Ships cross the Panama Canal by climbing a staircase made of water.
The canal cuts through Panama, but the land in the middle is higher than the sea. So instead of digging down to sea level, engineers built the route as a bridge over the hills.
How it works:
1. A ship enters the first lock chamber and the gates close behind it.
2. Water flows in from the next chamber through huge culverts under the floor. No pumps, just gravity.
3. The ship rises with the water level, then moves to the next chamber. Three steps lift it to Gatun Lake, about 26 m (85 ft) above the sea.
4. On the other side, the same trick runs in reverse and the ship is lowered to the ocean.
Every ship that goes through uses a huge amount of fresh water from the lake, which is why the canal depends on rainfall.
The timelapse and animated map show the whole climb from above.
馃帴 @wonderofscience
The blue morpho butterfly has no blue pigment in its wings.
What you see in the video is structural color. The wing scales are covered in microscopic ridges shaped like tiny Christmas trees, and the gaps between their layers are about the size of the wavelength of blue light.
When light hits them, the waves bouncing off different layers interfere with each other. Blue waves reinforce one another and shoot back to your eye. Most other colors cancel out or get absorbed by the dark pigment underneath.
That's why the color shifts as the wing turns. Change the angle, and you change the distance light travels inside the structure, so the shimmer flickers between vivid blue and dull brown.
The underside of the wings is brown with eyespots. Flap, and the butterfly flashes bright blue. Fold, and it vanishes.
Engineers have copied this design for anti-counterfeit marks on banknotes and for color-changing materials that need no dye.
A hockey-puck capsule just flew US Air Force cargo to orbit and dropped it into the Pacific.
SpaceX's Starfall launched on a Falcon 9 on Oct. 20 as part of the Bandwagon-5 mission. Inside were cargo pods built by the Air Force Research Laboratory, packed with medical equipment, food ration packs, a SATCOM tracker, batteries and a reverse osmosis desalinator. The kind of kit you'd want if you were stuck somewhere nobody can reach by road, sea or plane.
AFRL says the pods were recovered after splashdown. Each carried sensors logging temperature through launch, reentry and landing, plus the shock and vibration hitting the cargo.
The idea is point-to-point rocket cargo: launch from one place, land almost anywhere on Earth within about an hour, instead of flying for a day.
What's public about Starfall comes from an FAA environmental assessment - about 10.2 feet (3.1 m) across, up to 2,200 pounds (1,000 kg) of return cargo, a carbon fiber heat shield, gas thrusters for attitude control and parachutes for precision landings.
The Air Force wants to get to tens of tons per launch, far beyond Starfall. Its larger candidate, Starship, lost its Johnston Atoll landing pad plan in July 2025 over environmental concerns.
Source: Space
Switch off one protein called "Mitch" and human cells in a dish start burning more fat and stop turning into new fat cells.
Weizmann Institute researchers deleted the gene for MTCH2 in lab-grown human cells. The cells consumed more fuel, fats and carbohydrates alike, and immature cells had trouble developing into fat cells. The work was published in The EMBO Journal.
It started with mice. When the team switched Mitch off in muscle, the animals burned more energy, resisted obesity on a high-fat diet, had better endurance and improved heart function.
How it works: mitochondria, the cell's power plants, constantly merge into networks and split apart. Mitch helps them fuse. Without it the network fragments, and cells end up in a more energy-hungry state, so they burn more fuel.
Why anyone cares: drugs like semaglutide and tirzepatide work mostly on appetite, and a 2026 analysis of 20 trials with 15,782 people found lean mass made up roughly 25% to 39% of the weight lost on several of them. A way to burn fat while sparing muscle is the thing researchers want.
This is early lab and mouse work. No one has tested a Mitch blocker in people.
Source: Weizmann Institute of Science
Scientists drilled the same Tibetan ice cap 23 years apart and got nearly identical climate records. The ice may be more than 128,000 years old.
The cores come from the Guliya Plateau in northwestern Tibet, drilled in 1992 and 2015. Earlier work suggested the record only reached back to the middle of the Holocene, which began nearly 12,000 years ago. The new analysis pushes it back by more than 100,000 years, making these the oldest ice cores ever recovered from this part of Tibet.
How do you date ice that old? The team used radioactive beryllium and chlorine, which build up and decay at known rates. They picked out the signature of the Laschamp Geomagnetic Excursion, a disturbance of Earth's magnetic field more than 41,000 years ago, which left chemical traces in the ice. Then they matched the oxygen isotope pattern against nearby Tibetan cave deposits. Both lines of evidence point to ice well beyond 128,000 years.
