Sabine
patreon
Science News Oct 17
🕑 Added 2023-10-17 16:10:01 +0000 UTC[This is a transcript with links to references.]
Welcome everyone to this week’s science news. This week we’ll talk about three new laws of nature, one of which supposedly shows that we live in a computer simulation. We’ll have a first look at intergalactic filaments, talk about building roads on the moon, whether chatbots understand what they chat about, a new type of qubit with a low error rate, sound waves in crystals, how to deflect lasers with nothing but air, dunes on mars, the first hurricane prediction market, and of course, the telephone will ring.
This has been the New Laws of Nature week! There was not one but three papers that put forward new laws of nature and they were all pretty much on the same topic, the growth of complexity.
They’re trying to address a long-standing mystery. It’s that complexity in the universe at large and in smaller systems within the universe seems to increase, under certain circumstances. Simply put, the universe made us, and we don’t know why.
The second law of thermodynamics seems to imply that complexity must eventually decrease because entropy will wash it out. But we seem to be missing a natural law that tells us in which situations complexity arises in the first place. The problem starts with even trying to quantify what we mean by complexity.
The three papers are of very different quality. The first one is about something the author calls “the second law of infodynamics”. It’s an idea he proposed in an earlier paper. In the new paper he claims that this law is fulfilled and that supports the idea that we live in computer simulation.
The problem is the way that he defines his new law it’s just identical to the second law of thermodynamics. It’s not wrong but it’s not new either and it’s nothing to do with computer simulations.
The second paper comes from a group of philosophers. On the upside it makes more sense that the first paper, because it’s specifically about systems that undergo some kind of evolution. On the downside it’s mathematically vague.
They propose to measure complexity by a quantity called “functional information” that was introduced by another author about 20 years ago. It tells you loosely speaking how good a system is at fulfilling a certain function. In the new paper they now call their idea the “law of increasing functional information”. So, systems improve how they fulfil certain functions.
The problem is, as they write themselves, that this functional information can only be calculated when you specify the function of a system, which moves the burden from figuring out what complexity is to figuring out what a function is.
The authors of the third paper don’t explicitly claim they introduce a new law of nature, they’re a little bit more modest, but address the same question. They do it with an idea they call “Assembly Theory”. The idea is that the complexity of an object can be measured by how difficult it is to assemble and how well it can make copies of itself. The good thing about this idea is that it’s mathematically well-defined. You can actually compute this quantity, at least theoretically. They look at some examples from chemistry to explain how it works.
But just because you have a mathematically well-defined quantity doesn’t mean it explains anything, so we’ll have to see if this idea is actually good for something. There’s much more to say about those papers, let me know if you want me to make a longer video about this. For now I’m sorry to say there’s no evidence that we live in a computer simulation and you’ll not earn two extra lives by shooting down that floating cabbage.
Scientists from the United States and Australia have for the first time seen intergalactic gas filaments. These gas filaments are mostly made of hydrogen. Because hydrogen is the simplest element, a lot of it was created in the early universe. If it clumps enough, it forms stars and solar systems and galaxies. But where it doesn’t clump it just lingers around and it’s hard to see. And yet, seeing this stuff is important to confirm that our model of the universe is correct.
Measuring this hydrogen is really difficult. They did it by looking for a particular emission line of hydrogen, known as the Lyman alpha line. If hydrogen atoms wiggle, this is one of the wavelengths they emit. The Lyman alpha line is in the ultraviolet when emitted, so we can’t see it. But the universe expands while the light travels towards us, so the wavelength stretches, and it’s shifted into the visible range. These emission lines are faint and difficult to tease out of the data from the rest of the universe’s light.
They did it with instrument called the Cosmic Web Imager at the Keck Observatory in Hawaii using a sophisticated background removal technique called nod-and-shuffle. This entails shifting the focus of the instrument from the source you want to image to its background and tracking how the combination of both changes. Then you can identify and subtract much of the background.
The sources they looked at were at redshift around 2 point 5, so about 10 billion light-years away. The volume that their observations covered is a slice of roughly 3 million light-years wide and 600 million light-years long. So it ain’t small.
If dark matter exists, which it may not, then it should fit with the structure of these filaments, so measuring them is another way to probe dark matter. It’s an important test because most of the *normal matter in our universe is actually not in stars, but floats around as such barely visible gas, either inside of galaxies or between them.
Space is really a bit like society, the stars attract all the attention, but the real power is in the dark web.
By the way, this video comes with a quiz on quizwithit.com, the fastest and easiest way to make your new knowledge stick.
