Trilobites Invented Advanced Optics First | November 06 2026, 22:43

It turned out that the lenses in trilobites’ compound eyes were made of mineral. Each lens was a single crystal of calcite—the same stuff limestone, chalk, and marble are made of. The problem is that calcite strongly birefracts light: a ray passing through the crystal splits in two. Trilobites got around this by orienting the crystal in each lens so that its optical axis ran strictly along the line of sight, and there is no birefringence in that direction.

The second problem is spherical aberration. In a simple convex lens, rays passing through the edge converge somewhere other than the central rays, and the image gets blurred. A living eye can at least somewhat compensate for this through the shape and non-uniformity of the lens, but a rigid crystal cannot.

In one group of trilobites—the phacopids (Ordovician–Devonian)—the lenses were large and separate, and each consisted of two layers of calcite with different magnesium content: a convex upper part and an intralensar bowl” underneath (see picture). The boundary between the layers matches the profiles of aplanatic lenses described by Descartes and Huygens back in the 17th century. This shape cancels out spherical aberration—the rays passing through the edge of the lens converge at the same point as the central ones. There is even a hypothesis that these lenses were bifocal—the center and the periphery focused at different distances, so a fixed crystal could work for both near and far vision at the same time.

It’s pretty cool that evolution arrived at this solution roughly 400 million years before Descartes and Huygens, while neither of them lived to see the day when people figured out trilobite ophthalmology.

Unveiling BloomSpan: Revolutionizing Text Sequence Analysis in Large Datasets | October 12 2026, 04:10

I want to share our fresh work — the BloomSpan algorithm for finding the longest frequent sequences of words in an array of texts. I named it BloomSpan.

What is it, in human terms. Imagine you have a library of 50,000 books, and you need to find all the long phrases that are repeated word for word, without omissions, in many different books. Not “similar” phrases, but exactly identical chains of words, and as long as possible — such that if you add another word to the left or right, the phrase no longer occurs frequently enough. The task sounds simple, but with billions of words, it stumbles on memory and time. Classic algorithms either build huge indexes or go through too many candidates.

Where is it needed? For example, extracting patterns from documents (such as uniform “headers” in letters or contracts), searching for borrowings and plagiarism, highlighting stable phrases and clichés in large text arrays.

So, here’s what we got. The algorithm. Honestly, BloomSpan is not exactly an algorithm from scratch but a careful assembly of known ideas which, when combined, have a strong effect. At its core – different ideas from BIDE+, FHK, and on top of that: (a) The Bloom filter — a compact structure that answers the question “has this combination of words occurred frequently enough?”, without storing the combinations themselves. It can occasionally make a mistake in the direction of “yes”, but never in the direction of “no”, ensuring we never miss a phrase we need (b) “Depth-first” search directly through the text — we find a short frequent core and gradually expand it into a long phrase, without creating intermediate copies of the corpus. (c) An occupancy map — marking positions in the text already “covered” by found phrases, to avoid checking the same thing many times.

The results. On the Gutenberg Project corpus of 50,000 books (about 3 billion words), BloomSpan operates 11-15 times faster than the classic FHK algorithm and consumes up to twice as little memory. It is the only tested method that manages to complete the entire corpus within 40 GB. Moreover, where they do finish, the results fully match those of the precise algorithms — meaning the speedup does not compromise quality.

On synthetic data with an “unnatural” distribution of words, good old FHK is faster. BloomSpan is specifically tailored for living natural language, and the article discusses why this is so.

The preprint and code (Java) — in the comments

Exploring the Black Box: AI’s Struggle with Novelty Creation | September 14 2026, 20:19

I’ve devised a simple illustration of where current AI still falls short.

Imagine a black box with two knobs sticking out. You move the first one — the second one moves too, but according to its own rules, like with delay, inertia, threshold, however you like. In essence, the box transforms f(t) into f'(t). The question is: what’s inside? It’s necessary not only to devise an efficient mechanism but also to fit it within the box’s dimensions.

Formally, this is almost like programming. Because one can formally describe input-output and come up with an “alphabet” for mechanical components inside. But unlike programming, there’s a lot of nonlinear aspects – the mechanisms are bulky, and one must take into account numerous factors from materials to friction, and importantly, it’s not so simple to test a prototype before creating it. But the main thing is not even this – it’s understanding the problem. The only way to find out what’s inside the box is to put forward a hypothesis and devise a knob movement in which this hypothesis and an adjacent one yield different results. A good experiment here is more important than a good answer. And models are trained on a corpus of solved problems, where mistakes have already been cleaned up, and there nothing trains the ability to choose the next question.

