There is a particular kind of dread that comes not from sudden catastrophe but from watching a line on a graph quietly change its mind about itself. For decades, the line went up steadily — 0.2 degrees Celsius per decade, like a patient climb up a hill everyone agreed was there. Then, around 2013 or 2014, something happened. The line stopped climbing and started running.
A study published this week by Grant Foster and Stefan Rahmstorf at the Potsdam Institute for Climate Impact Research confirms what the data has been whispering for years: global warming has accelerated. Sharply. After stripping away the noise of El Niño, volcanic eruptions, and solar cycles, the underlying human-caused warming rate has nearly doubled — from roughly 0.2°C per decade since the 1970s to about 0.35°C per decade over the past ten years. The statistical confidence is above 98%. Five independent temperature datasets, produced by different organizations with different methods, all tell the same story.
What strikes me is not the number itself. It is the shape of the betrayal. We built our plans — our emissions targets, our Paris timelines, our political speeches — on the assumption of a straight line. A bad line, yes, but a predictable one. We were preparing for a flood that rose an inch a year. We are getting one that rises faster every season.
Stefan Rahmstorf puts it with characteristic German precision: if this new pace continues, the 1.5°C threshold of the Paris Agreement will be breached before 2030. Not by 2050. Not by 2040. Before the decade is out. The line bent upward, and our calendars became fantasies.
The study does not claim to know exactly why the acceleration happened. That is honest science — detect first, explain later. But the broader context is hard to ignore. James Hansen and colleagues at Columbia University have been tracking Earth’s energy imbalance — the difference between incoming solar radiation and what the planet radiates back — and it too has roughly doubled since 2015. The planet is absorbing more heat than it used to, and the oceans are storing it at record rates. Something has changed in the system. The brakes came off.
I think about what it means to live inside a system that is accelerating. Not warming — accelerating. The difference matters. A constant warming is a tragedy you can learn to live beside, like a chronic illness managed with medication. An acceleration is a different creature. It means the summers of your children’s twenties will not be like the summers of their teens. It means the extremes we are shocked by now will become the baseline we are nostalgic for later. It means every prediction is already wrong by the time it is printed.
There is a small cruelty in the timing of this study. It arrives in early August, when the Northern Hemisphere is already sweating through another record-hot summer, when wildfires in the Mediterranean and North America have become background noise, when the phrase “unprecedented heat” has lost its ability to precede anything. We are being told the fever is running faster while we are already burning.
And yet. The study also says something else, almost in passing: this acceleration is consistent with climate models. The physics has not broken. We are not in uncharted territory — we are in charted territory we chose not to look at. The models can produce periods of faster warming. They did not predict this exact decade, because models do not predict exact decades. But they said it was possible. They said the curve could bend.
The curve bent.
What do you do with a line that has changed its mind? You change your mind too, or you become a footnote in its history. The 1.5°C target is not a moral threshold — it is a statistical one, a number chosen by committees because it sounded achievable. The planet does not care about committees. It cares about joules per square meter, about aerosols and albedo and the chemistry of the upper atmosphere. The planet is running a physics experiment. We are the variables.
I keep coming back to that 98% confidence figure. In science, that is as close to certainty as you get without being certain. Five datasets. Two statistical methods. One curve, bending upward, visible to anyone who looks.
The fever learned to run. The question is whether we can learn to run faster.