The case for delaying solar geoengineering research

Tl;dr:

Ar­gu­ment:

1. So­lar geo­eng­ineer­ing is not fea­si­ble for the next few decades.

a. So­lar geo­eng­ineer­ing poses ma­jor gov­er­nance challenges.

b. Th­ese gov­er­nance challenges are only likely to be over­come in at least 50 years’ time.

2. So­lar geo­eng­ineer­ing re­search is a moral haz­ard, and re­search might un­cover dan­ger­ous weather ma­nipu­la­tion meth­ods.

3. Given this risk and given that we can de­lay re­search with­out ob­vi­ous costs, there is a good case for de­lay­ing so­lar geo­eng­ineer­ing re­search at least for a few decades.

Epistemic sta­tus: Seems cor­rect to me, but some ex­pert dis­agree (though I don’t think they have been ex­posed to these ar­gu­ments).


So­lar geo­eng­ineer­ing is a form of cli­mate in­ter­ven­tion that re­duces global tem­per­a­ture by re­flect­ing sun­light back to space. The best stud­ied form—strato­spheric aerosol in­jec­tion—in­volves the in­jec­tion of aerosols, such as sul­phur, into the strato­sphere (the higher at­mo­sphere). This mimics the effects of vol­ca­noes, which can have globally sig­nifi­cant effects via the same mechanism. For ex­am­ple, the Pi­natubo erup­tion in 1991 cooled large parts of the Earth by about half a de­gree. Com­puter mod­el­ling stud­ies have sug­gested that, if done in a cer­tain way and in cer­tain cli­matic con­di­tions, so­lar geo­eng­ineer­ing could elimi­nate many of the costs of global warm­ing with­out hav­ing se­ri­ous side-effects.[1] Th­ese mod­els are of course limited and crude, but they do sug­gest that so­lar geo­eng­ineer­ing could be use­ful tool, if it could be de­ployed and gov­erned safely.

Con­se­quently, in­ter­est in the tech­nol­ogy is in­creas­ing, as dis­cussed in this Economist ar­ti­cle. The Open Philan­thropy Pro­ject has in the past funded so­lar geo­eng­ineer­ing gov­er­nance re­search and com­puter mod­el­ling efforts.

Here, I will ar­gue that we should de­lay so­lar geo­eng­ineer­ing re­search for a few decades.

1. So­lar geo­eng­ineer­ing is not fea­si­ble for the next few decades

a. So­lar geo­eng­ineer­ing poses se­vere gov­er­nance challenges

In my view, so­lar geo­eng­ineer­ing is only likely to be used once warm­ing is quite ex­treme, roughly ex­ceed­ing around 4 de­grees. The rea­son for this is that so­lar geo­eng­ineer­ing would likely be ex­tremely difficult to gov­ern. I out­line some of the gov­er­nance challenges in sec­tion 3.4 of my pa­per on so­lar geo­eng­ineer­ing.

So­lar geo­eng­ineer­ing, if done us­ing the strato­spheric aerosol in­jec­tion method, would af­fect the weather in most or all re­gions.[2] So­lar geo­eng­ineer­ing would there­fore poli­ti­cise the weather in all re­gions, and would have di­verse re­gional effects. Ad­verse weather events would likely be blamed on so­lar geo­eng­ineer­ing by af­fected coun­tries, even if they were not in fact caused by so­lar geo­eng­ineer­ing. Public anger at such weather events would likely be se­vere if they thought a mas­sive in­ter­na­tional weather al­ter­a­tion scheme were at fault. Com­puter mod­els could at best offer highly im­perfect at­tri­bu­tion of weather events to cli­matic causes.

This sug­gests that for so­lar geo­eng­ineer­ing to be fea­si­ble, all ma­jor global pow­ers would have to agree on the weather, a highly chaotic sys­tem. Se­cur­ing such an agree­ment would be ex­tremely difficult in the first in­stance and also ex­tremely difficult to sus­tain in the longer-term. States would also fore­see the prob­lems of sus­tained agree­ment, dis­in­cen­tivis­ing suc­cess­ful agree­ment in the first place.

b. Th­ese gov­er­nance challenges are only likely to be over­come in at least 50 years’ time.

In light of this, so­lar geo­eng­ineer­ing is only likely to be used once cli­mate change is very bad for all re­gions. Judg­ing when this point will oc­cur is difficult, but my best guess hav­ing looked at the cli­mate im­pacts liter­a­ture in some depth is that this would only likely hap­pen af­ter about 3-4 de­grees of warm­ing.

We have had about 1 de­gree of warm­ing thus far and, ac­cord­ing to an IMF re­port, a fur­ther 1 de­gree of warm­ing would be eco­nomi­clly pos­i­tive for many re­gions, es­pe­cially Canada, Rus­sia and Eastern Europe, and even po­ten­tially China (IMF re­port page 15).

(Note that even this mod­est cli­mate change is bad over­all for the world.)

Rus­sia is a cru­cial fac­tor here: global warm­ing seems likely to bring nu­mer­ous eco­nomic benefits for Rus­sia, free­ing up the Rus­sian Arc­tic for ex­plo­ra­tion and thaw­ing po­ten­tial farm­land. It is very un­likely that they would agree to a global scheme that would likely dam­age their eco­nomic prospects. Without agree­ment from Rus­sia, I find it difficult to see how so­lar geo­eng­ineer­ing could ever be im­ple­mented.

