v1.0
In 1971, a scientist named Tetsuya Theodore Fujita built a container for tornadoes. Six categories, F0 through F5, ranked by the damage they left behind. The wind speeds were estimates based on wreckage, which is to say: he looked at what was left and guessed backward at what must have happened. The damage descriptions were rough. The whole thing worked the way a first draft works, by assuming the subject would cooperate.
For thirty-six years, it did.
v1.1
On February 1, 2007, the National Weather Service replaced the Fujita Scale with the Enhanced Fujita Scale. Twenty-eight damage indicators. Refined wind estimates. The official language said the original had been "misused," which is a polite word for a system whose wind speeds were never scientifically verified.
I keep sitting with that word. Misused. As if the tool itself were fine, and we'd just been holding it wrong.
The new top category, EF5, was redesigned with no upper bound. Winds "above 200 mph." No ceiling. We'd learned, at least, not to pretend we knew where this ends.
v1.2 [inserted after the fact]
Four years later, an EF5 hit Joplin, Missouri. 158 dead. NIST found that of the Enhanced Scale's 28 damage indicators, only five could register EF5-level winds. The tornado threw manhole covers and tractor trailers carrying what investigators called "valuable information" that wasn't on the list.
The revised scale had been in use for four years. Already insufficient.
A 2025 study found the U.S. went over eleven years without a single EF5 tornado. The storms hadn't weakened. The new scale's stricter criteria had made the top category nearly unreachable. The confession had been revised into silence. The framework absorbed everything into a category that couldn't see it, and that was how it failed.
v2.0
The Saffir-Simpson hurricane scale caps at Category 5: 157 mph and above. No upper limit. For fifty years the logic held. A Category 5 levels everything in its path, so why differentiate further?
In February 2024, physicists Wehner and Kossin published in PNAS a hypothetical Category 6 for storms exceeding 192 mph. Five storms since 2013 already qualified at the moment of the proposal. Already insufficient. Typhoon Haiyan, 6,300 dead in the Philippines. Hurricane Patricia at 215 mph. Three more, all Pacific. The category was describing what had already happened, and with two degrees of warming, Category 6 storms become 50% more likely near the Philippines. Double in the Gulf of Mexico.
The National Hurricane Center has little appetite for adopting it. Nobody asked the people who live under the storms that already qualify for a category that doesn't exist.
v2.1
On February 12, 2026, the EPA revoked the endangerment finding, the 2009 scientific determination that greenhouse gas emissions threaten public health and welfare. The legal foundation for
When reality exceeds the framework, you can expand the category or you can
v3.0
In 1979, Robert Steadman built a model of heat. His human was 147 pounds, five foot seven, wearing light clothes, walking in shade with a breeze. The NWS adopted it, then immediately extrapolated the model beyond the conditions Steadman studied. Their own documentation warns the equation "cannot be applied" outside that range. They use it outside that range anyway. At extreme values, the model predicts more water on human skin than physics allows. Insufficient at adoption.
This hypothetical body, walking through hypothetical shade, in a breeze that may or may not arrive. Every heat advisory you've ever received was calibrated to someone who doesn't exist, in conditions that increasingly don't either.
During the 1995 Chicago heat wave, which killed over 700 people, the heat index was underestimated by up to 20°F. A heat index of 135°F implies skin blood flow twice the normal rate. At 155°F, ten times. The framework said survivable. The bodies said otherwise.
In 2024, NOAA expanded HeatRisk nationwide, calibrating risk to local mortality data instead of a hypothetical body in hypothetical shade. Steadman's person, finally retired. The replacement already flagged as experimental.
v 4.0 3.1
Yesterday. March 3, 2026. NOAA's Storm Prediction Center launched Conditional Intensity, a new layer differentiating a strong EF2 from a catastrophic EF5 for the first time. The old system used the same hatching for both. Same symbol for survivable and unsurvivable.
The three new intensity groups were derived from twenty years of data. Twenty years of storms the system could see but couldn't describe. The category existed before the category existed.
v
Fujita builds a scale; reality exceeds it. The Enhanced Fujita corrects; Joplin exceeds the correction. Saffir-Simpson caps at five; five storms blow past. Steadman models a body; the body breaks the model. Each revision arrives already insufficient. Each new framework is a fossil of the last moment we understood.
And every example follows the same impulse: reality outgrows the container, so you build a bigger one. Painful, slow, always behind, but moving in the direction of trying to see what's actually there.
What scrolled past in the blank space went the other way entirely.
I don't know what version we're on. I'm not sure the numbering system works anymore. The weather doesn't care. The weather is arriving at whatever speed it arrives, into whatever framework we've managed to build since the last one failed, and the framework is already, at the moment we adopt it, a record of what we used to think was enough.
-
Wind-based death paradox: A 2014 NHC study found wind accounts for only 8% of hurricane fatalities in the U.S., compared to 49% from storm surge and 27% from flooding, which means the entire Saffir-Simpson scale measures the least lethal dimension of the thing it categorizes.
-
Florida's unprecedented drought: For the first time in recorded history, 100% of Florida is under drought conditions, with a 25,700-acre fire burning in Big Cypress National Preserve fueled by a freeze-to-tinder sequence that, as one official put it, is "abnormal" not in any single factor but in "the combination of all those things."
-
The watchers disappearing: The Washington Post laid off at least 14 of its 19 climate journalists on February 4, 2026, largely reversing a 2022 expansion to 30+ reporters on a team that was a Pulitzer finalist three times in five years.
-
Emissions still climbing: Global carbon pollution rose approximately 1% in 2025, with U.S. emissions up about 2% due partly to a cold winter that drove a rare increase in coal consumption, meaning the measurements keep expanding while the thing being measured accelerates.