There is plenty of evidence that climate change will pose increasingly existential threats to the most vulnerable individuals in society; to low-lying coastal cities and island nations; to indigenous cultures and ways of life; and to numerous plant and animal species, and perhaps even entire ecosystems. Such consequences are well-supported by the existing evidence, are already starting to emerge in certain regions, and should be of paramount concern. But even these very dire outcomes aren’t equivalent to the “end of human civilization”.
Daniel Swain, Comments to Climate Feedback
1. Recap
Many authors think that humanity currently faces high levels of existential risk, often between 10-50% in this century. In this series, I argue that leading estimates of existential risk are often exaggerated.
Part 1 introduced the series along with key concepts such as existential and catastrophic risk. Part 2 looked at Toby Ord’s claim that there is a 1/1,000 chance of irreversible existential catastrophe from climate change by 2100, and argued Ord gives us little persuasive reason to believe that existential risk from climate change in this century is high. That is not to say that climate change will not be catastrophic, only that we have not yet seen evidence for the further claim that climate change poses an existential risk to humanity.
A recent report by John Halstead, Climate risk and longtermism allows us to go further. Halstead examines all of the most plausible ways that climate change could lead to existential catastrophe by 2100, and argues that they are extremely unlikely.
Part 3 of this series looked at two climate threats: heat stress and agricultural change. Today, I want to give an opinionated summary of two more topics discussed by Halstead: rising sea levels and tipping points. I will also use Halstead’s discussion of paleoclimactic data to revisit Ord’s positive argument for climate risk from Part 2.
2. Rising sea levels
As the world warms, sea levels begin to rise. This, Halstead notes, carries many threats to human civilization. Most obviously, coastal areas can become permanently submerged underwater. Perhaps less obviously, coastal areas will also see increased flooding, erosion, ecosystem loss, salinization of water and soil, and impaired drainage of coastal waterways.
Some cities have already experienced a preview of what is to come. In fact, Halstead notes, many cities have seen rises as sharp as 2-4 meters in average sea levels. For example, here are some of the figures that Halstead cites:
| Location | Period | Relative sea level rise |
| New Orleans | 1900-2000 | 2.5m |
| Tokyo | 1900-2000 | 4m |
| Global coastal population | 1993-2015 | 2m |
These changes have been difficult to cope with. Global warming may have contributed to the catastrophic flooding in New Orleans after Hurricane Katrina, and Tokyo has taken costly action to adapt to rising sea levels. But they have not killed, nor even displaced, the vast majority of city residents in either city.
To put these numbers in perspective, here are the IPCC projections for sea level rise, not just by 2100, but all the way out to 2300, across a variety of warming scenarios:
This is horrible. In particular, it illustrates a point nicely stressed by Halstead: we should not put so much focus on climate impacts before 2100, because many impacts, such as rising sea levels, will get significantly worse beyond 2300 even if we get emissions under control.
But it is also survivable. We saw that many cities have already coped with sea level rises comparable to the absolute worst-case rise in global mean sea levels projected by the IPCC, not just in 2100, but in 2300.
This is not to say that everyone will cope. In fact, many of the world’s poorest will be forced from their homes and cities. Here Halstead cites Here Halstead cites projections by Lincke and Hinkel (2021) of migration during the 21st century caused by sea level rise across best, worst and median warming scenarios:
| Country | Median migration (millions) | Min-max migration (millions) |
| Vietnam | 5.7 | 1.2-12.0 |
| India | 5.4 | 3.0-14.5 |
| Indonesia | 1.4 | 0.8-4.5 |
| Bangladesh | 1.2 | 0.2-8.0 |
| Philippines | 1.0 | 0.6-2.5 |
| Brazil | 0.9 | 0.5-1.8 |
| Myanmar | 0.9 | 0.4-3.0 |
| China | 0.8 | 0.2-3.3 |
| South Korea | 0.7 | 0.5-1.3 |
| Japan | 0.6 | 0.2-2.3 |
This is a great human tragedy, concentrated largely in Asia and disproportionately affecting the world’s poorest nations. But it is not an existential catastrophe. Human civilization will not end if millions, or tens of millions are forced to migrate.
It may seem that some experts expect rising sea levels to pose an existential risk in this century. Lobbying by effective altruists has spread the language of existential risk as far as the UN Secretary General’s Common Agenda. And it appears that even the IPCC has been moved to use the term:
As the scale and pace of sea level rise accelerates
IPCC Sixth Assessment Report, Technical Summary, p. 94
beyond 2050, long-term adjustments may in some locations be
beyond the limits of current adaptation options and for some
species and some locations could be an existential risk in the
21st century (medium confidence).
