Analysis: Science & Technology — 07 October 2026
Francis Halzen wins the 2026 Nobel Prize in Physics for IceCube high-energy neutrinos
The Royal Swedish Academy of Sciences awarded the 2026 Nobel Prize in Physics to Francis Halzen of the University of Wisconsin–Madison for the IceCube Neutrino Observatory and the discovery of high-energy neutrinos of astrophysical origin. The prize is 12 million Swedish kronor. IceCube, finished in 2011, instruments a cubic kilometre of South Pole ice and has reported a neutrino flux that must come from outside the Solar System.
Neutrinos rarely interact and carry no charge, so magnetic fields do not bend their paths and they arrive with direction and energy largely intact. Halzen proposed in 1988 that a neutrino striking a nucleus in clear Antarctic ice would produce a flash that buried sensors could track. Cosmic accelerators reach energies up to about a million times those of laboratory machines. Only neutrinos preserve a clean record of those sites.
Source identification is still incomplete. IceCube has reported evidence from the blazar TXS 0506+056, the galaxy NGC 1068, and the Milky Way, but most of the high-energy flux is not tied to individual objects. A solo prize also sits beside a collaboration of about 450 scientists at 58 institutions. Halzen called the award a surprise and said he hoped it would help a new proposal; what that instrument would add is not yet public.
Sources: Royal Swedish Academy of Sciences; Phys.org; IceCube.
An internal OpenAI model claims solutions to 377 open mathematics problems
On 6 October, OpenAI said an unnamed internal model had solved 377 open questions across nearly every mathematical subdiscipline, released as 722 papers on GitHub. The company states that the model wrote the papers, with a human edit on only one, on a zero-free region of the Riemann zeta function. About half the results have been checked in Lean. OpenAI reports no model errors in those formal proofs, though a handful exposed mistakes already in the literature. The model failed on most of roughly 4,000 attempts.
The compute profile differs from OpenAI’s September Navier–Stokes run, which used about 10,000 agents and millions of dollars of computing. Here the firm says the average result used the equivalent of roughly three hours of ChatGPT Pro. Two exceptions used more: the zeta-function work and a special case of the Hodge conjecture, both adjacent to Millennium Prize problems. If that accounting holds, the change is from one expensive campaign to many attempts at modest marginal cost.
Volume is not the same as acceptance. Lean covers about half the claims, and the rest are not yet formalized. A correct proof can still be a thin advance if it does not yield a reusable method. Until independent mathematicians check the informal papers, 377 remains a company claim rather than a settled addition to the literature.
Sources: Quanta Magazine.
Oxford study moves a possible origin of animals back by about 200 million years
A University of Oxford-led team reports that animals may have originated between 800 and 700 million years ago, about 200 million years earlier than estimates tied to a younger fossil site. The paper, published 2 October in Science Advances, does not present a new animal body fossil. It challenges the use of China’s roughly 590-million-year-old Weng’an Biota as a maximum-age bound, on the argument that the absence of animals there has been read as proof that animals had not yet evolved.
The comparison is Mongolia’s Kheseen Biota, more than 40 million years younger and similarly well preserved. Animals are already known from other sites of that age, including Namibia and South China, yet none of more than 140 Kheseen samples examined by scanning electron microscopy can be confidently identified as an animal. If exceptional preservation can miss animals that existed, Weng’an’s silence cannot set a hard upper limit. Using older deposits in Norway, Australia, and Arizona, dated about 850 to 730 million years ago, shifts the molecular-clock estimate backward.
That window overlaps the Cryogenian glaciations that began around 720 million years ago, so early animals, if real, may have lived before or during Snowball Earth. Chemical biomarkers already hint at sponges by at least 650 million years ago, but pre-Ediacaran animal body fossils are still missing. First author Orin Lole Durbin states that the analysis does not prove animals existed 800 million years ago. The date stays a calibration problem until body fossils, traces, and biomarkers agree.
Sources: ScienceDaily; Science Advances.