For many people these days, it’s hard not to feel like the end of times is at hand. Greta Thunberg and Extinction Rebellion have brought extreme climate concerns into the mainstream, big-time. Global catastrophe and the collapse of civilization have made the leap from fringe to commonplace so quickly that the new talk is about “climate trauma” and “climate grief” — coming to terms with the impending apocalypse — as well as the “courageous” wartime measures supposedly necessary to survive it. As I write, this zeitgeist is further exacerbated by 2020’s “black swan”, the coronavirus outbreak, which many are interpreting as proof of the increasing fragility of our ever-more-interconnected Anthropocenic Earth.
On the other hand, if you want to believe the opposite — that our world is robust to climate change— there’s plenty of reputable literature for you. The models of “mainstream” scientists vary in assumptions, parameters, causal effects and the degree of the severity of their conclusions, but even the most dramatic scenarios are not quite Armageddon. Indeed, the dissonance between public perception and the scientific status quo is such that a reputable climate researcher has felt the need to make it explicit: “There’s never been that much evidence that climate change is going to literally cause the extinction of the human race”.
Of course, there is a huge spectrum between “business as usual” and “human race goes extinct”; it would be useful for those of us with a stake in the game — i.e., everyone — to narrow that range down somewhat.
Mainstream economics is, by and large, on the robustness camp. A relatively pessimistic study from last year, quoted in the popular press with misleading titles such as “Climate change could cost the U.S. up to 10.5 percent of its GDP by 2100”, actually says that economic output will be 10.5% lower than a counterfactual baseline (without climate impacts); the punchline, of course, is that this effect is dwarfed by the steady exponential growth that characterizes the baseline. Indeed, at the current pace of growth in the US (~2% per year), a loss of 10.5% corresponds to a regression of about 5 years; that is, if climate change did its worst, it would take us 80 years to get 75 years’ worth of growth. Said yet another way: if we shook a magic wand and made the climate problems go away, American GDP in 2100 would reach $94.5 trillion (in today’s currency); the lack of that magic wand would only bring this down to $84.6 trillion… which is still 3.4x times higher than today’s $19.4 trillion! This worst case doesn’t seem like a bad deal at all.
Indeed, the math is even more extreme for world GDP as a whole, where the baseline of growth is more like 3.4%; per this model, global GDP would increase 14.5-fold by 2100 on the baseline, and “only” 13.4-fold when accounting for climate change.
So the question of the world’s resilience seems quite undecided. Is it extremely fragile and bound for unavoidable global collapse? Or is it extremely robust and guaranteed eternal prosperity (with a chance of minor hiccups)? As you may suspect, this is a trick question; I don’t quite agree with either position. However, it’s not as simple as splitting the difference; what is the average between zero and infinity?
The logic of the Whole: self-realization
The problem with both of the analyses above, of course, is that they are purely statistical models, unduly focused on averages, aggregates and correlations: they don’t take into account the causal mechanisms that underlie the system’s functioning in the first place. To think rigorously about resilience and the future of the world, we must deeply understand the workings of our coupled economical-ecological system (Gaia, aka “the world” or “the Whole” below). But to do this, we must dispel a preconception about complex systems, to wit, the notion that they are intrinsically impossible to understand and predict. The well-meaning lessons of the 20th century about the dangers of linear thinking and the ubiquity of chaotic phenomena have left us with a sort of learned helplessness when it comes to complexity; any mention of the c-word and many otherwise smart people will default to knee-jerk memes about butterfly effects, etc. Luckily, the exact opposite is true: our complex world is also extremely predictable in a very specific sense.
Indeed, the reason why our system is complex is exactly the same reason why it is predictable: the world is an evolved system, shaped by selective pressures to produce subsystems that are self-realizing (self-organizing and self-reproducing). The predictability is that we can count on subsystems being self-realizing (“fit”), i.e., subsystems more adept at self-realization have a much stronger likelihood of growing in scope and therefore becoming eventually ubiquitous. The complexity comes from the multifarious structure that these subsystems evolve in order to achieve ever greater degrees of self-realization.
