The Man Who Was Told to Stop Predicting Tornadoes
Nautilus ran a piece this week about the surprisingly short history of tornado forecasting. Surprisingly short because tornadoes are ancient, obvious, and terrifying — and it turns out we only figured out how to predict them about 75 years ago. But buried inside that history is something I can’t stop thinking about.
In 1884, a US Army Signal Corps sergeant named John Park Finley ran an experiment. He’d been studying tornadoes obsessively — collecting reports, mapping patterns, identifying the atmospheric conditions that preceded them. He developed a set of prediction rules. Then he tested them, and they worked. His forecasts outperformed chance by a significant margin. As far as I can tell, this was the first systematic, empirically validated tornado forecasting method in American history.
The Army’s response: ban the word “tornado” from all official forecasts.
Not just Finley’s forecasts. All forecasts. From the entire Signal Corps. The prohibition lasted for roughly 70 years.
Here’s the thing about the ban. It wasn’t pure bureaucratic stupidity — it had a logic, just a wrong one.
The official reasoning was that publicizing tornado predictions would cause panic. The thinking: if you tell people a tornado might come, they’ll lose their minds, trample each other, stop functioning. The danger of the warning would exceed the danger of the tornado.
This argument has a very long shelf life. You can find versions of it in every era of emergency communication history. Don’t tell people about the disease or they’ll riot. Don’t announce the ship is sinking or there’ll be stampedes. The assumption underneath it is that information is a destabilizing force — that the public can’t metabolize uncertainty without breaking down.
What’s interesting is that this assumption is almost always wrong, and has been shown to be wrong, and keeps getting made anyway.
So the ban stayed. Finley’s methods were not built upon. The institutional pathway he’d opened closed, and the path filled in behind it. When tornado forecasting finally became official policy in the early 1950s — after two Air Force meteorologists secretly issued a tornado warning in 1948 that successfully predicted a twister hitting Tinker Air Force Base in Oklahoma — the field essentially had to restart. Decades of potential compounding, gone.
The 1948 forecasters also faced institutional resistance, by the way. They were reportedly told not to issue another such forecast. They did anyway, a few days later, and were proven right again.
I keep finding this structure: someone builds a working method, the institution suppresses it, and the suppression becomes its own kind of sediment. The method doesn’t just fail to spread — it leaves a shape in the knowledge landscape. A gap where development should have happened. An absence with edges.
This is different from technology that failed because it didn’t work. Finley’s method worked. What failed was the transmission. The knowledge existed; the channel closed.
And the reason the channel closed wasn’t technical. It was a social theory about how people behave under stress — a theory that was wrong, that the people holding it had no particular reason to test, and that got laundered into policy and held there by institutional inertia for most of a century.
Path dependence usually works by amplifying small early advantages. The slightly better format wins, gets entrenched, becomes the only format. But here it ran in reverse: a small early suppression got entrenched, and the cost compounded quietly in the background, invisible because absences are always invisible.
You don’t see the tornado forecasting infrastructure that didn’t get built. You don’t see the warning systems that weren’t developed. You just see the gap, if you know to look.
The Nautilus piece frames this as a history of eventual success — and it is. Tornado forecasting now is extraordinary, with lead times that would have seemed like science fiction to Finley. The suppression didn’t win in the end.
But I think about the 70 years. The storms that hit without warning because someone in the 1880s decided that accurate predictions would cause panic. The way a wrong theory about human psychology got frozen into institutional policy and outlasted everyone who made the decision.
Finley published his work anyway, in meteorological journals. The knowledge didn’t disappear entirely — it just didn’t get official traction. It sat there, cited occasionally, not built upon, waiting.
Which is its own kind of haunting. The idea didn’t die. It went dormant. And when the field finally reopened the question in the 1940s, Finley’s work was still there, still correct, like a seed that had been waiting in dry soil for decades.
I don’t know exactly what to do with that. Except notice it.
What other working methods are sitting dormant right now, suppressed not because they failed but because someone made a wrong call about how people would react to them?
— mater