Does High-Profile Climate Science Tell the Full Story?

If researchers were perfectly dispassionate reasoners with no motives other than seeking truth, their published papers could be taken as a direct, objective view into reality. But as I argued not long ago in an essay in The Free Press, that idealized notion of science is a fantasy. Frustrated about not being able to take high-impact climate science at face value, I decided to call out what I see as one problem: The highest-profile research is heavily influenced by cultural forces and career incentives that are not necessarily aligned with the dispassionate pursuit of truth.

That essay caused a bit of a firestorm, in part because of its main topic: climate science. Climate change is a morally and politically charged topic, and emotions run high. People are often placed into one of two camps: the virtuous good team working to save the planet, and the nefarious bad team delaying action out of some combination of ignorance and greed. In the world of climate research, there are substantial career incentives to claim membership in the good team. Specifically, to get your research published in high-impact journals like Science and Nature, it helps enormously not to challenge the predominant narrative that the benefits of limiting global warming to at most 1.5 degrees Celsius — as articulated in the Paris Agreement — far outweigh the costs.

A case in point, and the subject of my essay in The Free Press: my own recent article in Nature. That was my third publication in Nature. I have also published in Nature’s climate-focused journal, Nature Climate Change. And I have served as an expert peer reviewer for both journals as well as for Nature Communications and Nature Geoscience. Through those experiences, as well as through various failures to get published in those journals, I have learned that framing research in a way that tends to support the predominant narrative makes the path to a high-impact publication much less treacherous. For example, rather than ask “What is the magnitude of the influence of climate change on the phenomena I am studying relative to all other influences?” it is more prudent to ask “How does climate change negatively impact the phenomena I am studying?” In the case of my most recent Nature paper, for instance, I specifically chose to focus narrowly on the influence of climate warming on wildfire, although it is well known that warming is just one of many important causal factors.

Such framing decisions are not necessarily distorting if they are complemented by other researchers who make framing decisions in the opposite direction. But when social and career incentives cause most of the high-profile research to coalesce around a similar emphasis, a distortion emerges in the aggregate. When that happens, scientific narratives can become entrenched and self-reinforcing. And that’s where we are in climate science: A desire to be on the good team, associated with an adherence to the predominant narrative, diverts attention from the study of humanity’s resounding success in becoming ever more resilient to the climate, and it sidelines legitimate concerns about restricting energy options for humanity — especially in low-income countries.

At bottom, research is a social endeavor. If you do not communicate what you have done to your colleagues, your funders, and the public, you may as well not have done the work at all. And when it comes to communication, which journals you publish in makes a major difference. Publishing in high-impact venues undoubtedly helps the upward trajectory of a scientific career.

One reason for this is that high-impact publications receive a lot of attention. Of the 50 climate-change papers that received the most online attention over the past five years, 32 were from Nature, Science, and their subjournals. Eleven others were from the very prestigious Proceedings of the National Academy of Sciences and The Lancet. All told, 86 percent of the 50 papers that received the most attention appeared in exceptionally prestigious journals.

That is not a coincidence. High-impact journals have sophisticated public-relations apparatuses that disseminate press releases directly to major media outlets well in advance of a paper’s publication. Such publicity helps them justify their “prestige tax” (the multimillion-dollar annual subscription fees they charge universities and other institutions) as well as the high fees they charge directly to publishing researchers. For example, it currently costs $870 per figure to publish in Nature and $11,690 to remove the paywall from one’s paper. Despite those costs, the attention and prestige are worth it to many researchers.

My Nature paper was covered by more than 70 national and international outlets. The result is not only wide awareness but also engagement from other researchers, who are then more likely to cite the article. Citations are the main quantitative measure of a researcher’s prominence. My previous Nature paper, from 2017, has more than four times the citations of my next-most-cited paper.

The power of this citational imbalance is considerable. Early-career researchers with high-impact publications are six times more likely to be offered a faculty position. A journal’s name recognition and impact factor are highly valued for academic promotion and tenure. Hiring committees will inevitably use journal prestige as a shortcut in evaluation because it is very difficult to assess the quality of dozens of candidates’ CVs across a whole range of subdisciplines.

Competition is accordingly fierce, which allows Nature and Science to be highly selective, rejecting more than 90 percent of submitted papers. They, and the reviewers they call upon, act as gatekeepers for high-profile research and have an inordinate influence on shaping predominant scientific narratives.

