Comment by MatthewDahlhausen on The Next EA Global Should Have Safe Air · 2022-09-19T16:53:57.539Z · EA · GW

Perhaps I'm reading into it too much, but the implication in the title is that the indoor air at EAG was somehow unsafe.

I don't want people to get that impression, or in the worst case, avoid EAG because they think the air is unsafe.

Venues that have hosted EAG have commercial ventilation. These spaces provide at least a continuous 20 cfm per person of filtered, conditioned air, and the committee that sets those standards is acutely aware of the tradeoff between energy use from higher ventilation rates and health.

There is far greater risk from indoor house after-parties, as older residential homes don't have ventilation and even if they do they were never designed to handle that many people.

We could ask the venue to make sure the filters on the air handlers were changed recently. And/or sprinkle in-room air cleaners throughout the meeting areas. People that have a much lower risk tolerance should just wear masks.

Comment by MatthewDahlhausen on Climate Recommendations in EA: Giving Green and Founders Pledge · 2022-09-19T16:38:52.933Z · EA · GW

UPDATE: The $85/ton 45Q credit passed in the Inflation Reduction Act. This credit was one of Manchin's top priorities.

There are several planned coal and gas power plant projects that owe their existence to the increased credit. Capture efficiencies range from 60-80%, many for enhanced oil recovery. These are not "low carbon" projects.

As one detractor of a new gas plant in West Virginia says, “if the objective is to decarbonize our energy sector at the lowest possible cost, with the greatest reduction of greenhouse emissions, and with the greatest amount of job creation, the development of renewable resources would do a much better job of all three.”

Comment by MatthewDahlhausen on Who would you have on your dream team for solving AGI Alignment? · 2022-08-25T15:39:32.432Z · EA · GW

One design ideation method is instead of trying to think of good ideas, try to think of the worst possible idea.

With that in mind, encourage the writers of "It's Always Sunny in Philadelphia" to do an episode "The Gang Solves AGI Alignment".

Comment by MatthewDahlhausen on Effective altruism's billionaires aren't taxed enough. But they're trying. · 2022-08-25T01:14:40.766Z · EA · GW

Yeah, I think the basic argument holds at surface-level considerations.

It's not just an argument against raising billionaire taxes, but also an argument for reducing them. It raises the question - what is the appropriate tax rate?

At one extreme is zero tax (or even subsidizing billionaires, which one could argue happens quite a bit). Would the world be better off with no billionaire tax and correspondingly much, much smaller U.S. government discretionary spending? Some people with certain political persuasions would say yes. But I think that undervalues the role of shared funding via government services. I'm skeptical billionaires would fill in the funding gap. Plus it ignores how wealth influences politics and wealth inequality influences well-being.

Comment by MatthewDahlhausen on Effective altruism's billionaires aren't taxed enough. But they're trying. · 2022-08-24T16:58:10.089Z · EA · GW

Congress sets the budget, not the president. If you look at proposed white house budgets for the last few decades or so compared to what is enacted by congress, the final budget hews closer to congressional priorities. The exception to this is funding for military endeavors which constitute a significant portion of the discretionary budget. A chunk of military spending is determined by the president and their propensity for wars (see: Bush and the Iraq war, Obama and drone strikes).

The reasoning behind EAs not wanting to tax billionaires seems to be:

  • The average dollar spent by an EA billionaire is much more effective than the average dollar spent by the U.S. government.
  • It seems unlikely the U.S. will change its outlay substantially towards utilitarian values soon
  • Therefore, EAs should not encourage taxing billionaires

I think this is too narrowly focused. Some other considerations:

  • Billionaires are as wealthy as they are because they have an outsized role in setting the political agenda and writing the laws (especially since citizens united).
  • The U.S. government spending is not as effective as it could be because it prioritizes billionaire political preferences over utilitarian ones.
  • If billionaires were taxed more, the marginal increase in government spending would likely be spent on things preferred by the political group that was able to enact greater taxes on billionaires. My impression is that this political group would spend it in ways much more effective than the current average spending.
  • While returns to capital exceed the rate of economic growth, those worried about the well-being of the poor will eventually need to address income inequality. This is especially acute if economic growth cannot continue forever and wealth generation gets closer to being zero-sum. Given the political power wealth buys, it gets harder to correct wealth inequality the longer it goes on and concentrates except with destructive revolutions, if such revolutions are even possible.
  • At extreme levels of wealth inequality (arguably the U.S. is at such levels), wealth inequality reduces economic growth and competitiveness.
  • Well-being is not just absolute but also relative; people's perception of their well-being is heavily influenced by how wealthy they are compared to others. Reducing inequality (to a point) can improve well-being. This is effect is non-linear and dependent on cultural values and on absolute well-being. I'm not sure what the "ideal" Gini coefficient is (, but in general people [prefer a far more equal distribution of wealth than present] (

My take on this is:

  • Taxing billionaires more is necessary to reduce income inequality.
  • Reducing income inequality is a valid instrumental goal. In developed countries, it is likely better at improving well-being than absolute wealth gains.
  • Taxing billionaires should be accompanied by efforts to direct the extra tax dollars effectively.
Comment by MatthewDahlhausen on Historical EA funding data · 2022-08-15T15:00:16.977Z · EA · GW

I find this website helpful for picking colorblind friendly color schemes:

Comment by MatthewDahlhausen on Announcing: EA Engineers · 2022-07-06T16:07:30.237Z · EA · GW

HVAC is mechanical engineering. ASHRAE is the main professional body of mechanical engineers that works on ventilation standards and guidance for reducing aerosol transmission.

Comment by MatthewDahlhausen on The Role of Individual Consumption Decisions in Animal Welfare and Climate are Analogous · 2022-06-10T16:19:00.222Z · EA · GW

I don't think 1) is correct - specificity matters in moral frameworks that aren't purely utilitarian. Consider two scenarios:

  1. I kill a chicken eat it, then offer to buy my neighbor a vegan lunch the next 3 days who would otherwise have gotten chicken.
  2. I release a canister of R22 into the atmosphere, then pay tradewater to acquire and destroy a canister's worth of R22 that would otherwise have been emitted.

In both cases you can say on net there was zero impact. But there is a clear sentient being harmed in one case and not the other. That distinction is similar to why people aren't okay with doctors killing a healthy patient and giving their organs to others, even if it on net saves lives. Put another way, it is possible to wholly offset one's emissions in a that has zero impact. The same can't be said for non-vegan diets.

Comment by MatthewDahlhausen on Is Mineral Resource Scarcity a Risk Anyone is Researching? Worth Researching? · 2022-06-10T15:38:06.698Z · EA · GW

Our local group held a meeting on longtermism this week and one of the questions was "what are time-sensitive (<100yr) longtermism related actions that are NOT related to existential risk?". The two we came up with were the philosophy around space governance/travel and element/mineral resource drawdown. The effective environmentalism subgroup of EA brings up resources drawdown occasionally, but I haven't seen an in depth report on it yet within the EA/EE sphere.

Comment by MatthewDahlhausen on Deference Culture in EA · 2022-06-07T17:01:02.176Z · EA · GW

Thanks for detailing this aspect of EA. I think much of the deference culture is driven by early EA orgs like GiveWell, as you mention. There is a tendency to map the strong deference that GiveWell merits in global health onto other cause areas where it may not apply. For instance, GWWC recommends giving to several funds in different cause areas. The presentation suggests that funds are roughly equal in quality for their respective cause areas. GiveWell has about ~5x more staff than Animal Charity Evaluators, ~10x+ more staff compared to the EA Infrastructure Fund, and ~10x+ more staff compared to Founders Pledge's climate team. To the extent that a larger team size means more research hours and more research hours means better funding decisions, there is a significant difference in funding quality among the different fund recommendations. This difference isn't communicated in public-facing EA media like the GWWC webpage and videos.

As someone who is an expert in a cause area where the EA fund has comparatively little analytical capacity (climate change mitigation), I find the deference and marketing of the climate fund as the most effective giving option a continual source of frustration. I've written about that here and here. I'm also worried about people mapping weak deference onto causes where they should have greater deference: many people early in their EA engagement care about climate change as a cause area. If they have some level of expertise, they may find the climate fund recommendations underwhelming and then incorrectly assume funds in other causes areas have similarly low levels of research behind them. There may be some attrition in getting more people more involved in EA because of this, though it is a tiny niche. I don't think the answer to the comparative deference problem is to do something like delist fund options from the GWWC page. But we do need some way to communicate the differential level of rigor.

Comment by MatthewDahlhausen on UVC air purifier design and testing strategy · 2022-06-01T15:25:47.584Z · EA · GW

A study similar to what you are proposing was first done in the 1930s in Philadelphia as an intervention to prevent the spread of measles. See Nicholas Reed, "The History of Ultraviolet Germicidal Irradiation for Air Disinfection".

There is a substantial knowledge base on the effectiveness of UGVI, associated risks, and design considerations. See Kowalski's "Ultraviolet Germicidal Irradiation Handbook", NIOSH guidelines for UVGI systems, and ASHRAE technical resources on UVGI. ASHRAE is the professional society most responsible for the science, design, and testing of UVGI systems, in coordination with other professional bodies. They have two chapters on UV systems in their design handbooks.

UVGI is one of several control options. Filtration is usually cheaper and doesn't have associated health and safety risks like UVGI. UV-C doesn't provide much additional benefit beyond a MERV 13 filter. UVGI does work nicely and can be cheaper in settings with limited mechanical ventilation, and where additional disinfection is welcome such as in healthcare settings or dense spaces. It is a design challenge that needs to account for things like environmental conditions, ceiling reflectance, and the type of UV-C.

There are underexplored research questions around UVGI, and if this seems interesting to you I'm happy to connect you to some of the scientists and engineers working on this - just PM me. I suggest you read through the technical literature first to be able to talk with an informed perspective on a narrowly targeted research question.

My hunch is that the added disinfection benefit of UVC is not worth the added risk and cost compared to just a MERV-13 air purifier. In the U.S. it may be better to focus on helping schools access the funds in the American Rescue Plan already available for schools to purchase air purifiers. Many school districts just don't have the skills, willingness, or knowledge to know what to buy and manage the program process.

Comment by MatthewDahlhausen on Does it make sense for EA’s to be more risk-seeking in earning to give? · 2022-05-22T01:32:17.310Z · EA · GW

Thanks for laying out the math here. Given the high variance we've seen in the community, it does suggest E2G'rs should go for high-risk high-reward choices.

There is one further consideration that I think needs to be added to the higher variance scenario which comes from concentrating donations from fewer people.

If all the E2G people donate to the same charity, then it makes sense to have higher variance in giving as you laid out. However if the E2G people give to different charities, then donations are now skewed towards the preferred charity of the lucky entrepreneur.

One way to think of this is in terms of dollars donated per thought-hour. Assume each donor spends 10 hours thinking about where to donate, and that the lucky entrepreneur spends 20 hours deciding where to donate. In the lower variance scenario, there are ($10,000 * 10 / (10 * 10) hours) = $1,000 dollars donated per thought hour. In the higher variance scenario, there are ( (($40,000 / (10 hrs * 8 people)) * $40,000 + ($275,000 / 20 hours)* $275,000) / $315,000 = $12,067 dollars donated per thought-hour. We've traded a 3.15x increase in donations for ~4x (12067 / (3.15*1000)) less thoughtfulness.

So while it's great to have more money to donate, it'd be nice for the E2G givers to pre-commit in advance to a target charity, fund, donor lottery, or collective decision making process to not dilute the thoughtfulness behind donations. Another option is that the lucky entrepreneur could simply allocate donation decisions for a small portion of their giving to the other E2G people (say $5k each) and it would gain back all the lost thoughtfulness.

Comment by MatthewDahlhausen on Some potential lessons from Carrick’s Congressional bid · 2022-05-19T01:22:35.048Z · EA · GW

Thanks for clarifying. I agree with you that if the main reason you are supporting a candidate is their potential impact on long-term future oriented policy then the opposing candidate doesn't matter much beyond a simple estimate of their electoral chances vs. your candidate.

Comment by MatthewDahlhausen on Some potential lessons from Carrick’s Congressional bid · 2022-05-18T21:50:32.643Z · EA · GW

Can you elaborate?