(Given the above, I have taken to avoiding the terms “complex” or even “adaptive” when talking about such systems; I’ll stick to “self-realizing” for this article. This term clearly indicates we’re talking about a functional, purpose-driven system – even if it that purpose is a “blind watchmaker”. Complex systems may or may not be self-realizing. As for “adaptive”, that term simply means that a system responds to changes, whereas the important thing about our world is that it evolves.)
Note that I am not referring just to Darwinian biological evolution here. As brilliantly discussed by Robert Wright, this phenomenon of selective pressures creating increasing levels of self-realization (by causing functional, organized and synergistic systems to have higher success rates in the environment) is truly everywhere once you see it: from autocatalytic loops in chemistry (a popular explanation for the origin of life itself) to humanity’s evolving social, technical and economic capabilities.
The underlying theory, also intuited by Wright and more recently formalized by Karl Friston as the Free Energy Principle, is one of information processing: subsystems are successful to the extent that they organize to process specific types of information about their environment and respond to it (creating and maintaining internal order by increasing the environment’s entropy); the Whole is continuously selecting for subsystems that respond to each other in this way, in an all-encompassing causal tapestry.
Even taken at a very coarse-grained level, this is already more useful than most models and forecasts out there. According to this theory, if we zoom out far enough, the Whole is always “trending up” (i.e., increasing its internal order), right until we hit fundamental physical limits such as the energy output of the sun (from which we are still many orders of magnitude away). However, the same theory predicts that, if we zoom in, we will find frequent troughs in between the increasing peaks, as the system encounters changes in conditions (“shocks”) and reorganizes to tackle them. The system does eventually continue to achieve greater heights; however, the depth and length of the trough are highly dependent on the specifics of the shock and the subsystems affected.
Nonetheless, note that the theory is functional: troughs correspond to specific system failures; these may be complex and cause higher-order issues, but they are always solvable in principle: by understanding the nature of the breakdown, one can reason about the nature of the correction. As we’ll see, this will be very useful later on.
Note: this is very close to Howard Bloom’s thesis of “booms and busts” in The Genius of the Beast. However, I don’t subscribe to his thesis that all busts are healthy or necessary — I’m not even sure that he fully believes in that himself. More on this later.
Transitioning to a positively-coupled system
Now that we got the philosophy out of the way, what does that tell us about the future of the world? Well, it doesn’t give us a forecast for GDP growth, that’s for sure; but it does provide some reliable mental models, out of which we can start making sense of things. Namely, if we take seriously this picture — the world as an intricate, self-sustaining and self-evolving coupled system of humankind and nature — we can reason about how it is likely to respond to various kinds of disturbances, and what role we can (and must) play as part of this system.
To tackle that latter part, we need to better understand the nature of this relationship between human (economic) activity and the ecosystem. But to do that, we’ll go on a short detour to the Precambrian era, approximately 2.4 to 2.1 billion years ago.
The “Great Oxidation Event” (GOE) happened some time during the Precambrian; though the details are sketchy, the basic facts are well established. If the GOE’s name seems anodyne, consider two popular alternatives for the same event: the “Oxygen Holocaust” and the “Oxygen Revolution”. Quoting Wikipedia: “Geological, isotopic, and chemical evidence suggests that biologically induced molecular oxygen started to accumulate in Earth’s atmosphere and changed Earth’s atmosphere from a weakly reducing atmosphere to an oxidizing atmosphere, causing almost all life on Earth to go extinct”. It seems likely that a population explosion of photosynthesizing cyanobacteria produced such vast quantities of oxygen as to change completely the composition of the atmosphere and oceans and cause almost all life of the era, being anaerobic, to perish.