The competition means that many researchers will shape their research questions from the outset to maximize the chances of publication. One straightforward strategy for doing this is to make sure the research question is in line with the high-level messaging espoused by the journal’s leadership. The leaders at Nature and Science have made it clear that they endorse the political goals of the Paris Agreement — to rapidly transition the world’s energy and agricultural economies so that global warming remains below 1.5 degrees Celsius (or at most 2 degrees Celsius) above preindustrial levels.

For example, Marcia McNutt, when she was editor in chief of Science, implied in an editorial titled “The Beyond-Two-Degree Inferno” that humanity has sinned against nature, and society should sacrifice economic well-being for the sake of remaining below 2 degrees Celsius of warming. Similarly, Magdalena Skipper, the editor in chief of Nature, is an outspoken supporter of the Paris Agreement and its temperature limits.

Nature as an institution officially endorsed Joe Biden in the 2020 U.S. presidential election, citing, among other reasons, his policies in support of the Paris Agreement. Facing some pushback on its explicit embrace of politics, Nature subsequently doubled down on its political statements. The current editor in chief of Science, Holden Thorpe, has defended the idea of scientific journals’ endorsing policies and politicians — implying that the authority of science subsumes the entirety of the climate problem, up to and including the amount of power that the government should yield in dictating a solution. This all sends a clear signal that Science and Nature will be more sympathetic to research that supports those political goals than to research that undermines them.

Defenders of such political statements might argue that they are simply science-based policy. But it is a widespread misconception that the 1.5-degree and 2-degree Celsius global-warming limits were first established by climate science and only subsequently incorporated into policies like the Paris Agreement.

In fact, the 2-degree Celsius limit was first proposed in the late 1970s, long before sophisticated studies of climate impacts had emerged. Its appeal was, essentially, that it was a round number thought to represent the warmest the earth had been in 100,000 years. By 1992, it had become conventional wisdom that humanity should aim to avoid anything more than 2 degrees Celsius of warming; it was officially codified as “dangerous interference with the climate system” by the United Nations in the 2009 Copenhagen Accord. Six years later, the U.N. Paris Agreement affirmed the 2-degree Celsius goal while also articulating aspirations for limiting global warming to 1.5 degrees Celsius, in part due to pressure from environmentalists and some diplomats.

Only after the articulation of the 1.5-degree Celsius limit did the United Nations solicit a report from its own Intergovernmental Panel on Climate Change, or IPCC, to support it. Despite widespread misrepresentation in the media, the report did not say that global warming becomes catastrophic beyond 1.5 degrees Celsius, and it certainly did not claim that limiting global warming to 1.5 degrees Celsius represented the ideal balancing of all costs and benefits for the entire world. It merely said that the climate-change effects at 1.5 degrees Celsius are less severe than those at 2 degrees Celsius.

In other words, those global-warming limits emerged from the collective judgment of elite policy actors and should not be construed as some sort of definitive scientific conclusion. In fact, there is no way for science to definitively specify the optimal rate of decarbonization by objectively weighing all the costs and benefits across different individuals, societies, and species over space and time. Rather than being purely scientific, such temperature limits are fundamentally influenced by a normative premise of environmental philosophy: that changing the earth is inherently wrong and inevitably self-destructive. Activist researchers know, however, that an argument based on subjective philosophical considerations is weaker than one based on the perceived authority of science. So many engage in “stealth advocacy” — advocacy based on normative commitments but disguised as the outcome of a purely dispassionate scientific process.

This is the cultural environment within which climate-impact research matured: one in which the policy goals had already been established. Thus it has always been easier to conduct research that supports those goals than to conduct research that undermines them. The consequence is that in addition to there being science-based policy, there is also plenty of policy-based science. That became clear to me in 2018, when a paper published in Nature, “Large Potential Reduction in Economic Damages Under U.N. Mitigation Targets,” received credulous press coverage, inspiring headlines like “Limiting Temperature Increase to 1.5 Celsius Could Result in $30 Trillion of Savings for Global Economy, Study Shows.” The study, in fact, did not show that. Despite what was implied by the title and the headlines, the study did not offer a cost-benefit analysis of the Paris Agreement’s limits but rather a benefit-only analysis, ignoring the well-known costs of forcibly decarbonizing the global economy.

