Science Matters

Tom Chivers writes with insight as Unherd’s science editor Can we trust the climate scientists?  Excerpts in italics with my bolds.

The reaction to Steven Koonin’s book highlights just how toxic this debate has become

There’s a problem with writing about science — any science — which is that scientists are human like the rest of us. They are not perfect disembodied truth-seeking agents but ordinary, flawed humans navigating social, professional and economic incentive structures.

Most notably, scientists, like people, are social. If they exist in a social or professional circle that believes X, it is hard to say not-X; if they have professed to believe Y, they won’t want to look silly and admit not-Y. It might even be hard to get research funded or published if it isn’t in line with what the wider group believes.

All this makes it very hard, as an outsider, to assess some scientific claims. You can ask some expert, but they will be an expert within the social and professional milieu that you’re looking at, and who will likely share the crony beliefs of that social and professional milieu. All of which often makes it hard to disentangle why scientists do and say the things they do. Especially when it comes to scientific claims that are politically charged, claims on hot-button topics like race, sex, poverty — and of course climate.

I couldn’t help thinking about that as I was reading Steven Koonin’s new book, Unsettled. Koonin is (as it says, prominently, on the front of the book) the “former Undersecretary for Science, US Department of Energy, under the Obama administration”. The publishers are obviously very keen to stress the Obama link: “…under the Trump administration” might not have carried the same heft.

Koonin came to public attention a few years ago, after he wrote a controversial opinion piece for the Wall Street Journal headlined “Climate science is not settled”. It was a response to what he considered the widely held opinion among policymakers and the wider public that, in fact, climate science is settled. His particular concern was that we can’t yet accurately predict what the future climate shifts will be. The book itself is best thought of as the extended version of that op-ed, with added graphs.

Climate Science is not Settled:  We can break down his thesis into, roughly, three areas.

One, is that despite “the mainstream narrative among the media and policymakers”, it is hard to be sure that the climate has changed in meaningful ways due to human influence. In particular, floods, rainfall, droughts, storms, and record high temperatures have not become more common, and although the climate is unambiguously warming and sea levels have gone up, it’s hard to confidently separate human influence from natural variability.

Two, he says, climate models are highly uncertain and struggle to successfully predict the past, let alone the future, so we shouldn’t trust confident claims about the climate future. And if we do accept the IPCC’s predictions, they aren’t of imminent catastrophe. Instead, they point to slow change to which humanity can easily adapt, and, broadly speaking, to humanity continuing to prosper.

And three, he continues, there is basically nothing we can do about it anyway, partly because carbon dioxide hangs around in the atmosphere for so long, but mainly because the developing world is developing fast, and using ever more carbon to do so, and actually that’s a good thing.

These are — according to Koonin — all, by and large, only what the IPCC assessment reports and other major climate analyses say.

The public conversation, which he says is full of doom and apocalypse and unwarranted certainty, has become unconnected from the state of the actual science. And he blames scientists — and policymakers, the media and the public — for that disconnection.

So is he right? Certainly he has a case when it comes to Point One: I think he is correct that the media narrative about climate change is not especially well correlated with the IPCC’s own central assessments. For instance, I think it’s fair to say that the recent floods in London, China and Germany have been held up as examples of a changing climate. But the IPCC’s most recent assessment report, 2014’s AR5, found studies showing evidence for “upward, downward or no trend in the magnitude of floods” (see p214 of the AR5 Physical Science Basis document; be warned it’s a big PDF), and concluded that they were unable to be sure whether, globally, river floods had become more or less likely.

Similarly, I think there is a perception among many commentators and policymakers that storms, hurricanes, and droughts are all more common as a result of climate change, but the IPCC’s own report (see p.53 of AR5) has “low confidence” that those things are more common than they were 100 years ago. I know some scientists think the IPCC is overoptimistic, but it is the closest we have to consensus climate science.

That said, there is some fairness in accusing Koonin of cherrypicking. He spends a lot of time arguing about extreme daily temperatures, convincingly (to my mind) debunking a claim in the 2017 Climate Science Special Report (CSSR), the flagship US government climate science assessment, that US extreme daily temperature records have gone up. In fact, CSSR is comparing the ratio of extreme high temperatures to extreme low temperatures, and what in fact has happened is that extreme low temperatures have become less common. Which is interesting.