Outside the polar regions, it's the only mountaintop ice core going back that far. Records from Huascar谩n in Peru only reach about 30,000 years.
It also settles a question: big ice masses in Tibet did survive the transition out of the last ice age. As lead author Lonnie Thompson of Ohio State put it, "If we only had Holocene-era ice in Tibet, that would mean that the big ice sheets from the last ice age didn't persist through the early Holocene. But evidence shows that it did."
The study is in Science Advances. The team plans to analyze more samples from the cores against this new timeline.
Source: Ohio State University
Space has five parking spots where gravity from two giant bodies cancels out, and a spacecraft can hang there almost for free.
They're called Lagrange points. The James Webb telescope sits at one of them, about 1.5 million km from Earth. So does NASA's Nancy Grace Roman Space Telescope, which launched on 30 August 2026 and headed to the same region.
How it works:
1. Closer to the Sun than Earth, an object normally orbits faster than we do. Between us and the Sun, Earth's gravity pulls back on it and slows it down.
2. At one exact distance, called L1, the object's year becomes exactly as long as Earth's. It stays lined up with us instead of drifting away.
3. Behind Earth, the logic flips. An object there would normally lag behind, but Earth's extra pull speeds it up. At L2, the year again matches ours.
4. L3 sits on the far side of the Sun. L4 and L5 are the corners of two equilateral triangles built on the Sun-Earth line, and they're the stable ones: objects get pulled in. Jupiter has more than a million trojan asteroids around its own L4 and L5.
Why L2 is perfect for Webb: Sun, Earth and Moon all sit in the same direction, so one sunshield blocks their heat and light together, with no need to rotate it. Solar panels never fall into Earth's shadow, little fuel is needed to hold position, and Earth stays in direct line of sight for data.
The first three points were found by Leonhard Euler around 1750. Joseph-Louis Lagrange found the other two about a decade later, and published his essay on the three-body problem in 1772.
Which would you park something at: the Sun-facing L1 or the shadowed L2?
Source: Wikipedia
Finland has opened a criminal investigation into Google over a data centre site in Muhos, and the suspected crime is environmental damage.
The project is part of Google's 13 billion euro investment in Finland. The Oulu Police Department says that, based on current information, there were irregularities and shortcomings in drainage and water management at the worksite, which is run by Google's local subsidiary Tuike Finland. Five separate requests for an investigation are being handled in one probe, and other charges may be added as it goes on.
It began with the Finnish Supervisory Agency. It ordered Tuike Finland to halt work on two data centre projects, Muhos and Kajaani-Otanm盲ki, because the required environmental impact assessments were still incomplete. The company confirmed on 14 October that preparatory work had stopped at both sites.
The water plans were the sticking point. The agency found them insufficient to prove that water impacts can be controlled, especially with heavy rainfall, snowmelt and the acidic black shale soils in the area. Disturbed black shale can release acid and metals into runoff, which is why drainage at a construction site like this matters so much.
By 5 November the company has to submit a detailed plan with structures, timelines and independent water monitoring. If that falls short, the agency can move to enforcement, including a conditional fine.
Police expect the probe to take a good deal of time and aim to finish it during 2027.
Source: Yle
Mistral just released the weights of a 1-trillion-parameter model. It's called Mistral Large 4, nicknamed Le Chonk, and anyone can download it.
The numbers are the interesting part. Three weeks of extra reinforcement learning after the public preview lifted its Terminal-Bench 4.0 score by 13 points, to 41%. That's the top result among open-weight models, with 30% fewer output tokens.
On the Artificial Analysis Coding Agent Index it scores 48%, effectively tied with the leading open model GLM-5.3, but it uses 47% fewer tokens per task on average. Fewer tokens means a cheaper bill every time an agent runs. On Cognition's private FrontierCode 2.0, which asks whether a repository maintainer would actually merge the agent's change, it leads with 19.7%.
In Surge Code's blind pairwise tests, people preferred it over GLM-5.3, Kimi K3 and DeepSeek V4.1 Flash. Only Claude Opus 5.5 ranked above it.
Security is the other half. It scores 51% on the Artificial Analysis Cyber Index (find a vulnerability, reproduce it, patch it without breaking anything). Mistral says it refuses 96.6% of malicious cyber requests, the highest it measures among open models.
Mistral also says it saw potential for real attack uplift in live traffic and is tightening the guardrails. That's the catch with open weights: once the file is out, anyone can run it on their own hardware.
There are three reasoning modes (low, medium, high) so you can trade accuracy for cost. The model was trained from scratch on European infrastructure.
Source: Mistral AI
The same hit, 10 feet closer, and the hard hat still won.