But the term has already been qualified (for some locations, and some localities). In the rest of this passage, the IPCC clarifies exactly what it means.
Nature-based interventions, for example wetlands and salt marshes, can reduce impacts and costs while supporting biodiversity and livelihoods but have limits under high warming levels and rapid sea level rise (high confidence). Ecological limits and socioeconomic, financial and governance barriers will be reached first and are determined by the type of coastline and city or settlement (medium confidence). Accommodation can reduce impacts on people and assets but can address only limited sea level rise. Considering the long term now will help to avoid maladaptive lock-in, to build capacity to act in a timely and pre-emptive manner and to reduce risks to ecosystems and people.
IPCC Sixth Assessment Report, Technical Summary, p. 94
That’s quite a bit tamer than what effective altruists mean when they speak of an existential risk.
There is every cause for concern about rising sea levels. That is especially true for the inhabitants of coastal areas, some of whom may be forced to abandon their settlements in this century. But nothing approaching human extinction or the permanent curtailment of our potential for future development has even a slight chance of resulting from rising sea levels in this century.
3. Lessons from the paleoclimate
Chapter 3 of Halstead’s report looks at the paleoclimate, that is, the climate in earth’s historic past. At times, Earth has been much cooler than it is today, and at times it has been much warmer.
There have been five mass extinction events in the earth’s history, and some scientists project that we are in the midst of a sixth (Thomas et al. 2004). Climate change has been connected to many of these mass extinctions. What can the history of climate change and extinction teach us about existential risk from climate change today?
Paleoclimate data is normally used to make predictions about the fate of nonhuman animals and plants due to climate change. I am a bit skeptical of using paleoclimate data to predict rates of human extinction, for two reasons:
- Humans are different: Looking at how nonhuman species fared in the past doesn’t tell us much about what humans can survive. We can adapt to climate change in a number of ways that nonhumans can’t, ranging from air conditioning and changes in agriculture all the way up to decarbonization. So there is no clear inference from past nonhuman extinctions to future human extinction.
- Mechanisms are exhaustive: Once we review all of the plausible mechanisms (heat stress, sea level rise, etc.) by which climate change could kill us, it’s not clear what additional value is added by looking at history, unless that history can help us to understand why we have underestimated the extinction risk from some particular mechanism. Moreover, insofar as many of the mechanisms linking climate change to historical extinctions were found to pose a low risk of human extinction, this may give us good reason to discount arguments from paleoclimate data to high levels of climate risk.
But I will bite. I think Halstead is quite convincing in arguing that the paleoclimate data does not give us good reason to expect climate change to bring about irreversible existential catastrophe any time soon, and probably gives us good reason to expect that it won’t.
The first thing to note is that the earth has been much hotter in the past than even the most dire warming projections predict it will soon become. During the Paleocene-Eocene thermal maximum (PETM) 55 million years ago, the earth was an astounding 14-17°C above the pre-industrial average, but species extinctions were limited. Here is Halstead:
Overall, in the PETM, temperatures were upwards of 17°C higher than pre-industrial levels and the only species that went extinct that we know of was a single-celled marine organism, and on land it was a time of ecological flourishing, persistence and diversity.
But if the scorching temperatures of the PETM did not lead to existential catastrophe, why believe that lower levels of warming will now destroy humanity? We saw in Part 2 that even Ord seems concerned by this thought:
The best argument against … unknown mechanisms [for extinction] is probably that the PETM [Paleocene–Eocene Thermal Maximum] did not lead to a mass extinction, despite temperatures rapidly rising about 5 degrees Celsius, to reach a level of 14 degrees Celsius above pre-industrial temperatures.
Toby Ord, The precipice
However, Ord suggests that even though the PETM was hotter than the 21st century will be, warming during the PETM was slower than the rapid warming we are now experiencing:
Most importantly, anthropogenic warming could be over a hundred times faster than warming during the PETM, and rapid warming has been suggested as a contributing factor in the end-Permian mass extinction, in which 96 percent of species went extinct.
Toby Ord, The precipice
Fair enough. But Halstead has something to say about that too.
Citing Willis and MacDonald (2011), Halstead reminds us that we have seen rapid warming before:
a. In Greenland, temperatures may have risen by 10°C, though this may be an error in climate proxies.
b. In the Swiss Alps and other parts of Europe, a warming of 2 to 5°C appears to have occurred in 200 years or less.
c. In the Sierra Nevada of California, rates of warming in the late glacial may have been 4 to 5°C every 500 years around 15,000 years ago.
d. At the higher latitudes of the Northern Hemisphere, there were increases of 5°C and more over a few decades (11,700 years ago).
e. Data from Greenland ice cores suggest that a >10°C warming may have occurred over 20 to 6 years (13,000 to 11,000 years ago).
f. In California, warming at the close of the Younger Dryas (13,000 years ago) may have been on the order of 3°C in less than 100 to 200 years.
g. For the entire Southwest US, a general warming of 4°C may have occurred in less than a century (13,000 to 11,000 years ago).
