The GiveWell Blog

Our goal with hosting quarterly open threads is to give blog readers an opportunity to publicly raise comments or questions about GiveWell or related topics (in the comments section below). As always, you’re also welcome to email us at info@givewell.org or to request a call with GiveWell staff if you have feedback or questions you’d prefer to discuss privately. We’ll try to respond promptly to questions or comments.

You can view previous open threads here.

• Deepak De on December 14, 2021 at 2:28 pm said:

I would love a post on how well donation matching is working towards getting more donations. Like is it mostly driven by new people? Are people donating less themselves since the overall number will be 2x if there is 1:1 matching? Would love to know

• Dan Turner on December 14, 2021 at 3:26 pm said:

Thank you for all the work you do.

Would it make sense to publish your models (e.g. https://blog.givewell.org/2021/11/19/malnutrition-treatment/) as RMarkdown, Jupyter, or other repeatable notebooks?

Others could iterate on the models and to find narrower estimates, better data sources, and so on. The tools might be more powerful than your current spreadsheets, letting you make decisions you could not previously. E.g. I see your current spreadsheet’s outputs are upper and lower bounds, but more info in between the end points might be useful for making decisions.

Obviously there’s a cost to writing up your models in this way. Do you see it as (not) worth the cost?

• Thank you for the work you do.

https://techcrunch.com/2021/05/12/vitalik-buterin-donates-1-billion-worth-of-meme-coins-to-india-covid-relief-fund/

Want to ask how money from Vitalik Buterin’s ETH donation to Givewell ($54M in Q2) got used? Curious because the donation didn’t seem to show up in Maximum Impact Fund. • We have been giving substantially to US food banks this year. It might make sense to create a separate research staff for US charities. We understand US issues may be more complex or less efficient, and there may also be a segment of donors who simply wont give to African causes but will give to US causes. Expanding the organization with dedicated research in proportion to and funded by US donations would expand existing bandwidth not compete for it. • Many small, grass roots non-profits do amazing work that is difficult to quantify. One of the reasons is that they cannot afford to do a comprehensive evaluation, as rarely is money budgeted for consultants. Most funds are put directly into services, and fundraising is difficult for programs that serve unpopular causes. Homelessness is a good example. It is the most extreme example of the US’s failure to deal with systemic poverty and all it’s causes. Yes, it’s complex. Yes, it takes time and careful effort to address the issue on an individual basis, case by case, or to do the advocacy needed to make systemic changes. The general public seems to be divided into three main factions: in it’s perception and response: 1) It’s the homeless people themselves who are to blame.Lock ’em up and that will fix things. 2) There’s nothing I can do, I feel helpless in the face of such suffering and community degradation. I’ll just look away. 3) I know what works and I’ll tell you, despite the fact that I have no experience with these issues or their potential solutions. If you aren’t doing things my way, you are failing and I won’t give to your program. Until comprehensive evaluations can be done free or cheap for small non-profits, and done in such a way that it wont impact staff time better spent working on the issues, there will be no way for major funders to see how one small non-profit is making deep changes in the massive maze of poverty and homelessness. Don’t discount these orgs because you don’t know what they actually do. Find out, and help them do it. I speak from the experience of founding and then running a grass roots non-profit for 20 years. We worked with homeless people disabled with mental illness, substance use, cognitive conditions, HIV/Aids, and PTSD who could not get services elsewhere. Our budget was small, ($400,00 per year) and we were effective in providing treatment, income, housing and on-going support. Lack of funding closed us.

• Sonia Albrecht on December 19, 2021 at 11:28 pm said:

Do you have more uncertainty about the impact on the recipient’s finances of GiveDirectly or your top health charities? How big is the difference in uncertainty?

• Maggie (GiveWell) on December 20, 2021 at 12:18 pm said:

Hi Deepak,

We wrote a bit about matching funds in this 2020 blog post and you can read a little more about our matching structure here. In short, matching campaigns have been a successful way for us to track what motivates new donors to give. Matching funds have been particularly effective in allowing us to track the results of our podcast ads, which have been highly successful. We’re grateful to the supporters who have funded matching campaigns, which have helped many new donors find our work. Because we target new donor acquisition with our matching campaigns, we don’t have a read on how the match impacts gift size (we don’t have a previous donation as a benchmark). Our initial impression is that the size of the match may cause people to give more rather than less. Gifts have tended to cluster around the size of the match: when the match was capped at $1,000 many new donors gave at that level, and when the match was capped at$250, many new donors gave at that level.

-Maggie

• Maggie (GiveWell) on December 20, 2021 at 2:10 pm said:

Hi MKC,

Thanks for sharing your perspective on nonprofits. We acknowledge that there are many causes that aren’t easily captured by our model for finding effective, tractable interventions for which impact can be measured. Domestic homelessness is unlikely to be a good fit for our mission given how much farther funding can go in low and middle income countries. I shared a bit more about our commitment to global causes in my response to PJC (above).

-Maggie

• Maggie (GiveWell) on December 20, 2021 at 2:13 pm said:

Hi Dan,

Thanks for your note! We have considered doing more explicit Monte Carlo analysis, which we see as the main reason to use programming languages. Currently, we continue using Google Sheets because staff are already familiar with the format, and because it integrates well with the information we input, including the data charities send us.

