Human Extinction from Natural Hazard Events
Human Extinction from Natural Hazard Events
- Anders SandbergAnders SandbergFuture of Humanity Institute, University of Oxford
Like any other species, Homo sapiens can potentially go extinct. This risk is an existential risk: a threat to the entire future of the species (and possible descendants). While anthropogenic risks may contribute the most to total extinction risk natural hazard events can plausibly cause extinction.
Historically, end-of-the-world scenarios have been popular topics in most cultures. In the early modern period scientific discoveries of changes in the sky, meteors, past catastrophes, evolution and thermodynamics led to the understanding that Homo sapiens was a species among others and vulnerable to extinction. In the 20th century, anthropogenic risks from nuclear war and environmental degradation made extinction risks more salient and an issue of possible policy. Near the end of the century an interdisciplinary field of existential risk studies emerged.
Human extinction requires a global hazard that either destroys the ecological niche of the species or harms enough individuals to reduce the population below a minimum viable size. Long-run fertility trends are highly uncertain and could potentially lead to overpopulation or demographic collapse, both contributors to extinction risk.
Astronomical extinction risks include damage to the biosphere due to radiation from supernovas or gamma ray bursts, major asteroid or comet impacts, or hypothesized physical phenomena such as stable strange matter or vacuum decay. The most likely extinction pathway would be a disturbance reducing agricultural productivity due to ozone loss, low temperatures, or lack of sunlight over a long period. The return time of extinction-level impacts is reasonably well characterized and on the order of millions of years. Geophysical risks include supervolcanism and climate change that affects global food security. Multiyear periods of low or high temperature can impair agriculture enough to stress or threaten the species. Sufficiently radical environmental changes that lead to direct extinction are unlikely. Pandemics can cause species extinction, although historical human pandemics have merely killed a fraction of the species.
Extinction risks are amplified by systemic effects, where multiple risk factors and events conspire to increase vulnerability and eventual damage. Human activity plays an important role in aggravating and mitigating these effects.
Estimates from natural extinction rates in other species suggest an overall risk to the species from natural events smaller than 0.15% per century, likely orders of magnitude smaller. However, due to the current situation with an unusually numerous and widely dispersed population the actual probability is hard to estimate. The natural extinction risk is also likely dwarfed by the extinction risk from human activities.
Many extinction hazards are at present impossible to prevent or even predict, requiring resilience strategies. Many risks have common pathways that are promising targets for mitigation. Endurance mechanisms against extinction may require creating refuges that can survive the disaster and rebuild. Because of the global public goods and transgenerational nature of extinction risks plus cognitive biases there is a large undersupply of mitigation effort despite strong arguments that it is morally imperative.
- Risk Assessment
Like any other species, Homo sapiens can potentially go extinct. Related hominin species have gone extinct, H. neanderthalensis just 40,000 years ago. Indeed, 99.9% of all species that have existed are extinct (Raup, 1986), making this a likely outcome in the long run.
Human extinction comes under the umbrella term of existential risk, defined as “premature extinction of Earth-originating intelligent life or the permanent and drastic destruction of its potential for desirable future development” (Bostrom, 2002, 2013). This covers both standard extinction and possible degenerate states; the remainder of this article will focus on extinction per se.
Much research on existential risk is focused on anthropogenic existential risk, especially technologically mediated risks such as nuclear war or bioweapons. This article will focus on hazards not directly caused by human activity, although the risk of aggravating existing hazards through human action will be discussed. It should be noted that many of the largest extinction risks at present are anthropogenic.
While human extinction has been discussed mainly from a philosophical perspective (how bad is it? How to reason about such unlikely and future events? What are the rational responses?), many of the subfields of natural hazard science inform analysis. Many of the extinction risks are due to extreme cases of already-occurring catastrophes. In addition astrophysics, geophysics, epidemiology, and systems science naturally show up with regard to particular hazards, as well as ecology and paleontology with regard to extinction dynamics. Much thinking about response and mitigation has links to economics, social science, and philosophy.
Despite its obvious interest and relevance to humans, rigorous study of human extinction have so far been relatively rare. There are more academic papers on dung beetles than the fate of H. sapiens (Bostrom, 2013).
This article is organized as follows: first, an overview of the history of thought on human extinction. This is followed by an overview of human extinction in general, and discussion of various particular risks, as well as compound systemic risks. This is followed by a discussion of extinction probabilities and mitigation options.
Humanity has entertained apocalyptic thoughts since the earliest history. For a long time considerations of the end of the world were framed as theological or cultural narratives promising a satisfying end of history, an end of suffering, revelation, or a way of structuring history (Amanat & Bernhardsson, 2002).
The early modern realization that the heavens were not unchanging and the spread of a progressive rather than static or cyclic view of history began the shift of end-of-the-world scenarios from pure theology into the realm of potential history.
The rise of catastrophism in the 19th century as an explanation of fossil distribution and geological features through past upheavals led to the possibility that new global disasters could occur. This coincided with the growth of astronomical understanding about meteors, comets, and asteroids that naturally led to the conclusion that comets or asteroids could hit the earth or otherwise interact with it. Although much of this was confined to scientific speculation and fiction, it occasionally produced scares such as the concern in 1910 that toxic gas from the tail of Halley’s Comet might poison the earth’s atmosphere (Bartholomew & Radford, 2011, chapter 16).
The development of evolutionary biology also led to the conclusion that humans were a species among others, and hence potentially could go extinct or evolve into a degenerate form. Finally, thermodynamics introduced the concept of the heat death of the universe, the realization that at some point everything would reach thermal equilibrium and life inevitably had to die out (both degeneration and heat death appear prominently in H. G. Wells’s The Time Machine ). Together these findings made human extinction an actual scientific (if speculative) possibility rather than a cultural story.
In the 20th century the threat of nuclear weapons (Kuznick, 2007) and environmental degradation made existential risk even more salient, but moved the focus toward anthropogenic risks. Now the risks were experienced as real, direct, and something that could happen within one’s lifetime (e.g., Ehrlich, 1968). Through the link to ecology and systems science an understanding of human extinction risks as being linked and systemic emerged, perhaps best exemplified by the Limits to Growth report where a mathematical model predicted overpopulation, resource use, and pollution would interact and eventually cause a future collapse (Meadows et al., 1972). The shift in focus away from natural risk also led to the realization that human agency could affect and reduce extinction risks through appropriate policy.
Near the end of the 20th century an interdisciplinary field of human extinction research began to emerge. This was partially due to philosophical interest triggered by considerations of the moral weight of extinction, as well as the conundrum of the “Doomsday argument” and how to handle extinction in terms of decision theory (e.g., Parfit, 1984; Leslie, 1998; Bostrom, 2003, 2013; Posner, 2004), partially by the recognition that there exists cross-cutting similarities and issues between different risks allowing them to be treated under the same heading (Rees, 2003; Matheny, 2007; Bostrom & Ćirković, 2011). Concerns about radical emerging technologies contributed to the rapid growth and diversity of the field (Joy, 2000; Bostrom, 2009). Organizations aimed at investigating and reducing such risks were set up: the Future of Humanity Institute at Oxford University in 2005, the Centre for the Study of Existential Risk at Cambridge University in 2012, and the Future of Life Institute in Boston in 2014.
