About ovr.news
What ovr.news shows
You don't control your brain, but you do control what you expose it to. ovr.news is a lens machine for seeing what works in the world. Each of its five lenses looks for one kind of story about what is working.
-
Thriving: People's lives getting better: health, safety and circumstances improving.
How Thriving is scored -
Belonging: Communities holding together: bonds, rootedness, living culture and heritage.
How Belonging is scored -
Recovery: Nature healing: ecosystems recovering, species returning, nature newly found, protection in force.
How Recovery is scored -
Solutions: Technology, policy and initiatives in use, with results to show.
How Solutions is scored -
Discovery: Knowledge recovered that changes how we see the past.
How Discovery is scored
A story can pass more than one lens. It is shown under the most specific one it passes. The overall method, and its limits, is on how we select stories.
Each story is sourced, dated, and linked back to its publisher. The lens tells you what kind of evidence you're looking at. You see the world differently because you have a different instrument for looking.
We don't want to be where you stop. Our summary is a way in, not the story itself. Every article links to the publication that reported it, and we'd rather you went and read them. Check what we left out. Disagree with us. A lens is worth having because you can look through it and then look past it, not because it saves you the trouble.
How we're built
ovr.news runs on open-source software and local AI. Our pipeline tracks over a thousand news sources, and stories from 175 or more of them reach the site on a typical day. Articles are scored by small models that were trained on labels from a larger cloud model, and summarized by open-source language models. Both run on local hardware. We count page views and clicks without identifying anyone.
What gets in
A story gets in when one of our scoring models rates it above the bar for a lens, for example clear evidence that a species is returning. Editorial rules then remove duplicates and stories we have already shown.
AI and people
No one reads each summary before it goes up. Every headline and summary is written by AI and labelled as such. We check samples by hand and with other AI models, and we fix or remove what readers flag.
Money
No ads, no paywall, no tracking cookies. For now the maker pays the costs. Later we will ask readers who value it for donations.
We are transparent about our methods. See how we work and our accountability page for the details.
Get in touch
Questions, feedback, or ideas? Email info@ovr.news. Publishers can find information about content use and opt-out on our publisher page.
Based in The Netherlands.