Debunked? A review of Steven Koonin’s book ‘Unsettled?’

October 4, 2021

Note added on October 5, 2021: In an announcement today very relevant to this post, the 2021 Nobel Prize in Physics was awarded in part to Syukuro Manabe and Klaus Hasselmann, for laying the physics foundations “for the physical modelling of Earth’s climate, quantifying variability and reliably predicting global warming.”

Note added on October 22, 2021: Steve Koonin has written a response to our review below. We now append our discussion of Koonin’s response to the end of this post.

I. general comments

The two of us who write the posts on this blog site have both been involved in basic research in nuclear physics for many decades. We are both well acquainted with the important contributions to that field that Steve Koonin has made both before and during his long tenure as a nuclear theorist on the faculty at CalTech. We expect that he is also well aware of our own contributions. We respect Koonin as a fine scientist. We have been asked by colleagues to review Koonin’s new book entitled Unsettled? What Climate Science Tells Us, What It Doesn’t, and Why It Matters (see Fig. 1), in advance of his upcoming (October 19, 2021) Presidential Lecture at Purdue University. The lecture is one among a series of talks and interviews Koonin is currently giving to publicize his book, in which he lays out the reasons for his skepticism regarding the causes, impacts and mitigations of ongoing climate change.

Figure 1. The cover page of Koonin’s book.

To begin with, we must say that the large question mark on the cover of Koonin’s book is a ruse, intended perhaps to convey that each reader can decide for himself/herself, on the basis of what Koonin believes is his balanced and fair-minded presentation, whether the science of human-caused climate change is, in fact, so unsettled as to delay any government choices about mitigation policy. Koonin himself is hardly agnostic on this issue, as summarized in statements in the opening chapter:

The deficiencies of climate data challenge our ability to untangle the response to human influences from poorly understood natural changes…the science is insufficient to make useful projections about how the climate will change over the coming decades, much less what effect our actions will have on it.”

Koonin’s descriptions of the basic science underlying climate change discussions are generally good, but the presentation is hardly as balanced as he would have a reader believe. As we will discuss further below, he omits some crucial observations and misrepresents others in support of his view.

Koonin is especially concerned about public misperceptions of the science that he feels are driven by overhyping by a few scientists, many journalists, and quite a few politicians. In evaluating these concerns, it is useful to take note that scientists, journalists and policy-makers view their roles as distinct. It is certainly true that some journalists, in the search for attention-grabbing headlines, do overhype indications from the science. But there is also another side to the interactions between scientists and journalists. Indiana University recently hosted a question-and-answer session with the Pulitzer Prize-winning environmental journalist Elizabeth Kolbert.  In answering a question concerning her interactions with scientists, Kolbert indicated that the scientists are often so careful to emphasize uncertainties and caveats that the central truths uncovered by their research are sometimes obscured. She viewed her role as penetrating the jargon and the caveats to report the central truths in language most laypersons could understand. Koonin does a good job of avoiding jargon, or at least explaining the jargon he does introduce, but on the issue of obscuring central truths through an excessive focus on uncertainties and caveats, he is Exhibit A for Kolbert’s description.

Policy-makers – that is, those politicians who actually care about policy rather than just about being re-elected – have a harder job than scientists. They often have to construct policy on the basis of incomplete data and models – whether that data comes from science or economics or espionage — and without the aid of established theories. In the case of potential crises, policy-making usually comes down to an assessment of worst-case scenarios: what are the worst-case consequences of a contemplated action? What are the worst-case consequences of inaction? Are there actions that mitigate the worst-case consequences of doing nothing, while creating more palatable worst-case consequences than the action we’re currently contemplating? In Koonin’s mind the broad range of possible future climate trajectories argues for policy caution, but if he is wrong – as we consider to be likely – the worst-case scenarios may be catastrophic.

In some of our previous posts we have noted that conspiracy theorists often connect random dots. Koonin tends toward the opposite extreme. Like most scientists, he is skeptical of connecting dots in the absence of a firmly established theory that tells you why those dots should be connected. On the other hand, the development of such theories or models is often stimulated in the first place by recognition of patterns or correlations among data. But rather than point out noteworthy correlations among different aspects of climate data and impacts, Koonin takes a “divide-and-conquer” approach, treating each such observation independently to advance his claim that there is nothing unusual in the behavior compared with previous historical episodes.

In addition to the phenomena reviewed by Koonin, there are several other developments that he refers to only in passing, if at all.  These include:

Had he analyzed these occurrences, Koonin would likely have treated them also as separate events with historical precursors.  However, he is neglecting the crucial element that all of these features and many more are occurring simultaneously now.