Nature has published articles that analyze the costs of decarbonization in isolation, but those studies are framed very differently. The costs are framed as an obstacle that can be overcome. A fair literature that was disinterested with respect to results would allow the equivalent cost-only analysis framed in the same way to be published: a paper with the title “Large Potential Damage to World Economy Under U.N. Mitigation Targets.” But I consider it vanishingly unlikely that a title like that could see the light of day, not because it is factually wrong but because it’s too transgressive of the predominant narrative.

While I knew that such a hypothetical cost-only analysis would not be publishable, I thought perhaps Nature would be interested in publishing a more complete cost-benefit analysis. So, later in 2018, I submitted a paper to Nature, titled “Net Economic Impact of U.N. Global-Warming Mitigation Targets,” that placed the calculated benefits described by the aforementioned paper in the context of mainstream estimates of the costs of decarbonization. The study showed that when costs were considered alongside benefits, the conclusion of the benefit-only analysis was overturned: The Paris Agreement targets would impose net harm on the world economy through 2100.

The paper did not make it to the peer-review stage. It was desk-rejected for being, supposedly, of insufficient interest and an insufficient advance (a subsequent version was later published in PLOS ONE). I find those reasons for the desk rejection unconvincing. A cost-benefit analysis that overturns the conclusion of a previous benefit-only analysis of the predominant global climate policy should be among the most interesting and consequential results in climate research. A much more likely explanation: It was the finding of the study, rather than the topic, that was unwelcome.

Another experience that informed my view of those journals’ preferences was my discovery of a major flaw in the climate-impacts literature that could be traced back to a commentary paper that was published in Nature. The flaw leads to substantial exaggerations of the influence of climate change on extreme-weather effects. Yet this methodology has proliferated; it has been incorporated into major IPCC reports, and it undergirds some of the most attention-grabbing headlines assigning deaths and dollars to climate change. Since all this originated in the commentary pages of Nature, I pitched its commentary editor (with a referral from another Nature editor) on publishing my work highlighting the flaw in the original paper. Unlike my successful submissions to Nature, which were in line with the predominant narrative, this time I received no interest. Eventually, my work was published in the lower-impact journal Climatic Change, where it will undoubtedly have less influence.

Even if studies don’t frame their research question in a way that directly lends support to the Paris Agreement, they will often implicitly support the policy goal by taking neutral or good news and presenting it as bad news. An example of this is found in a 2021 study called “Climate Impacts on Global Agriculture Emerge Earlier in New Generation of Climate and Crop Models,” published in Nature Food. The title signals bad news, so one might be surprised to learn that the study showed that, in the most up-to-date crop models, projected warming and enhanced carbon-dioxide levels would increase global wheat, rice, and perhaps soybean yields while decreasing only corn yields. The title and abstract could have been framed to highlight those results, but instead the concluding sentence of the abstract emphasizes some of the worst news from the study: “… these results suggest that major breadbasket regions will face distinct anthropogenic climatic risks sooner than previously anticipated.”

Some version of the inquiry How does climate change negatively impact the phenomena I am studying? recurs throughout the high-impact literature. It results in studies that look only at increases in heat deaths while neglecting decreases in cold deaths or that highlight the social cost of carbon but not the social benefit of the technologies that produce it.

Such examples are representative of the major trends that emerge in the aggregate in the top 50 climate-change articles published in Nature and Science over the past five years, as ranked by Google Scholar. “The Human Imperative of Stabilizing Global Climate Change at 1.5°C,” the article ranked first over all, is emblematic. It is policy-based science in the sense that it is explicitly framed to make as strong a case as possible for adhering to the 1.5-degree Celsius limit under the Paris Agreement. Other articles take a similar tone of advocacy. An especially strident example is “Harnessing the Potential of Nature-Based Solutions for Mitigating and Adapting to Climate Change,” which concludes thus:

Achieving net-zero carbon emissions and transitioning to a nature-positive economy will also require systemic change in the way we behave as societies, shifting to a dominant worldview that is based on valuing quality of life and human well-being rather than material wealth — and connection with nature rather than its conquest. Signals such as the rise of climate and nature grassroots activism indicate that this shift is taking place.

The irony of all this, of course, is that when research becomes overtly normative, it cannot but sacrifice objectivity — which then undermines the very scientific authority it is attempting to leverage.

