00–04:00 US Eastern) | No: a day late by design |
| A list a day behind | Updates all day, but ranks a window that trails publication by about a day | AP News (~20 hours), TMZ (~26 hours) | No: stories arrive as they age out of the right-now feed |
| Slow shelf | Keeps the same few stories for days | The Dispatch | No |
A fifth kind turned up among lists I still track: lists with no popularity order at all, The New York Times's 20 most-viewed stories and TechCrunch's Most Popular box. The methodology explains how I handle both.
The numbers
Each retired list's last 30 days on the site (France 24: its 26 days), against the other lists on its desk over the same days. The last column is the share of each list's recorded hours that fell within 24 hours of the story's publication: how much of its presence the right-now feed could show.
| Desk and source | New stories a day | Median stay (hours) | Median hours from publication to list | Listed more than a day after publication | Time on the list inside the 24-hour window |
|---|---|---|---|---|---|
| World News | |||||
| France 24 (4–30 Aug) | 5.3 | 42 | 34 | 100% | 0% |
| AP News (12 Aug – 11 Sep) | 9.2 | 12 | 20.5 | 24% | 37% |
| The New York Times¹ | 18.2 | 20 | 5.0 | 2% | 61% |
| The Guardian (World)² | 16.2 | 6 | 2.0 | 0% | 88% |
| BBC | 28.6 | 6 | 1.8 | 1% | 92% |
| Al Jazeera | 25.2 | 5 | 1.3 | 0% | 95% |
| US Politics (31 Jul – 30 Aug) | |||||
| The Dispatch | 1.8 | 72 | 48 | 80% | 1% |
| Politico (US) | 5.8 | 19 | 12.3 | 16% | 48% |
| Intelligencer | 2.9 | 23 | 10.9 | 6% | 28% |
| The Hill | 24.7 | 5 | 1.1 | 0% | 85% |
| Celebrity (27 Aug – 26 Sep) | |||||
| TMZ | 3.2 | 42 | 26.6 | 100% | 0% |
| Page Six | 9.9 | 8 | 2.9 | 0% | 96% |
| The Sun | 33.0 | 5 | 2.9 | 1% | 91% |
¹ The New York Times's list has no popularity order: its figures are the time to enter, and the time spent in, its top 20. ² The Guardian's figures use its World news list only; see the methodology. World News peers are shown for AP News's window; over France 24's window, each one's median hours from publication to list is within a tenth of an hour of the figure shown.
The Dispatch: a shelf, not a list
The Dispatch's list held five stories, and they barely moved. In its last month on the site it took in fewer than two new stories a day, and the typical story stayed 72 hours before leaving. Its five longest-running #1s each held the top place for between 95 and 100 hours: "Trump's 'Rigged Election' Claims Are Short on Evidence, Again" spent 100 of its 106 hours on the list at #1.
The stories also arrived late. Four in five reached the list more than a day after publication, a median of 48 hours, and new entries came in a batch: of the 113 stories that entered the list after I began tracking it, 74 did so between 00:00 and 04:00 US Eastern time.
That combination did damage twice over. In the right-now feed The Dispatch was almost absent: 1% of its time on the list fell inside the 24-hour window. In the category archives it was everywhere. In US Politics that month it held six of the ten top places, including the top two, with 26 of the category's 1,188 stories. The Hill, whose list took in about 25 new stories a day to The Dispatch's two, held none. On a list that takes in two stories a day, a long stay says more about the shelf than about how many people are reading.
France 24: yesterday's news, every morning
France 24's Most read list was rebuilt once a day. Of the 137 stories that entered it while I tracked it, 109 arrived between 06:00 and 08:00 Paris time. Every one of them reached the list more than a day after it was published — the median was 34 hours — and then stayed: the typical story lasted 42 hours. Its longest-running stories each arrived about a day after publication and held #1 for 48 hours. "Clamping down: China tightens exit rules" was published on 18 August, reached the list 25 hours later, and spent 48 of its 72 hours there at the top.
A morning digest of yesterday is a sensible product for a broadcaster's website. For a feed of the past 24 hours it is a day late by construction: none of France 24's time on the list fell inside that window. The monthly archives told the opposite story. France 24 held 5.0% of all category top-ten places with 1.6% of the stories, and five of the top ten in Middle East with 34 of that category's 478 stories. The desk's other lists — Al Jazeera, the BBC, The Guardian — take on a typical story within about two hours, and each held fewer top places than its share of stories.