I expected this critique when I wrote that claim. I think I understand why someone would see the other candidate as being insignificant. Let me know if I'm presuming the wrong reasons here:

It seemed that the Flynn campaign message was all about pandemic preparedness. At least that's how it was marketed in EA spaces. And it's mostly true that there isn't anybody in congress championing pandemic preparedness. If you are a single-issue voter on pandemic preparedness or AGI, I can see how the opposing candidate doesn't matter to you; your candidate will do more for the cause than any other candidate, regardless of party, who likely doesn't care or have an opinion on it. It's more of a binary. If you care more about existential risks much more than anything else, this reasoning make sense.

But if you care about other causes like animal welfare, local or global poverty, climate change, democracy health, etc., chances are the other candidate does have views on it. If they are a progressive democratic candidate like Andrea Salinas, EA-aligned poverty alleviation, climate change action, and voting reform are significant parts of their platform. Also, one of the key issues in the U.S. presently is whether we are going to retain a semblance of a democracy or if elections are going to be decided by super PACs and gerrymandered state legislators. There is a significant party divide on support for EA-aligned voting reform and bans on alternative voting methods. If you care about being able to influence elections through public appeals, maintaining a functioning democracy matters even if you are a single-issue voter. There is a clear partisan divide. Given an equal chance of winning, would you rather the EA candidate run opposed to someone like Andrea Salinas or Madison Cawthorn?

Comment by MatthewDahlhausen on Some potential lessons from Carrick’s Congressional bid · 2022-05-18T19:05:24.418Z · EA · GW

I'm glad to see EAs running for political office explicitly as EAs. But I hope that the attitude and approach by the EA community towards the Flynn campaign doesn't become the norm. I felt that the campaign was intrusive and pushy, and the standard of care was much lower than what we expect for other causes/interventions.

Some points:

  • I got direct campaign emails from the Flynn campaign, even though I never signed up for campaign emails. Presumably some EA organization gave the Flynn campaign a list of emails or they scraped it off some EA website. I would prefer EA organizations to keep contact information private and adopt an "opt-in" policy for sharing emails. I don't want to get spammed by people asking money for causes or campaigns, especially if EA political campaigns become more frequent.
  • One of my local group co-organizers got a personal appeal from the Flynn campaign in the final days of the election asking them to fly to Oregon to do door-knocking for the campaign saying how it was high expected value. Not only is it a troubling sign that the campaign did not already have a large, local population of door-knockers, but the campaign didn't seem to consider the terrible optics of having people getting paid to fly in from out-of-state to do door-knocking for a few days. This seems anti-democratic.
  • This primary was flooded with billionaire Super PAC money. This is part of an ongoing trend of billionaires buying political power and is detested within the progressive community. It's undemocratic, and we should be cautious about engaging in politics through billionaire money, even if it is 'our' billionaire, and especially if the EA candidate is running in a progressive democratic primary. Even if you think democracy is just an instrumental good you should be worried about the capacity for billionaires to heavily influence elections.
  • The campaign language and EA posts about it, including this one, center entirely around Flynn and not the winner Andrea Salinas, who is also an excellent candidate. The values and views of other candidates the EA candidate is displacing should be a significant consideration in whether to support the campaign. It may be more successful to make EA a constituency for lawmakers, rather than just supporting EA candidates running against progressives.

Furthermore, I'm not sure the information value alone was worth the millions spent on this campaign by the EA community. The 'lessons learned' listed in this forum post seem obvious. I googled "tips for running for congress" and in 10 minutes read through several resources that gave most of these same lessons learned. I expect a 30 min call with a Democrat strategist, of which there are several in the EA movement, would have also given the same lessons learned, and probably would have given a more accurate prediction on the election outcome than the prediction markets cited in this post. Flynn got ~half the vote of the leading candidate, which is more of a blowout than as suggested by the prediction markets. I frequently see parts of the EA community think they've found some new fascinating insight (EA movement learns about 'X') when in fact they are just columbusing knowledge from other communities. It's as if some piece of knowledge must be blessed or learned directly by a well-known EA before it's accepted by the community at large. A little less hubris and a little more humility towards other knowledge domains would save quite a bit effort and resources when learning about things like running for congress.

Comment by MatthewDahlhausen on EA and the current funding situation · 2022-05-12T15:54:16.860Z · EA · GW

Here's a prediction: In the not-too-distant future, someone who calls themselves an effective altruist is going to purchase a private plane or helicopter and justify it saying the time it saves and the amount of extra good they can do with that saved time is worth the expense. The community is going to have a large population that disagrees and sees it as a wasteful extravagance, and a smaller but vocal population that will agree with the purchase as a worthwhile tradeoff, especially if that person is part of a sub-community within EA that is ok with more speculative expected value calculations. Instead of there being a clear, coordinated response disavowing the purchase as extravagant, the community is going to hesitate and argue about the extent to which it is good to feed utility monsters and be muted in its outward response. But that's not going to stop the wider media picking up the story. A small fraction of the population will then henceforth liken EAs to the pastors at megachurches with private jets who use do-gooder justifications for selfish purposes. And yes, you could construct some sort of hypothetical where someone needs a helicopter to more quickly fly between trolley levers to save a bunch of people. But the much more likely scenario is that someone wants a helicopter and is fine using an iffy, cursory justification for it and the trolley brakes are working just fine.

Comment by MatthewDahlhausen on Do you offset your carbon emissions? · 2022-05-05T18:22:41.090Z · EA · GW

I encourage people to offset, but think of offsets as just intentionally paying the costs of externalities from their lifestyle rather than as part of their charity budget.

If you do decide to offset, know that offsets differ greatly in permanence and risk.

You need to buy ~10x the "planting trees" offsets to equal the permanence and risk of a permanent direct air capture and geologic sequestration offset. Use this calculator to compare:

My current recommendation for offsets is, which purchases and destroys high global warming potential refrigerants. It's a permanent counterfactual offset with few displacement/additionality concerns and they are relatively cheap (~$17/ton).

Comment by MatthewDahlhausen on Against immortality? · 2022-04-28T20:43:52.741Z · EA · GW

Humanity and society are weird. By some cosmic fluke involving brains and thumbs we figured out how to mold the landscape to grow our food and later on figured out how to access million-year old energy deposits in the lithosphere.

We are less than two centuries out from the beginning of industrialized society and we have no clue how to balance energy and resource flows to sustain civilization beyond a few more centuries. And now some of us apes are thinking, "hey, how about we don't die?" as if the current weird state of things somehow represents some new normal of human existence.

There has been ample debate around "strong sustainability" vs. "weak sustainability", which centers on how much technological substitution can overcome increasing environmental pressures. People have been using specific, limited examples of weak sustainability being true (see debates around Limits to Growth) to argue against strong sustainability. Its one thing to argue that we can change planetary limits / carrying capacity, and another to say that those limits don't exist. Limits exist; that falls out of some basic thermodynamics.

Pursuing life extension beyond a few centuries seems reckless without figuring out how to do strong sustainability first. With limits, resources are zero-sum beyond some geologic replenishment rate; people living longer trade off against other people, non-human animal, and plant life, or it buys down the resources available to people in the future. I would expect longtermists to be especially cautious about how reckless life-extension could be given limits.

Comment by MatthewDahlhausen on Avian influenza is causing farmers to kill millions of chickens · 2022-04-18T18:48:34.641Z · EA · GW

Vox's most recent Future Perfect newsletter linked to this piece of investigative journalism by the Intercept on the use of VSD: [EDIT: warning that the article includes a video that shows a chicken suffering as it dies]

Comment by MatthewDahlhausen on An uncomfortable thought experiment for anti-speciesist non-vegans · 2022-04-18T15:55:18.458Z · EA · GW

"...seems morally okay as long as the clothes allow you to have more positive impact with your career."

Utilitarian calculations need to be justified beyond just piling up more things in the "positive" bin than the "negative" bin. An often used thought experiment is asking if it is ok if a doctor kills a healthy patient in a hospital to donate their organs to five other needy patients so that they may live. While utilitarians may justify this in the way you did, this justification looks unfounded if there is a recently deceased organ-donor in the morgue at a nearby hospital who could provide all those same organs. How is killing the healthy patient justified then? Would we see the utilitarian doctor as still justified if they said "It's annoying to have to drive over to the other hospital, fill out paperwork, get the organs, then drive back. It is still a net positive to kill the healthy patient here, and it's easier for me, so I'll just do that."?

Considering your analogy, it is easy to buy clothes that didn't require slave labor, and even if not, it is tenuous to see how a specific set of slave-produced clothes would have an overall positive benefit to your career greater than the suffering they incurred.

Bringing it back to animals, the equation isn't the negative of animal suffering against the positive of your career, it's the negative of animal suffering against the marginal career cost, if any, of switching to a vegetarian or vegan diet, which is much lower. You can understand why many would see the claim that the animal suffering is worth it in comparison to marginal personal inconvenience it saves as dubious and particularly self-serving.

Comment by MatthewDahlhausen on A review of Our Final Warning: Six Degrees of Climate Emergency by Mark Lynas · 2022-04-15T18:26:25.349Z · EA · GW

I came away with the same impression when I read it. Thanks for taking the time to highlight specific examples of misinterpretation and lack of nuance. And for running it by the original study authors.

After reading quite a bit of climate doomer literature like Six Degrees, I've become less interested in the extent of exaggeration and portrayed helplessness and more interested in why people are telling the climate story this way. It seems counter-productive. It gives fodder to opponents of action to say the problem is exaggerated. And for the scrupulous it creates noise and the possibility for over-correction or over-reaction. I'm worried the EA movement will develop a well-founded bias to dismiss or ignore studies of potentially serious climate impacts because of the extent of media exaggeration of scientific studies. Looking forward to your climate risk report which I hope will mitigate some of the effects of bad climate science writing.

Comment by MatthewDahlhausen on Response to Kurzgesagt: How We Can Fight Climate Change · 2022-04-14T04:00:14.802Z · EA · GW

Excellent production quality and explanations as usual. I liked the embedded links to the other videos as you reference them.

My feedback is that the donation recommendations don't necessarily represent what I think to be the current best advice in the effective environmentalism (EE) community.

I've been a part of the EE community for 7 years now and have been closely following the development of climate change research and giving recommendations from Founders Pledge and Giving Green (EE orgs). I still don't suggest their recommendations to friends looking to donate to climate change. Let me explain by way of analogy...

Looking with hindsight, if it was 2007, would you have given money to GiveWell's top recommendation Population Services International? Or would you wait, knowing that in just a few years their analysis capacity would expand 10-fold and they would produce much better recommendations such as AMF or SCI?

Founders Pledge and Giving Green are similar to GiveWell circa 2007 in terms of number of employees, money moved, and research depth. The depth and breadth of their analysis is limited. There are whole economic sectors and angles to the climate problem they have not yet investigated - the demand side of the energy equation, alternative proteins, adaptation, and natural feedbacks are examples. In hindsight, during GiveWell's early years, I'd say organizations like the Gates Foundation or JPAL would have given as good or better recommendations if asked. I think the same is true with the current state of climate philanthropy - places like the ClimateWorks Foundation have dozens of experts in a wide range of fields. Their teams on specific interventions are larger than the sum of climate researchers at Founders Pledge and Giving Green. ClimateWorks has considered a wider breadth of interventions and at greater depth than the EE orgs and I think they're covering some of the most impactful interventions (e.g. the cooling collaborative). And that's just one major climate foundation. If I were looking to donate, I'd start there, and I think donating to one of their grantees is likely to do about as much good as following an EE org recommendation.

The EE orgs are a bit different in that they are optimizing to minimize $/ton CO2e avoided (Giving Green) or trying to reduce climate damage with technology back-stops (Founders Pledge) source. GivingGreen has a ClimateWorks Foundation employee on their board, and they are actively trying to see how their work can be complementary rather than repetitive. Given the early stage of these EE orgs, I'd be much more inclined to fund Giving Green directly to hire another researcher than to give money per their recommendations.

Another piece of feedback is that I don't think finding the lowest marginal $/ton CO2e mitigation opportunities is necessarily the right angle. Climate change is characteristically different from other problems like extreme poverty. GiveWell is still oriented towards finding the best health and poverty interventions. And this works because the scale of the problem is so large compared to the funding. But their recommendations would likely change if they had 100-1000x more money and were aiming to eradicate extreme poverty entirely. Climate change is a problem we (humanity) are trying to solve in its entirety, no just at the margin. Several governments have drastic emission reduction targets, and there are massive, established markets to trade offsets. Climate mitigation is valued in a way that helping the poorest isn't. If you find a cheap offset, there is a good chance that funding it isn't fully additional - someone else may come along and fund it anyways. And if you are for sure going to offset, it may make sense to fund higher priced offsets instead to scale them and bring down their costs to the rest of the market. See the discussion and markets around RECs vs. hourly RECs for example. Even then, offset funding isn't really a limitation - there is currently a multi-year backlog for many of the highest priced robust offsets such as direct air capture with mineralization.

