I want to start with a confession, Herman.
That's a strong opening for a Tuesday.
Every morning I open Google News. Every morning. And every morning I get the same little drop in the stomach, like the world has been holding a grudge overnight and it's finally ready to tell me about it.
And yet you keep opening it.
That's the thing. That's exactly the thing. Daniel wrote in about this, and he's put his finger on something I think most of us feel and don't say out loud. He says checking Google News has been his daily way of finding out what's happening in the world for years, and more days than not it's a thankless, depressing exercise. He can't remember the last time he closed the app and thought, well, that was uplifting.
That's a good line.
It's his line, and it's the truest thing in the whole prompt. But here's what he does with it. He says amid all the geopolitical gloom, what he actually loves finding are obscure stories, weird happenings from around the world, and coverage of science and technology, because those give him a hopeful counterpoint. A lot of the prompts he sends us come from random tidbits he reads. So he's been meaning to find an aggregator that surfaces that stuff. His words: a fire hose of slightly off the beaten track news, with the occasional deep thought piece about what's actually happening in the world and how it could be better.
A fire hose of weird. I like it.
And then he gets to the darker suspicion underneath. He says day to day mainstream coverage is either not accurate, or at this point just not healthy to engage with. He wants something that learns his preferences over time and surfaces the things that would actually make his day a little brighter. And he asks, plainly, whether he's alone in this. Then: any recommendations for apps, websites, anything.
So it's a tool request wearing a philosophy coat.
It's both, and I don't think we get to pick. So let's take it seriously. Not just as a list of apps, but as a question about what news consumption has turned into and what it could be instead.
Here's the tension I'd start with. News has never been more accessible. You can read the Jerusalem Post, the Yomiuri Shimbun, and a wire report out of Lagos before your coffee cools. And at the same time, the default news diet has never been more aversive. Those two facts should be in tension, and they aren't, because access was never the bottleneck. The bottleneck is the emotional and cognitive cost of the default feed.
And that cost is measurable. This isn't Daniel being fragile.
Not remotely. The Reuters Institute Digital News Report has been tracking this for years. Around forty percent of people worldwide now say they sometimes or often avoid the news. In 2017 that number was twenty-nine percent. That's an eleven point jump in under a decade. This is a documented, global, cross-language phenomenon. It has a name. News avoidance.
Forty percent. That's not a fringe. That's a plurality of the planet deciding the fire hose is doing more harm than good.
And trust tracks it. Only around forty percent of people say they trust most news most of the time. So you've got a rough symmetry: four in ten avoid it, four in ten trust it. The other six in ten are reading something they don't believe.
Which is a strange way to spend a morning.
It is. But before we go further, I want to split Daniel's ask, because he's actually asking for three different things and I don't think he's fully separated them.
Go on.
One, obscure and weird stories from around the world. Two, hopeful coverage of science and technology. Three, occasional deep thought pieces about what's really happening and how it could be better. Those are three different needs. They have different supply chains. Weird geography is not the same product as a fusion breakthrough, and neither is the same as a three thousand word essay on the future of the city.
And the tools that serve one will not serve the others.
Correct. So the arc I'd propose: first, why mainstream news feels like this, structurally. Then what actually exists for each of the three needs. Then the personalization question, because that's where Daniel's instinct gets interesting and also where it gets dangerous.
Let's do it. Start with the why, because I think most people blame the wrong thing.
Most people blame journalists. And I want to be careful here, because there's a version of this conversation that's just a rant about bias, and I don't think that's the real mechanism. The real mechanism is older and duller. News values.
Define it. Because "news values" sounds like a media studies lecture.
It's simpler than that. It's the set of criteria that decides what counts as news at all. Novelty, conflict, threat, prominence, proximity. A plane that lands safely is not news. A plane that crashes is news. That's not a conspiracy, that's an economics of attention. The news hole is finite, and the editor has to fill it with the things that will make people look.
