Defusing the mitigation obstruction argument against geoengineering and carbon dioxide removal

by kbog11 min read29th Apr 20219 comments


Climate engineeringClimate change

Geoengineering is deliberate alteration of Earth's climate. It's usually proposed as a way to address global warming: the Earth is getting too hot, so let's do things to make it cooler (the most common idea is to inject sulfate aerosols into the atmosphere).

One of the most prominent arguments against geoengineering is the mitigation obstruction argument, which says that geoengineering would discourage actors from taking other measures to reduce pollution, like reducing fossil fuel use. The same argument also gets used sometimes against carbon dioxide removal.

Halstead (2018) and Morrow (2014) have reviewed the argument, and their arguments outline three general conclusions which I think will be eminently sensible to most readers, including thoughtful geoengineering advocates like David Keith:

  1. Geoengineering will presumably cause some mitigation obstruction, but we have no idea how much.
  2. Depending on the amount of mitigation obstruction and the bigger context of environmental policy, this mitigation obstruction might be legitimately justified (see Morrow section 2), could be bad without outweighing the good done by geoengineering, or could be so bad that it outweighs the good done by geoengineering ("pernicious mitigation obstruction"); the first option might be ruled out given the systematic shortcomings of environmental policy but we really have no idea whether the second or the third will come true.
  3. Given this uncertainty, we should act more cautiously with geoengineering than we would for any other notable environmental policy, but if we pursue limited kinds of research and follow some ethical guidelines then we will keep the risk of harm low enough to be acceptable.

Note that while Halstead focuses on pernicious mitigation obstruction, Effective Altruists (such as Halstead) should still take some interest in the possibility that mitigation obstruction is "bad without outweighing the good done by geoengineering", because that should lower the priority that we assign to geoengineering even if we still think it is a good idea. The mitigation obstruction argument is typically applied only to geoengineering, so that would lower its priority relative to other avenues of environmental policy.

But I would like to revise our view as follows:

  1. Geoengineering poses the same risk or less risk of mitigation obstruction compared to other environmental policies.
  2. We can say with high confidence that geoengineering, like nearly all environmental policies, would cause mitigation obstruction that is bad but not pernicious.
  3. Given this prediction, the mitigation obstruction argument does not provide a good reason to delay geoengineering projects or to be unusually cautious with geoengineering advocacy and research.

Or, put more simply, just forget about the mitigation obstruction argument.

The same applies to carbon capture.

(There are still other issues with geoengineering, see Halstead's paper and EA forum post)

Generalizing the mitigation obstruction argument

As far as I know, no one has taken a critical and consistent look at the mitigation obstruction argument. Mitigation obstruction could be hypothesized for any environmental policy, yet most literature and discourse seems to treat it like something which is either unique to geoengineering or shared only with one or two other interventions (like carbon capture).

The most basic logic of the mitigation obstruction argument is as follows:

  1. Actors become more or less environmentalist depending (among other things) on how much damage they perceive humans to be doing to the environment.
  2. When we do something that solves part of our environmental problems but not all of them, actors become less environmentalist, but there is still a need to protect the environment.
  3. Actors are irrational and/or parochial in ways that encourage them to be insufficiently environmentalist, so we don't want to discourage them from being environmentalist.

"Something that solves part of our environmental problems but not all of them" could be said about all kinds of green policies. When the government pays people to buy electric cars, that solves only part of our climate change problem. So why is the mitigation obstruction argument deployed so discriminately?

Why should mitigation obstruction apply here but not there?

Below I address three reasons people might think of geoengineering and carbon dioxide removal as being particularly prone to mitigation obstruction. It may be that these technologies change the incentives for marginal decarbonization. It may be that people are overly concerned about climate change and leave dirty air relatively neglected. Or it may be that people are prone to irrationally overestimate the effectiveness of geoengineering. But all three of these arguments fail.

Maybe it's about the changes to marginal incentives

The three cases where I have seen people raise the mitigation obstruction argument - geoengineering, carbon capture and 'clean coal' - all have one thing in common: they could be naively perceived as undercutting the environmental benefits of giving up a given amount of dirty energy.

Right now there is a lot of value in going from 10% clean energy to 11% clean energy. If humanity converts to 50% clean energy, this won't change - it will still be equally beneficial to convert to 51% clean energy. But if humanity geoengineers away most global warming or removes most carbon dioxide, then changing 1% of our energy sector is no longer so valuable. Because if we do geoengineering, we're going to geoengineer away most of the excess warming caused by that 1%, and if we do carbon dioxide removal, we're going to capture most of the excess carbon emitted by that 1%. Right? If true, the critic would say, this will be bad because it will reduce the incentive to become fully green.

