Hide table of contents

Important note

This is a rough draft I wrote between October 2019 and April 2020. It’s incomplete, and doesn’t reflect updates in my views in the 2+ years since I worked on it. 

I think there are serious downsides to sharing the draft publicly, because I think some parts of it are likely to be substantially wrong. I’m posting it anyway (with the hope that flagging the potential substantial wrongness will help people be especially skeptical of the conclusions) because I think the benefits outweigh those downsides. The benefits: 

  • Transparency: I’ve shared the draft with a number of researchers exploring civilizational collapse, and they’ve built off some of the research. It seems bad for an unpublished piece of research to be informing other research without being public (citable, scrutinizable) itself.
  • Potential insights: To the extent that not everything in this post is wrong, it seems good for people to be able to easily draw/build on any good arguments in it (rather than have to start from scratch).
  • Noticing bad arguments: As per Cunningham’s Law: “The best way to get the right answer on the internet is not to ask a question; it's to post the wrong answer.”

I think the probability of technological stagnation is somewhat higher than I did when I was working on this piece in earnest for a number of reasons — most of which I don’t have capacity to write up at the moment. 

The biggest reason is probably the risk of extreme, long-lasting climate change. It seems possible that anthropogenic climate change could cause global warming extreme enough that agriculture would become much more difficult than it was for early agriculturalists. Temperatures wouldn’t return to current levels for hundreds of thousands of years, so if the warmer temperatures were much less conducive to recovering agriculture and downstream technological developments, humanity might be stagnant for millennia. 


This research was funded by the Forethought Foundation. It was written by Luisa Rodriguez under the supervision of Lewis Dartnell, and draws heavily on research and conversations with Lewis and Haydn Belfield. Thanks to Max Daniel, Matthew van der Merwe, Rob Wiblin, Howie Lempel, Aron Vallinder, and Kit Harris who provided valuable comments. Thanks also to Will MacAskill for providing guidance and feedback on the larger project. And thanks to Katy Moore for editing this piece, and for drafting the summary. 

All errors are my own. 


In this post, I explore the probability a catastrophe that caused civilizational collapse might lead to indefinite technological stagnation (and eventual human extinction) — even if it didn’t cause extinction in the very short term (a topic I covered in What is the likelihood that civilizational collapse would directly lead to human extinction (within decades)?). To do this, I ask three key questions:

1. If we “re-ran” history, would we see the agricultural and industrial revolutions again? 

If a catastrophe caused a return to hunter-gatherer levels of society, we’d have to undergo the agricultural and industrial revolutions all over again to get back to our current levels of technological civilization. 

How likely is it that we’d overcome those hurdles again?

  • Because the first agricultural revolutions happened in multiple places following the stabilization of the climate after the last glacial period, I expect it’s very likely that we could expect to see subsequent agricultural revolutions within years (once the climate is suitable for agriculture).
  • I feel less confident, but still fairly optimistic, that another industrial revolution — which has only happened once in human history, and may have been the outcome of a variety of contingent factors — would be similarly likely to recur in a post-collapse world. 

2. Would technological progress look different in a post-collapse world?

My best-guess timescales for re-industrialization are greatly affected by the state of the world after the catastrophe, and to what extent survivors are able to take advantage of the remaining materials, technologies, skills, and knowledge for recovery. 

Dominant macroeconomic theory posits that modern economic growth is fueled by:

  • Natural and physical capital — natural resources like fossil fuels and water and timber, and the things we make out of those resources.
  • Human capital — the knowledge, skills, and people to do and invent things, plus the institutions and culture that enable and incentivize people to do and invent those things.

I use this framework to take something of an inside look at whether there would be any major barriers to achieving modern levels of economic output and growth in a post-collapse world. 

My best guess is that:

  • Natural resources seem likely to slow recovery down (there’ll be fewer accessible natural resources around).
  • Leftover physical capital seems likely to accelerate recovery — there’ll be stuff survivors can use (screws, plastic, vehicles) and learn from (engines, books, generators) lying around that early humans didn’t have. 
  • I expect human capital will also speed recovery up. Survivors will keep and pass on knowledge (memory, books) that should speed recovery up drastically, and human populations should be able to rebound quickly. I suspect cultural values, norms, and institutions would speed recovery up as well, but I found that much harder to reason about, so that feels more speculative.

3. What are the recovery timelines for a collapsed civilization?

Given that it took 200,000 to 300,000 years from the emergence of Homo sapiens to get to technological society, I consider 250,000 years a rough upper bound for thinking about how long recovery of current levels of technology would take if we lost the knowledge and skills critical to rebuilding civilization quickly (within 5–10 years of the initial catastrophe). 

I expect the physical and human capital that survived the catastrophe to enable recovery to happen approximately 3–90X faster than humanity’s initial industrialization, which translates to a best-case guess of 100–3,700 years, and a pessimistic guess of 1,000–33,000 years. 

If we think recovery time is limited by natural extinction threats, and we estimate an annual probability of natural extinction based on the length of time the genus Homo has survived natural threats (1/870,000), we can put the odds that humanity would recover technological civilization between 97% and 99.99%.


If we don’t go extinct, but we go longer than a human lifespan to recover, recovery looks much harder. Once all of the survivors who saw first-hand how human systems worked (and who ran those systems) have died, it seems much more likely that we’d lose much of the technology, complexity, and maybe even values of modern society. 

It’s not obvious exactly how far we’d get knocked back. Given this uncertainty, I’ll explore the most plausible pessimistic world: one in which we lose industry and domesticated agriculture and are reliant on hunting and gathering. To get back to technological civilization from hunter-gatherer levels again, we’d have to undergo the agricultural and industrial revolutions — technological advancements seen as pivotal to the human success story — all over again. Should we expect the descendants of the catastrophe’s survivors to overcome these hurdles a second time? Or would progress stall before they got to present levels of technological advancement?

We can start to think about this question by asking whether we would expect to see the agricultural and industrial revolutions again if we “re-ran” history. Obviously, this doesn’t completely reflect the reality in which those descendents would be trying to re-establish society. Even 100 years after an initial catastrophe, physical infrastructure that survived the initial catastrophe will only just be beginning to decay. Nonetheless, considering the extent to which the agricultural and industrial revolutions were inevitable is a good place to start. 

If we re-ran history, would we see the agricultural and industrial revolutions again?

The agricultural revolution happened in multiple places and at different times in human history, suggesting that the requisite discoveries and norms/values aren’t particularly rare (Price, 2009). 


Notably, these all happened at around the same time[1] in history, which could hint at the existence of “special circumstances” — circumstances that might be prerequisites for the emergence of the domestication of plants and animals. 

The leading hypothesis for why so many isolated regions had agricultural revolutions around the same time is that the domestication of plants required a stable and temperate climate — such that they could only come about after the end of the last glacial period (around 11,700 years ago). 

Given this, I would guess we could expect to see subsequent agricultural revolutions as long as the climate stayed human-friendly.

Was the industrial revolution similarly inevitable?

Unlike the agricultural revolution, we’ve only seen one industrial revolution.

Unfortunately, the fact that the industrial revolution only happened once (and only in the context of 18th century Great Britain) doesn’t tell us much about whether the industrial revolution was ‘special’ or not. One could assert that its uniqueness means that the factors necessary for something like an industrialization are very rare, and would only happen all at once on the scale of tens of thousands of years. 

On the other hand, given the fact that the world was already quite globalized by the beginning of the industrial revolution — there was a huge amount of transmission of ideas and goods, such that the successful elements of the industrial revolution spread widely relatively quickly — you would only have expected to see the industrial revolution emerge once. The fact that the agricultural revolution emerged in multiple regions is useful evidence that the ingredients that precipitated it weren’t unique and could be expected to happen again — but we would only expect to see this type of evidence in a world with isolated civilizations that couldn’t easily transfer ideas and technology. 

What’s more, some economic historians have argued that we got very close to an industrial revolution several times before the British industrial revolution. 

It’s been argued that the Roman Empire could have sustained something like an industrial revolution. Between the 1st century BCE and the 1st century CE, Rome experienced rapid population growth, increased coal production (and pollution), rising demand for consumer goods (including animal meat), and extensive trading — all considered precursors of sustained economic growth. The Romans also recognized the concept of property rights, which incentivized investment and trade.