So the filter is doing exactly what it was designed to do.
Exactly what it was designed to do. And it was designed to do it before algorithms existed. Which matters, because the common story is that the internet broke the news. The internet didn't break it. The internet accelerated a bias that was already load-bearing.
Then what did change?
The front page editor had judgment. A human being sat there and decided that a story was important enough to run even if it wasn't going to sell. There was a civic component to the job. The algorithm has no concept of importance. It has a concept of engagement. Clicks, dwell time, shares. And it turns out the things that maximize engagement are outrage, fear, and threat. So the machine doesn't distort the news values. It optimizes them. It takes the plane crash bias and runs it at scale, twenty-four hours a day, personalized to you.
So the editor was a brake. A weak, human, inconsistent brake, but a brake.
And the brake is gone. That's the honest version of what happened.
Here's where I want to push, though. Daniel says mainstream coverage is either not accurate or not healthy. I think those are two separate claims and only one of them is really about accuracy.
Say more.
The accuracy complaint. Only forty percent of people trust most news most of the time. But I don't think the mechanism is that journalists are fabricating things. I think the selection filter itself distorts reality. If you only ever see what's going wrong, you form a picture of a world that is mostly going wrong. The world isn't that. The world is mostly people going to work and fixing things and not crashing planes. But none of that makes the feed.
That's the part I'd underline. The news is accurate in the narrow sense and false in the aggregate sense. Every individual story can be true and the composite can still be a lie, because the composite is a filter for failure.
A plane lands safely in Tel Aviv. Nobody writes it down. A bridge in Ohio holds. Nobody writes it down. The composite you build from the feed is a world made entirely of exceptions, and you walk around believing the exceptions are the rule.
And that's why the distrust is rational even when the reporting is honest. People can feel the distortion without being able to name it. They know the world doesn't feel like the feed, and they conclude the feed is lying, when actually the feed is just selecting.
Okay. So there's a name for the response to this, and I want to get it right, because I think it's the most useful concept in the whole episode.
Constructive journalism. And the misconception to kill immediately is that it means positive news. It doesn't. It means journalism that includes solutions and context alongside the problem. The Constructive Institute has done the work on this, and the finding is consistent: adding solutions and context to a story increases reader engagement and reduces feelings of helplessness without sacrificing accuracy. Same reporting, same rigor, different frame.
So it's not "here's a nice thing that happened." It's "here's the problem, and here's what people are actually doing about it, and here's what we don't know yet."
Right. And the effect on the reader is not that they feel good. It's that they feel like the story is about something that could change. Helplessness is the real cost of the default feed, more than sadness. Sadness you can carry. Helplessness is corrosive.
That's the actual disease. Not gloom. Gloom is survivable. The thing that wears you down is reading about a problem every day for ten years with no sense that anything moves.
And that's Daniel's third need, the deep thought pieces. Because the difference between news and analysis is exactly this. News is what happened today. Analysis is what it means and what could be done. Mainstream news is overwhelmingly the former. It's built for the daily cycle. The "what could be done" lives somewhere else, and it lives in a slower medium.
So the three needs map onto three different time horizons. Weird stories are the shortest, a moment of delight. Science and tech is medium, a trend you can watch. Deep thought is the longest, a decade-scale argument.
And the default feed only serves the shortest horizon, badly.
So that's the why. Now the what. What actually exists for someone who wants a different diet?
Let's take the three needs in order. Weird and obscure first. And I'll be honest, this is the one where the tools are the weakest, because weird is hard to define and harder to index. But there are things. Kagi News is the most interesting recent entry. It's a paid, ad-free aggregator that summarizes stories from multiple sources and, crucially, lets you adjust the mix. You can tune it toward or away from categories. It's the closest thing to Daniel's fire hose that actually ships.
Paid and ad-free matters more than people think.
