Existential Risk and The Method of Collapsology

Apologies for the silence. The speak-first, elaborate later style of newsletters has taken some getting used to. From now on, Collapsology will appear weekly. I’m optimizing for variance and feedback. Feel free to post any in the comments.

Existential Risk research studies the upper sliver of risks that face our ostensibly undifferentiated humanity, risks that would destroy the entire world in an irreversible way: massive solar flares, the newly popular figure of the interstellar asteroid, bio and nanotechnology risks, and, of course, the risk from superintelligent (or just very powerful but stupid) AI. Because these risks are so extreme, and mostly more or less causally unrelated to one another, such research tends to parcel them out into distinct chapters in books on the topic, dealing with them one at a time.

In their very justified critique of the short-termism of conventional politics, such ‘longtermist’ thinking frequently conjures the stupendous number of people who could have future lives of happiness if we don’t all die in the relatively near future. As they say, if a 1-in-a-million risk can be reduced by a factor of ten and this allows for quintillions or more happy lives to be lived in the future, this is a worthy aim. It’s a compelling and serious vision, which demands that we consider why it is that as our society has become more and more technologically capable of laying the foundations for the distant future, it has nevertheless rapidly shortened its conceptual horizons. (We’ll leave aside for the time being the strangely conservative insistence on denominating the happiness of interstellar intelligent beings billions of years in the future in the actually-fairly-recent notion of the ‘individual’).

This mix of short-term risk (already being widely considered) and longtermism is the equivalent of a barbell trading strategy, where both very long-term and very short-term bonds are invested in, but not other bits in the middle of the risk distribution. For certain, other bits of the risk distribution need attending to: those in the upper 10% of damage and danger and not just those which form absolute bottlenecks to our intergalactic future. What might they be? Brutal climate change, widespread crop failures, the return of fascism at scale, ‘small-scale’ nuclear exchanges or intractable conventional war over a large area, the reduction of large parts of the world to ‘sacrifice zones’, the proliferation of new types of serious disease, is a decent start for a list of horrors. None of these is likely to terminate the human project entirely; they thus lack both the glamour and the suddenness of existential risks. But they absolutely must be attended to.

However, we can’t simply extend current existential risk research to encompass these other risks. These are not risks generated by the sudden exceptional spiking of a single parameter (a really massive volcano or an extremely capable and malevolent individual). They need a different method.

What is inadequate in the Existential Risk approach to address these kinds of risk? Bryan Caplan’s thin discussion of the global existential threat of ‘global totalitarianism’ from Nick Bostrom and Milan M. Ćirković’s edited volume on existential risk is a good example of a failure to address a complex sociological risk, on whose basic existence we and Caplan are agreed. His peculiar argument focuses on a single parameter, the mendacity of the totalitarian leadership itself, and is thus unable to account for such complex crises. Caplan argues that “Totalitarianism [in 1989 in the Soviet Union] ended not because totalitarian policies were unaffordable, but because new leaders were unwilling to keep paying the price in lives and wealth.” What is behind this peculiarly individualised conception of totalitarianism? Why are all the salient social factors except the internal commitments of a tiny group of extremely powerful individuals absent?

Another startling comparative quote is revealing: “One of the main differences between the Soviet Union and Nazi Germany was that the former became totalitarian very rapidly.” When did the Nazi Germany become totalitarian then, if not in the immediately post-1933 era of the proclamation of absolute dictatorial power, compulsory consumption of propaganda, death squads, pogroms, concentration camps, and the mass torture and killing of the enemies of the state? Only, apparently “in the last years of World War II,” presumably once the Nazi party and state had commandeered the means of production for the war effort.

As Quinn Slobodian has written, such a conception of ‘totalitarianism’, in which it is opposed not to democracy but to market liberalism, is derived straight from the writings of Friedrich Hayek. In the words of Wendy Brown, who expands on Slobodian’s analysis, Hayek argued that in order to prevent the risk of totalitarianism, ‘society must be dismantled’. If there is no mass society capable of sustaining totalitarian infringement on individual rights (here Hayek is more nuanced than Caplan), then ‘totalitarianism’ as an illiberal project of wealth redistribution or state control of the means of production can never arise. Brown uses such an analysis to show the process by which the neoliberal attempt to (incompletely) dismantle society and therefore prevent the return of fascism instead led to the proliferation of new forms of far-right anti-democratic politics that were antagonistic to the social liberalism of pluralistic societies in ways similar to classical fascism.

These analyses do precisely what Caplan’s cannot: address a multitude of social causes across the scales of the individual, institutional, intellectual, and political. They identify a knot of interrelated causes for a risk. And, perhaps most importantly, they address the ongoing tragic quality of human affairs: that the best-laid plans often end up producing variants of the exact thing they are attempting to prevent.

Of course, we should not dismiss Caplan’s concerns about totalitarianism. New communicative technologies absolutely do afford increased degrees of anti-democratic and authoritarian control, or at least stimulate an arms race between such control and its avoidance. But we need a much more nuanced conception of the risks and where they lie.

What fails in Caplan’s methodology - his focus on a clique of mendacious leaders - at the expense of the wider social system is also a problem in existential risk research more generally. What we need, if we are to turn to the broader category of not-quite-existential risks, is a more thorough engagement with sociology. We can’t simply widen the aperture of our risk sensor with no further changes.

So what can we say about the top 10% of risks beyond “it’s complicated”? First, it is increasingly reasonable to think of these upper 10% risks as not just qualitatively different from extreme risks, but as all deeply interrelated. They are not the risk of an extreme in a single parameter, but, as David Krakauer & Geoffrey West of the Santa Fe Institute wrote of the COVID-19 pandemic, a ‘complexity crisis’: “the failure of multiple coupled systems—our physical bodies, cities, societies, economies, and ecosystems” where the solutions involve “unavoidable tradeoffs, some of which amplify the primary failures. In other words, the way we respond to failing systems can accelerate their decline.”

Such a tragic view also stimulated the sociologist Karl Deutsch’s theory (here summarised by Jairus Grove) that power, far from being something that stabilizes politics, “is in fact pathological, leading to sclerotic and brittle states and organizations.” Why? Because “material power reduces the necessity of states and institutions to learn and adapt”. No need to adapt if you can simply bat away the problem for the time being. Both responding in a way that precipitates further damage and responding in a way that generates more and more brittle institutions are risks. Of course, as in the pandemic, so too in climate change: exponentially increasing damage will overwhelm any system inattentive to the need to build capacity.

There are existing institutions that engage with many of the above risks - the World Food Programme is right now loudly ringing the alarm over spikes in global food risk brought about by the war in Ukraine - but their interests are in mitigating risks in a world that remains essentially as is. This is another major difference of methodology with what we are calling Collapsology. Although Collapsology is not indifferent to suffering by any means, its more open to what Jairus Grove calls ‘Apocalypse as a Theory of Change’ (Grove will be a major topic in many future missives).

Collapsology is a project for the design of a politics that can carry something through collapse in full awareness of the complexity of the task, in the same way that recent innovations in military strategy de-emphasise achieving some specific set of aims (an impossible aim in a relentlessly chaotic environment) but focus instead on a ‘chaoplexic’ wrestling with an adversary. It is in this constrained design space - in scope between the annihilatory and the ‘glocal risk multiplier’ and in a style of engagement between morally abyssal and the technocratic - that an adequate response to collapse will have to be found. These constraints form the beginning of a methodological statement.

Before we start thinking about the possible drivers of contemporary collapse, or about its consequences for us as a species, a smattering of individuals, or as fragile societies, there is some ground clearing to be done. That will be our topic next time.