Edition of November 4, 2026

Amateurs still grind telescope mirrors by hand. They rub two glass disks together with grit in between. The strokes are sloppy and random. Yet both surfaces end up as matching spheres, accurate to tens of nanometres.

A physicist at SLAC and Stanford explains why, starting from a 1927 law of wear. Glass wears faster where it is pressed harder and rubbed faster.

The strokes point every which way. Only planes and spheres survive that symmetry. Small bumps fade within about an hour. Large ones spread out and even out, about 30 times slower.

Gravity rules out the plane: the top disk hollows, the bottom one bulges. The same maths explains tumbled pebbles. They turn shiny within weeks, but would need years to become spheres.

Source: From Rock Tumbling to Telescope Mirror Grinding: How Optical Perfection Emerges from Random Motion, https://arxiv.org/abs/2611.01673

The Choranche cave, in France's Vercors massif, holds thousands of "soda straws". These hollow stalactites reach 3 metres long, yet only 5 mm wide. Some may be 33,000 years old.

To rebuild the path to the cave, rock had to be blasted nearby. A Grenoble team first measured how the straws vibrate.

Each straw rings at its own pitch. A 15 cm straw rings near 175 hertz, right where the blasts peaked. Resonance can multiply the load fivefold. So the short straws were most at risk.

The team set a limit on how fast the cave wall may move when shaken: 2.4 mm per second. The works respected it, and no straw was harmed. An earthquake, which shakes more slowly, would threaten the long ones instead.

Source: Seismic study of soda straws exposed to nearby blasting vibrations, https://arxiv.org/abs/2611.02873

A dying massive star can leave behind a black hole or a neutron star. Two such remnants usually merge much later, at least about 100,000 years after the collapse.

On 10 September 2019, LIGO and Virgo flagged two faint signals, 7.1 hours apart. Their patches of sky overlap, at compatible distances. Unrelated alerts rarely line up this well: about 1 chance in 13,000.

Inside that overlap sits a supernova, SN 2019rsq. It peaked 24 days after the first signal. For its type, that is a normal delay after a collapse.

An international team suggests one star whose collapse left a disk that broke into pieces. One piece fell into the black hole. Two others merged hours later. Both signals remain marginal: chance is still possible.

Source: A temporally clustered gravitational wave pair and a Type IIb supernova, https://arxiv.org/abs/2611.00440

In Austria, physicists hung a 1-milligram mirror bar from a glass fibre 1 micrometre thick. In an ultra-high vacuum, it turns almost freely. Friction barely touches it: its damping rate, the pace at which its motion would fade, is a mere 7 nanohertz.

Nothing traps it. A laser tracks its angle without pause. Within 26 milliseconds, that tracking pins down its state. Measurement replaces the trap.

How well do they know its motion? Expressed as a temperature, 140 billionths of a degree above absolute zero. That is a hundred times colder than earlier tabletop experiments of this kind.

Such a quiet free mass can test proposed changes to quantum mechanics. Its stillness already rules out some new versions of them. It is also a step toward checking whether gravity can entangle objects.

Source: Taming a macroscopic free mass through continuous quantum measurement, https://arxiv.org/abs/2611.03159

Life's first chemistry may have started from hydrogen cyanide, HCN. Lab recipes need it concentrated, about 10 millimoles per litre. Comets carry plenty. Did they bring it to the young Earth?

Cambridge and Edinburgh researchers simulated 100,000 histories of comet impacts. Small comets burst in the air. The others hit so fast that the heat destroys most of their HCN.

In more than half the simulations, not one comet delivered enough. The rare successes came about 4.41 billion years ago. That is over 100 million years before the first hints of life.

Mars does far better: 744 successful impacts against 87 for Earth, before counting dry land. Its thin air and slower comets help. For Earth, the comet idea takes a heavy blow.

Source: Cometary impacts an unlikely pathway towards the origins of life on Earth, https://arxiv.org/abs/2611.02924

Above Antarctica, the stratosphere changes over decades. Yet the heart of the continent has very few instruments to watch it.

IceCube, a neutrino detector buried in the South Pole ice, also counts muons. These particles come from cosmic rays, over 2,000 every second. When the stratosphere warms, more muons reach the detector.