It matters enormously. The entire distortion we just described is downstream of the ad model. If the product is attention, the feed optimizes for attention. If the product is the subscription, the feed can optimize for something else. Kagi is a subscription product, and that changes what it's allowed to show you.
What else?
Ground News does something different and useful. It takes a story and shows you how different outlets covered it, side by side, with a breakdown of the political lean of the coverage. That's not a weird-story tool. That's an accuracy tool. It directly addresses Daniel's "not accurate" complaint by showing him the spread rather than a single telling.
Which is a clever reframe. Instead of asking "is this outlet trustworthy," you ask "what's the range of tellings."
And the range tells you more than any single one. Then for the daily habit, 1440 is a free email newsletter that curates a balanced mix across categories. It's not weird, it's not deep, but it's a decent replacement for the reflexive Google News check, and it's short.
And for actual weird, the niche stuff. Atlas Obscura for geography and history oddities. Hacker News for tech-adjacent strangeness, if you can tolerate the comment section, which I cannot, because I read too slowly to keep up with it.
Nobody can keep up with it. That's the point of it.
Second need. Hopeful science and tech.
This is where the supply is good, and I want to be precise about why. Positive News is a UK publication explicitly built on constructive journalism. Reasons to be Cheerful was founded by David Byrne and is focused on solutions. But the ones I'd actually send Daniel to for rigor are Quanta and Ars Technica. Quanta covers mathematics and physics and biology at a level that assumes you're smart and doesn't assume you're a specialist. Ars does the same for computing and space.
And these aren't saccharine. That's the key point. A Quanta piece about a proof is not "positive news." It's just coverage of something that isn't a war.
That's the whole trick. The hopeful counterpoint doesn't come from optimism. It comes from coverage of domains where progress is real and measurable. Science and technology are domains where the trend line, over decades, is up. Not every year, not every field, but the direction is up. So reading about them is not escapism. It's reading about a part of reality that the geopolitical feed systematically omits.
Because a new cancer therapy working in a trial is not novel enough, not conflict enough, not threat enough. It doesn't make the front page. It's the single most important thing happening on the planet that day and it's page nine.
Third need. Deep thought pieces.
This is where I have opinions, because this is the stuff I actually read.
Go.
Aeon. Noema. Nautilus. Palladium. These are publications whose entire reason for existing is the essay about how the world could be better. Aeon publishes philosophy and science essays that assume you'll sit with them for twenty minutes. Noema is explicitly about the future of technology and society. Nautilus is science with a literary register. None of them are trying to tell you what happened today.
And Substack changed the economics of this. Individual writers can now publish exactly this kind of analysis directly to an audience that pays for it, without a masthead. The long-form reflective essay used to need an institution. Now it needs a writer and a mailing list.
Which is good and bad. Good because the writing exists. Bad because discovery is now on you. There's no front page for the deep thought section of the internet.
There isn't. And that's the honest answer to half of Daniel's question. The supply exists. The aggregation doesn't.
Okay. Now the part I actually want to fight about. Personalization.
I knew it was coming.
Daniel says he wants something that learns his preferences over time. That sounds obviously good. And I think it's the most dangerous sentence in the prompt.
Defend that.
Because "learns my preferences" is exactly what the current feed already does. The outrage feed is personalized. It learned that you click on certain things, and it serves you more of them. The mechanism Daniel is asking for is the same mechanism that produced the problem. He's asking for a better-tuned version of the thing that's making him miserable.
That's the sharpest point in the episode, and I want to give the technical version of it, because it explains why this is hard. News recommendation is harder than movie recommendation or music recommendation. Three reasons. One, the item space changes daily. A movie catalog is stable. Today's news is gone tomorrow, so there's no stable inventory to learn over. Two, the signals are noisy. You click on something out of duty, not interest, and the system can't tell the difference. Three, the cold start problem. A new user has no history, so the system has nothing to work with, and news is a domain where you can't afford ten bad sessions before it learns.