Surveys of the public show that a large fraction—sometimes a majority—regard pessimistic scenarios, including dystopias or the imminent end of our existing way of life, as likely. In a survey 24% respondents believed there was 50% chance that humanity would be wiped out within a century (Randle & Eckersley, 2015). This is more pessimistic than many expert assessments. A key question is whether this outlook will be channeled into fatalism or activism.
Extinction of a species can be caused by something that disrupts the necessary parts of the environment (such as a direct disaster or habitat loss), directly harms individuals (a new predator, disease), or ecological change faster than the species can evolve (environmental change, a new competitor) (Smith, 1989). Predation or ecological competition is not a likely risk to H. sapiens at present, although it may have played a role in the extinction of past hominins. The species is also able to adapt faster to environmental changes through cultural change than through genetic change, making disruptive risks and direct harms more important than slow ecological change.
A special version of natural extinction is demographic collapse: if fertility becomes too low the population will shrink, and may eventually die out. The trend in many societies has been toward fertility rates below replacement; were this trend to continue universally, demographic collapse might occur (Bainbridge, 2009). The determinants of human fertility are a complex mix of biological, psychological, sociological, and economic factors (ESHRE, 2001). Predicting future fertility is an open question, and long-term decline cannot be ruled out (Lutz & Qiang, 2002; Lutz, 2009). Conversely, overpopulation is one of the classic scenarios for how resource stresses may cause a terminal implosion of the biosphere or human civilization (Ehrlich, 1968; Hern, 1990).
Over long timescales we may expect H. sapiens to naturally disappear through genetic drift or speciation. This may not count as extinction since there will be at least one successor species. Whether this should be regarded as desirable or not depends to a large degree on what the species evolves into (Bostrom, 2004).
Even in many extreme global disasters there will be survivors. Whether the population recovers or dies out depends on whether the survivors can form communities larger than the ecological minimum viable population (MVP). While small founder populations can be lucky and grow into large and stable populations, this is relatively unlikely. Populations smaller than the MVP are likely to become extinct due to further disasters or demographic, environmental, or genetic stochasticity.
Simulations suggest that for humans with mortality and fertility as hunter-gatherer societies MVP is on the order of a few thousand individuals. This is in line with other animal MVPs (Reed et al., 2003). Endurance mechanisms therefore need to ensure that survivor groups are large enough and can sustain themselves.
Astronomical and Physical Risks
Supernovas and Gamma Ray Bursts
Supernovas and gamma ray bursts (GRBs) represent rare but exceedingly powerful energy releases that could conceivably harm biospheres at astronomical distances. While supernovas have been recognized as a potential biosphere threat since the 1950s (Krasovsky & Shklovsky, 1957), gamma ray bursts were recognized as a risk first in the 1990s (Thorsett, 1995).
Unless the explosion is close enough to cause heating, the risk comes from radiation penetrating the atmosphere, a resulting UV flash, possible cosmic ray showers, and formation of nitrous oxides that deplete the ozone layer, produce acid rain, and cause multiyear cooling climate effects (Thomas & Melott, 2006; Galante & Horvath, 2007; Melott & Thomas, 2011; Horvath & Galante, 2012). The effects, if intense enough, could plausibly cause a mass extinction (Ellis & Schramm, 1995; Melott et al., 2004).
There is some uncertainty on what the damaging distance is, largely driven by uncertainty in biosphere response to radiation and ozone depletion. Supernovas radiate in all directions, making their effects decline with the square of the distance. GRBs are directional, beaming much of their energy in two narrow cones, hence having longer reach in some directions but not others. Estimates for supernova ozone depletion produce a risk radius of some tens of parsecs or less (Ruderman, 1974; Ellis & Schramm, 1995; Gehrels et al., 2003; Beech, 2011; Melott & Thomas, 2011). The UV flash may damage biospheres up to 150 kiloparsec from a GRB but would affect only one hemisphere. Ozone destruction may have a reach of 12–14 kiloparsec (Scalo & Wheeler, 2002; Galante & Horvath, 2007; Horvath & Galante, 2012).
There are at present no supernova or GRB candidates in the vicinity of the sun that will explode in the current epoch (Beech, 2011). The estimated rate of earth-affecting events is likely on the order of 1 in 100 million years, assuming random locations in the galactic disc and current observed rates (Melott et al., 2004; Beech, 2011). For smaller damage distance estimates the rate is an order of magnitude lower (Gehrels et al., 2003).
Earth is subject to impacts of Near Earth Objects (NEOs) and long-periodic comets. While the possibility of impacts causing disasters was suggested early by Halley, Laplace, and others, the serious possibility of global risk was first convincingly brought up in 1980 when Luis Alvarez suggested a link between an asteroid impact and the Cretaceous–Tertiary mass extinction event (Alvarez et al., 1980). Since then much research has tried to establish a causal role of impacts in past mass extinctions. While conclusive proofs have not yet been found, there is consensus that a large impact could cause a mass extinction and hence human extinction.
The effects of impacts depend to a large degree on impactor mass (although velocity and impact site can have relevant effects [Walkden & Parker, 2008]). For smaller (<1.4 km diameter) impactors the effects are local or transmitted through tsunamis. Beyond this size the main hazard is global cooling due to stratospheric ejecta and soot from wildfires, as well as harm the ozone layer through nitrous oxides. Ejecta from larger impactors may cause globally distributed fires and darken the sky enough to prevent photosynthesis for months, and injections of sulphate aerosols and water vapor into the stratosphere would change the climate over years (Toon et al., 1997; Pierazzo & Artemieva, 2012; Brugger, Feulner & Petri, 2017). At this scale infrastructure and agricultural collapse is to be expected (Chapman, 2003). Human mortality has been modeled as scaling linearly from 0% at 1.6 km to 100% for 10 km impactors (Stokes et al., 2003), although this is at best an educated guess. The most extreme impacts (440 km and upward) would sterilize the biosphere (Sleep et al., 1989); such events may have occurred in the early solar system but today only a handful of such bodies remain and are all in stable orbits.
The population of potential impactors has a roughly power-law size distribution (Malamud, 2004), with globally risky asteroids impacting once every few million years and multi-megaton locally devastating impacts such as the 1908 Tunguska explosion every 1,000 years (Brown et al., 2002; Harris & D’Abramo, 2015). At present the NASA Space Guard has mapped an estimated 90% of 1 km or larger NEOs, significantly reducing the remaining risk for the next century. The handful of known 10+ km NEOs are in safe orbits. All remaining risk is from undiscovered large NEOs. The residual human risk is dominated by tsunamis rather than global disaster or extinction (Harris, 2008; NRC, 2010). However, “new” long-periodic comets add a badly characterized risk that may be on the order of one 2+ km impactor per 5 to 10 million years. This is based on estimates from observed comet fluxes, correcting for observational incompleteness, rough models of comet size distributions, and the possibility of brief (2–3 Myr) comet showers due to gravitational interactions with passing stars (Chapman, 2003; Weissman, 2006).