Researchers at Harvard and Valencia simulated a decades-long record. The weak point is not chance, it is the detector. If its count drifts by 1% per decade, that fakes 2.8 degrees per decade. The rate must hold steady to 0.3% per decade.

Even snow piling up on top fakes some cooling. Still, with a stable detector, the cooling seen since 2011 is in reach. About 22 years of data would confirm it.

Source: Sensitivity of the IceCube atmospheric muon rate to decadal trends in the effective atmospheric temperature, https://arxiv.org/abs/2611.00918

Kreutz comets graze the Sun. C/2026 A1 (MAPS), found in January 2026, was due on 4 April. Its path would have taken it within 0.006 astronomical units of the Sun: under 1% of Earth's distance.

Over its last three days, a 30-metre radio telescope tracked it. The team was Spanish, French and American. They detected hydrogen cyanide. It is a first for this family of comets.

On 3 April, the comet's cyanide output jumped by over 75% in 17 minutes. Within roughly 90 minutes, that outburst stripped away 0.4 to 1.2% of its mass.

Less than a day later, two hours before its closest pass, the comet fell apart. The outburst may have helped break it. Its molecules may have already survived two earlier passes.

Source: HCN detection in the Kreutz sungrazing comet C/2026 A1 (MAPS) shortly before its disruption, https://arxiv.org/abs/2611.03417

In October 2019, Steam started asking players to review the games they had just played. Across the store, daily reviews jumped from about 17,000 to 70,000.

The question pops up after about three hours of play. Researchers studied 21.6 million reviews of 315 games. Just past three hours, reviews became 4.25 times as numerous as just below, and more positive.

Scores still rank games much as critics do. But they react less when a game takes a bad turn. One fixed rating tier caught 12% of downturns, down from 42%.

The information has not vanished. The extra reviews make up in number what each one lacks. Failures that developers admit still pull the score down. Platforms, the authors say, should recalibrate their labels.

Source: Ask More, Hear Less: How a Review Prompt Changes What Online Ratings Show, https://arxiv.org/abs/2611.02623

Colour every whole number red or blue. Call a number "same-colour" if it is the product of two reds or two blues.

Odd numbers red, even ones blue: three numbers in four qualify. Only 2, 6, 10 and their kind miss out. Erdős and Sárközy asked a sharper question. Can some colouring push that share as low as you like?

A mathematician at Dalhousie University, in Canada, says yes. The trick: a number and its "mirror" always get opposite colours. The work grew out of long exchanges with an AI model, GPT-6 Astra Pro.

The price is absurd. To get under 10%, the first 10^(10^41,823) numbers must all be red. The proof has not yet been checked by other mathematicians.

Source: A solution to the Erdős–Sárközy product-density problem, https://arxiv.org/abs/2611.01480

In the lungs, each hemoglobin molecule picks up four oxygen molecules. Textbooks say binding one makes the next easier. That is "cooperativity".

That picture imagines hemoglobin bathing in a sea of free oxygen. Physicists in Armenia point out a problem. A red blood cell holds about 30 times more hemoglobin than free oxygen. Each molecule mostly sees its own four.

They redid the maths for one hemoglobin with a few oxygen molecules. Cooperativity comes out much weaker. In Pauling's classic model, its key factor drops from 6.32 to 1.88.

Larger values turn unstable. Once one oxygen leaves, the other three tumble out together: a risk of oxygen poisoning. Oxygen might even travel by hopping from one hemoglobin to the next.

Source: Canonic hemoglobin-oxygen equilibrium: Reassessing the role of cooperativity, https://arxiv.org/abs/2611.00824

Are these two shapes the same? For shapes in four dimensions, no algorithm can always decide. That has been known since 1958.

Mathematicians pinned it down on one shape: k copies of S² × S², glued together. S² is the surface of an ordinary ball. The record shrank from 14 copies to 12, then 9.

Cameron Gordon, at the University of Texas at Austin, now gets down to 7 copies for shapes built from flat pieces, matching the record already reached without that condition. In the same shape, no algorithm can tell if a sphere is knotted.

The ultimate prize is zero copies: the four-dimensional sphere itself. Can it be recognised by an algorithm? Can a knotted sphere inside it be spotted? Both questions remain open.

Source: The recognition problem for 4-manifolds and the unknotting problem for 2-spheres in 4-manifolds, https://arxiv.org/abs/2611.00355