But the IPCC does think extreme daily temperatures have gone up globally (see p53 again). In his chapter on “Hyping the Heat”, Koonin doesn’t mention the IPCC, and the IPCC outranks the CSSR. His detective work is interesting, but he is fighting a henchman, not the end-of-level boss. Maybe the IPCC is wrong as well, but we don’t learn that here.

On Point Two, I don’t feel competent to assess the models; certainly it seems highly plausible to me that there are enormous uncertainties in predicting something as inherently chaotic as the climate, especially when to do so you first have to predict something as inherently chaotic as people. But my non-expert understanding is that broadly speaking the models have been getting it about right.

That said, I think he is right that, if you were to ask the average person in my social circle, you would hear that climate change will lead to catastrophe in the near future. And I think that is overstating what the IPCC reports actually say. For instance, it is true that the IPCC predicts more people will go hungry than otherwise would have: it says that almost 140 million children will be undernourished, in a world where climate change goes unmitigated, compared to 113 million in a world where there is no climate change (see p730 of this IPCC report). But that is still fewer than went hungry in 2000 – almost 150 million, out of a much smaller population. The IPCC predicts that a world with climate change will be worse than one without; but not so much worse that other things, such as economic growth and technological progress, won’t broadly keep the big things, like life expectancy and human health, improving. That does seem worth saying.

And Koonin’s Point Three is worth making too. If India were to increase its per capita emissions to those of Japan, “one of the lowest emitting of the developed countries”, he says, then that change alone would raise global emissions by 25%1. Realistically, we’re not going to be able to stop India — or China, or Brazil, or Mexico, or any of the other middle-income countries — from developing, and development at the moment means carbon.

More importantly: we don’t want them to stop developing. Richer countries have healthier, longer-lived citizens and are better able to cope with a changing climate. Even huge, swingeing cuts to Western emissions — politically unrealistic — would only go some way to offsetting the inevitable growth in the developing world. Those cuts may be worth doing, but there are limits to how much good they can do.

But even if Koonin is right about almost everything — if the best guess of the science is that we’re heading towards things merely getting better more slowly, rather than getting worse — then I think he’s missing a major point. That is, climate change models are uncertain. In fact Koonin claims they’re even more uncertain than we think. So they could easily be erring on the side of optimism.

And the one thing we should have learnt from the Covid pandemic is that it’s not enough to say “the most likely outcome is that it’ll be fine, so let’s act as if it’ll be fine.” The correct thing to say is “the most likely outcome is that it’ll be fine, but if there’s a 10% chance that it’ll be completely awful, then we need to prepare for that 10% chance.” Reducing greenhouse gas emissions in the developed world reduces the chance of some unforeseen but plausible disaster: as a happy bonus, it makes our cities more pleasant places in which to live. It will come at some cost, but hopefully not too high, because green technology is getting so cheap and effective these days.

Reviews by climate scientists have been unimpressed. “I would normally ignore a book by a non-climate scientist,” starts one review, which goes on to not ignore it. Another accuses him of cherry-picking his fights (not entirely unfairly, as I said). A third says the book is “distracting, irrelevant, misguided, misleading and unqualified”.

But none that I’ve read really addresses the nitty-gritty of his arguments — which is hard to do in a 900-word review, of course, but still. They usually pick some line out of the first chapter or two, disagree with it, and then say the whole book is therefore rubbish. But I wanted a bit more meat to the objections.

The third review, for instance, quotes Koonin as saying “The warmest temperatures in the US have not risen in the past fifty years,” and then asks “According to what measure?” Well, Koonin tells you the measure, at length: absolute record extreme daily temperatures. Maybe he’s wrong, but he does answer that in the book. (And your next sentence is “Highest annual global averages?” He’s talking about the US! You just quoted that bit!)

Similarly, it complains that Koonin says that the sea is only rising about a foot per century, saying “The trouble is that while seas have risen eight to nine inches since 1880, more than 30 percent of that increase has occurred during the last two decades.” But again: Koonin addresses this, for pretty much an entire chapter. His point is that most of the rest of the rise came during an (unexplained by climate models, according to him) period of rapid warming from 1910 to 1940, before human influence should have been relevant. That, he says, is good evidence that natural variation is driving the current acceleration. Is he right? I don’t know. But the reviewer is not attacking Koonin’s argument at its strongest point.