But geoengineering and carbon dioxide removal don't actually work like that. Their capacity is determined by their own industrial and environmental issues, not by the amount of fossil fuel that we are burning. If we can geoengineer or capture enough to offset 60% of our emissions in 2030, and then in 2031 we reduce our emissions by 1% (as measured at the smokestack), then the environmental damage will not fall from 40% to 39.6%, it will fall to 39%. So it's still a one-percentage-point change whether or not we do geoengineering and carbon dioxide removal.

The reality is more complicated because the consequences of global warming are not linear. Going from 2.0 degrees to 1.9 degrees of warming is less important than going from 4.0 degrees to 3.9 degrees (1, 2). So geoengineering and carbon dioxide removal do reduce the value of marginal improvements by putting us in a less critical situation. But only in the same sense that any green technology does! For instance, if we switch most but not all of our industry to perfectly clean sources, then we'll go from 4 degrees to 2 degrees, with the same result for the incentive to continue decarbonizing.

Pollution mitigation at the smokestack is a different story: it really can reduce the environmental incentive for marginal decarbonization. At the same time, it makes fossil fuels more expensive and therefore improves the economic incentive to decarbonize. And this is in addition to actually immediately reducing pollution. So this stuff definitely still seems like a net positive, even though it may have a political downside relative to other environmental efforts.

Maybe people don't care about dirty air

Geoengineering is a stark example of something that would solve only part of our environmental problems: it would reduce global warming, but wouldn't fix the concentration of pollutants in the atmosphere (which cause a variety of problems like respiratory illness, slowed cognition, and ocean acidification). You might say that mitigation obstruction is caused by misplaced priorities: perhaps people are too concerned about global warming relatively speaking, as they neglect dirty air. If you believe the world is making this systematic error, then you'll be worried that geoengineering will uniquely lead to bad mitigation obstruction.

But the idea that Americans don't care about ordinary dirty air flies in the face of our experience. We passed the Pollution Control Act and the Clean Air Act well before global warming was a mainstream concern. Particulate emissions from cars do not contribute to global warming but are strongly regulated. The Green New Deal resolution called clean air a basic need for every American. The Trump administration promoted coal pollution mitigation. Lead contamination in paint and water are reasonably weighty political issues where decent progress has been made recently, so clearly people care about pollution in its own right.

In fact, it is fairer to say that the American public is irrationally dismissive of global warming while being more reasonable about the other problems of air pollution. While Democrats emphasize climate change in their messaging, they also support the same policies for reasons of reducing air pollution; meanwhile, most Republicans are very dismissive of climate change but seem more open to appeals for clean and healthy air. If there is a forthcoming survey about this, I will bet $1,000 that it will show that Americans asked about a typical environmental policy will express more support when the policy is phrased as "to protect clean air" than when phrased as "to prevent climate change". This implies that geoengineering and carbon dioxide removal are unusually unlikely to cause mitigation obstruction.

It's worth remembering that people do not form their political opinions independently, they are influenced by activists, media and politicians. Global warming may be the defining environmental fear of our generation, but if it stopped being a big problem, then environmentalist opinion peddlers (like me) would just stop talking about it and would instead talk more about dirty air, and if dirty air is a convincing narrative (it is) then people will go along with it. This would merely be the reversal of a shift that happened several decades ago.

I know little about other countries but they seem to care a lot about healthy air too, especially China (which is a particularly big polluter). Correct me if I'm wrong!

Note that as we look at the international system, it may become even harder to argue that dirty air is the neglected side of environmental policy. Dirty air has local concentrated costs, whereas climate change is normally considered a collective action problem. We should expect polities to rationally clean their own air while selfishly deflecting their responsibility to reduce climate change. However, the idea of climate change as a collective action problem was seriously criticized in this recent paper, so maybe we should drop that idea.

Maybe it's just the nature of geoengineering

The final recourse of the geoengineering critic is to say that people will just feel irrationally satisfied by the particular characteristics of geoengineering.