Similarly, Eric Jones (author of Growth Recurring: Economic Change in World History) contends that China saw the beginnings of an industrial revolution during the Northern Song Dynasty (960-1127), during which the Chinese invented type printing, the blast furnace, mechanical water clocks, paddlewheel ships, the magnetic compass, water-powered textile machinery, and ships with watertight bulkheads that could carry 200-600 tons of freight and around 1,000 crewmembers. The result was increased agricultural output that exceeded population growth — which itself may have more than doubled in some regions during the 11th century CE[2] — and an explosion in the amount of cast iron produced.

But this so-called Chinese industrial revolution stalled in the mid-13th century with the fall of the Song Dynasty. 

Could these periods of industrialization easily have continued in some alternate version of history — perhaps leading to a period of growth comparable to the industrial revolution 1,000 or even 2,000 years earlier? Or did the growth stall precisely because both societies were missing some rare key prerequisite that only emerged in the British Empire? 

If the former, we can probably rest assured knowing that a collapsed society would likely re-industrialize pretty quickly. If the latter, if we got lucky with the British industrial revolution, we might worry that the necessary conditions wouldn’t arise quickly enough for the post-collapse civilization to re-attain modern economic growth trajectories. 

There’s no consensus among academics here. 

Some academics believe that market expansion — a growing population with sufficient income to demand consumer goods — was enough to lead to modern economic growth. These academics take the view that the conditions were ripe for industrialization several times in history, especially during the Roman Empire and Song Dynasty. Under such a view, the fact that we didn’t see something like the industrial revolution earlier is just bad luck. 

Other academics think that an abundance of natural resources was critical, such that an industrial revolution was only really possible in extractive societies, like Roman or European colonial empires. 

A third set of academics believes that the industrial revolution was even more contingent than that — requiring institutions and a culture which incentivize innovation in addition to an expanding market and abundant natural resources. Under this view, the conditions that led to the industrial revolution in Great Britain were more unique. 

For example, one of the camps in this third group[3] points to a host of contingent factors — all of which created a unique environment under which industrialization was possible — such as:

  • Physical and natural capital, including enormous benefits from the colonies, plus access to abundant sources of coal. On top of this, some have argued that the Black Death led to huge increases in individual wealth, as the deaths of between 23% and 60% of the European population left survivors with as much as double the resources they had previously.[4]
  • Human capital, not in the form of cheap labor (which, again, would have been greatly reduced because of the Black Death), but in the form of reasonably well-educated thinkers with the capacity to innovate. 
  • Economic incentives for innovation, created in part by cheap natural capital (especially energy) and high wages (itself the result of a combination of the labor shortages and increased individual wealth that followed the Black Death). This combination of cheap energy and high wages fueled incentives to invent technology that would mechanize labor, leveraging the availability of cheap energy and reducing reliance on expensive human labor. 
  • A culture of innovation, characterized by the “enterprising spirit,” which fueled inventions that multiplied the productivity of human laborers many-fold (and the higher wages of workers allowed them to consume more goods than ever before). On top of this, the embracing of the scientific method in the wake of The Enlightenment. 

To be conservative, we can adopt the view of the third group: the British industrial revolution was the result of a special set of circumstances that may arise infrequently enough as to be troubling from the perspective of civilizational recovery. Under this view, none of these other civilizations — the Roman Empire, the Song Dynasty, the Golden Age in Holland, nor any of the other so-called “efflorescences” — could have just as easily ended with a full-fledged industrial revolution. Instead, the British industrial revolution was truly the result of a special set of ingredients never seen by humanity before. 

But even accepting this view, it’s not clear just how infrequently we should expect those special ingredients to arise. It could be that the combination of circumstances isn’t inherently improbable in the scheme of things — that we should actually be surprised that we didn’t see them materialize until the 18th century. On the other hand, we might think the combination is extremely improbable — the product of dozens of low-probability events — such that we should be surprised that the conditions arose as soon as they did. In other words, we got really, really lucky that the industrial revolution happened as soon as it did. 

This is the problem with trying to draw base rates from historical events with an n of one. Gregory Lewis put it well when he wrote that reality is often underpowered.

Given this problem, I’ll consider 10,000 years my best guess for the frequency over which we should expect to see the conditions necessary for an industrial revolution to occur if history were run over and over again (this is something like my outside view base rate). If we make a few assumptions, we can estimate a 90% confidence interval for the frequency with which we’d expect to see the conditions necessary for industrialization arise. 

To do this, I assume that the conditions for industrialization are represented by an exponential distribution, and set the parameters to reflect a reality in which industrialization happened in 50% of “re-run histories.” When I do this, I find that industrialization should occur between 500 and 30,000 years after the emergence of agricultural civilization in 90% of “re-run histories.” 


But to be conservative, I’ll also give a pessimistic range. If we assume that the conditions for industrialization are represented by an exponential distribution, and set the parameters to reflect a scenario where we think those conditions were fairly unlikely, such that we would only expect to see the British industrial revolution as soon as we did in 10% of “re-run histories,” then we would expect to see industrialization arise 90,000 years after agricultural civilization in expectation (90% confidence interval: 4,400–280,000 years). 

But as I’ve alluded to above, there may be good reasons that the post-catastrophe conditions would be sufficiently different from the emergence of the British industrial revolution to drastically change the feasibility and timescales over which re-industrialization might happen. 

Would things be meaningfully different in the post-collapse world?

Are there ways that the post-collapse world would differ from the pre-collapse one, such that industrialization and further economic growth would happen slower or faster than it did the first time around? 

The inside view: What are the necessary inputs for modern economic growth? 

Dominant macroeconomic theory posits that modern economic growth is fueled by:

  • Natural and physical capital — natural resources and the things we make out of those resources.
  • Human capital — the knowledge, skills, and people to do and invent things, plus the institutions and culture that enable and incentivize people to do and invent those things.

I use this framework to take something of an inside look at whether there would be any major barriers to achieving modern levels of economic output and growth in a post-collapse world.

Natural and physical capital: Would we have the stuff we’d need to rebuild technological society?

Lewis Dartnell, a researcher and author of the book The Knowledge: How to Rebuild Our World from Scratch, looked into the feasibility of rebuilding civilization after a global catastrophe. He explored this topic from the perspective of the physical and natural capital that would remain after a collapse caused by a pandemic. The following summarizes the parts of Dartnell’s book I think are most relevant to rebuilding civilization. 


If the environment was suitable for it, survivors would have basically everything they’d need to practice some form of agriculture.

Civilization is built on specialization: the fact that humans develop specialized skills allows them to accomplish more (meet their basic needs, and then much more) than if they each had to meet all of their own needs using a generalized skillset. And a crucial step toward re-achieving specialization is improving the efficiency of agriculture so that labor is freed up to pursue other skills. 

In the short term, agricultural efficiency could be achieved with fairly basic knowledge and tools. Fundamental aspects of agriculture that survivors would need to re-acquire and re-master are: 

  1. Locating, synthesizing, and building the ingredients of productive agricultural systems: seed varieties for the most important crops; fertilizers containing nitrogen, phosphorus, and potassium; and tools and machinery to reduce the labor required to plant and harvest the crops using as little labor as possible. 
  2. Fundamental understanding of agrobiology: knowing how to make those fertilizers, which crops to grow in which soil types, the fundamentals of crop rotation to allow for the replenishment of nutrients.