Thus, it seems im­plau­si­ble that so­lar geo­eng­ineer­ing would be prac­ti­ca­ble at 2 de­grees of warm­ing, and 4 de­grees is a more plau­si­ble thresh­old, in my view.

How­ever, 4 de­grees of warm­ing will take many decades to oc­cur. On the high­est emis­sions sce­nario con­sid­ered by the IPCC, 4 de­grees of warm­ing would take at least 50 years to oc­cur (IPCC syn­the­sis, p59).

This means that so­lar geo­eng­ineer­ing is only likely to get used by around 2070, giv­ing us 50 years from now to find a solu­tion.

One po­ten­tial counter-ar­gu­ment would point to run­away feed­back loops that cause rapid warm­ing, such as re­lease of mas­sive amounts of methane from clathrates. I have looked at the ev­i­dence for this and the ev­i­dence over­all seems slim and the me­dian view in the liter­a­ture is that this is a neg­ligible risk for the next cen­tury at least. See sec­tion 4 of my write-up on cli­mate and ex risk for more on feed­back loops.

2. So­lar geo­eng­ineer­ing re­search is a moral haz­ard and re­search might un­cover dan­ger­ous weather ma­nipu­la­tion meth­ods

Re­search into so­lar geo­eng­ineer­ing it­self car­ries two main risks.

A per­sis­tent worry about so­lar geo­eng­ineer­ing re­search con­cerns moral haz­ard: the worry that at­ten­tion to plan B will re­duce com­mit­ment to plan A. Hav­ing so­lar geo­eng­ineer­ing as a backup will de­crease com­mit­ment to re­duc­ing car­bon emis­sions, which al­most all re­searchers agree to be the top pri­or­ity. The best dis­cus­sion of this is in Mor­row’s pa­per,[3] and I dis­cuss the con­sid­er­a­tions on moral haz­ard risk at length in sec­tions 4-6 of my pa­per. Over­all, I think this is a gen­uine risk with so­lar geo­eng­ineer­ing re­search and a rea­son not to carry out re­search.

Another risk of so­lar geo­eng­ineer­ing re­search is that it will un­cover new tech­nolo­gies that could desta­bil­ise global civil­i­sa­tion. I dis­cuss weapon­i­sa­tion risks in sec­tion 3.2 of my pa­per. For ex­am­ple, cli­mate re­searcher David Keith has dis­cussed the pos­si­bil­ity that a cer­tain type of nanopar­ti­cle could be much longer last­ing than or­di­nary so­lar geo­eng­ineer­ing and so could po­ten­tially pre­cip­i­tate an ice age if de­ployed for long enough. I don’t think this par­tic­u­lar tech­nol­ogy could ac­tu­ally be a fea­si­ble dooms­day weapon, but there is a con­cern that fur­ther re­search could un­cover dan­ger­ous un­known new geo­eng­ineer­ing tech­nolo­gies.

In a nut­shell, for those per­suaded by the Vuln­er­a­ble World Hy­poth­e­sis, re­search into tech­nolo­gies that could dra­mat­i­cally al­ter the weather seems like the kind of thing we should avoid if we can.

3. We should de­lay so­lar geo­eng­ineer­ing research

So­lar geo­eng­ineer­ing re­search has clear risks and, given that we can­not de­ploy it at least for the next 50 years, there is no need to in­cur these costs now. In­stead, the more pru­dent course seems to be to wait and see how well stan­dard miti­ga­tion efforts go and then, if these con­tinue to fail, start re­search­ing so­lar geo­eng­ineer­ing in earnest around the mid­dle of the 21st cen­tury. This would give us at least 20 years to cover the tech­ni­cal de­tails and a gov­er­nance frame­work. This seems to me like enough time, given that:

  • So­lar geo­eng­ineer­ing is prob­a­bly tech­ni­cally fea­si­ble with adap­ta­tions to var­i­ous differ­ent cur­rent tech­nolo­gies.

  • Ex­tant in­sights from the gov­er­nance of other pub­lic goods, free rider prob­lems, and free driver prob­lems, could in large part be ap­plied to so­lar geo­eng­ineer­ing, adapted to ac­count for the fea­tures of that tech­nol­ogy.

I at least don’t think that we need 50 years of for­ward plan­ning to figure this tech­nol­ogy out if we need to use it. Com­mit­ting re­search hours when we know it may ac­tu­ally be used makes more sense when re­search risks un­der­min­ing frag­ile com­mit­ment to miti­ga­tion, and risks dis­cov­er­ing dan­ger­ous new tech­nolo­gies.

Note that my view has changed on this and that in my pa­per on so­lar geo­eng­ineer­ing, I made a ten­ta­tive case for pri­mar­ily gov­er­nance-fo­cused re­search.

[1] For lay­man’s dis­cus­sion of a re­cent pa­per, see this Vox piece.

[2] The rea­son for this is that the par­ti­cles would be dis­tributed globally by strato­spheric winds.

[3] David R. Mor­row, “Eth­i­cal Aspects of the Miti­ga­tion Ob­struc­tion Ar­gu­ment against Cli­mate Eng­ineer­ing Re­search,” Philo­soph­i­cal Trans­ac­tions of the Royal So­ciety of Lon­don A: Math­e­mat­i­cal, Phys­i­cal and Eng­ineer­ing Sciences 372, no. 2031 (De­cem­ber 28, 2014): 20140062, https://​​doi.org/​​10.1098/​​rsta.2014.0062.