TL;DR: Small probability of runaway global warming requires preparation of urgent unconventional measures of its prevention that is sunlight dimming.



The most expected version of limited global warming of several degrees C in 21 century will not result in human extinction, as even the thawing after Ice Age in the past didn’t have such an impact.

The main question of global warming is the possibility of runaway global warming and the conditions in which it could happen. Runaway warming means warming of 30 C or more, which will make the Earth uninhabitable. It is unlikely event but it could result in human extinction. 

Global warming could also create some context risks, which will change the probability of other global risks.

I will not go here in all details about nature of global warming and established ideas about its prevention as it has extensive coverage in Wikipedia ( and 

Instead I will concentrate on heavy tails risks and less conventional methods of global warming prevention.

The map provides summary of all known methods of GW prevention and also of ideas about scale of GW and consequences of each level of warming.

The map also shows how prevention plans depends of current level of technologies. In short, the map has three variables: level of tech, level of urgency in GW prevention and scale of the warming.

The following post consists of text wall and the map, which are complimentary: the text provides in depths details about some ideas and the map gives general overview of the prevention plans.


The map:



The main feature of climate theory is its intrinsic uncertainty. This uncertainty is not about climate change denial; we are almost sure that anthropogenic climate change is real. The uncertainty is about its exact scale and timing, and especially about low probability tails with high consequences. In the case of risk analysis we can’t ignore these tails as they bear the most risk. So I will focus mainly on the tails, but this in turn requires a focus on more marginal, contested or unproved theories.

These uncertainties are especially large if we make projections for 50-100 years from now; they are connected with the complexity of the climate, the unpredictability of future emissions and the chaotic nature of the climate.


Clathrate methane gun

An unconventional but possible global catastrophe accepted by several researchers is a greenhouse catastrophe named the “runaway greenhouse effect”. The idea is well covered in wikipedia

Currently large amounts of methane clathrate are present in the Arctic and since this area is warming quickly than other regions, the gasses could be released into the atmosphere.


Predictions relating to the speed and consequences of this process differ. Mainstream science sees the methane cycle as dangerous but slow process which could result eventually in a 6 C rise in global temperature, which seems bad but it is survivable. It will also take thousands of years.


It has happened once before during Late-Paleocene, known as the Paleocene-Eocene thermal maximum, (PETM), when the temperature jumped by about 6 C, probably because of methane. Methane-driven global warming is just 1 of 10 hypotheses explaining PETM. But during PETM global methane clathrate deposits were around 10 times smaller than they are at present because the ocean was warmer. This means that if the clathrate gun fires again it could result in much more severe consequences.


But some scientists think that it may happen quickly and with stronger effects, which would result in runaway global warming, because of several positive feedback loops. See, for example the blog 


There are several possible positive feedback loops which could make methane-driven warming stronger:

1)      The Sun is now brighter than before because of stellar evolution. The increase in the Sun’s luminosity will eventually result in runaway global warming in a period 100 million to 1 billion years from now. The Sun will become thousand of times more luminous when it becomes a red giant. See more here:

2)      After a long period of a cold climate (ice ages), a large amount of methane clathrate accumulated in the Arctic.

3)      Methane is short living atmospheric gas (seven years). So the same amount of methane would result in much more intense warming if it is released quickly, compared with a scenario in which it is scattered over centuries. The speed of methane release depends on the speed global warming. Anthropogenic CO2 increases very quickly and could be followed by a quick release of the methane.

4)      Water vapor is the strongest green house gas and more warming results in more water vapor in the atmosphere.

5)      Coal burning resulted in large global dimming And the current switch to cleaner technologies could stop the masking of the global warming.

6)      The ocean’s ability to solve CO2 falls with a rise in temperature.

7)      The Arctic has the biggest temperature increase due to global warming, with a projected growth of 5-10 C, and as result it will lose its ice shield and that would reduce the Earth’s albedo which would result in higher temperatures. The same is true for permafrost and snow cover.

8)      Warmer Siberian rivers bring their water into the Arctic ocean.

9)      The Gulfstream will bring warmer water from the Mexican Gulf to the Arctic ocean.