AP News: always a day behind
AP News was on the site longer than any of the others: 241 days. Its list was the problem, and only in one way. It moved at an ordinary pace: the typical story stayed 12 hours, mid-table for the catalogue. But stories reached it a median of 20.5 hours after publication, and from April on only one or two in a hundred arrived within six hours. That never changed:
| Month (2026) | Stories | Median hours from publication to list | Listed more than a day after publication |
|---|---|---|---|
| January | 103 | 20.5 | 34% |
| February | 128 | 19.0 | 32% |
| March | 129 | 17.8 | 21% |
| April | 143 | 20.1 | 27% |
| May | 212 | 21.2 | 36% |
| June | 214 | 20.9 | 31% |
| July | 237 | 21.0 | 35% |
| August | 253 | 20.6 | 24% |
| September (to the 11th) | 71 | 20.1 | 24% |
A story that reaches its list 20 hours after publication has four hours left in a 24-hour feed. So AP was one of the most recognised names on the site and one of the least seen at the top of its desk: 37% of its time on the list fell inside the window, against 88–95% for the BBC, Al Jazeera and The Guardian beside it. Unlike the slow lists, AP didn't crowd the archives: its share of category top-ten places was almost exactly its share of stories (1.1×). Its presence was simply spent after the right-now feed had moved on. Even its five longest-running #1s were exceptions to its own pattern: each reached the list between 6 and 16 hours after publication.
AP was the hardest call, because its lateness is milder than TMZ's or France 24's and some lists I kept are slower (see below). What decided it was the desk. World News is the site's front page, and three of its lists cover the same beat and take on a typical story within two hours.
AP News appeared in the Q2 reports' comparison tables, where its record was patchier than any other source's. That isn't why AP left: its lag was the same in the months when its record was complete.
TMZ: first to the story, a day late to the list
TMZ broke stories first and ranked them a day later. Four stories from its last week on the site:
| Story | TMZ published (UTC) | Reached TMZ's own list | First on another outlet's list |
|---|---|---|---|
| Presley Gerber dies | 21 Sep, 02:42 | 26 hours later | ABC News, about an hour after TMZ published; Page Six and The Hollywood Reporter within about two hours |
| Chad Gilbert dies | 20 Sep, 19:12 | 26 hours later | NBC News, about seven hours after TMZ published |
| Jennifer Eckhart dies | 21 Sep, 16:08 | 26 hours later | New York Post, about five hours after TMZ published |
| Hayden Panettiere's cause of death | 22 Sep, 14:18 | 27 hours later | Page Six and Deadline, within the hour |
TMZ published first on three of these and within minutes of the others on the fourth. Those publication times agree with when other outlets carried the same stories, so the delay was in the list: TMZ's Most Popular list is consistent with a ranking window that trails publication by about a day. In its last 30 days every one of its stories with a known publication time reached the list more than a day after publishing, a median of 26.6 hours. Its five longest-running #1s each arrived 25 to 27 hours after publication and then sat at #1 for their entire 46- or 47-hour stay.
That is the paradox of a late list in its purest form. In the right-now feed TMZ was invisible: none of its time on the list fell inside the 24-hour window. In the monthly archives it was loud: 5.1% of all category top-ten places with 0.9% of the stories, and five of the top ten in Deaths & Tributes, including #1, with 12 of that category's 172 stories. The same day of delay, followed by two days at #1, produced both.
Page Six: a healthy list with no home
By every measure in this report, Page Six's list worked well. Stories reached its list a median of 2.9 hours after publication, none arrived more than a day late, and the typical story stayed eight hours: a well-behaved rolling list. It held fewer top places than its share of stories, not more (0.2×).
It left for a different reason: fit. By primary category, 84% of its stories in its last month were celebrity news and gossip. The desk built for that couldn't continue once The Sun moved (below), and none of the others fit it. The Hollywood desk is a trade desk, whose lists are mostly entertainment-industry news, with celebrity stories between a sixth and a bit over a quarter of each list; Page Six's own parent paper, the New York Post, puts one celebrity story in eight on its list, and none of the Post's list entries in that month linked to Page Six.
| Source (desk) | Celebrity | Entertainment | Everything else |
|---|---|---|---|
| Page Six | 84% | 7% | 9% |
| Variety (Hollywood) | 28% | 56% | 16% |
| The Hollywood Reporter (Hollywood) | 26% | 55% | 19% |
| Vulture (Hollywood) | 25% | 67% | 8% |
| Deadline (Hollywood) | 17% | 61% | 22% |
| New York Post (US) | 12% | 0% | 88% |
Share of each list's stories by primary category, 27 August – 26 September 2026.
Moving Page Six onto the Hollywood desk or the US desk would have put celebrity gossip among trade or general news, and a source can't sit on the site without a desk. The stories it covers haven't gone anywhere: the same deaths, splits and court cases reach the site through more than 30 other lists, as the Presley Gerber story did within hours.
Closing the Celebrity desk
The Sun's list covers the whole paper. Only three in ten of its stories were celebrity news; most of the rest were crime, accidents and sport, a mix closer to the UK desk's lists than to Page Six's or TMZ's, so The Sun fits the UK desk better and moved there. With TMZ a day late, that left Page Six as the only list the Celebrity desk's name described, and one list can't carry a desk. Neither People.com nor the Daily Mail offered a Most Read list I could use in their place, so the desk closed.