High Energy Physical Risks
An exotic set of conjectured risks to the earth are due to high-energy physical processes. These include cosmic ray interactions forming stable strangelets that convert normal matter into superdense matter, the formation of microscopic black holes that absorb the planet, or the decay of the current false vacuum into true vacuum, creating a bubble with different physical constants that expands at lightspeed. These risks have mostly been discussed in relation to human high energy physics experiments but were originally suggested in theoretical physics as possible events that could in principle be triggered naturally (Rees, 2003).
The overall risk appears small: cosmic rays have interacted with solar system bodies for billions of years with little effect (Ellis et al., 2008), a high rate of strangelet or black hole formation is not compatible with the observed rate of supernovas (Dar, De Rújula, & Heinz, 1999), and the relative young age of earth compared to other planets imply a low (less than one in a billion per year) risk of vacuum decay and similar expanding phenomena (Tegmark & Bostrom, 2005). Nevertheless, pinning down exotic and low-probability risks of this kind poses particular problems given the fallibility of scientific reasoning: the probability of an error in theory, model, or calculation is often larger than the risk probability, making risk assessments weaker in updating credences than for ordinary risks (Ord, Hillerbrand, & Sandberg, 2010).
Extreme natural disasters can only threaten a species if they affect the entire range of the species. In the case of humans this requires a global disaster, which rules out most localized categories (landslides, hurricanes, earthquakes, flooding, normal volcanism, etc.). The most plausible geophysical extinction risks are supervolcanos and climate-related disruption of food security and infrastructure.
In the past long-running volcanic eruptions have produced huge volumes of basalt lava, “large igneous provinces” (up to several million cubic kilometres in less than a million years). These events may have triggered mass extinction through global climate effects and releases of toxic and ozone-depleting gas (Whiteside et al., 2010; Sobolev et al., 2011). Such events are rare, with the geologic record suggesting return times of ten to a hundred million years.
Smaller (hundreds to thousands of cubic kilometres of ejected material) eruptions are more common, with a return time of approximately 1.4 million years for >1000 km3 supervolcanism and hundreds to tens of thousands of years for >100 km2 eruptions (Mason, Pyle, & Oppenheimer, 2004). Such eruptions devastate large areas by pyroclastic flows and deposit ash-falls over continent-sized areas that would disrupt human society. In addition, the sulphuric acid aerosols entering the stratosphere would cause ozone depletion and global cooling lasting for years, which would reduce agricultural yields (Self, 2006; Harris, 2008). It has even been suggested that super-eruptions represent a limiting factor for the longevity of technological civilizations (Rampino, 2002).
One reason for the interest in supervolcanism is the evidence that H. sapiens has passed through one or more genetic bottlenecks reducing genetic diversity, suggesting that the population was at some point low (possibly as low as in the hundreds [Relethford & Jorde, 1999]). This has been explained as a near-extinction due to a long-lasting volcanic winter resulting from the eruption of Toba ca. 73,000 years ago (Ambrose, 1998). The theory remains controversial due to inconsistent survival and genetic diversity of other primate species closer to the eruption (Oppenheimer, 2011), as well as mixed results in population genetics (Templeton, 2015).
Climate has a profound effect on where and how humans can survive, and natural climate change poses a set of potential extinction risks. These include concerns about an end to the interglacial returning earth to its ice-age state, global droughts, as well as global warming. The threat is primarily to food security rather than direct temperature effects, flooding or extreme weather events. Sudden cooling can occur for a variety of reasons and can threaten agriculture globally (Engvild, 2003). Since food stocks at present are smaller than yearly production more than a year of agricultural interruption poses a serious threat to the survival of most of the population.
In the past climate change has likely threatened the early H. sapiens by long periods of drought or ecological change, typically driven by the glacial cycles. Cooling periods have been associated with population contractions and extinctions (Foley, 1994; Gamble et al., 2004). Still, given human survival through these cycles they have been insufficient to cause species extinction.
Typical discussions of anthropogenic climate change focus on scenarios with a few °C of warming, because such scenarios are high probability, they are near-term and can be analyzed using current data and models. Such “normal” climate change (comparable to past temperature change during the Holocene [Marcott et al., 2013]) is expected to affect food security but not in a radical way. An increase in climate related extremes such as floods, droughts, cyclones, and wildfires can be expected but would do mainly localized damage. Reductions in agricultural productivity and water scarcity are to be expected. Effects depend on the vulnerability profile of different regions but are overall negative (Schmidhuber & Tubiello, 2007; IPCC, 2014). Still, despite serious local problems the overall situation appears to pose no major risk to the survival of the species.
Large temperature increases (4+ °C) would affect ecosystems at the same level as human land-use changes, cause substantial increase in extinction of non-human species, and increase the probability of large-scale tipping points (when a smooth change of parameters leads to an abrupt qualitative change in behavior) such as disintegrating ice sheets, methane release from undersea deposits (clathrates), and long-term droughts in some areas that act as reinforcing feedbacks for a changed climate and ecosystem. At this point crop yields can be expected to drop on the order of 20% or more (depending on crop and location), outdoor activity in the tropics becomes impaired, and global food security becomes less stable (IPCC, 2014). While this could seriously stress human societies or force a smaller population it is again not per se an extinction risk.
However, little is known about tail risks corresponding to higher temperature increases; uncertainty about climate sensitivity and future emissions is compounded by the possibility of positive feedbacks producing significantly more warming. Wagner and Weitzman (2016) argued, based on IPCC data, that depending on emission scenario there is a 3% to 10% risk of a 6+ °C increase. At this point large regions would be too warm for unprotected humans to survive in, and beyond 11–12 °C warming this would encompass most currently inhabited regions (King et al., 2015; Sherwood & Huber, 2010). While this does not ensure extinction, the global population would have a constrained and vulnerable habitat.
An even more extreme scenario would be causing a runaway greenhouse effect sterilizing earth. However, this is unlikely to be possible through adding greenhouse gases (Goldblatt & Watson, 2012). Conversely, triggering a snowball earth state (where the surface is nearly entirely frozen and the ice and snow maintains the low temperature by reflecting sunlight into space effectively) would require a 10% reduction in solar input, or a more modest reduction plus drastically lower CO2 levels (Yang, Peltier, & Hu, 2012). Although the biosphere is expected to destabilize eventually (dooming any terrestrial species) the expected lifespan is on the order of 1.6–2 billion years. The cause of extinction is thought to be the increasing solar luminosity, making conditions too hot, combined with CO2 levels becoming too low for photosynthesis, or water loss to space (Franck, Bounama, & von Bloh, 2006; Wolf & Toon, 2015).
