A craze, a quorum, and a clever bee

01A leaf priced like a luxury good

Matcha begins as tencha, a tea grown under shade for weeks before harvest, then dried and ground into powder between two slabs of granite. It has always been a small corner of Japanese farming. This spring the price of that leaf reached levels no one in the trade had seen: the Uji first-flush tencha average hit ¥13,972 a kilogram, roughly 1.7 times the prior record, while in Kagoshima tencha climbed toward ¥19,000.[4]

Exports tell the same story from the other side of the counter. Japan shipped 13,125 tonnes of green tea in the year to March 2026, up 42 percent by volume, and the value of those shipments more than doubled to ¥84.7 billion; powdered tea, most of it matcha, made up around seventy percent of what left the country.[3] A decade ago the export trade was a fraction of that size.[6]

The houses closest to the leaf reacted the way anyone rations something they cannot restock. Marukyu-Koyamaen, an Uji maker with roots in the Edo period, capped how much matcha a customer could buy and noted plainly that its leaf can be gathered only once a year.[1] Ippodo, founded in 1717, told customers it was retiring several everyday bancha and hojicha lines because farmers were pulling plants over to tencha, and that its own matcha prices had risen more than two-and-a-half-fold.[2]

02Why matcha can't just scale

The price absorbed the whole shock because almost nothing about matcha bends quickly. Tencha is only about six percent of Japan's tea, and the best of it is cut in a single spring flush.[11] A traditional granite mill turns out only about forty grams of matcha an hour, roughly ten cafe lattes' worth, where an industrial mill would manage that in a moment.[9]

The people who grow it, meanwhile, are ageing with few successors behind them. Japan's harvestable tea area has fallen to about 25,400 hectares, down some thirty percent in a decade, and the average tea farmer is now sixty-five.[7][10] Land, labour, leaf and the grind itself all move on their own slow clock, and none of them can answer a demand curve that doubled in a year.

03The gap fraud fills

When real matcha runs short and dear, imitations move in. Investigators in Japan have documented Chinese-made powder sold online under Uji makers' own brand names, a yellowish product passed off as premium green tea, and because “Uji matcha” is not trademarked in China, the makers have little recourse.[8]

The boom is also draining the rest of the tea shelf. As land and leaf shift toward matcha, everyday sencha has been caught in the same squeeze; one Shiga auction this spring drew a top bid nine times the previous year's.[4]

What to watch

The fraud makes the headlines, but the structural story is the one to keep an eye on. Supply cannot answer within a season or two, so the likeliest path is not a spike and crash but a lasting split — a protected premium tier of Uji and ceremonial grades, and a globalised culinary tier increasingly ground in China and elsewhere.[7] The real exposure sits with the growers converting to tencha at the very top of the market. If a craze built substantially on social feeds cools the way wellness fads do, the newest entrants could be left holding leaf the world has stopped queuing for.

01A landslide that changed nothing

The referendum asked two things. Whether to scrap the lifetime payment recently granted to two-term prime ministers and parliament speakers, a benefit that until then belonged only to former presidents; and whether to restore the Office of the Special Prosecutor and the elite police unit known as NAKA, both dismantled by the government in 2024.[4][10]

Among those who voted, more than ninety percent said yes to both. It made no difference. Turnout was 16.13 percent, just 705,227 people out of an electorate of about 4.36 million, and the constitution validates a referendum only when a majority of all eligible voters take part, a bar this vote missed by more than thirty points.[3][2]

This was Slovakia's tenth national referendum since independence in 1993, and only one has ever counted: the 2003 vote to join the European Union, which cleared the same bar with 52 percent turnout.[7] The closest recent echo came in 2023, when a referendum on early elections won 98 percent approval and still died on turnout.[6]

02The quorum trap

A turnout quorum, a rule that requires half the electorate to show up for the vote to count, sounds like a guard against frivolous ballots. In a divided country it works the other way. Whoever expects to lose the argument can win the result by keeping supporters home, so the referendum turns into a contest of mobilisation with the wind at the back of whoever benefits from apathy.[8]

That is roughly what happened. Prime Minister Robert Fico did not vote and dismissed the exercise as pointless and politically motivated, a stance that worked less as a call to vote no than as a signal to sit the day out; the opposition party that gathered more than 350,000 signatures to force the vote could never supply, from its own base, the roughly two million ballots the quorum demands.[3][5]

The reading that fits

Set against that rule, the government's line, that low turnout proves Slovaks don't care about corruption, is the weaker one. When nine in ten participants back a measure, the binding constraint is turnout, and turnout here was held down on purpose.[3] The more consequential effect is abroad. A failed vote lets Bratislava tell Brussels that citizens themselves declined to reverse the rollback, even as the European Commission weighs cutting EU funds over those same dismantled institutions, a quarrel that pulls Slovakia closer to Viktor Orbán's Hungary.[9] What quietly strengthens with each failed referendum is the incumbent's hand: the next opposition learns the tool is dead.

01What the bees actually did

The setup borrows a classic test of animal wit. Wolfgang Köhler once watched a chimpanzee stack boxes to reach a banana hung out of reach; here the banana was a blue ring the bees had learned to link with sugar water, lifted to the ceiling of an arena too shallow to fly in, and the box was a small ball they had to shove underneath and stand on.[7][8] The design was built to rule out a bee simply chasing a reward it could see.[3]

The bees were not blank slates, and that matters. Beforehand they had learned two separate facts — that the blue flower meant a reward, and that the ball was movable and harmless. What no one showed them was how to put those facts together. Doing exactly that, unprompted, is what the authors call spontaneous.[2] The trick has a lineage: back in 2017 the same researchers taught bees to roll a ball to a target for a treat, and even to improve on what they were shown.[6]

The clinching experiment hid the goal completely. With the flower tucked behind a screen, 23 of 30 bees still moved the ball to the right place, and most went straight there without first trying the wrong side — the opposite of blind trial and error.[1] Even a simpler version, with a single barrier over the flower, was solved by 16 of 22.[9]

02Why a small brain is the whole point

The headlines reached for chimpanzees and elephants, and the comparison is the authors' own, carried over from that box-and-banana test.[7] It is also where the coverage overshoots. Doing a task shaped like a chimp's is not the same as having a comparable mind, and the researchers were careful to say so — they decline to claim human-like reasoning, or even to call it tool use.[2][8]

The genuinely subversive claim is the quieter one. If a brain this small can recombine what it already knows to crack a new problem, then flexible thinking may need far less hardware than a vertebrate-centred science assumed. The task, not the animal, is what turns out to be cheap.[4]

How strong is “insight”?

The experiment is clever, and the core finding will probably hold. The label is carrying more than the evidence can bear. The strong result rests on a single test of thirty bees, a related control returned nothing, and “spontaneous” really describes the recombination, since the pieces were trained in first.[1] There is a deflationary reading, a learned map plus a practiced push stitched together, that this study narrows but does not kill, and tellingly the definitive version of that critique came from the same research lineage a decade ago.[5] Why it matters beyond a puzzle box is the stakes: if insects genuinely solve problems, the quiet harm pesticides do to their learning stops looking like a footnote, since one common neonicotinoid pesticide slows a bee's learning by a quarter to a third.[10]