Home Artists Posts Import Register

Content

[This is a transcript with links to references.]

Welcome everyone to this week’s science news. Today we’ll talk about why coin flips are not fifty-fifty, starquakes, the most powerful solar storm ever, space debris, IBM’s new brain-inspired computer chip, an exoplanet collision, a new tectonic plate, projections for AI power consumption, and of course the telephone will ring.

A coin flip is fifty-fifty, right? Not quite. A pretty big group of physicists just put out a pre-print on the arXiv saying that coin tosses aren’t entirely fair.

This is a follow-up study on a paper from 2007. Back then, another team of physicists developed a model for coin flips. They said that the result of a coin flip depends on which side is up when you throw it, and the angle between the direction of throw and the angular momentum. They found that, when averaged over reasonable initial conditions, the probability that the coin comes down with the same side up as if it was thrown isn’t fifty percent but 51.

For the new paper now, a group of 48 scientists came together and, well, flipped a lot of coins. Altogether they made it to more than three hundred and fifty thousand coin tosses before getting tired.

They found that coins are indeed slightly more likely to land on the same side they started on the odds are 50 point eight percent and 49 point two percent, compatible with the prediction from 2007.

Next time someone suggests tossing a coin to you, I think you should explain in great detail why that isn’t exactly fair, it’s going to make you really popular.

Researchers at the University of Tokyo might have figured out the reason for fast radio bursts. Their idea is that they are “starquakes.” 

First, a quick primer on fast radio bursts. These are incredibly intense bursts of energy out there in the cosmos. They last fractions of a second up to a few seconds and have wavelengths in the radio range. Curiously enough, the sources seem to repeat those bursts in irregular intervals.

Fast radio bursts are a fairly new discovery. They were seen for the first time in 2007 and their exact origin has remained a mystery. But you know how physicists are. They’ve put forward lots of idea for what causes fast radio bursts. On the conservative end we have collisions of neutron stars or black holes. Others have suggested some sort of supernovae, pulsars collapsing due to dark matter, alien signals, or white holes.

According to this new paper, it’s none of those but instead it’s tectonic activity on neutron stars. Yes, neutron stars are believed to have a solid crust. This crust, so the idea, sometimes breaks, and that releases huge amounts of energy.

They don’t have any direct observations for this though. Rather they analysed nearly seven thousand fast radio bursts from three of the most active sources. They found that their energy distribution and how much time passed between them was like that of earthquakes, and not like that of solar flares.

This is incredibly interesting, but the time-distribution of earthquakes is a classic example of a Poisson distribution, which is a very common event distribution. Other classic examples for a Poisson distribution are radioactive decay, calls to telephone hotlines, or the weekly number of births in a hospital. So fast radio bursts just gained a lot of potential for creative explanations.

Our sun by the way also has quakes that sometimes go along with big solar flares, though they look more like ripples in a pond.

A one million kilometre pond of nuclear plasma. But speaking of big solar flares, this is what we’ll talk about next.

Scientists in France and the UK have found evidence of the largest solar storm we’re ever heard of.

A solar storm is the aftermath of a big solar flare, in which plasma filaments erupt from the sun and blast out lots of radiation and charged particles. If the blast happens to go into direction Earth, we get showered by those charged particles. Large solar flares can create currents in electronic devices which can cause significant damage. It’s a problem especially for satellites, long-range communication, and old power grids.

The biggest previously known solar storm that hit earth was the Carrington event in 1869. At that time the biggest problem was that some telegraphers got electric shocks.

If the same thing happened today, we wouldn’t get away that easily. In twenty thirteen the insurance market Lloyd’s estimated that if a solar storm similar to the Carrington event hit Earth today, the damage to the electric grid would cost between 1 and 2 trillion US dollars to repair.  

Understanding how frequent big solar storms are is super important to figure out how urgently we need to upgrade the electric grid. And scientists have suspected for some time that Earth’s past has seen bigger events than the Carrington event.