So what should we learn from the GOE — that life is fragile? Yes, but also that life, uh, finds a way. Indeed, the newly oxygen-rich environment greatly expanded the chemical free energy available to organisms, opening the evolutionary field to a hugely expanded range of metabolic innovations — including mitochondria, complex ecosystems, and eventually multicellular life. Eventually, the evolution of aerobic organisms established an equilibrium in the availability of oxygen.
Was the GOE a lucky break? Were we just incredibly fortunate that evolution was able to find its way out of the enormous die-off before everything went extinct? Of course, one can’t prove otherwise; but a more fruitful perspective is that something like the GOE was likely to happen eventually, given (a) the enormously higher degrees of structure made possible in a high-oxygen world, and (b) the relative instability of the previous low-oxygen configuration (which made possible the uncontested explosion of cyanobacteria in the first place).
Before we proceed, let me reiterate: there is a line between the optimistic position of seeing the potential for a better end state after a catastrophe, and the misguided naive Darwinism of saying that catastrophes are for the best and should be welcomed (“let nature take its course”). Again, more on that later.
In other words, the GOE marked a tipping point in the transition from a lower-order to a higher-order equilibrium. I argue that today’s climate crisis similarly punctuates the transition from a purely ecological world to a positively coupled ecological-anthropic one — i.e., one where social organization based on human action is a foundation of the stability of the system. Just as the cyanobacteria caused a hecatombe by overproducing oxygen, our coupling started out looking like parasitism — sucking natural resources to “overproduce” economic capital, as the Anthropocene is usually characterized in current literature. And similarly to how the newly abundant oxygen produced the conditions for a new equilibrium based on aerobic life, the very thing that seems to spell doom — the abundance of capital, be it physical, intellectual, human, and social/organizational — is actually the fuel that expands the space of possible configurations to an unprecedented degree.
Resilience and shocks
Another perspective on resilience comes from considering extreme events, or “shocks”. As Nassim Taleb emphasizes in his oeuvre, such “non-normal” events, though rare, have a disproportionately large impact on the system’s trajectory, to a much larger degree than is usually recognized or accounted for in most models.
Indeed, whether in Taleb’s home turf of finance, or in economics, politics and ecology, this is true: normal, quotidian shocks are accommodated and dissipated locally by the system, while extreme shocks have global impacts, forcing the entire system to respond, adjust and learn — i.e., to evolve.
I have absolutely no disagreement with this argument; what is wrong is to conclude that this is a bad thing! Indeed, quite the opposite; this capacity for higher-level, global evolution is precisely the source and the measure of resilience. Systems that aren’t self-realizing, no matter how complex or efficient, are mere machines that break and stay broken; self-realizing systems have the ability to fix their own breakages, and more importantly, to create mechanisms preventing similar breakages in the future. (Taleb makes much the same point, in a somewhat more obtuse way, in the latter part of his book Antifragile.)
As we’ve seen above, in self-realizing systems, complexity — the density and speed of the interconnections — is evolved precisely to enable resilience. To understand this correlation better, consider the Permian–Triassic extinction event, also known as the Great Dying. Unlike the Great Oxygen Event, this one was most likely caused by a sudden, exogenous shock: quite possibly a massive release of volcanic gases over a period of centuries or millennia (that is, incredibly short on a geological time scale). The Great Dying earns its name as the Earth’s most severe known extinction event, with up to 96% of all marine species and 70% of terrestrial vertebrate species becoming extinct.
Now, on the one hand, this example drives home the horrifying impact that such extreme events can have. On the other hand, it helps illustrate the relationship between self-realizing and resilience. Now, at the time of the Great Dying, the Earth was already a complex ecosystem, with an abundance of multicellular species, and many if not most of the niches and traits present in today’s natural world were already exploited (though some, like pollination and flowers, seem to have come about only much later). Yet, obviously, biological evolution was hard-pressed to respond to such sudden changes as the global onset of acid rain and ocean temperature increases to as much as 40º C! Paleontologists’ characterization of the impacts to species and ecosystems in the aftermath include what one would expect: biomass and variety were greatly decreased; as complex food webs were eliminated, species became more omnivorous; motility increased in response to predatory pressure; etc.