Human populations are subject to natural pandemics where “new” pathogens spread across large areas. These can often be lethal: the Black Death killed 72–200 million people in Eurasia (18%–50% of the world’s population), with a mortality rate ranging from 20%–60% in different locations (Benedictow, 2004; Ziegler, 2013). The Columbian exchange of diseases between the New and Old world led to a catastrophic population decline in the Americas (Alfani & Murphy, 2017). The 1918 flu pandemic resulted in 50–100 million deaths (Johnson & Mueller, 2002) (2.7%–5% of the world’s population). Emerging diseases are likely to cause pandemics in the future, and this may contribute to extinction.
It may appear unlikely that a pathogen could cause extinction of its host, since lack of hosts would naturally limit the pathogen population. However, this does not apply if the pathogen has a reservoir in another host species. A pathogen with a reservoir species that acts as a stable carrier state for the pathogen, a high potential for infecting susceptible species (especially critical age groups), and hyperlethality (mortality in the range of 50%–75%) may cause repeated outbreaks that gradually reduces the fitness of the species until eventually it succumbs to other random environmental events (MacPhee & Greenwood, 2013). Amphibian chytridiomycosis may be a good example, where the pathogen has a large host range but is lethal to particular species (Berger et al., 2016). On the other hand, Tasmanian devil facial tumor disease may cause an extinction in an already threatened species because both a low density threshold and low genetic diversity among the hosts makes all individuals susceptible (McCallum, 2012).
As an example, avian influenza H5N1 represents a recognized potential pandemic that could cause massive damage if a mutation would add the ability of easy transmissibility between humans (at present human infections are not transmissible). It also exists within a large pool of bird hosts. Given that past influenza pandemics have infected between 24%–38% of the population, H5N1 has a case fatality rate ranging from 1% to 60%, a rough estimate suggest that a pandemic could kill between 16.8 million and 1.7 billion people (Cotton-Barratt et al., 2016). A pandemic flu model estimated 21–33 million deaths globally for a modern re-run of the 1918 flu but noted that it did not represent a worst case scenario (Madhav, 2013). While unlikely to be an extinction threat on its own, it could clearly weaken a vulnerable population.
Viruses with longer incubation times, higher infectiousness, and case fatality rates are known: while a super-pandemic combining all these properties may be unlikely, it does not seem biologically impossible. In addition, deliberate alterations of different viruses have successfully increased transmissibility and lethality, or reduced treatability: the major biological extinction risk may be deliberately engineered pathogens rather than natural.
Localized disasters or slow-moving risks are unlikely on their own to spell doom for H. sapiens. It may appear that an unlikely intense global event or confluence of disasters need to occur in order to cause extinction. However, many risks are potentially systemic: a sequence or combination of disasters may reduce resiliency and the ability to recover, especially when interacting with the human systems.
A model of how compound risks can act is the synchronous failure model of Homer-Dixon et al. (2015). Multiple stresses (such as climate change, resource shortages, or conflicts) can interact and accumulate in a social-ecological system, pushing it toward a state where its coping capacity is diminished. Different subsystems become coupled because they require support from each other to function in the stressed state. When a crisis occurs (either externally triggered or because an internal component finally fails) it rapidly cascades through the system, spreading between subsystems and causing the whole to fail. Simultaneous damage is often multiplicative in severity.
Many human systems such as food, energy, finance and communications are global, densely interconnected systems where failures can cascade rapidly (Helbing, 2013). They have developed in a locally rational way: the gains in efficiency and reliability have been significant. However, the probability of global failures also has increased compared to more local, modular and redundant systems (Goldin & Vogel, 2010). While societal collapse does not imply extinction, humans are dependent on complex societies and their high productivity, and any long-term collapse would reduce the human carrying capacity significantly.
A stressor such as climate change may increase the probability and severity of global failure, and once this occurs vulnerability to further risks increases. Various example scenarios can be constructed where plausible events produce gradual deterioration of the human system before it can recover; see, for example, Tonn and MacGregor (2009) and other papers in the same issue.
Another example is sudden geoengineering cessation. If, as a response to climate change, solar radiation management geoengineering is used to maintain temperature, this will require ongoing technological maintenance. If a global disaster disrupts civilization, besides the damage from the primary disaster there would also be a rapid temperature change to close to what the un-modified climate would have been. This will likely produce massive disruptions of agriculture and other human systems at the time when vulnerability is maximal (Baum, Maher, & Haqq-Misra, 2013). In this case a risk mitigation effort adds to systemic risk.
Systemic effects are hard to predict (trade can both strengthen human societies by providing an adaptive system of distribution, prosperity, and incentives for innovation as well as destabilize them due to market bubbles, dependencies, and spread of pathogens). Taking uncertainty into account is possible but tends to lead to conservative policies (Weitzman, 2009). Another approach is to engineer human systems so they are naturally redundant, modular, and otherwise resilient to systemic stresses (Helbing, 2013).
Estimating existential risks can be done in many ways, each with their own merits and drawbacks; see (Tonn & Stiefel, 2013) for a review.
It is possible to place upper bounds on extinction risks due to natural disasters by considering the fossil record. This can be done in several ways; the following will be based on the work of Toby Ord (2017). The simplest bound is based on the observation that H. sapiens has existed for 200,000 years: this observation would be unlikely if the extinction risk was higher than about 1 in 3,000 per century. One can say that an extinction rate of 0.15% or higher per century is ruled out at a 95% confidence level.
Another bound uses now-extinct related hominin species as a reference class, producing estimates in the range 0.001% to 0.05% per century. This is in line with survival times for mammalian species, which typically is 1–2 million years (Raup, 1978) but shorter than for the entire fossil record where lifetimes of 5–10 million years are typical (Raup, 1986; May, Lawton, & Stork, 1995).
H. sapiens is an unusually populous, well-dispersed, and adaptable large mammal species. However, it also has high food requirements and a long generation time. It may then be that the most likely risk to lead to extinction would be a mass-extinction level risk. Large mass extinctions occur at a rate of about 1 in 100 million years, producing a risk estimate of 0.0001% per century.
One issue is that we are still discovering new kinds of existential risks. As noted above, supernovas have been recognized as a risk since the 1950s but gamma ray bursts were recognized as a risk first in the 1990s. High-energy physics risks were suggested in 1970s and later. Recognition of supervolcanism as a risk dates to the 1990s, in turn based on the models of nuclear winter in the 1980s. “Big rip” early endings of the universe were noticed in 2003 (Caldwell, Kamionkowski, & Weinberg, 2003). Since the rate of discovery does not seem to have slacked off, it is plausible that more natural hazards exist that we are unaware of, yet could pose a threat. At the same time, the above estimates bound the total risk: we are merely refining our understanding of what hazard categories exist.
