TLDR: The 80K advice on solar geoengineering on their climate profile isn’t great. They imply they are carrying out a risk-risk analysis, but fail to do this, and their conclusions aren't massively supported by the literature. The source they cite to support their claims don't support this claim. The two key problems they identify with solar geoengineering are serious, but they fail to engage with the risks of not doing it. They conflate research and advancing of solar geoengineering without justification, and fail to grapple with the risks of not researching solar geoengineering. Their failure in the profile as a whole to utilise complex risk assessments simplifies the complexity of the way both climate and solar geoengineering interact with X-Risk. Finally, they fail to engage with the marginal difference EAs may make given the current research landscape.
Of course, summarising something so complex in a few lines is very hard, so I don't blame 80K for a lot of the problems, but nonetheless, they exist and have a real impact. They should emphasise the deep uncertainty and complexity of the topic more, and "meta-research" to answer whether and what kinds of solar geoengineering research we should do to reduce X-Risk would likely be a high impact career for those who have good personal fit.
80000 hours is a highly influential organization in EA, and their positions on problem areas are often taken as “accepted wisdom” by the community. By advising many EAs on their career paths, they have a large amount of influence which seems to get strongly translated into impact. However, as a great man once said, with great power comes great responsibility (3). When 80K gets things wrong, this can affect the communal understanding of problem areas at large, which could reduce the impacts EA can have over potentially important cause areas, and can make life harder (and thus impact reduced) for those who work in those cause areas. When an organisation has as much soft power as 80K, they have a responsibility not to take an easy cop out as they do in their problem profile, and own up to the difficult uncertainties which exist around this topic. At present their advice fails to do this. In EA, we are trying to engage with some of the world’s biggest problems, and it won’t be easy. But we cannot shy away from doing it.
In May, 80K released a problem profile on climate change (4). While I have various problems with this problem profile, a larger critique is for another time. Here I specifically want to focus on what they said on solar geoengineering, and what I think they get wrong/could do better. This isn’t necessarily a critique of their approach, as I am aware 80K’s problem profiles are meant to be summaries and the people who write them aren’t experts in it, but rather I hope to highlight some things I nonetheless think it does wrong, either by fault of the writers, the format or the fact that 80K are a generalist organization so won’t get everything right.
This is not a post arguing that we should do solar geoengineering now, or ever. It may be too dangerous to ever do. It also isn’t suggesting that we should do any specific type of research. The RESILIENCER Project (5) which I have started is researching this exact question, and in general it’s clear that solar geoengineering has many problems which may mean that many types of research programmes do more harm than good. Instead this blog post is highlighting that the arguments made in the problem profile and 80K’s general approach to solar geoengineering are somewhat weak, and don’t really support the strength of conclusions they make.
I make the case that the sources cited and arguments made in the sections on solar geoengineering fail to support the assertions made or the conclusions that 80K readers should not research solar geoengineering. I also highlight some key points that 80K miss out. Finally, I conclude with what I think the advice should say.
If you are unfamiliar with what solar geoengineering is, and its possible impacts related to climate change, I recommend checking out some of the resources at https://www.c2g2.net/solar-radiation-modification/ Also please do look at the references in this article, particularly (6) to get an understanding of impacts, (2) (7)(8) for the literature around solar geoengineering and GCRs/X Risks so far.
A quick intro to solar geoengineering (skip if you know): Solar geoengineering is a proposed set of technologies that may reduce some of the worst climate impacts by reflecting sunlight, often by injecting aerosols into the stratosphere to mimic volcanoes or by brightening clouds to mimic ship’s tracks. More explanation here (9) It, therefore, doesn’t address the root of climate change, but rather only the symptoms, and has been proposed as a band-aid solution given we are not mitigating climate change fast enough, or in the case we might hit tipping points etc. However, in general, it cannot be considered a replacement for mitigation, adaptation and carbon removal, which are the only long term solutions to the climate crisis. I won’t go into the in depth description of why this is the case here, as this is for another time.
Firstly, below is the two sections of the problem profile that refers to solar geo engineering: ”It’s also possible that we could be driven to develop destabilising technology to change our climate with the intention of averting catastrophe — e.g. solar geoengineering. But this poses its own risks, as it will be near impossible to carry out experiments on the global scale we’d need to act in order to verify the safety of our technology. And technology to change the weather could in turn lead to conflict between (or within) states over things like induced droughts or rainfall.”