It sure could feel more scalable than it actually is. If you give someone a ten second explanation of stratospheric aerosol injection, they will presumably recognize that dirty air will still be a problem, but they probably would not infer the limitations of geoengineering in terms of mitigating severe global warming or mitigating it over the long run - "just add more aerosols" is the intuitive and shortsighted response. As Lin (2014) puts it, "geoengineering could be inaccurately perceived as a comprehensive insurance policy against climate change".

Additionally, many people will not recognize or lament that because fossil fuel companies will not be hurt by geoengineering (or carbon dioxide removal), they will have an easier time exerting political influence against environmental policy in the future, compared to how they would act if we focused on things like carbon taxes or electric vehicle subsidies which can cause layoffs in the fossil fuel industry and thereby undermine their long-run political power. This political economy issue degrades geoengineering's long-run effectiveness against air pollution. (Yes, unfortunately, the politics of climate change in the United States are so broken that simply harming a major sector of the economy may be considered a reasonable political goal. I realize this is a bit nasty.)

But the problem of excessive optimism isn't unique to geoengineering. Give someone a ten second explanation of solar panels, and they won't recognize that putting solar panels on top of homes and schools is very expensive and not a serious way to reduce fossil fuel dependence, nor will they recognize that the big transmission lines required to move power across the country from the sun belt are politically infeasible due to NIMBY obstructions. Give someone a ten second explanation of thorium reactors, and they won't recognize that their flashy safety and waste disposal features do not make up for their lack of economic viability (1, 2, 3). Give someone a ten second explanation of nuclear fusion, and they won't recognize that it takes one of the bigger difficulties with fission power - the high capex - and blows it up by an order of magnitude. I would take dark pleasure in going on like this but you can take my word for it that there is a general pattern here. Halstead uses the stories of coal pollution mitigation and carbon dioxide removal as examples of environmental technology that have been overhyped in right-wing circles in ways that may displace important solutions, but you can also find examples of left-wing people arguing rightly or wrongly that some conventional climate solutions obviate others, like saying that solar and wind make nuclear unnecessary or that mass transit and livable cities make electric automobiles unnecessary. Ramez Naam notes, "there’s a tendency for climate and energy wonks – and legislators – to focus on electricity and cars when discussing climate policy... [but] our hardest climate problems – the ones that are both large and lack obvious solutions – are agriculture (and deforestation – its major side effect) and industry". To use Lin's phrasing of the problem with geoengineering, the latest generation of power and transportation technology has been "inaccurately perceived by many people as a comprehensive insurance policy against climate change".

Also, whereas things like building electric cars and closing polluting industries actively makes people feel like the environment is being protected, geoengineering has little such effect on a psychological level. If anything geoengineering makes most people perceive the world as even more screwed-up and interfered-with than before, even if the actual global warming problem has shrunk. So geoengineering could actually uniquely fail to take the wind out of the sails of the environmentalist political movement.

An underrated way of determining how people will respond to geoengineering is to ask them what they think about geoengineering. A 2012 poll found that 69% of respondents believed that the harm of geoengineering would outweigh the benefits, but a 2016 poll found that 67% of Americans support solar geoengineering. Adjust this for the big polling lesson of 2016 and 2020, that voters who are low on social trust are underrepresented in polling, and you will see a stiffer response. Still, such polls mean little because opinion will change in the context of real policy debate and implementation. Advocates will steal the term geotherapy to make it sound nice and then tell people that the planet is doomed if we don't spray aerosols in the stratosphere, and detractors will scream all sorts of nasty things about how the real climate doom is the aerosols we sprayed along the way. The net effect? We can't say for sure, but as David Shor has described on numerous occasions, survey-based estimates tend to give excessively rosy expectations for the political viability of ambitious left-wing programs.

Elites are still less open to geoengineering than the general public: only half of climate experts support its use even when it is limited to being an emergency response. Additionally, nearly every country has agreed to an indefinite moratorium on geoengineering until the science is better understood. Nor is it ever proposed seriously in American politics. (The rogue 2020 presidential candidate Andrew Yang called for the mere study of geoengineering, and he caught flak for it from the elite activist class.)

This raises the question: why would you think that people will irrationally treat geoengineering as a panacea, when currently they incorrectly treat it as something that's worse than it actually is? If the history of nuclear power teaches us anything, it's that it takes more than a few decades for people to warm up to climate solutions that perturb them. Ironically, the mitigation obstruction argument, in being applied so discriminately against geoengineering, helps refute itself by showing that actually people are unusually inclined to argue against the potential of geoengineering.