The point is this. It seems that the challenge for AI (we’re talking about LLM; the field is broader)— isn’t the assembly of the known, it handles that increasingly well. That is, with an “alphabet” for describing the task and laws of mechanics, AI can solve the box problem through combinations of the known, rather than through a fundamentally new component. The challenge is inventing something new. Although, to be honest, most human inventions, when analyzed, also turn out to be an assembly from far-flung pieces. Yet, consider the invention of the escapement in clocks, where all the components — a wheel and an oscillating fork, and even the very idea that the oscillating system shouldn’t be driven by a spring, but should itself regulate when the spring is allowed to release a burst of energy, reverses causality in the device, and took centuries to achieve. And the lockstitch in sewing machines. It’s pointless to copy a hand with a needle, instead, it’s necessary to pull a second thread through the loop of the first from the back, and the needle must have the eye at the tip. All this isn’t just assembly from components, it’s changing what is generally considered the task, and that’s why such things are so poorly derived by mere combination, whether by humans or models.

An inventor has feedback from reality and a cost for mistakes, so they learn to test before believing. Within one answer, a model has nothing to pay with for a wrong guess. And besides, if you replace the mechanics with particle physics, everything has to be redone because it all works differently there.

Interestingly, this is not cured by the size of the model, but by the structure of the task. Let the box respond to movements — and the task transitions from textual to experimental. Let’s see who learns to play this game first.

Mapping the Minds of Search & RecSys: A Comprehensive Directory | August 10 2026, 03:56

84 researchers, 40+ laboratories — who’s who in Search & RecSys, all on one page.

https://www.testmysearch.com/who-is-who.html

Google, Meta, Netflix, Spotify, Amazon, CMU, UMass, UCSD, and others — for each person links to their website, Google Scholar, LinkedIn, key papers. Plus the h-index and number of citations from Semantic Scholar.

Made it for myself, because it’s more convenient to quickly see who publishes what. Couldn’t find anything ready-made — had to compile it myself. Seems useful, I hope not only to me.

https://www.testmysearch.com/who-is-who.html

https://www.testmysearch.com/who-is-who.html

Exploring the Depths of Unicode’s Complex Symbols | August 09 2026, 12:22

It’s amazing that Unicode includes such complex “characters” as shown in the picture. This is a single character. When the Unicode table had accommodated all the alphabets of the world, arrows, circles, and there was still a ton of free space left that needed filling, the capacity was used up. It has space for about 1.1 million characters, and font designers are doing their best) Chinese and Korean occupy 71% of the existing characters.

Also, Unicode includes some cool ligatures. For example, the following text is one “character” ﷽ (Basmala). Here on Facebook, it doesn’t display quite right because of the fonts, the full version can be seen here: https://www.urbandictionary.com/define.php?term=%EF%B7%BD where ﷺ and ﷻ also live

Moreover, Unicode includes hieroglyphs that do not exist. An investigation in 1997 discovered, for example, the origin of 妛: the compilers wanted to record “山 over 女,” but could not print the composite sign — so they printed 山 and 女 separately, cut them out with scissors, glued them on a sheet and made a copy. On the copy, the seam line of the two pieces of paper looked like a stroke and was included in the hieroglyph. Now it remains so forever

A Close Encounter with a Giant Horsefly: Insights and Discoveries | August 05 2026, 01:33

I just had a huge, about three and a half centimeters across, loud buzzing black fly come into my room. I caught it with a blanket and shoved it into a vase, inspected it closely and Googled it. It turned out to be a black horsefly (Tabanus atratus), female. The females are vampires. They need fresh blood. For the eggs. And their bites are very painful. Essentially, a horsefly’s mouth is like scissors. A pair of serrated mandibles which they use to saw into the skin, opening up capillaries, and then they lick up the blood from the resulting pool. It’s said that their bites hurt a lot because you are literally being cut by these scissors. An anticoagulant is injected into the blood to prevent it from clotting, so the wound oozes for a long time after the horsefly has flown away. They can easily bite through fabric, so a shirt offers no protection.

So, it turns out that horseflies are possibly the fastest flying insects on the planet. Males of the genus Hybomitra (which also live here) are credited with speeds up to 145 km/h, and this, by some estimates, is the record for horizontal flight among insects (though this figure is indirectly obtained: an entomologist made a male chase a bullet shot from an air rifle, it intercepted it in the air, meaning it was flying faster than the pellet). Furthermore, horseflies are attracted to polarized light. This is how they find water — a smooth surface reflects light horizontally polarized, and for the horseflies, this is a beacon “here you can lay eggs”. Interestingly, this is also why white horses get significantly fewer bites, because they almost do not polarize reflected light. For these experiments, Gabor Horvat and his colleagues received the 2016 Ig Nobel Prize. Among other things, they smeared full-size horse models with glue and painted dummies to thoroughly check everything.

By the way, I released my captive.

Exploring Science and Myth through the Engineering Zen Group | August 03 2026, 15:37

I remind you that I have a group called Engineering Zen on Facebook and Telegram. Recently, there have been posts about:

– How for half a century science condescendingly explained to grandmothers that colds are caused by viruses, not drafts, — and in 2022, it found a mechanism by which grandmothers were partly right.