It should be noted that using past geological or fossil records to estimate risks that could have influenced the emergence of the species doing the risk estimation requires some care: risks that would have precluded the emergence of the species would naturally be underrepresented (Ćirković, Sandberg, & Bostrom, 2010). It is also clear that the peculiarities of the current situation may exacerbate some risks (e.g., pandemics) while reducing others (e.g., local disasters); these estimates merely show the risk magnitude for the earlier stages of the species’ history. The current probability is dynamically changing depending on human action.
Probability estimates are on their own irrelevant: the point of risk assessment is to motivate rational risk management. This includes prioritizing mitigation efforts (typically toward the largest, most urgent, and most controllable risks) and research to reduce uncertainty and find more options.
Human extinction is an unusual risk since it can only occur once. Mitigation efforts need to succeed every time.
Mitigating extinction risk can be done by reducing the probability of sufficiently severe hazards occurring, improving resilience mechanisms to reduce the damage, and endurance mechanisms to ensure that survivors can rebuild and repopulate.
Many astrophysical extinction risks, supervolcanism and the emergence of new diseases are likely impossible to prevent, requiring resilience strategies. Impacts from near earth objects or comets can in principle be prevented given enough lead time and the right technological level (NRC, 2010). The amount of impulse needed to avoid an earth collision scales inversely with the lead time and proportional to the impactor mass: with enough time, even a high-precision weak intervention can move large objects. Managing atmospheric emissions and possibly intervening with geoengineering can influence climate risks (Wigley, 2006; Moreno-Cruz & Keith, 2013). Human systems can be designed to be resistant to various forms of systemic risks (Helbing, 2013).
Prediction of extreme events is often impossible since they are the outcome of cascades in noisy, chaotic systems with hidden variables, and past data of less extreme cases often does not constrain models of phenomena of this magnitude. This requires using robust strategies taking large uncertainty into account (Weitzman, 2009). Although exact prediction may not be possible, rapid and improved response is possible and can enhance the resiliency against many of the listed threats. This includes better risk surveillance, preparation of responses and resources, as well as intergovernmental coordination.
Many extinction risks have joint pathways. For example, supervolcanism, large meteor impacts, and nuclear winters (not discussed in this article) do most of their harm by precluding agricultural/fishing over a span of years leading to widespread starvation (Engvild, 2003). While they also cause other harms this particular shared pathway can be dealt with by emergency food stores or alternative food sources (Denkenberger & Pearce, 2014). Shielding in space against radiation sources could in principle mitigate the risk from supernovas, GRBs, superflares, and similar risks (Ćirković & Vukotić, 2016). Improved resiliency against particular damage pathways can hence improve chances against a large set of risks.
Endurance mechanisms aim at ensuring survival, adaptation, and eventual recovery after a near-extinction disaster (Maher & Baum, 2013). An occasionally suggested endurance mechanism against extinction risks is the deliberate construction of refuges where people can survive (or the encouragement of natural refuges in isolated regions, nuclear submarines etc.). Ideally such refuges would be self-sufficient and independent of the earth’s surface (Baum, Denkenberger, & Haqq-Misra, 2015; Jebari, 2015). However, refuges only help against certain categories of disasters and their cost-effectiveness depends on the relative value of current and future generations (Beckstead, 2015).
Undersupply of Mitigation
Preventing extinction is important; at least as important as saving the lives of 7.2 billion people, and quite possibly far more important when taking future generations and their value into account (Parfit, 1984; Bostrom, 2003; Bostrom, 2013; Häggström, 2016).
Mitigating extinction risk is an undersupplied global public good. For example, traditional statistical life valuations suggest that a $16–$32 billion annual investment in asteroid defense would be cost-effective yet U.S. government spending on asteroid detection (with no mitigation) is around $4 million per year, orders of magnitude smaller than funding for hazardous waste sites per unit of risk (Gerrard, 2000; Matheny, 2007). The annual cost to the world due to pandemic influenza has been estimated to $570 billion per year or 0.7% of global income, comparable to estimates of the long-term costs of climate change (Fan, Jamison, & Summers, 2016): the global influenza vaccine market has been estimated to less than $4 billion per year (Kaddar, 2013). These estimates merely take lives saved into account, not the value of future generations.
Since existential risk mitigation is non-excludable and non-rivalrous there is a free-rider problem (non-participants gain the benefit without having to pay) and each producer of risk reduction would only gain a fraction of the total benefit. This is amplified by the transgenerational nature of risk reduction: most of the benefit will accrue to future generations. In principle the value to them of our present preventing extinction is near-infinite, but they cannot pay us any compensation (Matheny, 2007; Bostrom, 2013).
Beside the normal logic of undersupply and lack of global coordination mechanisms there are also cognitive and cultural factors making existential risk mitigation rare. Part of the problem may be discomfort with the topic leading to willful denial or ignorance (Epstein & Zhao, 2009). Part of the problem is the difficulty to fit the topic with human cognitive biases (Yudkowsky, 2008; Wiener, 2016). Humans have heuristics that provide quick and adequate answers for many situations but lead to systematic biases in many situations removed from our ancestral everyday ones. For example, since extinction has not occurred in the past, the availability heuristic (“probabilities of events are roughly proportional to how easy examples of past events come to mind”) will underestimate likelihood. Scope neglect makes us relatively insensitive to the number of lives affected, making the willingness to make an effort scale sublinearly with the size of the problem. In general, without rich context information people are generally bad at judging differences between low probability events (Kunreuther, Novemsky, & Kahneman, 2001).
Risks are judged not just by probability and severity but also by psychological aspects such as outrage and dread (Slovic, 1987). This can sometimes support efforts to mitigate global risks (since they tend to score highly on dread) but makes the focus strongly dependent on what is and is not discussed in public (Yudkowsky, 2008). This makes constructing risk management strategies that are resistant to behavioral biases vitally important for extreme risks (Kunreuther & Heal, 2012; Wiener, 2016).
There is clear evidence that natural events could cause the extinction of H. sapiens. While astronomical risks may be the most dramatic, geophysical risks to food security and pathogenic risks appear to be more significant. It is unlikely that a single disaster will be severe enough to directly cause extinction, but it is plausible that it could place the species in a vulnerable situation for a long time, during which other risks could lead to further vulnerability and extinction.
The overall probability of H. sapiens going extinct due to known natural disasters in the foreseeable future is relatively low. However, this is not a reason for complacency. First, new kinds of risks have been discovered in recent decades: there is a fair probability that some important risk categories have not yet been discovered. Second, many of the natural disaster risks interact strongly with human systems in such a way that they may act as triggers or components in larger disasters where human activity contributes to lethality.
Third, the total non-anthropogenic risk to the species is likely far smaller than anthropogenic risk (Cotton-Barratt et al., 2016). These risks, from nuclear war over bioweapons to runaway technologies, can develop quickly and are integrated in the techno-social systems of the species. While this means they can also be mitigated through human actions the complexity of doing so in the face of economic, national, and other interests can be daunting. The irony is that many of these risks are emerging as a result of increasing technological and economic sophistication that helps mitigate many non-anthropogenic risks.
