## South Asian Air Quality

This is a writeup of a medium investigation, a brief look at an area that we use to decide how to prioritize further research.

## In a nutshell

• What is the problem? South Asia experiences some of the world’s highest levels of population-weighted PM2.5 air pollution. Our understanding is that poor air quality contributes significantly to negative health outcomes for the more than 1.8 billion people in the region, and that reducing the levels of particulate matter present in the air could save millions of lives.
• What are possible interventions? Possible interventions, many of which require coordinated state action, include improving air quality monitoring programs and crafting and implementing sector-specific policies to reduce emissions. A philanthropist interested in reducing South Asia’s air pollution levels could support efforts to inform the design, implementation, and enforcement of more effective air pollution abatement policies, such as funding monitoring programs, research, and technical assistance.
• Who else is working on this? Philanthropic interest in South Asian air quality appears to be limited but growing rapidly, although many major philanthropic actors seem to address air pollution as a climate concern rather than as a health issue. Outside of the philanthropic sector, we think it’s likely that governments are the biggest spenders on improving air quality. We remain substantially uncertain about the exact levels of funding South Asian governments are directing toward the issue.

## 1. The problem

South Asia – and in particular the Indo-Gangetic Plain that covers parts or all of India, Pakistan, Bangladesh, and Nepal – experiences some of the world’s highest population-weighted air pollution levels.1 Our understanding is that poor air quality contributes significantly to negative health outcomes for the more than 1.8 billion people in this area.2 Of the pollutants present in South Asia’s air, we focus on PM2.5 – particulate matter no larger than 2.5 micrometers in diameter – which we understand to be associated with the greatest health costs.3 In total, the State of Global Air report – a collaboration between the Health Effects Institute and the Institute for Health Metrics and Evaluation’s Global Burden of Disease project – attributes approximately 71.4 million disability-adjusted life years (DALYs) across South Asia annually to air pollution.4 According to the Institute for Health Metrics and Evaluation, air pollution in South Asia accounts for nearly 3% of all DALYs worldwide – i.e. eliminating dangerous levels of air pollution in South Asia alone would reduce the number of prematurely lost years of healthy life by 3% each year.5

Exposure to PM2.5 air pollution can occur outdoors or within households, with the two settings associated with different concentrations, health outcomes, and interventions. Sources of outdoor, or ambient, air pollution in South Asia include brick kilns, vehicles, coal power plants, and crop burning.6 According to the State of Global Air report, South Asia’s average experienced ambient air pollution in 2019 was 78.2 µg/m3, a concentration higher than both the World Health Organization’s recommended standard of 10 µg/m3 and its intermediate standard of 35 µg/m3.7

While we have not investigated the overall evidence base thoroughly, we have encountered widespread agreement that long-term exposure to ambient PM2.5 pollution can result in significant negative health effects, such as chronic respiratory and cardiovascular diseases, that reduce life expectancy. The State of Global Air report, for example, estimates that, in 2019, almost 40 million DALYs in South Asia were attributable to ambient PM2.5 air pollution.8 This number appears to be stable in some South Asian countries and increasing in others.9 While we have not independently vetted this or other mortality and morbidity estimates, it seems reasonably plausible to us based on South Asia’s population-weighted air pollution levels and what the literature we’ve reviewed says about air pollution’s role in chronic illnesses.10

Concentrations of household (as opposed to ambient) air pollution in South Asia appear to be far more difficult to measure, with estimates we found ranging from 35 µg/m3 to over 2,000 µg/m3.11 We found more certainty, however, that household air pollution is widespread: one source estimates that roughly 60% of people in South Asia use solid cooking fuels, the primary source of household air pollution.12 This percentage is apparently decreasing as people switch to cleaner energy sources.13

The lack of reliable household PM2.5 concentration data makes it difficult to confidently discern health effects. The available evidence indicates that negative health outcomes of household air pollution in South Asia may include low birth weight, preterm birth, and other conditions that are correlated with an increased risk of infant death.14 The State of Global Air report, for example, attributed approximately 95,000 infant deaths within the first month of life to household air pollution in South Asia in 2019, estimating an overall impact of approximately 30 million DALYs within the region for that year.15

Of the nations comprising South Asia, India appears to experience among the highest average annual population-weighted ambient air pollution levels – 83.2 µg/m3 – and to contain the greatest number of DALYs attributable to both ambient and household air pollution – 31.1 million and 20.9 million, respectively.16 South Asia’s growing and aging population means that the burden from air quality – all else equal – is rising. In the case of household air pollution, this appears to be more than offset by people switching to cleaner cooking fuels, reducing the burden over time.17 However, the number of DALYs attributable to ambient air pollution appears to be increasing as outdoor air quality is worsening, accentuating demographic trends.18 The outsize impact of air pollution in India relative to other South Asian nations suggests to us that improving India’s air quality could greatly reduce South Asia’s population-weighted annual PM2.5 concentrations and DALYs resulting from air pollution.19

#### 1.1 Why believe these estimated harms

We often have concerns about the quality and reliability of non-experimental social scientific evidence, and prefer to be able to replicate key inputs to our calculations ourselves. That is not possible with the State of Global Air report, which does not have open data and code. So we start with some skepticism that these large DALY estimates should be taken at face value. However, we did a review of the underlying literature and – while, as with all social science literatures, we think there could be room for improvement – we came away thinking that we would probably not want to adjust the State of Global Air burden estimates downward by more than a factor of two.

More specifically, biological mechanisms appear to support the conclusion that exposure to air pollution results in significant negative health effects, including mortality, in humans. Both the American Heart Association and the Lancet Commission on pollution and health, as well as epidemiologists we have spoken with, state that breathing particulate matter generates inflammation and vascular damage. These effects in turn are linked to conditions such as atherosclerosis and high blood pressure, which are known to cause life-threatening diseases such as ischemic heart disease and ischemic stroke.20 In infants, the proposed pathway seems to be that particulate pollution causes lower transmission of nutrients to fetuses, resulting in lower birth weight and nutrient deficiencies associated with higher infant mortality and lifelong health complications.21

There are various animal and human RCTs and studies on these biological mechanisms. The studies generally find that particulate pollution causes vascular inflammation, atherosclerosis, and low birth weight.22 More recent animal studies, however, do not seem to use mortality as an outcome of interest, and some much older studies found null effects of air pollution exposure on mortality.23 According to Open Philanthropy’s scientific research team, the null mortality results in animal models in the older studies are not necessarily evidence against mortality effects in humans, largely due to innate differences in biology and lifespans, though we do take them to be a mild negative update.

Outside of studies on the relevant biological mechanisms, we have found various natural experiments conducted by economists that attempt to isolate the causal effect of particulate pollution on mortality. Ebenstein et al., 2017 in particular examines the health effects of air pollution in conditions similar to those in South Asia, although we’re skeptical of this paper’s headline mortality effects.24 Other quasi-experimental papers, many of which focus on short-term exposure to particulates, generally find meaningful effects on mortality on both infants and adults.25 These papers have reassured us that the non-experimental social science literature we have found is likely not detecting the mortality effects of a confounding variable.

We have not found any meta-analyses that look for publication bias in the quasi-experimental evidence mentioned above. There is an epidemiological literature, however, that contains funnel plots that aim to identify publication bias. In a literature with no publication bias, one would expect to see a symmetric, triangle-shaped pattern of dots in the scatter, with the lower-powered analyses equally likely to fall on the right or left of the high-powered analyses. The plots within Pope et al., 2020 (specifically Figure 4), which examines epidemiological papers on the causal effect of air pollution on mortality in cohort studies, appear to have some asymmetry in the middle of the funnels.26 We very tentatively believe that a publication-bias adjustment based on these charts would reduce the mortality effect size to a number slightly to moderately below the consensus in the epidemiological literature.27

## 2. Possible Interventions

#### 2.1 Government action

Many potential air quality improvements require coordinated state action. The following abatement policies are some of the ones that we thought had a mix of potentially addressing a large portion of the pollution problem and were likely administratively feasible.28

#### 2.1.1 Retrofitting and building efficient brick kilns

20% of clay bricks are produced in South Asia, although PM2.5 emissions attributable to the sector seem to vary by country and be concentrated in urban areas.29 A report by the World Bank estimates that the brick sector is the second-largest PM2.5 contributor in Bangladesh and Nepal, responsible for 11% and 3% of PM2.5 emissions, respectively.30 In India, meanwhile, the share of PM2.5 emissions attributable to brick kilns appears to be comparatively lower, although we have encountered substantial uncertainty around this point. The Health Effects Institute offers one of the lower estimates we found, tracing approximately 2% of India’s PM2.5 pollution and 2 to 3% of its PM2.5-related deaths to brick kilns.31 The World Bank’s report has the highest estimate of the sources we gathered, attributing 8% of India’s PM2.5 emissions to the brick sector.32 The World Bank estimates that retrofitting existing kilns could reduce PM2.5 by 30-50%, as well as improving energy efficiency.33

Despite the uncertainties around emission levels, we think it is plausible (but by no means decisive) that a government-championed effort (e.g., regulations and/or subsidies) to retrofit and build efficient brick kilns would be administratively feasible and could meaningfully reduce PM2.5 pollution from the brick sector.34

#### 2.1.2 Implementing and enforcing a ban on older vehicles

At least since 2015, government bodies in India have indicated an interest in limiting the use of older vehicles.35 There are some regional bans, but it’s unclear to us to what extent they have been implemented or enforced.36 This existing – if inconsistent – interest in banning older vehicles, along with what appears to be a low number of vehicles over 10 years old (meaning political/economic costs of a ban are smaller), suggests that this is a potentially promising area for further government action.37

We’re uncertain about the percentage of the population-weighted PM2.5 pollution in South Asia that is vehicular, although it appears fairly significant. A report by India’s Ministry of Environment, Forest and Climate Change estimates that vehicles contribute approximately 28% of population-weighted PM2.5 emissions in Delhi during the winter and 4% nationally when accounting for all modes of transportation.38 The Energy and Resources Institute attributes 50% of Bangalore’s PM2.5 load to automobile emissions.39 A source apportionment study of Mangalore attributed 70% of particulate pollution to vehicles.40 Older vehicles in particular appear to be a significant contributor to vehicle emissions, with one estimate we found claiming that vehicles older than 15 years account for 15% of total vehicular pollution, and tend to pollute 10 to 25 times as much as newer vehicles.41 Based on these numbers, we think it is likely that a ban on older vehicles could reduce total PM2.5 pollution, although we’re very uncertain about the total reduction we could reasonably expect and how enforceable (and beneficial) a ban would realistically be.

#### 2.1.3 Mandating and enforcing coal scrubbers

Most of the estimates we found attribute approximately 15% of India’s PM2.5 emissions to coal power generation.42 It seems plausible to us that coal is a significant source of PM2.5 emissions, given the prominence of coal in India’s electricity generation and CO2 emissions.43

One report we saw claims that installing wet coal scrubbers in power plants could reduce PM2.5 emissions by as much as 98% and newer fabric filters can reach efficiencies as high as 99.7%.44 While we have not independently vetted this estimate, if accurate, it indicates to us that coal scrubbers could significantly improve India’s air quality.45

The Indian government has already mandated that plants install coal scrubbers to limit emissions, although compliance appears to be limited.46 Given the apparent magnitude of coal power emissions and the government’s existing interest in pursuing mitigation measures, additional efforts to install coal scrubbers might be a promising intervention.

Below, we share our rough back-of-the-envelope calculations (BOTECs) on the potential cost-effectiveness of philanthropic support for the installation of coal scrubbers.

#### 2.1.4 Reducing crop burning with better targeted tractor subsidies

Our impression is that crop burning is a relatively minor source of emissions in India; one article claims that it constitutes an average of 5% of annual PM2.5 pollution in Delhi, although it reaches up to 40% at certain points in the year.47 The vast majority of farmers appear to burn their crops, with only an estimated 20% using tractors to till their fields.48 It seems plausible that better targeting tractor subsidies to increase the percentage of farmers using tractors, while decreasing the percentage of those who burn stubble, could moderately improve Delhi’s air quality.49 We are unsure of the potential impact tractor subsidies might have on air quality across the broader South Asian region.

#### 2.1.5 Better targeting liquified petroleum gas subsidies

From what we have found, solid cooking fuels – still used by approximately 60% of households – account for roughly 40% of the health burden from PM2.5 pollution in South Asia.50 We tentatively assume that substantial reductions in solid cooking fuel use could lead to large reductions in health impacts. The main replacement for solid cooking fuel (e.g. wood, agricultural refuse, charcoal, etc.) is liquified petroleum gas (LPG).

The Indian government already subsidizes LPG use, currently entitling each household to 12 LPG cylinders per year.51 The subsidies, however, do not provide significant discounts to the market price, suggesting that LPG cylinder prices may remain too high for many poor households to afford.52 As a means of increasing the subsidies available to the poor, the government has unsuccessfully attempted to convince wealthier households to voluntarily pay for unsubsidized LPG.53 Better targeting the subsidies by increasing availability and subsidy size for poorer households could plausibly help reduce the number of households that use solid cooking fuels.

#### 2.2 What could a philanthropist support?

We have encountered widespread uncertainty around the share total and population-weighted PM2.5 that is attributable to different sources in India and across South Asia. Addressing this information deficiency seems to be crucial to appropriately targeting abatement strategies. As such, it seems likely that philanthropic efforts may be able to productively focus on 1) improving the information ecosystem for local decision makers and other stakeholders and 2) increasing the technical capacity of key governmental agencies to address air quality. A philanthropist interested in supporting either of these two outcomes might pursue any of a variety of activities, some of which we’ve listed below.

#### 2.2.1 Source apportionment studies

As we mentioned previously, we have found that the deficiency in pollution source apportionment data has made it difficult to gauge the potential impacts of available interventions. Source apportionment studies are scientific studies that attempt to measure what share of the total PM2.5 concentration in a given city or region can be attributed to different sources, e.g. transportation, power generation, other industrial sources, etc.54

Source apportionment studies could be conducted in partnership with interested cities needing technical assistance.55 Such localized studies, in providing governments with rough estimates of their cities’ largest sources of air pollution, could potentially improve governments’ (and philanthropists’) abatement strategies.

Below, we share our rough back-of-the-envelope calculations on the potential cost-effectiveness of philanthropist support for source apportionment studies.

#### 2.2.2 Air quality monitoring

A philanthropist might fund either low-cost sensors or advanced monitoring stations. Low-cost sensors, which we have already supported, can be installed locally and could contribute data to real-time air quality maps that report shifts in pollution amounts. We assume that these maps could help improve public awareness of local pollution levels and precipitate minor behavior change, as well as enable governments and other entities to track the impacts of abatement methods. The limited accuracy of low-cost sensors may impede pollution measurements, however, as individual sensors may not be able to detect small changes in concentrations.56

Advanced monitoring stations are much more accurate – also significantly more expensive – and could be installed in each of India’s airsheds. Potentially combining the stations with sun photometers to measure the atmospheric column could allow for significantly more accurate and frequent satellite measurements of air pollution sources and concentrations.57 These measurements could, in turn, provide governments with more precise pollution targets, enable the effects of abatement policies to be tracked, and contribute to general air quality reporting.

We think that air quality monitoring could be a fairly large source of philanthropic spending in the short term, with smaller ongoing costs after the initial implementation. From our conversations, we also received the impression that Indian air monitoring is comparatively well-funded, and that supporting monitoring in other South Asian regions might generate more impact on the margin. We have not independently vetted these claims.

#### 2.2.3 Research on the abatement curve and air pollution health effects in India

We perceive research on the abatement curve (the graph describing the financial costs and volumes of PM2.5 reductions by intervention pursued) and air pollution health effects in India as having a variety of benefits. A better defined abatement curve data could serve as a menu of options for interested philanthropists or policymakers. Research on health effects could provide more targeted data on the health effects of PM2.5 pollution, including potentially distinguishing between the health effects of different types of pollutants.

In addition, such research could help drive awareness among governments and the public of the extent of the problem and accordingly encourage the adoption of targeted abatement measures (particularly if this research identifies a more narrow set of lower cost policy changes that could solve a large share of the total problem). We have heard that this may be more effective if the research is based at national institutions that also provide expertise to local governments or non-governmental organizations working on this issue.

#### 2.2.4 Technical assistance

Providing technical assistance to government entities could improve the outcomes of pollution abatement measures by increasing governmental capacity to implement, enforce, and monitor air pollution abatement measures. A funder interested in this outcome could, for example, work with outside consultants to provide technical assistance to India’s pollution control boards, which, for a number of reasons, have struggled to enforce air quality regulations.58 We have found conflicting estimates of the pollution control boards’ current spending, although it seems to be between $100 million to$300 million a year, split between air pollution, water pollution, noise pollution, and waste management.59

#### 2.2.5 Policy outreach

The interventions outlined in the section above are largely under the purview of the government. As such, philanthropic efforts might focus on providing decision makers with data and resources to craft effective air pollution abatement policies. Potential funding areas could include source apportionment and concentration research, real-time air quality maps, and reporting on air quality in local news outlets. Other means of increasing the salience of air quality might include funding programs like the Clean Air Fund’s Doctors for Clean Air, which raises awareness of the health impacts of air pollution, or supporting air quality programs at universities.

#### 2.3.2 Source apportionment studies

By our rough calculations, a source apportionment study would need to accelerate a reduction of 0.8 µg/m3 in pollution by 10 years for a city of 5 million to reach our 1,000x bar.61 This calculation assumes that:

• The cost of a source apportionment study would scale as a function of population size. We roughly estimate that a study in a city of 5 million would cost $500,000. • Source apportionment studies would only measure, and thus impact, ambient air pollution levels. • The PM2.5 concentration in cities is proportional to the national concentration. The average annual population-weighted concentration of ambient air pollution in India is 83.2 µg/m3. Approximately 31,140,452 DALYs in India are due to ambient air pollution, and every DALY is valued at$50,000.62
• India’s population is 1,366,000,000.63

#### 2.3.3 Coal scrubbers

According to a report by the Disease Control Priorities Network, installing coal scrubbers in all power plants would cost approximately $1.7 billion.64 The same report estimates that to retrofit the plants with the lowest cost per life saved would cost$615 million, although other sources we’ve encountered estimate costs that are more than an order of magnitude higher.65 If the $615 million figure were correct, paying to install coal scrubbers could reach and perhaps surpass our 1,000x bar, assuming the following conditions are true: • As we discussed above, coal power generation contributes approximately 15% of India’s PM2.5 emissions. • Installing scrubbers reduces PM2.5 emissions by at least 80%.66 • The selected coal power plants are responsible for 75% of the sector’s DALY costs.67 • The health effects of air pollution in India cost approximately$2.68 trillion/year.68
• Given that the government is already mandating the installation of coal scrubbers, our funding speeds installation up by five years.

Under these conditions, we would estimate an ROI of $2.68 trillion (total cost of ambient air pollution) × .15 (power sector share of total PM2.5) × .75 (selected plants’ share of power sector DALYs) × .8 (reduction in PM2.5 from scrubbers) × 5 (years of speed-up) /$615 million (cost of scrubbers) = ~1,960x, though again we do not know these assumptions to be correct and have seen much higher cost estimates in the literature.

#### 2.4 What scale could a program in this area possibly reach?

Based on our understanding of the available funding opportunities, we think there is a high likelihood that a program in this area could spend at least $25 million per year on activities such as air quality monitoring, abatement and source apportionment studies, technical assistance, scaling existing organizations working on air quality, and policy outreach, at a cost-effectiveness level comparable to other funding opportunities we pursue. We think there is a lower likelihood of significantly more than$25M/year of capacity in opportunities we would consider quite cost-effective.

## 3. Who else is working on this?

#### 3.1 Philanthropic organizations

Philanthropic interest in South Asian air quality appears to be limited but growing rapidly: an estimate by the Clean Air Fund, which was cited to us in multiple conversations, puts the philanthropic spending in this area at roughly $7 million in 2019, up from$1 million in 2015.69 We have not vetted the report’s estimates and would guess there are structural underestimates because the report is based on self-reported data from foundations, some of whom may not participate in data sharing, but the estimates are broadly consistent with what we heard in conversations.

The international philanthropic actors working on South Asian air quality that we have heard mentioned most frequently are Bloomberg Philanthropies, the Children’s Investment Fund Foundation, ClimateWorks, the IKEA Foundation, the MacArthur Foundation, the Oak Foundation, Pisces Foundation, and the William and Flora Hewlett Foundation. Some major Indian funders, such as Ashish Dhawan, have also come up in our conversations with experts and funders in the field. We do not believe that this is an exhaustive list: we would guess that we have accounted for the largest philanthropic funders working in this area, but we are certainly missing smaller investments from non-profits and activists.

Many of these major philanthropic actors appear to address air pollution as a contributor to climate change rather than in terms of direct negative health effects from particulates. Climate-focused philanthropic spending on air quality is part of a broader effort to mitigate emissions in India, with philanthropic annual spending on emissions reductions that we think is on the order of $100M-$350M.70

It is unclear to us to what extent treating air quality as a climate concern versus as a health issue would result in significantly different funding strategies. There is definite potential for overlap between climate and air quality spending, as many interventions that reduce greenhouse gasses tend to reduce PM2.5 emissions as well (e.g., limiting reliance on coal for electricity generation). But the two goals can also come apart (e.g., flue-gas desulphurization units on coal plants help improve air quality for health, but as far as we know do not mitigate climate impacts). Overall, we do not think that the presence of significant climate funders mitigates the need for more focused work to improve air quality from a health perspective.71

#### 3.2 Government

We found it difficult to find reliable estimates of governmental spending on air quality. According to one source we found, in the 2019-2020 budget cycle, the Indian government created and allocated a fund of Rs 44 billion (approximately $609 million at the time of conversion) to address air pollution in large cities.72 Additionally, a 2020 report released by the Council on Energy, Environment and Water and UrbanEmissions notes that the National Clean Air Plan, which directs cities to create action plans to reduce particulate matter concentrations by 20-30% by 2024, receives Rs 4.6 billion (approximately$63 million at the time of conversion). However, the report also noted that there are no penalties for non-achievement or “legal mandate for reviewing and updating plans.”73 In fact, only nine cities seem to have noted the costs of execution, which ranged from Rs 890 million to Rs 160 billion (approximately $11.9 million to$2 billion at the time of conversion, respectively).74 We remain substantially uncertain of the accuracy of these estimates and recognize that it’s plausible that there may be additional state funds we do not know about. Overall, we think it’s likely that the government is the biggest spender on improving air quality, but that the current spending is substantially lower than the amount required to adequately reduce air pollution.

