I won't have time to dig into the details here, but wanted to clarify one thing on the top level conclusion which is that:
(1) I agree that climate interventions are not the top intervention to reduce existential risk on the margin (I am also on the record saying that in the GWWC Climate write-up).
(2) Whenever I am asked about allocation of marginal resources (people or funding) between climate and longtermist priority areas, I tend to nudge people away from climate.
The goal of the Climate Fund is to optimize spending that is not cause-neutral, to have as big as possible a positive impact given constraints, it is not intended or marketed as "the top bet on reducing existential risk" and we are careful to not crowd in resources that would otherwise go to areas we think of as higher marginal priority.
PS: I tried to convey that in the Introduction by saying that "mitigating existential risk as cost-effectively as possible does not correspond to FP's mission[2]".
PPS: I have now edited the Introduction to point to your and Matt' comments, and thus better describe FP's goals. I no longer mention that "mitigating existential risk as cost-effectively as possible does not correspond to FP's mission".
Thanks for all of your work on this analysis, Vasco. We appreciate your thoroughness and your willingness to engage with us beforehand. The work is obviously methodologically sound and, as Johannes indicated, we generally agree that climate is not among the top bets for reducing existential risk.
I think that "mitigating existential risk as cost-effectively as possible" is entailed by the goal of doing as much good as possible in the world, which is why FP exists. To be absolutely clear, FP's goal is to do the maximum possible amount of good, and to do so in a cause-neutral way.
A common misconception about our research agenda is that it is driven by the interests of our members. This is most assuredly not the case. To some degree, member-driven research was a component of previous iterations of the research team, and our movement away from this is indeed a relatively recent change. There remain some exceptions, but as a general rule we do not devote research resources to any cause area or charity investigation unless we have a good reason to suspect it might be genuinely valuable from a strictly cause-neutral standpoint.
Still, FP does operate under some constraints, one of which is that many of our 1700 members are not cause-neutral. This is by design. We facilitate our members' charitable giving to all (legal and feasible) grantees in hopes that we can influence some portion of this money toward highly effective ends. This works. Since our members are often not EAs, such giving is strictly counterfactual: in the absence of FP's recommendations, it simply would not have been given to effective charities.
Climate plays two roles in a portfolio that is constrained in this way. First, it introduces members who are not cause-neutral to our way of thinking about problems and solutions, which builds credibility and opens the door to further education on cause areas that might not immediately resonate with them (e.g. AI risk). This also works. Second, it reallocates non-cause-neutral funds to the most effective opportunities within a cause area in which the vast majority of philanthropic funds are, unfortunately, misspent. As I have tried to work out in my Shortform, this reallocation can be cost-effective under certain conditions even within otherwise unpromising cause areas (of which climate is not one).
Finally, I do want to emphasize that the Climate Fund does not serve a strictly instrumental role. We genuinely think that the climate grants we make and recommend are a comparatively cost-effective way to improve the value of the long-term future, though not the most cost-effective way. I don't see any particular tension in that: every EA charity evaluator (or grantmaker) recommends (or grants to) options across a wide range of cost-effectiveness. From our perspective, the Climate Fund is better than most things, but not as good as the best things.
Thanks, Matt! I think the above is a very valuable explanation of FP's mission, and have edited the last 2 paragraphs of the Introduction to point to your and Johannes' comments.
This is fantastic to finally have a cost-effectiveness model for the long-term future for climate change!
This conclusion would hardly change due to including effects of removing GHG emissions which do not lead to trajectory changes.
I do think that trajectory changes such as making it more likely that worse values end up in AI, perhaps caused by an abrupt climate change catastrophe triggering a cascade, could be a significant multiplier for cost-effectiveness. I don't think you clarified whether when you say tonnes of carbon whether you actually mean carbon (which academia typically uses), or CO2 (which industry/popular typically uses). I think it makes sense that you do cost-effectiveness, e.g. of CCF as (2 kt/$), because of issues dealing with uncertainty. But I think it's worth pointing out that this cost-effectiveness is around five orders of magnitude higher than typical climate actions, such as subsidizing renewable energy or electric vehicles (~$100/tC). For the cost-effectiveness bar (bp/G$) estimates, it seems to me that the people were answering a different question than Open Phil, where the former were referring to the bar now, and the latter was referring to the last dollar. I think the last dollar of Open Phil is more appropriate (especially because the last dollar of EA would be significantly lower cost effectiveness than this), and would allow climate change to be about two orders of magnitude less cost-effective and still meet the bar.