”Research into carbon removal technology (but not solar geoengineering) Carbon removal technology, like negative emissions tech or carbon capture and storage, seems pretty neglected compared to green energy (read a popular overview), and could be a crucial way of reducing the effects of our emissions on the climate. Removing carbon in this way is a form of geoengineering — deliberate intervention in the climate. The other primary form of geoengineering is solar geoengineering (deliberately deflecting sunlight away from Earth to cool the planet down). Solar geoengineering poses potential risks to humanity in itself, given the unprecedented scale of the intervention and the fact that, once in use, solar geoengineering can’t be left untended without disastrous effects. These risks could be larger than the risks from climate change itself, so we think it’s potentially harmful to do work that could advance solar geoengineering. Geoengineering research of all kinds is mainly done in academia. The Oxford University Geoengineering Programme conducts research into the social, ethical, and technical aspects of geoengineering.”
Issue 1: Terminology
The 80K profile uses the term geoengineering to refer to both solar geoengineering (otherwise known as Solar Radiation Modification/Management) and carbon removal. Whilst this is indeed used in the literature (10), increasingly these two are being viewed as fundamentally different things, with the geoengineering label increasingly not being applied to carbon removal. Even the CDR primer (11) linked to in the profile is explicit in not considering carbon removal to be geoengineering. It isn’t “wrong” per se to conflate the two, but does make 80K appear somewhat out of step with the state of the academic and policy conversation around carbon removal and solar geoengineering, which as an organisation focused on public communication is something important.
Although this isn’t focused on solar geoengineering, in this same section of the profile they claims that carbon capture and storage(CCS) is a form of carbon removal. CCS is not considered a form of carbon removal, as it doesn’t remove CO2 from ambient air but rather from concentrated fossil fuel sources. Once again, the CDR Primer (11) linked to makes this clear. CCS can be used in carbon removal, such as Bioenergy with Carbon Capture and Storage, but CCS in the “bland” form is removing CO2 from smokestacks after burning CO2
which is a fundamentally different thing from carbon removal, and it is unhelpful to characterize them as the same thing.
Moreover, in the first quote provided in the above section, the profile seems to conflate solar geoengineering with ”technology to change the weather.” These are two different things with similar but distinct issues, and it is unclear if this conflation is intended or not.
Issue 2: Risk-risk trade off- lack of evidence
I’m going to skip to the last reference to solar geoengineering now, as I hope the reasoning becomes obvious later “These risks [from solar geoengineering] could be larger than the risks from climate change itself, so we think it’s potentially harmful to do work that could advance solar geoengineering.” This sentence claims to be making a risk-risk tradeoff between solar geoengineering and climate change. Given 80K’s advice is that people shouldn’t work on solar geoengineering, it is actually making a hidden stronger claim, that the probability that solar geoengineering risk is larger than climate change risk is unacceptably large. In fact, it is not claiming this in general, but rather that specifically two risks from solar geoengineering “the unprecedented nature of the intervention” and termination shock are more of a risk than all the risks from climate change. The problem is no sources are cited to support this claim. In fact, the only source cited in the section was (7) merely illustrating the idea of termination shock, so the assumption is that this evidence is from there. However (7) doesn’t argue either of these two points. It suggests the risk from termination shock is small (something I somewhat disagree with). It also suggests that “unknown environmental existential risks of SAI are negligible.” This isn’t to suggest that John Halstead doesn’t think there aren’t other risks of SRM that make it bad to do, or that he hasn’t changed his mind since, just that the only cited source by 80K doesn’t support the claims made.
One might then turn to (2), which analyzes in considerable depth the contribution of SRM to GCRs, including by effects on health, conflict, agriculture and latent risk. However, this paper doesn’t carry out a risk-risk analysis either, making the confidence of the claims made in the 80K profile unevidenced.
In fact a recent report on risk-risk analysis between climate change and solar geoengineering (6) certainly indicated that decisions on SRM should be taken in a risk-risk framework, but also indicated that it may be possible that SRM could be less risky than the alternative of a dangerously warmed world. Certainly, very little in that report justifies the 80K confidence that the risks of solar geoengineering are unacceptably large compared to climate change, and if anything, the report suggests that most risks seem to push the other way. However, uncertainties are still too large to be able to say much at all about this. I would recommend reading that report in depth for a good review of the literature on the (predominantly physical) risks and benefits of solar geoengineering- I will write these up at another time, but that is not the focus of this post. Instead, the focus is whether the literature in general supports
80K’s claim, which I suggest it does not. More of this will be discussed later. Ultimately, whilst seemingly claiming to be functioning within a risk-risk framework, the 80K profile fails to evidence the claims that it makes. The need for applying such symmetric precaution that a risk-risk analysis helps fulfill was outlined by Edward Parsons in a recent Science editorial (12), and the fact that this essentially has not been done on climate and solar geoengineering and their relation to X-Risk means that drawing the conclusions that 80K does seems presumptuous.