## 4. What have we done so far?

Air quality monitoring stood out to us as a particularly tractable abatement strategy that has the capacity to absorb immediate funding. We have accordingly recommended grants totalling $3 million to support a three-year collaboration between Professor Joshua Apte of UC Berkeley, the Indian Institute of Technology Delhi (IIT Delhi), and the Council on Energy, Environment, and Water (CEEW) to install a network of low-cost air quality sensors in South Asia. The aim of the collaboration is that the data from the sensors will inform the design, implementation, and enforcement of more effective air pollution abatement policies. Additionally, we also see this project as an early learning opportunity for the testing and deployment of low-cost sensors across South Asia; if successful, we predict that the sensors could have spillover effects on the speed at which other low-cost sensors are deployed, although we have not consulted experts on this point. Both of these outcomes could plausibly result in significant reductions to South Asian air pollution levels. For our back-of-the-envelope calculations on the potential cost-effectiveness of these grants, see above. ## 5. Potential risks and downsides We have identified a number of potential risks and downsides to funding air quality improvements efforts in South Asia, including: • Immediate spending capacity appears to be limited. We have identified a few abatement activities – such as air quality monitoring and certain forms of technical assistance – that could benefit from immediate funding. We have otherwise struggled, however, to identify areas that have the capacity for large-scale recurring support, and we expect that a philanthropist hoping to make South Asian air quality a long-term funding area may need to consistently find novel grantmaking opportunities.75 • Air quality interventions largely depend on government regulation and enforcement. Accordingly, most philanthropic efforts in South Asian air quality would be limited to activities that inform government policies but that may not directly impact air quality (for example, funding air quality monitoring stations that in turn provide decision makers with data to craft effective abatement measures). It will likely be difficult to predict what the likely impact of these efforts might be. • There are risks and restrictions specific to funding work in India. The Indian government has historically regulated foreign grantmaking within India and recently implemented additional restrictions on foreign funding to Indian NGOs.76 It is our impression that all of the potential philanthropic efforts listed in this writeup are permissible under current laws, but there is a risk that the Indian government could implement additional restrictions that would change that. • Philanthropic efforts could lead to overly restrictive policies in some domains, which in turn could create space for corruption or potentially slow economic growth. While this writeup does not take into account risks from air pollution beyond mortality, we recognize that some abatement strategies may not be economically feasible or desirable after accounting for their costs. • Philanthropic interest in South Asian air quality appears to be growing, so additional funding now could risk crowding out other funders who would otherwise enter. ## 6. Our process and next steps We talked to a number of experts and major funders in the field in the process of researching South Asian air quality. The following individuals agreed to being named as sources for this report, though this should not be interpreted to mean that any experts named here endorse our conclusions in part or in total: • Aaron Van Donkelaar • Ambuj Sagar • Amita Ramachandran • Arden Pope • Avijit Michael • Brikesh Singh • Dan Kass • Ishwar Gawande • Jarnail Singh • Josh Apte • Kanchi Gupta • Matt Whitney • Melanie Hammer • Michael Greenstone • Pallavi Pant • Randall Martin • Reecha Upadhyay • Rohini Pande • Sam Ori • Sangita Vyas • Santosh Harish • Siddarthan Balasubramania • Vinuta Gopal We continue to be open to learning about more opportunities in this space and may make additional grants in the future. ## 7. Sources  Air Quality Life Index, “India Fact Sheet” Source Anderson et al. (2005) Source Apte et al. (2018) Source Arceo et al. (2016) Source Belis et al. (2014) Source Berger (2020) Source BreatheLife, “Cities at the Centre of India’s New National Clean Air Programme” Source Brook et al. (2009) Source Brook et al. (2010) Source Burnett et al. (2018) Source Business Insider, “Indian Government Will No Longer Pay Out Direct Benefit Transfer for Cooking Gas — Subsidy Eliminated as Oil Prices Fall” Source Centre for Science and Environment, “What Will India Do With Its Old Vehicles?” Source Chay and Greenstone (2003) Source Chen et al. (2013) Source Clancy et al. (2002) Source Clean Air Fund, “The State of Global Air Quality Funding” Source Correia et al. (2013) Source Council on Foundations, “New Indian FCRA Amendments Impact Foreign Grants to Indian NGOs” Source Cropper (2016) Source Cropper et al. (2017) Source Currie (2013) Source Deryugina et al. (2019) Source Doctors for Clean Air, “Homepage” Source DW, “India Pollution: How a Farming Revolution Could Solve Stubble Burning” Source Ebenstein et al. (2017) Source Eil et al. (2020) Source EPA, “Particulate Matter (PM) Basics” Source Ganguly et al. (2020) Source Gao et al. (2018) Source Gardener (1966) Source Ghosh (2021) Source GiveWell, “Interpreting the Disability-Adjusted Life-Year (DALY) Metric” Source Goel et al. (2013) Source Greenstone et al. (2015) Source Haryana State Pollution Control Board, “Budget Estimate” Source Health Effects Institute, “Burden of Disease Attributable to Major Air Pollution Sources in India” Source Health Effects Institute, “Household Air Pollution and Noncommunicable Disease | Summary for Policy Makers” Source Heft-Neal et al. (2020) Source Hindustani Times (2021) Source Johnson et al. (2020) Source Kalaiarasan et al. (2018) Source Koshy (2019) Source Landrigan et al. (2017) Source Maharashtra Pollution Control Board, “Budget 2019-2020” Source Martin et al. (2019) Source McCormick (1985) Source Menon (2016) Source Mohan (2020) Source Myllyvirta et al. (2016) Source Narayan (2020) Source National Clean Air Programme, “Final Proposal” Source Open Philanthropy, “Scientific Research” Source Peng et al. (2020) Source Police et al. (2018) Source Pope et al. (2009) Source Pope et al. (2016) Source Pope et al. (2020) Source Rakshit (2020) Source Roeyer et al. (2020) Source Sharma and Dikshit (2016) Source Sharma and Kumar (2016) Source Sharma and Nagpure (2019) Source Snider et al. (2015) Source SS Rana & Co (2020) Source State of Global Air 2020, “Explore the Data” Source State of Global Air 2020, “Homepage” Source Task Force on Hemispheric Transport of Air Polution, “Questions and Answers” Source The Financial Express, “To Curb Stubble Burning, Pay Attention to EPCA on Making Straw Management Machines Affordable” Source The New Indian Express, “Three Years on, Not Many Willing to Give Up LPG Subsidy” Source The World Bank, “Population, Total — India” Source Times of India, “Centre Cuts Pollution Control Budget, Draws Flak From Experts” Source Times of India, “Foreign Contribution Regulation Act” Source Tripathi (2020a) Source Tripathi (2020b) Source Tuli (2020) Source Varadhan (2019) Source Veras et al. (2008) Source Zhang (2016) Source ## Research and Development to Decrease Biosecurity Risks from Viral Pathogens This is a writeup of a medium investigation, a relatively brief look at an area that we use to decide how to prioritize further research. ## In a nutshell What is the problem? We think natural, and to a greater extent engineered, pathogens have the potential to cause global catastrophes. We expect that as biotechnology advances, the risk of dangerous outbreaks will increase. Our impression is that viral pathogens seem especially likely to contribute to catastrophic pandemics because they have the potential to be highly virulent and transmissible compared to other pathogen types, and there are very few broad-spectrum therapeutics for use against pathogenic viral outbreaks and they have undesirable side effects. This report focuses primarily on reducing the risk from viral pathogens through scientific research and development, especially on vaccines and therapeutics. How could the problem eventually be solved or substantially alleviated? We believe that if a subset of the following abilities/resources were developed, the risk of a globally catastrophic pandemic would be substantially reduced: • A better selection of well-stocked, broad-spectrum antiviral compounds with low potential for development of resistance • Ability to confer immunity against a novel pathogen in fewer than 100 days • Widespread implementation of intrinsic biocontainment technologies that can reliably contain viral pathogens in the lab without impairing research • Improved countermeasures for non-viral conventional pathogens • Rapid, inexpensive, point-of-care diagnostics for all known pathogens • Inexpensive, ubiquitous metagenomic sequencing • Targeted countermeasures for the most dangerous viral pathogens A deeper understanding of the immune system also seem useful for its potential to expose new potential threats and countermeasures, though we see this as a source of potential important “unknown unknown” considerations rather than having a specific vision for how the research will lead to alleviating the problem. This report focused on vaccines and antivirals because we investigated them in relatively greater depth. We did that because they seemed like broad and important areas where we guessed that we might be able to uncover particularly promising projects related to averting catastrophic viral pandemics. We didn’t look as deeply into the other areas listed above because our briefer investigations indicated to us that a deep investigation was relatively less likely to be fruitful, generally because the areas seemed less important and/or less neglected. spreadsheet we drafted summarizes our overall views on this subject. What are the possible research interventions? There are a wide variety of methods of conferring passive and active immunity to pathogens. As computational models and gene editing techniques have become more advanced, new strategies involving these technologies have become increasingly feasible. Research into novel and technologically advanced vaccine and passive immunoprophylaxis (which we here categorize with vaccines) development methods, such as ab initio antigen and antibody design, and vectored immunoprophylaxis, currently appear especially promising for their potential to expand the range of pathogens against which immunity can be conferred. Thus far, we have identified only a few specific promising projects in this space, and many of the most promising-seeming lines of research may be fully funded already. Within research and development related to antivirals, host-directed antiviral compounds (i.e. antivirals that target part of the hosts’ cellular machinery, rather than targeting the virus) appear promising to us since some inhibit machinery used by a large number of viruses, making them likely to be relatively broad-spectrum, and making it seem less likely that individual pathogens will develop resistance to them. We think these compounds are unlikely to prove fully efficacious against all viruses in humans, but that they merit further investigation, and note that more extensive research on their antiviral effects in vitro, in animals, or on humans could be funded. Who else is working on it? Our Scientific Research Program Officers’ general impression is that there are many academics and companies working on vaccine and diagnostics research and development. We are unsure how much of this work is relevant to understanding and mitigating the risk of globally catastrophic pandemics (as opposed to developing improved vaccines for known pathogens with less pandemic potential). We speculate and have seen anecdotal evidence that companies may not be incentivized to focus on work related to rare but potentially catastrophic outbreaks, because those areas generally provide weaker and less reliable revenue streams than work related to chronic conditions (e.g. HIV, hepatitis). However, we encountered several efforts that appear especially relevant to the effort to develop vaccines and therapeutics specifically for the purpose of countering novel and/or potentially pandemic viral outbreaks. Of the areas discussed above, our impression based on our research is that: • Broad-spectrum antivirals are receiving limited attention. We’re planning to fund work in this area. • Vaccine R&D is generally a crowded space, though it seems possible to us that a deeper investigation into more of the specific subtopics would reveal additional opportunities. • Diagnostics R&D seems highly crowded. ## 1. Our process We decided to investigate scientific research and development that could assist with our Biosecurity and Pandemic Preparedness Program. Much of the research was conducted by our Scientific Research Program Officers, Chris Somerville and Heather Youngs (“Chris” and “Heather” throughout the rest of this writeup), who are biochemists and scientific generalists with no prior expertise in this topic. Former Open Phil scientific advisor Daniel Martin-Alarcon also contributed to this research. We asked them to survey the fields of vaccine and antiviral research and development, identify promising projects that were not being pursued, and help us understand how much progress is being and seemingly could be made on the development of rapid vaccines and broad-spectrum antivirals if various potential research projects were successful. Chris wrote an analysis of what steps could be taken to create a vaccine (or multiple vaccines) against a novel pathogen in approximately 100 days or fewer, and what scientific advances this would require (this may already be possible in some cases). They also briefly investigated the topic of the development of diagnostics for potential pandemic pathogens. We chose those topics because we thought they had the potential to be most relevant to preventing or reducing harm from pathogens with the potential to be globally catastrophic. Chris, Heather, and Daniel conducted literature reviews and collectively wrote about 150 pages of rough material on this area, which was shared internally. Topics investigated included viral zoonosis, identification of new human pathogens, antigen discovery, candidate vaccine development and testing, animal models, systemic scientific issues, clinical trials, and technologies for scale-up and distribution. They declined to publish these materials without doing a significant amount of work to clarify and refine them, and Open Phil decided it would not be a worthwhile use of their time. In aggregate, Chris and Heather spent about 7 weeks each researching the topics in this report, and Daniel Martin-Alarcon spent approximately 2 weeks. Claims not cited are generally based on the internal report produced by, and subsequent conversations with, Chris and Heather. In some cases throughout this text, citations are provided as examples of support for the associated claims, but may not be the primary or original reason we believe the claims to be true (often, our belief is based on information and impressions conveyed to us by Chris and Heather, which stems from both their general understanding of many of these topics and their speculation based on a wide range of readings, which we expect would be unduly time-consuming and ultimately unsatisfying to attempt to cite fully). We may continue and extend this investigation in the future. Nick Beckstead and Claire Zabel reviewed the materials produced by Chris and Heather. I, Claire Zabel, drafted this page, and it was reviewed by Chris and Heather and some other Open Phil staff before it was published. Note that this report does not constitute a comprehensive overview of our thoughts on this area. We omitted information when we thought the public discussion of the topic (either of particular types of risks or countermeasures we find promising because they might address those risks) could contribute substantially to the risks while not offering commensurate benefits. ## 2. What is the problem? Our shallow investigation into biosecurity describes the broader problem as we see it.1 In brief, we think natural and engineered pathogens have the potential to cause global catastrophes. We expect that as biotechnology advances, the risk of deliberate attacks or accidental releases of dangerous pathogens will increase. If scientists and health professionals had the capacity to quickly and accurately identify pathogens, had access to reliable broad-spectrum therapeutics, and could rapidly develop effective vaccines against novel pathogens, it seems like many of the biosecurity risks we are most concerned about would be substantially smaller. We’re particularly concerned about pandemic risk from viruses because of (i) their potential for high transmissibility and virulence, and (ii) the lack of effective therapeutics for many viral diseases. Because of this, we’re interested in research interventions that could be useful against a variety of potentially dangerous pathogens (especially viral pathogens), and could make medical countermeasure development faster and more effective. ## 3. How could the problem eventually be solved or substantially alleviated? This section focuses on imagining how scientific advances could eventually make it possible to prevent or substantially reduce the risk of a catastrophic pandemic. A spreadsheet we drafted summarizes our overall views on this subject. We go into greater detail on the subjects which seemed most promising to us once the initial research review was completed. Those subjects include: • Establishing a portfolio of strategies to rapidly and reliably develop efficacious and safe vaccines. If researchers had many different ways to stimulate immunity to novel viral pathogens, and thus develop vaccine candidates within months, that seems likely to substantially reduce the chances that pathogen could cause a globally catastrophic event, compared to worlds in which it takes many years to develop a vaccine or researchers are unable to develop vaccines against some pathogen types. • Investigating and developing efficacious broad-spectrum antivirals against which it is unlikely that a virus could evolve resistance. If these antivirals were successfully developed, they could be deployed in the event of a dangerous viral outbreak, possibly as soon as the outbreak was announced. We speculate that government offices such as the Biomedical Advanced Research and Development Authority (BARDA) might stockpile the antivirals in advance for this purpose, if the antivirals were available. It seems plausible to us that substantial scientific progress could be made in the two areas listed above within years or a few decades. Research in the following areas might lead to the recognition of additional risks or strategies for reducing risks. Because these areas are more exploratory, we have found it more difficult to anticipate which concrete positive outcomes they might lead to and timelines on which those might be realistic, and we don’t have particular concrete visions for how this could happen. We see these areas as sources of unknown unknown considerations with the potential to change our understanding of the risk landscape in important but unpredictable ways. • Basic research in immunology: greater knowledge of how the immune system works might aid in the design of more effective and safe immunogens (molecules that stimulate an immune response in the host), as well as open up new lines of research into other potential countermeasures. Chris and Heather’s impression was that scientists still lack understanding of many aspects of the human immune system, and thus they have limited ability to predict which methods stimulate immunity to different pathogens (i.e. it is very difficult to generate a good vaccine). Further research could lead to insights into the human immune system, which we imagine could enable scientists to identify new sources of risk and better predict which strategies for creating new therapeutics and vaccines are likely to succeed. We focused on topics where we thought it was most likely we could identify neglected but broadly significant research areas related to averting viral pandemics. However, some other themes we did not investigate and report on as thoroughly are listed below: • Preventing accidental release and making deliberate misuse more difficult with biosafety measures that don’t interfere with research. This is based mainly on our speculation about the possibility of altering pathogens to make them usable and safe for research in the lab but inviable outside it. For example, Benjamin tenOever and his colleagues at the Mount Sinai Medical Center devised a method for engineering flu viruses to “carry a 21-base-pair-long sequence that complements miR-192, a microRNA found in human and mouse lung cells but not in the respiratory tract of ferrets….” This microRNA binds with influenza RNA transcripts, flagging them for destruction within the cell. Viruses engineered with this method caused symptoms in ferrets but not mice and, by extension, presumably not humans. This method, which they call “molecular biocontainment” could potentially be used to create viruses that could be studied realistically in model organisms but would be unable to harm humans if they were released from a lab.2Further work in this area could test whether strategies that have already been proposed would interfere with research, or lead to the development of new molecular biocontainment strategies. We speculate that if molecular biocontainment tools were robust and in widespread use, the risk of accidental release of dangerous pathogens would be substantially reduced. However, it seems plausible that uptake would in fact be low and some or many labs might continue to engage in more dangerous practices. More experiments could be done to determine whether this technique could interfere with experimental results.3 • Platform technology for diagnostics: Research and development of diagnostics could help healthcare workers reliably and cheaply identify common, rare, or novel pathogens. This could be valuable for quickly identifying the presence and spread of outbreaks, and ensuring that infected individuals receive appropriate treatment, if it’s available. Chris and Heather’s impression is that: • A wide variety of rapid diagnostics are currently available or under development and the field is well funded at the moment. • Many companies are already working in this space to improve on current technology and reduce costs. But they also note the following limitations: • Some diseases that don’t provoke a strong immune response and/or reside in certain relatively inaccessible tissues (e.g. brain tissue) remain difficult to diagnose. However, our Scientific Research Program Officers suspect that highly infectious pathogens will be relatively more straightforward to find with a diagnostic because those pathogens will likely be shedding large amounts of virus into bodily fluids. • Some diagnostics are relatively imprecise (e.g. it might be possible to identify that a patient suffers from influenza, but not to easily identify the strain). • Some diagnostics require a relatively long time (days) to yield results, which can make treating individuals and identifying potentially pandemic viral outbreaks at the outset of the outbreak more difficult. • Some diagnostics require access to equipment that is expensive and/or hard to use in the field. It seems to us that work on this area is likely to be broadly useful for diagnosing more common pathogens as well as those that might cause dangerous pandemics, and so we would guess this area is less likely to be neglected than research and development in areas that have fewer common applications. So, we have tentatively decided against prioritizing this area as highly as the ones listed above for conventional grantmaking. This line of reasoning suggests, however, that if there are types of diagnostics that are mainly useful for identifying pathogens likely to be involved in potentially catastrophic pandemics (for example, novel ones), those types of diagnostics might be relatively neglected. • Non-viral therapeutics (antibiotics, fungicides, etc.): Work on therapeutics for non-viral pathogens could also lead to the development of new countermeasures. Focusing on antivirals seems in expectation more impactful to us because (i) it’s our impression that conventional non-viral pathogens are less likely to be responsible for catastrophic pandemics and (ii) broad-spectrum therapeutics exist for many non-viral pathogens (e.g. antibiotics are effective against many types of bacteria), though resistance to existing therapeutics sometimes renders them ineffective.4 However, we have not investigated this area deeply, and we otherwise restricted this writeup to R&D related to viral pathogens only. • Medical countermeasures aimed at addressing specific pathogens: Research and development using known techniques could expand the range of medical countermeasures available to target specific pathogens of concern. Examples of this kind of work might include creating influenza vaccines that are effective against the most virulent forms of influenza. Our understanding is that certain projects along these lines, if they are aimed at providing fairly robust defenses against some of the pathogens that seem most dangerous, may be highly impactful, but that projects aimed at pathogens that seem less concerning are lower-priority for funders with our focus. • Metagenomic sequencing for enhanced surveillance: The cost of metagenomic sequencing (sequencing from environmental samples which may contain genetic material from diverse organisms) might fall and systematic sampling, e.g. at airports, might be established such that it becomes feasible to rapidly and reliably identify pathogens with pandemic potential. We anticipate that that would make it substantially easier to contain dangerous outbreaks, but have deprioritized the area because our strong impression has been that there are many actors focused on the goal of reducing the cost and difficulty of metagenomic sequencing. Establishing a system for detecting outbreaks early and reliably is an area of interest to us, but does not seem directly related to the focus of this report (scientific research and development related to potentially catastrophic viral outbreaks). ## 4. What are the possible research interventions? Additional research could be pursued on all of the topics listed above. However, we focus below on impressions about and future research directions that seem promising related to: • Platform technologies and strategies for the rapid development of vaccines (i.e. technologies and strategies that might be useful for developing many potential vaccines quickly, not only ones directed against one or a few existing pathogens). • Broad-spectrum antivirals We focused on those because they seemed the most likely to be useful against a globally catastrophic biothreat in the near future, and we thought a systematic review of the literature might turn up promising giving opportunities for a new funder. However, we also think additional research in the other areas described above could prove valuable, and we have sufficient uncertainty that it would not surprise us if research on those topics proves as or more valuable. ## 5. Platform research and development related to vaccines #### 5.1 Background on vaccine development Developing and using vaccines has several established advantages over other types of medical countermeasures; namely, vaccines often only need to be used once or relatively rarely to protect an individual from a disease, and (partly because of this) vaccines are often cheap enough to be widely deployed in the developing world. In addition, once a vaccine is developed a population can often be preemptively vaccinated, meaning there is a relatively large window of time in which this intervention can be usefully deployed if the threat can be identified in advance. These advantages, while substantial, seem relatively less important for addressing the biosecurity threats we are most concerned about (which may involve novel threats that are only likely to arise once or rarely, and which are difficult to predict with sufficient specificity to immunize the general population in advance), than they are in the context of most work on public health. Nonetheless, if it were possible to create new vaccines rapidly, it seems likely that they’d prove invaluable tools against potential viral pandemics, because they can be customized to provoke the immune system to provide strong protection against specific pathogens of concern. This is in contrast to therapeutics like antivirals (which seems likely to be less efficacious and accompanied by more severe side effects, based on our general understanding of the track record of these types of medical countermeasures). Chris and Heather believe that eventually immunization is likely to be possible against many or all dangerous pathogens, and that rapid vaccine development against most pathogens of concern is slightly more likely than not to become possible in the next 20 years. Chris and Heather broke the process of vaccine development for a novel pathogen into four steps: 1. Sequencing the pathogen’s genome: Chris and Heather report that sequencing the pathogen is likely to be straightforward and rapid (though working with dangerous pathogens often requires substantial protective gear and specialized facilities), and there are already significant extraneous pressures to reduce the cost and increase the speed of sequencing, so we did not think searching for giving opportunities at the sequencing stage of the process was likely to be as impactful as work on the next two stages. They noted that during the 2014 Ebola outbreak, in-field sequencing of the samples was completed within 24 hours.5 2. Antigen discovery and design: The process of identifying antigens, molecules that stimulate the production of antibodies and other components of an immune response against the relevant pathogen, and possibly designing antigens that provoke a strong immune response that neutralizes the pathogen. This step may be unnecessary if conventional vaccine development methods, such as injecting deactivated or weakened forms of the pathogen, are effective and safe. 3. Vaccine candidate formulation: The process of developing candidate vaccines. Generally, vaccine development involves delivering the relevant antigens in some form to the relevant population so that the patient’s immune system produces the necessary antibodies. However, short-term immunity may in some cases be achieved by delivering antibodies produced in a lab in cells from another organism instead (this is called passive immunization, and sometimes is not counted as a type of vaccine development, though we group it here for simplicity). Multiple vaccine development strategies could be deployed simultaneously. 4. Testing: The process of testing a vaccine candidate for safety and efficacy, generally first in animals and then in humans. We expect that this process might be substantially abbreviated in the event of a sufficiently severe outbreak. Steps two and three seemed the most likely to have neglected-yet-impactful opportunities for improvement from a scientific R&D perspective. Below we enumerate some parts of the antigen discovery and vaccine candidate formulation process that we investigated. #### 5.1.1 Antigen discovery This is the process of discovering a pathogen’s antigens, the molecules on the pathogen that stimulate an immune response in the host. This step is primarily used in the development of vaccine design in cases in which conventional vaccine development methods, such as using an attenuated (weakened) or deactivated pathogen, is infeasible or ineffective. Below are some areas of research related to this process that could improve or hasten vaccine development. • Biosensor platforms: Researchers could develop better biosensor6 platforms for detecting the binding of antibodies to an antigen. This might allow them to better distinguish the immunogenicities of different antigens. Chris and Heather report that many platforms are already available and we think this is unlikely to be the bottleneck on the development of effective vaccines and prophylactics, though there may be room to incrementally improve the data quality of high-throughput devices. • Structural/computational protein design: Research in this area could advance vaccine development in a few different ways: • Once the amino acid sequence of an antigen is known, it is not necessarily the case that if it is synthesized separately from the rest of the original pathogen, it will form the same shape as the antigens in the virus and continue to bind to the relevant antibodies. This increases the difficulty of creating effective vaccines using methods of vaccine development that don’t involve the entire virus (either deactivated or attenuated). Computational tools could be used to predict protein interactions and design delivery platforms (e.g. nanoparticles or virions) for proper antigen display. • Chris and Heather speculate that those tools could help researchers design antigens ab initio that are better at stimulating the appropriate immune response than the antigens in the original pathogen.7 Artificial antigen design might be worthwhile because some diseases, such as influenza, do not naturally present antigens to the body that are capable of stimulating a strong immune response (instead, the antigens they present mutate rapidly, so immunity to influenza is usually fleeting and restricted to only some strains of the disease). These tools are being applied in the lab, but the research is still at the relatively preliminary stage and they have not yet led to vaccines which are approved for use in humans. • Alternatively, Chris and Heather theorize that in the future researchers may be able to use information about antigen-antibody interactions and computational tools to predict the optimal antibodies for binding to the antigen.8 In that scenario, those antibodies could then be tested, synthesized (if successful), and injected to deliver passive immunity (discussed in more detail below). This could be useful in a scenario in which there have not been instances of successful immune responses clearing the pathogen outside the lab. However, Chris and Heather are uncertain about whether computational models have advanced sufficiently to be able to routinely achieve these goals. Overall, it seems like further research into the development of methods for identifying and presenting antigens that will lead to the synthesis of improved antibodies has the potential to be useful in the event of the release of a dangerous pathogen, especially one engineered to escape the natural immune response. Funding work on improving computational models seems like a promising opportunity for a philanthropist interested in this area, and we have funded one project in this space.9 #### 5.1.2 Vaccine candidate formulation There are several well-established methods of developing new vaccines, including using live attenuated or inactivated versions of the pathogen, among others.10 We don’t focus on these in this writeup, because it was our impression that additional research on these methods would be less likely to have substantial impact. We made this judgment because we believe that there is likely to be more low-hanging fruit related to new strategies and because we believe the new strategies are more likely to expand the range of pathogens that can be vaccinated against. However, some ideas involving the application of new gene-editing technology to older methods also seem promising to us.11 Our Scientific Research Program Officers looked at some less-established (partially overlapping) strategies for vaccine (and passive immunoprophylaxis) candidate development. These include: • Nucleic acid vaccines: Nucleic acid vaccines involve delivering nucleic acids (DNA or RNA) coding for the antigens to cells, not the antigens themselves (as is the case with conventional vaccines). Once the antigens are produced by the patient’s cellular machinery, their immune system (hopefully) produces antibodies, generating immunity to the disease. No vaccines of this type have been approved for use in humans, though several DNA vaccines are in use to prevent diseases in nonhuman animals12 and trials on many nucleic acid vaccines are ongoing.13 There are many DNA vaccines in development, including the recently approved vaccines for Zika, however, the effectiveness has been lackluster in clinical trials.14 Improvements in delivery and adjuvant activation are ongoing and may result in effective DNA vaccines. RNA vaccines may be more efficacious because they don’t need to be delivered to the cell nucleus. Our Scientific Research Program Officers’ overall impression is that RNA vaccine testing thus far indicates good results in animals and appears promising in humans. • Viral vector delivery: DNA or RNA coding for the antigens of viral pathogens (but not the other, harmful parts of the virus) could be integrated into a virus that is generally not pathogenic in humans, or encapsulated in a viral coat so that it can deliver the nucleic acid into human cells with high efficiency. Then people could be infected with that (non-pathogenic) virus and the viral machinery could induce the infected people to create the antigens, and subsequent immune response, conferring immunity.15 If effective, this could address some of the delivery problems with nucleic acid vaccines raised above. Adeno-associated viruses (AAVs) are among the most-studied candidates for this method of delivery, though there are others.16 While this strategy involves risks such as not working in the fraction of the population that had already been exposed to the relevant virus, it nonetheless seems possible to us that it would be one of several useful strategies to pursue in the event of the emergence of a potentially catastrophic disease outbreak. • Passive immunoprophylaxis: Instead of delivering antigens and thus stimulating a full immune response, researchers could inject the relevant antibodies (which the immune system produces in response to the antigens). These antibodies might be delivered in the form of antisera (human or nonhuman blood serum containing antibodies from a survivor of the pathogen), or produced in a recombinant cell line or organism (e.g., plant). Once the antibodies have been delivered and if the process is effective, the host will have short-term immunity against the pathogen of concern but the immunity will be limited because no memory B cells (cells which produce the relevant antibodies and which are indirectly stimulated by the antigens) will be produced. Another limitation of this strategy is that some diseases require T-cell immunity or other aspects of immune response, in addition to the production of neutralizing antibodies, which this approach doesn’t deliver. However, this strategy has the following advantages: 1) it could be used to rapidly immunize people who are incapable of producing effective antibodies 2) in the future it’s possible that this strategy could be used to administer artificial antibodies that are superior to the ones human produce naturally. • Vectored immunoprophylaxis: Researchers could engineer DNA or RNA that codes for the creation of antibodies, then integrate that DNA or RNA into a (relatively safe) virus (for example, an AAV), as is described in the case of viral vector nucleic acid vaccines (above), except with antibodies instead of antigens. Susceptible groups could then be deliberately infected with the non-pathogenic virus. If effective, this would cause the body to produce the antibodies, temporarily protecting the vaccinated person against the pathogen of concern if he or she becomes infected.17 Similar limitations to the ones described above for passive immunity (e.g. impermanence) apply, though it may be substantially easier to immunize large numbers of people this way. This is because we expect it to be substantially easier and cheaper to produce the nucleic acid sequence coding for an antibody and insert it into a virus than it would be to produce and administer the antibody protein itself en masse. There are several other lines of research on vaccines that seem like they could plausibly be impactful, including the ones listed below (though, none of the below both strongly attracted our interest and were not already being pursued). We note them here only briefly with the purpose of representing more of the breadth of possible research topics: • Adjuvants: vaccines that stimulate an insufficient immune response alone are sometimes accompanied by compounds called adjuvants that increase the immune response to the vaccine. Further research on adjuvants could lead to stronger responses to vaccines with otherwise low efficacy. • Vaccine production platforms: different vaccines are produced using different platforms. For example, whole animals, eggs and cell lines can be used to generate whole (live, attenuated, or killed) vaccines, whereas cell lines or plants are more appropriate for recombinant vaccines and/or virus-like particles. Not all systems are suitable for all pathogens. New platforms could facilitate more rapid production of vaccines. • Effective surrogates for safety and immunity. In order to make more effective or new vaccines, it would be useful to have a non-human organism that has an identical immune response to humans and similar sensitivities for toxicity in which to test the safety and efficacy. Because of the development of new genome editing technologies it is is now theoretically possible to strongly modify the immune systems and other biochemical pathways of animals to more closely resemble that of humans. • Polysaccharide antigens: Like other antigens, polysaccharides and glycosylated protein antigens can elicit part of innate immune responses, in addition to stimulating the production of neutralizing antibodies. And, Chris and Heather note that mutations that alter glycosylation may alter presentation of classical antigens, reducing vaccine efficacy. However, most of the new approaches to vaccine design, such as producing nucleic acid vaccines, are not expected to be useful for this, and the immune responses to polysaccharides or protein glycans are poorly understood and difficult to study, and tools for artificial synthesis are lacking. #### 5.2 Vaccine R&D in preparation for a potentially catastrophic pandemic In the event of an outbreak of a highly virulent and transmissible pathogen, we would guess that multiple lines of research might be pursued simultaneously. For example, researchers might in parallel attempt to: 1. Identify antigens that can be used to prepare a nucleic acid vaccine 2. Use computational methods to produce recombinant vaccines in cell cultures 3. Clone antibodies that might be useful for generating passive immunity sera 4. Create a live attenuated vaccine Chris and Heather report that in some cases it’s possible to develop an initial vaccine candidate within six months or faster (although in most circumstances completing and evaluating the clinical trials necessary for the vaccine to be approved by the FDA takes years). We know of several examples of this, although they are importantly disanalogous to the likely situation we’d expect would occur with a novel viral outbreak: • Flu vaccines are prepared twice annually in preparation for seasonal flu,18 although our understanding is that the development processes depend on the vaccine backbones and organizational infrastructure developed in previous years (facilitating accelerated development). • Related to the above, Dormitzer et al. (2013) reports that “[w]ithin 6 months of the [2009 H1N1 outbreak] pandemic declaration, vaccine companies had developed, produced, and distributed hundreds of millions of doses of licensed pandemic vaccines.” although it also notes that these vaccines were for the most part not ready until after the pandemic had already entered a natural decline.19 • The World Health Organization (WHO) announced the Zika outbreak in February of 2016, and by August 2016 several Zika vaccine candidates were in clinical trials.20 We do not know when development of those vaccine candidates began, and the WHO announcement may not be a good indicator of when vaccine development began. This space seemed fairly crowded to our Scientific Research Program Officers, and they did not encounter many gaps in the research being pursued. Their overall impression is that there are not substantial obstacles remaining to developing vaccines in under 100 days, depending on the pathogen type,21 although to the best of our knowledge it has never been done and both (potentially severe) scientific obstacle we haven’t identified and logistical issues could arise. However, this would not include testing for safety and efficacy. They commented that, based on the level of sophistication of the science, they were surprised that relatively few new and effective vaccines have been developed in recent years, and they did not know why that was the case. They are reasonably confident that further advances in vaccine development will result in novel useful vaccines in the coming years, and that emergency scenarios would (if they occurred) spur more rapid development of relevant vaccines, as evidenced by the recent responses to Zika and Ebola. However, we are uncertain about whether this would substantially increase investment and progress in platform technology related to vaccine development. Overall, our impression is that the following seem particularly likely to be valuable to prepare in advance of the emergence of a potentially globally catastrophic pathogen: • Ab initio protein design for improved antigens and antibodies • Vectored immunoprophylaxis But thus far, we have identified only a few specific promising projects in this space. More generally, our Scientific Research Program Officers think developing a more mechanistic (i.e. coming from an understanding of the mechanisms involved in stimulating immunity) rather than empirical (i.e. trying out different approaches and checking if they succeed) process for developing vaccines is likely to result in more reliable and rapid production of new vaccines. ## 6. Antivirals for pandemic preparedness #### 6.1 Background on antivirals In contrast to vaccines, it’s our impression that relatively little work has gone into antiviral research and development in recent years. And, it seems to us that the majority of antivirals currently available are not likely to be useful countermeasures in the event of a potentially globally catastrophic pandemic. That’s because: • Many of them are not broad-spectrum; most only treat one or a few different viral diseases.22 • Existing antivirals have variable effectiveness, and over time viruses have been evolving resistance to them.23 #### 6.1.1 Host-directed antivirals Some compounds which exhibit antiviral activity are host-directed, meaning that they target proteins in the host that the virus depends on rather than the virus itself. The ones we know of are inhibitors of chaperone proteins, proteins which assist in the folding of other proteins, including viral proteins. Our Scientific Research Program Officers’ reading of the literature suggests that these types of antivirals may be unsuitable for long-term use, since their mechanism of action relies on interfering with host-cell machinery, and thus they may cause relatively severe side effects.24 However, some of them (Hsp90 inhibitors) have been tested for medium-term use as therapeutics for cancer patients, and have been tolerated.25 Hsp90s are “chaperone” proteins. It appears that dependence on Hsp90s among viruses is widespread, and may be universal.26 Hsp90 inhibitors appear particularly promising to us because the ones we know of seem likely to be relatively broad-spectrum and difficult for viruses to evolve resistance against (because the antivirals target relatively conserved virus-host interactions). Hsp70s are also chaperone proteins with inhibitors which we think may be promising, although Chris and Heather report that there is some weaker evidence of broad-spectrum antiviral activity among Hsp70 inhibitors and less evidence indicating that they may be safe for human use. Overall, we think these compounds merit further investigation, though we think they’re unlikely to prove fully efficacious and safe against viruses in humans due to the low overall base rate of discovery of new, highly effective pharmaceuticals. If these compounds were effective against highly virulent viruses at similar doses to those used for treating cancer patients, it seems likely to us that these side effects would be considered acceptable. Chris and Heather speculate that these antivirals would only be employed for short-term (i.e. on the scale of days or weeks) use against a virus; the idea would be to use them to mitigate the effects of the virus and “buy time” for the immune system to launch an immune response and clear the virus. Chris and Heather thought it was unlikely that viruses could evolve or be easily engineered to have resistance to host-directed antivirals, because the pathway of protein folding is both complex and very fundamental, and so they said that many mutations would be required obviate the requirement for a given chaperone protein. There are other proteins involved in the chaperone protein complex with inhibitors that may also prove useful as broad-spectrum antivirals, but the Hsp90 inhibitors seem the most promising because those are the only chaperone protein inhibitors that have been studied extensively in humans and have been found to be relatively safe. A funder could fund studies on these antivirals’ efficacy against different viruses in vitro or in animal models or humans, for viral pathogens against which they have not yet been tested. We are in the process of investigating funding opportunities in this space. ## 7. Who else is working on it? Our Scientific Research Program Officers’ overall impression is that there is a lot of commercial activity related to infectious disease vaccine and therapeutic development, and there are many companies working on the development of vaccines and diagnostics for use against viruses with greater economic potential in the developed world (e.g. influenza, hepatitis). However, we found it challenging to evaluate what proportion of the work is likely to be relevant to addressing pathogens with the potential to cause globally catastrophic pandemics. We did not look deeply into this topic because we thought it would be more useful to first identify more specific research topics that seemed most promising in this space, then investigate those. Thus, this section is relatively perfunctory compared to similar sections in other cause reports of ours. It seemed plausible to us that this area might be neglected relative to its importance because the relevant people and groups might not see work on it as their responsibility or in their best interest; for example, we speculated that pharmaceutical companies might not be incentivized to develop vaccines and therapeutics designed specifically for use in a catastrophic pandemic, especially if they have side effects that make them unsuitable for use against less severe diseases, or diseases for which less dangerous alternative treatments exist. That’s because developing pharmaceuticals is generally very expensive,27 and we think that the relevant opportunities are relatively unlikely to deliver continual and reliable revenue streams (because the risks are rare and disproportionately associated with societal disruption), compared to e.g. treatments for chronic and common diseases. We did encounter some potentially relevant efforts in this space that we thought were worth highlighting: • The National Institutes of Health (NIH) allocated an estimated$1.7B for “biodefense” research in 2016, $1.4B of which went through the National Institute of Allergy and Infectious Disease (NIAID).28 • The Coalition for Epidemic Preparedness Innovations (CEPI), a group formed in 2016, states that it intends to develop vaccines against known pathogens that may have epidemic potential.29 In January 2017, it was reported that approximately$500M in funding had been committed to CEPI.30
• In 2013, The Joint Science and Technology Office for Chemical and Biological Defense worked on post-infection antivirals including antibodies and FDA-approved drugs.31
• In February 2017, the Defense Advanced Research Projects Agency (DARPA) announced a four-year initiative called the Pandemic Prevention Platform (P3) program to prepare nucleic acid vaccines within 60 days of the identification of a novel pathogen.32