In this Black & Veatch demo, the first hit lands on a bare target and does serious damage. Then they put on a hard hat, move 10 feet closer, and run it again. The impact is so loud the crew says it sounds like a gunshot.
What the shell does in that split second:
1. The hard outer shell spreads the force of a small, hard impact over a much larger area.
2. A suspension harness inside leaves a gap between the shell and the skull, so the shell can flex without touching the head.
3. The harness absorbs part of the energy before it reaches the head.
The shell is built to take one big hit. After a serious impact it should be replaced, even if it looks fine.
馃帴 Black & Veatch
One mountain in Switzerland, two completely different skies: thick cloud on one side, clear blue on the other.
The video was shot on the Alpstein massif in the Swiss Alps and shows the orographic effect in action.
How it works:
1. Wind carries moist air toward the mountain.
2. The slope forces that air to climb. As it rises it cools, because pressure drops with height.
3. Cool air holds less water vapor, so the moisture condenses into cloud and often rain or snow on the windward side.
4. Once over the ridge, the air has lost much of its water. It sinks, warms up and dries out, so the cloud melts away on the leeward side.
That dry zone behind a mountain range is called a rain shadow. It's the reason the same ridge can hold a wet forest on one side and near-desert on the other, and why a hiker can step over a crest and go from fog to sunshine.
馃帴 Jess Wandering
This is a dead star's explosion, still expanding, filmed across three years by JWST.
The clip stitches together Webb images of the supernova remnant Cassiopeia A (Cas A), taken from November 2022 to October 2025. The shell of gas and dust is growing between frames, and you can see it on a screen.
How can it move that visibly? Cas A is a young remnant, about 11,000 light-years away, and its debris is flying outward at thousands of kilometers per second. At that speed, even across that distance, a few years is enough for the glowing knots and filaments to shift.
The video is the same 2-second sequence looped four times.
Processed by Yuval Harpaz from JWST data.
Volvo held a patent that saves about 15,000 lives a year in the US alone. In 1968 it let every competitor use it for free.
The man behind it was Nils Bohlin, a Swedish engineer who spent 16 years at Saab designing aircraft and helping develop ejection seats. In 1958 Volvo hired him as a safety engineer, and he spent about a year on one problem: keeping a person inside the car during a crash.
How the three-point belt works:
1. One strap crosses the chest and shoulder, another crosses the hips, and both lock into a single buckle at the side.
2. In a crash the body wants to keep moving forward at the car's speed. The belt catches it at the strongest parts of the skeleton: the pelvis and the rib cage.
3. The load is spread across a wide area instead of one point, so the force isn't concentrated on the soft belly or the neck.
4. The upper body is held back too, so the head and chest don't hit the wheel or dashboard.
Bohlin got his first patent (US 3,043,625) in 1962, after filing in August 1959. Then he did the part that made it matter. He studied 28,000 accidents in Sweden and presented the results at the 11th Stapp Car Crash Conference in 1967. Unbelted occupants were killed across the whole speed range. No belted occupant died in a crash below 60 mph. And no belted occupant died if the passenger compartment stayed intact.
The study led the U.S. Department of Transportation to require three-point belts in American cars. In 1968 the design was opened to the public. Today most industrial countries require occupants to wear seat belts by law.
The US National Highway Traffic Safety Administration says belts save about 15,000 lives a year in the US.
Do you buckle up on a 5-minute drive to the store?
Source: Wikipedia
Orion flew around the Moon with more than 10,000 tiny crystals inside two human-shaped mannequins. The radiation they recorded was higher than in Earth orbit, but not by as much as scientists might expect.
The detectors were part of the MARE experiment on the uncrewed Artemis I mission (late Nov to early Dec 2022). Among them were lithium fluoride crystals made at the Institute of Nuclear Physics of the Polish Academy of Sciences in Cracow.
How they work: a charged particle passes through the crystal and breaks its atomic structure, leaving defects called color centers. Shine blue light on the crystal and those spots glow red. Under a microscope, you can see the track of each single particle.
The hard part is reading them. In space, radiation comes from every direction, with many kinds of particles at wildly different energies, so tracks point every which way and blur together. The team wrote correction software to strip out the background glow. Their results landed within 15% of conventional track detectors, and they also caught weak particle tracks that older detectors miss.
The work was published in Radiation Measurements. These are preliminary measurements, but they show how much radiation astronauts can expect on the way to the Moon.
Source: Institute of Nuclear Physics
The limits are real: a small sample, only young men who already trained, and just eight weeks. The authors say longer trials with more diverse groups are needed. For people short on time, it's an early sign that the same work in three small pieces can count just as much.
Source: Journal of Exercise Science