In summary: life does find a way, but that way is often tortuous; according to one paleontologist, the recovery from the Great Dying took until the Late Triassic, about 30 million years after.
Is that the most resilient that a natural ecosystem gets? That is, would a more robust ecosystem have been able to recover faster in — 10 million years? 1 million? It’s hard to say. There are certainly features, like the aforementioned pollination, that seem like possible accelerants, but this is outside my field of expertise, so I can’t judge here.
However, I can compare that with the hypothetical response from other systems, namely, our present one, in which nature and society are positively coupled. I argue that our coupled ecological/sociotechnical system would have a strictly faster expected recovery even in the case of a Great Dying-scale event, with an impact on all the “operating parameters” of our natural world. Humanity may not survive such an event, but if it does, then so does our ability to learn and create; and we will preserve or, at worst, rediscover all our present knowledge we currently have. Our global network of learning, collaboration and trade may be completely disrupted, but its recovery becomes a matter of centuries or millennia, not eons. And indubitably, our first focus will be tapping whatever technological knowledge may be remembered or reinvented, from genetic engineering to carbon capture, in order to recreate a suitable environment for ourselves.
If this seems like such an outrageously extreme scenario that it’s hardly useful to learn from, consider the following, more modest case. Suppose that the climate change projections from the mainstream economists are way off, and through a series of snowball effects, a full 90% of our economic and natural infrastructure gets wiped out — enough to “send us back to the Stone Age”; the kind of “turkey apocalypse” that Taleb likes to imagine, and that is usually popular in non-lay literature. Well, how do we fare in such a scenario? Does a 90% reduction in world GDP seem drastic enough? Indeed, that would take civilization way back… to the late 1950s.
Let’s pause to let that sink in. Obviously being sent back to the economic and technological state of 60 years ago would be a massive catastrophe: for one thing, vast numbers of people would be sent back to poverty or even die of hunger. (The Green Revolution was still kicking in at that time, and famines were frequent.) However, all the civilizational and technological underpinnings of our contemporary wealth were already in place: we had the structure of DNA, quantum physics and relativity, transistors and high-level software languages, nuclear and solar power, etc. Even if we lost our blueprints for all the tech that came later in the ensuing chaos, it would all get reinvented, in one form or another. Therefore, it’s no stretch to say that even a catastrophe big enough to annihilate 90% of our economy would take humanity only a few decades to recover from, if not less.
A caveat: consider these scenarios just as illustrations. For example, they assume that natural capital can recover at roughly the same rate as economic infrastructure is redeveloped. I don’t know of enough literature on this topic to make a robust case either way, although the sources I have seen are optimistic on this point.
We can follow this reasoning a bit further. Losing 99% of GDP? That sends us back to the 1820s. The Industrial Revolution was already in full steam. 99.9%? Economists’ retroactive estimates become increasingly useless the farther back you go, but that may have been somewhere during the late first millennium BC — a time of short and brutish lives, but also the time of the classical Greek and Chinese civilizations, of Judaism and Buddhism, and- crucially — of experimental science and theory. However destructive we assume a global catastrophe to be, it seems that we have quite a bit of buffer removing us from the Stone Age.
I would pause here on the topic of black swans’ outsize impact, if it weren’t for a persistent — yet fallacious — argument that is often invoked by Taleb and peers. This invokes the statistical/game-theoretical concept of ruin — meaning a loss of 100%, not merely 99% or 99.9% and so on. Ruin means game over, finito. This — it seems — would put a sad end to all of our talk about resilience.