Given the offset market, EE orgs are pushed to more speculative mitigation opportunities such as policy change and technology innovation advocacy that can't be as easily calculated or monetized. Pretty much all EE funding to date has gone to one of these two interventions. Once you are in speculative land, there is a different value proposition - similar to how GiveWell's maximum impact fund and the EA Global Health and Development Fund are different. Two things are important in speculative land. First, it no longer makes sense to focus on a specific $/ton CO2e number. It's all about distributions and hits-based giving. And second, the distributions are highly subject to the biases and assumptions that go into making them. Median $/ton CO2e estimates from two different researchers for the same intervention can easily be orders of magnitude apart. Worse, if two researchers share assumptions and biases, the distribution is likely to be too narrow, probably wrong, and possibly even wrong directionally. This is the problem with only 1 or 2 people with similar biases doing the calculations and is the case in the current EE orgs. People donating on advice from the EE orgs now are putting a lot of faith in a few speculative expected value calculations done by just one or two people. I think that faith is unwarranted given the reasons above. And I think the mantles of "recommendations" or a "fund" masks how uncertain and unreliable these donations are compared to other recommendation organizations and cause-area funds in the EA community that easily have 10-100x more research hours put into them as well as greater thought diversity. To me, "A rough conservative estimate of the cost of avoiding one tonne of CO2 by donating to the Clean Air Task Force is just one dollar" is too much like "save a life by buying bednets for just $25". I don't find the particular calculation referenced in the script to be conservative at all - the values and the structure of the analysis are wildly optimistic from my perspective.

I think a more accurate assessment of the best advice from the (EE) community for climate mitigation is "we don't know what the best philanthropic opportunities are; give us money to help find them". I realize that is underwhelming - we want to give concrete recommendations now. But I don't think we can confidently claim that the current EE orgs represent the "biggest impact per dollar donated". At least not yet. My concrete advice for those who only want to give to climate change interventions and only to the "best" interventions at that is to wait. Or fund intervention analyses.

Comment by MatthewDahlhausen on Go Republican, Young EA! · 2022-04-14T03:30:48.524Z · EA · GW

Thanks for clarifying. I think there are few EAs aligned with the Republican party because the party takes an opposite stance to EAs on most issues. Animal welfare, climate change, healthcare, eradicating extreme poverty, support for democracy, inclusion for LGBTQ+, women, immigrants, and minorities are a few.

I see your point that'd it be much easier to rise to power and prominence within the party given the talent vacuum. I think the potential for harm is much greater - would an EA in the republican party work against other EAs in many of the areas I listed, just to advance some more neutral policy goal? How far does that justification go? Is the loss of democracy and lives of many marginal people worth it? I think it is difficult to understate how extreme the party has become over the last 30 years, especially in the last 6 years; it has taken a hard turn towards religious ethno-nationalism and staged an almost successful coup just over a year ago.

I've had engineers describe to me how if they didn't design cluster munitions or land mines, someone else would and they can make them less bad. At what point in that race to the bottom does replaceability stop being a valid justification?

Comment by MatthewDahlhausen on Go Republican, Young EA! · 2022-04-13T14:53:19.122Z · EA · GW

Perhaps it is the upbeat tone of the writing, but I can't tell if this post is meant to be sarcastic or earnest. Can you clarify?

Comment by MatthewDahlhausen on US Taxes: Adjust Withholding When Donating? · 2022-04-12T18:05:39.872Z · EA · GW

Just want to say thanks for posting this. I donate a substantial portion of my income, so I just filled out a new W4 with my employer. I'll get a more even cash-flow and money earlier instead of $10k+ annual refunds.

Comment by MatthewDahlhausen on Where are the actors who are trying to advance the development of nuclear energy in effective ways? Why don't most people trying to advance the development of nuclear energy care about effectiveness? · 2022-04-11T04:09:42.437Z · EA · GW

Who are the actors trying to advance the development of nuclear energy?

Primarily, the U.S. DOE through the Office of Nuclear Energy, with funding of about ~$1.5 billion per year. ~$1.5 billion is about equal to the combined spending on all spending on renewable energy technology innovation (wind, solar, water, geothermal, ~$700 million), advanced manufacturing ($400 million) and building efficiency innovation ($300 million, buildings are ~75% of grid demand). (This is EERE budget minus transportation and the weatherization program). Note the ~$1.5 nuclear estimate does not include the several billions in basic nuclear physics research including fusion funded through the Office of Science; the Office of Nuclear Energy is focused primarily on near-term demonstration of advanced reactor technologies.

To a smaller degree, other countries including Canada, the U.K, Russia, and China are also funding advanced reactor demonstrations.

Why don't most people trying to advance the development of nuclear energy care about effectiveness?

For the most part, those involved in the industry do. I think it is easy to misinterpret the voice of the industry with the din of nuclear bros turned keyboard warriors complaining how their one true hero was unfairly brought low by overly-risk averse regulation and unscrupulous environmentalists.

The reality is that advanced nuclear receives broad public support, and has for some time now . Safety regulations are becoming streamlined. Regardless, the nuclear-island and safety are a small component of overall plant costs, typically <20%, with most of the cost in the non-nuclear components of the cooling system necessary for any large rankine-style generator. See the EIRP report and the CATF report on advanced nuclear costs. The most optimistic projections are that new advanced reactors will costs $40-60/MWh in a decade or more. Compare that to current bulk power purchases at $15-30/MWh for wind and solar+batteries which are on 20-40% learning curves.

New reactors will likely never be cheap enough to compete directly with the combination of renewables, short-duration storage, and transmission on cost except in regions with particular climatic or geographic constraints. So that means they must compete in the firm generation market (supplemental power at peak times when renewables aren't producing) against a numerous array of other technologies, including efficiency, many with much lower projected cost floors. TerraPraxis is trying to get around the cost barrier by repurposing old coal plants that already have the non-nuclear island components built. They are relying on price drops of SMRs to make this feasible and are probably at least 5-15 years out from an initial demonstration project.

And why are they so rare?

Most of the people who are primarily concerned with decarbonizing the grid have looked at the economics of nuclear power and decided to pursue other grid technologies that they think are more promising. Those that have stayed with nuclear are particularly attached to it because they've spent their careers in the industry and don't want to see it die, think it is an important option to pursue for the sake of having options, or have some dogmatic reason why they prefer it as a technology.

Thinking more broadly, of all sectors that need to decarbonize, the electric power sector is the one that has been the fastest to decarbonize and has the most optimistic outlook. At the margin the most important innovation is happening in agriculture, industry, transportation, and carbon capture. Strategic thinkers see that and move to work on problems in those sectors.

Comment by MatthewDahlhausen on Aviation-Oriented Environmental Organization Looking for Input · 2022-03-21T22:19:15.353Z · EA · GW

First a general word of caution: There is an enormous amount of private and public sector investment in sustainable aviation. You are unlikely to find excellent, underfunded opportunities in a market-driven industry with billions in annual venture capital and R&D spending. Furthermore, focusing on emissions from just aviation is analogous to an interest in global health but only focusing on interventions to address micronutrient deficiencies. It may be worthwhile to investigate, and you may find some giving opportunities, but don't be surprised if the best emissions mitigation options in the aviation sector are much less cost-effective than those in other sectors.

That said, if you do find excellent philanthropic opportunities in the aviation sector, they could be listed among general mitigation-focused climate recommendations (e.g. Giving Green). It does make sense to have a sustainable aviation expert investigating giving opportunities for EA-aligned climate funding organizations. The EA climate orgs do not have the bandwidth or technical expertise to holistically investigate giving opportunities across all emissions-intensive industries.

Realistically, I think it will take a few years to get up to speed on the sustainable aviation industry and develop enough technical depth to make funding recommendations. Recommendations will likely be in the form of lobbying governments to fund specific subprograms at federal research agencies. If you have key industry experts in your group, this timeline can be compressed. If you don't have a full-time connection to R&D in the aviation industry, it's likely the industry will move faster than your capacity to follow along and you won't be able to come up with good funding opportunities. Also without industry expertise, you may get captured by an institution. Institutional capture happens quite a bit in philanthropy. It's where a funder relies too much on the expertise of a grantee and subsequently conflates the grantee's industry knowledge and perspective with a genuine belief that the particular grantee is doing the most important work in the space.

Where you'll find expertise:

  • National Academies research reports on aviation (start here)
  • ARPA-E technology managers on aviation
  • NASA's Aeronautics Research Mission Directorate
  • DOE labs like NREL or ANL
  • Technology experts at sustainable aviation start-ups
  • Aviation technology experts at VC firms

Lastly, on your criteria. You say you are open to all technologies and innovations, and short-listed an MIT lab working on new propulsion technologies. But you also list "target commercial date" as a key criteria in your selection process. Basic R&D and bench prototypes often take decades to go from lab to commercialization, if they ever make it there (most don't). A desire for immediate emissions reductions and a focus on R&D are conflicting considerations. If immediacy is your goal, it will likely be focused on displacing emissions from short-haul and regional flights with electric aircraft / eVTOLs, though I think you won't find any philanthropic opportunities here. If R&D is your goal, consider the much broader research areas that could influence aviation such as lightweight materials, batteries, renewable fuels, and even research on competing transportation modes.

If I had to pick interventions in the aviation space that are likely underfunded, I'd focus on atmospheric research to better predict and understand the impact of contrails on radiative forcing, and real-time aircraft re-routing / altitude adjustments to reduce contrail formation (see: Our World in Data post on aviation emissions, research on contrail formation). One could image an extension of air-traffic control that uses contrail formation risk to direct planes much as they do with real-time weather data for flight risks and to avoid turbulence.

Comment by MatthewDahlhausen on When did EA miss a great opportunity to do good? · 2022-03-09T17:17:31.764Z · EA · GW

I'm continually haunted by this wired article on the early WHO and CDC guidance on COVID:

CDC and WHO downplayed the importance of masks and airborne transmission in the first few months of the pandemic despite evidence of airborne spread and promising data from other countries showing masks reduced transmission.
It took them over a year to officially acknowledge and recommend serious countermeasures to aerosol transmission, by which point we already had the vaccine.

There are a few ways EA could have helped 1) funding work 5-10 years ago to help public health organizations more quickly and accurately identify high risk viruses and disease vectors, 2) making sure there were a range of different professionals including aerosol scientists, engineers, and primary care providers on the key committees at public health organizations that gave official guidance, and 3) supporting a precautionary public health messaging blitz at the first evidence of aerosol transfer, before the strictest lockdown procedures were lifted.

Open Philanthropy is the primary (only?) major funder of biosecurity preparedness in the EA community, and gave ~$65 million to the cause area pre-pandemic, $20 million of which was to Johns Hopkins Center for Health Security (CHS) in September 2019 which in hindsight was very apt timing.

Comment by MatthewDahlhausen on Cost Effectiveness of Climate Change Interventions · 2022-02-16T18:27:16.218Z · EA · GW

As you noted with McKinsey's GHG abatement curve, there are many interventions that have a negative abatement cost (though some many nitpick and say we need to account for the opportunity cost for the time it takes to change a lightbulb).  Marginal mitigation at present is something we can influence, with the best interventions being <$1/ton.

$30/ton seems like the right order of magnitude if we integrate over all abatement costs to get to net zero emissions.  Emission reductions are cheap now but will become much more expensive later; that suggests we may want to use the average abatement cost in our calculations assuming we are committed to solving the problem in its entirety.  That leaves a high cost per life saved.

I think there is much more we can do on the abatement side.  I remember the discussion when the WHO report came out and also read the recent DICE report.  These reports assume deaths with no adaptation.  Adaptation is something we can influence.  Most of the deaths in the WHO report come from issues of poverty: undernutrition, malaria, dengue, diarrhoeal disease.  Heat is also listed, but since the mortality rate from cold temperatures is much greater than from hot temperatures, 1-3C of warming may reduce overall temperature-related deaths (  Eradicating malaria in the next several decades removes ~1/4 of the mortality risk in the WHO report.  Eradicating extreme poverty gets rid of almost all the mortality risk through better nutrition, medical services, and cooling equipment.  These are things we can address and are cheaper than mitigation on a per-life-saved basis.  This strongly suggests that if your primary concern is human impacts of climate change, it's best to spend your money now on global health and anti-poverty development.

