Color any map so no two bordering countries share a color, and four colors are always enough. Nobody has a short explanation for why — but there's now a much faster way to check it.
Take any flat map — countries, counties, a stained-glass window — and you can always color every region with just four colors so that neighboring regions never match. Mathematicians have known this is true since 1976, when two researchers proved it by having a computer grind through 1,482 different map shapes one by one, a proof so long that no human has ever fully checked it by hand, which has bothered mathematicians for fifty years since. A team including Mikkel Thorup at the University of Copenhagen and Carsten Thomassen at the Technical University of Denmark, with Ken-ichi Kawarabayashi and Bojan Mohar, found a new proof that checks even more shapes — 8,202 of them — but checks huge batches of them at once instead of one after another, cutting the computer work needed from roughly n² steps down to about n log n for a map with n regions. It's still a computer-checked proof, not the fully by-hand version Thomassen has spent decades hoping for. "Here we have a problem that even a child can understand," Thomassen said. "I think that's the reason why it has been such a big challenge."
Quanta Magazine, reporting work by Mikkel Thorup, Carsten Thomassen, Ken-ichi Kawarabayashi, and Bojan Mohar · Sept 10, 2026 · quantamagazine.org
§
Why it's hereIt's one of the only famous math results you can test yourself with a map and four crayons — and this is a genuinely faster, checkable improvement on the 1976 proof, not just a retelling of it.
For hundreds of millions of years after the Big Bang, the universe was completely dark. One faint radio whisper survives from that time — and a suitcase-sized spacecraft is being designed to finally hear it.
For roughly the universe's first 400,000 years, there was no light at all — just a hot soup of hydrogen gas, in a stretch astronomers call the "cosmic dark ages," which lasted until the first stars switched on and lit everything up. That early hydrogen still gives off one extremely faint radio signal, at a wavelength of 21 centimeters, but it's nearly impossible to pick up from Earth because our own radio noise — TV, phones, satellites — drowns it out completely. A team led by Eloy de Lera Acedo at the University of Cambridge, working with Portsmouth University, STFC RAL Space, and the UK's Surrey Satellite Technology, is designing a small satellite called CosmoCube that would park itself behind the Moon, use the Moon's entire body as a radio shield, and listen in the 10–50 megahertz range for that leftover signal, giving scientists their first real look at how gravity and dark matter pulled the universe's first structures together. The mission isn't funded for launch yet — it's currently proposed to the European Space Agency, with a target cost under €50 million, and researchers think it could fly within five years if approved. "The far side of the Moon is really the only option," de Lera Acedo said. "It solves multiple problems at once, opening a clear window to the very early universe."
ScienceDaily, reporting University of Cambridge / University of Portsmouth · Sept 28, 2026 · sciencedaily.com
§
Why it's here"Listening for the time before the first stars" sounds like it should be impossible, and technically nobody's done it yet — this is a real, currently-proposed mission sitting right at the edge of getting funded, not a result that's already in.
Splitting water into hydrogen fuel usually needs an expensive metal catalyst. A new material skips the metal entirely — it just tears its own sulfur bonds apart in sunlight.
Hydrogen is a genuinely clean fuel — burn it and you get water back out — but making it today usually means either splitting water using pricey metals like platinum, or making it from natural gas in a process that releases carbon dioxide, which defeats the point. Chemists at Oregon State University, led by Kyriakos Stylianou, built a new framework material called BVR-19 that sidesteps the metal problem entirely: it contains an unusual sulfur-to-sulfur bond that breaks apart when hit with light, creating highly reactive sulfur fragments that pull electrons out of water and release hydrogen gas. The material forms on its own at room temperature while just sitting in water, so it skips the high heat and rare metals that make existing "green" hydrogen methods expensive — green hydrogen currently runs about $5 per kilogram, versus roughly $1.50 for hydrogen made from natural gas. BVR-19 itself isn't about to replace either method; it's proof that a cheap, metal-free material can do this job at all. "Our work provides a blueprint for designing better materials that can bring down the cost of green hydrogen," Stylianou said.
ScienceDaily, reporting Oregon State University / Journal of the American Chemical Society · Oct 4, 2026 · sciencedaily.com
§
Why it's hereClean hydrogen only matters if it's actually cheap to make, and this goes after the single biggest reason it isn't — the metal catalyst — rather than offering another small efficiency tweak.
One had been hiding in plain sight for decades, mistaken for a more common look-alike. Its eggs come with a built-in lid that's never been seen in any stick insect, ever.