The stuff survivors would need to rebuild efficient agriculture

  • If agriculture had been completely lost during the grace period, survivors would need to find seeds for heirloom crops in one of the many seed vaults located across the world. 
  • Crop productivity is boosted substantially by the addition of fertilizers, the main ingredients of which are nitrogen, potassium, and phosphorus. 
    • Potassium can be extracted from wood ashes. Manure is nitrogen-rich, and even human excrement can be treated (covered in sawdust or straw) and composted to be used as manure (as is done in e.g. India and Austin, Texas). 
    • Phosphorus is the most difficult nutrient to access, in part because, unlike nitrogen and potassium, much of the phosphorus in fertilizer is lost in runoff. At relatively small scales, phosphorus can be added to fertilizer by sprinkling in crushed-up animal bones.
    • Eventually, survivors will need to work out how to react bonemeal with sulfuric acid to make it more easily absorbed. Synthesizing sulfuric acid requires relatively complex chemistry techniques, though the necessary materials — pyrite rocks (fool’s gold) and sodium chloride (which can be extracted from wood ashes) — would be readily available, and the chemistry techniques aren’t prohibitively difficult. 
    • To eventually reach industrial-scale agriculture, survivors would need to relearn enough geology to extract phosphorus and potassium from mineral deposits (as is done today). Contrary to some claims, phosphorus continues to be abundant enough that it would not meaningfully limit agricultural production in the next several centuries.
  • Soils can be further enriched by using crop rotation. Crop rotation schemes involve rotating the types of crops planted to take advantage of their different biological properties (nutrient composition, helping with pests/weeds, etc.).
  • Survivors would need to relearn the use of agricultural tools — first basic tools like the hoe and scythe, and then later, the plow and combine harvester. The simpler tools could be easily fashioned from basic materials, while the more complicated ones would have to be scavenged from museums or forged from scratch (the process of forging could be relearned, as I discuss below).

At this level of agricultural sophistication, none of which is prohibitively difficult or resource-constrained, survivors would approach the level of efficiency reached following the British agricultural revolution. If done well, this level of agriculture would require just one agricultural worker to support five others — in other words, one person could free up five people to pursue other activities.

But survivors would need, at a minimum: 

  • A basic understanding of agrobiology.
  • The ability to do some kind of advanced chemistry.
  • Likely, some level of competency in building mechanical tools (for example, with gears, to build a seed drill) and in blacksmithing (for example, to forge a plowshare). 


Cooking is important for a bunch of reasons, like killing parasites and bacteria that would make us sick, as well as making the nutrients in food easier to absorb. From a technical perspective, maintaining or re-attaining the ability to cook food with heat seems pretty simple. During the grace period, matches, lighters, camping fire starters (which don’t even require fuel), and camping stoves could be used to heat food. As those supplies became harder to find, survivors would relearn how to light fires without modern aids — not an easy feat, but certainly doable.

Most people get a majority of their calories from grains, some of which can be eaten with minimal processing (like corn and rice), while others require a lot of processing (like cereals, which have to be milled). Milling is very time-consuming but not technically difficult. From there, survivors can make flatbreads without yeast (think naan and tortillas), or can scavenge or culture yeast to make leavened breads. 

In addition to cooking, survivors would need to relearn the fundamentals of food preservation, which would be critical for maintaining a food supply that can be kept pretty consistent across the seasons. There are lots of ways to preserve food, including drying, salting, smoking, adding sugar, adding acids (e.g. vinegar), and fermenting, among others — and civilizations have developed these techniques repeatedly throughout history. The fact that there are so many approaches means that survivors in different environments would be able to learn/practice at least some of the techniques with relative ease: desiccation would be easy in hot/dry environments though it’s a bit inefficient, salting in areas near the ocean, smoking basically anywhere (though it’d take some work to figure out how to build a smokehouse), adding sugar in regions where sugarcane is grown, adding vinegar/fermenting basically anywhere. 

We can also achieve food preservation without altering its chemistry by refrigerating it. Modern refrigerators require electricity, but there are other refrigeration techniques that don’t (e.g. absorption refrigerators), and could pretty easily be built without advanced technology/systems.   

And survivors could relearn to fire clay as a complement to these preservation techniques, allowing them to make lidded ceramic receptacles (and bricks for construction). To make watertight and air-resistant ceramic, the survivors would have to build kilns from scratch. This is somewhat difficult, because the kiln itself has to be built from materials that can withstand extreme heat. The neat thing here is that you can build a basic kiln from wood-fired clay bricks, then build a more robust kiln with the kiln-fired bricks and so on — an example of how learning basic techniques and principles will have compounding effects, allowing for even more complex ones.


Throughout most of history, infectious disease has been one of the main causes of human mortality. If the survivors of the catastrophe maintain the basic principles of germ theory, human longevity likely won’t be reduced to pre-industrial levels. That’s because many life-saving practices are quite low-tech — achievable with very basic inputs and knowledge. 

The main prevention and curative techniques to master will include water purification, sanitation, and hygiene; oral rehydration therapy; and eventually, the production of penicillin. 

Early on, water purification can be achieved by boiling it, leaving water in a clear container in the sun, or by making a water filtration system (this can be done using simple materials: a bucket, charcoal, sand, and gravel). Sanitation refers to the use of latrines (easily built) to keep defecation away from drinking water and food. And hygiene mostly refers to washing hands with soap, which can be made easily by mixing plant oils (coconut, olive, palm, etc.) with an alkali like sodium or potassium carbonate (these alkalis can be extracted from the ash left after burning wood or seawood, or by fermenting urine to get ammonia). 

Historically, diseases that cause diarrhea have been among the most deadly, as sufferers regularly died of dehydration. Oral rehydration therapy is an enormously effective intervention, and can be made quite easily by mixing sugar, salt (from the ocean or salt deposits), and pure water.

Finally, somewhat more difficult is the extraction and refinement of penicillin from Penicillium, which is actually one of the most common molds found in the environment, and the most common cause of food spoilage.

Sadly, childbirth will almost certainly become a leading cause of death again, though survivors that are able to recover birthing forceps from society’s ruins will be much better off. Other interventions can be improvised using materials left behind after the catastrophe. For example, Design that Matters has designed a makeshift neonatal incubator that can be built almost entirely from scavenged automobile parts, including headlights, the dashboard fan, and motorcycle battery. 


Keeping vehicles running will be possible for at least decades, and maybe indefinitely, though roads will become more and more difficult to navigate over time. Ethanol (fermented vegetables) or methanol (fermented wood) can be substituted for gasoline and diesel when those run out or expire, though the amount of land required to keep any one car running would be immense. Alternatively, survivors can pump gas to the engine, either from gas-filled bags or from wood gasifiers that sit on top of the vehicle, the way many European countries did during the fuel shortages of the first and second World Wars (one of many examples from the World Wars where we see humans successfully innovating around resource shortages). 

While spare engine parts and tires can be sourced from abandoned vehicles, landfills, and tire stockpiles, both the vulcanization of rubber (the process that makes tires so indestructible) and the precision manufacturing required for certain car parts could be inaccessible to survivors for some time. To maintain mechanized automobiles, humans will have to relearn metalworking and organic chemistry.[5] Until the survivors are able to source new rubber, vulcanize it, and produce/shape the alloys necessary for the car engine, they’ll have to rely on “analog” transport: horses, oxen, and wind-powered ships. 

There are no technical reasons this is unachievable (though, importantly, rubber is only produced in a few places in the tropics, so would require sufficient trade). The ingredients would be readily available, and the tools and techniques can be constructed and mastered from a very low baseline.[6] An (imperfect) proof of concept comes from a machinist who built an entire metalworking shop in the 1980s using just clay, sand, charcoal, and scrap metal. This feat has now been accomplished many times over as others have been inspired to do the same (you can now buy a seven-book series on building your own metalworking shop from scratch on Amazon).

Making stuff

And of course, the value of metalworking goes far beyond facilitating transportation. Metalworking would be necessary for making forging tools, gears, and a bunch of other things. 

In addition to metalworking, there are other critical skills we’d need to relearn. Importantly, just a few key ones (described below) would unlock a huge number of goods and tools.

Glass for windows and lightbulbs can be made by heating sand in a kiln, and blown using basic techniques that have been practiced for millennia. 

Producing lime (chalk) and quicklime are critical. Both are upstream of a bunch of different processes, like making mortar and concrete, whitewashing walls, treating sewage, improving soil efficiency, and making explosives (for mining). At small scales, survivors near coastlines can produce lime by simply crushing up coral and seashells. At larger scales, survivors will have to identify and mine chalk deposits. Quicklime is then produced by roasting lime in a kiln and adding water. 

