The risks of communicating extreme weather forecasts

For decades, climate change researchers and activists have used dramatic predictions in an attempt to influence public perception of the problem and as a call to action on climate change. These predictions were frequently for what could be termed apocalyptic events because they predict cataclysmic events resulting from climate change.

In a new article published in the International Journal of Global Warming, David Rode and Paul Fischbeck of Carnegie Mellon University argue that such forecasts are crap. “The truly apocalyptic forecasts can only be seen in their failure, that is, the world did not fall apart as expected,” says Rode, assistant research professor at the Carnegie Mellon Electricity Industry Center, “and observing a series of repeated failures of doomsday predictions can undermine public confidence in the underlying science.”

Rode and Fischbeck, professor of social and decision-making sciences and engineering and public policy, collected 79 climate-caused apocalypse predictions dating back to the first Earth Day in 1970. Over time, many of these predictions expired since; looking at all the fateful dates we should have died like cockroaches, well the sun shone as usual and the pigeons continued to fuck each other happily. In fact, 48 (61%) of the predictions have already expired at the end of 2020.

Fischbeck noted, “From a forecasting standpoint, the ‘problem’ is not only that all expired forecasts were wrong, but also that most of them never admitted any uncertainty about the date. About 43% of the forecast in our data set made no mention of uncertainty.”

In some cases, forecasters were both explicit and certain. For example, the absolute embodiment of misfires and real crooks for their predictions are Stanford University biologist Paul Ehrlich and Prince Charles, repeatedly expressing a high degree of certainty about doomsday weather events.

Rode commented “Ehrlich made predictions of environmental collapse going back to 1970 which he described as having near certainty” with a new final end date. Their predictions have been repeatedly apocalyptic and very certain … and so far they have also been wrong.

The researchers noted that the average time horizon before a climate apocalypse for the 11 predictions made before 2000 was 22 years, while for the 68 predictions made after 2000, the average time horizon was 21 years. Despite the passage of time, little has changed, through half a century of forecasting; the apocalypse is still around 20 years old.

Fischbeck continued, “It’s like the boy who cried wolf over and over. If I see many successive forecast failures, I may not be willing to take future forecasts seriously. ”

This is a problem for climate science, say Rode and Fischbeck.

“The science behind climate change has a lot of solid results,” explains Fischbeck, “the problem is often the leap in connecting predicting climate events to predicting the consequences of those events. Human adaptation and mitigation efforts, along with the complexity of socio-physical systems, mean that predicting sea level rise, for example, may not necessarily lead to doomsday flooding.

“By linking the climate event to the potential consequence of a dramatic effect,” Rode noted, “failure to observe the consequence may unfairly call into question the legitimacy of the science behind the climate event.”

With the new Biden administration making climate change policy a top priority, reliance on scientific predictions of climate change is more crucial than ever, but scientists will need to be wary of qualifying their predictions.

By measuring the proliferation of forecasts via search results, the authors found that forecasts that did not mention uncertainty in their doomsday date tended to be more noticeable (i.e., they had more than search results available). Making sensational predictions about the fate of mankind, while scientifically dubious, has always proved tempting for those wishing to make headlines.

The problem is that scientists, by virtue of their training, tend to make more cautious statements and more often include references to uncertainty. Rode and Fischbeck found that while 81% of predictions made by scientists referred to uncertainty, less than half of predictions made by non-scientists did.

“It’s not surprising,” said Rode, “but it’s troubling when you consider that forecasts that refer to uncertainty are less visible on the web. This means that the most visible voices are often the least qualified. ”

Rode and Fischbeck argue that scientists should be extremely careful in reporting high profile events. When it comes to climate change, the authors advise “think small”. That is, focusing on less grandiose and shorter-term predictions. “If you want people to believe big predictions, you have to convince them first that you can make small predictions,” says Rode.

Fischbeck added: “We need forecasts of a wider variety of climate variables, we need them done regularly and we need expert assessments of their uncertainties so people can better calibrate themselves. depending on the accuracy of the forecaster.”