One line of evidence they have pursued is tree rings. This is because if the charged particles from the sun hit the upper atmosphere, they do, among other things, create a radioactive isotope of Carbon, Carbon-14, with a half live of about 6000 years. Trees absorb this isotope. You can therefore use old tree rings to look for spikes in Carbon-14.

In 2012, the Japanese researcher Fusa Miyake found high levels of carbon-14 in Japanese trees dating to the 8th-century.

The same was later seen in tree rings from several other countries, including Germany, Russia, and the United States. And not only this, but scientists also have since found evidence for several more such events in tree rings. By now they count six of what is now called a “Miyake event”.

Some of those events, however, seem to last too long to plausibly have been solar flares. Some scientists have suggested it might have been a supernovae blasted off nearby.

The authors of the new paper looked at more than 14 thousand years old tree rings from France and report a massive spike in Carbon 14 around that time. It seems to indicate that back then a solar storm about 10 times stronger than even the Carrington event hit earth.

Good thing they didn’t have smartphones back then. How do you even survive in a cave if you can’t order takeout?

Researchers at the American National Oceanic and Atmospheric Administration have found microscopic bits of space debris in the upper atmosphere.

Satellites and rockets burn up when re-entering the atmosphere. Or at least they’re supposed to.This spectacle happens mostly in very far up layers of the atmosphere called thermosphere and mesosphere. But there’s no guarantee that what happens there stays there.

Indeed, some of the stuff rains down on earth, and there have been studies about it. But in a new paper, they looked at small particles, called aerosols, that spend a long time lingering the stratosphere, that’s the layer below the mesosphere.

They found that about 10 percent of aerosol particles up there contained metals that almost certainly come from space debris burning up in the atmosphere. These particles are called reentry aerosols.

It's not just satellites and rockets that burn up in the atmosphere of course, there are also meteors. But their chemical composition is different. Man-made stuff tends to contain more lithium, lead, aluminium, and copper.

We’ve known for a long time that this happens, but so far we didn’t have good data on how much of the stuff is up there. The question is now, what are the consequences?

As you may remember, spraying tiny particles into the stratosphere is the currently most popular method of geoengineering to combat climate change. These particles reflect sunlight, which counteracts the warming from carbon dioxide increase. However, the particles would have to be constantly replenished in the stratosphere, and it wouldn’t help with other problems caused by rising carbon dioxide levels, for example ocean acidification.

It’s rather unclear at the moment exactly what this powdered space debris will do, if anything. But maybe Elon Musk’s starlink satellites are really just a ploy for geoengineering?

Hello

Hi Elon,

A new premium subscription on X? Nice! If I sign up for the XL plan, can I get tweets delivered to my door with a SpaceX rocket?

Ohh. Okay. Wait until you hear about the Xmas jokes!

Love you too, bye.

IBM has unveiled a chip with new computing architecture that’s modelled after the human brain. A “neuromorphic” chip as it’s sometimes called. Their results from testing the chip just appeared in the jornal Science last week.

The new chip architecture, named NorthPole, has a whopping 256 cores and its special feature is that it can both compute and store memory with the same hardware. That’s just like the human brain, which also does information processing and memory storage in the same system. NorthPole has only 224 MB of memory but that’s plenty for applications in some places, like for example self-driving cars.

Scientists hope that neuromorphic chips will one day bring down energy consumption, but this one doesn’t quite fulfil that dream. It’s going to be a special purpose chip to speed up certain calculations, especially those requiring artificial intelligence, and I guess IBM will continue to try and further reduce energy consumption.

I’m not sure I want chips in self-driving cars to be inspired by the human brain. 

If it seems unlikely to you that we’d have missed an entire tectonic plate, think again.

Tectonic plates are large, mobile sections of Earth’s surface that fit together like a jigsaw puzzle. They should not be confused with continents which are geographical regions loosely describing connected landmasses. The motion of tectonic plates relative to each other creates geological features and frequently causes earthquakes. There are seven major plates and several smaller ones, or so we thought until now.

