8

Active Prevention and Deterrence

The workshop’s final session sought to tie together the lessons from the previous systems by addressing approaches to actively prevent and deter unanticipated rare events of major significance. The three speakers in this session were Terik Daly, senior scientist at the Johns Hopkins University Applied Physics Laboratory; Seth Baum, co-founder and executive director of the Global Catastrophic Risk Institute; and Robert Axtell, professor of computational social science at George Mason University. Justin Kasper, deputy chief technology officer for BWX Technologies and a member of the planning committee, moderated the final discussion after the three presentations.

ASTEROID IMPACTS—THINKING ABOUT AN UNCERTAIN THREAT

Large asteroid impacts are rare, but they are a legitimate concern, announced Terik Daly, who illustrated his point with a map of fireballs reported by U.S. government sensors. While most of the asteroids that enter Earth’s atmosphere explode with little consequence other than a bright flash, the 20-m diameter asteroid that exploded over Chelyabinsk, Russia, in 2013 broke glass, caused structural damage across the metropolitan area, and sent a couple thousand people to the hospital. An event of that size, observed Daly, probably occurs every few decades to centuries.

In the early 1900s, an asteroid of diameter 60 to 190 m exploded in the atmosphere with the equivalent energy of 5 megatons of dynamite. This event occurred over an area that was unpopulated at the time, but it would have been a significant event if it had occurred over a population area. An event of this size likely occurs every few centuries to millennia, Daly noted. Going back 65 million years, the Chicxulub asteroid, which was 10 to 15 km in diameter, devastated Earth and led to the demise of the dinosaurs. This size event may occur every few hundred million years.

In thinking about this threat and how to evaluate it, Daly said it is important to remember that the likelihood and consequences of an asteroid impact varies with asteroid size. Small asteroids, about 4 m in diameter, enter Earth’s atmosphere around once a year and are not a concern. Estimates place the number of these small asteroids at around 500 million, with fewer than 0.1 percent discovered. At the other end of the size spectrum lie the four huge asteroids that are 10,000 m or larger in diameter and the 900 or so asteroids between 1,000 and 10,000 m in diameter. These asteroids would cause global devastation, but they are also not a concern because astronomers know where more than 90 percent of these are and know they are not headed to Earth.

The asteroids of primary concern are those that fall between 50 and 1,000 m in diameter because these are large enough to cause regional devastation, they happen frequently enough—every 2,500 to 20,000 years or so—and most of the estimated hundreds of thousands of them remain undiscovered. “We think there are 200,000 asteroids 50 m or so that could come near Earth, and we have found 8 percent of them,” remarked Daly. As an example of what a 50-m asteroid would do, he noted that an asteroid that size collided with Earth in what is now Arizona and left a crater that was 1.2 km wide and 180 m deep. Modeling shows that the devastation from this asteroid impact would have extended as far as 40 km. The fireball produced by the impact would have incinerated everything within about 10 km, killed or wounded all the large animals out to 24 km, and produced hurricane-force winds out to 40 km.¹ If a similar event were to occur over Washington, DC, the entire city would be incinerated in the fireball, with significant casualties throughout the surrounding areas in Maryland and Virginia.

Daly explained that the asteroid threat has been known for some time, and the understanding of that threat has not changed much over the past couple of decades. However, in 2016 NASA established a Planetary Defense Coordination Office, and in 2018 the federal government released a national strategy.² The strategy has five goals:

1. Enhance near-Earth object (NEO) detection, tracking, and characterization capabilities.
2. Improve NEO modeling, predictions, and information integrations.
3. Develop technologies for NEO deflection and disruption missions.
4. Increase international cooperation on NEO preparation.
5. Strengthen and routinely exercise NEO impact emergency procedures and action protocols.

In 2021, the White House released a report on NEO impact threat emergency protocols.³

Addressing the threat of an asteroid collision boils down to three things, explained Daly: find the asteroids, develop ways to prevent an impact or reduce its effects if it cannot be prevented entirely, and decide how to act in the face of massive uncertainties. For the first step, ground-based telescopes around the world are scanning the skies looking for NEOs. Over the past 8 years or so, the number of discoveries per year have plateaued even though new telescopes have come online during that time. “We have reached the limits of what we can do from the ground,” he added. “At this rate, it will take decades before we can find all those asteroids large enough to cause regional devastation.”