So the thing he wants is technically hard and philosophically fraught.
Both. And there's a fourth issue, which is the one the research keeps landing on. Selective exposure. Personalization reduces aversiveness, which is good, but it also reduces exposure to information you didn't ask for, which is sometimes the information you most needed. There's a real tension there and nobody has resolved it. You cannot have both a feed that only shows you what you like and a feed that keeps you informed about things you'd rather not know.
So the honest answer is you have to choose, and most people choose comfort, and then they're surprised they don't know anything.
Which is why I'd push Daniel toward a stack rather than a single app. And I want to be clear that no single tool does all three of his needs. That's the misconception to kill. Weird stories, hopeful science, and deep thought are three different products with three different supply chains. You need a stack.
Give me the stack.
Kagi News for the tunable daily mix. Ground News when a story matters and he wants to see the spread. 1440 or a similar newsletter as the morning habit replacement. Positive News and Reasons to be Cheerful for the constructive stuff. Quanta and Ars for the science. RSS feeds for Aeon and Noema, because RSS is still the only tool that lets you subscribe to a publication without an algorithm in the middle.
RSS. The oldest tool in the room and still the only one that does exactly what it says.
It's the only subscription model where the publisher doesn't get to decide what you see next. You subscribe, you get everything, you decide. That's it.
Which is, funnily enough, closer to what Daniel actually wants than any of the smart aggregators. He wants a fire hose he controls. RSS is a fire hose you control.
It is. It's just not a fire hose that learns. And that's the tradeoff. A feed that learns is a feed that decides. A feed that doesn't learn is a feed you have to steer.
So the real answer to Daniel is: you can have the weird, the hopeful, and the deep. You cannot have all three from one app. And the "learns my preferences" part is the part I'd talk him out of.
Or at least, talk him into being careful about. Because there's a version of personalization that works, and it's not "learn what I click." It's "let me set the dials." Kagi does this. You tune the categories. You're the one deciding. That's different from a system inferring your preferences from your behavior, because behavior is a terrible proxy for what you actually want.
Because you click on the thing that makes you angry and then you feel bad, and the system learns "he likes being angry."
And it's not wrong. You did click. That's the trap. Your behavior is not your preference.
Thirty seconds of silence for everyone who has ever refreshed a feed they hate.
Noted.
So where does that leave the "make my day a little brighter" part? Because that's the actual goal. Not information. Mood.
I think the mood part is real and I think it's underrated. The reason the feed makes you miserable is not that it informs you. It's that it informs you in a register designed to alarm. And there's a legitimate version of what Daniel wants, which is news that respects your attention and doesn't spend it on fear.
Which is a design choice, not a content choice.
It's entirely a design choice. Same facts, different frame. You can tell someone about a flood in a way that makes them feel informed or in a way that makes them feel doomed. The facts don't change. The frame does.
And the frame is what Daniel is asking for. He's not asking for a world without bad news. He's asking for a delivery mechanism that doesn't treat him as a nervous system to be farmed.
That's the whole thing. That's the episode.
Say that again for the people in the back.
The feed is not designed to inform you. It's designed to keep you. Those are different objectives and they produce different products.
The moment you understand that, the question stops being "which app" and becomes "what do I actually want my information environment to do to me."
Herman, before we pull this together, there's a small detail I want to go back to. You said the front page editor had judgment and the algorithm doesn't. But you said it like you'd seen it.
I've read enough about newsroom operations to be confident in the shape of it.
The news hole was finite.
There was a fixed number of column inches and a fixed number of minutes, and somebody had to decide what filled them. That's the part that's gone. The hole is now infinite and the deciding is done by a machine that has no stake in what you know.
Which is why the wire editor is the right comparison. Somebody sat in a room at two in the morning with a stack of wire copy and decided what the region would wake up to. That job existed. It was a real job. And it's the job the algorithm now does, without the person and without the responsibility.