In the remaining sections, we point out what we consider to be specific flaws in Koonin’s treatment of basic questions regarding human-caused climate change. We rely on some of the material we have used and referenced in our previous posts on climate change, as well as on a detailed critique that selected climate scientists have provided about many of Koonin’s claims. Our judgments are intended to respond constructively to Koonin’s invitation: “this book issues a challenge and solicits, indeed welcomes, informed argument and disagreement.”

II. is ongoing global warming primarily caused by human activities?

Koonin acknowledges that human activities have contributed something to ongoing warming of the planet, but he considers it uncertain whether that contribution is the major source of the warming. In particular, he notes:

Past variations of surface temperature and ocean heat content do not at all disprove that the ~1°C (1.8°F) rise in the global average surface temperature anomaly since 1880 is due to humans, but they do show that there are powerful natural forces driving the climate as well, and they illuminate the scientific challenge of understanding those natural influences well enough to confidently identify the climate’s response to human ones.”

By indicating that such understanding relies on very complicated and unsettled global climate models, Koonin overcomplicates the attribution to human activities. In our view, that attribution rests on three simple model-independent observations, only two of which Koonin acknowledges:

  1. Long-established physics tells us that greenhouse gas molecules (especially water vapor, carbon dioxide and methane) absorb infrared radiation over a considerable portion of Earth’s emission spectrum and re-radiate in all directions, thereby retaining a substantial fraction of the heat in the lower atmosphere or returning it to Earth’s surface.
  2. The human burning of fossil fuels has led through the 20th and 21st centuries directly to a substantial (~50%) increase in the atmospheric concentration of greenhouse gases over their pre-industrial presence, as seen in Fig. 2. While carbon dioxide concentrations have varied historically, as Koonin notes, the rapidity of the current rise is unprecedented (Fig. 3). This increase enhances the thermal “blanket” the atmosphere provides for the Earth.
  3. The rise in global mean temperatures since the pre-industrial era has a time-dependence that very closely tracks the unprecedented time-dependence of the measured rise in atmospheric greenhouse gas concentrations. This extremely strong correlation is illustrated in Fig. 4 by the Berkeley Earth project’s comparison of the measured surface temperatures with a scaled version of the measured CO2 concentration, interrupted occasionally by few-year cooling episodes associated with sunlight-scattering aerosol injections into the stratosphere by known volcanic eruptions.
Figure 2. Worldwide carbon dioxide emissions from fossil fuel burning (blue points, left-hand axis) and global atmospheric concentration (brown points, right-hand axis) vs. year from 1750 to the present. The figure is from the U.S. Energy Information Administration.
Figure 3. Data gleaned from Antarctic ice core samples, showing Earth climate cycles over the past 800,000 years, as reflected in concentrations of carbon dioxide (CO2, blue curve and scale) and methane (CH4, red curve and scale) and in temperature (black curve and right-hand scale). Note that CO2 concentrations are specified in parts per million (ppm) while those of CH4 are in parts per billion (ppb). The unprecedentedly rapid rise in both greenhouse gases seen at the right edge of the plot continues, driven by ongoing fossil fuel burning.
Figure 4. A comparison of Earth surface temperature records since 1750, extracted by the Berkeley Earth team from worldwide meteorological stations, with a simple scaling (solid black curve) of the independently measured global atmospheric CO2 concentrations, with superimposed occasional negative few-year spikes associated with known (and labeled) major volcanic eruptions.

Koonin clearly acknowledges points (1) and (2) above in his book, but doesn’t stress the unprecedentedly rapid rise of greenhouse gas concentrations in Fig. 3 or even mention the correlation described in point (3) and Fig. 4. By ignoring those features, Koonin claims that human-induced effects on global and local climate will be small and slowly varying.  He thus concludes that we have ample time to study climatic effects in more detail.  However, Fig. 3 shows that both CO2 and methane levels are at their highest point in the past 800,000 years, and that they are rising at an extremely high rate, a rate consistent with human activity but inconsistent with other ‘natural’ causes.  And Fig. 4 indicates that global temperatures are tracking that unprecedented rise.