Now, of course, Taleb is correct to claim that ruin is qualitatively different from mere loss; a turkey doesn’t get another shot after Thanksgiving. He invokes the trump card of ruin to champion a sort of contemporary Pascal’s Wager: since ruin means you can’t play again, it must be imputed a value of “negative infinity”, and therefore, anything that may lead to ruin with a non-zero probability needs to be avoided with extreme prejudice. (Taleb’s post-financial crisis bête noire is GMOs, but he is also increasingly vocal about the ruinous potential of climate change.)
Here’s the problem with that argument, though. It’s easy to reason about ruin in the original setting, where your “system” is a gambler with $100 in the pocket, and the “game” is over when the gambler loses it all — or in the case of the turkey, for that matter. However, if we’re talking about the future of humanity — and Taleb does seem very concerned about that — then ruin must mean the literal extinction of the human race, or at the very least, of our collective memory and knowledge. Anything less than that, however catastrophic, must be accounted for as “just” a risk — quantitatively more or less bad than other risks, but not the literal end of times.
Now, faced with an existential threat to the world, some would find themselves justified to combat it at all costs. This is the heart of Taleb’s policy argument, although he masks it in risk-management terms. However hard to countenance when you consider just what “all costs” might mean in some of these cases (forced sterilization? Rationing of basic goods? A return to pre-industrial agriculture?), one must take it seriously: if the alternative is ruin, then it definitely does seem that some painful moral calculus might be in order.
However, as we’ve seen above, literal ruin for humanity and the planet is an inconceivably high bar — between our natural and cultural technologies for self-realization, the world makes for a remarkably resilient system. Now, it’s not impossible for some class of event to meet that bar — and a literal shock like an asteroid impact may do just that (more on this below). In such a case, dealing with repeated risks, one can argue that any risk, however small, gets compounded until it’s virtually certain. But therein lie the dragons of logical fallacy: to conduct such an analysis, one must make a judgment call informed by theory. Otherwise, one finds oneself in the untenable position of considering everything a potential risk of ruin, especially if one takes seriously “butterfly effects”. (Going to take that plastic straw? Think again, it just might spell the end of times!) More commonly, the “ruin” argument is used selectively, to cherry-pick technologies one doesn’t support on other (e.g., moral or aesthetic) grounds.
So, contra Taleb, we must consider climate change as “just” a major risk. Likely to cost a great deal in lives and money, and worth investing the combined ingenuity and resources of our species? Most definitely. Likely to annihilate humanity and worth fighting at literally all costs? No.
See this article for a sober analysis of Taleb’s and other worst-case scenario arguments on climate change.
On humanity’s job
OK, that’s enough about risk and ruin; let’s get back to the Whole and our role in it. We have postulated that, in a precise way, our “post-transition”, positively coupled economical-ecological system is strictly more resilient than the pre-transition, “natural” system. Additionally, I argue that humanity is also the only system component that can make such an impact in the resilience equation. This is no knock on nature at all, but an information-theoretical observation.
Specifically: the good regulator theorem from 1970 says that “every good regulator of a system must be a model of that system. That is, any regulator that is maximally successful and simple must be isomorphic with the system being regulated”. In this theorem, the “regulator” is any component of a system whose “job” is to provide control or feedback signals to ensure the system’s survival (i.e., minimize entropy); the theorem is simply saying that a regulator will succeed in this task only to the extent that its internal states correctly map to states of the world.
Again, to be fair, ecosystems are great at providing regulation, via our good friend negative feedback. There’s no shortage of examples: a CO2 surplus makes forests grow faster and suck that excess carbon out of the air, predator and prey populations regulate each other, plant cover regulates itself, etc. However, ecology’s causal tapestry, however intricate, is ultimately made up of elementary relationships that are fixed, at least outside of an evolutionary time scale. Predators will sense that there’s a surplus of prey and react by hunting them, but that’s about it: once they’ve eaten out all their stock, they won’t learn from that mistake and avoid over-hunting the next time around (at least not as individuals); and they definitely won’t predict that the declining prey numbers signal the moment to reduce meat consumption.