Synthesizing key acids to be used in fertilizers, battery acids, explosives, ink, bleach, detergents, lubricants, and synthetic fibers (among other things) will also be key. Unfortunately, sulfur — the main input into sulfuric acid (the most-produced acid in today’s chemistry industry and a precursor to many other useful acids) — will be much harder to find than it was the first time around. Most of the sulfur in easily accessible volcanic deposits has already been extracted. There’s a solution though, as sulfuric gas can be created by cooking pyrite rocks (fool’s gold), which can then be reacted with chlorine gas and charcoal. This process requires a well-developed understanding of organic chemistry, but getting hold of the inputs themselves wouldn’t be insurmountable. 

Wood pyrolysis, which literally means burning wood, is probably the most difficult technique survivors will have to master early on. Survivors will need to build a system to burn wood and separate and capture the resulting vapors. From Lewis Dartnell’s The Knowledge

An ideal stepping-stone for a recovering society to leapfrog would be to bake wood in a sealed metal compartment, with a side pipe drawing off the released fumes and coiling through a bucket of cold water to cool and condense the vapors… The collected condensate readily separates into a watery solution and a thick tarry residue, both of which are complex mixtures that can be teased apart by distillation. The watery part is acetic acid, acetone, and methanol… The crude tar sweated out of the roasted wood can also be separated by distillation into its major constituents: thin, fluid turpentine, thick dense creosote, and dark viscous pitch.  

But the products are incredibly useful: acetic acid can be used to pickle food; acetone can be used for degreasing and in explosives; pitch can be used to make torches and to make things water resistant (caulking seams), and methanol can be used as antifreeze, as a biofuel, and as a chemical solvent (as can turpentine, another product of wood pyrolysis). 


The simplest communication tools we’ll relearn to make are paper — which can be fashioned out of wood pulp and the alkalis discussed above (made from wood or seaweed ash) — and pen, which can be fashioned from a bird’s feather. Ink made of plant dyes or berries can be made permanent using an iron compound plus plant galls (the growth a tree forms when a wasp lays its eggs on it). And once we’d relearned metalworking as described above, we could make a printing press (again, the Chinese did this over a millennium ago). 

And for communication at a distance, we could make makeshift radios — a task that is surprisingly accessible once we re-figure out electricity (more on this in the next section). Again, an example from the World Wars demonstrates the power of human ingenuity when particular resources are in short supply. Lewis Dartnell explains in his book: 

“During the Second World War, both soldiers holed up at POW camps built their own makeshift radio receivers for music or news of the war effort. These ingenious constructions reveal the sheer variety of scavenged materials that can be jury-rigged to create a working radio. Aerial wires were slung over trees, or disguised as clotheslines, and sometimes even barbed wire fences were appropriated for the task. A good grounding was achieved by connecting to cold-water pipes in the POW barracks. Inductors were constructed by winding coils around cardboard toilet rolls, the scavenged bare wire insulated by candle wax, or in Japanese POW camps by applying a paste of palm oil and flour. Capacitors for the tuning circuit were improvised out of layers of tinfoil or cigarette-pack lining, alternating with newspaper sheets for insulation; the wide, flat device was then curled like a jelly roll to make a more compact component. The earphone is a trickier component to improvise and so was often salvaged from wrecked vehicles…. Perhaps the most ingenious improvisation of all, however, was in creating the all-important rectifier, needed to demodulate the audio signal from the carrier wave. Mineral crystals like iron pyrite or galena were unobtainable on the battlefield, but rusty razor blades and corroded copper pennies were discovered to serve just as well. The blade or coin was fixed to a scrap piece of wood alongside a safety pin bent upright. A sharpened graphite pencil was firmly attached to the point of the safety pin… allowing fine readjustment of the pencil graphite across the metal oxide surface until a working rectifying junction was found.”


Soon after the collapse, I expect survivors to work out how to harness mechanical energy using water wheels, windmills, and steam engines,[7] each of which can be built with rudimentary tools and don’t require refined gasoline, diesel, or natural gas. 

To begin to generate and distribute electricity again, the survivors will relearn how to make batteries, generators, and electrical transformers. In all likelihood, survivors won’t be working from scratch. Another example from the context of war shows how ordinary people have figured out how to improvise complex devices in the face of resource scarcity. During the Bosnian War, civilians jury-rigged floating generators using scavenged car alternators and other basic parts.


But figuring out how to create and harness electricity from scratch wouldn’t be prohibitively difficult even if survivors had to do so from zero. Electricity is one of those things that seems to most people like magic, especially because it took humans so long to be able to make and use it reliably. It’s actually not (and in fact, there’s evidence that batteries were developed and used to electroplate gold onto jewelry as early as 200 BCE). Electricity can be generated by spinning a magnet inside a wire coil, and electric motion can be generated by placing a magnet near a wire. Particular arrangements of wires and magnets scaled up is the basis for the electric motor that drives the majority of our electronics today. 

And once survivors understand electromagnetism well enough, they can build generators that harness the mechanical power from windmills and waterwheels. As a proof of concept, researchers fitted a traditional windmill with a generator and found that the system produced >50,000 kilowatt hours (kWh, a measure of energy consumption) of electricity per year. Water turbines, the technology underlying the Three Gorges Dam and Hoover Dam, are even better at this. This energy can then be distributed across a system of cables and electrical transformers, neither of which are particularly difficult to produce once the survivors have mastered metalworking. 

But to return to modern technological society will require a lot of energy. It takes about 90,000 kWh of energy to support the lifestyle of the average US citizen for a year, and 40,000 kWh of energy to support the average European. Lewis Dartnell points out that, to achieve this level of energy consumption during the Middle Ages (so using only human and horse power), it would require the physical labor of 14 horses and >100 humans for 24 hours a day, 365 days a year.  

Several key processes survivors will need to recreate — producing steel, brick, mortar, and cement — require huge amounts of thermal energy, much of which we get from burning fossil fuels. It’s been pointed out that the efficiency of fossil fuel extraction has gotten worse and worse as we’ve already depleted some of the easiest-to-reach sources. 

While this is true, we should still have more than enough energy to re-industrialize, even if we don’t initially have the technology necessary for advanced fossil fuel extraction like fracking. To illustrate, consider that in 2018, the North Antelope Rochelle Mine in the United States (the largest coal mine in the world) produced about 100 million tons of coal — double the amount produced by Great Britain in 1850, 100 years into the industrial revolution. And the North Antelope Rochelle Mine is a surface mine, meaning that we can get at the coal without particularly advanced technology. 

North Antelope Rochelle Mine

Further, fossil fuels could be supplemented with charcoal (pyrolyzed timber), which is superior to coal in a bunch of different ways (it burns hotter, it’s renewable, it produces fewer pollutants) but which became less popular than coal during the industrial revolution because coal was easier to access than timber. Survivors in areas suitable for growing timber (or naturally forested ones) could produce massive amounts of charcoal. 

Given all of this, natural resources seem likely to slow recovery down, while leftover physical capital seems likely to accelerate it. My guess is that physical capital wins, and these factors would push recovery faster on net, but it’d be great if someone else gave this more thought. For now, I’m guessing the multiplier is somewhere between 0.5X and 1.5X. 

Importantly, while we might have enough natural resources to recover civilization once or even a few times, if human civilization collapsed many times, it’s much less obvious to me we’d have the natural resources necessary to rebuild — at least using technologies remotely similar to the ones we have now. 

What about human capital? Would human capital pose a bottleneck on the recovery of industrial civilization like natural resources would? Would it speed it up, like physical capital?

Would we have the human capital to rebuild industrial society?

For the purposes of this post, I consider human capital to include these factors:

  • Knowledge and skills
  • Education
  • Population
  • Institutions
  • Cultural norms and values

I expect several of these to persist beyond the collapse, thus accelerating the timeline over which we should expect society to re-industrialize. 


In a world where 90-99.9% of the population has died, it’s possible that no one left alive would have some of the requisite knowledge to actually implement the reconstruction described by Dartnell in the previous section. This seems extremely important, given that, historically, ideas and knowledge have been more of a bottleneck on economic growth than natural resources and physical capital.