Scientists have come up with way to build roads on the Moon: By melting the surface with powerful beams of light. As we get closer to living and working on the Moon, we’ll need roads because the moon is very dusty, and all that dust would damage vehicle engines. But building roads on the Moon isn’t as simple as it is here on Earth. Because the Moon has a small gravitational pull, shovelling soil there will kick up a lot of dust and gravel. That’s not the greatest working conditions.
So these scientists came up with the idea of using concentrated light to just melt the soil. On the moon they would use a 2-meter lens to concentrate sun light. But down here in the laboratory they tested it by using a laser beam with roughly the same power as the concentrated sunlight would have.
They tested their method with mock Moon dust that chemically resembles the real thing and was developed especially for purposes like this. Their laser was indeed capable of melting the dust into a liquid, which then solidified as a single structure. They studied the result with a scanning electron microscope and found that the crystal structure would make it strong enough to carry vehicles so that’s quite promising.
The nice thing about this moon autobahn is that you don’t have to worry about speeders. If they hit a bump they’ll just float off into outer space.
Max Tegmark, a physicist at MIT, best known for his idea that all of mathematics is real, and his collaborator Wes Gurnee just put out a new preprint. They set out to check whether artificially intelligences trained solely on language understand space and time. It turns out the answer is yes.
The two looked at Llama-2, that’s Meta’s large language model which is open source. They probed the network activation of the model for thousands of names of cities, global landmarks, famous people, headlines, song names, movie names, book titles, and anything that’s plausibly related to a location and a time. Then they ran an analysis on the network activation patterns to figure out how well they fit on a two dimensional map. It turns out the answer is the do so amazingly well and indeed the spatial relations strongly resemble the correct ones.
I found this to be very interesting because I argued in a video in March that the relations between words that we use to describe the real world necessarily capture some information about the physical world itself. Unless of course, we live in a computer simulation in which case I guess this isn’t the real world and none of that matters anyway.
Physicists at Harvard have made a big step forward with a new method of quantum computing. They are using single atoms as qubits, that’s the units of computation in a quantum computer.
These atoms are trapped with optical tweezers, that are laser beams which can hold and manipulate the atoms. I just told you in an episode the other week that atoms in tweezers are one of the newcomers in quantum computing. Compared to trapping ions, trapping electrically neutral atoms is more difficult, but if you manage to do it, they’ll disturb each other less.
In their new paper which was just published in nature they report that they set up an array of sixty atoms, and managed to entangle pairs with a fidelity of more than 99 point five percent. This basically means they entangled what they wanted to entangle and nothing else.
--This is comparable to the fidelity of quantum computing approaches that have been around for much longer like ion traps or superconducting circuits. So this makes atoms in tweezers quite a competitive approach now.
Hello,
Shoes that listen to music. Nah, I already have two teenagers. Jaja, Thanks for calling, bye.
Researchers in the United States and Denmark have for the first time seen how sound waves spread in crystals. Here’s how they did it.
They shot laser pulses at a gold film on the surface of the crystal. The light caused the gold film to heat up and expand, which created sound waves inside the crystal. Then they used an X-ray beam to probe the crystal and measured its reflections at intervals less than a picosecond, that’s 10 to the minus 12 seconds.
It's interesting because the way sound waves bump around in materials reveals their internal structure, kinda like talking to my husband.
An interdisciplinary research team in Germany has found a way to redirect laser beams with air. They have now applied for a patent for their device.
Lasers are usually redirected with optical gratings, that are flat surfaces etched with little parallel lines. But because most gratings are made of solids, like glass or silicon, they get damaged over time.
The researchers instead used pressure created by high intensity ultrasound to modulate the air density. This redirects lasers for the same reason water redirect light, the refraction index is not equal to one. In their recent paper for Nature Photonics, the researchers write that the grating can deflect light up to a power of twenty gigawatts, that’s the equivalent of roughly 2 billion LED bulbs, with fifty percent efficiency.
This is only a proof-of-concept device, but it’s a pretty cool method of contactless laser manipulation. It’s also an idea that the military will either be very interested in, or very interested in making disappear.
The European Space Agency shared some gorgeous footage from Mars Express, its 20-year-old Mars exploration orbiter. The footagecaptures Noctis Labyrinthus, a network of valleys about as long as Italy. Noctis Labyrinthus, “the labyrinth of night,” is believed to have formed when volcanic activity forced the planet’s surface to stretch, producing cracks up to 30 kilometres wide and 6 kilometres deep.
We owe this beautiful footage to the high-resolution stereo camera of the orbiter. It’s capable of filming at up to two-meter resolution. And because it can also do three D imaging, we get to see the topographical details.
--This isn’t the first time Mars Express has captured footage of this area, but it’s stunning nonetheless. Just a few decades ago, getting video this detailed was a dream on Earth, now we’re doing it on Mars, so at least something is making progress.