Other information that seemed relevant to us:

• The NIH does not report spending on antivirals as a distinct category.33 However, 55 grants related to “broad-spectrum antivirals,” representing ~$17M, were reported by Grantome in 2015.34 • Chris and Heather’s impression during this investigation was that there are a variety of funders and other actors involved in platform tools for vaccine development, such that the topics they investigated did not seem highly neglected. • Chris and Heather contacted companies and researchers that had previously been involved in the development of Hsp90 and Hsp70 inhibitors, as well as some other groups in the field we thought might have insight into this, and did not find evidence of ongoing research on the development of these inhibitors as broad-spectrum antivirals. • If countermeasures were developed, the Biomedical Advanced Research and Development Authority (BARDA) might stockpile them. BARDA’s stated mission is to develop and procure medical countermeasures that address public health threats, including pandemic influenza and other infectious diseases.35 The Department of Defence may also manufacture relevant medical countermeasures.36 • There may also be non-public governmental research related to pandemic pathogen countermeasure R&D ## 8. Questions for further investigation • How neglected are the various themes discussed in this document that relate to vaccine development (e.g. “computational protein design,” “vectored immunoprophylaxis,” etc.)? What are the most promising unfunded projects related to these themes? • On what timescales could we expect to achieve advances in vaccine production such that vaccines against the most dangerous pathogens can reliably be developed in 100 days or fewer? What would the likely positive consequences be in a pandemic if vaccines could be produced that much earlier? • What types of viral pathogens with the potential to produce globally catastrophic pandemics could not be addressed with the advances discussed in this writeup? What research would help address the risks posed by those? • Are host-directed antivirals relatively safe and effective in humans? What viral pathogens would they not be sufficiently effective against, if any? • What types of research into immunology would be most likely to yield insights that prove useful for preventing catastrophic disease outbreaks? ## 9. Sources  BARDA: BARDA unveils path forward in the BARDA Strategic Plan 2011-2016 Source (archive) Carolson, 2016 Source (archive) CEPI: Approach Source (archive) Clinicaltrials.gov keyword “DNA vaccine” Source (archive) Clinicaltrials.gov keyword “hsp90” Source (archive) Clinicaltrials.gov keyword “RNA vaccine” Source (archive) Cohen 2017 Source (archive) Correia et al. 2014 Source (archive) DARPA: “Removing the Viral Threat” 2017 Source (archive) De Clercq and Li 2016 Source (archive) Department of Defense Chemical and Biological Defense Annual Report to Congress, 2014 Source (archive) Devitt 2013 Source (archive) Dormitzer et al. 2013 Source (archive) Geller, Taguwa, and Frydman 2012 Source Grantome.com “broad spectrum antivirals” Source Hasson, Al-Busaidi, and Sallam, 2015 Source Howe and Haystead, 2015 Source Kutzler and Weiner, 2008: Table 2 Source (archive) Morrison 2016 Source (archive) NIAID Fiscal Year 2017 Congressional Budget Justification Source (archive) NIH Categorical Spending 2017 Source (archive) Overview of the Department of Defense’s (DoD) Advanced Development and Manufacturing (ADM) Facility and Capabilities, 2017 Source (archive) Quick et al. 2016 Source (archive) The Open Philanthropy Project’s grant to the University of Washington for “Universal Flu Vaccine and Improved Methods for Computational Design of Proteins” November 2017 Source The Open Philanthropy Project’s non-verbatim summary of a conversation with Gigi Gronvall, October 6, 2014 Source The Open Philanthropy Project’s non-verbatim summary of a conversation with Wendy Barclay, October 2, 2014 Source Shaw 2017 Source (archive) Tripp and Tompkins 2014 Source (archive) Tufts Center for the Study of Drug Development, 2014 Source Sanders and Ponzio 2017 Source (archive) Vaccines.gov “Types of Vaccines” Source (archive) Willis et al. 2013 Source (archive) World Health Organization: Vaccines Against Influenza, 2012 Source (archive) ## Mechanisms of Aging This is a writeup of a medium investigation, a relatively brief look at an area that we use to decide how to prioritize further research. ## In a nutshell • What is the problem? Aging is a major contributor to cardiovascular disease, cancer, diabetes, neurodegenerative diseases, and other causes of death and impairment. Better understanding, and being able to mitigate, the basic mechanisms of aging could therefore contribute to reduced age-related mortality and impairment for a very large number of people. • How could the problem eventually be solved or substantially alleviated? We think it’s plausible that age-related diseases and impairments could be alleviated if scientists achieved specific research objectives in a number of areas (prevention/correction of epigenetic errors, senescent cell removal/prevention/reprogramming, reversing/addressing stem cell exhaustion, and organ/tissue regeneration and replacement). This list is not exhaustive and is populated with areas that seem to us to be especially fundamental and/or dynamic and potentially broadly applicable, and therefore seem like plausibly useful areas for subsequent research.We are highly uncertain how large the potential gains might be if the above objectives were achieved, but think several years of healthy life extension (and possibly more) could plausibly be made in the next 10-20 years in some areas (such as senescent cell removal), while areas that could yield substantially greater gains will likely require multiple decades of progress in enabling areas like neuroscience, selective delivery of agents to cells and/or organelles, and epigenetics. However, we don’t have a strong opinion about whether supporting the nearer-term or longer-term objectives is likely to have greater expected benefits per unit cost. • Who else is working on it? The NIH reports spending$2.7 billion per year on aging-related research. We are unsure how much of this is relevant to understanding, preventing, and mitigating the basic mechanisms of aging. We are aware of several foundations and nonprofits focused on aging research with collective expenditure around $80 million per year (likely an underestimate of philanthropy in the area). During this investigation, we became aware of several companies explicitly focused on aging with about$1 billion in investment collectively.

Open questions include: which themes are neglected; what levels of healthy life extension might be realized if the different research goals are accomplished; and which long-term research directions are most promising and neglected.

## 1. Our process

Our scientific advisors, Chris Somerville and Heather Youngs, are biochemists and scientific generalists with no prior expertise in aging research. We asked them to survey the field of aging, divide it into subfields, identify promising projects that were not being pursued, and help us understand the potential impact on healthy lifespan if various potential long-term research projects were successful. The latter question was not discussed in the literature, and we had to approach it very speculatively. We excluded investigation of science mainly relevant only to one specific age-related disease (e.g. research on cancer, cardiovascular disease, and Alzheimer’s) but included some topics relevant to many age-related diseases (e.g. developing the ability to grow organs from induced pluripotent stem cells).

Our advisors conducted literature reviews, spoke with several people in the field, and wrote rough internal memos for other staff to review. This was their main priority for roughly one and a half months.1

Nick Beckstead drafted this page and it was reviewed by our scientific advisors and a few other Open Philanthropy Project staff before it was published.

## 2. What is the problem?

Aging is a major contributor to cardiovascular disease, cancer, diabetes, neurodegenerative diseases, and other causes of death and impairment (some of which, such as muscular atrophy, loss of teeth, and damage to joints, are so common that they are not generally considered “diseases”).2 Proposed basic mechanisms are various and have disputed levels of comparative importance, but include “genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, deregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, and altered intercellular communication.”3 Additional potential mechanisms are presented below.

