(This is part 1 of the Stages of the Anthropocene, Revisited” Series (SotA-R).)

In Stages of the Anthropocene (2018), I argued that if one aims to make mid- or long-term predictions of the effects of climate change on human civilization, it is useful (if not necessary) to distinguish three phases or stages within the time period known as the “anthropocene”. There is some disagreement about when the anthropocene started – the industrial revolution (mid 18th century) and the advent of the atomic age (mid 20th century) have been suggested, and an argument has even been made for the emergence of agriculture – but fixing the beginning of the anthropocene (and thus of stage 1 therein) doesn’t matter much here, so we can fortunately sidestep this controversy. What does matter, on the other hand, is when stage 1 ends and stage 2 – as I defined it – begins. The definition of that boundary is the point in time when net artificial emissions of carbon and other (major) pollutants return to zero. If carbon-neutrality by 2050 would be a realistic possibility (and other major sources of pollution could also be terminated by the same date), then 2050 would be the end of stage 1 and the beginning of stage 2. That next phase of the anthropocene, stage 2, is characterized by the adaptation of the Earth system and its parts to the changed atmosphere and its consequences (caused in stage 1). Stage 2 ends (and stage 3 begins) when climate change processes that have been set in motion in stages 1 and 2 stabilize or become so slow that species and ecosystems can adapt to them.1

The significance of this division of the anthropocene in three “stages” is that these are defined by very different processes, and even more importantly, perhaps, that the nature and circumstances of stages 2 and 3 (and thus of most of the anthropocene) depend primarily on the amount of pollutants released in stage 1. Hence, mid- and long-term prediction of the effects of climate change, then, must start with a realistic model of stage 1. If we want to know what the lives of our grandchildren’s grandchildren might be like (assuming that those will be living in stage 2, if at all), we must first predict what will happen in stage 1 and when it will end.

Although I still believe that this aspect of the approach I took in Stages of the Anthropocene is right (if not essential), there is another fundamental defect in the method used to make predictions in that article. As I pointed out in Predictions (which could be seen as a preface to this article/series) attempts to predict the future often underestimate social feedbacks. Humans respond to changes in the world around them, adapting their institutions and behavior in various more or less subtle ways, and if such social and behavioral adaptations are insufficiently taken into account then there will be major discrepancies between scenario and reality. For example, people changed their behavior in response to the Corona pandemic, slowing down the virus’s spread. A model that doesn’t take such a behavioral change into account would predict an unrealistically fast spread throughout a population.

The aforementioned fundamental defect in Stages of the Anthropocene is exactly this: it takes insufficiently into account that the CO₂ we emit changes things and that such changes may affect how much CO₂ we will emit until the end of stage 1 (and when and how stage 1 will end). And again, how much CO₂ we emit until the end of stage 1 determines much of what comes afterwards. A realistic prediction of the mid- and long-term future cannot start with just guessing the total amount of CO₂ we’ll emit in some kind of business-as-usual scenario (because the effects of emissions may make “business as usual” impossible), but must dynamically jump back and forth between emissions, behavioral changes, policy changes, and other effects and intermediaries. For that reason, it cannot jump decades into the future at once either. If expected effects of atmospheric CO₂ are such that important aspects of the world in 2030 will be very different from the present situation, then skipping that time point will not produce a useful prediction.

Furthermore, for largely the same reasons realistic predictions must be radically interdisciplinary as well. Some experts on climate change seem to believe that they are the only ones qualified to predict the effects of climate change, but that assumption is nonsensical. As I have argued before, if you want to assess the effects of climate change on national security, for example, you’d better ask a national security expert, because in that case it isn’t climate change itself which is modeled – rather, climate change is then an independent variable and what you’re really looking into lays far outside the area of expertise of the average climate scientist. Things get a bit more complicated when what you are interested in is the interplay of climate change, (inter-)national security, ecology, economics, sociology, and a lot more. One could, of course, ask experts in all of these fields, but most likely the result will be much like the famous Indian parable about the blind men and the elephant – one thinks that an elephant is like a tree, another that is is like a rope, and so forth, because they are all focusing on one small part and no one can see the whole.

Obviously, I can’t see the whole either. I lack the expertise in all the relevant fields for the simple reason that no one can be a specialist in climate change, national security, economics, and so forth, and so one, all at once. What I can do, however, is start with some simple predictions and gradually refine those in a process of repeatedly revisiting them, adding new data and new connections in every iteration. That – in a sense – is what I’m doing already, as this post is the beginning of a revisiting of Stages of the Anthropocene (but also of The 2020s and Beyond).

a starting point

According to a projection published in Nature in December 2018,2 Earth will be on average 1.5°C warmer in 2030 than it was at the start of the industrial revolution, and we will reach 2°C around the middle of the 2040s. I’ll mostly refer to the years 2030 and 2045 in the following, but both references should be read as involving uncertainty margins, and the uncertainty margin for 2045 is obviously much wider. We might reach 2°C sooner (in 2040, for example), or a bit later, or possibly not at all, if due to unforeseen developments we cut down CO₂ emissions faster than even the most optimistic scenarios. However, aside from global nuclear war or a major asteroid impact could, I don’t see any plausible way of achieving that. If implausible scenarios and black swans are ignored, then the +1.5°C for 2030 and +2.0°C for 2045 (with aforementioned margins) are virtual certainties. This being the case, predictions and scenarios for the unfolding of stage 1 of the anthropocene can take these two data points as a given, and should start with an assessment of their likely effects.