Halstead, Climate change and longtermism
Just as history gives us abundant evidence of large mammals surviving temperatures hotter than the earth is projected to be by 2100, history also gives us evidence of survival at rates of warming comparable to the worst-case scenarios in the IPCC report.
Even the worst-looking scenarios give cause for comfort. For example, in the beginning of the Holocene (~10,000 BCE), climate change did seem to play a role in the extinction of megafauna (large animals). But there is little evidence of plant extinctions, and many megafauna survived. In particular, Halstead notes, humans lived, and by all appearances thrived during this time. And as for the megafauna who perished, a live scientific hypothesis is the overkill hypothesis: they died because we killed and ate them. (Halstead assigns this 90% credence, but notes that many scientists may be closer to 50%). All of this gives us every reason for confidence that today’s humans, with greater numbers, technology, and forewarning than our ancestors, can survive ongoing climate catastrophe.
We saw in Part 2 that the only sustained argument which Ord makes in favor of climate risk is his discussion of unknown mechanisms in the context of paleoclimate. But, if Halstead is right to suggest that the paleoclimate data gives us every reason for confidence in humanity’s survival, then we should not have much confidence in Ord’s argument.
4. Tipping points
As the world warms, we will begin to approach tipping points, large-scale climate discontinuities that could lead to major increases in total warming (Lenton et al. 2019). Could tipping points put us on an irreversible course towards existential catastrophe by 2100?
The first thing to say is that, as with the paleoclimate, tipping points need to be contextualized within our understanding of specific mechanisms, such as sea level rise, heat stress and crop failure, by which climate change could lead to existential catastrophe. We have seen that even the most outlandish warming scenarios do not pose a credible threat of existential catastrophe through any of these mechanisms. So even if tipping points raise the probability of extreme warming, it is not clear that they provide a new and credible pathway towards existential catastrophe.
More to the point, Halstead reviews each of the proposed tipping points in the diagram above and argues in detail that none of these tipping points presents a credible threat of human extinction. I will review a few representative examples below. (Interested readers are invited to consult Chapter 8 of the Halstead report).
Begin with dieback of the Amazon rainforest in the tropics and boreal forests near the arctic. Both forests play an important role in trapping carbon, preventing it from reaching the atmosphere.
Halstead suggests (plausibly) that boreal forest dieback is likely to be something of a wash. As the world warms, the southernmost reaches of boreal forests will become too warm to support existing forests. But at the same time, boreal forests will expand northwards into land that was previously too cold to support them. For this reason, boreal forest dieback is not expected to pose a significant existential threat anytime soon.
Amazon rainforest dieback is a bit more concerning. Indeed, Halstead suggests, there is a good chance that human-caused deforestation will continue to devastate the Amazon rainforest alongside the destruction caused by climate change. However, Halstead reminds us, most tropical forests fluorished during the Eocene period, in which periods were more than 10°C hotter than they are today. So there is no ready inference from present-day warming to a catastrophic collapse of the Amazon rainforest.
As Halstead reminds us, the IPCC’s latest assessment report estimates that even on pessimistic warming scenarios (4°C warming combined with some pessimistic modeling assumptions), Amazon dieback is expected to contribute about 0.3°C of additional warming. A catastrophe? Yes. But an existential catastrophe? Certainly not.
What about arctic permafrost? A great deal of carbon is currently trapped in the arctic permafrost, and as the world warms, permafrost will gradually melt, releasing carbon into the atmosphere. At Halstead reminds us, the IPCC currently projects that each 1°C of warming will lead to 18 gigatons of carbon (GtC) released from arctic permafrost. That is roughly two percent of the amount projected to be released by 2100 on a middle-of-the-road scenario (RCP 4.5) by the IPCC. As before, this is a regrettable catastrophe, but hardly an existential threat.
What about the most speculative scenarios? A rather controversial paper by Will Steffen and colleagues, “Trajectories of the earth system in the Anthropocene” predicts that once we reach 2°C of warming, the world will enter a `Hothouse Earth’ phase, in which even with little additional use of fossil fuels, cumulative warming will reach 6°C above preindustrial levels.