#### 2.1 What currently available interventions can address this problem?

There are a large number of symptoms associated with aging. Some are widely recognized as diseases and are subject to a variety of treatments (e.g., neurodegenerative disorders; heart disease); others are not considered “diseases,” and there are generally few if any treatments targeting them (e.g., normal muscular atrophy). While it is conceivable that there could be treatments addressing aging “in general” (e.g., addressing all or a large proportion of associated symptoms via a single mechanism), such treatments have not been conclusively demonstrated and may not be possible. There are approaches that have been hypothesized to fit in this category, such as caloric restriction. While some of these have been tested in model systems, they have not been tested in humans for the purpose of extending healthy lifespan, and we would guess that they would not have radical effects on healthy lifespan if they were tested (but plausibly could be substantially positive). We did not carefully consider nutrition and lifestyle interventions (except for caloric restriction) because we have a strong prior that available data is inconclusive and our science team’s expertise is more in the direction of molecular biology and biochemistry than nutrition.

## 3. How could the problem be substantially alleviated?

This section focuses on imagining, very speculatively, how scientific advances could eventually make it possible to prevent or substantially alleviate some problems associated with aging. Claims not cited are generally based on the internal memos produced by, and subsequent conversations with, our advisors (along the lines of the process described above).

This list highlights some imaginable scientific advances that attracted the interest of our scientific advisors because of their potential to extend healthy lifespan. The first three attracted the interest of our scientific advisors because they appear to address “basic mechanisms” of aging that might account for a large proportion of aging-related symptoms, whereas the last is less basic in this sense but seems especially dynamic and potentially broadly significant. The list is not exhaustive. With those caveats and clarifications in mind, we would guess that healthy lifespan might be extended if scientists eventually were able to:

• Prevent the accumulation of epigenetic errors associated with aging, or restore more youthful epigenetic states in cells. Various alterations of epigenetic state4 are correlated with both chronological age and symptoms of aging, and there are theoretical reasons to expect that these alterations would cause symptoms of aging.5 Interventions on the epigenetic state of mice have shown results consistent with the points previously stated6 but we anticipate that multiple aspects of inducing “epigenetic rejuvenation” would be challenging.7
• Solve the problem of senescent cell accumulation. As animals age, senescent cells (i.e., cells which neither grow and divide nor apoptose) accumulate. Research suggests that senescent cells contribute to damaging inflammation and may also suppress tissue regeneration by stem cells. In lab experiments, mice who had a portion of their senescent cells removed lived about 25% longer than mice in the control group that were born at the same time.8 We imagine that selectively killing senescent cells might become possible through a number of generic strategies.9 We have a less specific sense of how reprogramming senescent cells or preventing them from becoming senescent in the first place might work, but imagine that advances related to epigenetics (discussed above) or advances related to several of the other directions discussed in this document could be helpful. We mean to raise the above generic strategies only as plausible possibilities and do not have confidence in the feasibility or timeline for success of particular approaches.
• Reverse stem cell exhaustion. Somatic stem cells are induced by factors such as growth, normal senescence, and tissue damage to divide and replenish other cells. As adult humans age, their stem cells appear to become depleted or increasingly less active, which is thought to decrease the body’s capacity to replace and repair damaged tissue.10 It’s unclear how much of what we call aging is attributable to decreased stem cell activity, and what causes the apparent decrease in stem cell activity, but numerous factors have been implicated.11 Additionally, as noted above, cytokines released by senescent cells may play a role. We see a few (speculative) possibilities for addressing stem cell exhaustion, and we discuss three of them in a footnote.12
• Learn how to use induced pluripotent stem cells (IPSCs) to regenerate and/or replace tissues and organs damaged by aging and aging-related diseases Progress in getting IPSCs to differentiate into other cell types and organoids may allow repair or replacement of organs such as liver, pancreas (islet cells), lungs, kidneys, spinal cords, eye, and heart, and also some cell types in the brain, and could contribute to management of aging-related diseases affecting these cells/tissues/organs/organoids, including heart disease, diabetes, liver disease, vision loss, ALS, and Huntington’s disease.13

We are highly uncertain about, and do not have internal consensus regarding, the potential extension in healthy lifespan that might result if 1-2 of the above objectives were accomplished. Some of us see several years of healthy life extension as the plausible potential upside and others see larger possible gains, but all of us involved in creating this report expect that any increase in healthy lifespan would keep average lifespan within the range of natural lifespans observed in humans today (barring a historically exceptional increase in the rate of scientific progress). We would guess that much more radical life extension would likely require a larger number of successes like these and likely multiple successes that are not listed here, and we accordingly assign it much lower probability in the next few decades (with some caveats). We held this view about the difficulty of radical life extension prior to this investigation. Our findings fit with this prior view, and the investigation did not strongly affect our views on the matter.

Some other themes are also potentially important to aging, but they are covered in this write-up in less detail because we have focused on topics that seemed more basic, dynamic, and/or potentially broadly significant to us. The themes covered in less detail here include: genomic instability, telomere attrition, loss of proteostasis, deregulated nutrient sensing, mitochondrial dysfunction, altered intercellular communication, decline of the immune system, inflammation, neurodegeneration, the microbiome, and damage to individual cells (e.g. antioxidants and DNA repair). We may investigate inflammation and decline of the immune system more thoroughly in the future because these topics caught the interest of our scientific advisors. We compare our list of highlighted topics with a plan proposed by researchers at the SENS Foundation in a footnote.14

## 4. What are the possible research interventions?

Common obstacles to achieving the goals stated above include lack of ability to selectively deliver agents to desired cell types, measure and control the epigenetic state of cells, and understand and control differentiation and functioning of stem cells (plausibly closely related to the previous item). Therefore, progress on these more general themes may assist with extending healthy lifespan. Other types of relatively general research may also be helpful or necessary for substantial extension of healthy lifespan, such as general progress in neuroscience, improved biomarkers for various aspects of aging, and/or improved model organisms for aging.

We think substantial progress on many of the themes mentioned above is likely to require decades of work, so our intuition is that long-term, basic research (with an emphasis on tool development) in areas like neuroscience, selective delivery of agents to cells and/or organelles, and epigenetics is likely to be the most important work for making the greatest possible progress relevant to age-related disease and impairment in the long run (though we have a limited sense of which tools and research directions are likely to be most promising and/or neglected). Our reasoning for thinking this work will be more important in the long run can be made clearer by reference to a thought experiment: imagine that 30 years ago, a funder were working toward understanding and mitigating the fundamental causes of aging. Our intuition is that they would want a large share of their effort to go toward supporting very basic work on gene sequencing, microscopy, the areas of cell biology that have led to the field of epigenetics, and topics in cell differentiation that led to the discovery of stem cells – rather than work that might fall under the auspices of “aging research” per se. We suspect something similar is still true today. One observation we can offer in support of this is that many of the important questions relevant to extending lifespan could not even have been asked 30 years ago (e.g. some questions about stem cells and epigenetics).

Some of the themes listed above do not seem to have as many basic obstacles as others. For example, it seems plausible to us that some of the above objectives related to senescent cell removal and heterochronic parabiosis could be achieved in the next couple of decades. In these cases, we don’t have a strong opinion about whether supporting research primarily relevant to these objectives or long-term objectives is likely to have greater expected benefits per unit cost.

## 5. Indefinite vs. moderate healthy life extension

We think the best case for this cause involves the prospect of healthy life extension within the range that some humans currently live. In contrast, some people who are interested in the mechanisms of aging have promoted the idea of “curing” aging entirely. Some thoughts on this:

• Our default view is that death and impairment from “normal aging” are undesirable. However, we would have some concerns about indefinite life extension, mainly related to entrenchment of power and culture. We don’t have internal consensus on whether, and to what extent, such indefinite life extension would be desirable, and don’t consider it highly relevant to this write-up.
• We don’t see promising life science research that would result in indefinite life extension in the next few decades, barring a historically exceptional increase in the rate of scientific progress.
• When we consider possible transformative technologies that could result in indefinite healthy lifespan, some staff members think some kind of identity-preserving digital emulation is more likely than radical scientific advancements related to physiological aging, but others are more skeptical about the relevance/feasibility of that idea.15

Our program officer Nick Beckstead offers the following forecast to make the above more precise/accountable: By January 1, 2067, there will be no collection of medical interventions for adults that are healthy apart from normal aging, which, according to conventional wisdom in the medical community, have been shown to increase the average lifespan of such adults by at least 25 years (compared with not taking the interventions). (Subjective probability: ≥93%)

The prediction is called off if some other innovations cause a historically exceptional increase in the rate of scientific progress during this period (such as the development of transformative AI capabilities). The prediction excludes diet, exercise, and lifestyle, as well as existing medical interventions for healthy people (such as currently available vaccines).

## 6. Who else is working on it?

The NIH reports spending $2.7 billion per year on aging research in 2015.16 In the 2015 budget request,$510 million per year is tagged as “neuroscience” and $177 million per year is tagged as “aging biology.”17 We have heard in various conversations that this research is mainly relevant to addressing particular symptoms associated with widely-recognized diseases (e.g., Alzheimer’s disease), rather than on understanding the basic mechanisms that cause aging. This is plausible to us, but we haven’t seen any convincing evidence for it and we do not take it for granted. We sometimes hear the sentiment that research on aging is neglected because of an attitude that “curing” it isn’t desirable, but we haven’t seen any evidence that the NIH takes that attitude. A brief Google search revealed the following non-profit organizations working in the space, with all funding totals reflecting amounts dedicated to aging-related research: the Buck Institute ($35 million total annual budget in 2014);18 the Glenn Foundation ($11 million total in grants in 2014);19 the SENS Foundation ($1.5 million in grants and $5 million in total expenses in 2014);20 the American Federation for Aging Research ($7.7 million in grants and $10 million in total expenses in 2015);21 the Larry L. Hillblom Foundation ($6 million in grants and $6.8 million in total expenses in 2015).22 The Ellison Foundation is leaving the field.23 Some aging-focused companies working in this area that we became aware of in the course of this investigation include Calico ($500 million in disclosed investment and agreed upon potential for $1B more);24 Human Longevity, Incorporated ($300 million in investment);25 Unity Biotechnology ($119 million in investment, currently focused on senescent cell removal);26 Alkahest ($53.5 million in investment, focused on neurodegeneration and interventions related to heterochronic parabiosis);27 and Ambrosia (investment figures not readily available online).28

Another overview of funding in this area, made in 2015, is available here.29 Our survey of funders, non-profits, and companies in this area is incomplete and is somewhat skewed toward research topics highlighted in this write-up. We have a limited understanding of the pharmaceutical industry’s spending on research and development related to aging.

We have a limited sense of the absolute and relative neglectedness of the various categories of research discussed in this report. However, our scientific advisors identified specific unfunded projects related to the following themes:

• Understanding the mechanism(s) driving regeneration associated with heterochronic parabiosis: Experiments have indicated that the blood of older animals can have deleterious effects on younger ones, and that the blood and organ functioning of younger animals can improve the functioning of old ones, though to date the hypothetical increase in healthy lifespan has not been tested.30 Understanding the biology responsible for the observed effects might eventually lead to interventions that address health problems associated with aging.
• Aging and epigenetics: Documenting correlations between tissue-specific epigenetic states and signs of aging with a longitudinal cohort study and/or systematic examination of cadavers of people dying at various ages could yield valuable information that could lead to treatments to address aging-related health issues.

## 7. Questions for further investigation

• How neglected are the various themes discussed in this document (e.g. “epigenetics and aging,” “senescent cells,” etc.)? What are the most promising unfunded projects related to these themes?
• What is the comparative potential upside of accomplishing the core objectives related to these various themes for extending healthspan?
• With what probability and on what timescale could such successes be achieved?
• How likely is it that general-application tools and basic research areas that might not be thought of as part of “aging research” (analogous to epigenetics, stem cells, neuroscience, and drug delivery) will be bottlenecks to accomplishing the core objectives described above? What tools and/or research directions under these headings are most neglected relative to their promise, for the purpose of addressing these bottlenecks? What other general-application tools and basic research areas might be important for accomplishing these core objectives? Would interventions focused on these more basic/general themes have greater or smaller effects on the time by which such objectives might be achieved?
• What research programs could help scientists discover all aspects of the epigenetic state of cells and make it possible to measure and intervene on those aspects of cells? To what extent are the most important research programs of this nature being pursued already?
• How likely is it that advances in drug delivery (including delivery of other agents to cells) would be required for effective senescent cell removal or interventions to correct or prevent the accumulation of epigenetic errors? If such advances are needed, what are these advances? What research programs could lead to these advances? To what extent are the most important research programs of this nature being pursued already?
• What are the most important mechanisms of aging that were not investigated in this write-up?
• To what extent are the hallmarks of aging traceable to a few basic mechanisms, vs. a large number of distinct mechanisms that could not plausibly be addressed together except by many separate interventions?

## 8. Sources

DOCUMENT SOURCE
AFAR 990 for 2015 Source
Ambrosia website Source
American Diabetes Association, Type 1 Diabetes Source
Baker et al. 2016 Source
Buck Institute 990 for 2014 Source
Calico Press Release 2014 Source
Chen et al. 2016 Source
Conboy et al. 2005 Source
Cosgrove et al. 2014 Source
Crunchbase, Human Longevity Source
Crunchbase, Unity Biotechnology Source
Farr 2013 Source
Friedman and Friedman 2002 Source
Glenn Foundation 990 for 2014 Source
Horvath 2013 Source
Horvath et al. 2016 Source
Koyuncu et al. 2015 Source
Hillblom Foundation 990 for 2015 Source
López-Otín et al. 2013 Source
Lowe 2017 Source
Nelson 2015 Source
NIH Categorical Spending 2016 Source
NIH FY 2015 Congressional Budget Justification Source
Ocampo et al. 2016 Source
Pitchbook, Alkahest Source
Sandberg and Bostrom 2008 Source
Sen et al. 2016 Source
SENS Foundation 990 for 2014 Source
Social Security Administration, Period Life Table, 2013 Source
Sweatt 2010 Source
Taimen et al. 2009 Source
Trounson and DeWitt 2016 Source
Villeda et al. 2011 Source
Wikipedia, List of the verified oldest people Source
Zealley and de Grey 2013 Source

## Animal Product Alternatives

This is a writeup of a medium investigation, a brief look at an area that we use to decide how to prioritize further research.

## In a nutshell

• What is the problem? More than eight billion land animals are raised for human consumption each year in factory farms in the U.S. alone. These animals are typically raised under conditions that are painful, stressful, and unsanitary. Chickens account for a large share of the animals affected. Currently, there are no alternatives to meat, dairy, or eggs that have succeeded in displacing a large fraction of the market for animal-based foods.
• What are possible interventions? Successfully developing animal-free foods that are taste- and cost-competitive with animal-based foods might prevent much of this suffering. A for-profit investor could invest in companies developing plant-based alternatives to animal-based foods (such as Hampton Creek and Beyond Meat); cultured egg and dairy products (such as Clara Foods and Muufri); plant-based foods that use bioengineering to mimic the taste and texture of meat (such as Impossible Foods); or cultured meat products (such as Modern Meadow, though Modern Meadow is developing a novel, ‘meat-based’ product that is not a direct substitute for traditional meat). A philanthropic funder could support academic work aimed at developing cultured ground meat or meat with a more complex structure (such as steak or chicken breasts) in hopes of bringing the products closer to development and making them more attractive to profit-motivated investors. Our impression is that developing cultured egg whites will be substantially easier than developing ground meat, which (in turn) will be substantially easier than developing meat with complex structure (such as steak or chicken breasts). Based on cost analyses, comparisons with tissue engineering and biofuels, and discussion with scientists who have experience with cell cultures and tissue engineering, we currently see developing cost-competitive cultured muscle tissue products as extremely challenging, and we have been unable to find any concrete paths forward that seem likely to achieve that goal. We have not closely investigated the challenges associated with creating plant-based alternatives to animal products.
• Who else is working on this? There are well-funded companies developing plant-based alternatives to animal-based foods. The largest potential gaps in this field seem to be for cultured meat and cultured eggs. There are no companies developing cultured meat as a replacement for animal-based foods (though Modern Meadow makes cultured leather and “steak chips”); this problem receives little attention from governments and philanthropy (we estimate < $6M in funding in the last 15 years), and much of the work is unlikely to be done by people in neighboring fields (such as tissue engineering). Because we do not see a concrete path that could lead to commercially viable cultured meat in the short term, we would guess that cultured meat is also a poor fit for profit-motivated investors. We are aware of only one company developing cultured egg whites (Clara Foods), and it has received approximately$1.7 million in funding.

## 1. What is the problem?

See our overview of factory farming for more detail.

## 2. What are possible interventions?

#### 2.1 Overview and general issues

It seems to us that if there were plant-based or cultured alternatives to meat and eggs that were cost- and taste-competitive with animal-based foods, it could greatly reduce the amount of meat and eggs produced, and thereby greatly reduce pain and suffering of animals produced for food.

There are a variety of possible routes to developing high-quality alternatives to animal-based foods, including:

• Plant-based alternatives that simulate the taste and function of animal-based ingredients. For example, Hampton Creek and Beyond Meat are companies working in this space (see below).
• Fermentation (yeast-based production) of ingredients originally from animals using synthetic biology. For example, Clara Foods puts the genes coding for egg white proteins in chickens into yeast so that the yeast can produce the same proteins in fermentation processes.1 Muufri is taking a similar approach for milk.2
• Various types of engineered muscle tissue (aka “cultured meat”). This approach involves growing muscle cells from animals in cell cultures, and forming the cells into tissues that can be eaten. Different versions of this approach include:
• Millimeter-scale muscle strands grown in the lab, (hereafter “ground meat”) which are similar to muscle from an animal and might be used to replace ground meat. For example, Professor Mark Post is working on developing cheap animal-free cell culture media3 (i.e., liquids that provide the chemical environment and nutrients for animal cells to grow in the lab), incorporating separately-grown fat tissue into the ground meat product to improve the product’s taste,4 and working on overall manufacturing scale-up.5
• Large whole pieces of cultured muscle tissue (hereafter “slab meat” in contrast with ground meat). For example, Amit Gefen, funded by Modern Agricultural Foundation, is exploring the feasibility of making a whole piece of chicken meat, such as a chicken breast.6 We have not investigated what technical strategies they are considering.
• Novel meat-based products that would not directly substitute for conventional meat, such as Modern Meadow’s “steak chips.”7

Our investigation focused on cultured meat and yeast-based production of egg whites, rather than other possibilities listed above, because:

• There are a number of well-funded companies pursuing plant-based alternatives (see ‘Private companies’ under ‘Who else is working on this’?).
• There are many more chickens producing eggs than there are cows producing milk,8 so we guessed that the “fermentation” work focusing on eggs would be more promising than the “fermentation” work focusing on milk.

Within our investigation of cultured meat, we focused primarily on ground beef (rather than slab meat or ground chicken or pork) because:

• Work on cultured ground beef seems to be more developed than work on cultured ground chicken or pork.9
• The market for ground beef is much larger than the market for ground pork or chicken (more on this below).
• Our understanding is that progress toward cultured ground beef would be relatively transferable to cultured meat from other animals. Mark Post suggested that work on ground meat is highly transferable between species, noting that he was able to transfer from mice to pork and from pork to beef in his lab in roughly six months in each case.10 Other researchers also reported that the same (or a similar) protocol for culturing muscle cells has worked for multiple species, including mice, rats, and chickens.11
• Apart from an ongoing feasibility study by Amit Gefen (see above), we are not aware of anyone trying to make cultured slab meat. Moreover, we would guess that progress on ground meat would also help with developing slab meat because some of the challenges are shared (e.g., finding a low-cost, animal-free media to feed the cell culture).

We found limited evidence on the question of whether the public would buy cultured animal products and did not pursue the issue further.

#### 2.2.1 Potential impact of taste- and cost-competitive cultured meat

In 2014, ground beef accounted for 43.3% of total beef sales,12 but for other animals, ground meat accounted for a much smaller portion of the total meat market. For example:

• Chicken breasts made up 56% of the market for chicken in the U.S. in 2014, and ground chicken only claimed 1% of the market.13
• Ground lamb was only 6.2% of the market for lamb in 2014.14
• Ground pork was only 1.9% of the market for pork in 2014.15

We would therefore guess that the market for a cultured ground beef product would be much larger than the market for a cultured ground chicken/pork product.

It is possible that displacing a large fraction of the markets for these other types of meat (such as chicken) would require developing cultured slab meat. If so, that would make the value of successfully developing cultured slab meat much greater than the value of successfully developing cultured ground meat because a very large fraction of all land-based farm animals are chickens.16

#### 2.2.2 What is the state of the art for making cultured meat?

Our understanding is that Mark Post’s cultured beef hamburger represents the state of the art for cultured meat.17 The main steps in his process are as follows:18

1. Extraction of satellite cells, which are skeletal muscle stem cells, from a muscle sample of an animal taken using a needle biopsy.19
2. Proliferation phase to grow large numbers of cells from an initial batch by encouraging the cells to divide, producing copies of themselves.20
3. Differentiation of the muscle precursor cells into mature muscle cells. Growing or “conditioning” the cells in the right mechanical environments (attached to scaffold) that allow them to “get exercise” by contracting against structures that offer resistance. The muscle cell contraction boosts protein production in the cells and increases their size.21
4. Harvesting the muscle strands and combining them with color, flavor and texture enhancers, such as separately grown fat tissue.22

As discussed above, our understanding is that the process for making cultured ground meat from other species is/would be similar.

The process for making the “steak chips” that Modern Meadow is pursuing similarly begins with harvesting and growing muscle cells, however the final steps to process the product are different. The first step is a biopsy to harvest muscle cells from a cow and then create large quantities from that sample by growing them in the lab and allowing them to divide and produce more cells like themselves. After growing the cells, the next step is to harvest them by separating the cells from the liquid they are grown in (cell culture media). The final step is to combine the cells with pectin and flavorings (e.g., teriyaki or BBQ) and use a food dehydrator to make them into chips.23 We would guess that the technical challenges involved with making the steak chips are probably simpler than the challenges involved with making ground meat because making ground meat still involves making small chunks of tissue, whereas steak chips could conceivably be made without establishing much (if any) tissue structure. However the manufacturing scale-up challenges associated with cost-effectively producing a large number of cells still remain.24

Methods for making slab meat are currently unknown.25 Here it would seem important not only to have a process for growing and flavoring cells, but also to have a way of generating the (potentially complex) three-dimensional structure associated with non-ground meat.