The other way of interpreting 80K's claim is that it is not making a risk-risk trade off, but rather suggesting that, as there are risks from SRM research, it shouldn’t be done at all- an extreme version of the precautionary principle. I assume this is not the case, as this wouldn’t line up with what they suggest for other risks. For example, they promote research into AI Safety, such as interpretability research (13), which has its risks as it can be used to advance AI capabilities. Yet 80K (rightfully I think) suggest that the risks of not researching outweigh the risks of research, and so recommend its done. Thus, it would be odd in the case of solar geoengineering if they weren’t using a risk-risk tradeoff as they seemed to have done elsewhere. Hence why I have assumed in this section that when they say the risks from solar geoengineering could be larger than climate change, they mean to suggest they have indeed taken a risk-risk tradeoff.
I hoped to have shown the evidence provided in the 80K piece and in the literature more generally is grossly insufficient to support 80K’s conclusion, and that much of the evidence in the literature indeed may (tentatively) suggests solar geoengineering would likely (but we still need to know far more, so we shouldn’t be at all bullish yet) be less risky than climate change, given the same level of GHG emissions (6). It should be noted that all of these still require us to cut emissions, adapt and remove CO2; solar geoengineering, even in the best case scenario, is no replacement for these (14).
Issue 3: Unprecedented Impacts
Of the two risks of solar geoengineering which the 80K profile highlights, one of them is the “unprecedented nature of solar geoengineering.” Remember, however, we are functioning in a risk-risk framework, so the question posed must be: which has greater unprecedented impacts, solar geoengineering of climate change?” Of course, this is dependent on the specific scenario that we are in, including warming levels and modes of deployment, but it certainly seems that the unknown physical impacts of climate change are at least as large in magnitude, if not larger than those of solar geoengineering. At least 1K of cooling via stratospheric aerosols has precedent during human civilisation; after all, this was roughly the cooling that the eruption of Mount Pinatubo brought (15) (16). Whilst this isn’t a perfect like for like comparison, as Pinatubo’s impacts lasted about a year, whilst solar geoengineering programmes would be expected to last decades to centuries, nonetheless, there is some precedent. It is, however, arguable that there is less precedent for a severely warmed world. Tipping points and other unknown unknowns (17) could occur in that climate and whilst some of these may act as negative feedbacks, we should also expect positive feedbacks as well (1). Of course this is debatable, as we do have paleoclimatic data, but none of those paleoclimates with analogous warming was during the lifetime of human civilization. Moreover, whilst a solar geoengineered world is certainly more dangerous than not emitting in the first place, it seems like such a world could (depending on what deployment scheme is used) be closer to the present climate, certainly in terms of climate risks, temperature and precipitation, than a world with significant warming and without solar geoengineering (18) (19). Thus we should expect that the unknown risks would be less under a more similar climate than under the greater warmed world. Thus a world with solar geoengineering (alongside mitigation) may - may being the operative word here- be less dangerous than a world with similar mitigation levels but no solar geoengineering.
Whilst this section certainly doesn’t provide a confident case that the unprecedented risks from solar geoengineering outweigh the unprecedented risks from climate change, I hope it has provided a confident rebuttal of the argument that we know enough to reject solar geoengineering research on this basis.