To address this, NASA is developing a space telescope to accelerate the hunt for these asteroids. The NEO Surveyor has a launch readiness date of 2026 and is designed to find asteroids 140 m in diameter and larger that come near Earth. Daly explained that 140 m is the threshold above which there would be massive consequences should one strike a major metropolitan area. Currently, astronomers have found 39 percent of these objects, and the mission’s goal is to find 66 percent of them within 5 years of launch, with an aspiration of finding 90 percent of these objects within 10 to 12 years of launch.

He then explained how these numbers are derived and how astronomers know when they have hit these goals. “What we do is look at what we have found compared to what we think is out there,” he said. Estimates of the population of NEOs come from extrapolating from observations of asteroids exploding in the atmosphere and understanding the collisional evolution of the solar system.

In terms of developing ways to prevent an impact, there are a few techniques that might work, with the options depending on the size of the asteroid and how much warning time exists.⁴ For the largest asteroids with years

___________________

¹ D.A. Kring, 2007, Guidebook to the Geology of Barringer Meteorite Crater, Arizona (a.k.a. Meteor Crater), Lunar and Planetary Institute, Houston, TX, http://www.lpi.usra.edu/publications/books/barringer_crater_guidebook.

² National Science and Technology Council, 2016, National Near-Earth Object Preparedness Strategy, Interagency Working Group for Detecting and Mitigating the Impact of Earth-Bound Near-Earth Objects, December, https://www.nasa.gov/sites/default/files/atoms/files/national_near-earth_object_preparedness_strategy_tagged.pdf.

³ National Science and Technology Council, 2021, Report on Near-Earth Object Impact Threat Emergency Protocols, The Interagency Working Group on Near-Earth Object Impact Threat Emergency Protocols, January, https://trumpwhitehouse.archives.gov/wp-content/uploads/2021/01/NEO-Impact-Threat-Protocols-Jan2021.pdf.

⁴ National Research Council, 2010, Defending Planet Earth: Near-Earth-Object Surveys and Hazard Mitigation Strategies, The National Academies Press, Washington, DC.

of warning, a nuclear detonation either on or next to an asteroid could disrupt it enough to change its orbit. This approach should also work for smaller objects without much warning time. The kinetic impact option—taking a spacecraft and slamming it into the asteroid—can be effective over a wide range of warning times. With many decades of warning, a gravity tractor would work. This involves parking a spacecraft next to an asteroid and letting the gravity of the spacecraft over decades slowly change the asteroid’s orbit. If the NEO is small enough or there is not enough warning time for the above options, the option is civil defense involving preparing evacuation and effective response and recovery strategies.

On November 23, 2021, NASA launched the Double Asteroid Redirection Test (DART) mission to determine whether the kinetic impact approach can work. The spacecraft, whose core is about the size of a vending machine, will self-direct itself to crash into the 160-m Dimorphos asteroid that is orbiting around the larger Didymos asteroid. NASA is conducting this test at this double asteroid as a safety precaution, explained Daly, because the end result, if successful, will be to change the orbit of the smaller asteroid around the larger asteroid, not of the larger asteroid around the Sun. A small, shoebox-sized CubeSat called the Light Italian CubeSat for Imaging of Asteroids will fly over the asteroid and witness the impact. The goal is to change the orbital period of Dimorphos, as observed using ground-based telescopes, by about 1 percent.

In terms of deciding how to act in the face of massive uncertainties, Daly discussed a theoretical exercise conducted in 2019 that started with an asteroid discovered in March that had a 0.002 percent chance of impacting Earth in April 2027. Over the next month, additional observations increased the probability of a strike to 1 percent. At this early stage, the most likely thing to happen is that the asteroid does not hit the Earth, but decision makers will struggle to decide what to do to prepare for the unlikely event that it does hit Earth with major consequences. By July 2019, additional observations put the likelihood of impact at 10 percent, but it was not until 17 months later, in December 2021, that the impact probability had risen to 100 percent. At that point, decision makers had to decide how to reroute critical infrastructure and who to evacuate. In this exercise, the decision was made to try to deflect the asteroid, but that ended up being only partially successful. Impact occurred in 2027.