In 2013 it was reported to have cost Mark Post $325,000 to make a single hamburger with cultured ground beef in a university research lab.26 We do not have a detailed understanding of the costs for Mark Post’s prototype. However, major sources of cost for large-scale manufacturing in tissue engineering include the cost of cell culture media, facilities, maintaining sterile conditions in the facilities, and skilled labor.27 Obstacles to decreasing costs include lack of knowledge of how to make cheap animal-free media (researchers don’t know what ingredients would work), how to cheaply and efficiently grow muscle cells on scaffolding, and how to harvest and process this cultured tissue at massive scales. Researchers also don’t know how to grow cells quickly in culture – academic scientists have estimated one month to grow one batch of meat in a bioreactor.28 Longer production times increase the cost of production and increase the risk that any given batch is contaminated.29 In order to decrease the cost, a funder could support lab work aimed at: • Cheap animal-free media that supports muscle cell growth. Cell culture medium is the liquid that provides the chemical environment and nutrients needed for animal cells to grow in the lab – it is the liquid surrounding the cells in the Petri dish. Standard cell culture media for muscle cells contains fetal bovine serum (extracted from the blood of cow fetuses), which is not suitable for cultured meat production focused on reducing animal agriculture because it is likely that one or more cow fetuses would be required to make 1 kg of meat.30 Currently there are commercial animal-free media formulations that do not contain fetal bovine serum, but we’ve been told these are expensive and may not encourage high growth rates of muscle cells because current animal-free media formulations have not been optimized for cultured meat production. Hence, finding a cheap animal-free liquid medium to grow the cells in is essential to making cultured meat commercially viable.31 Developing a cheap animal-free medium would likely involve guided trial and error of many combinations of potential ingredients that could support good muscle cell differentiation and growth.32 • Established cell lines. When doing research experiments to improve cultured meat development, a researcher must have animal cells to work with. If each researcher uses cells harvested from a different animal, it may be difficult to get reproducible and comparable results across multiple labs, slowing progress in the field because it is harder for separate researchers to build on each other’s work. An “established cell line” is a standard lineage of cells that are capable of proliferating indefinitely (whereas normal mammalian cells can only divide a limited number of times).33 While established stem cell lines exist for humans and mice, they do not exist for agricultural animals such as cows and chickens. Nicholas Genovese is currently working on this. Established cell lines may also be used to form the starting material for manufacturing-level production, reducing the need to continually harvest from animals.34 • Efficient scaffolding designs for ground meat. To develop tissue structure and protein production similar to that of muscles in an animal, muscle cells grown in the lab need to grow on structures (aka “scaffolds”) that mimic the mechanical environment they would experience in an animal. Currently there are at least two main approaches for scaffolding for ground meat or processed meat products: growing tissue in thin sheets (e.g., Modern Meadow) or growing cells and tissues on scaffolds (e.g., velcro) in lab-scale dishes and bioreactors (e.g., Mark Post).35 We do not have a detailed understanding of the imperfections of current approaches to scaffolding, but a number of people we spoke with suggested that improved scaffolding could reduce the costs of producing cultured meat,36 and some people working in the stem cell industry have reported that optimizing scaffolding design made their process substantially more efficient (although we have not closely examined this claim).37 Developing improved scaffolding might involve varying the shapes and sizes of scaffolds and/or the scaffolding materials used, and testing which most improve cell culture growth rates (without introducing other issues). • Developing efficient methods of harvesting and assembling a large number of muscle cells from the scaffold/bioreactor into the final meat product; e.g., perhaps design scaffolding that dissolves or is edible and can be incorporated into part of the product. We are not aware of any existing process for this and have a limited understanding of what challenges might be involved. • Methods to maintain sterility at lower cost. Preventing or minimizing microbial contamination is important because other organisms like bacteria grow faster than mammalian cells and could quickly consume a bioreactor, ruining an expensive batch of muscle cells. Maintaining sterile manufacturing conditions is expensive because expensive fans and filters are needed to keep contaminants out.38 The current tissue engineering industry is already able to create sterile conditions, however maintaining sterility contributes significantly to the high cost of production.39 We have a limited understanding of potential methods for reducing these costs. • Optimization of bioreactor operating conditions. Cells are grown in vessels called “bioreactors,” which control temperature, replenishment, and circulation of the cell culture media, oxygenation, and other conditions that promote cell growth. According to Mark Post, cultured meat has only been grown in small bioreactors (~5L) so far, but will need to be grown in much larger bioreactors if production will be scaled up (~25,000L). Bioreactors will need to be optimized for this larger scale of production.40 Industrial scale-up is a non-trivial engineering challenge requiring significant resources, because the systems are expensive to build and test. However, the biotech and tissue engineering industry are currently working on bioreactor designs and scale-up that may be relatively transferrable to cultured meat production. Our understanding is that scaling-up is one of the most common modes of failure in industries such as synthetic biology which also involve large scale bioreactors.41 Our understanding is that this work would involve designing new bioreactors and testing them for improvements in metrics like cell growth rates and contamination rates.42 #### 2.2.4 Issues associated with taste and consumer acceptance People who tasted Mark Post’s burger said it tasted “close to meat,” but that it was “not that juicy,” had a somewhat unusual texture, and that the lack of fat was noticeable.43 In addition, cultured meat is currently missing some cell types and chemicals found in conventional meat, including blood and connective tissue.44 We would guess that it may be necessary to find the appropriate ratios and arrangements of these ingredients in order to replicate the texture and flavor of natural meat. Potential paths to improving the taste and texture of cultured meat include: • Trying different scaffold designs and media formulations, and testing the taste and texture of the results. • Culturing fat cells and experimenting with incorporating them into the meat. Mark Post is currently working to improve the quality of the cultured meat by adding fat into the burger.45 • Adding flavor enhancers, an approach which is likely already widespread in both ground and whole slab conventional meat production as evidenced by the USDA regulations on labeling of flavor enhancers in meat and poultry.46 Other approaches (which we have considered less and may be more speculative and challenging) include:47 • Experimenting with different ratios and arrangements of connective tissue (e.g., fat, collagen), muscle cells, and blood. #### 2.2.5 Possible approaches that are not, to our knowledge, being pursued within cultured meat research We are not aware of anyone pursuing the following approaches to cultured meat: • Hybrid plant-based and animal-cell-based approaches: These approaches would involve mixing plant-based and animal-cell-based food, and could potentially offer both better flavor/texture than pure plant-based approaches and lower cost than pure animal-based approaches. Plant-based approaches (e.g., Hampton Creek for egg products) are typically significantly cheaper than animal-based food products; however, the challenge has been in producing a plant-based product that tastes like it came from an animal.48 Developers could explore the minimum amount of animal-based product needed to provide a realistic meat-like flavor and texture, when combined with plant-based ingredients. • GMO approaches for cultured meat. Mark Post and Modern Meadow are not pursuing genetically modified approaches to cultured meat.49 According to Mark Post, his decision on this is driven by concerns about lower market adoption.50 However, we would guess that allowing genetic modifications of the muscle cells could help make manufacturing scale-up of cell culture more efficient, as it has in synthetic biological fermentation. For example, we would guess that GMO muscle cells might be made to reproduce faster, significantly reducing manufacturing times (and thus costs). #### 2.2.6 Will it be possible to make cultured meat cost-competitive with conventional meat? Some estimates of the possible future costs of cultured meat are below: ESTIMATE BY: COST OF CULTURED MEAT (USD) ASSUMED MANUFACTURING VOLUME YEAR Vandenburgh$5M / kg51 Small-scale production in laboratories52 200453
Exmoor €3300 – 3500 / tonne

(€3.3 – 3.5 / kg)

Scaled-up to large volume 2008
Van der Weele and Tramper €391 / kg assuming typical media cost of €50,000 / m³. One estimate of the lowest possible cost of media is €1,000 per m³, but we do not know what this estimate is based on. Plugging this assumption into Van der Weele’s model would imply that €8 of media is needed for 1 kg of meat.54 Scaled-up to large volume55 2014

The Exmoor estimate seems very optimistic to us. For example it assumed $0 cost for growth factors in media,56 but it could be one of the more expensive components.57 Note that the cost of production of beef at the time was ~€3600/tonne. 58 The study notes that cell culture media cost €7000-8000/tonne, and needed to reach < €350/ tonne for commercial viability.59 Isha Datar, now the Executive Director of New Harvest, said this estimate was preliminary and “could be largely inaccurate.”60 We are highly uncertain about the eventual cost per kg of cultured meat, and have not closely examined the above cost estimates. However, none of these estimates suggest a cost competitive with that of conventional meat. The Van der Weele and Tramper estimate is based on back of the envelope calculations by academic researchers. They argue that even if we reduce the price of animal-free medium to what they believe is its lowest possible cost—€1 per liter—it will be insufficient to make cultured meat cost-competitive with conventional meat.61 Van de Weele and Tramper do not explain why €1/L is the lowest possible cost of animal-free medium, but their claim that a cost of less than €1/L is required seems correct to us because meat costs a few dollars per pound, and we would guess that a minimum of a few liters of serum-free media would be required to produce 1 kg of meat. For reasons explained below, it seems to us that it will be extremely challenging to get the cost of animal-free media this low. Currently, a major cost of animal-free media is the cytokines (cell-signaling molecules) that encourage cell proliferation.62 Two approaches to reducing the cost of cytokines include: • Trying to find small molecule replacements for cytokines. While inexpensive small molecule replacements for some cytokines have been discovered, we were told that it has proven challenging to find replacements for others, and we would need to replace all the expensive cytokines used in order to get costs below €1/L. We are not aware of how many cytokines would need to be replaced, but would guess that manufacturers of animal-free media could answer this question. • Trying to culture genetically engineered yeast cells (or other host organisms) to produce cytokines through fermentation. We have a limited understanding of this approach. Steve Oh, a scientist who has worked on decreasing the cost of animal-free media, told us that it would be extremely challenging to get the cost of animal-free media below ~€1/L through either of these approaches.63 Even if the cytokines could be produced at zero cost, we would guess that that would not suffice to bring the cost of animal-free media below €1/L because, as of 2006, “basal medium,” a sort of medium without cytokines, still cost$1-4/L.64 We have not closely investigated the feasibility of decreasing the cost of the basal medium, but it may be challenging because many of its ingredients sound like commodities (based on a first-glance review of the list of major ingredients).65

#### 2.2.7 How long will it be before it is possible to make cultured meat cost- and taste-competitive with conventional meat?

We have seen only a couple of informed estimates of how long it might take to make cultured meat cost- and taste-competitive with conventional meat. Mark Post estimated 7-10 years,66 and a scientist in the tissue engineering field said that cost-competitive cultured meat would be very unlikely to be available in the next 10-15 years, absent a major technological breakthrough.67 We have a very limited understanding of what these estimates are based on.

We are highly uncertain on this topic because we believe there is essentially no industrial data around cost of scaling up cell production to these levels. Many non-trivial unknowns may exist associated with a new endeavor like this, making it very challenging to accurately predict the costs. For example, synthetic biology company Amyris (described more below) took longer to scale up biofuel production than anticipated.68

One way to consider likely future cost reductions is to compare cultured meat with progress in the closest industry analogues we’ve been able to identify in tissue engineering and synthetic biology:

• Organogenesis (a tissue engineering company) was founded in 1985 and had its first skin graft product, Apligraf, approved in 1998.69 Their product, Apligraf, is a wound care patch (75 mm diameter circular disc that is 0.75 mm thick), is functionally similar to skin, and is used to cover wounds and speed up the healing process in patients with certain leg and foot ulcers.70 Manufacturing Apligraf is similar to manufacturing cultured meat because it involves culturing multiple cell types and combining them with collagen and other chemicals into a tissue substitute.71 Unlike cultured meat, there is much lower cost pressure on medical products, and Apligraf had to pass FDA regulatory approval. We have a limited understanding of the state of their program in 1985, but it was early-stage academic research, and the early years of the company were heavily research-based.72 The product volume required for that product is much lower than for meat (making for a less challenging scale-up problem) and the cost pressure is much lower (they can charge a premium for high-value medical product, rather than competing with a commodity like conventional meat). Since then, they have further reduced the real cost of their skin grafts by roughly a factor of three.73 Based on rough back of the envelope calculations, we would guess that the manufacturing cost of Apligraf is on the order $90,000/kg.74 • The synthetic biofuel company Amyris was founded in 200375 and started fuel production in Dec 2012.76 We would guess that at founding, Amyris was mainly based on early stage academic work and had done little work in manufacturing scale-up.77 Moreover, we would guess that biofuel may be a simpler product than cultured meat because the final product, a liquid biofuel, does not require cells or three-dimensional tissues.78 Despite$700M investment in the company, Amyris has not been able to compete on cost with conventional fuels.79

Judging on the basis of the above two examples, the challenges involved in dramatically reducing the cost of animal-free media, and our holistic assessment of the challenges involved in reducing the cost of cultured meat, discussion with scientists who have experience with cell cultures and tissue engineering, we currently see developing cost-competitive cultured meat products as extremely challenging, and we have been unable to find any concrete paths forward that seem likely to achieve that goal.

#### 2.3 Cultured slab meat

We investigated cultured slab meat less closely because it seems more challenging and we are not aware of anyone who is working on it now, except for a group doing a feasibility study on chicken (see below).

Slab meat poses additional challenges because it would likely require cell cultures with multiple types of tissues that grow in appropriate complex formations.80 In addition to the interventions listed above—which would largely help with the development of cultured slab meat as well—a philanthropist interested in accelerating the development of cultured slab meat could support:

• The development of scaffold/bioreactor designs able to feed and support thick three-dimensional structures of cells. Thicker multi-layer tissue structures may require vascular systems to deliver nutrients deep within the tissue.81
• Feasibility analysis of slab meat, e.g., as Amit Gefen is doing for chicken.
• There may also be approaches involving self-assembly and 3D-printing, however we have not closely considered these possibilities and would guess that they would be difficult to implement in the near future.82

We would guess that developing cost-competitive cultured slab meat is a challenge in the same rough ballpark of difficulty as developing transplant organs in vitro because both would likely require growing large, complex, multi-layered 3D structures and vascular systems. We would guess that cultured slab meat would be technically easier to achieve without cost constraints (since it would not be necessary to fully replicate as many natural functions), but also that price pressure on in vitro organs would be much lower.

#### 2.4.1 Potential impact of taste- and cost-competitive cultured egg whites

In 1996, approximately 12% of the total eggs produced in the U.S. went towards making processed egg whites,83 out of the 65 billion eggs produced in the U.S. overall.84 In 2014, the number of eggs produced in the U.S. had risen to 100 billion.85 We would guess that the development of cultured egg whites that were cost- and taste-competitive with traditional egg whites would significantly reduce the amount of eggs produced.

#### 2.4.2 What methodology is currently used to make cultured egg whites?

We are only aware of one company developing cultured egg whites: Clara Foods. Clara Foods is a biotechnology company founded in 2015 working on developing cultured egg whites. It is still in development stages. Its process involves:

• Putting genes for chicken egg white proteins into yeast cells using genetic engineering techniques.
• Growing the yeast cells that produce the egg white proteins in a fermentation process.
• Letting the yeast produce the egg white proteins.
• Separating the egg white proteins from the yeast through a purification process.86

Clara Foods is working on protein purification methods to efficiently extract the egg white proteins from the “soup” of yeast fermentation.87 Their founders have created a prototype to demonstrate that only a subset of egg white proteins are needed to functionally substitute for egg whites. To make the prototype, they extracted a subset of proteins from real egg whites and showed that it was possible to make meringues using this subset of proteins. All of these steps are aimed at making the cultured egg whites cost competitive with conventional egg whites.88

#### 2.4.3 Will it be possible to make cultured egg whites cost-competitive with conventional egg whites?

Our understanding is that the main hurdle for cultured egg whites is in achieving scaled-up manufacturing that is cost-competitive with conventionally farmed eggs. We would guess that it will be significantly easier to meet this challenge with cultured egg whites than with cultured meat because:

• Cultured egg white protein production is more similar to already demonstrated industrial biotech processes (e.g., synthetic patchouli).89
• Rather than producing tissues (which are arrangements of entire cells), producing egg whites likely only requires manufacturing a small number of proteins (which is much simpler).90

#### 2.5 Plant based alternatives

We did not closely investigate this topic because our impression is that work in this area is well-funded. See below.

## 3. Who else is working on this?

#### Key Companies Developing Alternatives to Animal-Based Foods

COMPANY PRODUCTS IN DEVELOPMENT CELLS / GENES FROM ANIMALS? MAINLY PLANT BASED DIRECT SUBSTITUTE FOR CONVENTIONAL ANIMAL PRODUCTS FUNDING
Modern Meadow Leather, possibly steak chips91 Yes (cells)92 No No $10.4M93 Breakout Labs,94 ARTIS Ventures, Francoise Marga, Healthy Ventures, Horizons Ventures, Interplay Ventures (Partner: Mark Peter Davis), Karoly Jakab, Iconiq Capital, Sequoia Capital95 Clara Foods Egg white produced by yeast via synthetic biology96 Yes (genes)97 No98 Yes99$1.7M100
David Friedberg, Gary Hirshberg, Ali and Hadi Partovi, Scott Banister, SOSventures101
Impossible Foods Plant-based meat and cheese alternatives using bioengineering techniques to combine select plant proteins and molecules to replicate the experience of animal-based foods.102 No Yes103 Yes104 $183M Bill Gates, Horizons Ventures, Jung-Ju (Jay) Kim, Khosla Ventures, UBS, Viking Global Investors105 Beyond Meat Extruded plant-based chicken substitute106 No107 Yes108 Yes109 Undisclosed amount, completed Series D in 2014. Bill Gates, DNS Capital, Kleiner Perkins Caulfield & Byers, Obvious Ventures, S2G Ventures110 Muufri Milk produced by yeast via synthetic biology111 Yes (genes)112 Some plant-based fats113 Yes114$2M + $30k SOS Ventures, Horizon Ventures115 Hampton Creek Plant-based versions of products containing eggs, such as mayonnaise, cookie dough, scrambled egg mixture (in development)116 No117 Yes118 Yes119$120M
Ali and Hadi Partovi, Ash Patel, Brian Meehan, OS Fund, Collaborative Fund, Demis Hassabis, Eduardo Saverin, Far East Organization, Founders Fund, Horizon Ventures, Jean Piggozzi, Jerry Yang, AME Cloud Ventures, Jessica Powell, Kat Taylor, Khosla Ventures, Marc Benioff, Mustafa Suleyman, Scott Bannister, Tao Capital Parters, Tom Steyers’ Eagle Cliff, Uni-President Enterprises Corporation, Velos Partners, WP Global Partners120, Bill Gates 121

#### 3.2 Academic researchers and non-profit organizations

Key Academic Groups Developing Alternatives to Animal-based Foods

RESEARCHER WORK APPROACH DEVELOPING MEAT/EGG ALTERNATIVE?
Mark Post Cheaper animal-free cell culture media, adding fat to cultured meat to improve taste122 Cultured animal tissue123 Yes124
Nicholas Genovese Establishing new lab in stem cell biology for meat production125 Developing stem cell lineages for cultured meat126 Yes127
Amit Gefen Feasibility study on making whole cultured chicken meat Cultured animal tissue128 Feasibility study129

#### 3.3 Potential gaps in the field

Plant-based alternatives appear relatively well-funded (see ‘Key Companies Developing Alternatives to Animal-Based Foods’).

We see cultured meat and cultured eggs as the largest potential gaps in the field. To our knowledge, Clara Foods is the only organization working on a cultured egg product (egg whites specifically), and it has received about $1.7M in funding since they were founded in 2015 (see Key Companies Developing Alternatives to Animal-Based Foods’). Cultured meat also receives a limited amount of funding (we estimate <$6M over the last 15 years),130 though it has a longer history and a larger number of people involved. We discuss it in more detail in the following section.

#### 3.4.1 Current activity

We are aware of only one company working on cultured meat today: Modern Meadow. As noted in the table ‘Key Companies Developing Alternatives to Animal-Based Foods’, they are currently focused on leather and possibly steak chips, which would not translate immediately into alternatives to meat. However, we would guess that there could be substantial overlap in research and development pathways (such as developing less expensive animal-free media, finding low-cost methods of maintaining sterility, and optimizing bioreactor operating conditions) but perhaps less overlap related to scaffolding.

There are a few academic groups worldwide working on cultured meat,131 but our understanding is that very limited grant funding is available for cultured meat132 and others we spoke with suggested that the work is considered fringe by the academic community.133 Mark Post has talked about how his family was disappointed when he switched his research to cultured meat.134 Among the limited amount of work on cultured meat, beef is receiving more attention than chicken or pork as seen in the tables ‘Key Companies Developing Alternatives to Animal-Based Foods’ and ‘Key Academic Groups Developing Alternatives to Animal-based Foods’.

New Harvest is a small non-profit that makes seed grants to entrepreneurs and academics working on cultured animal products.135 In 2013, their annual expenditures were about $60,000.136 Mark Post plans to start a company focused on commercial production of cultured meat, in hopes of replacing conventional ground beef.137 #### 3.4.2 Past activity There has been work on cultured meat at least since Willem van Eelen began promoting research efforts in the late 90s.138. Other past projects include: • In 2002, NASA funded Prof. Morris Benjaminson of Touro College to culture goldfish muscle tissue.139 This was very preliminary work – he took samples from animals and showed that their size increased after growing in a dish, in order to investigate the potential of culturing meat for human consumption in space.140 • As an art project in 2003, Oron Catts cultured frog muscle cells into a small “steak” which was eaten in public (four of eight tasters spat it out).141 • Vladimir Mironov performed research on cultured meat (although his U.S. lab closed) and collaborated with Nicholas Genovese who is continuing work on cultured meat.142 • Mark Post began work on cultured meat in 2008. His work stems academically from early cultured meat proponent Willem van Eelen.143 • In 2010, a startup company called Mokshagundum Biotechnologies was interested in making genetically modified meat by harnessing tumor-growth-promoting gene, but we found no record of this company since then.144 • In 2013, Singularity University produced a startup team called LifeStock aimed at producing animal-free meat, specifically focused on the scaffolding component.145 However, we have not found recent activity (2014 and onwards) mentioning LifeStock, and the website listed on their Facebook page does not appear to be active.146 There have been at least four symposia/workshops on cultured meat since 2008: • The first cultured meat symposium was held in 2008 in Norway.147 • In 2011, there was a workshop by the European Science Foundation held in Gothenburg, Sweden, with 25 attendees.148 • In 2012, a panel on tissue engineered nutrition was held at the TERMIS (Tissue Engineering and Regenerative Medicine Int’l Society) conference in Vienna.149 • In May 2015, New Harvest and the IndieBio accelerator hosted an event in San Francisco called “Edible Bioeconomy”, convening those involved with the development and promotion of animal product alternatives.150 • There is also a symposium planned for the fall of 2015, the “First International Symposium on Cultured Meat”, focused on discussing tissue engineering for cultured meat, which is organized in association with Maastricht University and New Harvest.151 #### 3.4.3 Related work in tissue engineering Over$4.5B of capital investment went toward tissue engineering efforts worldwide between 1990 and 2002, and over 90% of the investment came from the private sector.152 Tissue engineering efforts focus on many of the same challenges facing cultured meat: developing scaffolds,153 scale-up, and tools to accelerate the research.154 However, we would guess that some of the challenges associated with making cultured meat cost- and taste-competitive with cultured meat are unlikely to be addressed by people working in tissue engineering unless they are done specifically for manufacturing cultured meat. Two such examples include:

• Creating extremely low-cost animal-free media: Tissue engineers in the medical field already have incentives to use animal-free media because it improves reproducibility.155 Our understanding is that cell media is a significant cost for tissue engineering companies,156 but because the prices of tissue engineering products are much higher than the price of meat,157 we would guess that tissue engineering companies may not be incentivized to lower costs of animal-free media to the levels that would be required for cultured meat to be commercially viable.
• Scaling up production of cultured meat to vast quantities: The volume of tissue production required for biomedical applications (such as skin grafts and organ replacements) is significantly lower than the volume required for meat.158 We would therefore guess that they will not need to develop as large-scale production facilities as may be required for mass-produced cultured meat in the near future.

#### 3.4.4 Availability of funding from for-profit investors

Our impression is that venture capitalists generally seek investments that could provide a significant return within several years.159 However, as discussed under above, we would guess that it will take longer than that for cultured meat to become cost- and taste-competitive with natural meat. One potential goal of philanthropic funding would be to allow work to progress without expectations of a near-term return on investment in hopes of de-risking cultured meat to the point where a profit-motivated investor would be willing to finance further development.