Comment by MatthewDahlhausen on Some thoughts on vegetarianism and veganism · 2022-02-16T17:15:41.910Z · EA · GW

It seems like many of the disagreements in the comments comes from disagreements about the moral status of animals.

In an expected value calculation, it is fine to offset greenhouse gas emissions with true counterfactual offsets because there isn't a clear case of one moral actor causing harm to a moral patient, and the harm can be completely compensated for.  But most people would object to the idea of killing someone, selling their organs on the black market, and then donating the money to prevent other people from dying from easily preventable diseases.  Here there is a clear case of a specific moral actor causing specific, serious harm to a moral patient.  The harm can't be undone.  While the first greenhouse gas example doesn't have a clear justice angle, the second organ harvesting one does.

Vegans tend to see a strong justice case involved when eating animals - there is clearly a specific moral patient being harmed (eaten).  Non-vegans may grant that animals can suffer, but don't fully elevate them to a moral status where justice concerns matter.  If there isn't a justice angle, then expected value logic applies and the desire to reduce suffering can be fluidly traded off against perceived inconvenience.  The moral severity of taking a life doesn't register.

The exclusion of justice considerations is speciesist (, and we should be cautious about denying full moral consideration simply based on species membership.  If we are broadly worried about reducing suffering and moral circle expansion, keeping justice concerns for animals at the forefront of conversations about eating meat is important.

Comment by MatthewDahlhausen on The Culture of Fear in Effective Altruism Is Much Worse than Commonly Recognized · 2022-02-07T19:10:07.355Z · EA · GW

I'm not sure how common it is for people to experience hostility to criticism within the EA movement.  I think apathy is a much more common response, one that I've experienced several times, and is no less damaging.  If you don't share the perspective as the EA organization you are criticizing, then you have an enormous burden of proof to get them to even pay attention.  Whereas someone with the same perspective can write a few sentences including words like "epistemic status confident" or "likely" and it will be absorbed with enthusiasm.  It's easy to give up; how many times would you try walking into a church and persuading the priest to convert to another religion?

EA organizations aren't unique in this flaw; there is a general human propensity to engage in motivated reasoning.  It is elevated in the EA community because everyone is trying to work on "the most important thing".  That is identity forming, so criticisms strike personal - questioning value systems and capacity to reason.

Being able to openly, honestly, and enthusiastically accept criticism is a rare skill.  It requires being as dedicated to being unsure as being right or "less wrong".  But being unsure doesn't inspire action, so those ending up in leadership roles tend to be more obstinate than their waffling counterparts.  I think organizations could correct for this by employing a jester or red team with equal status and decision making power as organizational leadership, but I've never seen this done well.

Comment by MatthewDahlhausen on Climate Recommendations in EA: Giving Green and Founders Pledge · 2022-01-19T20:22:49.137Z · EA · GW

I don't know how to contextualize your statement that 45Q is worth supporting "even at risk of marginal emissions increases in some edge cases" given your own report.  CATF's report "Carbon Capture & Storage in The United States Power Sector" which only considers coal and gas power production CCS was a primary input to your cost-effectiveness model of CCS globally (page 117 and footnote 35 of your Nov 21 report "A guide to the changing landscape of high-impact climate philanthropy").  I don't see how this is either a "marginal emissions increase" or "edge case" when it was the primary justification for CATF to support 45Q at the $85/ton level.  Have you significantly changed your cost-effectiveness model for CCS since you released your report in November?

We can have some nuance about 45Q here.  Why not just say to CATF "lobby for only industrial CCS, keep the level at $30-$40/ton, and don't lobby to remove the 75% capture requirement"?

Comment by MatthewDahlhausen on Climate Recommendations in EA: Giving Green and Founders Pledge · 2022-01-19T19:43:17.146Z · EA · GW

Quick calculation here:
emission rates per kWh:
(1 kg CO2 per kWh) * (0.8 capture rate) * ($85/1000 kg sequestered) = $0.068 / kWh.  That's a big enough generation subsidy that it could keep some coal plants operational.  And there are plants that are explicitly using this reasoning to justify to regulators that they should stay open.

CATF could have lobbied to have the tax credit apply only to industrial facilities, or that the credit amount should be lower - more like $30-40/ton.  But CATF targets $85/ton partly because they think that is the threshold necessary to make CCS projects for fossil fuel electricity production viable:
Founder's Pledge assumptions are primarily based on coal CCS and its impact on retrofitting locked-in coal power plants globally.

The risk exists for other sectors too - that the tax credit could be used to keep fossil incumbents operating longer with lower but far from zero emissions.  I do think it makes sense to have some carbon sequestration tax credit for certain industries.  If I were drafting 45Q, I would be careful to balance the sunset clause on the tax credit to be long enough to encourage projects in key industries, but short enough to remove it if it is crowding out true zero-emission alternatives on an industry-by-industry basis.  I also think we should be incredibly cautious and risk averse doing lobbying work in this space because of the potential of lock-in effects and white-knighting the fossil fuel industry.  Unlikely other areas in the climate space, there is a clear and significant risk of doing damage here.

Also, I think "R&D" is thrown around too loosely.  The conventional, common definition of R&D that DOE and academia uses refers to technologies on a 1-9 technology readiness level scale  Coal CCS is a mature technology, not on that scale, and there is an operating plant.  It's just enormously expensive and still has emissions.  I think what you and others are referring to when using the phrase R&D in this context is driving cost reductions through scaling and learning curves, not new technology (R&D).  This is generally referred to as deployment, not R&D.  Please do correct me if you think I'm misunderstanding and misrepresenting your terminology here.  In the context of scaling and learning curves, one does not do this with "the hope that this becomes a viable technology down the road when we really need it".  Learning curves and scaling require consistent and significant growth of a sector; you can't just do a bit of deployment, shelve the industry, and then come back and expect it to be at the same price point.  If we want to deploy CCS to be at scale and low cost, that requires a massive developing and growing industry.  In that context lobbying for 45Q one way or another is risky and path dependent, and I agree with you that the distribution has really wide bounds encompassing zero.  For coal CCS specifically, I think that distribution is nearly entirely negative and there are clear downsides (at least 20% of continued coal emissions) with small potential for an upside.  Why not just focus on industrial CCS only?  That removes a significant know risk and learning probably happens across a wider array of industries anyways.

Regardless of whether you think CCS is important or not, I don't see a case for why EAs should fund CATF doing advocacy in this space.  CCS is an existential lobbying priority of the fossil fuel industry which massively outspends most other industries and environmental groups.  45Q got extended and increased as pork for Manchin and republican support on BBB and the infrastructure bill.  Why is CATF necessary here?

Comment by MatthewDahlhausen on Climate Recommendations in EA: Giving Green and Founders Pledge · 2022-01-19T07:14:03.550Z · EA · GW

Thank you for laying out your thinking on how Giving Green (GG) and Founders Pledge (FP) think about the problem differently.

I've got several comments on the strategy, but I wanted to highlight one major instance where I think CATF is likely to cause more emissions than would otherwise be emitted. CATF is a top recommendation by Giving Green and Founders Pledge, and is the top recipient of Founders Pledge climate fund grants.

It concerns CATF's support for the 45Q tax credit as mentioned in the post:
"CATF – Using a conservative and backward-looking back-of-the-envelope calculation, Founders Pledge estimates that CATF’s work on passing the 45Q tax credits for carbon sequestration reduces CO2e at the cost of $0.11 per metric ton in expectation. Founders Pledge estimates that CATF’s work in the US removes CO2e at roughly $0.30 per metric ton in expectation. (Giving Green is currently undertaking our own CEA for CATF.)"

Some factual background:
- The 45Q tax credit provides a financial incentive for carbon capture and sequestration.  It is described by the congressional research service here:  Extension and expansion is part of the climate provisions in the Build Back Better bill that is currently stalled in congress.

- CATF is supportive of the use of 45Q for the power sector.  John Thompson is their team lead on this, and they've laid out their reasoning for why they support it in this report:

- Founders Pledge agrees with the form and structure of this analysis, but does assume 50% less savings to be conservative: "Note that we are not taking this study at face value, we are assigning a 50% probability that 45Q has zero effect." (Founder's Pledge recent report,, footnote 35 page 117).  They do not entertain the possibility that 45Q could increase emissions.

- Last year 45Q lobbying support was a substantial part of CATF's lobbying agenda:  They were alone (as far as I can tell) among environmental groups lobbying to increase and extend the 45Q tax credit, joining a mix of coal, fossil fuel, and utility companies.  45Q was the primary lobbying objective of coal companies last cycle.

- CATF's lobbying efforts have been both to increase the size of the tax credit and to remove the requirement that a 45Q recipient must capture at least 75% of its carbon emissions to qualify  CATF advocates for increasing the credit from ~$30/ton to ~$85/ton

- Coal CCS projects do not capture 100% of their carbon emissions.  The CATF report assumes 80% capture.  As noted above, the industry is seeking to qualify projects that capture any carbon, regardless of the percentage of emissions it represents.

- Captured CO2 is primarily sold for enhanced oil recovery.  To account for this increase in fossil production and monetary benefit to the CO2 producer, the 45Q gives a lower tax credit for EOR projects versus saline storage projects.

- The only currently operating coal CCS project has had equipment issues and is only capturing 44% of its targeted 90% capture rate

- The U.S. does not have a successful coal CCS plant operating, but that is not without trying.  A bit over a decade ago, as part of the ARRA funding, the U.S. government gave $684 million to eight coal projects to build out coal CCS.  None of them succeeded, and the financial waste was so bad that the government accountability office wrote a report about it.  John Thompson of CATF has responded to the GAO report defending the funding, saying that the ~$400 million that went to industrial projects was successful and that the 45Q tax credit should be increased to $85/ton to make coal CCS viable

Now the analysis:
If the competition for coal CCS were coal plants without CCS, then a tax credit would make sense to offset the cost of capture.  But this is not the case.  The primary competition for coal CCS is cheaper efficiency, wind, and solar.  They aren't just cheaper, they are much cheaper: new wind and solar costs are cheaper than the operating costs of existing coal plants:  This, along with cheap gas, has been the main driver of coal plant retirements in the U.S..

In this context, the 45Q tax credit, if large enough, is a financial lifeline to coal plants that keeps them from retiring sooner.  This is a key point made by Jeremy Fisher of the Sierra Club and others  In this context, the appropriate comparison is between wind or solar at zero emissions and a coal plant at 20% emissions (comparable to gas), assuming it runs perfectly.
This was evidenced in a front-page article in the Washington Post today (18 Jan 2022) regarding the Coal Creek Power plant in North Dakota.  The plant was going to retire and be replaced with wind, but the Republican government worked hard to keep the coal plant open.  The new developer plans to keep it running, calling "the several billion dollars in tax credits the federal government offers for carbon sequestration “essential” for making that happen.", despite having no experience with coal CCS, and no successful projects evidenced in the U.S.
I cannot emphasize enough how mindbogglingly stupid it is to keep a coal plant running instead of replacing it with wind in a region that has among the best wind quality resources in the world:

The CATF study supporting coal and gas CCS suggests that most projects are likely to occur in Texas, Oklahoma, Kansas, Missouri, and Arkansas, as these are among the only places that have a suitable geologic reservoir to sequester CO2.  Again, this part of the country has among the best wind resources in the world:  The CATF study mentions that "carbon-controlled fossil generation does not displace generation from renewable sources," meaning a 45Q tax credit doesn't displace renewables, primarily because renewables are cheaper even with the tax credit support for coal.  They do not consider it the other way around - that without the tax credit,  renewables (namely wind) would completely obviate the need for any coal CCS by being cheaper.  This is baffling to me, and highlights that one of Founders Pledge's "conservative" assumptions is not conservative at all; a "conservative" assumption would be that 45Q prevents cheaper, zero-emission wind and solar generation from coming online and that support of 45Q in the U.S. is in fact negative emissions reductions per ton.

Founder's Pledge then goes on to estimate that 45Q will kick off a learning curve in coal CCS of ~10% cost reduction per year that can be applied globally.