The national plan, noted Daly, provides guidelines on how to make decisions in the face of these large uncertainties based on four factors: how large, how likely, how soon, and how feasible is it to do anything to change the asteroid’s trajectory. Looking at the interplay of these four factors leads to different actions, such as simply cataloging the asteroid, warning the public of an impending impact, preparing emergency responses, launching a mission to study the asteroid and more accurately determine its mass and orbit, or launching a prevention mission.

In closing, Daly added that there are no known asteroids that pose a threat to Earth for at least 100 years. However, he added, “we know we do not know where most of the asteroids are that could cause regional devastation.” As a result, policy makers need to make decisions about how to deal with the threats that acknowledge both of those truths. “We have to allocate resources in a way that recognizes that right now there is no immediate threat, but also that we do not know where the asteroids are,” finished Daly. “Once we do, which hopefully will be in the next decade, we will definitely know whether or not we need to be concerned about asteroids, but until that time, we have to make decisions and allocate resources in a way that addresses both what we know and what we do not know.”

THE CHALLENGES OF ADDRESSING RARE EVENTS AND HOW TO OVERCOME THEM

From a big picture perspective, life on Earth will no longer be able to exist in about a billion years as a result of the Sun’s natural stellar evolution that will have it become larger and warmer, announced Seth Baum. The interesting question, then, is what happens between now and then. There might be good developments such as advanced transformative technology breakthroughs and colonization of space, and there might be extreme catastrophes, such as the collapse of agriculture or industry, that could harm human civilization or lead to human extinction.⁵ Discussions about global catastrophic risk, remarked Baum, focus on those catastrophes that cause harm on that scale. “There is a strong case for attention to this class of risk based on the important role that it can play in the big picture future of human civilization,” he observed.

___________________

⁵ S.D. Baum, S. Armstrong, T. Ekenstedt, O. Häggström, R. Hanson, K. Kuhlemann, M.M. Maas, et al., 2019, “Long-Term Trajectories of Human Civilization,” Foresight 21(1):53–83, https://doi.org/10.1108/FS-04-2018-0037.

Addressing global catastrophic risk, Baum detailed, first requires understanding the possible decision options and how well they would work at avoiding that risk and then putting those options into action. The second step requires either motivating people or institutions to care about these extreme, catastrophic events or achieving action without motivating people to care about human civilization in the distant future.

Identifying and evaluating decision options is a significant challenge because of how rare and extreme catastrophic events are given that modern, global human civilization has never been destroyed. Baum remarked that this is obviously a good thing, but it does create a data problem, and it eliminates the option of making a data-driven, quantitative risk analysis. Fortunately, he noted, this type of analysis is not always needed. For instance, there are scenarios in which a nuclear terrorist attack could result in a nuclear war even if that was not the intention for many of the parties. This might happen if the victimized country misinterprets the attack as one from a rival country or one sponsored by a rival country and launches its own nuclear weapons in what it believes is retaliation, but is in fact, the start of a nuclear war.⁶

This scenario, Baum explained, is perhaps the only way that an extreme global catastrophe would result from nuclear terrorism. As a result, the policy conclusion should be that if there is going to be a nuclear terrorist attack in the future, there should be guardrails in place to ensure it is not an attack from another country so as to not accidentally initiate a nuclear war for the wrong reasons. “This is a policy conclusion that is not sensitive to doing a quantitative risk analysis,” said Baum. “It is just a fairly obvious conclusion from this particular circumstance.”

An example of when a quantitative analysis is needed involves nuclear power. Nuclear power is helpful for reducing greenhouse gas emissions, but at the same time, there are concerns that it can lead to nuclear weapon proliferation, which was the concern regarding Iran’s nuclear program. These two factors require analyzing whether the benefits of reducing greenhouse gas emissions offset the harms related to nuclear weapon proliferation.⁷ Quantitative analysis is important in this case to assess which of these two different effects is the stronger one and point to a clear answer as to whether nuclear power is a good thing or bad thing in any particular circumstance, Baum observed.