Stick insects are already masters of disguise, built to look exactly like the twigs they sit on, and Australia's rainforests hide some genuinely giant versions of them. A team led by Braxton Jones at the University of Sydney used DNA testing, museum specimens, citizen-science photos, and captive breeding to sort out two species that had been tangled up with look-alike relatives for decades: Anchiale robusta, a large species that had gone unrecognized as its own species, and the brand-new Anchiale mabiensis, found only in a small, critically endangered patch of Queensland rainforest called the Mabi rainforest. Stretched out, Anchiale mabiensis measures 16.3 centimeters — about the length of a dollar bill — and lays eggs with what the researchers describe as "a structural lid unlike anything previously recorded in stick insects or their closest relatives." Sorting out exactly which species is which isn't just trivia: some of these insects occasionally swarm eucalyptus forests in huge "plague" numbers, and foresters need to know which species they're actually dealing with to manage it. "These findings demonstrate the importance of protecting threatened ecosystems," Jones said, "and documenting Australia's extraordinary biodiversity before we lose species without us having even discovered them."
University of Sydney news, reporting research published in Austral Entomology · Sept 22, 2026 · sydney.edu.au
§
Why it's hereA giant insect that's gone unrecognized for decades, right in people's backyards, with an egg feature nobody's ever documented before — and a real practical reason (swarm management) for sorting the species out, not just taxonomy for its own sake.
A gene called BC200 normally works quietly inside human neurons. Twice in the last 100,000 years, it also slipped itself into the DNA of a virus that still infects people's skin today.
Deep in human brain cells sits a small, mobile stretch of DNA called BC200, which seems to help neurons control which proteins get built — though scientists still don't fully understand its job. BC200 descends from an ancient "jumping gene" that first copied itself into primate DNA about 40 million years ago, and new research shows it never really stopped moving: at least twice in the past 100,000 years, it jumped out of the human genome and into molluscum contagiosum virus, a common poxvirus that still causes small, harmless skin bumps in people today. Cedric Feschotte's lab at Cornell, with first author Pu Gao and collaborators in Beijing and Texas, traced the jumps to skin cells — the only place that virus infects — and showed the gene still behaves like a jumping gene once it's inside the virus's own DNA. It's a strange discovery because BC200 appears to be doing two jobs at once: a real, useful task in brain cells, and an entirely separate, parasitic existence riding along inside a virus. As Feschotte put it, "BC200 was itself created from a mobile element but has retained its mobility and yet it is also clearly serving a cellular function."
Cornell University, reporting research published in Science · Sept 24, 2026 · news.cornell.edu
§
Why it's hereJumping genes are old news, but one jumping out of a working human brain gene into a virus that still infects people today — twice, independently — is a genuinely new and slightly unsettling twist on the idea.
This fall, an AI system beat human mathematicians to part of one of math's seven $1-million problems. Now mathematicians are arguing, in public, over whether that's the start of a golden age or the end of their field.
In September 2026, OpenAI announced that one of its AI systems had produced a result related to the Navier–Stokes equations, one of math's seven Millennium Prize problems — a claim that's still contested, but one that landed in a field already anxious about what AI proofs mean for mathematicians' jobs and sense of purpose. The worry isn't just "will AI take mathematicians' jobs" — it's something stranger: AI systems can now generate proofs that are correct but effectively unreadable, which breaks a lot of what mathematicians actually value, since the point of a proof is usually the human understanding it hands you, not just a confirmed "yes, it's true." Some mathematicians are responding by avoiding the problem entirely — posting fewer open conjectures online so AI systems can't scrape and solve them — while arXiv has started limiting submissions to deal with a flood of AI-generated papers. The views genuinely split: Scott Aaronson at UT Austin says mathematicians have been "dethroned as theorem-proving entities," and Ken Ono warned a room of Berkeley students they might be graduating into "a nonexistent profession," while others, like Daniel Litt at the University of Toronto, argue the field just needs new incentives built around communication and big-picture thinking instead of raw problem-solving speed. A new group, the Association for Human Mathematics, has formed specifically to work out what mathematics should reward now that speed is no longer a uniquely human advantage.
Why it's hereNot abstract hand-wringing: named mathematicians at real universities are genuinely divided over whether their field survives this in anything like its current form, which is a sharper and higher-stakes debate than most "AI changes everything" pieces manage. Worth reading skeptically either way — some of the loudest voices here have an obvious stake in the outcome.
Sources this issue
Quanta Magazine, reporting Mikkel Thorup, Carsten Thomassen, Ken-ichi Kawarabayashi & Bojan Mohar — math
ScienceDaily / University of Cambridge / University of Portsmouth — physics
ScienceDaily / Oregon State University / Journal of the American Chemical Society — chemistry
University of Sydney / Austral Entomology — biology