Issue 4: Termination Shock
Termination shock, as per (20), refers to “a rapid and substantial rise in global temperatures following a cessation of SRM [solar geoengineering] deployment.” This is due to the fact that solar geoengineering merely ‘masks’ the warming from greenhouse gases by increasing the Earth’s reflectivity (albedo). As this is done by injecting aerosols into the atmosphere, with lifetimes on the order of weeks (tropospheric aerosols) or months (stratospheric aerosols), whereas CO2 has lifetimes on the order of centuries, if you stop injecting the aerosols for a significant amount of time you get an increase in temperature eventually to the equilibrium temperature given the GHG concentrations in the atmosphere. This is a real danger. However, it is only a danger under a specific set of circumstances, namely:
• Solar geoengineering would have to be carrying out substantial cooling
• It must be interrupted for a significant period of time
• The termination of solar geoengineering must be rapid
Given the relative cheapness of solar geoengineering, and the fact that it could, if necessary, be carried out unilaterally, it seems like termination may only be likely in the case of a global catastrophe. This is because there is enough of a lag-time to rebuild physical infrastructure before termination shock if state capacity to do this still exists. And due to the vastly negative nature of termination shock, and the free driver effect, for a country or coalition of countries to electively stop solar geoengineering they would need to be able to exert that power over every other capable state. This mostly seems implausible (20). Such a GCR followed by termination shock would lead to a “double catastrophe” (21), where not only does the global catastrophe cause vast damage, but then the damaged and likely collapsed civilisation would be faced with rapidly rising temperatures due to termination shock, which may make near term recovery very hard. However, even this doesn’t necessarily support the position that people shouldn’t work on solar geoengineering. For example,(2) highlights a possible cause of termination shock could be space weather events, so making solar geoengineering systems resilient to these events, or having more redundancies in place, may reduce the chances of double catastrophe. It also seems unclear that such research would increase the probably solar geoengineering were deployed. Moreover, the literature on double catastrophe may have suggested that it could occur over a wider range of events than it might in reality. (2) and (21) both discuss nuclear war, however, in the case of a nuclear winter, it is unclear whether termination shock would even be a bad thing. Soot aerosols have e-folding times of 4.6 years (22), whereas sulfate aerosols have an e-folding time of only 1 year (22), meaning that whilst the magnitude of the forcing is more on a geoengineered world than a non-geoengineering world, the delta of the forcing by the end of year 1 is either negligibly different or smaller on the geoengineered world vs a non-geoengineered world. However, this is only from back of the envelope calculations so should be taken with a massive pinch of salt. More research ought to be done on climatic responses and which are most important for civilisational recovery, but it appears that we have reason to think termination shock under a nuclear war scenario may be (most likely) negligible and perhaps even slightly less bad than without solar geoengineering termination. It should be noted this likely wouldn't be the case for other cooling e.g. by volcanos or asteroid strikes. So, whilst it is clear that termination shock is a real danger, 80K may be overplaying its case by suggesting it, in concert with the unprecedented risks, may pose enough of a risk to reject researching solar geoengineering. There still seems to be too much uncertainty to make such a statement. Certainly, it still seems very unclear that termination shock outweighs the benefits of solar geoengineering, which may reduce the probability of a ”single catastrophe” occurring. Moreover, as such little research has been done into termination shock after a global catastrophe, more research ought to be done as to how serious this would be for such a collapsed civilisation’s survival or recoverability.
Issue 5: Conflating research and advancing solar geoengineering
The headline of that section suggests that people shouldn’t research solar geo engineering, however in the piece it suggests it could be dangerous to advance solar geoengineering. It is true that there are significant links between research and deployment (23) (24) (25), and viewing research as purely neutral and isolated is (in most cases), dangerous and false. However, it is also clear that much research into solar geoengineering or similar technologies needn’t make deployment more likely (26), and some of the previously proposed pathways, like individual moral hazard, to some degree seems to not be evidenced in experimental literature (27). It may, however, be reasonably argued that one cannot predict whether one’s research may make deployment more likely, however, as there is the opposite case seems equally plausible, an appeal to uncertainty here doesn’t seem convincing evidence against research. In fact, in Halstead’s appeal to deep uncertainty (7), he suggests there is likely a tentative case for research. This doesn’t necessarily mean all research should therefore be promoted or accepted, but that steps should be taken to carry out research to reduce our uncertainty on the linkages between research and deployment. However, there does potentially seem like possible pathways for research that might not significantly increase the probabilities of deployment if done correctly, whilst may have an outsized impact on reducing GCRs and X-Risks. I have a paper I am working on currently somewhat talking about how we might recognise these, but I expect that research specifically focused on how to reduce the GCRs/X-Risks that solar geo engineering poses, as well as risk-risk assessments of solar geoengineering and climate change under heavy tailed scenarios, are unlikely to increase the probabilities of deployment. These also may be exceptionally valuable at actually answering the question of whether specific types solar geoengineering research is something that could be very high impact. Doing this sort of research is also massively neglected, with the overall core community of solar geoengineering researchers about 100 people (28) and the number of people who work on solar geoengineering and GCRs/X-Risks is about 5 people, and so for people with very good personal fit I could see it being very high impact intervention. Moreover, solar geoengineering research may do many things, including promoting better governance of solar geoengineering by building the capacity of more countries to participate in negotiations(29). Once again this may (or may not) lead to it being less likely to deploy, or less likely to deploy dangerously. Furthermore, solar geoengineering research may allow for better attribution (30), which may reduce the likelihood of conflicts over natural disasters during the period where solar geoengineering is deployed (if it is at all). However, this idea may ignore the role of actors who rely less on science and expertise when making decisions, something which was seen a lot during the COVID pandemic. It is also unclear what solar geoengineering research actually means, something I am also writing (the same as before actually) paper on at the moment. However, certain types of research, for instance that research which promotes governance or even outright bans for physical science experimentation (31), or those characterizing possible risks of solar geoengineering (32), don’t seem to necessarily constitute “advancing solar geoengineering,” and yet still would be considered solar geoengineering research. Scientists who seem very skeptical of carrying out solar geoengineering, such as Alan Robock (33), have certainly engaged in solar geoengineering research (34), perhaps further providing evidence that not all solar geoengineering research makes deployment more likely. This is important, as not all research is equal. Some research may indeed prove very damaging, pose a severe moral hazard and increase likelihood of deployment. Other research may not and may do the opposite. Thus, research cannot be considered a monolithic abstract, but rather what sorts of research are not recommended and those which are should be specified.