## 4. Our process

We initially decided to investigate the cause of alternatives to animal-based foods because:

• We believe that, for someone who cares about the welfare of farm animals, the treatment of animals in industrial agriculture is an extremely important problem. For more detail, see our overview of industrial animal agriculture.
• It seems to us that if there were plant-based or cultured alternatives to meat and eggs that were cost- and taste-competitive with animal-based foods, it could greatly reduce the amount of meat and eggs produced, and thereby greatly reduce pain and suffering of animals in industrial agriculture.
• The area seemed unlikely to get financial support from traditional sources that support research and development in the life sciences.

We spoke with 8 individuals with knowledge of the field, including:

• Isha Datar, Executive Director, New Harvest
• Mark Post, Professor of Vascular Physiology and Chair of Physiology, Maastricht University
• A scientist with 18 years experience in the tissue engineering industry
• Five individuals who spoke with us off-the-record.

In addition to these conversations, we also reviewed documents that were shared with us and had some additional informal conversations.

## 5. Questions for further investigation

We have not deeply explored this field, and many important questions remain unanswered by our investigation.

Among other topics, our further research on this cause might address:

• How quickly and in what directions are commercial and academic tissue engineering advancing? How transferable might this work be to cultured meat?
• If funded, what is the potential impact of additional academic research in cultured meat?
• Is it possible to make a realistic cost model in which cultured meat/egg products are cost-competitive with traditional meat/eggs? What advances would be needed to put such a model into practice?
• How long will it take to develop cost- and taste-competitive cultured meat/eggs?
• How could additional funding advance the development of cultured egg whites?
• What could be done to advance the development of cultured slab meat? How much more challenging would it be to develop cultured slab meat in comparison with cultured ground meat?
• Would the public buy cultured animal products?

## 6. Sources

DOCUMENT SOURCE
2014 Consumer Perishables Databook Source (archive)
Amyris company history Source (archive)
Angel.co company profile, Hampton Creek Source (archive)
Angel.co company profile, Muufri Source (archive)
Apligraf website, How is it Made Source (archive)
Apligraf Medicare Product and Related Procedure Payment, 2015 Source (archive)
Apligraf website, What is Apligraf Source (archive)
Amyris 2013, Amyris and Total Announce Successful Demonstration Flight With Renewable Jet Fuel During Paris Air Show Source (archive)
BBC 2011, Grow your own meat Source (archive)
Benjaminson, Gilchriest, and Lorenz 2002 Source (archive)
Biofuels Digest 2011, Amyris opens first commercial facility Source (archive)
Bloomberg 2013, Modern Meadow Makes Leather and Meat Without Killing Animals Source (archive)
Breakout Labs portfolio company page, Modern Meadow Source (archive)
BusinessWire 2013, Beyond Meat Completes Largest Financing Round to Date Source (archive)
Category share of beef sales in the United States in 2014, by cut type Source (archive)
Category share of chicken sales in the United States in 2014, by cut type Source (archive)
CBInsights company profile, Modern Meadow Source (archive)
CNN 2009, Lab meat Source (archive)
Crunchbase company profile, Hampton Creek Source (archive)
Crunchbase company profile, Impossible Foods Source (archive)
Crunchbase company profile, Impossible Foods Investors Source (archive)
Crunchbase company profile, Modern Meadow Source (archive)
Danoviz and Yablonka-Reuveni 2012, Skeletal Muscle Satellite Cells: Background and Methods for Isolation and Analysis in a Primary Culture System Source (archive)
Datar and Betti 2010, Possibilities for an in vitro meat production system Source (archive)
DFJ Portfolio Source (archive)
Discover Magazine 2012, Steak of the Art: The Fatal Flaws of In Vitro Meat Source (archive)
Economist 2011. Global Livestock Counts Source (archive)
Edelman et al. 2004, In vitro cultured meat production Source (archive)
Exmoor In Vitro Meat Consortium Preliminary Economics Study, March 2008 Source (archive)
Fast Company 2012, THE RISE AND FALL OF THE COMPANY THAT WAS GOING TO HAVE US ALL USING BIOFUELS Source (archive)
Fastcoexist 2013, Lifestock: The Newest Player In The Growing Lab-Made-Meat Industry Source (archive)
First International Symposium on Cultured Meat, 2015 Source (archive)
Food Safety News 2012, Lab-grown meat? Source (archive)
FoodProcessing.Com.Au 2015, Producing Egg Whites Without Chickens Source (archive)
Forbes 2014, Bill Gates-Backed Food Startup Hampton Creek Source
Forbes 2014, Yum! Petri-Dish Beef Is like “Jerky Melting in Your Mouth” Source
Galef 2011 Source (archive)
Genetic Engineering & Biotechnology News 2014, Driving Down the Cost of Stem Cell Manufacturing Source (archive)
Genetic Engineering & Biotechnology News 2015, Biopharma Demand Is Driving the Cell Culture Market Source (archive)
GiveWell’s non-verbatim summary of a conversation with a scientist with 18 years experience in the tissue engineering industry, April 23, 2015 Source
GiveWell’s non-verbatim summary of a conversation with Isha Datar, March 10, 2015 and July 24, 2015 Source
GiveWell’s non-verbatim summary of a conversation with Mark Post, March 24, 2015 Source
GiveWell’s non-verbatim summary of a conversation with Steve Oh, October 7, 2015 Source
Gizomodo 2011, The first lab-grown hamburger will cost $345,000 Source (archive) Guardian 2012, Fake meat Source (archive) Heartblog on growth factors Source (archive) Hochfeld 2006 Source Hunsberger et al. 2015, Manufacturing Road Map for Tissue Engineering and Regenerative Medicine Technologies Source (archive) i24news 2015, Israeli researchers developing first lab-grown chicken Source (archive) IEET 2014, Interview with Nicholas Genovese Source (archive) Impossible Foods website Source (archive) Independent 2013, Lab meat taste test Source (archive) Jacklenec et al 2012, Progress in the Tissue Engineering and Stem Cell Industry Source (archive) Jochems et al., The use of fetal bovine serum: ethical or scientific problem? Source (archive) Lees 2008, The taste of test tube meat Source (archive) Lifestock Facebook Page Source (archive) Maastricht University, Cultured Beef Project Fact Sheet Source (archive) Maastricht University, Mark Post bio Source (archive) MacKay 2006, Bioactive wound healing, bioaesthetics and biosurgery Source (archive) MIT Technology Review 2014 on Modern Meadow Source Modern Agriculture Foundation Press Release, Jan 2015 Source (archive) Modern Meadow FAQ Source (archive) National Chicken Council, Per Capita Consumption of Poultry and Livestock Source (archive) National Geographic 2014, Meat grown in a lab Source (archive) Nature Bioentrepreneur 2005, Commercializing synthetic biology Source (archive) Nature News 2010, Food: A taste of things to come? Source (archive) Nature News 2011, Meat-growing researcher suspended Source (archive) NBC News 2013 on Brin and Post Source (archive) NBC News 2013, It’s (not) alive! Franken-meat lurches from the lab to the frying pan Source New Harvest 2014, World’s First In Vitro Meat Symposium Source (archive) New Harvest 2015, Edible Bioeconomy Event–Recap Source (archive) New Harvest 2015, Willem Van Eelen obituary Source (archive) New Harvest Expenditures Source (archive) New Harvest Profile of Marianne Ellis Source (archive) New York Times 2013, Building a$325,000 Burger Source (archive)
New Yorker 2011, Test-tube burgers Source (archive)
Oklahoma Food Safety Division, How Much Meat? Source (archive)
Organogenesis company profile Source (archive)
Pavillon 35, In vitro meat as art Source (archive)
Post 2012, Cultured meat from stem cells Source (archive)
Post 2014, An alternative animal protein source: cultured beef Source (archive)
Post and Courier 2011 on lab meat Source (archive)
Q+A with Modern Meadow CEO, Andras Forgacs Source (archive)
SBIR 2012 grant, Engineered Comestible Meat Source (archive)
Schwartz 2015, NT Cattlemen’s Association to hear that ‘cultured meat’ could end their industry within decades Source (archive)
Science Daily 2011, Growing meat in the lab Source (archive)
Science Insider 2014, PETA Abandons $1 Million Prize for Artificial Chicken Source (archive) Scientific American 2013, Test-Tube Burger Source (archive) Techcrunch 2015, Clara Foods Cooks Up$1.7M in Funding Source (archive)
Technologist 2014, From stem cells to Big Macs Source (archive)
Technology Review 2012, Why Amyris is Focusing on Moisturizers, Not Fuel, for Now Source
The Guardian 2014, Would you eat lab grown meat to save the environment? Source (archive)
Time 2014, The Surprising Reason ‘Pink Slime’ Meat is Back Source (archive)
U.S. Poultry & Egg Association 2015, economic data Source (archive)
USDA Grant for Engineered Comestible Meat Source (archive)
USDA, Egg Products Processing and Distribution Module Source (archive)
USDA, Natural flavorings on meat and poultry labels Source (archive)
USG 2007, Advancing Tissue Science and Engineering Source (archive)
Van der Weele and Tramper 2014, Cultured Meat: every village its own factory? Source (archive)
VentureBeat 2014, Hampton Creek’s data scientists team up with chefs to find the holy grail of plant proteins Source (archive)
Verge 2013, Lab meat Source (archive)
Wall Street Journal 2013 on Steve Jurvetson Source (archive)
Wall Street Journal 2014, The Secret of These New Veggie Burgers: Plant Blood Source (archive)
Walpole et al. 2012, The weight of nations: an estimation of adult human biomass Source (archive)
Washington Post 2014, Can this company do better than the egg? Source
Washington Post 2014, Man-made cow’s milk Source
Wikipedia Growth Medium page 2015 Source (archive)
Wired 2013, Alton Brown on the End of Meat as We Know It Source (archive)
Wired 2014, Forget GMOs. The Future of Food Is Data—Mountains of It Source (archive)

The CrunchBase pages we archived are covered under the Creative Commons license.

## Macroeconomic Policy

This is a writeup of a medium investigation, a brief look at an area that we use to decide how to prioritize further research.

## In a nutshell

#### What is the problem?

The recent Great Recession points to the large humanitarian costs of business cycle downswings. Going forward, it seems reasonable to expect recessions to cost the global economy an average of hundreds of billions of dollars a year in lost output due to idle capacity. To the extent that better stabilization policy is possible, it could carry large humanitarian benefits.

#### Who already works on this issue?

Central banks, including the U.S. Federal Reserve, are generally the most active players, as both economic policymakers and researchers. Academic economists outside of central banks do a significant amount of research on policy-relevant questions, and there is some philanthropic support for research. A few think tanks and international organizations also work on the topic.

#### What could a new philanthropist support?

A philanthropist could focus on advocacy or research. Advocacy work might involve supporting think tanks, educating the public, or building interest group coalitions. There seem to be a number of unresolved research questions of substantial policy relevance, and a funder aiming to facilitate progress on them could pursue any of a number of approaches.

## 1. What is the problem?

Economic recessions can carry enormous humanitarian costs. For instance, the 2008 financial crisis and the associated Great Recession appear to have:

• Cost the US economy roughly ten trillion dollars, and the rest of the global economy a comparable amount, in lost output due to idle capacity.1
• Reduced potential output over the long run by forcing some people out of the workforce, reducing investment, and delaying productivity improvements.2
• Caused millions of people to suffer bouts of long-term unemployment, which may carry significant psychosocial and health costs and result in permanent income losses.3

Despite these harms, the impacts of the recession could plausibly have been much worse had the response of policymakers been less aggressive.4

The 2008 financial crisis prompted an unusually deep recession, and it is difficult to assess the likelihood of similarly deep recessions in the future. A very rough approximation, based only on the occurrence of the Great Depression and the Great Recession, might be to expect such events twice per century, which would suggest that recessions carry annual global costs in the hundreds of billions of dollars range.5 Accordingly, the possibility of reducing the frequency or depth of recessions would carry significant humanitarian value, though the extent to which such reductions are possible or desirable is disputed amongst economists.6

It is not clear to what extent we should expect certain features of the recent recession to recur in the medium to long term. The relatively stable economic conditions observed in the U.S. during the decades prior to the recession may have prompted an underestimate of the magnitude of variability in the economy – deep recessions may be more likely than the recent experience prior to the Great Recession would suggest.7 Real interest rates also appear to have been declining throughout the developed world for much of the last thirty years,8 increasing the probability that economies will reach the “zero lower bound” on nominal interest rates, as they have recently, with greater frequency in the coming years. (The “zero lower bound” presents a particularly challenging situation for monetary policy, discussed more below.)

Note: this page focuses on macroeconomic issues related to business cycle stabilization. We may investigate other macroeconomic policy issues separately at a later date.

## 2. Who already works on this issue?

The Federal Reserve (“the Fed”), the United States’ central bank, plays an enormous role in U.S. macroeconomic policy and research.9 Its “dual mandate” is to promote maximum employment and stable prices.10 During recessions, the Fed typically uses monetary policy tools, e.g. lowering short term interest rates, to attempt to return the economy to full employment. Recently, having hit the “zero lower bound” on short term nominal interest rates, the Fed has been pursuing “unconventional policies” including large scale asset purchases (“quantitative easing”) and forward guidance about the future trajectory of interest rates that aim to stimulate the economy using other channels. Economists are not in full agreement about the likely impacts of these policies.11

It is difficult to accurately estimate the resources that the Fed devotes to monetary policy research, but the Board of Governors budgeted $64 million for “research and statistics” in 2013, while the twelve regional banks budgeted a total of$602 million for “monetary and economic policy” in 2013.12

There are also hundreds or thousands of economists, primarily based in universities, who conduct research on macroeconomics, though they do not necessarily produce research that aims to influence policy.13 Some people we spoke with noted that the outsize influence of the Fed extends to academic macroeconomists, who tend to be hesitant to criticize it too stringently for fear of alienating friends or harming their careers.14 Accordingly, it may not be correct to view academic economists as a fully independent check on the research and decisions of the Fed.

There is relatively little philanthropic engagement in macroeconomic research and policy.15 Funders of macroeconomic research or advocacy include:16

• The National Science Foundation (a U.S. government institution), which spends roughly $40 million a year on grants for economic research (most of which is not for macroeconomics).17 • The Institute for New Economic Thinking (INET), which was initially funded by a$50 million, 10-year grant from George Soros.18
• The Washington Center for Equitable Growth, which seems to focus primarily on inequality and growth rather than fiscal or monetary policy, expects to fund a few million dollars a year of research.19
• The Russell Sage Foundation, which spent $4 million on a program (now ended) on the impacts of the Great Recession.20 • The Alfred P. Sloan Foundation, which makes grants to support research on “economic institutions, behavior, and performance” including the economic implications of the Great Recession.21 • The Peter G. Peterson Foundation, which spent$12.8 million in 2012, predominantly focuses on efforts to limit the federal budget deficit (as opposed to other macroeconomic research or policy focus areas, such as unemployment).22

Think tanks working on these issues include:23

• The Brookings Institution, which houses the new Hutchins Center on Fiscal and Monetary Policy and the Brookings Papers on Economic Activity.
• The Peterson Institute for International Economics (PIIE).
• The Center on Budget and Policy Priorities’ Full Employment project.

Our discussion has focused primarily on research and policy in the United States, but macroeconomic research and policy is highly globalized. A number of international organizations, like the International Monetary Fund, play important roles, as do central banks in other countries.

## 3. What could a new philanthropist support?

Philanthropic efforts to reduce the frequency or depth of recessions would likely have to aim to ultimately change (or prevent changes to) policy or the behavior of institutions such as the Federal Reserve. Toward this end, a philanthropist might support advocacy for particular policies and behaviors or research to help determine which policies and behaviors would be the best ones to adopt. Though the distinction between advocacy and research may not always be clear, we discuss them separately below.

A number of people who we spoke with noted that most advocacy on monetary policy tends to come from people who are skeptical of the Federal Reserve and want to focus on limiting inflation or return to the gold standard, and they argued that supporting groups that are more concerned about unemployment to engage in debates around monetary policy could be valuable.24 An example of such an opportunity might be to support more liberal/progressive think tanks to be more involved in debates over monetary policy.25 One risk from supporting progressive advocates on monetary policy is that they might continue to advocate for looser monetary policy even when it was appropriate for the Fed to tighten, potentially leading to worse policy in the long term.26

• Attempts at improving public communication around macroeconomic policy in order to help people understand the issues and tradeoffs.
• Supporting think tanks or advocacy groups to work on proposals for other countercyclical policies, such as automatic aid to states during recessions or other automatic stabilizers.27
• Creating a “shadow” Federal Open Market Committee to offer informed commentary and alternative policy proposals after Federal Open Market Committee meetings.28
• Supporting advocates for stricter financial regulation, which may reduce the risk of recessions caused by financial crises like the one in 2008.
• Collecting and disseminating the consensus views of economists on macroeconomic policy questions, whether through surveys or credible nonpartisan research institutions.29
• Mobilizing business groups to convey points of economic consensus to a conservative audience.30
• Designing a model stimulus bill to have prepared in case of a future recession.
• Advocacy for more federal funding for macroeconomic research.

As with other advocacy efforts, it is difficult to predict what the likely impact of these efforts might be.

To pursue these strategies, a philanthropist might support think tanks like Brookings or PIIE, or more explicitly ideological groups like the Center on Budget and Policy Priorities’ Full Employment project, the Economic Policy Institute, or the Center for American Progress.31 We aren’t aware of many other organizations advocating on macroeconomic policy (while placing a strong emphasis on the importance of reducing unnecessary unemployment as opposed to focusing on controlling inflation).

#### 3.2.1 Strategies for supporting policy-relevant research

A philanthropist aiming to eventually improve macroeconomic policy by supporting research might pursue any of a variety of strategies:

• Conventional research grants, similar to those provided by the National Science Foundation to economists.32
• Supporting journals, such as the Brookings Papers on Economic Activity, which focus on policy-relevant questions and aim to have some influence.33
• Summer programs to bring young economists up to speed on the state of a field and encourage them to pursue research in it.34
• Offering economists training or support to communicate their work to a policy audience.35
• Increasing the representation of women and people of color in economics.36
• Supporting awards for policy-relevant macroeconomic research or public communication.
• A fellowship or sabbatical program to fund a semester or year of paid leave for young macroeconomists interested in policy-relevant research.
• Funding economics PhD graduate programs to train more students.
• Supporting a conference or an edited volume on a topic of particular interest.37

We expect that these different approaches to supporting research might have very different cost-effectiveness profiles, though we do not have much sense of which are likely to get the strongest returns.

It is generally quite difficult to determine how much impact funding for research has on research output. At the margin, we would guess that grants allow academics to devote more of their time to a research project, but a number of people we spoke with mentioned the possibility that grants end up “supporting” research that would have occurred anyway, which intuitively strikes us as plausible.38

#### 3.2.2 Important unresolved research questions

There appear to be a number of potentially important policy-relevant questions about macroeconomics that are, to the best of our knowledge, unresolved:

• Which securities are best to purchase under a quantitative easing policy? Is it better to simply purchase the securities or explicitly target longer-term interest rates?
• What specific short-run fiscal policies have the best tradeoff of economic impacts and political plausibility? What kinds of automatic stabilizers might be adopted to reduce the need for discretionary fiscal policy in the future?39
• Should the Fed adopt a higher inflation target?40
• Should the Fed target a price level or an inflation rate?
• Should the Fed target a Nominal Gross Domestic Product (NGDP) level?
• Should long-term debt contracts be structured in non-traditional way to reduce the likelihood of future financial crises?41
• To what extent do unconventional monetary policies (such as quantitative easing) increase risks of financial instability?
• In measuring economic slack, should policymakers focus on short term unemployment, long-term unemployment, or the labor force participation rate? To what extent are declines in the labor force participation rate cyclical or demographic? How do high unemployment rates affect wage growth for people who are employed?42
• What would be the costs and benefits of moving to a system of electronic money (which could conceptually overcome the zero lower bound problem), and how politically feasible might such a system be?

To the extent that there is a literature on these questions (which varies), it is often difficult to find credible, unbiased syntheses of the literature that are accessible to a non-economist reader. We are not aware of any institutions that regularly publish authoritative, unbiased systematic reviews of the economics literature on questions like these ones.

#### 3.2.3 Novel methodological approaches to macroeconomic research

Critics often claim that traditional macroeconomic research tools are not well-suited to reaching conclusive answers to such questions, and call for novel approaches.43 It is difficult to anticipate in advance what kinds of new approaches might be helpful, but an incomplete list might include:

• Novel data collection efforts, whether qualitative (such as Alan Blinder or Truman Bewley’s research on wage stickiness), quantitative (such as the Billion Prices Project), or historical (such as digitizing archival bankruptcy records).44
• Using large-scale multi-player online games to experiment with macroeconomic phenomena.45
• Using agent-based models to develop a more sophisticated understanding of the impact of various economic policies.46
• Incorporating approaches from other disciplines, such as economic history, into macroeconomic research.47

We do not have a strong sense of whether support for more conventional research approaches or these (or other) more radical new approaches are likely to be more valuable.

#### 3.2.4 Potential grantees

Potential grantees for a funder aiming to support research might include:48

• The National Bureau of Economic Research (NBER)
• The Brookings Institution, which houses the new Hutchins Center on Fiscal and Monetary Policy and the Brookings Papers on Economic Activity
• The Peterson Institute for International Economics (PIIE)
• Academic centers that conduct research on fiscal or monetary policy
• Individual academics or research projects

We do not have a strong sense of which of these potential grantees are likely to be most promising. Funding individual academics or research projects would presumably give a funder more control but would also require more internal capacity.

## 4. Questions for further investigation

Our research in this area has been fairly limited, and many important questions remain unanswered by our investigation.

Amongst other topics, further research on this cause might address:

• What is the macroeconomic policy and research situation in other countries? What impacts do unconventional policies that may be beneficial for developed countries have on emerging market economies?
• How effective is marginal funding in producing policy-relevant economic research? Are there cases where important research projects have not occurred because of a lack of funding?
• What kind of advocacy efforts are most likely to be effective in promoting better policy?
• To what extent do economists agree about the appropriate macroeconomic policies to adopt? In areas of disagreement, is further research likely to result in a consensus? Do important unresolved questions represent a natural process of research progress or a more problematic shortcoming?
• How likely is the zero lower bound to be an ongoing problem? Is additional research over and above that likely to occur in the status quo necessary in order to develop optimal policy responses to the zero lower bound problem?
• To what extent are monetary policy disagreements partisan ones? To the extent that they are, how should we weigh intervening in them?

