How hot could the world get?
25:58 Share

Silicon Valley Bank is still the SVB you know and trust. The SVB that delivers human-focused, specialized lending and financial solutions to their clients. The SVB that can help take you from startup to scale-up. The SVB that can help your runways lead to liftoff. The only difference? Silicon Valley Bank is now backed by the strength and stability of First Citizens Bank. Yes, SVB. Learn more at svb.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.com.

Support for this show comes from Amazon Business. We could all use more time. Amazon Business offers smart business buying solutions, so you can spend more time growing your business and less time doing the admin. I can see why they call it smart. Learn more about smart business buying at amazonbusiness.com.

We just had a very hot summer. Umair Irfan, Box's climate reporter and resident ray of sunshine. 2024 broke all kinds of records. Much of the world saw the hottest May ever, the hottest June ever, and the hottest July ever. Which puts this year on course to being the hottest humans have ever measured. But don't worry, it's only going to get worse.

The evidence we have suggests that the planet is only going to continue getting hotter. Years like this year will soon become the normal and may eventually even be a relatively cool year. So it's looking fairly grim. The question is, how grim exactly? Like, how much hotter will the world get?

And how fast? These are questions that a lot of different climate scientists all around the world are racing to answer. The answers that they have come up with don't always agree with one another. But at least until recently, they had a pretty effective way to sort of come together and present the world with some best guesstimates.

These rough numbers that governments and architects and activists and, I don't know, insurance agents could use to make plans for the future. But now, these climate scientists are rethinking their approach. And things are getting a little bit heated. So, this is Unexplainable. I am Bird Pinkerton. And today on the show...

It is getting hot in here, but we're trying to figure out how hot it might get and how we decide which guesses we should take seriously.

The quest to figure out our future starts with a lot of scientists all around the world building climate models. Climate models are very complicated physical models of the Earth's climate, and they're run on some of the world's fastest supercomputers. Zeke Hausfather is a climate scientist who studies climate models and has also worked with the Intergovernmental Panel on Climate Change.

The IPCC, the big four-letter agency of saving the planet. They evaluate climate models, so Zeke knows a lot about them. Climate models, the way they broadly work is they take from the bottom of the ocean to the top of the atmosphere and break it up into little boxes. Each box might be 25 kilometers by 25 kilometers, and that may be half a kilometer high.

So it's a box that would encompass basically the greater New York City area or much of the important bits. And so you get, you know, thousands and thousands of these boxes all around the world. And the climate model simulates what happens in each of those boxes. To do this, they go back over 100 years and they take what we know about what's gone on in the atmosphere. So

how much methane and CO2 humans have put up there, sort of what volcanic events have added up there, what changes there have been and how much energy is reaching the Earth from the sun. And they pop all that data into their models. They share a lot of the underlying code with the models that produce your weather forecasts.

physical equations that themselves, you know, hopefully have relatively small uncertainty. But then scientists also have to make some assumptions here as well. These assumptions are where models really start to diverge. Because different scientists make different assumptions. Like,

Let's say they're looking at clouds. Some clouds will reflect heat back into space. Some clouds will trap heat on Earth. And both of those mechanisms are working at the same time. The question is, which one is more important and under what circumstances? And different scientists, for instance, will come to different conclusions about how important that is and how to weigh that in their models.

So how a scientist thinks about a cloud could affect whether her climate model predicts a livable future world or a dramatically terrible one. And clouds are just one example, right? There are lots and lots and lots of different variables that can be considered or dismissed or calculated differently by smart, reasonable people, right? There's no clear answer about whose approach is correct.

Which means that every climate model is the result of sort of lots and lots and lots of big and small decisions and predictions about how the Earth might react to a changing climate. The result of all this, of course, is that there's going to be a pretty wide spectrum of outcomes. Let's say we double the amount of carbon dioxide in the atmosphere from pre-industrial levels. There are some models that expect roughly two degrees of warming Celsius in that scenario.

There are others that say potentially there are feedback mechanisms that we might trigger that will cause more of a runaway heating event. And so maybe we're looking at 3, 4, 5, 6 degrees Celsius of warming. And there is a huge difference between 2 degrees of warming and 6 degrees of warming. We know that essentially 2 degrees of warming is something that we would probably be able to survive based on some of our historical records.