My view on whether geoengineering will cause mitigation obstruction

I believe that most voters are not stupid.[1] I think that the majority of people are concerned about climate change and concerned about dirty air, and will still remain somewhat concerned even if the issues are partially solved. No matter what sorts of partial solutions we adopt, people will accurately decide that continuing to transition to green technology is just moderately less important than it was before.

Geoengineering will not actually be implemented in the near future, though. If it actually makes it through domestic politics in the United States, I think it will be in something like either of the following two scenarios:

  • The climate situation has grown so dire that a strong political movement arises that is willing and able to do literally anything to protect civilization. (Think ecofascism)
  • Scientists and bureaucrats resolve the scientific worries and then, having come to internal agreement, implement their solution under the cover of benign bipartisan neglect. (Think the Federal Reserve, CIA, or fluorine in the tap water)

I don't know about other countries, things could be very different there. Please comment if you can identify a difference.

In the first case, climate panic - I mean real climate panic, not what you get from Greta Thunberg or progressive politicians - would be strong enough to motivate people to try all kinds of things to stop emissions. If by then it's not legitimately too late to make a difference by switching to clean technology (or if we haven't already switched to clean technology), people will support that in parallel with geoengineering. And some of their other solutions to climate change may be a lot darker than geoengineering so we should perhaps be relieved if geoengineering obstructs mitigation for them.

In the second case, the optimistic assumption is that the public will remain dismissive and skeptical about geoengineering, as opposed to explosive backlash. Either way, clearly the public won't stop being environmentalist. As for elites, they will now be people who deeply understand geoengineering and accepted it after a long intellectual process, thus are less likely to overestimate its effectiveness. Of course, we all know (since COVID-19 or the invasion of Iraq) that government agencies can be biased and terrible, but still they usually only do this in ways that give them more funding and prestige, which means they will probably advocate too much climate action not too little.

Mitigation obstruction considered consistently

I've sketched some reasons to think that environmentalist policies generally cause mitigation obstruction, and geoengineering and carbon dioxide removal are not worse for mitigation obstruction compared to other programs. *All this is limited by the fact that my knowledge is mostly confined to the United States.

But what are we to make of the idea that most environmental policies cause comparable amounts of mitigation obstruction? We must conclude that mitigation obstruction, generally speaking, is not pernicious. If pernicious mitigation obstruction were a real thing, then implementing better environmental policies would overall lead humanity to emit more pollution in the long run. But that's absurd. Is there anything but strong evidence that our decades of environmental policy have been a net success?

Note that this logic can also apply to many other policy areas besides the environment. Public health insurance can reduce the drive to lower drug prices. A larger EITC can reduce the drive to lower housing costs. Stricter IRS auditing can reduce the drive to raise taxes. In all cases, we may know in the back of our heads that partial solutions weaken the political demand for other solutions, but we never seriously consider it a reason to be more timid about some particular policy.

  1. Although I do think it is a little too easy to scare them away from untested technologies. ↩︎


9 comments, sorted by Highlighting new comments since Today at 11:39 AM
New Comment

I ultimately agree with you but I think you miss the best argument for the other side. I think it goes like this:

  1. Humans are particularly bad at coordinating to reduce harms that are distant in time or are small risks of large harms. In other words out of sight out of mind. We are much better at solving problems which we experience at least some current harm from but prefer to push off harms into the future or a low probability event.

The argument for this point is buttressed by the very fact that we aren't doing anything about warming right now.

  1. Geoengineering takes the continuous harms of from increasing temperatures and renders them discontinuous and increases the risk of sudden major negative effects.

The argument here is that Geoengineering let's us eliminate all negative effects as long as it is effective but if the Geoengineering mechanism ever fails we experience all the built up warming at once. Maybe we get hit by a bit solar flare and can't launch our sunshade or shoot our sulfur into the stratosphere.

I'm not sure if immediacy of the problem really would lead to a better response: maybe it would lead to a shift from prevention to adaptation, from innovation to degrowth, and from international cooperation to ecofascism. Immediacy could clarify who will be the minority of winners from global warming, whereas distance makes it easier to say that we are all in this together.

At the very least, geoengineering does make the future more complicated, in that on top of the traditional combination of atmospheric uncertainties and emission uncertainties, we have to add uncertainty about how the geoengineering regime will proceed. And most humans don't do a great job of responding to uncertain problems like this.

But I don't think we understand these psychological and political dynamics very well. This all reminds me of public health researchers, pre-COVID, theorizing about the consequences of restricting international travel during a pandemic.

