How ThinkingNews works
Every claim on this page is checkable against the site itself. If something here does not match what you see, that is a bug and we want to know.
One story per event
Most aggregators show you one card per article, so a single event appears ten times from ten outlets. We group reports that describe the same event into a single story.
Grouping is done by comparing the meaning of each headline and summary, not by matching keywords. Two reports join the same story only when they are close enough in meaning and were published near each other in time — a story stops accepting new reports 36 hours after its first one, so that unrelated events on a recurring subject do not collapse together.
The account is written across every report
Once two or more newsrooms have covered an event, a language model writes a short account from all of their reports rather than from any single one. It is instructed to use only facts present in those reports, to prefer facts more than one outlet states, and never to add figures, dates or names that do not appear in the source material.
Where that account has not been written yet, the card shows the lead outlet’s own summary instead and says so, in place, rather than presenting one newsroom’s reporting as a collective view.
Every report is named and linked
Each story page lists every outlet that covered it, in the order they filed, with the time of filing and a direct link to the original. We do not reproduce full articles, we do not paywall-bypass, and we do not claim authorship of anyone else’s reporting. The reading happens on the publisher’s site; our contribution is the grouping and the cross-source account.
Translation
Story headlines and accounts are translated by machine into eight languages. Translated pages say so. Original headlines in the “how it was reported” list stay in the language they were published in, because those links lead to the publisher.
What this method gets wrong
Automated grouping sometimes puts two related-but-different events in one story, and automated summarising can flatten a nuance the original reporting was careful about. Neither is rare enough to pretend otherwise.
We do not report, interview or verify independently. Nothing here is a substitute for the original journalism, and where an account matters to you, follow the links and read the outlets themselves.
Corrections
If a story is wrong, mis-grouped, or attributes something to an outlet that did not report it, write to contact@thinkingdbx.com and we will correct or remove it. Our corrections policy explains what happens next.