A research team in the Netherlands claims that Earth has a previously unknown tectonic plate. They’ve called the plate Pontus, and say it was once a quarter of the size of the Pacific Ocean.

Suzanna van de Lagemaat, a geologist at Utrecht University, and her team predicted Pontus’ existence more than 10 years ago from anomalies they claimed too see in Earth’s mantle.

Her group now reviewed existing geological and geophysical data from the Junction Region, where the North American Plate, the Eurasian Plate and the African Plate intersect. They also incorporated new data from northern Borneo. They then ran these findings through a software program and that spat out the additional Pontus plate.  

Well that’s interesting. Personally, I’ve found that a good way to find lost plates is checking the fridge for leftovers.

Astronomers believe they have seen the afterglow of an exoplanet collision and it wouldn’t have happened without twitter.

The group was originally observing a sun-like star roughly one thousand eight hundred light years away from Earth. That star had been dimming, which they thought was interesting. But when they shared the light curve of the star on twitter two years ago, someone in the comments pointed out a peculiar brightening in the infrared that had happened earlier.

The astronomers had a closer look and now believe that what happened in this system is this. There were two giant exoplanets orbiting around a star, and they collided. The collision created super-hot, big remnants that produced the infrared flare. Meanwhile, debris from the impact formed a big cloud which caused the dimming.

The researchers expect the debris to start spreading along the orbit of the collision remnants within the next few years, so we’ll get to continue to watch this space drama.

The French energy company Schneider Electric has published a white paper with a projection for power consumption by AI driven internet search. The numbers are staggering.

According to the international energy agency, the number of internet users worldwide has more than doubled in the past decade, and global internet traffic has increased more than 25 times. Data centres now account for roughly one and a half percent of global electricity use. That’s something in the ballpark of 300 Tera Watt hours energy per year, which is about as much as the energy consumption of the entire united kingdom.

That’s already a lot, but AI driven internet search and other chatting with your fav bot is more energy intensive than the usual lookup, so the question is what’s going to happen. Well according to the new whitepaper, AI currently runs on approximately 4 point 3 gigawatts, that’s roughly the output of four nuclear power plants.

They estimate that by 2028, the power requirement will go up more than a factor of three. That will be just about 20 percent of the total power consumption of data centres, but still the energy to supply that power will need to come from somewhere.

It's not just that AI is yet another burden we’ll have to carry around when trying to reduce fossil fuel use, the industry will also have to adapt. Data centres are likely to have to grow and increase their computing resources, which is going to make their management more expensive.

Of course, it’s all well and fine to talk about ethical implications of AI and how they’re going to enslave us, but if you want to get people’s attention, just tell them it’ll be really expensive.

Take the quiz for this video here. 

Files

(No title)

Comments

Anonymous

And now for something completely different: A group at Cambridge University (in the UK, that Cambridge) has a paper that is work-in-progress, but three of the group have been bold enough to post "as is" in ArXiv. ("Constraining Dark Energy from Local Group dynamics", by David Benisty, Anne-Christine Davis and N. Wyn Evans.) They expect that better observations with the JWST of the relative motion of the Andromeda galaxy and our Milky Way towards each other should produce, "in the near future", much better bounds both on the galaxies masses and on the value of Lambda, the cosmological constant in the field equation of General Relativity. Providing a better estimate of Dark Energy, because the Cosmological Constant is thought to be related to the density of Dark Energy, while also helping sort out which one is best among proposed patches Newton's gravity/dynamics such as MOND. Newton's has already been patched for many decades to get "post-Newtonian" gravity and dynamics (with "relativity corrections") I am happy to work with, as I never have to do with things far beyond the Solar System or closer to the Sun than Venus, or maybe Mercury, with those "corrections". And all of that is a re-patched patch on GR. For the PDF: https://arxiv.org/pdf/2306.14963.pdf

Aleksei Besogonov

It's surprisingly hard to make fair dice or lottery balls.