Extinction risks can be caused by a broad spectrum of hazards, ranging from well understood to speculative. There are profound uncertainties attached to these hazards and handling them requires expertise from a range of different disciplines. As a result, there is a strong need for interdisciplinary investigation of extinction-level risk in order to find better mitigation policies.
- Bostrom, N., & Ćirković, M. M. (2011). Global catastrophic risks. Oxford: Oxford University Press.
- Häggström, O. (2016). Here be dragons: Science, technology and the future of humanity. Oxford: Oxford University Press.
- Leslie J. (1998). The end of the world: The science and ethics of human extinction. London: Routledge.
- Rees M. (2003). Our final century: Will civilisation survive the 21st century? London: Arrow.
- Alfani, G., & Murphy, T. E. (2017). Plague and lethal epidemics in the pre-industrial world. Journal of Economic History, 77(1), 314–343.
- Alvarez, L. W., Alvarez, W., Asaro, F., & Michel, H. V. (1980). Extraterrestrial cause for the Cretaceous–Tertiary extinction. Science, 208 (4448), 1095–1108.
- Amanat, A. & Bernhardsson, M. (2002). Imagining the end: Visions of apocalypse from the ancient Middle East to modern America. London: I. B. Tauris.
- Ambrose, S. H. (1998). Late Pleistocene human population bottlenecks, volcanic winter, and differentiation of modern humans. Journal of Human Evolution, 34(6), 623–651.
- Bainbridge, W. S. (2009). Demographic collapse. Futures, 41(10), 738–745.
- Bartholomew, R. E., & Radford, B. (2011). The Martians have landed! A history of media-driven panics and hoaxes. Jefferson, NC: McFarland.
- Baum, S. D., Maher, T. M., & Haqq-Misra, J. (2013). Double catastrophe: Intermittent stratospheric geoengineering induced by societal collapse. Environment Systems & Decisions, 33(1), 168–180.
- Baum, S. D., Denkenberger, D. C., & Haqq-Misra, J. (2015). Isolated refuges for surviving global catastrophes. Futures, 72, 45–56.
- Beckstead, N. (2015). How much could refuges help us recover from a global catastrophe? Futures, 72, 36–44.
- Beech, M. (2011). The past, present and future supernova threat to earth’s biosphere. Astrophysics and Space Science, 336(2), 287–302.
- Benedictow, O. J. (2004). The Black Death, 1346–1353: The complete history. Martlesham, U.K.: Boydell & Brewer.
- Berger, L., Roberts, A. A., Voyles, J., Longcore, J. E., Murray, K. A., & Skerratt, L. F. (2016). History and recent progress on chytridiomycosis in amphibians. Fungal Ecology, 19, 89–99.
- Bostrom, N. (2002). Existential risks. Journal of Evolution and Technology, 9(1), 1–31.
- Bostrom, N. (2003). Astronomical waste: The opportunity cost of delayed technological development. Utilitas, 15(03), 308–314.
- Bostrom, N. (2004). The future of human evolution. Death and anti-death: Two hundred years after Kant, fifty years after Turing. Ann Arbor, MI: Ria University Press, 339–371.
- Bostrom, N. (2009). The future of humanity. In E. Selinger & S. Riis (Eds.), New waves in philosophy of technology (pp. 186–215). Basingstoke, U.K.: Palgrave Macmillan.
- Bostrom, N. (2013). Existential risk prevention as global priority. Global Policy, 4(1), 15–31.
- Bostrom, N., & Ćirković, M. M. (2011). Global catastrophic risks. Oxford: Oxford University Press.
- Brown, P., Spalding, R. E., ReVelle, D. O., Tagliaferri, E., & Worden, S. P. (2002). The flux of small near-earth objects colliding with the earth. Nature, 420(6913), 294.
- Brugger, J., Feulner, G., & Petri, S. (2017). Baby, it’s cold outside: Climate model simulations of the effects of the asteroid impact at the end of the Cretaceous. Geophysical Research Letters, 441, 419–427.
- Caldwell, R. R., Kamionkowski, M., & Weinberg, N. N. (2003). Phantom energy: Dark energy with w<− 1 causes a cosmic doomsday. Physical Review Letters, 91(7), 071301.
- Chapman, C. R. (2003). How a near-earth object impact might affect society. In Commissioned paper presented at OECD Global Science Forum for “Workshop on Near Earth Objects: Risks, Policies, and Actions,” Frascati, Italy.
- Ćirković, M. M., Sandberg, A., & Bostrom, N. (2010). Anthropic shadow: Observation selection effects and human extinction risks. Risk analysis, 30(10), 1495–1506.
- Ćirković, M. M., & Vukotić, B. (2016). Long-term prospects: Mitigation of supernova and gamma-ray burst threat to intelligent beings. Acta Astronautica, 129, 438–446.
- Cotton-Barratt, O., Farquhar, S., Halstead, J., Schubert, S., Snyder-Beattie, A. (2016). Global Catastrophic Risks 2016. Global Challenges Foundation and the Global Priorities Project. Retrieved from http://globalprioritiesproject.org/wp-content/uploads/2016/04/Global-Catastrophic-Risk-Annual-Report-2016-FINAL.pdf
- Dar, A., De Rújula, A., & Heinz, U. (1999). Will relativistic heavy-ion colliders destroy our planet? Physics Letters B, 470(1), 142–148.
- Denkenberger, D., & Pearce, J. M. (2014). Feeding everyone no matter what: Managing food security after global catastrophe. San Diego, CA: Academic Press.
- Ehrlich, D. (1968). The population bomb. New York: Ballantine.
- Ellis, J., Giudice, G., Mangano, M., Tkachev, I., & Wiedemann, U. (2008). Review of the safety of LHC collisions. Journal of Physics G: Nuclear and Particle Physics, 35(11), 115004.
- Ellis, J., & Schramm, D. N. (1995). Could a nearby supernova explosion have caused a mass extinction? Proceedings of the National Academy of Sciences, 92(1), 235–238.
- Engvild, K. C. (2003). A review of the risks of sudden global cooling and its effects on agriculture. Agricultural and Forest Meteorology, 115(3), 127–137.
- Epstein, R. J., & Zhao, Y. (2009). The threat that dare not speak its name: Human extinction. Perspectives in biology and medicine, 52(1), 116–125.
- ESHRE Capri Workshop Group. (2001). Social determinants of human reproduction. Human Reproduction, 16(7), 1518–1526.
- Fan, V. Y., Jamison, D. T., & Summers, L. H. (2016). The inclusive cost of pandemic influenza risk (No. w22137). Cambridge, MA: National Bureau of Economic Research. Retrieved from http://www.un-influenza.org/sites/default/files/Inclusive_cost_NBER_Working_Paper.pdf.
- Franck, S., Bounama, C., & Von Bloh, W. (2006). Causes and timing of future biosphere extinctions. Biogeosciences, 3(1), 85–92.