A few things wrong with this:
First, learning curves do not apply to every technology. Solar PV is small and scalable, coal CCS is a mega infrastructure project.  Solar PV realizes its learning curves by having many dozens of plants from dozens of different manufacturers distributed globally cranking out hundreds of gigawatts per year.  Coal CCS has a few specialized mega engineering firms with many unique considerations per project doing 1 GW a decade so far.  Mega infrastructure projects do not have as good of learning curves as smaller, scalable tech.  Some technologies, like nuclear, have negative learning curves.  So Founder's Pledge again falsely assumes that their learning curve estimate is conservative: "Even if one assumes a modest learning rate of 10% per doubling of capacity – for reference, solar is approximately 30%, so this is a fairly conservative estimate" (FP report, p.117).  Relatedly, it's not clear that learning rates for large infrastructure projects in the U.S. are transferable to other parts of the globe, again because of the mega infrastructure nature of coal CCS and unique project characteristics depending on the geologic sink is paired with.

Secondly, and more importantly, the same market competition with renewables applies.  Coal CCS globally will need substantial government support to retrofit and operate coal plants at a price comparable to just replacing them with new renewables.  Founder's Pledge says that coal CCS may be critical in retrofitting coal plants that are "locked in", but again new renewables are cheaper.  RMI did a big report on that this year  I broadly agree with RMI's arguments here, and I think their forecasting accuracy is better than most in the energy space.

Founder's Pledge main goal here is to fund technologies in case the main ones don't work out.  But in the case of coal CCS, two things must happen for that reality to pan out: 
1) governments are willing to spend vast sums of money to keep their coal industries alive by preventing renewables from replacing coal and 2) in that scenario they still care enough about climate that the are willing to pay even more to reduce emissions of the coal plants by like ~80% at best.  I don't think that is a plausible reality, and it also still entails substantial emissions.

Third, renewables are being installed at hundreds of GW annually and are quickly stranding coal and even gas plants globally.  Coal CCS, even if governments want to save their few thousand coal miner jobs at the cost of millions of dollars per job, are not going to be able to build coal CCS at a rate even within an order of magnitude of the present rate renewables are being installed.  Coal CCS projects are massive and take years to build.  I'd bet we will have a mostly decarbonized the power sector by the time we see even 20 GW of coal CCS built or retrofitted globally, if we ever do. (20 GW is ~1% of global coal capacity).

In summary, I think CATF's report on carbon reductions from coal CCS is irredeemably flawed.  Given that, CATF's support for expanding the 45Q tax credit and reducing it's capture requirements is unjustified and will likely increase emissions (see: North Dakota example).  Regardless the support is non-additional given the amount of lobbying dollars that are being spent to support it anyways by the fossil fuel industry and utilities.  If you find yourself the lone environmental group lobbying alongside Peabody, Arch coal, and FirstEnergy and against the Sierra Club (who has been THE leader in coal plant retirements with their Beyond Coal campaign), that should send a red flag that maybe you should reconsider your analysis.  Founder's Pledge compounds this error by making "conservative" assumptions that aren't actually conservative at all because they fail to red-team the structure of the analysis and entertain reasonable counterarguments.  And then they scale up the assumptions globally, assuming that what is learned doing mega infrastructure coal CCS in the U.S. is directly transferable anywhere. All while disregarding the reality that renewables are driving retirement of coal plants in the U.S., and out-competing coal for new generation globally.  Founder's Pledge would benefit by hiring someone with a different perspective to red-team their reports and catch these sort of mistakes.

My general advice from this is that if you do give to CATF, you should target your donation to just their Super Pollutants program.  Their coal CCS program is likely damaging.  Their other programs are likely neutral, but I'll save that comment/post for another day.

[EDIT: enabled links and fixed some typos]

Comment by MatthewDahlhausen on Democratising Risk - or how EA deals with critics · 2021-12-28T21:27:49.790Z · EA · GW

The IPCC reports have hundreds of authors from all over the world:  It is misleading to say the IPCC is homogeneous and that the authors are "disproportionately white very well-educated males in the West".  Every country and a large variety of civil institutions are represented at the conference of parties, and they use a consensus process.

Comment by MatthewDahlhausen on How Do We Make Nuclear Energy Tractable? · 2021-11-11T19:13:24.493Z · EA · GW

Nuclear energy is expensive. This is true of all rankine cycle style generators, which use heat to generate steam to turn a turbine.  In most places, the cost of new renewable energy is cheaper than the ongoing operation and maintenance cost of rankine cycle generators.  Nuclear is particularly expensive among rankine style generators, as it requires lots of capital and long build times, and is the only major generating technology with a negative learning curve over the past several decades.  The cost issue is the main problem with nuclear power.

While rankine cycle generators can't compete on cost with renewable energy now, there is a need for excess renewable energy, storage, or other firm capacity at high levels of variable renewable energy (80%+) on the grid.  This is decades away.  Ideally, you'd want a technology with a low capital cost and high ramp rate.  The theory supported by nuclear advocates is that smaller modular nuclear reactors (SMRs) may be cheaper to build than existing light water reactors and can provide that firm capacity.  Nuclear however, is just one of many competitors to provide that grid service.  It has to compete with hydropower, high capacity factor offshore wind, various forms of storage, targeted building efficiency, demand response, expanded transmission, and even overbuilt renewable energy with curtailment as costs get cheap enough.  Nuclear isn't the ideal solution to firm capacity, as it is expensive and has low ramp rates.  Optimistic cost projections for SMRs are still multiple times that of renewables.  Nuclear plants want to run at high capacity factors (85%+) to offset their cost, and take many hours to ramp up or down.  Some have suggested pairing electrolyzers onsite with nuclear to convert the constant power output into a one with more variable energy and higher ramp rate.  But it's almost certain that renewable energy paired with electrolyzers or other renewable gas will be cheaper and do the same thing.  Here is an example case study comparing generating technology mixes for a 100% renewable U.S. power grid:  In all decarbonization scenarios, VRE accounts for 70%+ of capacity and 90% generation.  Nuclear and coal phase out as the highest-cost generating options.

Compounding the cost problem is that the nuclear industry has had significant issues of corruption, fraud, and other criminal activities which undermine safety and increase costs.  Some of these issues are nuclear-specific, but most are related to having vertically-integrated or state run utilities building large generators instead of cheaper alternatives and having to have some mechanism of public debt repayment to make up the cost.  The safety issues that arise from this are unique to nuclear power, and justifiably worry anti-nuclear activists.

Nuclear, even some SMRs, will likely be part of the grid in 30 years, if for no other reason than because the utility and nuclear lobbies are quite politically powerful and engage in both legal and illegal bribery.  Even coal CCS is getting substantial government R&D investment, which will never be economically viable.  But SMRs will never play more than a supporting role for the grid, at best reducing capacity costs ~20% for a 100% RE scenario several decades from now.  Beyond keeping existing nuclear plants open that are still in good operating condition, there isn't much of a role for nuclear in the next several decades.  The most important thing to focus on in the next several decades is building lots more wind and solar, electrifying heating and transport, and building efficiency.  Everything else is secondary.

Comment by MatthewDahlhausen on This Can't Go On · 2021-08-24T18:39:43.777Z · EA · GW

A bit late to the discussion here, but I wanted to re-emphasize the slightly nearer term limits that Holden brought up by Tom Murphy in his "Do The Math" blog, first presented a decade ago:

This is formalized in Chapters 1&2 in a textbook he just released:

The argument goes 1) there is a finite amount of energy available, limited by solar capacity, 2) decoupling of energy from the economy is ultimately limited; we can only grow non-energy intense parts of the economy so much as a percentage of the overall economy, 3) space colonization is extremely difficult if not impossible given how inhospitable space is, and therefore 4) economic growth will eventually cease given limited energy and resources.

He also mentions thermodynamic limits to surface heat rejection via blackbody radiation.  We may be able to get around that by moving most computation to space.  Though I imagine that being a harder collective action problem to solve than climate change presently.

Galactic expansion through digital minds may be a possibility that gets around the human biological limits to space expansions, but it's unclear that the timing will line up conveniently.  Our rate of economic expansion may hit hard energy and resources limits before a galactic digital mind expansion, regardless of whether it is possible.

Ultimately I don't see how it's possible to get around the energy limitations.  As long as there are energy requirements for people - biological or otherwise - then it doesn't seem possible to indefinitely shrink the energy intensity of the economy to an arbitrarily small value.

Do we really imagine the future of the economy to be increasingly faster trading of multi-trillion dollar cat holograms?  Is that humanity's destiny?

Comment by MatthewDahlhausen on Interactive Graph of Climate Change Intervention Effects + Reflections · 2020-05-17T20:35:56.486Z · EA · GW

En-Roads is a systems dynamics model similar in purpose though much more complex and well-referenced than the World3 model of Limits to Growth fame.

Reference Guide:

is probably roughly accurate for the sector models of buildings and industry, growth, land and industry emissions, carbon removal, and non-electric transport. The electrification and energy supply sector model is not reliable, as the model simulates on a 0.125 year timescale which misses the market dynamics of electricity production. Furthermore, it calibrates the electric market projections against the IEA world energy outlook (WEO) which has continually underpredicted renewables and overpredicted coal capacity to an embarrassingly large degree: For example, solar capacity additions in 2016 were ~50 GW, WEO 2016 estimated solar additions would be roughly constant at 50 GW/year in 2018, reality was over 110 GW. WEO 2019 still assumes there will be substantial buildouts of coal plants in the U.S. and EU in the "current policies" scenario. The GIGO (garbage-in, garbage-out) applies here for the En-Roads electricity sector model - they calibrate against garbage energy supply projections from IEA, so the En-Roads analysis of electric sector policies will be grossly inaccurate.

This model error could significantly change your inference on points 3 and 4 once corrected. The rest of your points of inference are probably accurate.

Comment by MatthewDahlhausen on [Link] Updated Drawdown now available, incl. 2020 Review · 2020-03-29T18:13:28.390Z · EA · GW

I want to make one more point, separated from the disinterested economic and technological arguments above. Understand that lobbying for nuclear research - as you and Let's Fund have promoted - is not politically neutral and does not occur in a vacuum. The U.S. through the DOE and NSF is by far the largest R&D supporter of nuclear and energy research in general, worldwide. The current administration is heavily focusing on financial support and R&D for coal and nuclear at the expense of other energy technologies and energy efficiency. The are using dishonest tactics and misleading information, including some of the points you mentioned above, as talking points to promote these interests. As an example, read the DOE grid reliability report that concluded grids would be fine without coal and nuclear, but had its abstract edited by former DOE head Ricky Perry to claim the opposite result. Two years ago, the administration tried to pass an emergency declaration for subsidies for power production with on-site fuel storage (coal and nuclear). And the most recent main thrust of R&D at DOE is for coal CCS and nuclear. This is at the expense of many other areas of R&D. The current administration has tried multiple times to cut ARPA-E, a highly successful technology development program across the whole energy sector, and the office of energy efficiency and renewable energy (EERE). EERE provides technology R&D and market commercialization for energy efficiency (the cheapest form of energy - "negawatts") and was the source of funding that helped make solar and wind as cheap as they are today. To suggest precious EA dollars go to supporting nuclear R&D in the U.S. is ignorant of the political reality and potentially damaging to other energy and energy efficiency efforts, insofar as it shifts funds instead of increases the pie. I understand this is surely not the intent of Founders Pledge and Let's Fund recommendations. But ignoring this reality is not an excuse for what is, in my opinion, a bad philanthropic recommendation.

Comment by MatthewDahlhausen on [Link] Updated Drawdown now available, incl. 2020 Review · 2020-03-29T18:12:40.492Z · EA · GW

Responses to your points above:

1. IPCC Integrated Assessment models don't dictate technologies. By design, they assume many different future scenarios and calculate impacts from those scenarios. Some scenarios use ample amounts of BECCS to achieve negative emissions to hit a 2C target by trading off more short term emissions with expensive negative emissions in the future. This isn't a determination of what is needed, just an example of a technology scenario that hits an emissions target. Massive amounts of BECCS would be extremely expensive; IAMs don't factor in these economic factors. However BECCS may be needed to get negative emissions. Nuclear can't do that, and will need to compete economically for energy production. If you think nuclear is absolutely necessary, please send me the particular IAM that states that and the economic assumptions compared to other electric grid build-outs.

When I speak of the academic community here, I'm referring to the community doing resource planning and grid modeling - the people that are making the decisions about what grid resources, transmission, and R&D to pursue. In this community, nuclear is not recognized as a substantial contributor to short term or long term decarbonization.