In addition, it should be noted there seem to be opportunities whose cost-effectiveness is above the bar of 2 bp/G$. Denkenberger 2021 and Denkerberger 2022 estimate the following 5th and 95th percentiles[14] (in bp/G$):
To be clear, these resilience cost-effectiveness estimates include not just the amelioration of climate catastrophes, but also nuclear and others (and they include trajectory changes).
I don't think you clarified whether when you say tonnes of carbon whether you actually mean carbon (which academia typically uses), or CO2 (which industry/popular typically uses).
I meant tonnes of GHG, namely CO2e. I have now clarified this in the 1st footnote.
But I think it's worth pointing out that this cost-effectiveness is around five orders of magnitude higher than typical climate actions, such as subsidizing renewable energy or electric vehicles (~$100/tC).
I agree. The 1st bullet point of the Discussion now addresses this.
I think the last dollar of Open Phil is more appropriate (especially because the last dollar of EA would be significantly lower cost effectiveness than this), and would allow climate change to be about two orders of magnitude less cost-effective and still meet the bar.
I think the cost-effectiveness of the last dollar of OP's longtermist projects is similar to that of the last dollar of other organisations in the EA space funding longtermist projects, because I believe the money of OP and such organisations is fungible to a reasonable extent. I would say OP's estimate of 0.05 bp/G$ is lower than the others mostly because it is an underestimate.
I'm asking to get a sense of what the current margin of funding looks like, as a way to help researchers and others prioritize our efforts.
So I don't think they were answering for last dollar. I believe Open Phil said that it is willing to fund anything that meets its last dollar bar now, but their current bar could be higher now. I guess you could argue that there is uncertainty in what the last dollar bar will be, so we should stay at the current bar for the foreseeable future. However, if something is urgent, that is we could miss out on the X-risk reduction if we don't do it in the next few years, I think that we should be using the last bar for current decisions. The case of climate change is complicated because the impacts are mostly a century or so in the future, but you could argue that there are opportunities to decarbonize now that will be gone in the future because that carbon will already be emitted. I think the case is clearer for resilience to catastrophes that could happen in the next few years, such as nuclear war, that they are urgent. Another way to think about it is if you would eventually fund these things because they are above the last dollar bar, you get a better benefit to cost ratio by doing it now because you have more overall X-risk reduction (assuming the interventions are long lived).
Here is a simple point estimate illustrating how one could get the cost-effectiveness of decreasing GHG emissions from a longtermist perspective: "cost-effectiveness of decreasing GHG (1 kt/$)"*"increase in the value of the future if all risk from climate change was eliminated (0.1 bp)"/"reduction in GHG emissions to eliminate roughly all risk from climate change (4 Tt)" = 0.025 bp/G$, which would be half of OP's conservative bar.
The results were obtained with this Colab, and the key ones are summarised in the table below[1] (for more, see Results and Discussion). Comments about how to interpret them are welcome.
The cost-effectiveness bar for reducing existential risk (8 kt/$) is estimated to be 3 times as high as (an overestimate for?) the cost-effectiveness of CCF (2 kt/$) (see Discussion). This suggests the best interventions to fight climate change are not amongst the most effective ways of reducing existential risk.
Result
Mean
Existential risk due to climate change (bp)
1.00
Cumulative GHG emissions between 2020 and 2100 (Tt)
3.76
Existential risk reduction caused by removing GHG emissions (bp/Tt)
0.273
Cost-effectiveness of removing GHG emissions via CCF (t/$)
2.34 k
Cost-effectiveness of CCF (bp/G$)
0.640
Cost-effectiveness bar for reducing existentiat risk (bp/G$)
2.17
Cost-effectiveness bar for removing GHG emissions (t/$)
7.94 k
Acknowledgements
Thanks to Alexey Turchin, Anonymous Person, David Denkenberger, Johannes Ackva, Luke Kemp, and Nuño Sempere.
Introduction
Removing greenhouse gas (GHG) emissions decreases heat-related deaths: Bressler 2021 estimated the 2020 mortality cost of carbon to be 2.26*10^-4 life/t. However, most of the benefits of reducing GHG emissions respect the decrease in existential risk due to climate change.
This analysis estimates the existential risk reduction caused by removing GHG emissions, and the cost-effectiveness of CCF, which is compared with an estimate for the cost-effectiveness bar for reducing existential risk. Nonetheless, it should be noted that, according to this comment from Johannes Ackva[2]:
The goal of the Climate [Change] Fund is to optimize spending that is not cause-neutral, to have as big as possible a positive impact given constraints, it is not intended or marketed as "the top bet on reducing existential risk" and we are careful to not crowd in resources that would otherwise go to areas we think of as higher marginal priority.