For example, if we thought that the survivors and their descendents would lose even very basic knowledge of biology, they could fall into a Malthusian food productivity trap. In a food productivity trap, a population’s limited understanding of methods for boosting agricultural efficiency limits the size of the population able to be sustained. This in turn further limits improvements to agricultural output: people are the generators of ideas, so fewer people means lower probability of agricultural innovation. And perhaps more importantly, even if the population does conceive of testable innovations, because its agricultural output is just enough to support the existing population, no one is willing to jeopardize the existing outputs by testing new methods. 

Traps like these probably slowed us down historically, and it’s likely we’d get temporarily stuck in them if we lost much or nearly all of our knowledge. In a recovery scenario in which we have to rediscover all of the knowledge we’ve accumulated over thousands of years, achieving industrial civilization could easily be as slow or slower than it was the first time around.

I strongly suspect that agricultural innovation would not be a bottleneck on recovery. The fact that we saw multiple agricultural revolutions extremely quickly after the Earth’s climate stabilized makes me think that agricultural civilization is not incredibly contingent. But we don’t actually have to settle this particular debate, because, crucially, survivors rebuilding society after collapse wouldn’t have to do much innovation to re-attain agricultural complexity — they would have examples of the technology they were trying to recreate lying around, along with books explaining the biology and technology left in libraries. 

How much of this would survive an initial catastrophe? And how long would it last before even what remains would be destroyed?


Obviously, the physical stuff we’ve produced over centuries has widely varying lifespans. Structures built thousands of years ago are still standing, but modern skyscrapers would last less than a century, and our houses just a few decades. 

Consider that the ruins of ancient civilizations, while sparse, still tell us enormous amounts of information about how those civilizations lived: what tools they had, what they ate, even how their society was structured. 

And modern construction is more durable than ever before. Unlike ancient civilizations, we have concrete, steel, and iron. Things made of iron and steel would last longest, with some lasting as long as thousands of years. Reinforced concrete and stone would last more like several hundreds of years (with unreinforced concrete lasting a bit longer). Finally, brick and wood deteriorate the fastest, lasting only tens of years (wood), sometimes a hundred (brick).

There’s been a similar trajectory in most of our goods, many more of which are made out of durable goods like plastics and metals rather than wood or clay. 

Thus, we can probably expect to see something like the demonstration effect, whereby a group adopts a technique or practice after seeing another group with that practice.[8]

There are troubling exceptions — instances where a society or region lost knowledge or skills despite having examples to look at for guidance. W. H. R. Rivers wrote about these cases in his “The Disappearance of Useful Arts” (1912). He gives the example of the society that inhabited the Vanuatu archipelago, which lost the ability to build seaworthy canoes for generations (therefore becoming trapped on one of the archipelago’s islands, Torres Island). 

In an even more troubling example, human civilization mostly lost the ability to build concrete structures until 1824, despite the fact that the Romans figured out how to do it 2,000 years earlier — and indeed left the still-standing Pantheon and Colosseum as models future societies could study and replicate.

Crucially though, modern civilization has mastered something the Torres Islanders and the Romans hadn’t: books. We have virtually perfect copies of our knowledge in dedicated information warehouses all over the world. 


We have 2.6 million libraries in the world. If we imagine one of the most physically destructive catastrophes we’re aware of, nuclear war, we can get a rough sense of whether and how many libraries would survive. Given that there are 900,000 libraries in countries without nuclear weapons, and about 700,000 libraries in countries without nuclear weapons or nuclear alliances, we can feel confident that even in a very destructive catastrophe, many hundreds of thousands of books would survive the initial destruction.[9] 

Some books would be destroyed pretty soon after the collapse — burned up in fires and explosions,  disintegrated in rain and floodwaters, or just rotted by humidity. But books and microfiche in dry climates could last centuries. There are about 80,000 libraries in countries with extremely dry regions that would be unlikely to be involved in a nuclear war. With around 80,000 surviving libraries, it seems likely that at least some groups of survivors would be able to relearn a variety of basic knowledge and skills from a range of disciplines — things like germ theory, hygiene, crop rotation, even basic physics and chemistry. 

After that happened, those learnings would almost certainly become more widespread.

There are certainly many skills and practices we’d struggle to learn from books and artifacts. Much of the tacit knowledge required for important skillsets, like medicine or science, wouldn’t be easily learned through books. But I strongly suspect that the most useful knowledge — that which is directly relevant to our survival and population growth — can be learned pretty easily from books. And I think it’s more likely than not that we could learn a lot of more advanced science as well. 

All of that would be an enormous boost given the level of understanding held by human civilizations the first time around. I would expect this alone to speed recovery up by something like 1.1-2X. 


It seems exceedingly likely that, in a scenario where we’ve been knocked back to pre-industrial levels, we’d lose basically all formal education. However, as long as we maintain some form of communication — and I can find no plausible explanation for why we wouldn’t — the survivors would still have both verbal and written intergenerational knowledge transfer. 

Importantly, in his book on the role of collective intergenerational learning, The Secret to Our Success, Joseph Henrich points out that knowledge transferred in a small population (with only a few teachers) is liable to become diluted. As a younger generation learns from the older, most of the learners will be worse than their teachers (Henrich uses the analogy of a photocopy, which will typically be worse than the original). In a large population, some of the learners will outperform their teachers, either because they’re particularly diligent, or because they come up with an innovative augmentation of the process or skill. But in small populations, you’re less likely to see individuals who outperform their teachers, purely for probabilistic reasons. This can cause the degradation of knowledge over time. This is known as the Tasmanian effect, and it is a major barrier to the accumulation of knowledge over generations.

While the post-collapse population is small, I expect the Tasmanian effect to cause the loss of knowledge over time, especially tacit knowledge. But as the population grows, cultural evolution is able to counteract the Tasmanian effect. Over time, this informal transmission of knowledge will be formalized into something recognizable as education. Once the survivors developed formal education again, the gains from old and new knowledge and ideas would be further multiplied, accelerating growth even more. 

This points to another critical input to rebuilding technological society, and economic growth more broadly: a sufficiently large and growing population.


People are key to economic productivity and growth. There are certainly drawbacks to having larger populations: more pollution, more disease, and overcrowding, among others. But historically, these seem to be clearly outweighed by the benefits. Concretely, large populations mean more innovators, a market for consumer goods created by innovators, and economies of scale that ease the cost of supporting all of these people. 

By definition, population will fall during the collapse of civilization. And while this will initially have unintended positive consequences — fewer people competing for the leftover resources that will enable their survival — low population would be a detriment over time. 

However, given that I’m very optimistic that the survivors and their descendents will be able to relearn the basics of agriculture and medicine through books and by studying the ruins of the previous civilization (if they lose the basics at all, which I’m not sure they would), there shouldn’t be major constraints keeping population low. 

Importantly, while hygiene and basic modern medicine will improve physical health, the fact that the overall disease burden may be higher than it was in pre-modern society may push in the other direction.[10]

My guess is that human longevity and productivity (nutrition and frequency of illness) would still be much better, even taking into account a higher disease burden. This belief mainly comes from the fact that many of the staggering gains in life expectancy (from ~35 in 1800 to over 70 years today) can be attributed to low-tech public health improvements like improved public sanitation (toilets, handwashing) and medical interventions like the use of oral rehydration therapy and antibiotics. 

Some slightly higher-tech interventions like water chlorination, pasteurization, and sunscreen would be more technically difficult to recreate in a post-collapse society, but even just the understanding of the underlying science behind those interventions (water and milk have germs in them; exposure to the sun can be harmful) would get survivors and their descendents a fair bit of the way to re-attaining the ability to implement the interventions again. 

All of this makes me think that population will grow as fast (and likely much faster) than it did early on in human history, speeding up the second industrialization by 1.1-3X.

Institutions, norms, and values

Finally, it’s widely agreed that political, economic, legal, and social institutions have profound effects on economic growth. These include things like property rights, political stability, honest government, a dependable legal system, and competitive and open markets. It also includes even fuzzier things, like a culture of innovation and placing value on the scientific method. 

Interestingly, these may have been highly contingent, but also seem to be fairly long-lasting compared to physical stuff and technology. Indeed, hundreds of tools and technologies and even our understanding of the fundamentals of science have become obsolete over time. In contrast, many of our institutions and cultures can be traced back centuries and even millennia (consider how many of our norms have origins in religion, or that modern Western political institutions look a lot like early Roman ones). 