- Foley, R. A. (1994). Speciation, extinction and climatic change in hominid evolution. Journal of Human Evolution, 26(4), 275–289.
- Galante, D., & Horvath, J. E. (2007). Biological effects of gamma-ray bursts: Distances for severe damage on the biota. International Journal of Astrobiology, 6(01), 19–26.
- Gamble, C., Davies, W., Pettitt, P., & Richards, M. (2004). Climate change and evolving human diversity in Europe during the last glacial. Philosophical Transactions of the Royal Society B: Biological Sciences, 359(1442), 243–254.
- Gehrels, N., Laird, C. M., Jackman, C. H., Cannizzo, J. K., Mattson, B. J., & Chen, W. (2003). Ozone depletion from nearby supernovae. Astrophysical Journal, 585(2), 1169.
- Gerrard, M. B. (2000). Risks of hazardous waste sites versus asteroid and comet impacts: Accounting for the discrepancies in US resource allocation. Risk Analysis, 20(6), 895–904.
- Goldblatt, C., & Watson, A. J. (2012). The runaway greenhouse: Implications for future climate change, geoengineering and planetary atmospheres. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 370(1974), 4197–4216.
- Goldin, I., & Vogel, T. (2010). Global governance and systemic risk in the 21st century: Lessons from the financial crisis. Global Policy, 1(1), 4–15.
- Harris, A. (2008). What spaceguard did. Nature, 453(7199), 1178–1179.
- Harris, A. W., & D’Abramo, G. (2015). The population of near-earth asteroids. Icarus, 257, 302–312.
- Harris, B. (2008). The potential impact of super-volcanic eruptions on the earth’s atmosphere. Weather, 63(8), 221.
- Helbing, D. (2013). Globally networked risks and how to respond. Nature, 497(7447), 51–59.
- Hern, W. M. (1990). Why are there so many of us? Description and diagnosis of a planetary ecopathological process. Population & Environment, 12(1), 9–39.
- Homer-Dixon, T., Walker, B., Biggs, R., Crépin, A. S., Folke, C., Lambin, E., et al. (2015). Synchronous failure: the emerging causal architecture of global crisis. Ecology and Society, 20(3).
- Horvath, J. E., & Galante, D. (2012). Effects of high-energy astrophysical events on habitable planets. International Journal of Astrobiology, 11(4), 279.
- Häggström, O. (2016). Here be dragons: Science, technology and the future of humanity. Oxford: Oxford University Press.
- Intergovernmental Panel on Climate Change. (2014). Climate Change 2014–Impacts, adaptation and vulnerability: Regional aspects. Cambridge, U.K.: Cambridge University Press.
- Jebari, K. (2015). Existential risks: Exploring a robust risk reduction strategy. Science and Engineering Ethics, 21(3), 541–554.
- Johnson, N. P., & Mueller, J. (2002). Updating the accounts: Global mortality of the 1918–1920 “Spanish” influenza pandemic. Bulletin of the History of Medicine, 76(1), 105–115.
- Joy, B. (2000, April) Why the future doesn’t need us. WIRED.
- Kaddar, M. (2013). Global vaccine market features and trends. Presentation at Workshop on Business Modeling for Sustainable Influenza Vaccine Manufacturing. Washington, DC, January 14–16, 2013. Retrieved from http://www.who.int/influenza_vaccines_plan/resources/session_10_kaddar.pdf
- King, D., Schrag, D., Dadi, Z., Ye, Q., & Ghosh, A. Climate change: A risk assessment. Cambridge, U.K.: Cambridge Centre for Science and Policy, 2015.
- Krasovsky, V. I., & Shklovsky, I. S. (1957). Supernova explosions and their possible effect on the evolution of life on the earth. Proceedings of the USSR Academy of Sciences 116, 197–199.
- Kunreuther, H., & Heal, G. (2012). Managing catastrophic risk (No. w18136). Cambridge, MA: National Bureau of Economic Research. Retrieved from http://www.nber.org/papers/w18136.pdf.
- Kunreuther, H., Novemsky, N., & Kahneman, D. (2001). Making low probabilities useful. Journal of Risk and Uncertainty, 23(2), 103–120.
- Kuznick, P. J. (2007). Prophets of doom or voices of sanity? The evolving discourse of annihilation in the first decade and a half of the nuclear age. Journal of Genocide Research, 9(3), 411–441.
- Leslie J. (1998). The end of the world: The science and ethics of human extinction. Routledge: London.
- Lutz, W. (2009). Editorial: Towards a world of 2–6 billion well‐educated and therefore healthy and wealthy people. Journal of the Royal Statistical Society: Series A (Statistics in Society), 172(4), 701–705.
- Lutz, W., & Qiang, R. (2002). Determinants of human population growth. Philosophical Transactions of the Royal Society B: Biological Sciences, 357(1425), 1197.
- MacPhee R. D. E., & Greenwood, A. D. (2013). Infectious disease, endangerment, and extinction. International Journal of Evolutionary Biology, 2013, 1–9.
- Madhav, N. (2013, February 21) Modeling a Modern-Day Spanish Flu Pandemic. AIR Currents. Retrieved from http://www.air-worldwide.com/Publications/AIR-Currents/2013/Modeling-a-Modern-Day-Spanish-Flu-Pandemic/
- Maher, T. M., & Baum, S. D. (2013). Adaptation to and recovery from global catastrophe. Sustainability, 5(4), 1461–1479.
- Malamud, B. D. (2004). Tails of natural hazards. Physics World, 17(8), 25.
- Marcott, S. A., Shakun, J. D., Clark, P. U., & Mix, A. C. (2013). A reconstruction of regional and global temperature for the past 11,300 years. Science, 339(6124), 1198–1201.
- Mason, B. G., Pyle, D. M., & Oppenheimer, C. (2004). The size and frequency of the largest explosive eruptions on Earth. Bulletin of Volcanology, 66(8), 735–748.
- Matheny J. G. (2007). Reducing the risk of human extinction. Risk Analysis, 27(5):1335–1344.
- May, R. M., Lawton, J. H., & Stork, N. (1995). Assessing extinction rates. In J. H. Lawton & R. M. May (Eds.), Extinction rates. Oxford: Oxford University Press.
- McCallum, H. (2012). Disease and the dynamics of extinction. Philosophical Transactions of the Royal Society B, 367(1604), 2828–2839.
- Meadows, D. H., Meadows, D. H., Randers, J., & Behrens, W. W., III. (1972). The limits to growth: A report to the club of Rome (1972). New York: Universe.
- Melott, A. L., Lieberman, B. S., Laird, C. M., Martin, L. D., Medvedev, M. V., Thomas, B. C., et al. (2004). Did a gamma-ray burst initiate the late Ordovician mass extinction? International Journal of Astrobiology, 3(01), 55–61.
- Melott, A. L., & Thomas, B. C. (2011). Astrophysical ionizing radiation and earth: A brief review and census of intermittent intense sources. Astrobiology, 11(4), 343–361.