Maybe not necessarily well. We may not be all happy about it. But this still seems like something that perhaps we know how to manage. But once we get outside that range, we start dealing with temperature scenarios that human civilization has never seen before. Maybe we could survive it with technology and a lot more upgrades and making some drastic changes. But at that point, we don't know.

But whether we're headed for a somewhat predictable, albeit very unpleasant, two degrees of warming or like a wildly unpredictable six degrees of warming, policymakers need to know, right? Like architects and investors need to know. All of us need to know or at least have a sense of what might be ahead of us so that we can plan for our futures. And if all these models are making different predictions...

How do we figure out which models to go with? Which of these numbers to use? That's a thorny question that the community has wrestled over a lot and it's tied up a little bit in the politics of the climate modeling exercises that are done. Basically, every six or seven years, the IPCC issues a report. So again, the IPCC is like this UN panel of different countries coming together to figure out what to do about climate change.

And in this report that they put out, they look at models from all around the world. They look at the model's predictions, and then they have to decide on a kind of range of predictions that everyone can work with. And the way that they have done this surprised me. Historically, there hasn't really been a desire to say that, oh, we think that the CSM2 model that America produces is better than the model that Russia produces.

And so there's a bit of a challenge historically with model democracy, where they want to sort of include everything, no matter how good it is, throw it all in the pot, stir it around and see what the outcome is. The IPCC has actually kind of treated all the models as if they're the same. Everyone's a winner. All of them are folded into the average and given equal weight. And then from there, they take the output and they give us a range of how much warming we can expect. But I guess...

Aren't some models better than others in this equation? Like, should all models be treated equally? So some models do, in fact, have more sophistication, more computing power behind them, or are just better constructed. They do a better job of imitating or replicating history, which sends to be a good signal that they'll do a good job replicating the future. But there's sort of a diplomatic...

consideration here as well. You know, you want buy-in by all the different member countries. You want, you know, the Russians and the Chinese and the Italians, you know, everyone else to feel like their voices and their scientists are being heard and included in these processes. And so rather than pushing anybody away from the table, they tell everybody to contribute what they can and try to incorporate everyone's contribution into helping shape the conclusion. This might seem like a kind of a

wild way to do science, right? Like just weight everyone's models equally so they don't

take their ball and go home. But there's also a scientific reason to do it. Different models are also just good at different things. Maybe one model is really good at predicting warming in the future. It's optimized for that, but less good at predicting rainfall and snowfall. Another model might actually be really good at predicting precipitation, or a third model might be really good at predicting wind speed. When you mix them all together, they balance each other out, smooth out each other's weaknesses.

You know, there's a lot of interesting papers that look at how the average of all the models often outperforms any individual model across a bunch of different metrics. Basically, it's the wisdom of the crowds, right? Everyone's guesses, all blended together, are better than any one individual's guess. And the end result here was still a relatively broad range of potential temperatures for the Earth.

But it did narrow things down somewhat. It gave people confidence in the range of warming that they can expect. It doesn't really rule out completely any of the extreme scenarios where we see maybe zero warming or 10 degrees of warming or something like that. But it really helped people know what was more likely, what was plausible. It seemed like it was working. Until recently.

Before the latest roundup of climate models, the one that came out in 2023, Zeke says that this wisdom-of-the-crowds approach started to break down. And it broke down a bit because there was a subset of models, about 20% of them, that was running really hot.

Basically, there were a bunch of models that were predicting that if you doubled CO2 in the atmosphere, the Earth would warm by five degrees or more. Very high. So high that researchers like Zeke started calling them hot models. Maybe you're like, what's wrong with a hot model? Daira Banks likes hot models. We should celebrate hot models everywhere.

I think it sets an unreasonable beauty standard for other models. I think it's unfair to hold just one as a platonic ideal. All models are beautiful. Do you know that? And then you come in and you treat this like a joke? They can, I don't know. I've never watched this show. I'm out of cultural references here. Tiffany, I'm extremely disappointed in you. Okay, but actually, after the break, we're going to look at why these hot models are such a source of controversy. We were all right!