2. Break out the capital cost figures. Licensing and regulation isn't the major difference. That accounts for less than 10% of the cost; The World Nuclear Association (nuclear lobbying group) puts it at 5% of the cost. The $3k figure in China is because labor is much cheaper there, which is also a reality for renewables.

On cost - I'll make the point again - even with heroic capital cost reductions, nuclear won't be competitive in the market. The O&M and fuel costs associated with rankine-cycle based power production cannot compete with VRE. Even the most nuclear pro lobbying groups can't claim nuclear is competitive using recent data (the link above compares to 2012 prices). Nuclear plants are closing now because even with O&M costs, they are more expensive than new VRE.

The same economics apply to coal plants - large rankine based producers with substantial fuel costs.

The slight difference is that coal also has to contended with large capital expenditures on criteria pollutant emissions controls (in most countries), whereas nuclear has larger capital cost requirements for the reactor.


a) On LCOE vs system LCOE. Marginal LCOE is what determines what new generation resources get built. In fact, utilities have an obligation to consumers to pursue lowest cost generating resources as overseen by the public utility commission. No one uses total system levelized cost in planning, and it's not clear how one would even do that. If this was Sim City and you could plan out the eventually 50 year grid from the beginning could you do it? Maybe, but it may not be centralized generating resources. As for high VRE costs, there isn't "betting" on the cost inflection point for VRE- we have very accurate models of the electric grid and build-out concerns. Electric battery storage is already cost-competitive with gas peakers in some grid regions. And as VRE increases, the marginal LCOE will tilt in favor of load shifting, DR, and storage assets instead of gas peakers, gas CCs, and certainly baseload coal and nuclear plants which have to earn revenue in the production and capacity markets to stay viable.

In a high VRE scenario, it's not clear that added nuclear or any baseload generator is the cost-preferred option to extra VRE with curtailment or even existing storage costs. Saying baseload generators are a solution needs supporting evidence, especially including how the market would need to be restructured to keep these plants viable.

b) On the point of political feasibility - eminent domain is an issue regardless of generating source. Gas lines, transmission lines, siting uranium or coal mines all have political pushback. Local opposition is much stronger against nuclear and fossil generating facilities in general.

c) Value deflation is an issue for solar, though not so much wind with a higher capacity factor and 24/7 power production. This is where load shifting and DR technology in buildings is likely to become cost-competitive with new generation. Building codes in California are already account for this using a TDV (time-dependent valuation) metric in design, rewarding energy savings during peak evening hours a lot and daytime savings comparatively little.

d) Germany heavily subsidized solar, providing the market incentive that brought the price down considerably for everyone else. And now subsidies are no longer needed to make wind and solar competitive - in general, they are the cheapest generating source on their own. But to claim the subsidies failed looking only at historical solar build out in Germany alone in comparison to the total German subsidy cost is to ignore the massive price decrease it meant for solar globally. I could make a similar claim for nuclear if I weighed U.S. nuclear program costs vs. the first 5-10 years of nuclear production in the U.S.

e) Sure, historical experience, especially recent historical experience should carry weight. No advanced economy has decarbonized. The fastest rates of decarbonization in absolute terms are from VRE, and nuclear is nowhere close. France, used as the common example, is building out VRE and retiring older nuclear plants. Nuclear prices have increased in every developed economy in the last 2 decades.

4. Exxon, Shell, BP are very interested in zero carbon fuels. NREL has a $100 million research project on next gen VRE to biofuels "electrons to molecules" funded by Exxon. Here cost of energy is incredibly important; I'm not sure why you suggest nuclear here? Electricity to fuels is well-paired with renewables to absorb low cost solar and wind during periods of otherwise curtailment.

District heating is a specific application where cogeneration is preferable, and a potential U.S. of SMRs in a few major cities with central district steam systems (e.g. NYC). I suspect a cogeneration application is where we most likely see an SMR demonstration project. For newer district systems, there are competing technologies of heat-pump based ambient loops or four-pipe chilled water/hot water loops that are much more efficient than conventional steam district systems and have lower operating costs that steam-based systems.

As to your point that 45% of fossil fuel emissions are electricity and heat? I assume you got that from Fig 2 in:

The numbers heat 2% + combined heat + elec 5% + elec 26% + load following elec 12% + res/commercial heat 10% = 55%, transportation is 22%, cement 4%, iron & steel 5%, and other industry 14%.

Note that this graph includes other industrial non-energy related CO2 emissions and other gas emissions. Cement production involves emissions from limestone reforming in kilns, and steel mills from coke production. Other industry involves substantial methane emissions from refineries and oil and gas production, as well as ammonia.

Combined heat and energy - where future nuclear has potential economic competitive viability is <5% of this picture.

I don't agree with the view that an "all of the above" strategy that includes substantial support for nuclear R&D for electric production is the least risky from a climate perspective. Even if the R&D budget increases, these funds could be better spent on storage, integration, liquid fuels from electricity, direct carbon capture, market commercialization of several lab-proven technologies, or support for better building codes (75% of grid load). I see it as similar to "clean" coal CCS - unlikely to ever be viable and a distraction to less-risky efforts.

Comment by MatthewDahlhausen on [Link] Updated Drawdown now available, incl. 2020 Review · 2020-03-10T16:33:04.900Z · EA · GW

There are valid criticisms of Drawdown, but it's lack of consideration of nuclear and research into SMRs is not among them.

*Nuclear energy for electricity production will not make a meaningful contribution to addressing climate change*

I'm happy to go into more technical and numeric detail than the comments below.

1) Nuclear is too expensive (, ( The O&M costs alone (<20% of total cost of a nuclear build) are greater than new onshore wind and utility scale PV. Solar and wind costs have been declining at 5-7% per year, while nuclear costs have stagnated or gone up.

2) SMRs will still involve a rankine-cycle based power production with its massive water and construction requirements. Even if research eventually yields a 10x capital cost reduction in nuclear, by that time it will have to compete with even cheaper wind and solar and new offshore wind and solar perovskites. Simply, advanced nuclear doesn't have a chance at being competitive in the future, even with massive R&D. It's O&M costs will still be above current wind and solar. Most goals involve decarbonizing the electricity sector in developed countries by 2035. It'd be lucky if there were even 1-2 demonstration SMRs online by that time.

3) Variable renewable energy (VRE) is manageable, and grid integration studies from LBNL and NREL show VRE can easily reach 70-80% of production with little additional storage needs. New onshore wind is reaching capacity factors of 40-50%, and offshore wind at 60-70%, creeping into that mythological "baseload" generation. Buildings are becoming much more grid-responsive to control when and how they use power. Energy storage research is getting much more funding and showing similar price declines as early renewables. The challenges are not VRE amount, but rather the power electronics involved with switching from a synchronous-generator based grid to an inverter based grid. Even with current technology, existing energy storage plus overbuild and massive curtailment of VRE will be much cheaper than nuclear in hitting a 100% renewable grid.

The academic debate has moved beyond whether advanced nuclear power will be a relevant solution to addressing climate change. It's now at whether it would be better from a climate perspective to shut down certain LWRs early and spend the money on renewables and efficiency instead, or if a subsidy in the form of a carbon-price would help keep existing LWRs open a little bit longer. This is largely a function of carbon price and utility regulation in mandating that all cost savings from closing nuclear plants go towards efficiency and new VRE.

Instead of seeing nuclear research as promising, we should view it alongside its current R&D partner - CCS for coal plants. Even if it is neglected in total funding compared to other R&D doesn't stop it from being a wasteful money pit. I think there is an argument for SMR R&D for ship propulsion, space propulsion, and extreme security and redundancy military installations, but this is not a climate change consideration.

Comment by MatthewDahlhausen on Clean cookstoves may be competitive with GiveWell-recommended charities · 2020-02-10T19:00:52.723Z · EA · GW

You can track health impacts with HAPIT, the Household Air Pollution Intervention Tool . Use that directly instead of the assuming clean cookstoves eliminate the disease burden, which is far from the truth.

Clean cookstoves have a really high bar to clear to reduce the disease burden for several reasons:
1) Exposure is non-linear. The relative risk of dropping from 400 ug/m3 to 175 ug/m3 PM exposure is the same as going from 100 ug/m3 to 50 ug/m3. To reduce the disease burden to <1.5 requires exposure <50 PM ug/m3, which is very difficult for any biomass stove to accomplish.
2) For the disease burden to be reduced, nearly all stoves in a locality need to be replaced, otherwise the outdoor air will still be above the exposure threshold.
3) Stoves need to be used, used correctly, and maintained. The education and cultural habits to do this are very difficult to embed in a population.

For these reasons, clean cookstoves have historically been largely unsuccessful at reducing the disease burden. They are improving. See the "2019 Climate Action and Clean Cooking Co-benefits workshop presentations and discussions" presentation by the Clean Cooking Alliance. This group is setting standards and tiers for clean cookstoves to rank them on performance and targeting intervention locations based on many factors to determine where cookstoves will be most successful.

Even with optimistic assumptions of stove performance and uptake from HAPIT, it is likely that cookstoves will remain at least an order of magnitude more expensive than the best GiveWell recommended interventions for some time.

[Edit: fixed some spelling errors]

Comment by MatthewDahlhausen on Crowdfunding for Effective Climate Policy · 2019-07-12T19:58:04.609Z · EA · GW

I distinguish between R&D and economic/policy factors, because it matters where technology is at in the R&D pipeline. Solar and wind are mature technologies. There is some additional work that can be done in solar (e.g. perovskites) and wind (bigger blades, offshore), but the vast majority of the costs at this point are not associated with the technology itself, but rather the implementation, permitting, financing, etc. At this stage in technology development, costs get driven down by expanding the market, not so much additional early stage R&D. There can be more investment in >6 hr energy storage and zero-carbon liquid fuels, as many of the solutions are in early stage research.

Therefore, I'm more inclined to say that clean energy deployment in developing countries is economic and policy limited, not technology limited, given relatively low deployment rates and maturity of the most applicable technologies. I still agree with more R&D, but I don't think that is limiting factor in a lot of countries right now. Major climate philanthropy seems to agree - focusing on policy around development, energy efficiency, and deployment of exist tech, rather than early stage R&D funding. Perhaps they don't because the government already funds R&D at level greater than the philanthropic sector could ever meet. But if as you say, the R&D is more funding limited and has the better marginal return, then most of the philanthropic giving should be going to that.

I understand the desire to not dive into the specifics which technologies to focus on and in general just get more clean energy R&D funding. More clean energy R&D funding lifts all technologies. An analogy would be to global health. It would be good to get more general funding into global health, and most academics and EAs support that. But I think the EA angle could benefit from being more specific on which kinds of interventions/technologies, as like global health, the effectiveness of additional funds could vary greatly depending on where they are spent (e.g. energy storage vs. clean coal). This is a increase funding or use existing funding more effectively question. Your argument is that Clean Energy R&D funding is so low that it is much more important to increase the funding. I agree with you on this. I have a mix of thoughts on whether ITIF's specific lobbying priorities within Clean R&D are correct, but don't want to get into that too much.

I do want to address the points on adaptation and extreme warming.
Adaptation gets funded through the UNFCCC framework is the fund has given out $1 billion, with support of $4 billion from other sources. This is total, not per year.
I think R&D in adaptation is underfunded, and adaptation in one area is likely to be replicable in other places (making it a global public good, similar to clean energy R&D). This is about limiting the worst effects, and is neglected in the same way tropical diseases are neglected on the global scale.
An analogy on this is a expansion of pond metaphor. The water level is representing greenhouse gas levels in the atmosphere, and the pond is filled with adults (developed countries) and small adults or kids (developing countries). Some are already struggling. Mitigation will slow or stop the rise of water, but we should also spend some effort helping the smallest humans out before they can't touch the bottom (e.g. life preservers, rocks to stand on). We can presumably get better at figuring out ways to do mitigation (Clean Energy R&D) and adaptation (helping people in the pond to not drown). Right now, I think helping the smaller people is more neglected than lessening the rise of water level.

Lastly to the comments on extreme warming scenarios. While there is a lot of research improving climate models and projections (largely computation limited), there is still a lot to be done on translating those extreme scenarios to impacts, and also geoengineering responses to lessen those impacts. This needs funding on the ~$1 billion scale and is vastly underfunded (funding got cancelled in the US given the current administration). I'm more inclined to think that this or adaptation are likely to yield better returns from an EA perspective.