- Moreno-Cruz, J. B., & Keith, D. W. (2013). Climate policy under uncertainty: A case for solar geoengineering. Climatic Change, 121(3), 431–444.
- National Research Council. (2010). Defending planet earth: Near-Earth-Object surveys and hazard mitigation strategies. Washington, DC: National Academies Press.
- Oppenheimer, C. (2011). Eruptions that shook the world. Cambridge, U.K.: Cambridge University Press.
- Ord, T. (2017, forthcoming). Bounding the natural risks of human extinction.
- Ord, T., Hillerbrand, R., & Sandberg, A. (2010). Probing the improbable: Methodological challenges for risks with low probabilities and high stakes. Journal of Risk Research, 13(2), 191–205.
- Parfit, D. (1984). Reasons and persons. Oxford: Oxford University Press.
- Pierazzo, E., & Artemieva, N. (2012). Local and global environmental effects of impacts on Earth. Elements, 8(1), 55–60.
- Posner, R. (2004). Catastrophes, risk and resolution. Oxford: Oxford University Press.
- Rampino, M. R. (2002). Supereruptions as a threat to civilizations on earth-like planets. Icarus, 156(2), 562–569.
- Randle, M., & Eckersley, R. (2015). Public perceptions of future threats to humanity and different societal responses: A cross-national study. Futures, 72, 4–16.
- Raup DM (1978). Cohort analysis of generic survivorship. Paleobiology, 4, 1–15.
- Raup, D. M. (1986). Biological extinction in earth history. Science, 231, 1528–1534.
- Reed, D. H., O’Grady, J. J., Brook, B. W., Ballou, J. D., & Frankham, R. (2003). Estimates of minimum viable population sizes for vertebrates and factors influencing those estimates. Biological Conservation, 113(1), 23–34.
- Rees M. (2003). Our final century: Will civilisation survive the 21st century? London: Arrow.
- Relethford, J. H., & Jorde, L. B. (1999). Genetic evidence for larger African population size during recent human evolution. American Journal of Physical Anthropology, 108(3), 251–260.
- Ruderman, M. A. (1974). Possible consequences of nearby supernova explosions for atmospheric ozone and terrestrial life. Science, 184(4141), 1079–1081.
- Scalo, J., & Wheeler, J. C. (2002). Astrophysical and astrobiological implications of gamma-ray burst properties. Astrophysical Journal, 566(2), 723.
- Schmidhuber, J., & Tubiello, F. N. (2007). Global food security under climate change. Proceedings of the National Academy of Sciences, 104(50), 19703–19708.
- Self, S. (2006). The effects and consequences of very large explosive volcanic eruptions. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 364(1845), 2073–2097.
- Sherwood, S. C., & Huber, M. (2010). An adaptability limit to climate change due to heat stress. Proceedings of the National Academy of Sciences, 107(21), 9552–9555.
- Sleep, N. H., Zahnle, K. J., Kasting, J. F., & Morowitz, H. J. (1989). Annihilation of ecosystems by large asteroid impacts on the early earth. Nature, 342(6246), 139–142.
- Slovic, P. (1987). Perception of risk. Science, 236(4799), 280–285.
- Smith, J. M. (1989). The causes of extinction. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 325(1228), 241–252.
- Sobolev, S. V., Sobolev, A. V., Kuzmin, D. V., Krivolutskaya, N. A., Petrunin, A. G., Arndt, N. T., et al. (2011). Linking mantle plumes, large igneous provinces and environmental catastrophes. Nature, 477(7364), 312–316.
- Stokes, G. H., Yeomans, D. K., Bottke, W. F, Chesley, S. R., Evans, J. B., Gold, R. E., et al. (2003). Study to determine the feasibility of extending the search for near-earth objects to smaller limiting diameters. Report of the NASA NEO Science Definition Team.
- Tegmark, M., & Bostrom, N. (2005). Astrophysics: Is a doomsday catastrophe likely? Nature, 438(7069), 754.
- Templeton, A. R. (2015). Population biology and population genetics of Pleistocene hominins. In Handbook of paleoanthropology (pp. 2331–2370). Berlin: Springer.
- Thomas, B. C., & Melott, A. L. (2006). Gamma-ray bursts and terrestrial planetary atmospheres. New Journal of Physics, 8(7), 120.
- Thorsett, S.E. (1995). Terrestrial implications of cosmological gamma-ray burst models. The Astrophysical Journal, 444(L53).
- Tonn, B., & MacGregor, D. (2009). A singular chain of events. Futures, 41(10), 706–714.
- Tonn, B., & Stiefel, D. (2013). Evaluating methods for estimating existential risks. Risk Analysis, 33(10), 1772–1787.
- Toon, O. B., Zahnle, K., Morrison, D., Turco, R. P., & Covey, C. (1997). Environmental perturbations caused by the impacts of asteroids and comets. Reviews of Geophysics, 35(1), 41–78.
- Wagner, G., & Weitzman, M. L. (2016). Climate shock: The economic consequences of a hotter planet. Princeton, NJ: Princeton University Press.
- Walkden, G., & Parker, J. (2008). The biotic effects of large bolide impacts: Size versus time and place. International Journal of Astrobiology, 7(3–4), 209–215.
- Weitzman, M. L. (2009). On modeling and interpreting the economics of catastrophic climate change. Review of Economics and Statistics, 91(1), 1–19.
- Weissman, P. R. (2006). The cometary impactor flux at the earth. Proceedings of the International Astronomical Union, 2(S236), 441–450.
- Whiteside, J. H., Olsen, P. E., Eglinton, T., Brookfield, M. E., & Sambrotto, R. N. (2010). Compound-specific carbon isotopes from Earth’s largest flood basalt eruptions directly linked to the end-Triassic mass extinction. Proceedings of the National Academy of Sciences, 107(15), 6721–6725.
- Wiener, J. B. (2016). The tragedy of the uncommons: On the politics of apocalypse. Global Policy, 7(S1), 67–80.
- Wigley, T. M. (2006). A combined mitigation/geoengineering approach to climate stabilization. Science, 314(5798), 452–454.
- Wolf, E. T., & Toon, O. B. (2015). The evolution of habitable climates under the brightening sun. Journal of Geophysical Research: Atmospheres, 120(12), 5775–5794.
- Yang, J., Peltier, W. R., & Hu, Y. (2012). The initiation of modern ‘Soft Snowball’ and ‘Hard Snowball’ climates in CCSM3. Part I: the influences of solar luminosity, CO2 concentration, and the sea ice/snow albedo parameterization. Journal of Climate, 25(8), 2711–2736.
- Yudkowsky, E. (2008). Cognitive biases potentially affecting judgment of global risks. In N. Bostrom & M. M. Ćirković (Eds.), Global catastrophic risks (pp. 91–119). Oxford: Oxford University Press
- Ziegler, P. (2013). The black death. New York: Faber & Faber.