In fact, none of them seem to: they just want to dismiss the book. They attack Koonin’s credibility and credentials, his temperament. They say he was only hired by the Obama Energy Department because of his contrarian views; they call him a “climate denier”, which seems de trop since he accepts most of the central claims of the climate consensus. The response felt more like a circling of the wagons than a serious effort to counter a serious argument. After all, it is unpleasant to hear reasons why you might be wrong about something: cognitive dissonance is painful.

I started this book confident that climate change is a serious concern, and I finished it only slightly less confident; Koonin has not persuaded me. But I’m glad Unsettled, flawed though it is, has been written. As I said at the beginning, science in a politically charged environment is very hard to assess. Scientists are as prone to groupthink and motivated reasoning as anyone else, and I know very well that there are some who feel they need to keep heterodox views quiet. The reviews, which make so little effort to engage with the substance of the arguments, do not reassure me that climate science is a uniquely groupthink-free discipline.

One thing Koonin suggests is a so-called “Red Teaming” of climate scientists: getting scientists to act as adversarial critics of the existing consensus, a method used by superforecasters, among others, to improve their accuracy by actively hunting out flaws in their reasoning. Science can only progress if assumptions are tested. Red teams in climate institutions — any institutions — seem like a good idea, and I’d support them.

Whether it’s possible or not, of course, is tricky to say. The climate debate is so highly charged, so borderline toxic, that it might be difficult for any climate scientist to take on the red-team role without making their own life more difficult. According to Koonin, one senior climate scientist told him “I agree with pretty much everything you wrote, but I don’t dare say that in public.” The old “in my emails, everyone agrees with me” line is hardly a new one, but it wouldn’t surprise me if there’s a bit of truth in it.

But if the Catholic Church was able to stomach someone advocating for the Devil, then climate science should be able to stomach one doing it for the sceptics. And in the meantime, this book does an acceptable job.

Footnote:  I encourage you to read the comments at the Unherd website.  For example this from Norman Powers:

I feel that’s a bit unfair to Tom. Reading the arguments of people who disagree, thinking about them carefully and weighing up debate is a large part of critical thinking, and he does all those things in this article. Just because he hasn’t arrived at the same conclusions as you (yet?) doesn’t make it not critical thinking.

Moreover, the social factors that apply to scientists apply to journalists as well. Do you think it’s easy for Chivers to talk about climatology and criticize climatologists so directly? He basically says none of their reviews of the book even meet basic standards of coherence, let alone being convincing. There aren’t many media outlets that would pay for such journalism, regardless of the truth.

Beyond your criticism of the author I feel all your points are well made.

For me, what really shook my belief in climatology to the core was the discovery that the temperature record itself is the output of modelling. Yes, you read that right. Not merely predictions about the future or truly ancient temperatures come out of models. Temperatures recorded by thermometer, in Europe and the USA, in the past 100 years or even just the past decade, are also the output of models. Although the raw data is given as an input the models proceed to heavily modify it; the outputs are then presented as “the history of temperature” without making it obvious what’s happened.

One of the consequences of this is that temperature time series often have multiple “versions”, reflecting the fact that the model software evolves over time. These new versions invariably seem to create warming when the prior versions didn’t show it. This has been going on for decades. They have a variety of justifications, all of which sound plausible on first glance, some of which seem less plausible on deeper analysis.
But. At school I was taught in no uncertain terms that in science you are not allowed to edit your raw data. All the marks for science assignments were allocated to the methodological correctness, and if you did an experiment and the data didn’t line up with the theory but what you did followed the rules, you wouldn’t be marked down (of course, in practice, if you failed to replicate a simple and famous experiment you probably did make a mistake somewhere so the distinction rarely mattered).

This was their way of teaching us that the rules are there for a reason, and that scientists aren’t allowed to tamper with their data post-facto. That’s taboo. Except, not in climatology. The risks are obvious: climatologists only really have one theory, so data that shows temperatures not going up undermines the entire community. Once the Rubicon has been crossed and model outputs are being substituted for real data, it’s very easy for people to try lots of different ways to “fix” errors in the data and then select only the ones that line up with what everyone knows “should” be happening. Over time this process keeps repeating until the theories become unfalsifiable.