This makes me inclined to think that many important cultural, economic, and political institutions and values would survive a catastrophe. I expect that this would be a multiplier on the pace of recovery, though I’m very uncertain as to how large. I imagine it could be between 5X and 10X (if this seems like too much, consider how much quicker we could have made it through the Middle Ages if we’d understood and applied the scientific method at the time). 

But to be clear, this view is very speculative. 

One thing some people have wondered about is whether a catastrophe caused by technology could alter humans’ cultural acceptance and embrace of technology — aspects of culture key to rebuilding industrial society. In effect, even if it were technically possible to do so, the survivors of a technology-fueled catastrophe might not want to rebuild a technological society.[11]

I don’t find this concern very plausible. For this type of cultural shift to actually stop technological advancement, it would have to be universally adopted by all survivor groups. We can point to examples of times when many actors have coordinated to prohibit the use of some technologies — genetically modified foods, for example, which have been banned in 26 countries  — but it’s pretty hard to think of any that have been universally banned. And for those that have been banned, I can’t think of any that don’t have some kind of functional substitute. For example, while blinding lasers were basically universally banned as part of the United Nations’ Convention on Certain Conventional Weapons, “flashers” (lasers that cause temporary blindness) have been developed as a way around the ban

This points at another reason technological society would be hard to avoid: because technological society could likely be built on a range of technologies, deciding not to pursue some technologies isn’t enough. For example, if society collapsed as a result of anthropogenic global warming, the survivors could decide not to redevelop technologies that rely on fossil fuels. However, as described above, the elimination of fossil fuels wouldn’t necessarily make industrialization impossible — charcoal, hydropower, solar power, and wind power could be exploited to meet much of the successor society’s energy needs. The use of alternative energy sources, and possibly the discovery of new ones, could still fuel another industrial revolution. Given this, the survivors of a catastrophe would basically have to decide to avoid all technology to ensure that technological advancement didn’t eventually arise again. 

I find it really difficult to imagine that all survivor groups come to a consensus on this. I also find it hard to believe that this could be enforced through coercion, given that coercion on a truly global scale can probably only be done really well with pretty high levels of technology (weapons and surveillance). 

But even if we imagine that the survivors have all universally decided not to advance technologically (or are successfully coerced), it must be the case that this value lasts indefinitely (i.e. until the accumulated natural risk gets us) — a degree of lock-in that, ironically, can probably only be achieved with advanced technology. Given that there are pretty unignorable short-term benefits of technological development, it seems quite unlikely to me that everyone everywhere would reject technology for centuries or more. 

All of this makes me think that a cultural aversion to technology would be unlikely to stick in the longer term. It could slow things down a bit in the short term, but I would be very surprised if this amounted to anything more than a slowdown of 0.9X (so I expect the effect is between 0.9X and 1X).

Given all of this, where do I put the probability of the recovery of industrial society?

What are the recovery timelines for a collapsed civilization?

Given that it took 200,000 to 300,000 years from the emergence of Homo sapiens to get to technological society, we can consider 250,000 years a rough upper bound for thinking about how long reconstruction would take if we lost the knowledge and skills critical to rebuilding civilization quickly (but again, we only have an n of one — it actually might have happened much slower or much faster if we re-ran history again).

Again, I expect that the survivors would achieve something like agricultural civilization in a matter of years given a stable climate (which seemed like the main precipitator of the first agricultural revolutions). Based on this, I would consider just the time it took to get from agricultural civilization to industrialization a reasonable lower bound on how long reconstruction of industrial society would re-industrialize. 

This brings me to:

  • about 10,000 years in expectation in a best-case scenario (90% confidence interval: 500-30,000 years), and 
  • about 90,000 years in a pessimistic scenario (90% confidence interval: 4,400-280,000 years).

When I then take an all-things-considered view (which I admit is very speculative), I expect that recovery could happen approximately 3–90X[12] faster than industrialization. This translates to a best-case guess of between 100 and 3,700 years, and a pessimistic guess of 1,000 to 33,000 years. 


If we think recovery time is limited by… 

Base rate — If we assume that re-industrialization would take exactly as long as it did the first timeInside view — If we assume that existing physical and human capital would accelerate the speed of re-industrialization relative to the base rate

Agricultural rev. and industrialization take as long as they did the first time (~300,000 years)

Agricultural civilization returns quickly, industrial revolution takes as long as it did the first time (500-30,000 years)


Best-case guess1

(100-3,700 years)


Pessimistic guess(1,000-33,000 years)

The natural rate of human extinction — Homo sapiens base rate





The natural rate of human extinction — Homo genus base rate 





End of the biosphere
(1/~1 billion years)[15]






1. Assumes the British industrial revolution happened about when we’d have expected (100-3,700 years)

2. Assumes we got very lucky with the British industrial revolution (1,000-33,000 years)






Why I might be wrong

If I’m wrong about whether we’d stagnate permanently as a result of civilizational collapse, my guess is that it’s because of one (or several) of the following:

  • My arguments/research are based on the slice of time we’re currently in — they don’t account for the ways resource availability and technology (and other things) could change in the future.
  • Catastrophes we don’t know about have different properties than the ones we know about (in their deadliness, the types of people they affect, etc.).
  • There are many key resource constraints that I don’t know about.
  • Relatedly, I’m underestimating the likelihood that survivors fall into a food productivity trap (and stay there).
  • I’m overestimating the degree to which survivors would be able to relearn critical knowledge/skills from books and physical artifacts.
  • There is knowledge/skills that can’t be learned from books/artifacts that turn out to be critical.
  • We got lucky with the industrial revolution. The circumstances that led up to it the first time around were actually exceedingly unusual, and it would probably take much longer to happen if we re-ran history.
  • Norms against technology following a catastrophe are much stickier than I imagine them to be.
  • I expect too much continuity of culture/values/institutions, etc. before and after the collapse.


  1. ^

    5,000 years may seem far apart in the timescales we’re used to thinking about, but it’s actually very close in time when thinking about the scale of humanity's lifespan.

  2. ^

    “Nevertheless, one view shared by both Western and Chinese literatures is that the Northern Song Period was marked by its rapid population growth. The main evidence comes from officially registered household numbers, which grew from 6.2 million households in 980 AD to 17.5 million households by 1101 AD, an increase of 280 per cent…  An independent check is available to support this growth rate. According to the government record, from 995 to 1078 the Songs total marketed salt increase from 373,545 xiaoxi (small units) to 739,620 daxi (large units), or from 43.5 million to 103.5 million catties, with an annual growth rate of one per cent. Salt consumption is both price and income inelastic. It is a reliable barometer for population growth.” (Deng, 2013)

  3. ^

    There are other theories that probably fall in this category. For example, Frey argues that we should be skeptical of the Black Death–caused industrial revolution given that we only saw the industrial revolution in Great Britain, despite the fact that other European countries experienced the pandemic as well. Frey’s preferred theory is one where British industrialization is explained by the British government’s backing of capitalists over workers (who were being displaced by the capitalists’ machines), whereas France’s government, for example, backed workers. 

  4. ^

    This hypothesis is quite controversial, with at least one empirical paper directly contradicting it. Gregory Clark explores the changes in wages, land rents, and returns on capital in the centuries before and after the Black Death (Clark, n.d.). Clark finds that the gains from Black Death were short-lived, returning to pre-pandemic trends within a century.

  5. ^

    To vulcanize rubber, one has to melt the raw rubber, which itself may be difficult to acquire as it’s only grown in the tropics, and sprinkle in sulfur. Sulfur is hard to obtain from its most abundant source, pyrite rocks (fool’s gold), but can be done using two-pot distillation by heating the pyrite rocks to a high temperature. 