Baum presented a case he has previously studied—the use of nuclear explosives for asteroid deflection. In this case, the analytical question is whether the benefits of reducing the risk of an asteroid colliding with Earth offsets any harms that may arise from international security concerns. The conclusion he reached in his work is that while there is not a precise answer, there is much to learn from going through the analysis in terms of the various aspects of the risks and possible policy decisions that can better inform the decision-making process. He pointed out that in doing this research, he was not making any decisions about using nuclear explosives for asteroid deflection.⁸

Baum underlined that his group has been active in characterizing the risk of nuclear war and has developed a model for the probability of it.⁹ The model outlines two paths to nuclear war, one in which there is an intentional first strike, the other involving unintentional or inadvertent nuclear war scenarios, such as the terrorist scenario he previously discussed. To inform the model, Baum and his collaborators compiled data from past nuclear war incidents, including World War II, and a long list of near-miss incidents that went part of the way to nuclear war, such as the Cuban Missile Crisis. These near-miss incidents, he remarked, go part of the way to determining the probability of nuclear war developing from relatively normal conditions to an initial event such as the Cuban Missile Crisis, but they do not fill in the remaining steps from that initial event to the decision to launch nuclear weapons. Nonetheless, this type of model can provide some understanding of the factors that might lead to nuclear war. He noted that he has also created a model for the effects of nuclear war on human population but would not discuss it in the interests of time.¹⁰

___________________

⁶ S.D. Baum, 2018, “Uncertain Human Consequences in Asteroid Risk Analysis and the Global Catastrophe Threshold,” Natural Hazards 94:759–775, https://doi.org/10.1007/s11069-018-3419-4.

⁷ R.H. Socolow and A. Glaser, 2009, “Balancing Risks: Nuclear Energy and Climate Change,” Daedalus 138(4):31–44, https://www.amacad.org/publication/nuclear-energy-climate-change.

⁸ S.D. Baum, 2019, “Risk-Risk Tradeoff of Nuclear Explosives for Asteroid Deflection,” Risk Analysis 39(11):2427–2442, https://doi.org/10.1111/risa.13339.

⁹ S.D. Baum, R. Neufville, and A. Barrett, 2018, “A Model for the Probability of Nuclear War,” Global Catastrophic Risk Institute Working Paper 18-1, https://gcrinstitute.org/papers/042_nuclear-probability.pdf.

¹⁰ S.D. Baum and A.M. Barrett, 2018, “A Model for the Impacts of Nuclear War,” Global Catastrophic Risk Institute Working Paper 18-2, https://gcrinstitute.org/papers/043_nuclear-impacts.pdf.

Returning to the question of how to achieve action to address these types of risks, Baum first discussed the nuclear terrorism scenario. He noted he could make a case for de-emphasizing the risk of nuclear terrorism except when it could lead to nuclear war, and some might find this argument persuasive and turn their focus to nuclear war, while others would not and would remain focused on the risk of nuclear terrorism. When two groups do not have the same motivation to focus on the risk of nuclear war, one approach is to draw attention to the scenario of nuclear terrorism leading to nuclear war. This approach can highlight ways to reduce the risk of an extreme global catastrophe without changing either group’s primary focus.

As a second example, he discussed the trade-offs between the greenhouse gas benefits of Iran’s nuclear power program and the proliferation risk. Here, the trade-off is less clear, and there is no obvious good answer to this question. At the same time, if the goal is to reduce greenhouse gas emissions, there is a long list of other possible actions, including expanding nuclear power in the United States where there is no nuclear weapons proliferation risk since the nation already has nuclear weapons and nuclear power. The same would apply to Russia and China. In fact, there is a project to have the United States and China cooperate to expand nuclear power to offset the use of coal, which he said would be a good development in terms of addressing climate change.

At the same time, there may be individuals who are not concerned about climate change, nuclear weapons, and global catastrophe. Another way of grounding this policy decision would be to emphasize the local benefits of exchanging nuclear power for coal-burning power plants in terms of reducing air pollution. “This is another way that certain actions can be achieved without necessarily having the same sort of motivation to address the rare events.”