Issue 6: What they don’t say + Marginal Impact
There are lots of reasons to be skeptical of solar geoengineering research. Firstly, as mentioned above, doing research may reduce the likelihood of cutting emissions, may lock in unwanted futures, may cause a slippery slope, or may lead to other things that make undesirable deployment options more likely. Moreover, if deployed in the wrong way, solar geoengineering may lead to adverse impacts, and if suddenly terminated, may lead to the aforementioned termination shock. However, not researching also comes with risks, be it deployment carried out in an uninformed way without knowledge of the risks, the impetus for governance to be significantly less (35), a failure to build up resilience of the possible system to global catastrophes (2), or not being able to deploy solar geoengineering when it may have significantly reduced X-Risk. Failing to research also creates its own lock-in, and this must be considered. Nonetheless, none of these ideas are explored in the 80K piece, and its failure to explore the risks of not researching is arguably the 80K piece's biggest failure. Solar geoengineering itself could potentially act as a key defense given how the climate crisis may develop(18) (36), so rejecting research as dangerous so early without having fully engaged in a risk-risk analysis is itself dangerous.
Secondly, the risks of conflict from solar geoengineering are potentially very large, including it driving great power conflict (2). However, climate change also provides a significant stressor which may significantly increase conflict (37) that may cascade to a GCR (1) . Once again we are deeply uncertain as to what we should do and whether solar geoengineering increases or decreases conflict risk. More research into solar geoengineering and conflict may be desirable, as well as research into what governance measures could be taken (7). The failure of the 80K piece to mention how solar geoengineering may reduce conflict risk as well as increase it is again important, and yet another failure to take symmetric precaution (12) seriously.
Moreover, the piece makes no mention of the research and policy environment we are in at the moment, where solar geoengineering is being explored in international policy settings (38), research advised by the NAS (14) and endorsed by Nature(39), and the US Federal Government setting up a research programme (40). If 80K had assumed that solar geoengineering is indeed bad, then campaigning against solar geoengineering’s normalization may be able to do a lot of good, as such campaigns have in the past been successful (41). If solar geoengineering is a potentially impactful intervention, either positive or negative, just disengaging and ”not researching” should not be the sort of actions taken. The recognition that we need to make an impact on the margin is one of EAs’ key advantages, and yet in their advice on solar geoengineering 80K essentially seem to not mention and ignore the counterfactual. This lack of awareness of the field is made all the more obvious by them citing the Oxford Geoengineering Programme as a key organization to work for, despite them no longer being very active and not particularly impactful in the field anymore. It is clear that 80K here have failed to consider, given the landscape around solar geoengineering research right now, the marginal impact that one could have . The other explanation for this failure to consider actions that could be taken on the margin is a belief that we are so clueless about solar geoengineering that making a predictable positive impact shaping its development is too hard. However, 80K has shown a willingness to recommend careers that are speculative in their impact (13) or careers in researching what actions are those which make positive impacts (42). Thus, it seems strange to suggest we shouldn’t be re searching solar geoengineering at all, not even to work out whether and how to have a positive impact by either encouraging, blocking or changing the direction of research. This seems like a major oversight.
