January 27 2024, 22:46

Continuing to read Ed Yong’s Immense World. As always, sharing something interesting. THIS IS A LONG READ! For me — to research further, for you — I don’t know why you need this. But it will definitely be interesting.

It is known that the range of audible frequencies for animals is different from that of humans, but I didn’t realize just how different. Imagine the highest pitch imaginable – it would be just below 20kHz, as it is considered to be the upper limit of hearing. Both the upper and the lower limits tend to decrease with age. Most adults cannot hear sounds above 16 kHz. Anything over 20kHz we call ultrasound.

So, it turns out that our closest relatives, chimpanzees, can hear up to 30kHz, dogs up to 45kHz, cats up to 85kHz, mice up to 100 kHz, and moths even up to 300kHz. Just think about it, there are so many high-frequency sounds around us, and how rich their auditory world is compared to our restricted one. It would be interesting to walk around with headphones that compress the range from 20-40000Hz to 20-15000Hz. Many animals, such as mice, actively use ultrasound for internal communication, beyond the hearing range of their predators.

And of course, when it comes to ultrasound, one cannot overlook bats with their echolocation. It turned out to be wildly interesting.

Probably, everyone knows that bats successfully hunt in caves where no light penetrates, and they don’t crash into various stalactites and stalagmites. There’s a saying in English, blind as a bat, but in fact, bats do have vision. Some species have better, others worse. But let’s talk about echolocation.

In general, it’s just like radar. A bat screams, the sound bounces off a tree, returns to its ears, and it gets the information on how far away the tree is, and whether to stop or not. But, as they say, the devil is in the details. “Engineering” details.

Firstly, high-frequency sound attenuates quickly, so you need to scream very loudly for something to return from a distance of several meters. Beyond that, bats simply can’t “see”. So indeed, they scream very loudly and in a directed manner. Specifically, they’ve recorded at 138 decibels, which is the sound level of a jet engine if standing nearby. But in the ultrasonic range.

Secondly, when they scream so loudly, they need to plug their own ears to not destroy their sensitive apparatus. It turns out they have special muscles that block the inner ear during screaming.

Thirdly, they and their prey are actually moving very quickly and chaotically. Meanwhile, the speed of sound is about 343 meters per second. The bat’s brain must calculate the difference between the signal and the response, taking into account both its own motion in space and that of the prey. It was found that the vocal muscles of a bat can contract up to 200 times per second. Moreover, the frequency depends on the phase of hunting. 200 times – that’s the final phase, when the moth is right in front of them, and minor movements need to be tracked.

Fourthly, the bat’s brain also has to cope with not creating interference between what was screamed two moments ago and what was just screamed a moment ago. Considering that sound may reflect off a far wall and a nearby branch. Plus, there are waves from the screams of other bats, usually a lot in caves. For this, they seemingly throw slightly different modulations, plus their muscular system lets them “fire” very short pulses – a few milliseconds, and renew the pulses with their own frequency through very short intervals. Just think what kind of computer in their brains performs the inverse Fourier transform.

In conclusion, all this works rather well in small groups. But for example, Brazilian free-tailed bats live in groups of millions. Really, together 20 million mouths scream and wait for their echo from walls and each other. You just can’t pick modulation and frequencies simply, but they somehow manage. Not perfectly, and when they gather in really large numbers in a cave, their commuting to hunt and back in the cave is done “by memory” – presumably due to difficulties with echolocation. When a “door” was placed at the entrance to the cave, a bunch of bats smashed into it.

Fifthly, think about how they measure distance. You have to recalculate the difference between the sent signal and received one (amid a bunch of noise from other bats), and to hunt, this needs to be very accurately calculated. Of course, sound isn’t light, but 343 meters per second is still a lot. So, studies have shown that bats can recognize differences within millionths of a second, allowing them to determine distance in fractions of a millimeter. In other words, our eyes are much less precise than their ears.

Additionally, a typical moth is quite a complex 3D creature, which reflects sound differently with its various parts. Otherwise, bats would eat everything that moves. They discriminate. In complete darkness. A mouse’s scream contains a whole palette of frequencies, which reflect differently off parts of the moth, and the bat’s brain somehow manages to translate all this into a coherent picture. Moreover, for each of the constituent frequencies, the delay is different.

Then all this information is layered over time. Roughly speaking, a snapshot from one point combines with a snapshot from half a meter to the right, then half a meter forward and so on, many times, thereby “sharpness” and detail are enhanced. Overall, it’s the same with us – we only see the spot in front of us clearly, while the rest is constructed by the brain. But the bat’s brain weighs 1-2 grams compared to our half kilogram.

Now, think about it, flying with such a built-in radar, and in front of you are two branches at the same distance, which essentially produce the same response for their ears. To distinguish them and understand that it’s not one object but two requires a really advanced brain.

So, they send pulses lasting 1-20 ms, plus longer pauses between pulses. The pulses are complex in terms of frequencies, so such bats are called frequency modulation (FM) bats. But there are about 160 species where the scream lasts significantly longer – tens of milliseconds, but with short pauses, and instead of a complex range of frequencies, these use a pure “note”. Such bats are called CF – constant frequency. So, these bats have a problem with the Doppler effect – the increase in frequency as distance decreases. Since their brain is tuned to a strict frequency, say 87kHz for example, they might lose their prey if they receive a response shifted in frequency. And what they do – they scream at a sound speed lower, so that it ends up at the right frequency in the brain as a result of the Doppler effect.

Interestingly, their radar has two modes – forward and down, the replies from both are processed separately. The downward radar gives information about the position in space, while the forward radar gives the position in space of the prey.