The celebrity categories remain. In the 30 days before the change, Page Six and TMZ supplied about three in ten of the stories in Celebrity News & Gossip, and about a third of the stories that led it on a given day. The rest came from more than 30 other lists — The Sun above all, then Variety, The Hollywood Reporter, Vulture, Deadline, the BBC and Metro — and those categories now draw on them alone: the same events, reported by tabloid, trade and mainstream outlets.
Invisible today, loud next month
Put the two measures side by side and the pattern behind three of the retirements is plain. A list that runs a day behind is absent from the right-now feed and over-weighted in the month view, because the same delay produces both: its stories arrive after the right-now feed has moved on, then hold #1 for days while the monthly archive keeps counting them.
| Source | Time on the list inside the 24-hour window | Share of stories | Share of category top-ten places | Over-representation |
|---|---|---|---|---|
| The Dispatch | 1% | 0.6% | 5.0% | 8.8× |
| TMZ | 0% | 0.9% | 5.1% | 5.8× |
| France 24 | 0% | 1.6% | 5.0% | 3.2× |
| AP News | 37% | 2.4% | 2.6% | 1.1× |
| Page Six | 96% | 2.8% | 0.7% | 0.2× |
Each source's last 30 days. In each category, the top ten are the ten highest-ranked stories published in the window, as a This Month archive ranks them; 1.0× means a source's share of top places equals its share of stories.
AP was only late: its stories arrived after the right-now feed had moved on but didn't linger once there. Page Six was neither, which is why its reason was different.
Size made it worse for The Dispatch and France 24. A Most Read list ranks a site's stories against each other: it shows which stories that site's readers chose, not how many readers the site has. By Similarweb's traffic estimates, The Dispatch's website gets a small fraction of The Hill's or Politico's visits, and France 24's, counting all its languages, a fraction of Al Jazeera's. So their stories didn't only hold top places for days; they held them on lists with far fewer readers behind them. In the US Politics archive that month, The Dispatch's two top stories looked like the most-read political stories of the month, and nothing from The Hill made the top ten. Similarweb's estimates make it likely that both had far fewer readers than The Hill's biggest stories. TMZ is a big site in its own right: its 5.8× came from the long stays alone.
Aren't some lists you kept just as slow?
A few are slow, but none is in AP's position, beside three faster lists covering the same news on a 24-hour feed.
The i Paper is late for a different reason. Its late arrivals are mostly features and advice — "I'm a 54-year-old man without any male friends", "I'm a divorce lawyer – this is why you should stay single (for now)", "We built a £13,500 garden office" — that keep drawing clicks for days or weeks after they were written. A day-late news story has already been ranked elsewhere; a feature like these has no substitute on the site. Since The i Paper joined, it has supplied six of the UK desk's eight Family & Relationships stories and ten of its thirteen in Personal Finance.
The Japan Times rebuilds its What's trending list about once a night, the France 24 pattern, but the Asia desk runs a 48-hour feed, which catches 81% of its stories.
ScienceAlert is slower than its neighbours on the Science desk, but two in three of its stories reach its list within a day, well inside that desk's 48-hour feed.
TechCrunch is slow even for the Tech desk's 48-hour feed, but its startup and funding stories appear on no other list the site tracks, and it is one of the desk's two lists without a paywall, with Ars Technica.
Couldn't you just weight them differently?
No amount of re-weighting moves a story's publication time back inside the day, and no weighting changes a card that says "#1 on TMZ for 2 days" when those two days began a day after the story broke. In each case the problem was what the list measures, not how much it counts.
Is this a judgement of their journalism?
No. AP and France 24 are among the most respected newsrooms in the world, and TMZ published first on three of the four stories above. Each retirement is about the list a publisher exposes and how it meets a site built around the day's news — and, for Page Six and The Sun, about which desk a reader would expect to find them on. The events they cover still reach the site, through other lists.
About these numbers
- Windows. Each retired list is measured over its last 30 days on the site (France 24 over its 26 days), and its desk's other lists over the same days. Retired lists' records come from the readings I kept when each was retired; none of them is on the site any more.
- Median stay is the same measure as in the Q3 overview, and the Q2 reports' Median Duration: the hours a story was directly recorded on the list. Stories already on the list when a window opened, or still on it when tracking stopped, are left out of duration figures.
- Hours from publication to list run from the publisher's publication time to the story's first appearance on its list; stories without a known publication time are left out.
- Time on the list inside the window is the share of a list's recorded hours that fell within 24 hours of each story's publication (48 on the Tech, Science and Asia desks).
- Category top-ten places use each story's primary category, which a language model assigns. Sorting stories into categories and tagging them is all a model does on this site; it has no part in ranking them. Retired lists are ranked alongside everything else, on today's ranking method.
- Site sizes are Similarweb's estimates of each website's visits. I have no traffic data of my own, and they compare whole sites, not stories.
- The New York Times and The Guardian: see the notes under the first table.
Metric definitions and the full data-collection method are on the methodology page.
Corrections: errors are fixed in place with a dated note at the foot of this page. Spotted something wrong? Get in touch.