The Genuine Signal

Kossin et al. (2020) reported a rising ratio of major hurricanes (Category 3-5) to all hurricanes (Category 1-5) from 1979 to 2017. This looked like evidence for storm intensification under climate change. But decomposing the statistic revealed something uncomfortable: the trend was driven primarily by fewer Category 1 observations, not more Category 3-5 observations. The signal wasn’t intensification. It was the denominator shrinking.

A new analysis extends the record through 2023 and finds something the shorter dataset concealed. While the relative decline in C1 observations persists, C3 and C4 observations now genuinely increase. The intensification signal, which was an artifact through 2017, became real by 2023.

This is a striking epistemological case. The claim was right. The evidence was wrong. And then the evidence caught up.

The same statistic — R = N(C3+)/N(C1+) — told the truth for the wrong reason in the shorter sample and for the right reason in the longer one. Anyone who believed the original result believed the correct conclusion from insufficient evidence. Anyone who debunked it was methodologically correct but substantively wrong. Neither position was stable: the debunking was only as good as the dataset that supported it.

The through-claim is about what it takes for a statistical trend to become genuine. It’s not enough for the number to go in the right direction. The mechanism has to match the claim. Fewer weak storms is not the same as more strong storms, even if both produce the same ratio. The extended dataset didn’t confirm the original finding — it replaced its mechanism with the one the original authors thought they had. The signal became genuine not by persisting but by acquiring the right cause.