  6. ^

    Some tools would be more difficult to build from scratch than others. A simple but critical tool is the precision screw — considered an important enabler of the industrial revolution. Ideally, survivors should try to scavenge a long, threaded lead screw, as they are exceedingly difficult to cut from scratch, and can themselves be used to produce many more. Considering that the first precise metal screw thread took around 200-300 years of iterative improvements to make, this offers an example of how existing physical capital could enable a successor society to reindustrialize much more quickly than the initial industrialization process. Similarly, while our ancestors had to build smelting furnaces to purify the metals they worked with, the survivors of a collapse will be able to melt down scrap metal from abandoned cities. Once that scrap metal runs out, the survivors will eventually have to relearn how to build and operate blasting furnaces (as the Chinese did over a millennium ago), but this too will be made easier with the materials available for scavenging.

  7. ^

    Steam engines to transform between thermal energy and mechanical energy. Achieved by venting hot steam into the cylinder of a piston, which condenses from gas to liquid, reducing the pressure in the cylinder. The pressure from the atmosphere then pushes the piston down (the same mechanism that draws liquid up a straw when you suck the air out). Add a crank and you can perform rotational mechanical tasks like driving machinery or wheels. 

  8. ^

    Possible examples include the prehistoric diffusion of pottery techniques, artefacts brought back and distributed during Marco Polo’s expeditions to China, metallurgy techniques transferred between the East and West during the Middle Ages.

  9. ^

    Matthew van der Merwe pointed this out when reviewing this post: 

    As Toby Ord describes in The Precipice, physicists working on the Manhattan Project had unresolved concerns that the detonation of a nuclear weapon could ignite a fusion reaction with the hydrogen in the Earth’s oceans, or with the nitrogen in the atmosphere — either of which might have destroyed all complex life on Earth. They were unable to completely resolve these uncertainties before they conducted the Trinity Test (the first-ever detonation of a nuclear weapon).

    From pp. 331–2 of The Precipice

    James Conant, President of Harvard University, took the possibility seriously enough that when the flash at detonation was so much longer and brighter than he expected, he was overcome with dread: ‘My instantaneous reaction was that something had gone wrong and that the thermal nuclear transformation of the atmosphere, once discussed as a possibility and jokingly referred to a few minutes earlier, had actually occurred.’  … 

    When the war ended, he returned to Harvard and summoned its chief librarian, Keyes Metcalf, for a private meeting. Metcalf later recalled his shock at Conant’s request (Hershberg, 1995, pp. 241–2): ‘We are living in a very different world since the explosion of the A-bomb. We have no way of knowing what the results will be, but there is the danger that much of our present civilization will come to an end ... It might be advisable to select the printed material that would preserve the record of our civilization for the one we can hope will follow, microfilming it and making perhaps 10 copies and burying those in different places throughout the country. In that way we could ensure against the destruction that resulted from the fall of the Roman Empire.’

    Metcalf looked into what this would require, and prepared a rough plan for microfilming the most important 500,000 volumes, or a total of 250 million pages. But in the end they did not pursue this, reasoning both that its becoming public would cause significant panic, and that written records would probably survive in the libraries of university towns that would not suffer direct hits from atomic weapons.

    However, when Metcalf resigned from Harvard, he began a project of ensuring vast holdings of important works in major universities in the southern hemisphere, perhaps inspired by the conversation with Conant and fear of nuclear catastrophe (Hershberg & Kelly, 2017).

  10. ^

    The higher disease burden is itself caused by the domestication of animals, which has increased the rate at which diseases jump from non-human animals to humans. Consider that tuberculosis and smallpox came from cattle, the common cold came from horses, measles came from dogs and cattle, and many flu viruses still come from poultry and pigs.  

  11. ^

    Further, James C. Scott has argued that the early agricultural civilizations had worse living conditions (lower longevity, worse quality of life) for most people — perhaps suggesting that a transition from hunting and gathering to agricultural and industrial civilization is not desirable at the individual level.

  12. ^

    I get this by adding up all of the multipliers from earlier sections. Reminder that the multipliers represent acceleration of progress caused by the existence of additional physical and human capital.

  13. ^

    “Using only the information that Homo sapiens has existed at least 200,000 years, we conclude that the probability that humanity goes extinct from natural causes in any given year is almost guaranteed to be less than one in 14,000, and likely to be less than one in 87,000” (Synder-Beattie, Ord, & Bonsall, 2019).

  14. ^

    “Using the longer track record of survival for our entire genus Homo produces even tighter bounds, with an annual probability of natural extinction likely below one in 870,000” (Synder-Beattie, Ord, and Bonsall, 2019).

  15. ^

    “There is significant uncertainty in all of these estimates. We can be reasonably confident that the runaway and moist greenhouse effects pose an upper bound on how long life can continue to exist on Earth, but we remain uncertain about when they will occur, due to the familiar limitations of our climate models. Wolf & Toon (2015) find a moist greenhouse will occur at around 2 billion years, whereas Leconte et al. (2013) place a lower bound at 1 billion years” (Ord, 2020, p. 406). 

Sorted by Click to highlight new comments since:

This is such a fantastic piece of research - changed views or no, I'm so glad you've finally published it :)

A couple of follow-up questions:

I think the probability of technological stagnation is somewhat higher than I did

What do you mean exactly by 'technological stagnation'? Something like 'longer than your previous most pessimistic 33,000 years guess'?

If the answer to the previous doesn't cover this: what time ranges would you now guess, if you had to? ('cause if not you, then who?)

First of all, this is dope, great job two years ago you. 

I actually think you were probably originally pessimistic on the timescales. On your climate change update I haven't thought much about that stuff so can't comment. 

The main crux for me is if the period of ravenous competition after the collapse (assuming billions survive) will kill people in correlated ways (i.e. most of the people who would want to read books and rebuild things are murdered or outcompeted). 

The other main crux is if birth rates will actually go to back up from current. It seems possible to me that main reason we would go extinct is that people aren't even having replacement rate kids. This would definitely stop us from getting back to where we are as our global world requires a decent amount of people. 

In general I'm not really sure that re-running history is all that apt of a baseline approach though I think it's somewhat relevant information and I'm glad you thought through it. It would be much, much, much easier a second time around. You seemed to acknowledge this but didn't give it the weight that I would. People didn't understand basic physics or germ theory until incredibly recently (and on and on). We have decentralized electricity generation lying around everywhere. Books aren't the best way to learn, but they aren't bad. And there are books explaining everything. We'd pick up the important stuff if we wanted to. 

Technologically speaking, I'd assume we'd be well into the 20th century within 10 years of collapse (in the majority of different collapse situations), and in some ways we would have already returned to the 21st (e.g. we would probably be using computers again very quickly, though making them is a whole different story). I think looking at developing countries that leveraged catch up growth is a better baseline. 

Political and social organization is where things could change a ton, but it's unclear if we even have good  current versions of these things, so it might not be so hard to match or exceed current quality. 

Karma upvoted, agreement downvoted.

It seems possible to me that main reason we would go extinct is that people aren't even having replacement rate kids.

I can't see how that could plausibly be a reason for extinction unless we were already right on the edge. The problem just solves itself rapidly as cultural and actual evolution select for the people who have the most children.

Technologically speaking, I'd assume we'd be well into the 20th century within 10 years of collapse (in the majority of different collapse situations)

This seems incredibly optimistic. To get that level of technology we'd already used an enormous amount of cheaply available resources that subsequent civilisation has made much less available. For example, fossil fuels would be mostly gone, and rock phosphorus would be much harder to access - some of the energy from the former is available in the form of things we use it to construct on the surface, but it seems likely 90%+ of it would be gone. Rock phosphorus I'm less clear on, but we've flushed the majority of what we've used of it into lakes and oceans, and while plenty remains which is accessible to modern society, we generally take the easiest-to-access deposits first, so a rebooting society would find it substantially more expensive to get in quantity.

The problem just solves itself rapidly as cultural and actual evolution select for the people who have the most children.

You are definitely right here. I take back what I said, or at most I append it to saying that the length it takes for this evolution to occur is a crux of how long recovery takes. 

Re point 2: I think I might be a bit confused on what the end goal we are actually measuring is. Is the original post saying the amount of time it would take to push past our current level of technology, or until we have a stable society that is advancing technologically or until the world GDP per capita is approaching what it is today?

I'm not sure what I originally thought but now I'm having a tough time even understand what the time estimates are for. E.G. One might think Global supply chains could take decades - a couple hundred years to get back going as they once were but we might immediately be able to find new ways to gerry-rig much of the existing technology and upkeep it, even if we can't build more of it or permanently save it. We might even be able to incrementally improve some of it. 