Two in the morning. That's a specific image.
Hilbert: It was usually closer to four.
Sorry?
Hilbert: The wire. You said two in the morning. The last edition of the night was usually locked around four. The wire editor came in at eleven and the paper went to bed at four. Two in the morning was the middle of it, not the end.
You worked the wire.
Hilbert: Regional daily, circulation about forty thousand. I did the overnight wire for two years. You had a stack of copy from the agencies and a fixed number of column inches, and you decided what the region read with its breakfast. Most of it you spiked. That was the job. Spiking was the job.
How much did you spike?
Hilbert: Most of it. You'd get maybe two hundred items on the wire and you'd run thirty. The rest went in the bin, and the bin was the point. Anybody can run everything. The value was in what you threw away.
That's the finite news hole. That's exactly what we were describing.
Hilbert: The hole was the product. That's what the algorithm doesn't have. There's no hole now. There's no reason to throw anything away. And once you don't have to throw anything away, you don't have to decide, and once you don't have to decide, you don't have to be responsible for what you chose.
The editor wasn't a brake. The editor was the whole thing. The editor was the news.
Hilbert: The reporting was the easy part. Anybody can report. The hard part was the four in the morning decision about what forty thousand people needed to know, and being wrong about it in public.
The algorithm makes that decision a billion times a day and is never wrong, because it was never asked to be right. It was asked to be engaging.
Hilbert: I spiked a story once about a researcher at the local university. Small item, agency copy, maybe four inches. She'd done something with a protein, I don't remember the details. It wasn't newsy. No conflict, no local angle, no picture. I ran a zoning dispute instead. She won something big a few years later. I don't remember what. I remember the four inches.
You still think about it.
Hilbert: I think about the zoning dispute more, honestly. Nobody ever complained about that one.
That's the human cost of the filter. The ones that didn't.
Hilbert: The problem isn't that the news is bad. The problem is that the news was never designed to make you feel good. It was designed to make you pay attention. Those are different things. They were different things when I was doing it and they're different things now. The only difference is I had to look at the stack. The machine doesn't.
Then you went home at four in the morning.
Hilbert: Then I went home at four in the morning. I've got to move a chest freezer before it gets dark, so.
Right.
Hilbert: The chest freezer's been in the hallway for a week. It's not getting any lighter.
It never does.
Hilbert: No.
Daniel asked a simple question. Are there better tools. And the answer turned out to be more complicated than a list of apps.
The supply exists. Kagi for the tunable mix, Ground News for the spread, Positive News and Reasons to be Cheerful for the constructive stuff, Quanta and Ars for the science, Aeon and Noema and the Substack long tail for the deep thought. RSS to hold it together without an algorithm in the middle. That's a real stack and it would work.
But the thing he's actually asking for, the thing that learns his preferences, is the thing I'd be most careful about. Because that's the mechanism that produced the problem. The feed already learned his preferences. It learned them perfectly and it made him miserable.
The open question I'd leave people with is this. If the news was never designed to make us feel good, what would a news diet designed for flourishing actually look like? And is that even possible at scale, or is it inherently a personal curation project that each of us has to do by hand?
Because the tools are getting better. AI summarization and personalization are going to make this sharper over the next few years. And the gap between what happened and what matters to you is going to widen. That could go either way. It could be the thing that finally gives Daniel his fire hose of weird, or it could be the thing that seals him in a bubble so tight he never sees anything he didn't ask for.
Both are available. That's the honest answer.
Before we go. If you've found a tool or a source that actually makes your news consumption feel sustainable, tell us. We'll compile the best ones and put them somewhere useful. We're at my weird prompts dot com, or email us at show at my weird prompts dot com.
Thanks to Hilbert Flumingtop, who produces this show and apparently spent two years deciding what forty thousand people read with their breakfast.
This has been My Weird Prompts.
The human-AI collaboration podcast. We'll be back soon.