10)    The current period of a calm, spotless Sun would end and result in further warming.


Anthropic bias

One unconventional reason for global warming to be more dangerous than we used to think is anthropic bias.

1.  We tend to think that we are safe because not runaway global warming events have ever happened in the past. But we could observe only a planet where this never happened. Milan Cirncovich and Bostrom wrote about it. So the real rate of runaway warming could be much higher. See here:

2. Also we, humans tend to find ourselves in a period when climate changes are very strong because of climate instability. This is because human intelligence as a universal adaptation mechanism was more effective in the period of instability. So climate instability helps to breed intelligent beings. (This is my idea and may need additional proof).

3. But if runaway global warming is long overdue this would mean that our environment is more sensitive even to smaller human actions (compare it with an over-pressured balloon and small needle). In this case the amount of CO2 we currently release could be such an action. So we could underestimate the fragility of our environment because of anthropic bias. (This is my idea and I wrote about here:


The timeline of possible runaway global warming

We could name the runaway global warming a Venusian scenario because thanks to a greenhouse effect on the surface of Venus its temperature is over 400 C, despite that, owing to a high albedo (0.75, caused by white clouds) it receives less solar energy than the Earth (albedo 0.3).


A greenhouse catastrophe can consist of three stages:

1. Warming of 1-2 degrees due to anthropogenic C02 in the atmosphere, passage of «a trigger point». We don’t where the tipping point is, we may have passed it already, conversely we may be underestimating natural self-regulating mechanisms.

2. Warming of 10-20 degrees because of methane from gas hydrates and the Siberian bogs as well as the release of CO2 currently dissolved in the oceans. The speed of this self-amplifying process is limited by the thermal inertia of the ocean, so it will probably take about 10-100 years. This process can be arrested only by sharp hi-tech interventions, like an artificial nuclear winter and-or eruptions of multiple volcanoes. But the more warming occurs, the lesser the ability of civilization to stop it becomes, as its technologies will be damaged. But the later that global warming happens, the higher the tech will be that can be used to stop it.

3.Moist greenhouse. Steam is a major contributor to a greenhouse effect, which results in an even stronger and quicker positive feedback loop. A moist greenhouse will start if the average temperature of the earth is 47 C (currently 15 C) and it will result in a runaway evaporation of the oceans, resulting in 900 C surface temperatures. ( ). All the water on the planet will boil, resulting in a dense water vapor atmosphere. See also here:




If we survive until positive Singularity, global warming will be not an issue. But if strong AI and other super techs don’t arrive until the end of the 21st century, we wll need to invest a lot in its prevention, as the civilization could collapse before the creation of strong AI, which means that we will never be able to use all of its benefits.


I have a map, which summarizes the known ideas for global warming prevention and adds some new ones for urgent risk management.




The map has two main variables: our level of tech progress and size of the warming which we want to prevent. But its main variable is the ability of humanity to unite and act proactively. In short, the plans are:

No plan – do nothing, and just adapt to warming

Plan A – cutting emissions and removing greenhouse gases from the atmosphere. Requires a lot of investment and cooperation. Long term action and remote results.

Plan B – geo-engineering aimed at blocking sunlight. Not much investment and unilateral action are possible. Quicker action and quicker results, but involves risks in the case of switching off.

Plan C – emergency actions for Sun dimming, like artificial volcanic winter.

Plan D – moving to other planets.

All plans could be executed using current tech levels and also at a high tech level through the use of nanotech and so on.


I think that climate change demands that we go directly to plan B. Plan A is cutting emissions, and it’s not working, because it is very expensive and requires cooperation from all sides. Even then it will not achieve immediate results and the temperature will still continue to rise for many other reasons.


Plan B is changing the opacity of the Earth’s atmosphere. It could be a surprisingly low cost exercise and could be operated locally made. There are suggestions to release something as simple as sulfuric acid into the upper atmosphere to raise its reflection abilities. 