The direct effects of rising temperatures are climatological, of course. An increase of the global average temperature by 1.5°C will lead to more (deadly) heatwaves, more frequent and more severe droughts (and fires resulting therefrom), more extreme weather ranging from cold spells in winter to stronger and more frequent hurricanes and typhoons, and so forth. All of these could be called “disasters” – “natural disasters”, perhaps, although there is something rather unnatural about them – but such disasters differ in two very important dimensions: intensity and scale (or depth and width). The intensity (or depth) of a disaster concerns the extent of localized damage. Tornadoes destroy almost everything in their path, and thus have a very high intensity. The scale (or width) of a disaster consists of two further dimensions (or sub-dimensions): their temporal and spatial scale. The spatial scale concerns the area affected; the temporal scale concerns the length of time the main direct effects of the disaster last. The scale of a tornado, then, is limited – the area destroyed tends to be very small and tornadoes don’t last long either. Heat waves, cold spells, and drought are more or less on the opposite end of the spectrum: their spatial and temporal scales tend to be vast, while their intensity is relatively limited.

While high-intensity disasters obviously produce horrendous suffering for their victims, they tend to have limited spatial and temporal scales. Such disasters are usually local, and because of that, their effects on larger spatial scales tend to be limited or even negligible. Japan is hit by several typhoons each year, for example, and it seems that the damage they cause grows every year, but this damage is always very local, and the effects on Japan as a whole are, therefore, quite insignificant. Low-intensity disasters, on the other hand, tend to produce less acute suffering and receive, for that reason, far less media attention, but also often affect larger regions and longer time spans. It is not the low intensity itself that makes such disasters important, of course, but there appears to be a negative correlation between intensity and scale. This negative correlation is represented with the downward sloping orange line in the following diagram.

Dimensions of disaster.

Many types of disasters are on or close to that line. Heat waves and cold spells tend to affect far bigger areas (and longer time periods) than tornadoes and floods, for example, but their intensity is also much smaller. The exact positions of the various kinds of disasters in the diagram isn’t very important, by the way. What is important, however, is that the “severity” of a disaster is a combination of the two dimensions such that the higher the intensity and the larger the scale, the more severe the disaster. Or in other words, this added severity dimension crosses the orange hypothetical “line of best fit” at a (very!) roughly 90° angle. If the location of the various disasters in the diagram is approximately correct, and the angle is 90° indeed, then this would mean that tornadoes, floods, hurricanes, and heat waves are of comparable severity and that cold spells are slightly less severe. The main outlier, however, is drought. The spatial and temporal scales of droughts tend to large (as already mentioned), and while their intensity may be less than that of a tornado or flood, it tends to be far greater than that of a heat wave, for example.

Furthermore, droughts are an outlier for another reason: their secondary effects tend to be far worse than those of the other disasters, although this is largely a consequence of their scale. Tornadoes, floods, extremely heavy rain, and drought all destroy crops, but only droughts do so in areas of a significant size. Contrary to these other disasters, droughts often cause widespread water shortages and famine. Tornadoes, hurricanes, and floods destroy homes, but droughts can destroy the conditions that make it possible to have one’s home in the area affected in the first place. It is for this reason that droughts are the kinds of disasters that are often associated with famine, refugee flows, and violent conflict (including war and civil war).

Making predictions or sketching scenarios for stage 1 of the anthropocene, then, needs to start with a prediction of drought and its likely effects. This doesn’t mean, of course, that other kinds of disasters can be neglected, but the focus should be on drought, at least at first. That, therefore, will be the topic of the next episode in this series: expected drought and its effects in 2030 and 2045.

If you found this article and/or other articles in this blog useful or valuable, please consider making a small financial contribution to support this blog, 𝐹=𝑚𝑎, and its author. You can find 𝐹=𝑚𝑎’s Patreon page here.


  1. They won’t stabilize completely, because artificial CO₂ in the atmosphere will very slowly be reintegrated into the bio- and lithosphere, and thus CO₂ levels will very slowly fall throughout stage 3, leading to a gradual cooling of the planet. This will last many millennia and is unlikely to result in a return to the present global climate due to the fact that we’ll almost certainly pass some irreversible tipping points in the current century.
  2. Wangyang Xu, Veerabhadran Ramanathan, & David Victor (2018). “Global Warming Will Happen Faster than we Think”, Nature 564 (6 December 2018): 30-32.