Issue 7: Complex Risk Assessments
The 80K article, in drawing its distinctions between direct and indirect risks, as well as neglecting systemic risk, latent risk or other forms of more complex risk, takes a simple risk assessment model of climate and its relation to X Risk. This may be warranted, although there are reasons to think that it isn’t (2) (1). This focus may have a major impact on their assessment of solar geoengineering. If one only takes direct, extreme hazards as the key impact, and ignores the possibilities of cascading risks or systemic risks for example, the type of risk-risk assessment one would do with relation to climate and solar geoengineering would be significantly different than if one opted for a more complex risk assessment. Solar geoengineering is exceptionally complex, as at various different heavy tails it may increase or decrease the probability of X Risk. For example, its latent risk seems to increase probabilities of extinction following collapse, but its positive effects may reduce the likelihood of collapse by reducing climate stress. Depending on how it is deployed, it may increase or decrease conflict risk, which then may combine with either positive or negative impacts it has on the climate to increase or decrease the systemic risk. It adds greater complexity to society, potentially increasing vulnerability, but may decrease exposure. So whilst indirect-direct risks are in theory covering the whole space, by failing to utilize the other risk frameworks, normally they end up in practice missing out the key components of what makes each of these risks so serious. Such frameworks are used by virtue of their utility, and by failing to use a complex risk assessment framework, 80K fails to have an adequate conceptual framework to consider some of the more likely and more serious global catastrophic and existential risks which climate change and solar geoengineering contribute to. (1) (2). More of these ideas are discussed in (8), but the fact I am having to cite a podcast shows just how neglected and under explored complex risks of solar geoengineering vs climate change are, which is also discussed in (1)
People have, using simple risk analysis, criticized solar geoengineering as a ”techno-fix” and a ”band aid” solution. Both of which are true. But unless we do a complex risk assessment, with a risk-risk trade off, we can’t tell if a band aid is exactly what we need to reduce GCRs and X-Risks, or whether the fact it's a techno-fix and a band aid makes it hubristic and dangerous and mostly contributing to X-Risk.
I hope this has proven that the 80K remarks on solar geoengineering are unevidenced and unsupported at best, and at worst ignore key aspects of the debate around whether we should do solar geoengineering research or not. I hope they spend more time looking at solar geoengineering, or at least better evidence the claims they make. They should emphasize the uncertainty of the topic far more, as if they indeed had the certainty that implicitly comes across in the article, you would expect advice on how to stop solar geoengineering research. This they do not present. Note, this critique isn’t of 80K as an institution, or the quality of their research. If I had tried to summarize the entire solar geoengineering debate and give meaningful advice in just a few lines, I could probably write a critique this long of myself. Nonetheless, it is a strong critique of the certainty of the conclusion that people should not research solar geoengineering (or seemingly do other actions around this) that seems to come across in the article, as neither do I think the article, nor the literature in general, supports such a strong conclusion.
I do not think all solar geoengineering research is a good thing; in fact, much of it may increase X-Risks. For example, this post is not an endorsement of the desire to do outdoor research at present, or the desire to ban them. I do however think there seems to be research programmes which may significantly reduce X-Risks. Thus, doing meta-research to identify how solar geoengineering research relates to X-Risks, and so what research into solar geoengineering we ought to do to reduce X-Risks, may be a prudent course of action.
So, what should the advice be?
We are highly uncertain as to whether solar geoengineering research is a very good thing, a very bad thing or neither. We are highly uncertain whether the (in particular) systemic risks and conflict risks, as well as other risks, of climate change outweigh the latent risk and conflict risk, as well as other risks, of solar geoengineering. We are highly uncertain as to the circumstances in which solar geoengineering would reduce existential risk, as well as those under which it would increase it.
However, this isn’t necessarily a reason to think that we should or shouldn’t do specific types of research into solar geoengineering. In particular, trying to answer these questions of whether solar geoengineering research and/or deployment would increase or decrease existential risk, and under which circumstances, may be a very high impact thing to do. Particularly, given that the existential risks associated with climate change (1) and solar geoengineering (2) may be able to be reduced by specific types of research programmes, trying to work out if research can be done safely may have a very high impact. This question of whether solar geoengineering is an important consideration for people who care about X-Risks, and whether and what research can be done safely, is thoroughly neglected, and more people working on this question would be useful. At this stage we are unsure whether different types of solar geoengineering research are very positive or very dangerous, so more "meta-research" as to whether and what we should research is needed before we can make a committed statement either way.
Of course, whether or not to do solar geoengineering research isn’t a problem that lends itself to simple cause-effect thinking, and overconfidence in favour of solar geoengineering research may be very dangerous. Certainly, the 80K advice (4) is at least more conservative than the previous position 80K had taken on it. However, this is a situation where there are risks in not researching solar geoengineering, and risks in researching it. A blanket statement that “These risks could be larger than the risks from climate change itself, so we think it’s potentially harmful to do work that could advance solar geoengineering,” fails to engage with this complexity, as there is no easy way out of doing complex risk-risk analyses.
Thus, if anything, 80K should be encouraging people with good personal fit to work on this meta-question of "how does solar geoengineering research relate to X-Risks and GCRs?" and not saying that people shouldn’t work on research. What they say right now is just a cop out. When an organisation has as much soft power as 80K, they have a responsibility not to take an easy cop out, and own up to the difficult uncertainties which exist around this topic. In EA, we are trying to engage with some of the world’s biggest problems, and it won’t be easy. But we cannot shy away from doing it.