Comment by MatthewDahlhausen on EA Survey 2018 Series: Do EA Survey Takers Keep Their GWWC Pledge? · 2019-07-01T23:53:07.618Z · EA · GW

I just logged on to my GWWC pledge dashboard and noticed I was under 10%, even though I've been giving 10-20% the past 4 years. It seems that my reported incomes were each included twice, possibly leftover from the site migration.

I'm wondering if this happened to other people - can you check if there multiple or duplicate entries for the same date ranges?

Comment by MatthewDahlhausen on Crowdfunding for Effective Climate Policy · 2019-07-01T19:59:20.541Z · EA · GW

Thanks for putting this together, and sorry for the delay in posting a response.
It's great to see some more attention to effective climate interventions in the last year, starting with the founders pledge report:
There is a lot of open debate in the EA community on how much to focus on climate change as a cause area:'m going to skip over the questions of how much of a catastrophic risk it is, and what the appropriate split between mitigation and adaptation should be, and instead focus on the point you make of "What can we most effectively do to fight climate change?".

In getting from the broad goal of emissions reductions to specifically support ITIF's policy work, there are several key assumptions:

  • Energy related emissions from fossil fuels are the top priority
  • The priority should be on clean energy development in the most populous countries (India, China)
  • Reductions in energy related emissions are technology limited, rather than economic or policy limited; Adequate technologies do not already exist to in the most populous countries
  • Clean energy R&D can produce low or emissions-free technologies; there are substantial clean energy R&D opportunities with high impact on emissions reductions
  • Clean energy R&D is currently funding constrained, and there are high marginal returns for the next dollar to speed deployment of these technologies
  • The philanthropic sector, private/market sector, and governments have neglected clean energy R&D, or could substantially increase their giving in this area
  • Technologies developed in developed countries like the U.S. will lower costs and ease deployment in the most populous countries with negligible barriers to tech transfer
  • Lobbying to increase government R&D spending is likely to be successful, and specifically by ITIF
  • ITIF's target R&D areas are broadly correct, and will surpass the challenges above

Responses to those assumptions:

  • Energy related emissions from fossil fuels are the top priority.  Mild agreement.  Agricultural, land use change, methane, and F-gas emissions account for ~25% of emissions, and there is substantial uncertainty how much is being emitted.  There could be high impact interventions focusing on these emission sources.  However, mitigation will certainly require eliminating emissions from fossil fuels.
  • The priority should be on clean energy development in the most populous countries (India, China). Agree.
  • Reductions in energy related emissions are technology limited, rather than economic or policy limited; Adequate technologies do not already exist to in the most populous countries.  Mild disagreement here.  I think this is true for the hard to mitigate emissions ( from Cement, Iron, Steel, Aviation, Shipping, and the last ~10% of electric generation when 90% is served by renewables.  These account for ~15% of total energy related emissions. Otherwise, the technologies exist are largely driven by policy.  75% of electricity use is in buildings, and we already have the technology to make them low or zero-carbon.  It's a matter of adopting rigorous building energy code.  Electric vehicles are in the deployment stage, relying on infrastructure build out to support their manufacture and charging availability. Better batteries would help, and this is getting a lot of R&D interest from all sectors.  Lastly, urban planning is largely the biggest lever in reducing developing country emissions, as they can design out the need for high transportation energy use.  However, the ITIF fund is explicitly targeting the ~15% hard to eliminate emissions, so they are only focusing on the technology-limited emissions.
    • Additionally, I think the choice to use the German example of clean energy subsidies is unrepresentative.  In the U.S., which is responsible for 35% of clean energy R&D, the ratio is much closer to 1:1 (post-ARRA).  And while the subsidies in Germany didn't go to R&D, they encouraged learning in manufacturing and production, which greatly drove down the price.  While this isn't included as R&D, it has a similar effect of making renewable energy cheaper.  
  • Clean energy R&D can produce low or emissions-free technologies; there are substantial clean energy R&D opportunities with high impact on emissions reductions.  Agree.  Largely with batteries and liquid fuels.
  • Clean energy R&D is currently funding constrained, and there are high marginal returns for the next dollar to speed deployment of these technologies. Mild agreement.  This is true for zero-carbon liquid fuels.  New battery technologies are not funding constrained, and at a point where additional funding will not lead to faster development.
  • The philanthropic sector, private/market sector, and governments have neglected clean energy R&D, or could substantially increase their giving in this area.  Mild agreement, though I disagree with the characterization that the philanthropic sector has neglected developing countries.  In the philanthropic sector, the Packard and Hewlett foundations are the main funders in this space, and have made major contributions to the Energy and Climate foundations, who in turn have focused their grants on developing country emissions, largely around things the building code, vehicle electrification, and development policy.  If the criticism is that the philanthropic sector is under investment in R&D, it is largely because they think policy priorities in developing countries are a more impactful mitigation strategy.  
  • Technologies developed in developed countries like the U.S. will lower costs and ease deployment in the most populous countries with negligible barriers to tech transfer. Agree.  This has been demonstrated with solar and wind, though I would have appreciated more in the write-up on tech transfer.
  • Lobbying to increase government R&D spending is likely to be successful, and specifically by ITIF.  Neither agree nor disagree.  I have no way to judge how successful ITIF will be with their lobbying.  While they are well regarding, I'm not sure how much political power they have.  
  • ITIF's target R&D areas are broadly correct, and will surpass the challenges above.  Mix of agreement and disagreement.  There are 6 areas:
    • 1) Advanced Nuclear Energy, particularly on SMRs (small modular reactors).  The application here is for the last 10-20% of electricity generation that is hard to cover with renewables.  However, ~80% of the costs of rankine-based technologies for power production are from the capital and maintenance costs associated with the rankine cycle (cooling towers, concrete, etc.). SMRs, even small ones on the 50 MW scale, even if their nuclear component is vastly cheaper than LWRs, are unlikely to be able to compete with renewables and storage on cost.  There are other means of meeting this grid need with demand response or transmission, meaning SMRs will likely only find use in certain applications like shipping.  Therefore, I don't think greater funding in this area is climate-relevant.
    • 2) Long Duration Grid Storage, seasonal storage.  This is also for the last 10-20% of grid use.  I agree this could use more funding, though it is speculative and technology specific. 
    • 3) Carbon-Neutral Fuels.  This is the strongest R&D need. There are some recent big investments in the space, but it could get more attention.
    • 4) Carbon Capture, Utilization, and Storage (CCUS).  CCUS is not competitive in the electric sector.  The application will be for the industrial sector in cement and steel manufacturing, and for low-carbon liquid fuels.  I think the investment needed here is less on the capture technology, and more on the robustness of sequestration and storage to prevent leaks  
    • 5) Carbon Dioxide Removal Technology.  I agree with this as a government research priority as there isn't a market incentive.
    • 6) Basic Energy Research.  Basic energy research requires government investment, and gets a lot of the R&D share.  It's unclear to me what lobbying for this would entail.  The current administration is more inclined to basic energy research at the expense of all other areas of energy and climate, so lobbying for this in the next few years may actually be counterproductive.

Overall, I think ITIF is broadly correct in the need for government-funded R&D in carbon-neutral liquid fuels, CCUS in industry, CO2 removal, and somewhat long-duration storage. I disagree with their Advanced Nuclear and Basic Energy research goals for practical and current political reality reasons.  I don't think ITF's lobbying on their 6 focus areas rise to the level of "most effective climate interventions", as ~60-70% of emissions are policy-limited, not technology-limited.  And I think the U.S. climate philanthropic sector has largely correctly identified policy interventions in India and China as the highest priority.  From an EA angle, I think the most neglected climate-interventions are in adaptation in poor countries that are most vulnerable to climate change, and in understanding and adaptation to extreme warming scenarios.

Comment by MatthewDahlhausen on New research on effective climate charities · 2018-07-14T05:19:20.054Z · EA · GW

1 "Concerns about your gains from preventing deforestation being reversed should be accounted for in your marginal cost-effectiveness estimate." Well yes, if we account for the fact that current best marginal emissions reductions at present might fail in later years into our marginal cost-effectiveness estimate then we can still use the marginal cost-effectiveness estimate. If I did that, it would show rainforest work having mediocre cost-effectiveness because emissions reductions aren't robust. So we reach different conclusions on best interventions, despite claiming to adhere to the same principle. I agree with the position "we should act on the best marginal effectiveness". It's just that rainforest work is not independent of other interventions. Its cost-effectiveness is co-dependent on the cost-effectiveness of other interventions - needing to hit sub 3C century end warming. So my cost-effectiveness estimate cares more about the 80th percentile on the cost abatement curve from future projects, while a simpler analysis may just focus on the cheapest marginal abatement cost intervention at present yearly emissions.

This is heavily related to the concept of lock-in ( Even though some interventions may be more expensive than others, they may represent a substantial enough amount of emissions and lock-in threat that on a longer time horizon they become the best marginal cost-effective interventions at present.

Once we've accounted for lock-in, largest emissions interventions, robustness, etc., THEN we can start moving down our new inclusive-forecast-century-weighted abatement curve of best interventions. I just think this will look different from McKinsey's abatement curve and yield different interventions than the ones you've selected in the report.

2 I agree the aim is to decarbonize, not get as much renewables growth as possible. My statement wasn't cheerleading renewables, it was making the observation that in actual grid capacity purchases and planning at present - the current market - renewables are more bullish than they appear in the reports you reference. IEA World Energy Outlook has abysmal prediction accuracy on renewable install rates (, and the integrated assessment models in AR5 are similarly conservative in their estimates.
This is good news given the amount of emissions that come from the power and buildings sectors.

I'm confused by your comment "Also, this is only electricity not all energy, so other stuff like CCS and nuclear will be necessary to get us all the way to decarbonise" Where does nuclear contribute besides the power sector? Your "emissions averted by different energy technologies" has nuclear's impact only from displacing coal and gas electric power.

4 Yeah, I wish Drawdown was more explicit in their calculations. I found an error in their documentation on plant diets, but couldn't track down if that was just in the documentation on the calculation too. I only reference drawdown because it gives explicit GtCO2e estimates. For instance, it's nuclear estimate is 16 GtCo2e, while yours is 136 GtCO2e. It's wind and solar estimate in total is 171 GtCO2e, while yours is 135 GtCO2e. Not enough to change things on a log scale. Obviously, you have to look elsewhere for philanthropic neglectedness calculations.

On points 3 & 5 - Recent bids for renewables with storage are cost competitive with gas, and cheaper than nuclear, even at the $60/MWh quoted in the report.

The intermittent and non-intermittent power source debate is about a decade old, and doesn't reflect the reality that additional intermittent contributions to the grid have made the grid more reliable, not less, at least in the United States. Energy markets are structured to provide a reliable electric grid - they price capacity and when electricity can be produced - and nuclear isn't competing in this environment. Recently in the U.S. the nuclear lobby has hitched itself to the coal lobby to argue for emergency interference in energy markets by the government to require subsidizing large plants, precisely because they could not compete in the hourly capacity market.

Your comment about Germany doesn't seem applicable; renewables reduced emissions compared to the proper counterfactual where they hadn't been installed AND the nuclear plants were taken offline.

Again, this isn't me cheerleading renewables at the expense of nuclear. It's an observation of the current energy market that no one is even thinking about starting a new nuclear plant build because of how outrageously expensive they are compared to other options. There are 2 reactors under construction in the U.S., 2 recently abandoned, with the primary contractor filing for bankruptcy last year. All vastly overbudget.

Given the state of affairs with nuclear, CATF's large share of funding towards nuclear seems like pissing money away, especially since the government already funds this heavily through the DOE for reasons other than emissions reductions and competitive energy. I think the argument for CATF has the best donation target relies on their CCS work solely.

We can have different perspectives on this, and I share a different outlook, so I propose a bet: If a nuclear plant is built in the US: 1) at least 150 MW in size, 2) in the next 10 years, 3) with construction started 2019 or later, 4) and sells its power in a competitive bid process for an electric grid, I will pay you $100. If not, you pay me $100.

Also - can you respond or publish on why you didn't include an analysis of the other charities in the report that you include but do not recommend? Why were they rejected?

Comment by MatthewDahlhausen on New research on effective climate charities · 2018-07-12T06:37:09.042Z · EA · GW

Thank you John Halstead for putting this report together. Climate change is a significant issue in popular culture, and one of the most widely known catastrophic risks, with relatively little analysis in the EA community. So I'm glad you took it on.