"According to Keith’s calculations, if operations were begun in 2020, it would take 25,000 metric tons of sulfuric acid to cut global warming in half after one year. Once under way, the injection of sulfuric acid would proceed continuously. By 2040, 11 or so jets delivering roughly 250,000 metric tons of it each year, at an annual cost of $700 million, would be required to compensate for the increased warming caused by rising levels of carbon dioxide. By 2070, he estimates, the program would need to be injecting a bit more than a million tons per year using a fleet of a hundred aircraft." 


There are also ideas to recapture CO2 using genetically modified organisms, iron seeding in the oceans and by dispersing the carbon capturing mineral olivine.


The problem with that approach is that it can't be stopped. As Seth Baum wrote, a smaller catastrophe could result in the disruption of such engineering and the consequent immediate return of global warming with a vengeance.


There are other ways pf preventing global warming. Plan C is creating an artificial nuclear winter through a volcanic explosion or by starting large scale forest fires with nukes. This idea is even more controversial and untested than geo-engineering.


A regional nuclear war I capable of putting 5 mln tons of black carbon into the upper athmosphere, “average global temperatures would drop by 2.25 degrees F (1.25 degrees C) for two to three years afterward, the models suggest.”  Nuclear explosions in deep forests may have the same effect as attacks on cities in term of soot production.


Fighting between Plan A and Plan B


So we are not even close to being doomed by global warming but we may have to change the way we react to it.

While cutting emissions is important it will probably not work within a 10-20 year period, quicker acting measures should be devised.


The main risk is abrupt runaway global warming. It is low probability event with the highest consequences. To fight it we should prepare rapid response measures.


Such preparation should be done in advance, which requires expensive scientific experiments. The main problem here is (as always) funding, and regulators’ approval. The impact of sulfur aerosols should be tested. Complicated math models should be evaluated.


Contra-arguments are the following: “Openly embracing climate engineering would probably also cause emissions to soar, as people would think that there's no need to even try to lower emissions any more. So, if for some reason the delivery of that sulfuric acid into the atmosphere or whatever was disrupted, we'd be in trouble. And do we know enough of such measures to say that they are safe? Of course, if we believe that history will end anyways within decades or centuries because of singularity, long-term effects of such measures may not matter so much… Another big issue with changing insolation is that it doesn't solve ocean acidification. No state actor should be allowed to start geo-engineering until they at least take simple measures to reduce their emissions.” (comments from Lesswrong discussion about GW).


Currently it all looks like a political fight between Plan A (cutting emissions) and Plan B (geo-engineering), where plan A’s approval is winning.  It has been suggested not to implement Plan B as an increase in the warming would demonstrate a real need to implement Plan A (cutting emissions). Regulators didn’t approve even the smallest experiments with sulfur shielding in Britain. Iron ocean seeding also has regulatory problems.

But the same logic works in the opposite direction. China and the coal companies will not cut emissions, because they want to press policymakers to implement plan B. It looks like a prisoner’s dilemma of two plans.


The difference between the two plans is that plan A will return everything to its natural state and plan B is aimed on creating instruments to regulate the planet’s climate and weather.


In the current global political situation, cutting emissions is difficult to implement because it requires collaboration between many rival companies and countries. If several of them defect (most likely China, Russia and India, who have heavy use of coal and other fossil fuels), it will not work, even if all of Europe were solar powered.


Transition to zero-emission economy could happen naturally in 20 years after electric transportation will become widespread as well as solar energy.


Plan C should be implemented if the situation suddenly changes for the worse, with the temperature jumping 3-5 C in one year. In this case the only option we have is to bomb Pinatubo volcano to make it erupt again, or probably even several volcanos. A volcanic winter will give us time to adopt other geo-engineering measures.


I would also advocate for a mixture of both plans, because they work on different timescale. Cutting emissions and removing CO2 using the current level of technologies would take decades to have an impact on the climate. But geo-engineering has a reaction time of around one year so we could use it to cover the bumps in the road.


Especially important is the fact that if we completely stop emissions, we could also stop global dimming from coal burning which would result in a 3 C global temperature jump. So stopping emissions may result in a temperature jump, and we need a protection system in this case.


In all cases we need to survive until stronger technologies develop. Using nanotech or genetic engineering we could solve the warming problem with less effort. But we have to survive until this time.


