An independent magazine founded on a single conviction: the most important ideas deserve more than 800 words. We publish long-form reporting for curious minds who want to actually understand something.
Undercurrent was founded in 2021 by a small group of science journalists and editors who were frustrated by the state of science communication. Important findings were being summarized into misleading headlines. Complex ideas were being compressed until they became caricatures of themselves. Nuance was treated as an obstacle.
The alternative we built: articles long enough to actually explain things, written by specialists who know the material, and edited to be readable by anyone intelligent enough to be curious. We don't talk down to readers, and we don't oversimplify the science to avoid losing them.
We're independent โ no corporate parent, no venture capital, no advertising that shapes coverage. We're funded by subscriptions and the occasional institutional grant. This makes us small. It also means nobody tells us what to write.

Every Undercurrent piece is written by a specialist and edited by a journalist. The combination matters โ accuracy without clarity is useless; clarity without accuracy is worse.
Former science editor at The Atlantic. PhD physics, MIT. Covers particle physics and cosmology.
Cultural critic and essayist. Former staff writer at The Guardian. Covers technology's social effects.
Environmental journalist based in Oslo. Covers Arctic science, climate policy, and earth systems.
AI researcher turned journalist. Former Google Brain. Covers machine learning and its implications.
Articles are as long as they need to be to properly explain what they're explaining. Not a word more, but never artificially compressed to fit a format.
We don't have generalist journalists cover specialized topics. Science pieces are written by scientists. AI pieces are written by people who understand AI.
We don't run advertising. Our revenue comes from subscriptions. Nobody pays to influence our coverage, because nobody advertises with us to threaten.
Science is full of uncertainty, disagreement, and provisional conclusions. We represent that honestly rather than forcing false certainty to make a cleaner story.
Science and culture are not American phenomena. Our contributor network spans six continents, and our coverage reflects that.
We read every letter. We correct errors prominently when we make them. We engage with reader expertise. Our readers often know things we don't.
The categories are distinct, but the underlying question connecting them is the same: what is actually happening, and what does it mean?
All articles are free to read. Our newsletter โ a curated Sunday digest of the week's best articles โ is free to subscribe to. We don't have a paywall, and we don't intend to add one. We're funded by voluntary subscriptions (our readers can become supporting members) and occasional grants from journalism foundations.
Editorial decisions are made by the editorial team based on significance, interest, and our confidence that we can cover a topic accurately and interestingly. We don't follow the news cycle. We don't write about something because it's trending. We write about things when we have something genuinely worth saying about them โ even if the most important angle on a topic is different from what's currently being discussed.
Yes. We welcome pitches from specialists in their field โ researchers, practitioners, and subject-matter experts who have something specific and well-informed to say about a topic within our coverage areas. We do not accept pitches from generalist freelancers without domain expertise for technical subjects. Send pitches to hello@undercurrent.press with a brief description of the argument and why you're positioned to make it.
Yes, and we take them seriously. If you identify a factual error in an article, email corrections@undercurrent.press with the specific claim and the evidence. Confirmed errors are corrected with a visible correction notice appended to the article. We do not memory-hole mistakes.