The close agreement of the solid curve in Fig. 4 with the overall trend of the temperature data was sufficient to convince another prominent skeptic of human-caused climate change, Richard Muller, who established the Berkeley Earth project. Muller described his conversion in a 2012 New York Times op-ed:

“How definite is the attribution to humans? The carbon dioxide curve gives a better match than anything else we’ve tried. Its magnitude is consistent with the calculated greenhouse effect — extra warming from trapped heat radiation. These facts don’t prove causality and they shouldn’t end skepticism, but they raise the bar: to be considered seriously, an alternative explanation must match the data at least as well as carbon dioxide does. Adding methane, a second greenhouse gas, to our analysis doesn’t change the results. Moreover, our analysis does not depend on large, complex global climate models, the huge computer programs that are notorious for their hidden assumptions and adjustable parameters. Our result is based simply on the close agreement between the shape of the observed temperature rise and the known greenhouse gas increase… Just as important, our record is long enough that we could search for the fingerprint of solar variability, based on the historical record of sunspots. That fingerprint is absent.”

We agree with Muller that the established correlation does not prove causality. But it does provide a “natural” explanation for the ongoing warming, an explanation that is independent of any global climate models. In contrast, Koonin holds out any confirmation of human dominance here until we have far more successful global climate models that give us an accurate account of all the other climate drivers that nature offers. He is well aware that nuclear and particle theorists have a very strong preference for theories that meet a rather specific definition of “naturalness.” But his climate skepticism is based on the possibility of a quite “unnatural” account, in which a variety of nature’s drivers conspire to produce the same time-dependence as the single human driver of greenhouse gas emissions. Occam’s razor favors the human influence account.

The scaling factor applied to the carbon dioxide concentration to obtain the solid curve in Fig. 4 is, by the way, completely consistent with the account from more sophisticated global climate models. The solid curve indicates an ~1.5°C global mean temperature increase since pre-industrial times, while the measured carbon dioxide concentration has gone just about halfway toward a doubling of the pre-industrial value of 280 parts per million. That suggests that doubling the atmospheric CO2 – an eventuality we would reach during this century if greenhouse gas emissions continue at anything like their present pace — would introduce a human-caused temperature increase of about 3°C. That value, the so-called Equilibrium Climate Sensitivity, falls near the middle of the range of values suggested by not only the current suite of global climate models discussed by Koonin, but in fact by all serious climate modeling done over the past four decades. The sensitivity suggested by Fig. 4 is a proper cause of concern to policy-makers, even while global climate models fail to converge on a more precise value.

III. are we feeling impacts of climate change yet?

Koonin bemoans the “popular perception that extreme [weather-related] events are becoming more common and more severe.” He attributes this perception to sensationalized journalism that ignores detailed data showing that each class of extreme events, considered on its own, is exhibiting behavior that is in no way remarkable in the historical record.  In his opening chapter he includes several “headlines” which he judges that many readers will find surprising, but that he claims accurately reflect the data: “Record high temperatures are becoming rarer; humans have had no detectable impact on hurricanes over the past century; Greenland’s ice sheet isn’t shrinking any more rapidly today than it was eighty years ago; the global areas burned by fires each year has declined by 25% since observations began in 1998.” Each of these statements is at best misleading, in that it obscures, rather than reveals, the central truth. We find it somewhat irresponsible to make such statements without further discussion in the opening chapter, past which many prospective readers will not go.

Later in his book, he devotes an entire chapter attempting to demonstrate how misleading is the summary statement below from the U.S. government’s 2017 Climate Science Special Report (CSSR):

There have been marked changes in temperature extremes across the contiguous United States. The number of high temperature records set in the past two decades far exceed the number of low temperature records.”

Despite Koonin’s insistence on misreading it, that statement is carefully worded and accurate. It is not even inconsistent with Koonin’s mock headline about record high temperatures. This can be seen in Fig. 5, which we previously included in our debunking of the Heartland Institute 2017 booklet Why Scientists Disagree About Global WarmingThe figure is from a 2009 research paper by Meehl, et al. From whenever one starts record-keeping, the frequency of temperature records will fall off in subsequent years, simply because there are more prior-year results to contend with. In the initial year considered, every day and location records a record high and a record low. If the climate is stable and marked only by more-or-less random temperature fluctuations, one would expect record highs and lows to be only half as likely in the second year, one-third as likely in the third year, and so on. Over time, the record frequency should then fall as 1/N, where N is the number of years since recording was started. That falloff is indicated by the solid line in the upper frame of Fig. 5. Koonin’s headline that “High temperature records are becoming rarer” thus carries very little information about the climate.

Figure 5. The annual number of record high (red) and record low (blue) temperatures, and their ratio (lower frame), recorded across the U.S. between 1950 and 2008, and projected out to 2100 within a global climate model, from Meehl, et al. (2009).  Both high and low records decrease in frequency since the beginning of good record-keeping, but record highs were recorded at more than twice the rate of record lows in 2008. Note the logarithmic scale in both frames.