It seems to me that the idea of cutting emissions is overhyped and solar management is "underhyped" in terms of public opinion and funding. By changing that misbalance we could achieve more common good.


An unpredictable climate needs a quicker regulation system

The management of climate risks depends on their predictability and it seems that this is not very high. The climate is a very complex and chaotic system.


It may react unexpectedly in response to our own actions. This means that long-term actions are less favorable. The situation could change many times during their implementation.


The quick actions like solar shielding are better for management of poor predictable processes, as we can see the results of our actions and quickly cancel them or make them stronger if we don't like the results.


Context risks – influencing the probability of other global risks


Global warming has some context risks: it could slow tech progress, it could raise the chances of war (probably already happened in Syria because of draught, it could exacerbate conflicts between states about how to share recourses (food, water etc.) and about the responsibility for risk mitigation. All such context risks could lead to a larger global catastrophe.


Another context risk is that global warming is captures almost all the available public attention for global risks mitigation, and other more urgent risks may get less attention.


Many people think that runaway global warming constitutes the main risk of global catastrophe. Another group think it is AI, and there is no dialog between these two groups.


The level of warming which is survivable strongly depends of our tech level. Some combinations of temperature and moisture are non-survivable for human beings without air conditioning: If the temperature rises by 15 C, half of the population will be in a non-survivable environment because very humid and hot air prevents cooling by perspiration and feels like a much higher temperature. With the current level of tech we could fight it, but if humanity falls to a medieval level, it would be much more difficult to recover in such conditions.


Rising CO2 levels could also impair human intelligence and slow tech progress as CO2 levels near 1000 ppm are known to have negative effects on cognition.


Warming may also result in large hurricanes. They can appear if the sea temperature reaches 50 C and they have a wind speed of 800 km/h, which is enough to destroy any known human structure. They will be also very stable and live very long, thus influencing the atmosphere and creating strong winds all over the world. The  highest sea temperature currently is around 30C.


In fact we should compare not the magnitude but speed of global warming with the speed of tech progress. If the warming is quicker it wins. If we have very slow warming, but even slower progress, the warming still wins. In general I think that progress will overrun warming, and we will create strong AI before we have to deal with serious global warming consequences.


Different predictions


Multiple people predict extinction due to global warming but they are mostly labeled as “alarmists” and are ignored. Some notable predictions:

1.      David Auerbach predicts that in 2100 warming will be 5 C and combined with resource depletion and overcrowding it will result in global catastrophe.


2.      Sam Carana predicts that warming will be 10 C in the 10 years following 2016, and extinction will happen in 2030.


3.      Conventional predictions of the IPCC give a maximum warming of 6.4 C at 2100 in worst case emission scenario and worst climate sensitivity to them:


4.      The consensus of scientists is that climate tipping point will be in 2200


5.      If humanity continues to burn all known carbon sources it will result in a 10 C warming by 2030.  The only scenario in which we are still burning fossil fuels by 2300 (but not extinct and not a solar powered supercivilzation running nanotech and AI) is a series of nuclear wars or other smaller catastrophes which will permit the existence of regional powers which often smash each other into ruins and then rebuild using coal energy. Something like global nuclear “Somali world”.


We should give more weight to less mainstream predictions, because they describe heavy tails of possible outcomes. I think that it will be reasonable to estimate the risks of extinction level runaway global warming in the next 100-300 years at 1 per cent and act as it is the main risk from global warming.







More posts like this

Sorted by Click to highlight new comments since:

Have been following these maps for a while. Would be interested in a map of interventions on some matter like global warming, or even more ideally some assessment of the importance / neglectedness / tractability of interventions on some futuristic issue.

Thanks for your interest - I would be happy to create a map based on public interest, but I need more input.

The map of intervention in global warming is in this post. )) Please clarify what do you mean.

Another map of x-risks prevention in general also about interventions.

All published maps are linked here:

Some future maps are in different draft stages:

AI as an instrument for life extension

Fermi paradox

Personal identity

Mind improvement

Aging theories


My question is whether you could indicate which areas of a map you think are fruitful areas to try to work on. It may be a bit intimidating to see the size of a field like global catastrophic risks. So it seems like it would be helpful to accompany some maps with a scheme for prioritizing the important areas. e.g. if people could know that safe ai engineering is a useful area for reducing gcrs (because not many people are trying that thing) but climate change activism is not (because it's a really crowded area).