Hello
Hi Elon,
To Mars in three years?
Yes, if you could get me one in XS saying “Elon Musk went to Mars and all I got was this lousy T-shirt" I’d donate that to the red cross then, they’ll really appreciate your support.
Sure, talk soon.
A group from two UK universities is launching the world’s first hurricane prediction market.
The word “market” might make you think of fish, and while that isn’t quite right,it isn’t far off either. A prediction market is a way of collecting and evaluating information, not unlike figuring out which fish are good to eat, or the stock market. Fish, stocks, same thing really.
On the new hurricane market, users can bet on predictions and receive a reward if you’re right. To limit noise, the betting on such markets is usually confined to experts, in this case that’d be for example meteorologists, climatologists, and statisticians.
The market is run by CRUCIAL, an acronym for Climate Risk and Uncertainty Collective Intelligence Aggregation Laboratory. They’re now calling for participants in the betting market. So if some of your best friends are hurricanes, you might want to chime in and, who knows, soon enough you’ll buy twitter and fly to mars.
I need to tell you about this cute episode in the history of physics that I just recently learned about. It isn’t’ science news, but more like science olds, I guess, but it was news to me.
This story plays out in the 1960s in the United States. Albert Einstein’s theory of general relativity had become accepted, but it didn’t attract a huge amount of attention because no one really knew what to do with it.
One exception was the American physicist and astronomer Robert Dicke. He picked up on an idea by Paul Dirac, who had argued that the gravitational constant in Einstein’s theory isn’t constant.
The gravitational constant is the big G in Einstein’s Field Equations and it’s the same as Newton’s gravitational constant. If it was time-dependent, that’d change Einstein’s theory and have observable consequences.
Dicke developed this idea further with Carl Brans and it became known as Brans-Dicke theory. This Brans Dicke theory of gravity gives rise to a type of gravitational waves that do not exist in Einstein’s theory. They’re called scalar gravitational waves and they’re concentric.
Dicke now had the idea that such scalar gravitational waves could trigger earthquakes. And these earthquakes should be correlated with changes in the spinning rate of earth. This was a very distinct prediction and one that was testable. He set a student on the task.
Dicke’s idea was bunk, those funny gravitational waves were never found, and Brans-Dicke theory has by now been ruled out, but here’s where things get interesting.
The name of the student who he put on the topic was William Jason Morgan, who was at that time in his early 20s. He had studied physics, but now he had to learn a lot of geology.
Scientists had noticed around 1900 or so that the shapes of the continents fit together and had probably broken apart from one big piece in the past. This had given rise to the idea of continental drift, but they didn’t know how that worked or how it the plates drifted. Morgen keeps staring at images of earth and realizes that all that drifting can be described by spherical transpositions provided one also takes into account the oceans, not just the continental plates.
Morgan wrote his PhD thesis on the search for those funky gravitational waves, but would go on to publish his theory of tectonic plates, and developed secondary ideas from this, like that earthquakes and volcanoes preferably happen where tectonic plates meet and so on. Basically, all you’ve ever heard about tectonic plates goes back to Morgan.
Morgan just passed away some months ago at the age of 87. I learned about this in his obituary. I think this is an amazing story about how smartly pursuing a completely wrong idea led to groundbreaking scientific advances. Hope you liked it too!
This video doesn’t have a sponsor, so in my own mission I want to encourage you to check our new quiz app. We’ve developed this as the fastest and easiest way to learn. For just 2 and a half Euro a month you’ll get access to *all features, collect points, appear in rankings, see instantly whether your answers were right or wrong, see the full transcript with links to references, and soon we’ll also have some premium content. We’ll use your support to further develop the app and make it accessible to other creators, so you’re not just helping us.
Thanks for watching.
Comments
Thanks for the exciting episode again, the quiz has a fun-factor too, hope it is successful
Actually, of the two meanings of "function" in the "systems engineering" sense you have mentioned (and, by the way, though I very rarely, such as now, find a proper occasion to mention this, I have a doctorate in systems engineering) one of these meanings is what we intend the system to do, that is to say, what its function (noun) should be. The function (noun) of a telephone is to enable us to speak with people too far away to hear us and vice versa, because it has been designed to serve this very purpose. It's function (noun, again) is not to be a cooking utensil, such as a pot, so it won't be much good to try to make it function (verb) like one. And what was your relevant point, anyway?
"functional information can only be calculated when you specify the function of a system, which moves the burden from figuring out what complexity is to figuring out what a function is." Well, in Systems Engineering a function is a verb + noun. For example, a projector has a function called: generate light. Functions are what a system needs to do to accomplish its objective(s).