I am happy to respond to any criticisms or concerns of this piece.
 Kemp L, Xu C, Depledge J, Ebi KL, Gibbins G, Kohler TA, et al. Climate
Endgame: Exploring catastrophic climate change scenarios. Proceedings
of the National Academy of Sciences. 2022;119(34):e2108146119.
 Tang A, Kemp L. A fate worse than warming? Stratospheric aerosol
injection and global catastrophic risk. Front Clim. 2021 Nov;3.
 Cronin B. When we first met - when did Uncle Ben
first say “with great power comes great responsibility?”;
2015. Accessed: 2022-8-28. https://www.cbr.com/
 Hilton B. Is climate change the greatest threat facing humanity
today?; 2022. Accessed: 2022-8-28. https://80000hours.org/
 Home;. Accessed: 2022-8-11. http://resiliencer.org.
 Felgenhauer, T , Bala, G , Borsuk, M , Brune, M , Camilloni, I , Wiener
J B , Xu, J . Solar Radiation Modification: A Risk-Risk Analysis. C2G; 2022.
 Halstead J. Stratospheric aerosol injection research and existential risk. Futures : the journal of policy, planning and futures studies. 2018 Sep;102:63- 77.
 Reviewer 2 does geoengineering: Ok, Doomer! SRM and catastrophic risk - Tang on;. Accessed: 2022-8-28. https://podcasts.apple.com/us/ podcast/ok-doomer-srm-catastrophic-risk-tang/id1529459393?i= 1000576637605.
 Irvine P. Archimedes’ lever; 2022. Accessed: 2022-8-28. https:// peteirvine.substack.com/p/archimedes-lever.
 Royal Society (Great Britain), The Royal Society (London) , ) JSg. Geo engineering the Climate: Science, Governance and Uncertainty. Royal Society; 2009.
 Things O. Carbon Dioxide Removal Primer;. Accessed: 2022-8-28. https: //cdrprimer.org/.
 Parson EA. Geoengineering: Symmetric precaution. Science. 2021 Nov;374(6569):795.
 Hilton B. Preventing an AI-related catastrophe - Problem profile; 2022. Accessed: 2022-8-28. https://80000hours.org/problem-profiles/ artificial-intelligence/.
 National Academies of Sciences, Medicine. Reflecting Sunlight: Recommendations for Solar Geoengineering Research and Research Governance. Washington, DC: The National Academies Press; 2021.
 Proctor J, Hsiang S, Burney J, Burke M, Schlenker W. Estimating global agricultural effects of geoengineering using volcanic eruptions. Nature. 2018 Aug;560(7719):480-3.
 Soden BJ, Wetherald RT, Stenchikov GL, Robock A. Global cooling after the eruption of Mount Pinatubo: a test of climate feedback by water vapor. Science. 2002 Apr;296(5568):727-30.
 Lenton TM, Rockstrom J, Gaffney O, Rahmstorf S, Richardson K, Steffen W, et al.. Climate tipping points — too risky to bet against; 2019. Accessed: 2022-8-28. http://dx.doi.org/10.1038/d41586-019-03595-0.
 Irvine P, Emanuel K, He J, Horowitz LW, Vecchi G, Keith D. Halving warming with idealized solar geoengineering moderates key climate hazards. Nat Clim Chang. 2019 Mar;9(4):295-9.
 Reynolds J, Parker A, Irvine P, Sub O, Ucwosl PWeD. Five solar geo engineering tropes that have outstayed their welcome. Earth’s future. 2016 Oct;4(12):562-8.
 Parker A, Irvine PJ. The Risk of Termination Shock From Solar Geoengineering. Earth’s future. 2018 Mar;6(3):456-67.
 Baum SD, Maher TM, Haqq-Misra J. Double catastrophe: intermittent stratospheric geoengineering induced by societal collapse. Environment systems & decisions. 2013 Jan;33(1):168-80.
 Robock A, Oman L, Stenchikov GL, Toon OB, Bardeen C, Turco RP. Climatic consequences of regional nuclear conflicts. Atmos Chem Phys. 2007 Apr;7(8):2003-12.
 Tsipiras K, Grant WJ. What do we mean when we talk about the moral hazard of geoengineering? Environ Law Rev. 2022 Mar;24(1):27-44.
 Callies DE. The Slippery Slope Argument against Geoengineering Research. J Appl Philos. 2019 Aug;36(4):675-87.
 McKinnon C. Sleepwalking into lock-in? Avoiding wrongs to future people in the governance of solar radiation management research. Env Polit. 2019 Apr;28(3):441-59.