Some general comments on this report:

COMMENT 1 - IMPORTANCE VS. MARGINAL TRACTABILITY REGARDING CLIMATE CHANGE Something I've been puzzling over with climate change specifically is that I think funding at the margin might miss the largest emissions reductions needed to stay under a warming target.

For example, I've contributed to Cool Earth with the goal of rainforest protection. At the margin, this is an amazing bargain for carbon storage - and can be equated to very cheap carbon emissions reductions at $1 or so per tonne CO2e. However, there is a limited supply of rainforest protection we can do, and if other areas of emissions are unaddressed, the Amazon may turn into a savanna anyways at current projections of warming, releasing all that carbon. At higher expected warming, rainforest protection becomes less useful. This was not considered in the report. Relevant recent papers on that:

It's the sort of problem where you may get good returns at the margin for the first 10% of emissions reductions, but you need to hit 80% of the reductions or more to achieve the desired outcome. Another way to say it is that in most EA causes, marginal tractability counts more than overall importance/scale, but for climate change, these concerns are more equal.

I think an analogy to this is how energy is priced in a deregulated market - generators bid in their power at a price, the grid operator buys it, and everyone gets paid the price of the last MWh purchased, the highest price on the marginal cost curve that meets the total load.

I expect the best use of my marginal dollar for climate philanthropy will depend on the current landscape of funding and projections for how quickly we are reducing emissions. On our current track, we are set for 3-4C of warming, so I'm more inclined to put dollars towards adaptation efforts and economic development / global health for the worlds poorest to lessen the damage than I would be if we were farther along in our decarbonization efforts.

COMMENT 2 - RENEWABLE ADOPTION POTENTIAL A recent report out of the Electric Markets and Policy Group at LBNL ( and related work at NREL (disclosure: I'm an NREL employee) have found that we could probably push renewable generation up to 80% of total annual electric production without a significant cost increase. I think the 2016 reference cited in this report saying >50% would be cost prohibitive with storage is outdated. This year, renewables + storage beat out gas generation on cost in both California and Colorado markets. The last 20% of storage will be expensive, and there is ongoing work at the national labs to finds ways to reduce that cost or shift times of energy demand.

I'll need to review the sources used to do the importance calculation for renewables and EE (, but my initial read is that they are already 4-5 years out of date given recent research. (As is to be expected; academic meta-reviews lag a few years behind reviews, which lag a few years behind potential studies).

The electrification of the transportation sector is ignored in this report, but is a necessity towards greater emissions reductions goals, and will greatly increase the importance of renewables.

COMMENT 3 - MARKET ADOPTION AND COST Part of the reason why Nuclear is underfunded in the philanthropic sector is that it cannot compete on cost with renewables. /9 Philanthropists have largely moved on from funding this, except for some fundamental research in reactor designs, simply because market economics means that more plants won't get built, even with a carbon price. In this case, philanthropic neglectedness is a measure of the philanthropic sector's pessimism that more $ towards advocacy would result in greater nuclear power build-out. The scoring in the report treats neglectedness as a positive for nuclear advocacy when there is a strong reason behind the neglectedness that should decrease the score.

COMMENT 4 - COMPARISON TO RELATED WORK AND OTHER INTERVENTIONS Drawdown is another recent project that does a more thorough calculation of carbon mitigation potential for the interventions considered. It reaches different estimates of mitigation potential for some interventions. (10x difference for nuclear for example) I would have liked this report to have considered other GHGs besides CO2, namely methane and refrigerants. Refrigerant emissions reductions have made major progress in recent years with the Kigali Accord (, but there is much to be done in the space of recovering and controlled destruction of existing refrigerants, refrigerant alternatives, and compressor-free cooling designs. Clean and plant-based meats to replace animal agriculture and associated methane emissions was excluded from this report for lack of time to evaluate it. It has importance comparable to or greater than many of the interventions considered and is vastly underfunded in comparison. I hope it is included in a subsequent analysis.

COMMENT 5 - SELECTIONS OF SPECIFIC CHARITIES I think both Clean Air Task Force and Coalition for Rainforest Nations were excellent choices for funding between 5-20 years ago, and this report does a great job of synthesizing their impressive accomplishments. I do not share the outlook that their future work will be as impactful. Given that nuclear (even with new designs) and fossil fuel generation are losing the market competitiveness battle to renewables, and that nuclear, CSS, and power plant regulations are the vast majority of the funding gap for CATF, I expected CATF to have very little impact per $ of additional funding in their campaigns over the next several years. This is the opposite conclusion of the report.
I'm also cautious of the potential for Coalition for Rainforest Nations funding given the projections of future warming and the impact on rainforests as carbon sources/sinks.

I don't have strong candidates for charities I'd recommend in their place, but I'd be happy to contribute to a short-list for the next round of analysis.

Comment by MatthewDahlhausen on [deleted post] 2018-01-20T21:42:06.058Z

The causal chain you propose is:

A) peak oil -> energy scarcity -> humanitarian crisis from ?

If not A), then:

B) emissions -> climate change -> agricultural loss -> humanitarian crisis from famine (with land grabbing exacerbating the crisis)

Let's jump to the crux of Rander's update to the LTG model, since that is the most recent work most closely attached to the concept. The fundamental collapse prediction comes from the pollution - death rate linkage that I mention in the previous comment. What basis is there to assume the overall death rate will increase? And how does the model explain the decreasing death rates in that part of the world? Is it based on a presumed energy scarcity? "when we can assume with relative safety that in 20 years we can only extract half the oil relative to now and that the current pace of global energy transition would need to be multiple times faster to rectify the shortages thereby created" Where do you derive the assumption that oil production will be cut in half in 20 years for reasons of scarcity? U.S. EIA forecasts relatively flat curves. And how do you distinguish good substitutions from shortages? Concerns about peak oil presume a fixed consumption per person, meaning no fuel substitution or demand elasticity. I think this is incorrect. Oil consumption is responsive to price, and even in the least elastic sector where it is used (transportation), there is still a tradeoff in size vs. efficiency for cars people buy. You can go on Gapminder and see how the trend in oil consumption per person can vary quite a bit over several years. Electricity consumption per person (what Turner used as his proxy for "services per person") has actually been decreasing in the U.S. because of large-scale efficiency. I expect we'll see more of that in other sectors including transportation, with lower energy use but greater energy services overall.

Given the substitution and efficiency arguments, and how none of the climate-economic models in IPCC's modeling exercises show an energy scarcity or pollution induced collapse, I don't think causal chain A you propose is a reality we can expect.

So that leaves causal chain B. The World Bank report is for a 4-degree temperature rise, and is by no means a fait accompli. I think what we do now looks a lot like what EAs are currently funding in the region - improving health and encouraging inclusive development. When people have greater incomes and are less dependent on agriculture, climate change effects are less severe. This is the assessment of a follow-up report the World Bank did to the 4-degree report which is worth reading: Shock Waves Managing the Impacts of Climate Change on Poverty.

Comment by MatthewDahlhausen on [deleted post] 2018-01-20T07:27:11.824Z

There is quite a lot to respond to here. I used to be of the same mindset on limits. I followed the The Oil Drum while it was still running and attend a few limits to growth conferences (the ones where attendees called themselves "Doomers"). After engaging with that material, I don't think the projections are accurate and think the catastrophizing is unwarranted. In particular, I don't think resource limits are likely to be a significant issue to humanity in the 21st century. Peak oil concern just isn't a reality; resource economics just doesn't work like that, and demand is elastic. The Oil Drum shut down partly in recognition that Peak Oil wasn't a useful concept anymore, and academics had long since departed from it.

Some more specific points: Factor 1 - Land Grabbing Can you provide citations and sources for the % of population this is happening to?

Factor 3 - Climate "As a consequence, the Sahara will expand well over a hundred kilometres south, a process called desertification." Increasing desterification is a concern, but 100 kilometers advancement across such a large continent isn't going to make a difference. "vast parts of land will become unsuitable for agriculture and hence will force hundreds of people to leave their homes." I would not call hundreds of people a humanitarian crisis. It's not clear that the issues you cite are enough to trigger the catastrophic famine and migrations you prophesize. People respond to droughts and other agricultural challenges in different ways - switching crops, using different water sources, relying on more imports, and finding other income. EA fund a lot of efforts that help this part of the world, malaria eradication and deworming in particular, which yield significant economic gains and life improvement. Development and increasing incomes improves resiliency. You are suggesting that despite these efforts, the factors you describe will overwhelm all of the health and development work being done. You present limited information, and will need more data, models, and economic models to justify the level of concern you are raising.

On Limits To Growth (LTG). Statements along the lines of "The LTG collapse scenario has been fairly accurate to date" imply that there are real world metrics mapped to LTG variables, and that the they expect the underlying model dynamics to remain roughly accurate. I've been able to find one published piece of work where someone explicitly details the variables they use to match to LTG. That report was by Graham Turner, the author of the Guardian Piece you cite. The report is not peer-reviewed, and was published by the institute where Graham is a senior fellow. It is based on a 2008 paper that was peer-reviewed. The author picked per capita electricity consumption and literacy rates to represent global "services per capita". This is misleading. One can pick almost any available variable remotely tied to represent "services per capita" that matches the shape of the LTG model, scale it, and claim that the "LTG standard run is close to reality". There are so many spurious correlations out there. Even then, the majority of trends are 20%, 50%, 100%+ off from the "LTG standard run". The report does not include statistical fit or calibration statistics. How can one meaningful track global pollution? Or non-renewable resources remaining?
It was never the intent of the work to be a predictive forecasting tool. The variables are lumped together proxies to represent categories of real world things, and the authors were explicit when they made the report that these did not represent real world variables; they were to just trying to show the dynamics of their theoretical model. Subsequent updates to the LTG model haven't been able to resolve which collection of real world variables get weighted together to match to which LTG variables. It's easy to cherry pick data to match the trend, especially if you aren't precommitting what constitutes a fit. And even if there is a good match to trends, that doesn't mean that a specific model is the correct representation of reality; there may be many models with wildly different assumptions of the dynamics that produce the same result. Vaclav Smil's review of the LTG is a longer deconstruction of the LTG modeling exercise and worth a read.

More importantly, Dennis and Jorgen (living original authors who I've met) repeatedly say these forecasts are not to be taken literally. Jorgen Randers has a new (2014) forecast which looks very different from the "resource crisis" scenario in the 1972 LTG model. Jorgen now claims the climate crisis is the key concern and the driving force in the model. Even then, he still assumes the same overall model dynamics, but doesn't detail the mechanisms for how the variables will actually influence each-other. For example, in using carbon emissions as his pollution variable, he assumes climate change will greatly increase overall death rates, overwhelming all factors that reduce death rates. There are many models out there (30+) that make assumptions on how climate change impacts human society in the future of which Jorgen's new work is just one. None assume overall death rate increases as Jorgen does, especially in the near term. Be wary of projections from a single model/source. The point is, it is misguided to get doomy about older model forecasts from one model that the authors say are no longer reflective of reality, especially when there is a much wider variety of more complex, robust forecasting models in existence today that have different scenarios.

Comment by MatthewDahlhausen on The Moral Obligation to Organize · 2017-02-17T04:59:33.430Z · EA · GW

Key article from this forum: Developing positive impressions of EA is much more important than near-term growth. If we align with partisan political causes, we risk greatly limiting the eventual scope and impact of EA. Because our movement goal is inherently very different (long term size and positive impression, vs. immediate policy changes), I don't think the organizing knowledge is transferable/useful. Also, much of organizing around left the left is founded in a justice as the core value framework, rather than an impact as the core value framework. There are many posts/arguments in the social justice community explicitly arguing against impact (e.g. arguing against metrics for charity) because these can undermine more speculative causes and deprioritize grassroots/marginalized activists. If we align EA with these movements, we risk undermining the core quantitive and utilitarian values in EA. Because of these risks to EA, I'm partial a firewall between EA and social justice themed organizing, meaning EA orgs do not endorse partisan political causes. This isn't to say EAs should never participate in politics. As you pointed out, there is a lot in international aid that is nonpartisan or very weakly partisan, and the good from doing so is likely to overcoming the risks above.
If we engage in more controversial leftists political causes, EA work would be better spent in cause research, rather than direct political activism. Also, we can prioritize implementing laws that are already passed more effectively, rather than proposing new partisan legislation. This was the aim of the EA policy analytics project. I echo the above comments that elevating organizing to an "obligation" is inappropriate given the speculative nature of impact and possible externalities.