At the risk of beating a very dead horse, the first thing to say about this scenario is that it still would not kill us. We have reviewed almost every plausible direct mechanism (heat stress, crop failure, sea level rise, …) by which warming could lead to existential catastrophe. And we have seen that none of these mechanisms plausibly threatens anything like existential catastrophe even under the extreme assumption of 6°C warming.
More to the point, `Hothouse Earth’ is rather speculative science, and media coverage has stretched its conclusions far beyond those initial speculations. The model itself is highly uncertain and presents options for avoiding catastrophe. Here is Richard Bettes, Professor of Climate Impacts at the University of Exeter:
With some exceptions, much of the highest-profile coverage of the essay presents the scenario as definite and imminent. The impression is given that 2°C is a definite “point of no return”, and that beyond that the “hothouse” scenario will rapidly arrive. Many articles ignore the caveats that the 2°C threshold is extremely uncertain, and that even if it were correct, the extreme conditions would not occur for centuries or millennia.
Richard Betts, “Hothouse earth: Here’s what the science actually says“
And what exactly is the research-backed case behind `Hothouse Earth’? Here things get a bit hairy, because as Betts reminds us, Steffen et al. (2018) is an opinion piece, not a standalone research paper, and as such it really doesn’t do the kind of modeling that would be needed to advance its thesis to the realm of serious credibility.
One thing that strikes me about the scientific literature on “tipping points” is that there are a lot of review papers like this that end up citing the same studies and each other … There is a great deal of interesting, insightful research going on using theoretical methods and calculations with large approximations. However, we have yet to see an equivalent level of research in the highly-complex Earth System Models which generate the kind of detailed climate projections used for addressing policy-relevant questions by the Intergovernmental Panel on Climate Change (IPCC). Steffen and colleagues have made a good start at addressing such questions, going as far as they can on the basis of the existing literature, but their essay should motivate new research to help narrow down the huge uncertainties. This will help us see better whether “Hothouse Earth” is our destiny, or mere speculation.
Richard Betts, “Hothouse earth: Here’s what the science actually says“
At the risk of piling on, Halstead adds another objection to the mix – Steffen and colleagues seem to have forgotten to argue for the main conclusion of their paper: that climate feedbacks could generate six degrees of warming. Halstead collects together all of the feedback effects mentioned by Steffen and colleagues in the main text:
The problem? These add up to a mere 0.47°C of warming by 2100! If feedback effects are to take us from 2°C of 6°C of warming, Steffen and colleagues need to find an additional 3.53°C of warming after 2100 due to feedback effects.
Steffen and colleagues certainly claim that this additional warming will occur. But they are not terribly precise about the details. As Halstead puts it: “they do not explain what might cause the remaining 3.5°C of warming over millennia implied by their Figure 1”.
This sort of speculative modeling can be a good impetus for future work, but it should not be taken as a scientifically respectable ground for alarmist predictions of the likelihood of impending existential catastrophe due to climate change. There is simply not enough evidence to ground such predictions.
5. Looking ahead
In this post, we looked at three more ways to motivate the idea that climate change poses a significant threat of existential catastrophe by 2100: an appeal to rising sea levels; Ord’s appeal to warming in the paleoclimate; and models of climate tipping points.
We saw that even on the most dramatic scenarios, rising sea levels and tipping points do not pose a credible risk of existential catastrophe by 2100. We also saw that many of the most dramatic scenarios are substantially under-evidenced.
We saw that discussions of paleoclimate and tipping points must also be contextualized against earlier discussions of extinction mechanisms. To draw lessons from history or to speculate about future tipping points is not to pose a novel mechanism by which climate change could kill us all. That mechanism must have a name, such as rising sea levels of crop failures, and if we did not find this mechanism to be credible to begin with, there is only so much that reflections on history or tipping points can do to change our minds.
At this point, there is mounting evidence that Ord’s estimate of a 1/1,000 chance that climate change puts us on an irreversible course to existential catastrophe by 2100 is inflated by several orders of magnitude. More than that, we have struggled to identify scientifically plausible mechanisms that could have motivated such an estimate in the first place. When this is taken in tandem with Ord’s own silence on his reasons for estimating a high level of climate risk, we begin to get the impression that Ord’s numbers may not have had much science behind them in the first place.
That is not good. Facts matter. Science matters. Existential risk estimates must be put on a firm scientific basis.
The next part of this series will look at some final mechanisms by which climate change could pose an existential risk, then draw conclusions from our discussion of climate change. Then I will move on to discussions of other types of existential risks, where we will see a similar pattern.
In the meantime, are there arguments that I have neglected linking sea level rise, tipping points or the paleoclimate to existential catastrophe? Are there other mechanisms that you would like to hear about in my final discussion of climate risk? Let me know.
Leave a Reply