DEALING WITH EXTREMES IN ECONOMICS AND FINANCE

Economics and finance, asserted Robert Axtell, have a long and significant history of significant, rare events. Unlike the possibility of satellite collisions or the existence of rare biological species, in human social systems the probability of an extreme event, like a market crash, may depend on people’s beliefs about such an event. This makes forecasting and anticipation inherently different for social systems, compared to natural systems, and notions such as prevention and deterrence may even be problematical for economic systems, in which fluctuations are a common feature of normal operations. While such events cannot be prevented, they can be managed.

The financial world has always lived with large, rare events. The Great Depression was one, as was the period of high inflation and unemployment during the 1970s, the financial crisis of 2007 to 2008, and the economic downturn triggered by the COVID-19 pandemic. These large events are part of distributions of events by size that are not Gaussian in character and are not fully understood, explained Axtell. The lesson here, he revealed, is that extremes are regular features of economic and financial processes.

The economic world, which is where his work focuses, is often dominated by extremes. For example, in the United States, there is one business firm with a million employees and nearly a million firms with one employee. Approximately 60 percent of U.S. private-sector workers are employed by the largest 1 percent of companies, and 80 percent of all people working for businesses are at one of the largest 10 percent of firms. Furthermore, these firms are linked through buyer-supplier networks. What this means is that fluctuations that happen in a few large firms, whether on short or long time scales, can have ripple effects throughout the entire economy. Axtell referred to this as the so-called granular hypothesis, due to Xavier Gabaix, which suggests that fluctuations among firms at the extreme end of the firm size distribution can be amplified macro-economically.¹¹ This holds true, he said, in the broader social world as well, where there are highly skewed distributions of city size, income, and wealth, as well as in the internet and social media worlds, where heavy-tailed distributions are associated with web connectivity, the numbers of friends and followers, and many other network properties.

To illustrate how rare economic and financial events manifest themselves in important ways, Axtell discussed several specific events. As a first obvious case, the stock market crash of 1929 led to the Great Depression and a decade-long loss of economic output. Another case was the tech bubble of 2000 to 2002, which led to a recession and to the Federal Reserve maintaining low interest rates for a long period. The subsequent sub-prime mortgage bubble resulted in nearly $10 trillion of losses to U.S. households.

___________________

¹¹ X. Gabaix, 2011, “The Granular Origins of Aggregate Fluctuations,” Econometrica 79(3):733–772, https://www.jstor.org/stable/41237769.

While these are examples of how financial events can manifest themselves in the real economy and society generally, the opposite can also be true. For example, a decline in the output of silver mines in the early 1890s was partially responsible for the financial panic of 1893, which triggered bank failures in Chicago and New York. Other examples include the oil embargoes of the 1970s that led to dramatic swings in the value of commodities.

All of this is background, noted Axtell, to addressing the question of how to manage in a world with such extremes. He noted that for the first 100 or so years of U.S. history, booms and busts occurred every 4 to 5 years. Since the Great Depression and the empowerment of the Federal Reserve System, the period between recessions grew to 6 to 10 years, aided by various policy tools, like bank account guarantees, which effectively eliminated bank runs by depositors.

Axtell explained that many of these policies act in ex ante fashion—that is, they are actions taken ahead of time based on forecasts of how the economy is likely to respond to specific events and policies. When it comes to modeling social systems and forecasting policy actions, models have to represent how people will respond to policies. To cite one example, models of investors have to include policy maker behavior, and vice-versa, all conditional on how the world is likely to unfold. This back-and-forth between agents in the economy and policies makes rare economic and social events different from naturally occurring rare events.