The important take-away from Fig. 5 and other similar compilations is that the time trends of record high and record low temperatures are now diverging from one another. Record highs are falling in frequency less rapidly than the 1/N expectation, while record lows are falling much faster in frequency. The 2008 difference in frequency was more than a factor of 2, but that factor is projected in climate models, without mitigation of human greenhouse gas emissions or substantial changes in other climate drivers, to rise to about 20 by the end of the century, as the global mean temperatures would continue to increase. That is a clear signal of global warming. The warming we are seeing is quite rapid compared with the historical record, but it would take much more rapid global warming than we are now experiencing for the record high curve in Fig. 5 to begin increasing, instead of just remaining flat or decreasing more slowly than 1/N.

Why does the trend in low-temperature record frequency in Fig. 5 appear to deviate further from the 1/N curve than the high-temperature record trend? That occurs because nights are warming faster worldwide than days, though both are warming; the greenhouse gas thermal blanket impact is currently felt more strongly when the sun is not shining. Koonin uses basically this observation to claim that the Earth is simply growing “more temperate” rather than experiencing higher extreme temperatures. Even if the highest temperatures reached in the U.S. have not increased dramatically compared, say, to the Dust Bowl period of the 1930s and 40s, people experience a warming trend if those highs are reached in more locations (ask Alaskans), and over more extended time periods, than previously.

We are sure that Koonin would object to beginning the plot in Fig. 5 in 1950, because that bypasses the unexplained warming trend experienced in the 1930-40 time period, seen as a small blip above the solid curve in Fig. 4 and much emphasized by Koonin. But starting earlier would not eliminate the fact that record high temperatures are currently occurring much more often than record lows, just as stated in the CSSR. And the existence of previous slightly warm periods does not outweigh the observation that a number of extreme weather impacts appear to be occurring simultaneously now.

Among the predictions from various global climate models are indications that global warming should induce more extended heat waves, more severe hurricanes and storms arising over warmer seas, more extreme rainfall variations, with some regions seeing increased flooding and others seeing more extended droughts, more severe and extensive wildfires occurring in regions characterized by heat waves and drought conditions. For each of these effects in turn, Koonin uses technical data and substantial year-to-year fluctuations to argue that “there have been times before human influences became significant that were at least as active as today.” He attributes public perceptions of increased activity to sloppy and sensationalized journalism and anecdotal reports, noting that it is not possible to reliably attribute any specific event to human-induced global warming.

While he is correct in expressing skepticism regarding single-event attributions, he simply ignores actuarial data such as that in Fig. 6, which suggest that something significant is occurring in climate-related disasters. The figure reports the number per year, since 1980, of worldwide natural disasters of various types that have caused at least one human fatality and/or extensive financial damages, compiled by Munich RE, the world’s largest reinsurance agency. One sees substantial, correlated increases in frequency over the three categories Munich RE identifies for climate-related disasters, including among those categories severe storms, heat waves, floods, droughts and forest fires. Cumulatively, the severe climate-related catastrophes have more than tripled in frequency between 1980 and 2018. One might be tempted to attribute such a correlated and steady increase to a rise in population density or in expenses of damage repair, so that more and more events surpass the financial damage threshold imposed to count events in this figure. But the figure also includes geophysical events – earthquakes, tsunamis and volcanoes – unrelated to climate, and these show no comparable increase in frequency. Statistics like those in Fig. 6 are more closely related to public perceptions of impacts on humans than the technical measures on which Koonin relies.

Figure 6. Statistics maintained by reinsurance company Munich RE since 1980 on the number of substantial natural loss events recorded worldwide each year. While the frequency of geophysical events (brown bars at the bottom) unrelated to climate change, such as earthquakes, tsunamis and volcanoes, has remained roughly constant, the frequency of severe storms, floods, droughts, forest fires, and extreme temperature periods has more than tripled since 1980.

Koonin would again object to beginning Fig. 6 in 1980, since his examples of historical variability often predate the start of Munich RE actuarial statistics. But the most important take-away from Fig. 6 is the correlation among different classes of climate-related disasters. The more appropriate question is not whether previous historical periods with high human population density have seen roughly comparable increases in one or another type of activity, but whether they have seen similar correlations among the classes of events. By ignoring such correlations – by refusing to connect the dots – Koonin weakens his case that there is nothing unusual to see here. The statistics themselves, of course, do not necessarily indicate human impacts on climate, but the fact that they are occurring over the same time period as the increased greenhouse gas concentrations and global mean temperatures seen in Figs. 2 and 4 is properly cause for concern.