Re point 2: I think I might be a bit confused on what the end goal we are actually measuring is. Is the original post saying the amount of time it would take to push past our current level of technology, or until we have a stable society that is advancing technologically or until the world GDP per capita is approaching what it is today?

I'm not sure exactly what Luisa had in mind, but I think the vagueness about what 'recovery' looks like after a catastrophe is a real problem in the x-risk community. I've got a couple of posts on a related subject in the works, which will hopefully be ready in a week or so.

Since Luisa is only considering natural sources of extinction during recovery, I assume that recovery means until the new time of perils when we have anthropogenic existential risks, so something like 1945. Some of this content is in her 80K podcast.

I listened to Luisa's podcast a while ago but don't really remember it so thank you for the clarification. I will go ahead and say that might be a bad definition to use. If anyone who had any knowledge about the nuclear protocol of any country with an arsenal survived, then plausibly we could launch dozens + nukes once we got a few basic functionalities up and running. Similar things could be said about AGI, if we imagine it has been invented before the collapse. I guess this does start to get really in the weeds of each countries nuclear protocall and how easy it would be to access them just based on the systematic defenses in place. I am completely out of my depth here. It's possible that without a bunch of passcodes or people who remember certain things and have access to certain doors it's all impossible. But on the other hand if most people are dead you can just start dynamiting parts of the pentagon until you find the right room...

( I think it's an open question if nukes actually pose an existential risk until you start launching absurd amounts of them, something Luisa herself has done research on IIRC. Still, i'm mainly going off of getting to the level of destruction we could have created in 1945)

It seems like it is the creation of new objects and technology that will take a lot of time. Weapons of mass destruction will still exist, and perhaps be even easier to access for pernicious uses (but like I said I don't have a clear idea of this).

Nice I’ll keep an eye out for the posts.

Thanks this is great!

My intuition strongly suggests that it would take less than a thousand years to rebuild industrial society, even starting from hunter-gatherer levels. 
Argument to articulate my intuition:
--It would take at most something like 100 years to get to something like ancient Athens, a primitive agrarian economy but rich enough that some people could afford to think and learn etc., and with a strong culture of science/discovery/engineering. The italicized part is the hard part, I think -- it happened in Athens and various other city-states but not in most of them. However,  100 years after the collapse the cultural knowledge of the Before Times would still be present, so I think the average city-state 100 years after collapse would have levels of science/discovery/engineering culture that would be at least as high as Athens.
--Ancient Athens, if it had loads of artefacts from an industrial civilization and diagrams lying around for how to build them, probably could have figured out to get there in 100 years or so. The bottleneck historically was NOT the sheer clock time it takes to go through the various stages (iteratively improving a set of tools and processes and accumulating resources which you use to build the next tech level, which you then improve to build the tech level after that, etc.) but rather the clock time it took to invent the new technologies. (A major restriction was that people didn't generally think inventing things was a worthwhile pursuit; the governments of the time thought that building huge monuments was a better use of resources. Well, if they thought they could be FIRING CANNONS in 50 years and FLYING IN AIRPLANES BOMBING THEIR ENEMIES in 100 years, if only they put loads of money and effort into science and tech, they surely would!)
--Conclusion: Estimated ~200 year timeline to rebuild civilization starting from industrial era.

This estimate is way more optimistic than yours, I wonder why. Which of the premises above would you disagree with? I'm guessing the second one?

Also I don't see why it would take even close to 100 years to get to ancient Athens. I see you said at most so possibly we are in agreement.

I feel pretty confident that my father, who is a city dweller like myself, but who loves gardening and is an avid reader, could produce 1000x-10000x his own needs in food within 5 years if he had access to nearby farmland, tractors/corn picking machines, gas, a well, and some relevant books.  I don't even think he would need electricity(though I'm not sure ). We live in Chicago, which is only a few days walk or less from literal industrial scale cornfields with all the equipment you would need and also a wealth of libraries with any book he might need about how to farm at scale. 

This is someone who has a pretty deep understanding of cellular biology and moderate understanding of botany as well as lots of experience gardening and reading tips that have taken 1000s of years to develop. He's somewhat unusual but It's not like he's an incredibly rare person. There are lots of people alive right now who understand plants wayyyyyy better than Mendel, let alone anyone in Athens. Moreover we have GMO'd plants and TRACTORS and drip irrigation, and we understand crop rotation, etc. etc. etc. These things won't disappear and neither will the infrastructure created by mega farms, which have automated the vast majority of what they do (depending on the plant, but in the case of corn for instance, there is nearly no labor involved).

I just don't see it. Either there is a 100 year period where going outside or establishing a colony will get you insta murdered or there will be lots of people who will quickly prosper to levels of wealth completely unseen by all but the wealthiest people ever alive during 300 BC. There needs to be some specific mechanistic explanation for why someone like my father would not be able to safely farm a cornfield. 

I'm beating a dead horse here but based on an incredibly quick search 80% !! of ancient greece was engaged in agriculture. Currently in the US it's something like 1-2 %. Let's assume we lose an order of magnitude of our current agricultural productivity per capita. Then we are still producing 4-8x the food as them per person. And this seems like an incredibly conservative assumption. Because if civilization collapsed, we wouldn't be farming oranges and bananas and spinach and berries, etc. We'd be farming corn and potatoes, which are both more calorie dense and extremely automated/optimized and thus trending way above the general efficiency ratios noted above.

With this abundance of food we could immediately spend the majority of our time studying modern civilization and rebuilding. It's at this point where I feel much less clear of the timelines.

Well, if they thought they could be FIRING CANNONS in 50 years and FLYING IN AIRPLANES BOMBING THEIR ENEMIES in 100 years, if only they put loads of money and effort into science and tech, they surely would!) 

I feel like you really hit the nail on the head here of where I think the original post went off track. 

My opinion is that the original post is significantly underrating the value of the current zeitgeist, which is related  to cultural norms and values but not the same thing. There will be this sense in the air of what things can become/return to, and it won't be some vague notion of progress, it will be a very specific notion of planes, capitalism, computers, etc. It will be massively important that people understand that they can get to these things. I'd even go farther than you and say it's not inventing the technologies that takes so much time, but the combination of having to invent them and not knowing which technologies to invent in the first place. If we told the romans that they should invent electricity I feel like it's not that crazy to think they would have figured it out (depending on how much we specify what electricity is).

That's why I said in my other comment that it might make sense to look at developing countries that leveraged catchup growth rather then rerunning history as a baseline. These countries had a similar Zeitgeist (i'm sort of just using this as a placeholder word until I can specify what I mean more) in this dimension. (some) People in these countries went to developed nations and saw what things could become. Of course it is also important they have some idea how to get there, but I don't see that as much of a crux, since there should be books lying around everywhere. 

It's great to see your reasoning. It seems reasonable for rerunning history. However, for something like fossil fuels and mineral resources being more expensive, it might not just be a delay in getting industry, but instead not having rapid enough economic growth to exceed population growth and therefore escape the Malthusian trap. Even worse would be potential permanent changes in humanity, such as a change in genetics due to radioactive damage (I don't think we have enough nuclear weapons to do that now, but in some future scenario, we might). Similarly, perhaps the massive mortality of the catastrophe could select for people with less desirable characteristics, such as lower trust. If the catastrophe involved the loss of the grass family, that would have excluded all agricultural civilizations except for the Andean one. There could also be an endemic disease that prevents high population density, which is basically required for civilization. You mentioned long-lasting anthropogenic climate change as being a significant risk, but naturally we could go back into an Ice Age or just have a generally less stable climate than the last 10,000 years.

I think the bigger picture is how much a catastrophe reduces our long-term potential of humanity. Though not the question you're asking here, I am concerned about even if we do recover technologically, that our values might be worse than now, and that could get locked into AI.

For all of these reasons, I think the reduction in long-term potential of these global catastrophes is more like order of magnitude 10%.

I wonder why "technological stagnation" isn't a topic already. Is there a way to recommend it to moderators?

Curated and popular this week
Relevant opportunities