It is also possible to manage financial extremes in real time, ex post, as when a big event occurs and action must be taken, remarked Axtell. Note that during the Black Monday event of October 19, 1987, stock market circuit breakers kicked in to stop trading temporarily until market panic receded and prices were re-established. Regulators can also intervene when particular firms get in trouble, as happened during the Asian and Russian financial crises of the late 1990s, when Long-Term Capital Management needed to be recapitalized, and then again during the financial crisis of 2008 when regulators bailed out certain banks that were “too big to fail” and arranged sales of troubled institutions. Subsequently, in the early days of high-frequency stock trading there were so-called flash crash events when prices of specific issues fell dramatically for brief periods and then quickly recovered. Such events were managed by the Securities and Exchange Commission (SEC) and Commodities Futures Trading Commission (CFTC) and through trading curbs. In the 1980s, regulators prosecuted bad actors involved in the savings and loan (S&L) crisis, after the fact, and today, the SEC can invalidate transactions it believes occurred for “insider trading,” among other things. Axtell observed that there are a variety of ways to manage these rare events after they occur.

Another way to deal with financial extremes is to design systems to function in new ways in order to avoid extreme outcomes. In one case, the SEC ordered NASDAQ to get rid of trading in eighths and sixteenths of a dollar, hoping to eliminate so-called spread clustering by market-makers, among other objectives. At the time a “digital twin” of the NASDAQ was created in an attempt to forecast what would happen when the NASDAQ market decimalized.

Axtell pointed out that financial markets are in a constant state of evolution, creating growing complexity and meaning such markets can be hard to manage properly. For example, there are now “upstairs” markets in which big firms conduct large block trades without going through the trading floors of normal stock exchanges, meaning such transactions are often invisible to other market participants. Shadow banking is another relatively recent innovation in which financial activities occur among non-bank financial institutions and outside of the purview of normal regulatory authorities. Cryptocurrencies and blockchain capabilities are yet other innovations with the potential to disrupt normal market operations. All these innovations suggest that extreme economic and financial events are almost certainly as likely to manifest themselves in the future as they have in the past, albeit possibly for different reasons.

Beyond financial examples, Axtell observed that societies create institutions to manage rare economic events. Unemployment insurance payments help individual households and keep aggregate demand up during downturns. Bankruptcy laws exist to deal with the extreme event of a company going out of business. When the COVID-19 pandemic hit and restaurant and entertainment employees lost their jobs, a variety of legislation was put in place to help the people most affected.

To conclude his presentation, Axtell discussed whether reaction, mitigation, and adaptation might replace anticipation, prevention, and deterrence in the context of human systems. Modifying systems to prevent a bad outcome, can end up creating a system where there is a high chance that an even bigger failure can occur. As an example, the prevention of small fires in a forest may create the conditions for much larger fires to occur down

the road. “By eliminating a certain class of events, do we effectively guarantee the occurrence of more extreme events?” asked Axtell.

In economics and finance, the idea behind trying to manage a complex system sometimes means doing it from the bottom up. In this approach, social science-savvy, behavioral solutions replace “technical” solutions, as the latter tend to be brittle and narrow, often leading to unintended consequences. For policy, Axtell explained, such unintended effects should be studied as first-order phenomena that may jeopardize the welfare gains produced by the policies in question. In that regard, Axtell emphasized that it is important to consider social welfare functions existing on individual, organizational, and societal levels.

Regarding deterrence, Axtell observed that deterrence and its relationship to mutually assured destruction was center stage during the Cold War, but in asymmetric situations deterrence is a more nuanced concept that takes on a different character. Consider new technologies that, when introduced, are not hardened and thus are vulnerable to hacking by adversaries. As examples, he cited how Gutenberg’s invention of the printing press led to the Reformation, anonymous broadsheets attacking public figures, political strife, and yellow journalism; how advances in chemistry in the 19th century were “hacked” to produce chemical warfare during World War I; and how the invention of the internet led to social media platforms that are now organizing venues for extremists. These outcomes, resulting from new technologies having unintended consequences, are hard to see ahead of time or deter once they are used for hostile purposes.

Axtell then mentioned an idea circulating in economics and finance, concerning whether it is possible to create models analogous to weather forecasting that would take 70-plus years of economic experience and use supercomputers to provide a near-term forecast of possible outcomes. For weather, 2 weeks of look-ahead is now possible. High-fidelity modeling of the financial system might make it possible to computationally generate every type of economic or finance crisis possible, something that would be useful to agencies such as the Financial Stability Oversight Council or other institutions tasked with promoting robust market operations.