Over and above ignoring correlations, what is wrong with Koonin’s mock headlines from the opening chapter?

  • Record high temperatures are becoming rarer (or more accurately, no more frequent) – strictly true because of a counting effect – new records are harder to set the more previous years’ worth of records one has to compete against – having relatively little to say about global climate (Fig. 5).
  • Humans have had no detectable impact on hurricanes – while human impacts are hard to detect directly, recent data reveal a clear increase in the fraction of hurricanes since 1980 that become severe, i.e., that reach categories 3-5 (see Fig. 7).
  • Greenland’s ice sheet isn’t shrinking more rapidly – accurate but misleading, as airborne and satellite measurements reveal that the average annual ice loss from Greenland during the period 2003-2010 is about 2.5 times higher than the average from 1900-2003, and that the rate of ice loss has been increasing for the past several decades.
  • Global areas burned by fires are declining – accurate but misleading, because the global decrease is strongly driven by decreases in the intentional fires set to remove flammable vegetation for agricultural clearing, especially in tropical areas, as opposed to the uncontrolled climate-related wildfires that have increasingly devastated parts of Australia, California and Canada, for example, in recent years.
Figure 7. The fraction of hurricanes that have developed into major hurricanes (category 3-5) globally from 1979 through 2017.  Data are binned into 3-year periods. Though significant fluctuations remain, the solid average line suggests a 25% increase in severe storms over the analyzed period. The figure is from Kossin, et al. (2020).

Another important prediction of global climate models is that the combination of warming seas and melting ice sheets will lead to accelerating rise of global sea levels, with potentially devastating effect on coastal communities. Here Koonin notes that sea levels have been rising for a long time and the current rate of sea level rise is not a historical outlier; he implies that it is not accelerating. His claim about acceleration is refuted by extractions of the rate of sea level rise from a century’s worth of tide gauge measurements and from satellite measurements available over the past three decades. As shown in Fig. 8, from Dangendorf, et al. (2019), the rate of rise in mm/year has been increasing steadily over the five decades of the strongest global temperature increase – a clear indication of accelerating sea level rise. Again, Koonin is correct that the current rate of rise has been exceeded in the emergence from past ice ages, but the correlation of the ongoing acceleration in Fig. 8 with the recent 4-to-5-decade trends seen in Figs. 2, 4 and 6 should not be swept under the rug, as Koonin does.

Figure 8. The rate of global mean sea level (GMSL) rise in millimeters per year from 1900 to 2015, based on data from tide gauges (blue band) and satellites (red curve) compiled by Dangendorf, et al. (2019). The level has been rising throughout this period, but the rise is clearly accelerating over the past half-century.

The correlation of all the impacts discussed above during the past half-century provides a strong indication that we are already suffering impacts of human-caused climate change, despite Koonin’s attempt to “divide and conquer” the suite of impacts by misleading presentations of research results and accurate “half-truths.”

IV. What, if any, actions should be undertaken to mitigate climate change?

While Koonin is skeptical that human influences dominate ongoing climate change, he is downright pessimistic about doing anything about it:

Fifteen years ago, when I was in the private sector, I learned to say that the goal of stabilizing human influences on the climate was ‘a challenge,’ while in government it was talked about as an ‘opportunity.’ Now back in academia, I can more forthrightly call it what it is: ‘a practical impossibility’…”

That is a self-fulfilling prophesy – an excuse for doing nothing – and one fueled in part by a surprising, rather unquestioning, acceptance of economic projections, which contrasts sharply with Koonin’s dismissive attitude toward climate model projections. In fact, the range of outcomes between best- and worst-case economic impacts of climate change seems to us much broader than the range of projected global mean temperature increases.  Koonin shows projections of modest climate impacts on worldwide gross domestic product (GDP) by the end of the century. But do these include costs associated with possible mass migration from coastal regions, when humans have not handled recent mass migrations well? Do they include the impacts of climate-related disasters, such as those considered in Fig. 6, growing steadily in frequency? Do they include costs incurred by possible “climate” wars over control of scant resources? Economic impacts go beyond GDP, especially when one considers impacts on individuals and communities.

The asymmetry in Koonin’s treatment of climate science and economics is reflected in the following statement: “Decisions must balance the cost and efficacy of mitigation measures against the certainties and uncertainties in climate science.” In other words, he is suggesting not a cost-benefit analysis, but a cost-uncertainty analysis. Why not balance the uncertainties in cost of mitigation measures against the uncertainty in cost of inaction? What are best-case estimates for each? Worst-case estimates?