I'll think a bit more on this.

I like your post because it puts some more backstory behind an argument that many people usually accept for face value.

I don't quite understand this argument:

If we can geoengineer or capture enough to offset 60% of our emissions in 2030, and then in 2031 we reduce our emissions by 1% (as measured at the smokestack), then the environmental damage will not fall from 40% to 39.6%, it will fall to 39%. So it's still a one-percentage-point change whether or not we do geoengineering and carbon dioxide removal.

This assumes that geoengineering will cause effects equivalent to removing a certain amount of CO2 from the athmosphere. So each additional ton of CO2 still has the same effect. Is that a good model for the effects of geoengineering? I would naively assume that geoengineering could, by blocking sunlight through aerosols, reduce the impact that every ton of CO2 in the athmosphere has. If this is the case, then 39.6% would be accurate and the argument wouldn't work.

Hm, I suppose I don't have reason to be confident here. But as I understand it:

Stratospheric aerosol injection removes a certain wattage of solar radiation per square meter.

The additional greenhouse effect from human emissions only constitutes a tiny part of our overall temperature balance, shifting us from 289 K to 291 K for instance. SAI cuts nearly the entire energy input from the Sun (excepting that which is absorbed above the stratosphere). So maybe SAI could be slightly more effective in terms of watts per square meter or CO2 tonnes offset under a high-emissions scenario, but it will be a very small difference.

Would like to see an expert chime in here.

I fear that we need to do Geoengineering right away or we will be locked into never undoing the warming. Problem is a few countries like russia massively benefit from warming and once they see that warming and then take advantage of the newly opened land they will see any attempt to artificially lower temps as an attack they will respond to with force and they have enough fossil fuels to maintain the warm temps even if everyone else stops carbon emissions (which they can easily scuttle).

IMO this concern is more persuasive than the risk of trying Geoengineering.

But I disagree that Geoengineering isnt going to happen soon. All the same reasons we aren't doing anything about global warming now are reasons that we'll flip on a dime when we start seeing real harms.

I fear that we need to do Geoengineering right away or we will be locked into never undoing the warming. Problem is a few countries like russia massively benefit from warming and once they see that warming and then take advantage of the newly opened land they will see any attempt to artificially lower temps as an attack they will respond to with force and they have enough fossil fuels to maintain the warm temps even if everyone else stops carbon emissions (which they can easily scuttle).

Deleted my previous comment - I have some little doubts and don't think the international system will totally fail but some problems along these lines seem plausible to me

[+][comment deleted]15d 2

Great post, thanks! Some thoughts:

1. I agree that the mitigation obstruction argument is over-sold and not particularly strong, for many of the reasons you outline (in the real world, it seems to me that most pressure to do something on climate is not addressed by geo-engineering, so politicians face little incentive to over-emphasize geo-engineering, rather they will underemphasize it compared to potential given the demonization and the perceived strength of the obstruction argument in public discourse).

2.  That said, you do seem to miss one key aspect of solar geo-engineering compared to other environmental interventions, that it is incredibly cheap and quickly to implement, which is not true of other climate solutions with similar mitigation / damage prevention potential. So there is  a way in which the obstruction argument could be stronger "let's not do this expensive infrastructure investment into clean tech now, let's do cheap geo-engineering in 10 years if things go bad". 

3. I think the mitigation obstruction argument should not matter much either way for relatively small-scale marginal actions (such as supporting charities that advance geo-engineering research, or carbon dioxide removal, etc.). That is because those ideas exist and it seems implausible that the strength of mitigation obstruction arguments working in public discourse is very sensitive to whether we (societally, globally) spend 5 or 20 million a year working on this, either way you can point to solar geo-engineering as a reason to not act (this is related to 2, the low cost makes this more likely).

4. The worlds I am worried about the most are worlds where that mitigation obstruction argument did some work with obstructing (a) climate action and (b) action on geo-engineering, carbon dioxide removal, etc. and then we are actually not ready to deploy even if we wanted to. 

5. I think this is a real risk for carbon dioxide removal solutions (hence, supporting Carbon180) as finding things that work reasonably cheaply when we want to mass deploy requires targeted innovation and search now; I am less worried about solar geo-engineering because it can be researched and deployed quickly (that it is too expensive is not a concern with solar geo-engineering, it's all about risk trade-offs and governance) .


[+][comment deleted]16d 1