← Back to editorials
✍️

We measured how rare good news actually is. It's worse than you think...

By Rik Van Bruggen · 13 July 2026, at 05:49

We measured how rare good news actually is. It's worse than you think...

I want to show you a single number, and then spend a few paragraphs on why it kept me up a little.

Sixty-two percent.

That's the share of "source-days" in our data on which a news outlet published exactly zero clearly-positive articles. Not "a bit negative". Not "mostly grim with a silver lining". Zero. Nothing that our scoring model looked at and said, yeah, that's genuinely good news.

Let me be precise about what that does and doesn't mean, because it's easy to over-read. It does not mean that on 62 percent of days there was no good news anywhere - across all our sources combined, almost every single day had at least some. What it means is narrower, and in a way more telling: take any one outlet on any one day, and about six times out of ten, that outlet ran nothing positive at all. The good news exists, but it's scattered so thinly that no single publication reliably carries it.

Grid of 100 squares where 62 are grey and 38 are green. Each square is one outlet on one day; 62% carried zero clearly-positive articles

Let me back up and explain where that comes from, because I think the method matters. For the last few months, every article coming through Positron's feeds gets read by an AI model and scored from 1 to 10 on positivity. Seven and up is "clearly positive". I took 91 days of that - 37 news sources, more than 77,000 articles, from all over Europe and North America - and just started counting.

The headline stats look like this. The average positivity across everything is about 7 percent. Sounds low already, right? But the average is lying to you a little, because it's being propped up by a handful of dedicated good-news sites. The number that actually tells the truth is the median, and the median is zero. Zero percent. If you line up every outlet-and-day combination and walk to the middle of the queue, the one you land on contains no clearly-positive articles at all.

Now, I want to be really careful and fair here, because it would be easy to twist this into doom, and doom is exactly what we're trying to push back against. So two honest caveats.

First: a low positivity score does not mean the news is all negative. Our model sorts things into positive, neutral, and negative. A newspaper sitting at 1 percent positive is not 99 percent catastrophe - most of the rest is neutral. Factual. A council meeting, a match report, a quarterly result. So the picture isn't "the world is on fire", it's more like "the world is beige, with occasional flashes of red, and very rarely a bit of green".

Second: these scores come from an AI. It's consistent, but it has its own idea of what "positive" means, and it's not infallible. I'm not claiming three-decimal precision here. But the pattern is so strong, and so consistent across dozens of very different outlets, that I'm confident the shape of it is real.

And the shape is this: good news is not rare like a nice day is rare. It's rare like a four-leaf clover is rare. On most days, in most outlets, there simply isn't any.

Here's the bit that got me, though. I also looked at the sites whose entire reason for existing is uplifting news - the Positive News, the Good News Network. Surely they're at 90-odd percent? Nope. Even they clear the "clearly positive" bar only about half the time. Half! And that's not a criticism of them - they do lovely work. It's a measure of how demanding reality is. When a team of people whose whole job is finding good news still comes up with a coin-flip, you realise how much of what happens in the world genuinely resists a positive framing.

So what do I take from all this? Honestly, not despair - the opposite. This is more or less the whole thesis behind Positron in one spreadsheet. If good news is a one-or-two-percent trickle scattered across dozens of feeds, then no single person is ever going to stumble across enough of it to feel any different about the world. It's too thin, too spread out. You'd have to read everything to find the good bits, and nobody has time to read everything.

That's the job we've quietly given ourselves. Read everything, so you don't have to, and hand you back the two percent. Concentrate the green.

Turns out that trickle is even smaller than I assumed when we started. Which, weirdly, makes me more convinced the work is worth doing, not less.

There's more buried in this data that surprised me, and I'll share the best bits as I keep digging. But this was the finding I most wanted to start with.

Cheers

Rik
Share