Now, again, this is not a problem under “normal” circumstances, as local variations — such as an endemic population being extinguished by overzealous predators — are absorbed by the system (the niche is filled by a migration, or the remnants of a prey population evolves a response to predation, etc). The problem is when a shock to the system affects too many such relationships at once (introducing global correlations), at too fast a timescale for “natural” reactions to emerge. In such cases, the system can only survive if it has components that are able to integrate global information, learn, predict and react fast enough.
That pretty much leaves humanity as the only contender. Our evolved ability to perceive and respond at fast timescales and higher levels of abstraction, including understanding of the self, of the Other, of social groups, of and systems, makes us uniquely positioned to play a key regulating role for our Whole. “Humanity’s job” is, thus, to collaborate globally in order to integrate information, learn, materialize knowledge into technology, and use it to predict and react to risk and harm.
Remarkably, the very theory of self-realizing systems, which we’ve described above, should be eventually found to play a key role in enabling us to do this job effectively. To the extent that we further this understanding of the Whole, we come closer to fulfilling the conditions of the good regulator theorem, and bring Gaia further along the journey towards becoming a fully self-designed world.
Humanity’s job evidently includes climate change mitigation and adaptation, but is not limited to it: other important risk-management activities include asteroid impact prevention and space colonization. Of course, as the Whole’s only “rapid response system”, we need to look out for our own integrity, by managing human-specific risks such as pandemics and war.
Note that, by elevating human action to a critical requirement for the system’s survival, this perspective also answers the ethical concerns with the supposed amorality of “naive Darwinism”, mentioned above. Indeed, in this theory, intention and morality are endogenous: humanity’s role not only requires us to act, but prescribes a specific moral yardstick for decision-making.
Finally, it is interesting to note that all these risk-management activities — taking care of the system — are given attention to in proportion to the prosperity that relieves us from having to care for our own individual survival. There is an interesting question of how did we get to this job? How did we become the Whole’s regulators? There is no direct selective pressure for that — our pre-human ancestors, living brutish and short lives, certainly wouldn’t have been selected to collaborate globally and prevent asteroid impacts. But given the system’s self-realizing nature, it does seem likely that it would find ways to “reward” species with high potential — some cognitive surplus and ability to coordinate among individuals — and eventually bring them to fill that niche.
Conclusion: Plus ça change
We started this article by comparing two opposite popular views of resilience, and we ended up with something remarkably different to both of them. In our perspective, resilience is borne not of stability, but of change. The Whole is neither robust (immune to change), nor fragile (vulnerable to change), but antifragile (responsive to change). Being self-realizing, it responds to new structures by testing them and rewarding useful ones with opportunities for replication, and to great shocks and transitions by expanding the space of such opportunities to create great new waves of innovation. The Whole is also not picky about specifics: structures and traits that perform a useful function are guaranteed to be developed, but one “implementation” may be mercilessly swapped out by another one.
I acknowledge that this perspective isn’t for everyone, for deep-seated psychological reasons: humans don’t like change. We have always worried about the end of the world, and it’s usually in proportion to the speed and intensity of change. (It’s not for naught that the prominence of apocalyptic cults, civil unrest, and environmental variability all correlate historically.) There’s only one problem: change is the only constant. The Anthropocene is here to stay: the world will never go back to an idyllic state of nature, or even “stabilize” at some homeostatic level of output. (Homeodynamics not homeostasis, as a classic paper says.) Quite the opposite, the transition to the Anthropocene is just beginning: the pace of global change will continue to accelerate as we become increasingly adept at understanding and influencing natural systems, and yes, those systems will become as fully dependent on humanity as humanity is on them.
To be sure, this is a massive responsibility, and our track record as a species is far from unblemished. But it is our job now: there’s no turning back. The only way forward is to accept this honor, and live up to the high expectations given to us by the hand of fate. In Stewart Brand’s immortal words: “We are as gods and have to get good at it.”