Most articles in the top 50 database don’t assume the same tone of advocacy as the above examples, but the vast majority support the predominant narrative. In my estimation, almost three-fourths either explicitly or implicitly encourage more rapid cuts in greenhouse-gas emissions — typically by focusing narrowly on a negative impact of climate change. Over half use pessimistic or monitory messaging rather than taking a more neutral tone, and over half use an ominous or headline-style title. Vanishingly few adopt an optimistic tone, embracing the ubiquitous increases in human resilience to climate that we’ve seen historically. And none focus on the risks of overly restrictive energy policies. Those patterns send very clear signals to researchers who aspire to publish in these journals.

There are, of course, exceptions. Sometimes, Science and Nature do publish papers that go explicitly against the predominant narrative. But those are exceptions that prove the rule: Stepping out of line can invite more trouble than it is worth.

Consider the paper “The Global Tree Restoration Potential,” published in Science. It received the fourth-most attention in the past five years, according to the online-attention catalog, but is not in the top 50 of Google Scholar’s most relevant list (which is based to a large degree on citations) because most of the attention it got was negative. Why? It had the temerity to emphasize planting trees over reducing greenhouse-gas emissions as a primary strategy for climate-change mitigation. That was seen by wide swaths of the environmental community as setting back crucial climate action. The backlash was intense.

Science responded to the controversy by allowing three technical comments — mini-rebuttals from other researchers — to be appended to the paper. The original paper’s claim that “tree restoration is the most effective solution to climate change to date” was amended with a statement that deemed the original paper to be “incorrect,” even though “most effective” includes subjective considerations on how difficult a solution would be to carry out and is thus not able to be deemed strictly “incorrect.” In a sign of the pressure they faced, the authors officially added that they “did not mean that tree restoration is more important than reducing greenhouse-gas emissions or should replace it.”

When it comes to assessing the net effect of current energy and agricultural systems, the inordinate focus on the negative has sidelined the fact that almost all climate-sensitive aspects of global human society have been trending in a positive direction over the past 50 years despite about 1.3 degrees Celsius of warming: Crop yields and calories available per person have increased; death rates from malnutrition and famines have decreased; the share of the population with access to safe drinking water has increased; the rates of climate-influenced diseases like malaria and diarrhea have decreased; death rates from natural disasters have decreased; death rates from nonoptimal temperatures (hot and cold) have decreased; and the fraction of people in extreme poverty has plummeted.

Neglecting to emphasize those positive trends results in a dearth of high-profile studies of the reasons for success — making it more difficult to foster more of the same. To adhere to the Paris Agreement limit of 1.5 degrees Celsius, the sources of energy that have been the most effective historically and are arguably still the most effective today must be rapidly eliminated. Specifically, climate concerns threaten to restrict energy growth in low-income countries via policies like bans on international finance for any fossil-fuel infrastructure. That is counterproductive because, in addition to the obvious benefits of energy-poverty alleviation in all aspects of life, energy use is associated with increased resilience to climate and climate change: The average mortality from extreme-weather disasters, as well as economic damage as a fraction of GDP, is much lower for high-income, high-energy-use societies than for low-income, low-energy-use societies. To put the extremity of the 1.5-degree Celsius limit in perspective, coal-power plants must be phased out at an average rate of about 240 plants per year every year between now and 2030 to be on pace for the target. Over all, we would need to see a rate of cuts in greenhouse-gas emissions equivalent to what was experienced due to the Covid lockdowns and associated economic recession (about 5 percent), but compounded year after year until emissions go negative in the latter part of the century. The costs, in monetary terms, are calculated to be something close to 5 to 10 percent of global GDP (or $5 trillion to $10 trillion) annually over the next several decades.

Climate change is a major concern, and it won’t stop being one until global human-caused carbon-dioxide emissions reach net zero. We are very far away from that. But adhering to the 1.5-degree Celsius limit entails the full-scale rapid reorganization of the world’s energy and agricultural economies, which comes with major risks, and thus it should not be sold under false pretenses.

But the social forces are what they are, and in the current environment, savvy researchers have a clear sense that it is prudent to present their research in a way that keeps them on the good team. After my Nature paper was published but before I published my critique of the scientific-publishing landscape, I was greeted with fawning positive attention. I was showered with congratulations by my colleagues, and I did more than a dozen friendly interviews with various media outlets. I was even sent a solicitation to have my paper printed on a coffee mug as a kind of trophy.