We will also talk to a co-author of one of the hottest models around. Your business deploys AI pilots everywhere. But are they going anywhere? Or are they stuck in silos, exhausting resources, unable to scale? Maybe you don't need hundreds of AI pilots. You need a holistic strategy.

IBM has 65,000 consultants with Gen AI expertise who can help you design, integrate, and optimize AI solutions. So you're not just deploying AI, you're scaling it across your business. Learn more at ibm.com slash consulting. IBM, let's create.

Support for this show comes from Amazon Business. We could all use more time. Amazon Business offers smart business buying solutions so you can spend more time growing your business and less time doing the admin. I can see why they call it smart. Learn more about smart business buying at amazonbusiness.com.

Thank you.

It's time to cut through the noise and make a real impact. So tune into the future of marketing, a special series from the PropG podcast sponsored by Canva. You can find it on the PropG feed wherever you get your podcasts. So you have these hot climate models. They're considered hot because most of them think the world will warm by more than five degrees if we double CO2.

And they get there in various different ways. Some models have different ways of calculating cloud cover in the years to come, and that makes their predictions a bit more dire in terms of warming. There was a paper that came out last year that looked at aerosols, tiny particles of stuff like bits of air pollution, sulfur, sea salt in the atmosphere. And that also led to somewhat hotter findings. But what really makes some of these hot models so hot is

is their potential implications. If these models are right, the world gets very hot very fast. That could be very consequential. It could be dire.

That means people now are thinking about some much more dramatic interventions. There was a model that came out last year that got a lot of controversy because one of the co-authors suggested we might need to consider something called solar geoengineering. Solar geoengineering is where you literally spew huge amounts of aerosol, stuff like sulfur dioxide or sulfuric acid, into the atmosphere to create a sunshade for the planet, sort of like an umbrella to help cool it all down.

This is somewhat frapp. There's a long list of reasons why geoengineering is controversial. You know, we're looking at potentially irreversible changes that would require constant long-term interventions indefinitely just to maintain a certain level of cooling for the planet. It would be a huge step to take, which makes some people tread with caution before sort of running with the results from hot models.

And then, before the latest IPCC report, some papers started to come out questioning how accurate some of these models actually are. Some of the researchers here, and Zeke is one of those scientists, think that these models don't do a good job of replicating the past, which is one of the key ways that we actually test our models. To test a climate model, researchers will say, OK, if we go back to 1850, say…

Would your model then do a good job of predicting what actually happened over the next century or so? The logic being that the better it lines up with what we've seen in the past, the better it will probably do at predicting the future. So some researchers tested how well these hot models could predict one important variable, past temperatures. And the hot models, they didn't seem to be doing as good of a job of predicting the past in this way.

So imagine that you are the IPCC, right? You are responsible for sort of mixing up this climate stew to get people these numbers that they can work with to make plans for the future. And suddenly you have a bunch of these hotter models that are raising concerns. Basically, Zeke says the researchers were worried that if they tossed all of these models into their stew,

The stew would get too strong of one flavor. If you throw too much cayenne pepper, it would just be too spicy and you'd lose all those subtle flavors of sage and bay leaf. You would have so many outliers, basically, with these climate models. Your predictions might skew too hot and that would have consequences for all the other things that you're building on these models. The IPCC didn't want to just throw all these hot models out or kind of ignore them completely.

But in the most recent report, they did decide to sort of rework their approach a couple ways. One really important thing that they changed was instead of just throwing every model into the pot and taking an average, they decided to weight each model based on how well they predicted the past temperatures. The models, hot or not, had to go through this past temperature test. And if the test thought a model did a good job, it got a little more sway, but

But if the test thought a model did a less good job, it had a little less sway. The end result was that the hot models were turned down and had less weight overall. So the range of likely temperatures for the Earth that we wound up with wasn't quite as warm as it would have been if you just averaged everything out equally.

On the one hand, I get why the IPCC changed the formula here, right? I understand that this is an attempt to get at as accurate a set of predictions as possible. But on the other hand, there's something really appealing about the idea of model democracy, right? The idea that if we can't be sure who's right, we just give everyone the same voice. And in the interest of hearing other voices...