## 5. Our process

Our investigation of the field of macroeconomic research and advocacy has been relatively limited: we’ve followed a number of blogs on the topic, reviewed some of the academic literature, and spoken with people with knowledge of the field. Public notes are available from our conversations with:

• Joseph Gagnon, Senior Fellow, Peterson Institute for International Economics; former Associate Director at the Division of International Finance and Senior Economist, US Federal Reserve Board
• Mike Konczal, Fellow, Roosevelt Institute
• Josh Bivens, Research and Policy Director, Economic Policy Institute
• Robert Johnson, President, Institute for New Economic Thinking; Senior Fellow and Director of the Project on Global Finance, Roosevelt Institute; former Chief Economist, U.S. Senate Banking Committee
• Justin Wolfers, Professor of Economics and Public Policy, University of Michigan; Senior Fellow, The Brookings Institution
• Robert Bloomfield, Professor of Management and Accounting, Johnson Graduate School of Management, Cornell University
• Laurence Ball, Professor of Economics, Johns Hopkins University
• Scott Sumner, Professor of Economics, Bentley University

## 6. Sources

DOCUMENT SOURCE
Active NSF Economics Awards 3-26-14 Source (archive)
Amir Sufi explains how old consumer debt holds back today’s economy Source (archive)
Atkinson, Luttrell, and Rosenblum 2013 Source (archive)
Blanchard 1994 Source (archive)
Blinder 1994 Source (archive)
CBO 2013 Source (archive)
Economic Institutions, Behavior, and Performance Source (archive)
English, López-Salido, and Tetlow 2013 Source (archive)
Farmer and Foley 2009 Source
Federal Reserve Annual Report: Budget Review 2013 Source (archive)
Furman 2014 Source (archive)
GAO 2013 Source (archive)
Institute for New Economic Thinking Inaugural Grant Report 2010-2011 Source (archive)
International Monetary Fund 2014 Source (archive)
Krugman 2014 Source
Leonhardt 2013 Source
Notes from a conversation with Joe Gagnon on February 4, 2014 Source
Notes from a conversation with Josh Bivens on February 6, 2014 Source
Notes from a conversation with Justin Wolfers on February 26, 2014 Source
Notes from a conversation with Laurence Ball on April 17, 2014 Source
Notes from a conversation with Mike Konczal on January 23, 2014 Source
Notes from a conversation with Robert Bloomfield on April 4, 2014 Source
Notes from a conversation with Robert Johnson on February 18, 2014 Source
Peter G. Peterson Foundation 2012 Form 990 Source (archive)
Plumer 2012 Source (archive)
The Social and Economic Effects of the Great Recession: Recent Awards Source (archive)
Williams 2014 Source (archive)

## Criminal Justice Reform

Note: This writeup represents the state of our investigation into criminal justice reform as of May 2014, after some preliminary investigation. Since then, our views have changed noticeably, but they remain preliminary, and we have hired a Program Officer to lead our work in this cause going forward. If you have additional information on this cause that you feel we should consider, please feel free to get in touch.

## In a nutshell

#### What is the state of our investigation into U.S. criminal justice reform?

We have completed our medium-depth investigation of criminal justice reform.1 The investigation has continued to progress and we have made some grants in this area. However, we have not yet chosen criminal justice reform (or any other cause) as a long-term program area. In order to learn substantially more from this investigation we believe that we would need to commit to this cause for the medium-term (i.e., several years), so we have paused this investigation until we are ready to select U.S. public policy causes for that level of commitment.

#### Why are we making criminal justice reform grants?

Criminal justice reform seems like a particularly promising area because (a) we have identified a set of people who are promoting a particular viewpoint that is both appealing to us and appears underfunded, meaning that finding initial promising giving opportunities took less work in this area than we’d expect it would in many others; and (b) as evidenced by a wave of reform packages in over twenty states and the recent confluence of conservative and progressive interest in reform, this issue seems to strongly stand out from most political issues in terms of political momentum and tractability.

#### What is the problem?

The United States incarcerates a larger proportion of its residents than almost any other country in the world and still has the highest level of criminal homicide in the developed world.2 While confounding factors impede rigorous analysis of incarceration’s effect on the crime rate, it has been argued (and seems probably correct) to us that, while initial increases in the incarceration rate may have reduced crime, a rate this high does not have large benefits for public safety and represents indiscriminate incarceration of offenders whether or not prison is the appropriate punishment. In particular, it seems plausible that the United States incarcerates too many low-risk offenders for too long. Incarceration has large fiscal costs. We believe that it also has large human and economic costs. Community corrections (such as probation and parole), which are one of the main alternatives to incarceration, may also need improvement, as evidenced by the 40% revocation rate for persons on probation.3

#### What are possible interventions?

Proposals for criminal justice reform can broadly be divided into two categories. Front-end reforms affect individuals at their first point of contact with the criminal justice system. Back-end reforms affect individuals after they have already entered the criminal justice system.

Strategies to promote these reforms include policy research, legislative advocacy, technical assistance to policymakers or practitioners, litigation, communication and public education, direct services, and pilot projects.

Criminal justice reform is a broad field and we have so far focused on the particular set of interventions that initially got us interested in this cause — interventions that broadly seek to reduce incarceration while having neutral or even positive effects on public safety, by focusing prison beds on higher-risk offenders and improving community corrections (such as probation and parole) so that offenders for whom prison is not necessary are adequately supervised.

#### Who else is working on it?

Several foundations and the United States federal government are working on various aspects of criminal justice reform. Our perception is that organizations can roughly be divided into those that frame reform in terms of the rights of defendants and people in prison and organizations making a case that focuses on improving public safety and reducing public spending on incarceration by saving jail and prison beds for high risk offenders (and thus seeks to appeal to conservatives as well as progressives).

#### How much of an impact could we have?

We are very uncertain about how big of a humanitarian impact we could expect for a given level of funding for criminal justice reform. Based on claims in states that have already adopted reform packages, we would guess that reducing prison populations by about 10% while having a neutral or slightly positive impact on public safety could be a reasonable near-term goalpost. In the longer term, we’d guess that much larger reductions might be possible if community corrections are strengthened and the political dialogue around criminal justice policy continues to improve.

## 1. Why we are interested in criminal justice reform

We initially became interested in criminal justice reform because policy generalists we spoke with during our broad exploration of U.S. politics and policy advocacy indicated to us that it was an unusually promising — and in particular, unusually tractable — cause.4 Specifically, we heard from multiple sources that the combination of the adverse U.S. fiscal situation, declining crime rates, and emerging conservative interest in an issue historically supported by progressives may have created a “unique moment” for criminal justice reform with a limited window.5

Upon investigation, we learned that more than two dozen states, supported by the Pew Charitable Trusts’ Public Safety Performance Project (PSPP) and the federal government’s Bureau of Justice Assistance, have passed criminal justice reform legislative packages since 2007.6 We are unaware of any equally important U.S. public policy cause that can match this extensive wave of reforms nor of any other policy organization deeply involved in so much recent legislation, which reinforced our view that criminal justice reform stands out for its political momentum. The strong possibility of observing substantial policy change in the short-term, combined with the possibility to work in multiple states, suggested that criminal justice reform might be a particularly good opportunity to learn about policy advocacy.

Criminal justice reform also stands out because we quickly identified (a) a set of innovative ideas that seemed potentially high impact and underfunded and (b) a set of researchers, specialists, and technical assistance providers ready to promote these ideas with additional funding. In particular, we learned about certain proposals such as making responses to offenses more “swift and certain,” improving community corrections (for example, by using electronic monitoring or improved risk assessment), designing regulations for cannabis in Washington and Colorado (where recreational use is now legal according to state law), and reducing the prevalence of binge drinking and alcohol use disorders, that aim to reduce incarceration with a neutral or positive effect on public safety. Mark Kleiman (more) and Angela Hawken (more) are academics who work on these types of policies and we believe each could expand their work with more funding.

## 2. The problem

The United States is burdened both by a very high incarceration rate and by a high (but declining) crime rate. In 2012, there were about 1.5 million Americans in prison and 750 thousand in jails, for a total of about 2.2 million incarcerated Americans.7 The United States incarcerates more residents as a proportion of its population than almost any other country in the world.8 The incarceration rate is not necessarily driven entirely by criminal justice policy, however; the United States may also have more crime than other developed countries. The U.S. homicide rate is more than twice as high as the OECD average of 2.23 homicides per 100,000 people.9

As shown by the above figure, since the 1990s, violent crime in the United States has fallen dramatically while incarceration rates have continued to increase.10 Rising incarceration rates in the context of falling crime could be taken to suggest that the U.S. incarcerates too many residents, but it is also possible that increased incarceration contributed to declining crime by incapacitating those most likely to commit crimes. Criminologists have attempted to disentangle these factors, but confounding factors impede rigorous analysis and we have not seen good studies nailing down the high incarceration rate’s effect on crime.11 Still, it has been argued (and seems probably correct) to us that, while initial increases in incarceration may have reduced crime, at the current margin, higher incarceration does not have large benefits for public safety.12 In particular, we’d guess that sentences in the United States are too long and too inconsistently applied and that strong community supervision (combined with the threat of short stints in jail) is not adequately used as an alternative to incarceration for low-risk offenders.

We have not attempted to carefully quantify the costs of the high incarceration rate but we believe those costs to be quite high along many dimensions:

• The fiscal costs to states and the federal government13
• The suffering of people who are incarcerated.
• The economic costs of lost earnings by those prisoners who would have been employed had they not been incarcerated.14
• Economic costs due to decreased earnings by ex-offenders after they are released from prison.15
• The costs to the families and communities of those who are incarcerated. As of 2010, 2.7 million minor children had a parent behind bars, potentially leading to a host of problems.16 Additionally, as of 2007, one in nine young black men was behind bars.17 We’d guess that high incarceration rates might have particularly negative effects in communities where incarceration has become a normative experience, potentially impacting family formation. The costs of crime may also be concentrated in black communities.18

## 3. Possible interventions

Many criminal justice reforms fall into one of two categories. Front-end reforms affect individuals at their first point of contact with the criminal justice system. Back-end reforms affect individuals after they have already entered the criminal justice system. Some examples of front-end reforms include reducing or improving the use of pretrial detention, increasing the use of alternatives to incarceration, decriminalizing some activities (such as the possession or sale of drugs), improving the quality of legal representation for criminal defendants, changing policing policies, and shortening sentences. Some examples of back-end reforms include increasing eligibility for parole, minimizing the frequency or severity of probation and parole revocations (for example by increasing the swiftness and certainty of responses), and improving community corrections to try to minimize recidivism.

Additionally, improving prison conditions may reduce the suffering of those in prison and some policy reforms outside of the criminal justice system (such as reducing lead exposure or fetal alcohol exposure) may be able to reduce crime.

We have not deeply vetted the entire range of policy changes proposed by criminal justice advocates and do not have a firm position on which reforms are most beneficial. That said, because we believe that crime and incarceration both have high costs, we are most enthusiastic about reforms that seem to reduce incarceration while improving (or at least maintaining) public safety.

#### 3.1 Ideas promoted by Mark Kleiman

Mark Kleiman is a researcher who studies and promotes approaches to crime control with the potential to simultaneously reduce crime and incarceration.19 Dr. Kleiman also worked as a consultant to advise the state of Washington on implementing the legalization of recreational cannabis.20 We think that a set of reforms propounded by Dr. Kleiman, some of which have been studied by Angela Hawken, seem particularly promising. These reforms include:

• Swift and certain sanctions exemplified by Hawaii’s Opportunity Probation with Enforcement (HOPE) program. Mark Kleiman, Angela Hawken, and other criminal justice specialists believe that misbehavior can be deterred more strongly with less punishment if punishment is distributed more consistently and closer to the time the offender misbehaves.21 The current criminal justice system (particularly the probation and parole systems) attempts to use harsh punishments to make up for the inconsistency of enforcement. Advocates of swift and certain responses believe this approach is ill-suited to human psychology and particularly to the psychology of those most likely to commit crimes, who are likely to have trouble with long-term planning. The last fifteen years of a twenty year sentence, these advocates believe, have very high cost but very low additional deterrent effect.22 Swift and certain sanctions aim to improve the efficiency of the criminal justice system and to decrease crime with less punishment.In HOPE, Hawaii Judge Steven Alm implemented a version of swift and certain sanctions by frequently, yet randomly, drug testing probationers and, instead of revoking those who failed from probation altogether, sentencing them immediately to a jail stay of a few days.23 Initial evidence from a randomized controlled trial of HOPE, which we have not vetted, was very positive, though some have expressed uncertainty about whether the program is replicable and whether its effects will be sustained over the long-run.24 Studies of other programs implementing swift and certain sanctions may have also shown promising results.25Dr. Kleiman believes that, despite the initial success of HOPE and other programs, there is a lack of funding for research on and implementation of swift and certain sanctions.26
• GPS position monitoring. GPS position monitoring has the potential to reduce both crime and incarceration by enabling law enforcement to closely supervise offenders without putting them behind bars.27 For under five dollars a day, offenders could be given tamper-evident ankle monitors so that it would be immediately evident to their supervisors if they were at the location of a crime.28 This technology might enable some prisoners to be more quickly released back into the community (or to avoid incarceration altogether) without risking public safety. The large number of crimes currently committed by those on probation or parole might also be reduced by this technology.29 Lastly, in addition to its potential to improve the quality of supervision, position-monitoring also opens up the possibility of graduated punishments, such as curfews, that are less severe and have fewer collateral consequences than incarceration but that might still deter or incapacitate many offenders.30 However, Mark Kleiman stated to us that GPS position monitoring has not yet been subject to much testing and has a very bad reputation among practitioners because of its use for actively monitoring sex offenders.31
• Drug policy reforms with potential to reduce abuse and incarceration.
• Regulating cannabis in Washington State and Colorado. Dr. Kleiman believes that the cannabis markets in Washington State and Colorado will end up promoting themselves to heavy users like the alcohol market does unless advocacy and technical assistance to state governments lead to responsible regulation.32 Dr. Kleiman was part of a group that worked with Washington State on its implementation of cannabis legalization and has suggested multiple ways to try to limit the costs of legalization and decrease marketing to abusers.33
• Reducing the prevalence of binge drinking and alcohol use disorders. Dr. Kleiman told us that alcohol abuse is the most important public policy issue that [he] can think of that has no major advocate at the moment.34 According to Dr. Kleiman, reducing binge drinking by, for example, raising the alcohol tax or prohibiting drinking among past abusers would lower crime and have an immediate effect on the homicide rate.35 However, increasing regulation of alcohol could be politically difficult.36

#### 3.2 Ideas promoted by the Pew Public Safety Performance Project

The Pew Public Safety Performance Project (PSPP) supports various evidence-based corrections and sentencing reforms intended to help states maximize the return on their public safety spending. These reforms include:

• Risk and needs assessments of offenders to improve decisions about detention, incarceration, release, supervision, and treatment
• Accountability measures to ensure the use of evidence based practices, and/or develop data reporting requirements
• Good time and earned time credits that reduce the sentences or length of community supervision for inmates who behave well in prison or participate in programs designed to reduce recidivism
• Intermediate and graduated sanctions, often based on HOPE, to deter bad behavior
• Increased funding for community based treatment such as substance abuse programs and reentry plans
• Sentencing changes
• Mandatory supervision requirements to ensure that high risk offenders are supervised after being released from prison
• Problem solving courts, which are often focused on offenders with substance abuse or mental health disorders
• Streamlined and expanded parole to reduce incarceration while protecting public safety by expanding post-release supervision.37

While these are among the most common reforms, the specific reforms in each state are tailored to each state’s needs. PSPP’s population impact projections in the participating states estimate that these types of reforms can reduce state prison populations by 5% to 25%.38 We hope to discuss these reforms in more detail in an upcoming writeup.

#### 3.3 Strategy recommended by Steve Teles

Steve Teles is a political scientist at Johns Hopkins who has written about philanthropy’s role in political movements and in policy advocacy. Prof. Teles is the only person we’ve come across who has extensively studied the historical role of philanthropy in politics, and seems to have a broad view of the different ways in which philanthropy can influence policy. We have hired Dr. Teles part-time as a consultant and his work for us on criminal justice reform has informed many parts of our investigation.

Dr. Teles believes that with vastly intensified community corrections and alternatives to incarceration in place, incarceration rates could be safely reduced by up to 50%. To do this, we would need to substantially expand the advocacy infrastructure around criminal justice reform and might need to move beyond the specific interventions listed above.39

Various political and advocacy strategies, including policy research, legislative advocacy, technical assistance to policymakers or practitioners, litigation, communications and public education, funding direct services, and funding pilot projects could be used to pursue these policies. We are still learning about what might be the most fruitful strategies for achieving promising reforms. Two tactics that have struck us as particularly promising are providing technical assistance to state policymakers through the justice reinvestment process and engaging conservatives advocating reform. PSPP uses both these tactics, and we plan to elaborate in a future writeup.

We also believe that building an infrastructure to advocate for swift and certain responses to offenses and to provide technical assistance to policymakers and practitioners interested in those principles might be particularly effective. Our grants to Mark Kleiman (more) and Angela Hawken (more) are, in part, intended to provide tools to practitioners implementing or evaluating swift and certain sanctions.

## 4. Who else is working on it?

Several national foundations are involved in criminal justice reform.

FUNDER BUDGET FOCUS AREAS
The Open Society Foundation (OSF) Budget of slightly under $15 million per year on criminal justice reform and an additional$7.8 million per year on their Campaign for a New Drug Policy.40 OSF appears (based on our scanning their list of grants) to focus on the death penalty and drug decriminalization.
The Pew Public Safety Performance Project (PSPP). PSPP does not make its annual budget publicly available. We hope to discuss PSPP more in a future writeup. PSPP is a program of the Pew Charitable Trusts a public charity. PSPP focuses on providing technical assistance to states seeking to improve their return on public safety spending, engaging with nontraditional allies of criminal justice reform (such as conservatives, law enforcement groups, and victims’ advocates), and research and public education.41
The Laura and John Arnold Foundation (LJAF) LJAF made $5.5 million in criminal justice grants over three years from January 1, 2011 to December 31, 2013.42 LJAF may also perform some criminal justice related work directly using its operating budget. LJAF focuses on “the front end of the system, which runs from arrest through sentencing, and forensic science.”43 The Ford Foundation Ford appears to have spent about$2.8 million on criminal justice grants in 2013.44 The Ford Foundation appears (based on scanning their list of grants) to be focused on supporting public defense, building a criminal justice reform advocacy infrastructure, and Law Enforcement Assisted Diversion in Seattle.45
Atlantic Philanthropies Atlantic Philanthropies spent about $9.8 million on criminal justice related grants in 2012 (the most recent year for which we have data). We expect this number to go down over time because the foundation is winding down.46 Atlantic Philanthropies appears (based on scanning their list of grants) to focus heavily on capital punishment related work.47 Public Welfare Foundation Public Welfare’s Criminal Justice Program budget is about$6 million per year.48 The Public Welfare Foundation focuses on pretrial detention reform, sentencing reform, and racial disparities.49
The Smith Richardson Foundation (SRF) SRF spends about $1.5 million per year on criminal justice policy.50 SRF exclusively funds research.51 We believe that we have identified all of the large U.S. foundations with major criminal justice programs, but do not believe that this is an exhaustive list of all funding in the area. For example, many local foundations may work on criminal justice reform. We have identified somewhat less than$60 million per year in criminal justice reform funding by foundations (not including some spending that is strictly on drug policy). The federal government and state governments also promote criminal justice reform including through the Justice Reinvestment Initiative, a partnership with PSPP. We do not know how much the public sector spends on this cause. On the other hand, we used a very broad definition of criminal justice reform funding, so our estimate of total funding includes many projects that are not primarily intended to promote comprehensive reform. For example, the Ford Foundation and Atlantic Philanthropies support some direct legal services, such as public defense and capital defense, and the Open Society Foundation focuses much of its resources on opposing capital punishment.

We have heard from various public policy and criminal justice reform experts that the field as a whole could use additional funding, but we don’t yet have very strong beliefs about the size of an adequately funded advocacy field, so we are relatively agnostic about this question.

We have, however, identified some areas where there appear to be gaps in existing funding. For example, there is no organization dedicated to promoting and providing technical assistance to jurisdictions interested in implementing swift and certain responses to offenses.52 More generally, we believe that there is room to fund certain evidence-based criminal justice reforms that focus prison beds on the highest risk offenders, try to build on support from moderates and conservatives, and prioritizing public safety while improving the efficiency of corrections spending.

While many of the above funders may fund some organizations that share this perspective, only three major foundations – Pew, LJAF, and SRF – seem very concentrated in this area of focus. In addition, we’ve spoken with two nonprofits whose approach seems, at least in part, aligned with ours. The Vera Institute of Justice’s Center on Sentencing and Corrections (CSC) provides technical assistance to governments seeking to improve their public safety spending and has worked with PSPP to provide assistance to states implementing data-driven reform through the Justice Reinvestment Initiative.53 CSC’s staff includes about twenty employees.54 The Brennan Center for Justice is a “part think tank, part public interest law firm, part advocacy group, part communications hub,” whose Justice Program is focused on using “data-driven research” to reduce mass incarceration and reform the criminal justice system’s frontend.55 The Justice Program has about eight staff members.56 We do not know the budget of either of these organizations nor do we know the extent to which their funding comes from the foundations mentioned above.