He then posed the idea of adopting a perspective that treats ideologies as platforms for action in the same way that YouTube or Facebook are platforms. The New York Stock Exchange and NASDAQ, for example, are platforms on which different customers operate. In that view, Axtell wondered if it is useful to think of democracy and authoritarianism as two platforms, or to consider capitalism, market socialism, and communism as platforms. The idea then would be to conceptualize the transition from one platform to another as a rare event.

In closing, Axtell concluded that in social, economic, and financial systems, rare events are simply events that occur in the tails of heavy-tailed distributions. Such events are managed in both ex post and ex ante fashion, by policy professionals, with more or less success. He explained that reaction, mitigation, adaptation, and redesign are more common in these areas than anticipation, prevention, and deterrence, which suggests that, say, eliminating all forest fires would be less effective than a strategy of keeping small fires from becoming catastrophic events. Given that the world is an ecology of adversarial agents who are perpetually evolving, new strategies are hard to anticipate, making deterrence hard in practice. However, computational modeling with rich data may provide better insights into rare events and how to manage them than the more heuristic methods used historically.

DISCUSSION

The discussion started with Daly answering questions about whether an adversary could turn an asteroid into a weapon and whether deflecting an asteroid might cause it to hit one country over another. Theoretically, said Daly, it would be possible to change an asteroid’s path in such a specific way that it would strike a specific country, but from a practical context he said it was not obvious how that would work. Regarding the second scenario, that could be a possibility and a United Nations conference in 2020 actually wrestled with this issue.

Daly noted that the reason NASA runs the United States planetary defense system is so that these issues would be handled in the public eye rather than in secret through DoD. In fact, he stated, there are concerns about what China is doing in terms of planetary defense. The way to stay on top of what other countries are doing is to look at the technology developments that are occurring, he added.

Justin Kasper noted that plasma physics sees long-tailed distributions similar to those that Axtell discussed and added that they arise through long-distance interactions. He asked Axtell if economic models incorporate

long-distance interactions among people and if that complicates trying to predict what groups of people will do. Axtell replied that data on who is interacting with whom over what distance do not exist, but such phenomena do exist. The collapse of Long-Term Capital Management was such an event where several coupled events occurred that triggered the company’s default. He added that turbulence in financial markets is one of the few areas where the natural sciences and social sciences have a direct relationship. Kasper noted that this could be an example where the behavior of plasmas could be the analogous situation to inform finance and economic models since their statistics are similar.

When asked if there is a role for dystopian science fiction in developing scenarios of rare, high-impact events, Baum emphasized that the answer is definitely yes. One role science fiction can play is to help understand how some unprecedented catastrophes could play out. Another role is through public communication by making a particular risk more psychologically available in people’s minds. For example, Deep Impact, the asteroid collision blockbuster film, played a constructive role in raising the public profile of the asteroid risk and may have led to some constructive policymaking. The danger is that science fiction might give people the wrong impression about the risk.

Baum then replied to a question about how much of the probability of the risk of nuclear war is driven by geopolitical events transpiring and how much is driven by bad procedures with logical flaws that result in a human doing something wrong. He remarked that this is a subject of contention within the nuclear war community and that there is currently no clear answer to that question. This is an important policy question, though, that could benefit from more research.

Kasper then asked the panelists to comment on the role that high-fidelity digital twins might play in better understanding the particular risks they study. Daly replied that it is important not to trust the digital twin too much. He suggested that the asteroid community assumes in its simulations that asteroids are made of quartz and that spacecraft are made of aluminum, neither of which is completely true. He noted, too, that people outside of the community might look at the visualizations a digital twin produces and believe they must be right, when the people running the models know that is not true. Baum replied that he thinks the concept of a digital twin is unlikely to have much value for studying catastrophic risk because the systems are too poorly understood and too difficult to model.

Axtell noted that DoD often requires simulation models to assess the feasibility of military projects. Similar models play an important role in modern production processes, where the idea is to stress the system to see where it fails and learn where changes are needed. His guess was that by 2050 there will be enough data and computational power that nobody will do prototyping in hardware anymore, it will all be done computationally. His hope is that rare event modeling will also improve to the degree that it will be possible to simulate rare events and not have to live through them.