– People who are currently walking around with plutonium in their chests. Legally, registered with the nuclear regulator, with the instruction “to be returned to the state after death.” The capsule was seriously tested: cremation, plane crash, shotgun blast. It wasn’t radiation that killed the technology.

– Old sailors who poured oil overboard during storms. Not superstition, but working physics — and from it emerged, it seems, the first measurement of the size of a molecule in history. A teaspoon of oil, a pond under London, and Benjamin Franklin. Today, the same effect captures oil spills from satellites.

– Why almost all blue in the animal world is a deception: there’s no pigment at all. Neither in the wings of the morpho butterfly, nor in the feathers of a peacock, nor in your blue eyes. Mammals do not have any blue pigment. But it’s the opposite in plants.

– How the ocean floor is mapped… by looking at its surface from space. Underwater mountains betray themselves through “bumps” on the water — it turns out the sea is not flat at all. Only about a quarter of the sea floor has been measured with a sonar to this day.

– The MRI magnet, which is turned on only once in a lifetime — and then it circles current through a ring for years, without any socket. Even at night, when the clinic is closed. But if the cooling fails — a “quench” occurs, and tons of helium burst out through the pipe to the roof within seconds.

– The theorem by which a sphere can be cut into pieces and assembled into two identical spheres. Not a trick or sophistry — honest mathematics. The catch is that these “pieces” have no volume. At all. Like the coast of Britain has no length.

– Why in a language with three words for colors, the third is always red, and blue appears last. No need to know the language — just count. And why the go signal on Japanese traffic lights is legally “blue.”

– A rescue operation going on in space right now: a robot trying to catch a falling telescope, which has neither handles nor a docking unit — it was built in an era when no one thought of such things. Racing against the atmosphere.

– Lunar dust — the main nightmare of lunar bases. Microscopic sharp glass shards that eat away space suits within hours, stick to everything due to static, and — surprisingly — smell like gunpowder.

– A hypothesis that lasted 85 years and fell during the World Cup final — from a single line on Twitter. A counterexample is verified by a student in a minute, and it’s unclear why it wasn’t found earlier.

– The illusion of “stopped clocks,” which everyone has experienced: you shift your gaze to the second hand — and it has stopped. It turns out, the brain blinds you several times a second — and then edits time retrospectively.

– Ice that does not melt at +300°C — found trapped inside diamonds from the Earth’s mantle. And its even more strange black superheated relative seems to make up the interiors of Uranus and Neptune.

– Why the ads on the stadium boards, which you see on television, physically do not exist — viewers in Brazil, Germany, and Japan see three different sponsors at the same moment.

– A piece of wire mesh that tames fire where it is bound to explode — and on which the entire century of mining labor stood. The inventor gave it to the world for free.

– And for dessert — why a giant NASA rocket is transported not on asphalt, but on river pebbles. They tried asphalt — it went “in waves, like the sea.” Personally verified, I spent an hour looking at that gravel out the window.

Join us to not miss much more interesting stuff.

Engineering Zen

P.S. There is also a group in Telegram with the same name, but there the audience has stopped growing at all and stands at about 100 people, and until I figure out how to manage this, I am temporarily giving up on it.

Grandmothers Partially Right: The Cold Truth About Catching Colds | August 02 2026, 21:03

HEY, for half a century, science condescendingly explained to grandmothers that a cold is caused by a virus, not a draft, and in 2022 found the mechanism by which the grandmothers were right after all (partly).

Background: Salisbury, 1946-1990, Common Cold Unit, 20,000 people went through it – they were infected with cold viruses and frozen (cold baths, drafts, wet swimsuits). They did not get sick more often. Next – Douglas’s work in 1968. There were 44 people, all given rhinovirus, one group sat at 4°C. No difference was found, and the article was titled “Failure to demonstrate effect”. The conclusion spread through popular books – cold has no relation to getting a cold, period.

New: In December 2022, they published the mechanism. In short. The nasal mucosa has a frontline (only discovered in 2018). Sensing a pathogen, cells throw extracellular vesicles into the nasal mucus – the authors describe this as “kicking a wasp nest”. These vesicles feature the same receptors that the virus latches onto, denser than on the cells themselves – it sticks to the dummy and doesn’t reach the cell. Then the experiment – 15 minutes at 4.4°C – and the temperature inside the nose drops by 5°C. On tissue cooled just the same, nearly 42% fewer vesicles are thrown, antiviral microRNAs in each are halved, and up to 70% fewer receptors.

Conclusion: “getting a cold from cold” is wrong, but “easier to get infected in the cold if there is someone to catch it from” is very much so. The most interesting part is what they want to make out of this. Since the swarm can be summoned, the authors discuss a spray that commands the mucosa to “kick the wasp nest” in advance.