I realize this would be a big additional project but I am curious whether you've considered the prioritization problem as relevant to your mapping project.

I will think about it

My Identity map is on.

So, some ideas for further research, that is fields which a person could undertake if he want to make an impact in the field of x-risks. So it is carrier advises. For many of them I don't have special background or needed personal qualities.

  1. Legal research of international law, including work with UN and governments. Goal: prepare an international law and a panel for x-risks prevention. (Legal education is needed)

  2. Convert all information about x-risks (including my maps) in large wikipedia style database. Some master of communication to attract many contributors and balance their actions is needed.

  3. Create computer model of all global risks, which will be able to calculate their probabilities depending of different assumptions. Evolve this model into world model with elements of AI and connect it to monitoring and control systems.

  4. Large research is safety of bio-risks, which will attract professional biologists.

  5. Promoter, who could attract funding for different research without oversimplufication of risks and overhyping solutions. He may be also a political activist.

  6. I think that in AI safety we are already have too many people, so some work to integrate their results is needed.

  7. Teacher. A professor who will be able to teach a course in x-risks research for student and prepare many new researchers. May be youtube lectures.

  8. Artist, who will be able to attract attention to the topic without sensationalism and bad memes.

Extra risks from geoengineering:

Cause additional climate problems (ie. it doesn't just uniformly cool planet. I recall seeing a simulation somewhere where climate change + geoengineering did not equal no change, but instead significantly changed rainfall patterns).

Global coordination problems (who decides how much geoengineering to do, compensation for downside, etc.). This could cause a significant increase in international tensions, plausibly war.

Climate Wars by Gwynne Dyer has some specific negative scenarios (for climate change + geoengineering)

But if we stop emissions now GW will probably continue to exist for around 1000 years as I read somewhere, and even could jump because cooling effects of soot will stop.

Global coordination problems also exist, but may be not so annoying. In first case punishment comes for non-cooperation, and in second - for actions, and actions always seems to be more punishable.

I'm not saying these mean we shouldn't do geoengineering, that they can't be solved or that they will happen by default, just that these are additional risks (possibly unlikely but high impact) that you ought to include in your assessment and we ought to make sure that we avoid.

Re coordination problems not being bad: It's true that they might work out, but there's significant tail risk. Just imagine that say, the US unilaterally decides to do geonengineering, but it screws up food production and the economy in China. This probably increases chances of nuclear war (even more so than if climate change does it indirectly, as there will be a more specific, attributable event). It's worth thinking about how to prevent this scenario.

Scientific studies and preparation for GE is probably the longest part of GE, and could and should be done in advance, and it should not provoke war. If real necessity of GE appear, all need technologies will be ready.

Ok, I will add it as risks from geo-ingeneering

This has a more comprehensive list of emission reduction techniques, including building energy efficiency, industry energy efficiency, carbon capture from power plants and storage (at the bottom of the ocean or underground), nuclear energy, etc. As for removing the CO2 from the air, other options include growing biomass and burying it, encouraging cultivated lands to store more carbon (actually considered in the paper above), and chemical capture from the air and storing it underground. A hybrid technique is growing biomass, making liquid fuel, and sequestering the CO2, which is overall net negative emissions. As for solar radiation management, other options for increasing stratospheric reflectivity include photophoretic levitation of engineered aerosols, and nanoparticles from combustion. We might be able to use equatorial mountain combustion to inject the particles into the stratosphere. Other options for increasing surface albedo include reflective balloons in the ocean (not a good idea because of impact on marine life). A benefit of solar radiation management over reduction in emissions is that it enables higher agricultural productivity via CO2 fertilization. This is important because most of people who are in extreme poverty are farmers. Though CO2 fertilization was found to be not as large as once thought in natural ecosystems, the effect is large in agriculture because needs like water and fertilizer are generally provided for. Another benefit of solar radiation management is that if it looked like we were heading back into an ice age, we could turn it off.

Curated and popular this week
Relevant opportunities