The basic elements of his argument about practical impossibility are familiar. Since Earth temperatures are influenced by cumulative greenhouse gas concentrations in the atmosphere, and CO2 in particular lasts for many decades in the atmosphere, global greenhouse gas emissions have to be radically reduced to limit further increases in atmospheric concentrations. Energy demand grows with population and GDP, so economic development of currently under-developed, high population density, countries will increase worldwide demand going forward. Fossil fuels currently dominate worldwide energy production and renewables are increasing at only a modest rate.

But Koonin’s projections of increases here are more pessimistic even than those of ExxonMobil – hardly a climate activist organization. Koonin estimates that global energy demand will grow by 50% by 2050, and consequently, worldwide emissions of carbon dioxide will still be growing by mid-century. In their 2018 Outlook for Energy, ExxonMobil uses the same demographic and GDP trends in estimating that increasing efficiency in the usage of energy will moderate demand growth to about 20% and CO2 emissions should stabilize by 2040 – see Figs. 9 and 10. Stabilization of emissions is better than “business as usual,” but still insufficient to keep likely global mean temperatures below the 2°C increase from early 20th century values recommended in various IPCC reports. Doing better than the ExxonMobil projections to mitigate human-caused climate changes requires accelerating the growth of renewable energy production to take on a considerably larger role.

Figure 9. Graphs from the ExxonMobil 2018 Outlook for Energy, showing projected changes by 2040 in worldwide energy demand (left) and in household electricity usage (right) in OECD and non-OECD countries.
Figure 10. Graphs from the ExxonMobil 2018 Outlook for Energy, showing projected changes by 2040 in global annual CO2 emissions, organized by region (left) and by energy usage sector (right).

Koonin is skeptical this can happen based largely on political considerations: “The fact that any effective policy must cover all of the major emitting nations around the globe is the nub of the challenge in reducing human influences. The prosperous countries have the resources to reduce their emissions while maintaining their prosperity, and many have started on that path…Yet the developing world has a host of far more immediate and pressing problems facing it…Unless emissions-lite technologies are developed to the point where they are essentially no more costly than emitting technologies, or efforts like the Green Climate Fund become much more substantial, it’s natural to ask ‘Who will pay the developing world not to emit?’”

It seems to us that the proper answer to Koonin’s question is that it is in the self-interest of the prosperous nations to adopt policies and commit resources to accelerate the development of reliable and cost-competitive renewable energy production and electric vehicles, so that developing countries find these alternatives to be the preferred options as their population and GDP grow. The accessible forms of renewable energy – wind, solar, hydroelectric, biomass – vary in abundance from country to country, or even from state to state within the U.S. But an appropriate mix can be found for any region, and scientists and companies in the prosperous countries can consult with developing countries to highlight that appropriate mix.

Koonin’s rhetorical question ‘Who will pay the developing world not to emit?’ implies that this is very unlikely to be realized.  However, we have a precedent for such cooperation.  When it was decided to halt production of ozone-destroying chlorofluorocarbon (CFC) chemicals, a global fund was created to assist developing nations in this transition.  That effort, outlined in the Montreal Protocol, established lines of support for developing countries; it included monetary allocations and also technical and advisory assistance for Third-World countries.  It will take strong global leadership to reach a comparable agreement in the case of greenhouse gas emissions, where the economic scope is much greater than in the case of CFCs.

Koonin worries about the reliability of energy supply if we “rapidly” alter the distribution of energy production, but U.S. national laboratories have been planning for more than a decade already the necessary modifications of electrical grids that would be needed to handle the variability of renewable sources. This is not a new problem, and none of the envisioned shifts can be characterized as an “energy revolution.” We do, however, share Koonin’s skepticism regarding the generation-long political will, wisdom and global leadership that will be needed to adopt coherent, long-lasting, and optimally targeted government policies to direct the massive economic transformation that will accompany reduced global greenhouse gas emissions. The political track record to date is not encouraging.

Even in the best, but still realistic, case of managing greenhouse gas emissions, we are likely to see global mean temperature increases well above 2°C this century. It is possible that such increases will trigger one or more irreversible climate “tipping points,” leading, for example, to the eventual melting of the entire Greenland ice sheet or the permanent death of worldwide coral reefs. While we judge the probability of severe impacts to be far more significant than Koonin does, we agree with him about the importance of research into ways to adapt to the changes, to remove CO2 reasonably efficiently from the atmosphere and store it, and into geoengineering (modifying Earth’s reflection of sunlight) approaches to counteract at least some global warming. What Koonin does not mention is that understanding the downsides of atmospheric geoengineering well enough to proceed from small-scale to large-scale experiments will rely on even more unsettled climate models than those currently projecting significant 21st-century global warming, which Koonin sees as unconvincing and diverging.