Overall, we would guess that there is less than $20 million per year being spent by organizations (other than governments) on the types of reforms we are most interested in as detailed above. We have also heard that there is a lack of funding for policy reforms, for reforms to the nexus between immigration and the criminal justice system, for front-end reforms, for national-level reform, for interdisciplinary research, and for efforts to improve prison conditions, but we have not investigated these claims in detail.57 ## 5. How much of an impact could we have? We are very uncertain about how much additional policy change various levels of funding might lead to in this space, the magnitude of prison population reduction that would be achieved by various policy changes, and the effects various policy changes might have on other variables such as the crime rate or the economy. We are also very uncertain about how much we should value reductions in prison time served relative to other potential outcomes (for example, additional income or years of life, or the ability to migrate from a developing country to a rich country). We have, however, done back of the envelope calculations to get a sense of whether plausible near-term reforms would be important enough to justify our investment in this area. We do not yet know how ambitious our longer-term goals might be were we to invest in this area over multiple years. • Justice reinvestment. Justice reinvestment is a partnership through which PSPP and the federal government’s Bureau of Justice Assistance provide technical assistance to states so that they can improve the efficiency and reduce the costs of the criminal justice system and then use those cost savings to fund evidence based public safety practices, improving the overall state budget.58 Much of the criminal justice legislation passed since 2007 came out of the justice reinvestment process. Specifically, through 2013, twenty-nine legislative reform packages had been passed in twenty-seven states through justice reinvestment.59On average, PSPP forecasts that in the states in which it has worked, prison populations will fall about 11% in the five years after reform relative to what would have happened in reform’s absence. We hope to give our assessment of these forecasts in a future writeup focused on PSPP. Justice reinvestment’s track record suggests to us that a 10% further reduction in incarceration is ambitious but not infeasible.For an in depth overview of justice reinvestment, please see the Urban Institute’s assessment here. Our own full review of PSPP, including its justice reinvestment work, is near completion and should be published shortly. • Scaling swift and certain responses (exemplified by HOPE). As discussed above, HOPE is a program in which particularly high-risk probationers are transferred from normal probation into a swift-and-certain sanctions regime, in which they are subjected to randomly scheduled drug tests and immediately given brief jail stays for each probation violation60The randomized controlled trial of HOPE, which was restricted to drug-involved probationers who were not on supervision for domestic-violence or sex offenses, estimated that the average time spent in prison per probationer was reduced from about 134 days behind bars to about 69 days, an average difference of 65 days.61 In addition, no-shows for probation appointments, positive urine tests, new arrest rates, and probation revocation rates all decreased by half or more, suggesting that HOPE may also have positive effects on public safety.62About two million persons entered probation in the United States in 2012.63 It is difficult to guess at the coverage that could be reached through an ambitious but pragmatic expansion of swift and certain responses, especially because we do not know whether the intervention would have strong effects on probationers other than the specific population covered by the HOPE RCT (high risk, drug-involved probationers neither under supervision for sex crimes nor for domestic violence). However, we’d guess that a successful campaign for the adoption of swift-and-certain might scale to one fourth of probationers.64If HOPE achieved the same results after being scaled to one fourth of the people entering probation in a given year, it would reduce the number of people incarcerated at a point in time by about 90 thousand people.65There were about 2.2 million incarcerated people in 2012, so this would be a decrease of about 4%.66 It strikes us as possible that a major advocacy push combined with major investments in criminal justice infrastructure and administration might lead to a more extensive roll out of swift-and-certain or to an expansion of HOPE’s principles to contexts beyond probation.67 It’s also possible that additional research on HOPE would lead to improvements and better-calibrated responses to offenses. However, this possibility is quite speculative and it’s quite possible that the results of HOPE would not be replicated at scale.68 • Effects of a 10% reduction in incarceration rates. Based on the above points, we would guess that a 10% reduction in incarceration rates, with a neutral or positive effect on public safety, is a reasonable back of the envelope baseline for a very successful criminal justice reform campaign. Below, we discuss our very rough sense of the social impact of such a win. • Years of incarceration averted. In 2012, there were about 2.2 million people incarcerated.69 Reducing the incarcerated population by 10% for 10 years would therefore avert 2.2 million person years of incarceration.70 • Fiscal savings Total state corrections expenditures were about$48.5 billion in 2010.71 We’d guess that a 10% reduction in incarceration would lead to somewhat less than a 10% reduction in corrections spending because some of the reduction in incarceration will be achieved by strengthening community corrections and because the marginal cost of incarceration may be less than the average cost. We’d therefore very roughly guess that a 10% reduction in incarceration would lead to a 5% reduction in corrections spending and savings of $2.4 billion per year.72 Thus, over the course of ten years, we’d guess that successful, ambitious reform would save states about$24 billion on the costs of incarceration. However, the total impact on state budgets would probably be somewhat smaller because investments in improving community corrections would be required to facilitate these reductions in incarceration. It is possible that further reductions in incarceration could be safely achieved if large portions of these savings were spent on programs attempting to lower recidivism or alternatives to incarceration, but we have not carefully investigated the effectiveness of such programs.
• Diffuse effects and effects on other outcomes. We’d hope that funding groups that study and spread data-based criminal justice policies designed around effectiveness instead of designed to make politicians appear “tough on crime” would have additional diffuse effects on improving the U.S. criminal justice debate. However, we find it difficult to estimate the magnitude of these effects and do not factor them into this discussion. This analysis also does not focus on possible reductions in crime. We’ve seen some evidence that these effects might be substantial, but we have not tried to quantify these effects.
• The potential for much larger reductions. Prof. Steve Teles believes that for a sufficiently ambitious effort, including substantial expansion of advocacy infrastructure on many dimensions, impacts in the range of a 50% reduction in incarceration could be achievable.73

## 6. What have we done so far?

Criminal justice reform stands out to us in part because we found impressive people and organizations with innovative ideas who were struggling to get adequate funding. These opportunities suggested to us both the possibility of making an impact and also the possibility of learning quickly from our grantees.

We discuss below the grants that we have made so far.

#### 6.1 Grant to support Mark A.R. Kleiman, Professor of Public Policy at UCLA

We first decided to speak to Dr. Kleiman because of his work researching and promoting the concept of swift and certain sanctions, which Matt Stoller and Aaron Swartz found to be a potentially promising policy and because his work was recommended to us by Steven Teles.74

During our conversations with him, Dr. Kleiman presented us with policy ideas aimed at reducing crime and incarceration that struck us as innovative, potentially high-impact, and neglected. These ideas included:

• Swift and certain sanctions (more above)
• Position-monitoring (more above)
• Reducing alcohol abuse by, for example, increasing alcohol taxes (more above)
• Regulating cannabis in Washington State and Colorado. Some ideas Dr. Kleiman thought would be worth trying out or studying to prevent recreational cannabis legalization from harming heavy users included:
• Allowing home delivery of cannabis.
• Allowing individuals to set their own quotas (alterable only with thirty days notice) as a commitment mechanism to avoid using more than they intend.
• Setting limits on the amount of THC that could be produced.
• Studying the retail process and ways for the state to dissuade abuse such as labels or other forms of communication.
• Setting levels of cannabis taxation such that the price of legal cannabis will neither be much higher than its current, illicit price (which would incentivize a continued illicit market) nor much lower (which would increase abuse)75
• Refocusing international narcotics enforcement on violence prevention76
• Researching the possible beneficial effects of some illicit drugs77

These ideas struck us as potentially innovative, insightful, and pragmatic. Dr. Kleiman believes that, without regulation, legalized cannabis markets may be dangerous to heavy users (see above) or lead to a backlash.78

As our investigation of criminal justice reform moved forward, we heard that Dr. Kleiman is excellent at navigating the intersection of research and policy and we learned that there does indeed seem to be a lack of research funding and attention toward his ideas relative to our impression of their promise (see above for a discussion of existing funding for criminal justice reform).79

We asked Dr. Kleiman about his need for more funding and his team’s priorities. Dr. Kleiman told us that he could use $250,000 to$300,000 on immediate, time-sensitive research and technical assistance and could usefully spend up to $1 million per year if additional funding priorities were included.80 He also told us that his application to a foundation for$175,000 to work on these issues had recently been rejected.81

Dr. Kleiman also sent us a prioritized list of fifteen specific projects and their budget estimates. The top six projects, totaling $245,000, were: • Alcohol cross-elasticity Dr. Kleiman believes alcohol abuse is a serious threat to public safety and cannabis use could be a complement or substitute for alcohol use, so cannabis policy’s effects on alcohol use could be an important component of its costs or benefits. He proposed to “use the natural experiment created by difference in cannabis policy between Western and Eastern Washington to measure the impacts of cannabis availability on alcohol sales and on health and public-safety outcomes.”82 • Outcome list and data-gathering plan. In order to learn from the experiments with legalization in Washington and Colorado, it could be important to have baseline data from before legalization has had its effect. Dr. Kleiman proposed to identify the most relevant outcomes for evaluating legalization, determine how to measure those that are measureable, determine which must be measured before legalization is implemented, and then estimate the cost of carrying out time-sensitive data-gathering.83 • Online implementation tool for swift-and-certain sanctions programs. Dr. Kleiman proposed to create a website to provide information to jurisdictions interested in implementing swift and certain sanctions for probation and parole violations.84 • Swift-and-certain mechanism study: self-command and procedural justice. In order to learn more about the mechanisms behind the success of swift and certain sanctions and to improve program design, Dr. Kleiman proposed to “develop and field-test instruments to measure self-command, delayed gratification, and perceptions of fairness among offenders subject to swift-and-certain sanctions programs to determine which, if any, predict outcomes.”85 • Optimal cannabis taxation. Dr. Kleiman proposed to “[d]etermine the optimal level and basis of cannabis taxation for states now legalizing, balancing considerations of health, public safety, revenue, and administrative feasibility.”86 • User-determined quotas. Dr. Kleiman proposed to study the possibility of implementing user-determined quotas to help cannabis users avoid problem use.87 We were surprised to learn that, in a field with so much attention, Dr. Kleiman had not already found funding for his agenda, and was planning to allocate the same staff time to for-profit consulting if he could not find funding for this work. Good Ventures made a$245,000 grant to the Washington Office on Latin America, which will be used to support Dr. Kleiman’s research. While the grant amount was designed to be enough to fund the above six projects, it is unrestricted and Dr. Kleiman is free to use the funding for other research opportunities if they arise.

Dr. Kleiman’s descriptions of his proposed projects are available here.

## 7. Open questions

• What is the relationship between incarceration and public safety? How does this differ among various reforms? What incarceration rate and crime rate should we expect if the criminal justice system is functioning well?
• What are the expected impacts on crime and incarceration of the most plausible reforms that are currently on the table? What effect should we expect from broader advocacy strategies?
• What is the appropriate level of funding for different criminal justice reform strategies? How much of an effect on policy should we expect from differing levels of funding? How does this differ by the type of strategy? Is it possible that the field has developed far enough that reform will happen even without additional funding?
• How do moderate reforms affect the possibility of achieving large-scale reforms? Do moderate reforms make larger-scale reforms more likely by moving policy and dialogue in the right direction or do they subvert large-scale reforms by drawing away attention and resources?
• What could be accomplished by developing a strategy that attempted to more aggressively increase the attention paid to criminal justice issues, expand political possibilities, and design additional interventions?

## 8. Our process

Good Ventures began exploring drug policy reform, a field closely connected to criminal justice reform, in 2012 with a grant to the Drug Policy Alliance followed by a report on the topic researched and written by Matt Stoller and Aaron Swartz. We initially learned about Mark Kleiman through this report.

In the spring of 2013, we began to investigate the broad cause of policy-oriented philanthropy and spoke to U.S. public policy generalists including Steven Teles, Gara LaMarche, and Mark Schmitt, and gained the impression that criminal justice reform is a promising cause with a “window of opportunity” for policy change.95 Dr. Teles, in particular, argued that this cause represented the best available opportunity to find good U.S. giving opportunities quickly and make relatively short-term progress.96

Over the summer and fall of 2013, we had broad conversations about criminal justice reform with experts in the field and with criminal justice reform specialists and many of the field’s biggest funders. Public notes are available from our conversations with:

More recently, our conversations have focused on evaluating specific funding opportunities. Public notes are available from such a conversation with Mark Kleiman and Angela Hawken. We expect additional public notes to be forthcoming.

Our research has particularly but not exclusively investigated approaches to criminal justice reform focused on improving public safety and reducing state spending on incarceration by saving prison beds for high-risk offenders because we learned early on that this approach seems underfunded relative to its promise. Dr. Teles, whom we have hired as a part-time consultant, has also played a major role in helping us to design our criminal justice reform strategy and in sourcing opportunities. We continue to be open to learning about more opportunities in this space and may make additional grants in the future.

## 9. Sources

DOCUMENT SOURCE
Angela Hawken email to GiveWell on September 17, 2013 Unpublished
Angela Hawken homepage Source
BetaGov proposal Unpublished
Brennan Center website Source
Bureau of Justice Statistics. Correctional Populations in the United States, 2012 Source
Bureau of Justice Statistics. Criminal Victimization, 2012 Source
Bureau of Justice Statistics. Probation and parole in the United States, 2012 Source
Bureau of Justice Statistics. State Corrections Expenditures, FY 1982-2010 Source
GiveWell spreadsheet compiling criminal justice grants Unpublished
GiveWell’s notes from a July 26, 2013 conversation with the Open Society Foundation Source
GiveWell’s notes from a September 10, 2013 conversation with Mark Steinmeyer Source
GiveWell’s notes from an August 27, 2013 converstion with Matt Stoller Source
GiveWell’s notes on a July 2, 2013 conversation with Mark Kleiman Source
GiveWell’s notes on a June 12, 2013 conversation with Steve Teles Source
GiveWell’s notes on a November 12, 2013 conversation with Mark Kleiman Source
GiveWell’s notes on a September 16, 2013 with Angela Hawken Source
GiveWell’s notes on an August 22, 2013 conversation with Inimai M. Chettiar Source
GiveWell’s notes on an August 23, 2013 conversation with Adam Gelb Source
GiveWell’s notes on an August 28, 2013 conversation with Seema Gajwani Source
GiveWell’s notes on an October 31, 2013 conversation with Gara LaMarche Source
GiveWell’s notes on May 22 and June 14, 2013 conversations with Mark Schmitt Source
Hawken and Kleiman 2009 Source
LaVigne et al 2014 Source
LJAF website. Areas of Focus Source
LJAF website. Grants Source
Mark Kleiman email to GiveWell on April 28, 2014 Unpublished
Mark Kleiman homepage Source
Mark Kleiman. Proposed project list. Source
Mark Kleiman. Smart on Crime Source
Mark Kleiman. The Outpatient Prison Source
National Institute of Justice. Swift and Certain Sanctions Source
OECD. How’s Life? 2013: Measuring Well-Being Source
Pew Presentation to Good Ventures and GiveWell Unpublished
Pew Public Safety Performance Project. Sentencing and Corrections Reforms in Justice Reinvestment States Source
Pew. Collateral costs: incarceration’s effects on economic mobility Source
Pew. One in 100: Behind Bars in America 2008 Source
Schmitt, Warner, and Gupta 2010 Source
Steve Teles conversation with GiveWell on May 9, 2013 Unpublished
Steve Teles email to GiveWell on April 24, 2014 Unpublished
Steve Teles email to GiveWell on March 1, 2014 Unpublished
Vera Center on Sentencing and Corrections Website Source

## Geoengineering Research

Updated: October 2013

This is a writeup of a medium investigation, a brief look at an area that we use to decide how to prioritize further research.

## In a nutshell

• What is the problem? Geoengineering – i.e., large-scale interventions in the climate to attempt to reduce global warming or its impacts – could conceptually mitigate some of the most catastrophic impacts of climate change, but major questions remain regarding the feasibility, likely costs and benefits, and optimal governance of the various possible geoengineering approaches.
• What are possible interventions? A philanthropist could directly fund research on these issues, lobby a government to fund research, or fund the development of governance mechanisms to enable research.
• Who else is working on it? Although there appears to be significant academic interest in geoengineering as a research area, funding appears to be limited. There appears to be much more funding on carbon dioxide removal than on solar radiation management; carbon dioxide removal is likely less risky than solar radiation management, but also likely more expensive and less fast-acting.

## 1. What is the problem?

Unmitigated climate change is likely to have large negative humanitarian impacts across a range of outcomes, and may have disastrous impacts. We have written about the likely impacts of unmitigated climate change here, summarizing the findings of the Intergovernmental Panel on Climate Change’s 2007 Fourth Assessment Report. We have also written about less likely but potentially extremely harmful impacts of more extreme climate changes on a separate page.

Once emitted, carbon continues to warm the planet for decades and a portion remains in the atmosphere (keeping temperatures warmer) for many centuries.1 In the event that existing efforts to limit carbon emissions fail to adequately reduce emissions, technology to rapidly limit temperature increases could be quite valuable, particularly in the event of worse-than-anticipated outcomes.

Geoengineering refers to large-scale interventions in the climate to attempt to reduce global warming or its impacts.2 We have seen two broad types of geoengineering discussed:

• Solar radiation management (SRM): reflecting back more sunlight to cool the earth without directly affecting carbon dioxide concentrations. The particular strategy we’ve seen discussed most frequently is injecting sulfate aerosols into the stratosphere to reflect back some sunlight, but other proposals include using saltwater spray to brighten clouds or using large mirrors in space to reflect back more sunlight.3
• Carbon dioxide removal (CDR): there are a number of technical proposals for attempts to remove carbon dioxide from the atmosphere that we have seen discussed, including direct air capture with chemical processes, biochar, ocean iron fertilization, and bio-energy with carbon capture and storage.4

Most of our research to date on geoengineering has focused on stratospheric injection of sulfate aerosols. Our understanding is that, relative to other geoengineering approaches, stratospheric injection is likely to be faster-acting and cheaper (in simple financial terms, not necessarily in terms of risks or costs and benefits), making it a plausible candidate for use in response to a climate emergency.5 On the other hand, our understanding is also that SRM has a greater potential for causing harm than CDR6 and that research into SRM is not as well-funded as research into CDR.7 We believe there to be many open questions about the potential effectiveness and side-effects of SRM, the answers to which could inform the behavior of policymakers facing climate emergencies.8

We have not done any systematic comparison of the case for funding further research on SRM compared to further research on CDR, and we have looked at only a subset of all possible SRM approaches, for instance, not thoroughly investigating albedo modification or marine cloud brightening. We regard these as important questions for further investigation should we proceed further with this research.

#### 1.1 Background on stratospheric aerosol injection

Volcanic eruptions naturally produce aerosol that cools the planet by reflecting back sunlight. For example, the eruption of Mount Pinatubo in the Philippines in 1991 decreased the 1992 global mean temperature of the Earth by about 0.5ºC.9

These “natural experiments” raise the possibility of intentionally injecting sulfate aerosols into the stratosphere to offset the warming effects of climate change. Our understanding is that scientists believe it to be likely, though not certain, that such an effort would be feasible and would result in lower average global temperatures.10

Unlike many other approaches to climate change, such as emissions reductions or carbon dioxide removal, stratospheric injection of sulfate aerosols does not address the fundamental issue of elevated greenhouse gas concentrations. This means it would perform worse than other climate response strategies in many ways, including:

• Since it does not reduce carbon dioxide concentrations, sulfate aerosol injection would not address all of the results of high carbon emissions, such as ocean acidification.11
• Once started, rapidly halting aerosol injection would lead to far faster warming than climate change itself, with potentially more disruptive results.12

In addition, stratospheric aerosol also carries a variety of risks. For instance, some models have suggested that solar geoengineering could negatively affect precipitation, leading to droughts in some places.13 Some scholars have also pointed to the risk of conflict over control of geoengineering efforts as another potential negative outcome,14 and “unknown unknowns” are a central cause for concern.15

#### 1.2 Open questions

Despite its potential benefit as a form of insurance against catastrophic climate emergencies, we believe that there remain many unanswered questions about whether and how stratospheric injection could or should be deployed, and what the likely positive and negative effects of deployment would be.

Some of the open questions we see as most important are:

• Could stratospheric aerosol injection offset several degrees of warming on an ongoing basis? Although volcanic eruptions serve as precedent for small levels of short-term cooling, some have argued that there are limits on how much cooling sulfate aerosols could produce,16 and, in any case, there is the possibility of other limitations on the viability of sulfate aerosols.
• What are the likely humanitarian costs of conducting such an effort, and are there technical strategies that could be used to mitigate them? One example that has been cited is the possibility that sulfur aerosol injection would reduce the strength of the Asian monsoon.17 Another possibility we’ve seen discussed is that sulfate aerosols would harm the ozone layer.18
• How should a global geoengineering scheme be governed? What would the political implications of the availability of sulfate aerosol injection technology be? International violence arising from disputes over the appropriate amount of aerosol injection to employ could conceivably be much more harmful than aerosol injection itself.19

Researching this topic could be very valuable no matter the findings. In the event that large-scale stratospheric aerosol injection could not feasibly reduce temperature, or that deploying it would cause more harm than benefit, having that knowledge prior to attempts to deploy the technology in an emergency situation could be enormously valuable. If large-scale stratospheric aerosol injection could feasibly reduce temperature, and would be net-beneficial under some future set of adverse climate conditions, that also seems to be quite valuable knowledge for policymakers to have. In either case, effective and informed governance could be invaluable.

## 2. What are possible interventions?

The two main strategies we see for using philanthropic funding to help address these questions are:

• directly funding further research
• advocating for governmental funding of further research.

In practice, either overarching approach could involve a focus on governance discussions and research, as opposed to a focus on directly attempting to answer the kinds of questions mentioned above. Governance of geoengineering research itself appears to be an active area of research.20

We are not aware of any non-profit organizations currently raising money to systematically pursue any of these aims, and we do not have a strong sense of what the likely costs or returns to these approaches would be.

## 3. Who else is working on this?

Our understanding is that there is a significant amount of academic interest in stratospheric aerosol injection, but that government and philanthropic funding for research is limited:

• A September 2010 report by the United States Government Accountability Office assessed U.S. federal funding for geoengineering research in fiscal years 2009 and 2010, reporting $949,000 of research on solar radiation management and about$101 million on geoengineering research overall, the vast majority of which was on conventional mitigation approaches that could be relevant to geoengineering.21
• During a May 2013 conversation, Andrew Parker estimated that the total sum of global government research spending on ongoing solar geoengineering research projects was roughly $20-25 million; since many of the projects span multiple years, the figure does not represent an annual estimate.22 • Bill Gates has personally funded$4.6 million worth of geoengineering research, including but not exclusively focused on stratospheric aerosol injection.23

Building on a list compiled by Andrew Parker and David Keith (PDF), we have tried to identify funded projects and funding sources around the world that explicitly include a significant solar geoengineering component (XLS).24 Our total tally of funding for such projects (for which we have funding information) amounts to about \$11 million/year. This may be an overestimate of total resources directed to solar geoengineering, as it incorporates non-solar-geoengineering aspects of grants that are only partially devoted to solar geoengineering, but we believe it is more likely to be an underestimate of total resources because:

• research that is supported by general institutional resources (such as unrestricted funding to a university, graduate students stipends, or computing resources) is not accounted for
• some funded research that might be classified as solar goengineering may not be explicitly portrayed as such by the researchers or grant agencies
• our search strategy of explicitly enumerating all the grants that we know of and taking the sum of their funding means than any missed funding sources would entail an underestimate, and we believe that we missed at least some funding.25

For details of the projects included in our tally, see our spreadsheet on geoengineering research funding.

## 4. Questions for further investigation

Our research in this area has been relatively limited, and many important questions remain unanswered by our investigation. (These are meant to be distinct from the questions above, for which we believe further academic research is necessary. These questions are for our further research.)

Amongst other topics, further Open Philanthropy Project work on this cause might address:

• How would further research on sulfate aerosol injections compare with other research related to climate change, such as further monitoring of feedbacks, or with other types of geoengineering research, such as carbon dioxide removal or marine cloud brightening?
• How likely is it that funding research on geoengineering would cause harm (e.g. by undermining public support for optimal emissions reductions or by starting down a “slippery slope” towards deployment)? To what extent are policymakers considering geoengineering likely to respond to improved evidence?
• What type of research is likely to be most helpful for policymakers, and what is the best way to facilitate its creation? Should a philanthropist focus on directly supporting scientific research or on improving the governance of research, or both?
• How long is it likely to take to obtain the main benefits from a geoengineering research program? How likely are major funders to enter the field over that time horizon?
• What would the appropriate level of investment in a research program be, and how does this vary based on strategy (e.g. by whether a philanthropist directly funds research versus governance versus lobbying for more research)?

We believe that answering these questions would require a considerably deeper investigation than we have done to date.

## 5. Our process

We initially decided to investigate solar geoengineering as part of our more general shallow investigation of climate change as a potential philanthropic program area because we had heard about it in the popular press and because the Copenhagen Consensus report on climate change identifies geoengineering as a particularly promising mechanism for responding to the threat of climate change.26

Our initial investigation in mid-2012 consisted of reading a number of articles about geoengineering and speaking with several senior scholars who had written about the issue. We returned to have a few more conversations and to write up this review in the spring of 2013. Public notes are available from our conversations with:

We also attended a portion of the Fourth Interdisciplinary Summer School on Geoengineering at Harvard in August 2013, entitled “Solar Radiation Management: Exploring uncertainties and trade-offs.”

Our research has particularly but not exclusively focused on stratospheric injection of sulfate aerosols, one particular geoengineering approach, because we believe it may be worse-funded, relative to its potential importance, than other aspects of geoengineering research, but this is something we have not investigated deeply and regard as an important issue for further investigation.

## 6. Sources

 Blackstock et al. 2009 Source Archive Copenhagen Consensus on Climate: Findings of the Expert Panel Source Archive GAO 2010 Source Archive IPCC AR4 WGI Source Archive Keller conversation Source – Moreno-Cruz and Keith 2012 Source Archive Parker conversation Source – Rasch et al. 2008 Source Archive Ricke, Morgan, and Allen 2010 Source Archive Robock 2008 Source Archive Robock, Oman, and Stenchikov 2008 Source Archive Ross and Matthews 2009 Source Archive Solar Radiation Management Governance Initiative 2011 Source Archive Fund for Innovative Climate and Energy Research Source Archive