 Bodansky D, Parker A. Research on Solar Climate Intervention Is the Best Defense Against Moral Hazard. Issues Sci Technol. 2021 Jun;37(4):19.
 Irvine P. The results are in: discussing solar geoengineering doesn’t undermine emissions cuts; 2022. Accessed: 2022-8-24. https://peteirvine. substack.com/p/the-results-are-in-discussing-solar.
 M¨oller IM. Winning Hearts and Minds? Explaining the Rise of the Geo engineering Idea. In: Has it come to this? Rutgers University Press; 2020. p. 21-33.
 Putting developing countries at the centre of the SRM conversation; 2016. Accessed: 2022-8-28. https://www.degrees.ngo/.
 MacMartin DG, Irvine PJ, Kravitz B, Horton JB. Technical characteristics of a solar geoengineering deployment and implications for governance. Clim Policy. 2019 Nov;19(10):1325-39.
 Biermann F, Oomen J, Gupta A, Ali SH, Conca K, Hajer MA, et al. Solar geoengineering: The case for an international non-use agreement. Wiley Interdiscip Rev Clim Change. 2022 May;13(3).
 Carlson CJ, Colwell R, Hossain MS, Rahman MM, Robock A, Ryan SJ, et al. Solar geoengineering could redistribute malaria risk in developing countries. Nat Commun. 2022 Apr;13(1):2150.
 Robock A, Jerch K, Bunzl M. 20 reasons why geoengineering may be a bad idea. Bull At Sci. 2008 May;64(2):14-59.
 Tilmes S, Fasullo J, Lamarque JF, Marsh DR, Mills M, Alterskjær K, et al. The hydrological impact of geoengineering in the Geoengineering Model Intercomparison Project (GeoMIP). J Geophys Res. 2013 Oct;118(19):11, 036-11, 058.
 Reynolds JL. A Path Forward. In: The Governance of Solar Geoengineering: Managing Climate Change in the Anthropocene. Cambridge University Press; 2019. p. 196-220.
 Issues. Toward a responsible solar geoengineering research program; 2017. Accessed: 2022-8-28. https://issues.org/ toward-a-responsible-solar-geoengineering-research-program/.
 Mach KJ, Kraan CM, Adger WN, Buhaug H, Burke M, Fearon JD, et al. Climate as a risk factor for armed conflict. Nature. 2019 Jul;571(7764):193- 7.
 c2g. Recalibrating our work after the UNEA resolution; 2019. Accessed: 2022-8-28. https://www.c2g2.net/ recalibrating-our-work-after-the-unea-resolution/.
 Give research into solar geoengineering a chance. Nature. 2021 May;593(7858):167.
 Temple J. The US government is developing a solar geoengineering research plan. MIT Technology Review. 2022 Jul.
 Christina Henriksen, Johanna Sandahl, Mikael Sundstr¨om, Isadora Wron ski. SCoPEx Advisory Committee, editor; 2021.
 Duda R. Global priorities research - 80,000 Hours; 2016. Accessed: 2022-8-28. https://80000hours.org/problem-profiles/ global-priorities-research/.
I wrote the 80,000 Hours problem profile on climate change. Thank you so much for this feedback! I’m genuinely really grateful to see such engagement with the things I write - and criticism is always a welcome contribution to making sure that I’m saying the right things.
Just to be clear, when I said “we think it’s potentially harmful to do work that could advance solar geoengineering”, I meant that (with a fair degree of uncertainty), it could be harmful to do work that advances the technology (which I think you agree with) not that all research around the topic seems bad! It definitely seems plausible that some research on the topic might be good - but I was trying to recommend the very best things to do to mitigate climate change. My reviewers pretty much all agreed that, partly as a result of potential harmful effects, it doesn’t seem like SRM research would be one of those very best things, and so suggested that we stop recommending working in the area. In large part I’m deferring to this consensus among the reviewers on this.
Hope that helps!
Hey! Arden here, from 80,000 Hours. Thanks for this in-depth feedback! The author of our climate change profile is away right now, so we'll take a look at this in a couple of weeks.
That's ok! Thanks for the comment, and sorry for the such in depth response- I know it was only a few lines in the profile, but you guys are super important (and super cool) and so I think its important you get this right. I would be really happy to have a chat about this at any point as well, feel free to message me if this is something you would be interested for me to explain through any uncertainties you have with this feedback.
No problem of course - in-depth is great, and thanks for the offer to chat! Agree this is important to get right. I'll pass this on to the author and he'll get back in touch if it seems helpful to talk through : )