Koonin’s bottom line on mitigation efforts is described among his Closing Thoughts: “And for me, the many certain downsides of mitigation outweigh the uncertain benefits: the world’s poor need growing amounts of reliable and affordable energy, and widespread renewables or fission are currently too expensive, unreliable or both.” In essence, Koonin and the rest of the world are weighing the relative dangers of two massive social experiments going forward. One is learning whether humans can weather a large technology-driven transformation in worldwide economy and its influence on human behavior. We have experience in handling such shifts; for example, the development of the combustion engine and of miniaturized electronics led to rapid (few-decade) massive shifts during the 20th century.

The second experiment is learning whether humans and other species can successfully and smoothly adapt to the consequences of rapid growth in greenhouse gas concentrations and global temperatures. We have no cumulative experience as a species in dealing with that situation, because the climate has so far been much more stable throughout the development of human civilization. Koonin believes that the second experiment, for which we have no precedents, is the better bet. Count us among the many who are led to the opposite conclusion, not by media hype, but by our own informed perceptions of the science, the data, the technological challenges, and yes, even the uncertainties.

koonin responds:

Steve Koonin has posted a lengthy response to our above review of his book. We thank Koonin for the response, because we believe, along with him, that the responsible exchange of opposing views can help to illuminate the situation for confused readers. We do not propose to respond point-by-point to his response, which largely repeats claims from his book. Readers who care enough can read our review and his response and decide for themselves which points matter.

However, we do want to emphasize two points here. Some of his responses seem to us as semantic differences. For example, he objects to our labeling of him as a “skeptic;” he prefers “dissenter to the alleged consensus.” Fine. Call him a dissenter. He also objects to our characterization that “Koonin claims that human-induced effects on global and local climate will be small and slowly varying. He thus concludes that we have ample time to study climatic effects in detail.” Koonin is correct that he did not say precisely that. It was instead our attempt to summarize what we took away from the attitudes expressed in his book. Here is a direct quote from the book:

The impact of human influences on the climate is too uncertain (and very likely too small) compared to the daunting amount of change required to actually achieve the goal of eliminating net global emissions by, say, 2075…I would wait until the science becomes more settled – that is, until the climate’s response to human influences is better determined or, failing that, until a values consensus emerges or zero-emissions technologies become more feasible – before embarking on a program to tax or regulate greenhouse gas emissions out of existence or to capture and store massive amounts of carbon dioxide from the atmosphere.”

Given the strong doubts he expresses elsewhere in the book about the convergence of climate models or the readiness of renewable energy production, we interpret his advice to wait as a statement that human effects on climate are unlikely to be a serious problem for a long time, so we have ample time to continue to study the problem. We disagree with that advice because there are potential tipping points beyond which climate-induced change becomes irreversible on time scales shorter than several centuries.

If one is dealing with highly complex models with many parameters and diverging projections, as Koonin details in the book, then it makes sense to ask whether the data themselves tell us something useful about the influence of human activities on climate change. And this brings us to our main point. The dominant theme of our review above, one repeated several times, is that Koonin “refuses to connect the dots,” i.e., he disregards correlations among different measured quantities. He really doesn’t respond directly to this critique in his response but features of his response give us the impression that he has a blind spot in regard to correlations.

For example, he dismisses our Fig. 4 (originally prepared by Richard Muller) as “a cartoon.” That is, frankly, a bizarre characterization of what is essentially a data-to-data comparison. We did not present Fig. 4 as a proof that CO2 is the only driver of climate change. Rather we said it illustrates that the industrial-era global temperature trend exhibits the same overall time-dependence as the independently measured growth in atmospheric CO2 concentration, and even Koonin admits that the latter growth results from the human burning of fossil fuels. The correlation thus suggests that current global warming has a time-dependence bearing the stamp of human activities. Koonin does not find that correlation compelling. He says: “There is warming from 1910 to 1940 that is not reproduced by the solid curve which is as rapid as that in recent decades.” There is indeed a mid-century blip in Fig. 4 which deviates from the CO2 trend and probably results from climate drivers unrelated to greenhouse gases. But we doubt that an unbiased observer looking at Fig. 4 would describe that ~0.2°C blip as comparable to the global warming of the past half-century.