Umair and I wanted to know how the authors of these hot models felt about all this, like how they felt about their results being somewhat downplayed. So we ended up reaching out to Neil Swart, who co-authored a very hot model in the latest IPCC cycle. We had a joke that we're the hottest model. So, you know, phrase it as a positive thing. In our conversation, Neil raised a couple of big questions about the IPCC's decision to weight models.

He did say that, like, yes, it makes some sense to measure how well a model predicts past temperatures. You could...

make like a decent argument that this is a robust approach. That being said... What happens if I'm not interested in temperature as my primary thing? Maybe I'm a person that's interested in precipitation. Again, climate models don't just try to predict temperatures. They are also trying to predict stuff like future rainfall. You can't just apply the temperature weighting to precipitation because there's no saying that models that did better on the temperature metric are going to do better on the precipitation thing.

That was, again, one of the reasons for model democracy. Like, models have different strengths and weaknesses, and if you throw them all into the pot, they kind of balance each other out. Neil says before all this went down, his group realized that their model was going to come out seeming pretty hot.

And they actually asked themselves. Should we do something to try and adjust this? And we made a very conscious decision, no. So for us, we did not try at all to optimize on this metric. We put together the best physics that we can into the model. And, you know, the result came out with this kind of climate sensitivity or this warming just as it did. You know, the chips fell where they may. That's what emerged. And we went with that. But for the next round of modeling and IPCC reports...

It could be different. If you do start doing this and kind of like, let's call it punishing models for being outside of some metric like this, are we going to do that again for the next round? No, we're not. We're going to say, okay, we can adjust this. We have the power to adjust this.

And so it affects philosophically maybe how you approach the problem is if that pressure is too strong and you constrain them too much so that everyone tries to tune their model, you could actually artificially reduce the uncertainty that you're getting. And that would be a problem for the community because then you're overconfident in your predictions. So if everyone's taking a similar approach, maybe they wind up with a similar result, not because that result is actually right, but

but because they were all asking the question in the same way. I think these outliers, another way to say it is that the outliers provide some value. They help us remember just how uncertain things really are. There's room for debate here. Like, debate about how the scientific community should figure out these numbers. What do you do when you have lots of experts using the best available data that they have, the best available research, the best available model, in good faith?

coming to slightly different conclusions or very different conclusions. How do you start making judgments and assessing those? This is a hugely consequential field of research because we're making, you know, predictions and forecasts of the future. And the only way we can validate the future is by getting there.

This episode was reported and produced by Umair Irfan and me, Bird Pinkerton. It was edited by Jorge Just. Meredith Hodnot runs the show. Noam Hassenfeld is our host and does the music. Christian Ayala did the mixing and the sound design. Anouk Douceau did our fact-checking. Manding Nguyen is the fact that many birds have hollow bones. And we are always, always grateful to Brian Resnick for co-founding the show.

Thanks so much to Norman Loeb, Loretta Mickley, and Eliza Jean Harris for their help and their time.

If you have questions about this episode or thoughts about this episode or thoughts about future episodes that we should do, please send them to us. We are at unexplainable at Vox.com. You can also support the show and all of Vox's journalism by joining our membership program today. It would be very much appreciated. Go to Vox.com slash members to sign up. If you can't do that, for whatever reason, you can also support the show by leaving us a nice rating or a review.

Those things mean quite a lot to us. Unexplainable is part of the Vox Media Podcast Network. And we will be back next week. Support for this show comes from Amazon Business. We could all use more time. Amazon Business offers smart business buying solutions so you can spend more time growing your business and less time doing the admin. I can see why they call it smart. Learn more about smart business buying at AmazonBusiness.com.

Canva presents a work love story like no other. Meet Productivity. She's all business. The Canva doc is done. Creativity is more of a free thinker. Whiteboard brainstorm. They're worlds apart, but sometimes opposites attract. Thanks to Canva.

The data is in the deck. And now it's an animated graph. Canva, where productivity meets creativity. Now showing on computer screens everywhere. Love your work at Canva.com.