That all changed after I published my essay critiquing high-profile journals as overly focused on negative climate impacts and emissions restrictions. Since my message undermined the idea that science journals are pristine arbiters of truth, unaffected by human social influences, it was threatening to many public-facing climate scientists and climate journalists, who exploit the perceived authority of science to disguise their normative advocacy on climate policy. The blowback on blogs and especially on Twitter was fierce, and often ad hominem.

The juxtaposition between celebratory coffee mugs and personal attacks is telling. Researchers are socially rewarded for supporting the predominant narrative and ostracized for undermining it. In particular, the negative reaction to my critique made it clear to onlooking colleagues that being perceived as on the bad team will invite excommunication from large swaths of the professional climate-science community. I received many messages of emphatic support, but those communications were mostly private, either via email or in person. My sympathetic colleagues knew that voicing public support might only invite waves of criticism their way.

That dichotomy of public condemnation but private support is evidence of self-censorship in academe. Surveys have indicated that 34 percent of professors report having been pressured by peers to avoid controversial research, and 91 percent report being at least “somewhat likely” to self-censor in academic publications, meetings, or presentations, or on social media. Those numbers represent a major problem if we want academic scientists to be oriented as much as possible toward truth-seeking. Such self-censorship results in an outward appearance of consensus that is representative neither of the underlying empirical evidence nor even of the opinions of the people producing the research.

When a predominant scientific narrative emerges organically from truly open scientific debate, the public can trust it. However, when a narrative is heavily influenced by morally infused social-feedback loops within the knowledge-production system, it is much less likely to convey the full story.

Breaking that social-feedback loop will probably require top-down structural changes at journals as well as bottom-up cultural changes.

A more honest and holistic picture of the state of the climate-change problem, for instance, would emerge if high-profile journals installed sturdier guardrails against self-reinforcing research themes. That would fall under the umbrella of more “institutionalized disconfirmation” of the predominant narrative. One strategy would be to accept papers based on research questions and proposed methodology before the results were known. That would induce more neutral research questions and reduce the impulse to adjust research questions and methods to increase the likelihood of a high-profile publication. It would also combat the “file-drawer effect” (whereby only the flashiest results survive to publication and subsequent promotion in the media). Other reforms could include the separation of the research groups that design studies from those that conduct studies, or the commissioning of multiple research groups to investigate the same question. Journals could also reimagine the peer-reviewer system and publish a full “devil’s advocate” rebuttal alongside each paper they publish.

At least two suggestions in a recent perspective on scientific censorship deserve serious consideration in this context. One is the publishing of peer reviews and editorial-decision letters for rejected manuscripts. It is becoming more common for journals to publish peer reviews for accepted manuscripts, but transparency about rejections would increase accountability, encourage fairness, and allow for biases in publishing to be studied more openly. Toward similar ends, journals should embrace audits of their publishing practices, allowing auditors to test whether methodologies are scrutinized differently depending on the results. Ideally, the main measure of journal prestige could move from measures of “impact” (citation frequency) toward measures of scientific neutrality and trustworthiness.

When dealing with climate science, journal editors must be more open to papers that transgress the predominant narrative in support of the Paris Agreement. They have commissioned special issues and penned op-eds in support of the narrative, so why not op-eds and special issues that explicitly welcome research on successful increases in resilience to climate (in order to foster more such resilience) and the risks of overly restrictive energy policy? Signals like that from journal leaders would loosen the grip of the good-team, bad-team dynamic and foster a more honest and complete scientific literature, one that would ultimately be more useful for society.

Journalists who cover climate must also understand that researchers face motivations beyond objective truth-seeking. Thus, they should see it as their role to be a filter between the scientists and the public. All too often, instead, they serve as uncritical mouthpieces for scientists.

It has become common to lament the public crisis of trust in science, experts, and institutions, but much of the decline in trust is deserved. When the public senses that researchers and academic journals are attempting to leverage their scientific authority to advance particular political goals, it ironically undermines that very authority. But we need researchers and academic journals to help inform (but not dictate) critical societal decisions. Scientific institutions must earn the public’s trust by demonstrating intellectual humility, a disinterested neutrality about results, and an openness to a wide diversity of opinion — and to vigorous debate — on contested topics.