I encourage the readers to check this comment from Matt Lerner, FP's research director, to better understand FP's mission:
To be absolutely clear, FP's goal is to do the maximum possible amount of good, and to do so in a cause-neutral way.
Methodology
The (marginal) cost-effectiveness of CCF was estimated from the product between:
The existential risk reduction caused by removing GHG emissions, which was estimated from the ratio between:
The existential risk due to climate change.
The cumulative GHG emissions between 2020 and 2100 assuming current climate policies[3].
The cost-effectiveness of removing GHG emissions via CCF, which was estimated from:
The reciprocal of the cost to remove GHG emissions via CCF.
The assumption of current climate policies regarding the cumulative GHG emissions is consistent with the current Metaculus' cummunity prediction of 2.6 ºC for how much greater (in ˚C) will the average global temperature in 2100 be than the average global temperature in 1880. According to the Climate Action Tracker (CAT), current climate policies are predicted to result in a global warming between 2.5 ºC and 2.9 ºC (interval which contains 2.6 ºC). The modelling of the cumulative GHG emissions would ideally consider more scenarios. The exclusion of the emissions after 2100 until net zero, after which the existential risk due to climate change is arguably negligible, tends to overestimate the cost-effectiveness.
Existential risk due to climate change
The existential risk due to climate change was modelled as a beta distribution with:
Mean equal to 0.01 %, which was determined from the geometric mean between:
The 0.1 % guessed by Toby Ord in The Precipice for the next 100 years (2021 to 2120).
The upper bound of 0.01 % guessed by 80,000 Hours here[4].
The 0.001 % respecting the John Halstead's best guess presented here[5].
Ratio between the mode and mean equal to that of the total existential risk considered in Denkenberger 2022, which is defined as a beta distribution with parameters alpha and beta of 1.5 and 8 (see section 2.3).
This led to parameters alpha and beta having values of 1.73 and 17.3 k[6].
The distribution defined here should not be considered resilient. As discussed in Kemp 2022, bad-to-worst-case scenarios of climate change are underexplored.
Cumulative GHG emissions between 2020 and 2100
The cumulative GHG emissions between 2020 and 2100 assuming current climate policies were modelled as a lognormal distribution with 25th and 75th percentiles equal to 3.333 Tt and 4.136 Tt, which are the lower and upper estimates of CAT. This led to values in Tt for the mean and standard deviation of 3.73 and 0.595.
Cost to remove GHG emissions via CCF
The cost to remove GHG emissions via CCF was modelled as a lognormal distribution with 2.5th and 97.5th percentiles equal to the lower and upper bound of the 95 % confidence interval provided by Johannes Ackva (via personal communation): in $/t, 10^-4 to 10. These should be intended as informed guesses, not resilient estimates. FP is working to produce a more robust cost-effectiveness distribution.
The guesses were assumed to account for the value of removing emissions as a function of global warming. For example, assuming existential risk due to climate change increases quadratically with global warming, and that this increases linearly with cumulative emissions, removing 1 t at 4 ºC of global warming would be twice as valuable as removing 1 t at 2 ºC of global warming.
Results
The results are presented below for a Monte Carlo simulation with 10 M samples. I encourage the readers to make a copy of the Colab model, and select their preferred parameters. The model is fully commented (the inputs section is at the top), and could be run in 6 s for 10 M samples.
Result
Mean
Standard deviation
5th percentile
Median
95th percentile
Existential risk due to climate change (bp)
1.00
0.760
0.146
0.815
2.48
Cumulative GHG emissions between 2020 and 2100 (Tt)
3.76
0.606
2.85
3.71
4.83
Existential risk reduction caused by removing GHG emissions (bp/Tt)
0.273
0.215
0.0387
0.219
0.691
Cost-effectiveness of removing GHG emissions via CCF (t/$)
2.34 k
90.0 k
0.252
31.6
3.97 k
Cost-effectiveness of CCF (bp/G$)
0.640
34.3
39.1 μ
6.26 m
0.953
Discussion
The mean cost-effectiveness of removing GHG emissions via CCF of 2.34 kt/$ appears to be an overestimate:
7 k times as high as the cost-effectiveness of 0.3 t/$ for "reforestation".
50 k times as high as the cost-effectiveness of 0.04 t/$ for "wind energy subsidies".
200 k times as high as the cost-effectiveness of 0.01 t/$ for "concentrating solar power expansion (China & India)", "renewable fuel subsidies", and "livestock management policies".