In my research, I found that yes, after 20kHz humans hear nothing except for one exception — frequencies 2.4GHz and 10GHz, which belong to the microwave range. Yes, humans can “hear” these frequencies, but not through the ear, but “hear.” This phenomenon is called the microwave auditory effect or Frey effect. Initially, it was recorded by people working near radars during World War II, and the sounds they perceived were not heard by others. It turned out that with the impact of pulsed or modulated microwave radiation on areas around the cochlea of the ear, it is absorbed by the tissues of the inner ear, accompanied by their thermal expansio

January 26 2024, 16:56

It’s quite intriguing to look back about 40-50 years ago — how did people work in an office back then, assuming you aren’t scheduling appointments like a doctor might, nor are you a lathe operator who needs to physically craft things, nor a scientist, but just a manager. First off, it’s already hard to envision such a project that requires sitting in an office instead of a place where something tangible is being made — like constructing a building, a bridge, or planting trees. But let’s strain our imaginations and visualize it. Most probably, it would have been a government job, regardless of the country. In the office, you would have had a desk and a telephone. You arrive at work and sit at this empty desk. Then, over the course of 8 hours, the options are either talking to coworkers (about work or otherwise) or making calls and solving little issues, should something unexpectedly break down in the processes. It’s hard to imagine that distractions from work-related interactions could only be non-work-related interactions with colleagues, and nothing else. Thus, it’s very likely that you’d be burdened with some tedious nonsense – like sorting papers into three piles. It’s interesting to see how everything has changed. And how many managers now don’t need to step out from behind their computers in their offices.


preview

https://media.vanityfair.com/photos/5b58abac54e73b3d0270c056/master/w_2240,c_limit/google-workforce.jpg

January 25 2024, 16:27

Interesting from Anton Repushko:

“A lengthy text about how I’ve been going insane from despair over the last 8 months and started learning Ancient Greek.

Posts about incomprehensible white letters and jokes about lazy ancient Greek scribes have already flickered above. It’s time to put everything together into a coherent picture and explain myself. I can’t publish and explain the technical solution in detail due to the rules (it’s under review), but I’ll share various other things.

The competition is called the Vesuvius Challenge. When Vesuvius erupted and destroyed Pompeii, it also destroyed the city of Herculaneum at its base. We are interested in Herculaneum because there they excavated a villa that had a very rich library. And in this library, they have already found/expect to find about 4-5 thousand scrolls. Important clarification: from Antiquity, very few texts have survived to us: the Odyssey, the Iliad, all philosophy – just a small percentage of all texts that existed back then. Imagine how much knowledge and history are stored in these thousands of scrolls. But of course, there is a problem: they look like potatoes charred on coals due to conservation in lava and ash (but thanks to this they lay in the ground for 2000 years).

Enter Professor from Kentucky, Dr. Brent Seales, who spent his entire career pushing the following method: let’s make 3D CT scans of these scrolls and then try to cleverly restore and read them. The plan is reliable, like a Swiss watch, if it weren’t for one thing: the inks there are carbon-based and are invisible on MRI results. There were also attempts to scan the same scroll at different beam phases and by subtracting the results of this get some letters. The letters were enough for an article in Nature, but it looks pretty awful. What to do with this next?

Enter ex-CEO of Github, Nat Friedman: a philanthropist, organizes various AI grants, and invests in AI startups. A very nice fellow who helped organize this competition and stuffed the prize fund with money. Here’s his private fund and so far, to me, he seems like a healthy-person’s Elon Musk.

The competition itself started last February with a prize fund of 1M USD. It began on Kaggle, and then it all continued on Discord. There are 3D CT scans of the scrolls and the text needs to be read from there. The process of virtual unfolding of text proposed by the organizers is quite complicated and tricky, and now there’s about 7-8Tb of data, which sets a pretty high technical entry barrier.

After the Kaggle competition, things quieted down until in August, two students (one from the states, the other from Berlin) independently found some unclear reliefs with their eyes (!) that confirmed the initial idea: despite the ink being invisible, the paper remained deformed at the place of the written letters. Thus, the task became trying to programmatically learn to identify these deformed places (sometimes invisible to the eye) and read the letters. Around early August, that’s when I dived head-first into the competition.

So many sleepless nights, worries, and everything else throughout these months. I came and pestered anyone who could help me with advice/ideas (thank you all, friends. I couldn’t have done it without you). When discussing news and what happened during the day with my girlfriend, I thought dozens of times that nothing at all happened to me: I was solving the scrolls, just like yesterday, and will continue tomorrow.

And despite not meeting the official criteria for the main prize (four texts with 140 readable characters each), I think no one did. According to the organizers, they received about a dozen submissions. And for about a month now, we and the other participants have been waiting for the results of the papyrologists’ analysis. During these six months, I’ve met a lot of interesting people, won an intermediate prize of 5200 USD, started learning ancient Greek and greatly improved in solving such unclear problems.

Interestingly, during the competition, I felt incredibly fulfilled and it’s a very good alternative to all the races with LLM-research: you have a complex unclear problem, which intuitively seems solvable with current means. And the whole history of these scrolls (which were passed from hand to hand for decades until they could be read with new technical means) is a great example of human cooperation for something that can’t be measured in money. I really like it.

The organizers have already announced a new stage of the competition sometime in February, so there will be even more posts about it. Below is a picture with roughly the final results achieved: individual words and prepositions are readable, but without knowing Ancient Greek, it’s hard to read more. Let’s see what happens next.”