A correlation between two observables might be an accident telling us nothing terribly important. There is the fact, as we mention in our review, that the climate sensitivity suggested by the correlation in Fig. 4 to a doubling of atmospheric CO2 would be about 3.0°C, in the middle of the range projected by essentially all climate models. This suggests that the correlation is not purely accidental. Indeed, none of the global climate models is able to reproduce the past half-century of warming without attributing that growth predominantly to human drivers. And it does not mean that CO2 is the only relevant climate driver. The recent growth in atmospheric methane has the same basic time-dependence, as will secondary feedback effects caused by Earth’s response to the growing greenhouse gas concentrations.

But if one is unimpressed by a single correlation, look at other data. Predicted consequences of global warming such as accelerating sea level rise (our Fig. 8), heat stored in the oceans (which Koonin does discuss in his book), and polar and Greenland ice melt also are increasing dramatically over the past half-century. And though Koonin is correct that the IPCC reports place low confidence in significant growth in climate-related disasters, we maintain that public perceptions that such disasters are becoming more frequent are not simply caused, as Koonin implies, by misleading headlines and questionable attributions for individual storms. The actuarial data in our Fig. 6 demonstrate that high-impact (i.e., deaths and serious financial losses) climate-related disasters have dramatically increased in frequency over the past four decades, while climate-unrelated disasters have seen no comparable frequency change.

Koonin is largely dismissive of correlations. In fact there are only two mentions of the word in the text of his book. One notes that temperature fluctuations are correlated between two locations 42 miles apart. The other mention is in describing “the perils of correlation. A quick Google search reveals that both the number of fishing boats and the amount of movie violence are also among the things that have doubled since 1950, and certainly neither of these trends are due to changes in climate.” True, but there is considerable science relating warming to greenhouse gas concentrations and climate changes to warming. So those correlations cannot be so easily dismissed. Koonin presents a number of results he is sure his readers will be surprised by. In the same vein, we might ask if readers of his book might be surprised to see the results shown in our Figs. 4, 6 and 8, which do not appear and are not referenced in his book.

The confluence of all these correlations does not provide proof beyond a reasonable doubt that human burning of fossil fuels is the primary cause of serious global warming and climate consequences. But together with the still imperfect global climate models, they do provide a preponderance of evidence supporting serious, though still quantitatively uncertain, human influence on Earth’s climate. And “proof beyond a reasonable doubt” is not a standard we can afford to apply in the formulation of public policy when the potential consequences of waiting until there is quantitative precision are as serious as they are from warming continuing throughout this century.

Koonin is confident humans will adapt to the climate changes. So are we, but at what human (not just GDP) and other species’ cost? Humans have by now pretty much adapted to COVID-19, but at the cost of nearly 5 million (and still counting) lives. Many more lives would have been lost if the world’s governments had decided to follow the advice to aim for natural herd immunity, offered by some who found the science of vaccinations and treatments unsettled. Climate mitigation has considerable costs as well. So the critical question one must address is this: What are the costs of climate mitigation vs. the costs of simply waiting until the science is settled to Koonin’s satisfaction? What are best- and worst-case estimates of each?


S.E. Koonin, Unsettled: What Climate Science Tells Us, What It Doesn’t, and Why It Matters (BenBella Books, 2021),

M.P. Mills, Wall Street Journal Article Repeats Multiple Incorrect and Misleading Claims Made in Steven Koonin’s New Book ‘Unsettled,’

R.A. Muller, The Conversion of a Climate Change Skeptic, New York Times, July 28, 2012,

D.J. Wuebbles, et al., 2017: Climate Science Special Report,

G.A. Meehl, et al., Relative Increase of Record High Maximum Temperatures Compared to Record Low Minimum Temperatures in the U.S., Geophysical Research Letters 36, L23701 (2009),

J.P. Kossin, K.R. Knapp, T.L. Olander and C.S. Velden, Global Increase in Major Tropical Cyclone Exceedance Probability Over the Past Four Decades, Proceedings of the National Academy of Sciences of the U.S. 117, 11975 (2020),

K.K. Kjeldsen, et al., Spatial and Temporal Distribution of Mass Loss from the Greenland Ice Sheet Since AD1900, Nature 528, 396 (2015),

D.M.J.S. Bowman, et al., Vegetation Fires in the Anthropocene, Nature Reviews Earth and Environment 1, 500 (2020),

S. Dangendorf, et al., Persistent Acceleration in Global Sea-Level Rise Since the 1960s, Nature Climate Change 9, 705 (2019),

National Renewable Energy Laboratory, Renewable Electricity Futures Study (2012),