1 M times as high as the cost-effectiveness of 0.002 t/$ for "solar photovoltaics subsidies".
It is 2 k times as high as 1 t/$, which is arguably the bar FP considers.
It is 70 times as high as the "optimistic" cost-effectiveness of 31.4 t/$ estimated by FP here for the future projects of Clean Air Task Force[8] (CATF). However, Johannes thinks the 2018 cost-effectiveness analysis which produced this estimate "radically underestimated the real uncertainty in both directions".
However, the estimated mean cost-effectiveness of CCF is still smaller than most of the cost-effectiveness bars for reducing existential risk. These are summarised in the table below, and were taken from the answers to this question from Linchuan Zhang[9], or provided via personal message[10].
The mean cost-effectiveness of CCF of 0.6 bp/G$ only exceeds OP's conservative (lower) bar. The median cost-effectiveness bar of 2.17 bp/G$ is equivalent to 7.94 kt/$ (= 2.17/0.273), which is 3 times (= 2.17/0.640) as high as the estimated cost-effectiveness of CCF, and 8 k times as high as 1 t/$.
Assuming interventions funded by CCF are amongst the best opportunities to remove GHG emissions, the above suggests interventions to fight climate change are not amongst the most effective ways of reducing existential risk.
In addition, it should be noted there seem to be opportunities whose cost-effectiveness is above the bar of 2 bp/G$. Denkenberger 2021 and Denkerberger 2022 estimate the following 5th and 95th percentiles[14] (in bp/G$):
"Far future potential increase per $ due to resilient foods average over ∼$100 million S model": 20 and 20 k.
"Far future potential increase per $ due to resilient foods average over ∼$100 million E model": 30 and 80 k.
"Far future potential increase per $ AGI safety research at the $3 billion margin (same for both models)": 0.2 and 100.
This conclusion would hardly change due to including effects of removing GHG emissions which do not lead to trajectory changes. For example, the direct benefits of reducing existential risk due to climate change which result from removing GHG emissions may well be over 100 k times as large as those from decreasing temperature-related mortality between 2020 and 2100, supposing:
The benefits of reducing existential risk are 27.5 life/t, based on the following assumptions:
Existential risk reduction caused by removing GHG emissions of 0.275 bp/Tt (see Results).
Population size of 10 G (currently, it is 8 G).
Existence of 10 Gyear, which is given as a lower bound in Beckstead 2013 (search for “expected years of civilization ahead of us”).
Life expectancy of 100 year/life (currently, it is 70 year/life).
The benefits of decreasing temperature-related mortality between 2020 and 2100 are 2.26*10^-4 life/t, as given by the mortality cost of carbon estimated in Bressler 2021.
That said, we [80,000 Hours] still think this risk is relatively low. If climate change poses something like a 1 in 1,000,000 risk of extinction by itself, our guess is that its contribution to other existential risks is at most a few orders of magnitude higher — so something like 1 in 10,000.
With those caveats in my mind, my best guess estimate is that the indirect risk of existential catastrophe due to climate change is on the order of 1 in 100,000, and I struggle to get the risk above 1 in 1,000. Working directly on US-China, US-Russia, India-China, or India-Pakistan relations seems like a better way to reduce the risk of Great Power War than working on climate change.
The cost-effectiveness estimates were calculated from the reciprocal of the point estimates or the geometric mean of the lower and upper bounds of the values in $/t presented in Gillingham 2018.
The estimates of Kirsten Horton and Nuño Sempere were not included, as they were seemingly supposed to be underestimates of their cost-effectiveness bars.
This refers to OP's longtermist projects, and is based on the estimate of "$200 trillion per world saved" provided by Ajeya Cotra in this section of episode 90 of The 80,000 Hours Podcast. It concerns "meta R&D to make responses to new pathogens faster", and "[Open Philanthropy] were aiming for this to be conservative", meaning the actual cost-effectiveness bar is higher.
Geometric mean between the lower and upper bound provided. "I currently think it's probably somewhere between $1-100 trillion per existential catastrophe averted (I find this framing more intuitive than bp/G$)".
This is the median cost-effectiveness bar provided by Linchuan. According to "[Linchuan's] very fragile thoughts as of 2021/11/27": "I feel pretty bullish and comfortable saying that we should fund interventions that we have resilient estimates of reducing x-risk ~0.01% at a cost of ~$100M" (10 bp/G$); "I think for time-sensitive grants of an otherwise similar nature, I'd also lean optimistic about grants costing ~$300M/0.01% of xrisk, but if it's not time-sensitive I'd like to see more estimates and research done first" (3.33 bp/G$); "for work where we currently estimate ~$1B/0.01% of xrisk, I'd weakly lean against funding them right now, but think research and seed funding into those interventions is warranted" (1 bp/G$).
The quantiles are expressed in 1/$ in the articles, but in bp/G$ here. However, for loss of industry preparation and resilent foods, they do not apply (as accurately) to investments of 1 G$, as they were computed for investments between 30 M$ and 100 M$.
I used AI to fix transcription errors, rerrarange the ideas, and suggest tweaks to the title and some sentences.
Three of the most exciting projects to come out of EA in recent years are, in a vague sense, CEA spinouts:
* Kairos is directly a spinout of CEA and now handles most support for university AI safety groups. Basically everyone I've found who knows them is really excited about what they do
* NEST is an opinionated ideas-fi...
TLDR; To help the effective animal advocacy movement cost-effectively absorb greater amounts of funding in the near future, we are seeking expressions of interest from people who could found a new organization focused on:
* Highly neglected animals: insects, wild animals, shrimp, fish, etc, or
* AI and animals: AI alignment and governance for animal welfare, strategic actions considering transformative AI, AI for wild animals, etc.
* ...
Thanks for this, Vasco!
I won't have time to dig into the details here, but wanted to clarify one thing on the top level conclusion which is that:
(1) I agree that climate interventions are not the top intervention to reduce existential risk on the margin (I am also on the record saying that in the GWWC Climate write-up).
(2) Whenever I am asked about allocation of marginal resources (people or funding) between climate and longtermist priority areas, I tend to nudge people away from climate.
The goal of the Climate Fund is to optimize spending that is not cause-neutral, to have as big as possible a positive impact given constraints, it is not intended or marketed as "the top bet on reducing existential risk" and we are careful to not crowd in resources that would otherwise go to areas we think of as higher marginal priority.
Thanks for clarifying, Johannes!
PS: I tried to convey that in the Introduction by saying that "mitigating existential risk as cost-effectively as possible does not correspond to FP's mission[2]".
PPS: I have now edited the Introduction to point to your and Matt' comments, and thus better describe FP's goals. I no longer mention that "mitigating existential risk as cost-effectively as possible does not correspond to FP's mission".
(I am research director at FP)
Thanks for all of your work on this analysis, Vasco. We appreciate your thoroughness and your willingness to engage with us beforehand. The work is obviously methodologically sound and, as Johannes indicated, we generally agree that climate is not among the top bets for reducing existential risk.
I think that "mitigating existential risk as cost-effectively as possible" is entailed by the goal of doing as much good as possible in the world, which is why FP exists. To be absolutely clear, FP's goal is to do the maximum possible amount of good, and to do so in a cause-neutral way.
A common misconception about our research agenda is that it is driven by the interests of our members. This is most assuredly not the case. To some degree, member-driven research was a component of previous iterations of the research team, and our movement away from this is indeed a relatively recent change. There remain some exceptions, but as a general rule we do not devote research resources to any cause area or charity investigation unless we have a good reason to suspect it might be genuinely valuable from a strictly cause-neutral standpoint.
Still, FP does operate under some constraints, one of which is that many of our 1700 members are not cause-neutral. This is by design. We facilitate our members' charitable giving to all (legal and feasible) grantees in hopes that we can influence some portion of this money toward highly effective ends. This works. Since our members are often not EAs, such giving is strictly counterfactual: in the absence of FP's recommendations, it simply would not have been given to effective charities.
Climate plays two roles in a portfolio that is constrained in this way. First, it introduces members who are not cause-neutral to our way of thinking about problems and solutions, which builds credibility and opens the door to further education on cause areas that might not immediately resonate with them (e.g. AI risk). This also works. Second, it reallocates non-cause-neutral funds to the most effective opportunities within a cause area in which the vast majority of philanthropic funds are, unfortunately, misspent. As I have tried to work out in my Shortform, this reallocation can be cost-effective under certain conditions even within otherwise unpromising cause areas (of which climate is not one).
Finally, I do want to emphasize that the Climate Fund does not serve a strictly instrumental role. We genuinely think that the climate grants we make and recommend are a comparatively cost-effective way to improve the value of the long-term future, though not the most cost-effective way. I don't see any particular tension in that: every EA charity evaluator (or grantmaker) recommends (or grants to) options across a wide range of cost-effectiveness. From our perspective, the Climate Fund is better than most things, but not as good as the best things.
Thanks, Matt! I think the above is a very valuable explanation of FP's mission, and have edited the last 2 paragraphs of the Introduction to point to your and Johannes' comments.