-Maggie

• Maggie (GiveWell) on December 21, 2021 at 8:10 pm said:

Hi Sonia,

We’re far more certain about the impact of cash transfers (such as GiveDirectly), where the mechanism is straightforward and the evidence is very strong. The financial impact of our health charities is informed by quasi-experimental evidence on the long run effect of a negative health outcome (such as malaria) on earnings, which is less certain. We don’t have an explicit quantitative comparison of the level of uncertainty, in part because a lot of our uncertainty is model uncertainty (which we find more challenging to quantify) rather than parameter uncertainty. One of the main uncertainties in our model is the extent to which we are being consistent in how we account for flow-through benefits between interventions with direct economic benefits and interventions for which the primary benefit is health related. It’s possible we aren’t being as consistent as we could be, and that we might count the income-improving benefits of health interventions more than we count the health-improving benefits of income.

-Maggie

• Maggie (GiveWell) on December 21, 2021 at 8:15 pm said:

Hi PJC,

Thanks for taking the time to reach out. There is certainly a lot of interest in cause areas and geographies that GiveWell research doesn’t investigate, and we know there’s a growing demand for more evidence-based approaches to choosing domestic charities. That said, our mission is strongly committed to doing the most good per dollar, which is why we focus on global health and development work. Taking research staff time away from finding the most cost-effective funding opportunities globally in order to research geography-restricted interventions is unlikely to become a priority of ours in future years. You can read more about our commitment to global causes here and here.
-Maggie

• Provide An Update on Deworming Confidence?

GiveWell last wrote in depth about deworming in the 2017 time frame [1]. At the time, GiveWell advocated for deworming based on “weak evidence on the causal relationship between reducing worm loads and improved life outcomes.”

It’s been four years. Has anything happened [2] in that time to make this evidence base more solid?

[1]
https://www.givewell.org/international/technical/programs/deworming – last updated Jan 20218
https://blog.givewell.org/2016/12/06/why-i-mostly-believe-in-worms/ – written December 2016; says it was updated in July 2021, but I can’t tell what changed. (No revisions stand out to me.)
https://blog.givewell.org/2017/01/04/how-thin-the-reed-generalizing-from-worms-at-work/ – written January 2017; says it was updated in November 2017.
https://blog.givewell.org/2017/12/07/questioning-evidence-hookworm-eradication-american-south/ – written December 2017; updated January 2018

[2] new published research, new data from the top deworming charities, additional GiveWell analysis of existing data, etc.

• Maggie (GiveWell) on December 28, 2021 at 9:21 am said:

Hi Phil,

Thanks for your kind words! Vitalik Buterin’s gift was unrestricted, which means that it can support GiveWell’s operating expenses, subject to our excess assets policy. Because his gift was so extraordinarily generous, we only retained ~$8 million for our operating expenses and fees related to the gift, and designated the remainder for grantmaking. Of that, ~$16.6 million was distributed to AMF alongside our most recent Maximum Impact Fund allocation. We expect to grant the remainder soon.

-Maggie

• Claudia Wolfe on December 28, 2021 at 10:34 am said:

I BEG that Give Well to check out S.O.I.L in Haiti as an exceptional Charity to support. They get compost toilets into homes and communities, employ locals to pick up the waste, take it to composting site where it is turned into rich soil for community gardens. So it’s efficacious sanitation, while offering dignity and jobs to one of the poorest countries in the world.

• Hello! I am a long time GiveWell Donor and I was wondering what efforts GiveWell made regarding the rhetoric used to discuss Effective Altruism. It is concerning to me how it often plays so much on guilt, like “would you let a baby drown in a lake” and the devaluing emotional connection and individuality. I also wish it had a better way to emotionally connect people to the results of their giving. I don’t have easy answers to this but GiveWell is hugely influential in this area and I would love to see a more positive approach that seems more likely to work. I would like to see communication that is based in psychological science and have concerns that the current predominant approach is unlikely to be widely appealing.

• Dan Hare on December 29, 2021 at 7:39 am said:

A wonderful initiative.
There are far better tools for repeatable data acquisition, cleansing, modelling and benchmarking than Google sheets or, even worse, code.
However, despite our best efforts not for profits are the most resistant to change even when there’s no charge.
Would love to help.

• Maggie (GiveWell) on December 29, 2021 at 4:18 pm said:

Hi Claudia,

Thanks for the suggestion! Because our research process requires substantial involvement of the charities in consideration, we actually require that applications be submitted by formal representatives of those charities. If you’d like this charity to be considered by GiveWell, the best thing to do is to send them the instructions posted here and suggest that they proceed with an application.

-Maggie

• Maggie (GiveWell) on December 29, 2021 at 4:22 pm said:

Hi Emily!

Thanks for your longtime support. We definitely think very carefully about messaging. A central priority of ours is transparency of messaging – whether accurately communicating donor impact or the level of confidence we have in our research claims. We work hard to avoid any manipulation of donors, emotional or otherwise. We have tended to focus on positive messaging (i.e. highlighting the impact a donor can make if they are motivated to give, rather than drawing on guilt). We also prioritize respect for the individuals our top charities serve, and reflect their role as active agents of change. We are always working on our messaging, though, and I’d be curious to hear more of your thoughts in this area. I’m reaching out to you via email to see if you’d like to share more.

-Maggie

• Maggie (GiveWell) on December 29, 2021 at 5:16 pm said:

Hi Dan,

Thank you for the offer! Spreadsheets serve our current needs and are an accessible tool for the public to vet our research (which is very important to us). That said, we’re open to learning about new tools as we become aware of them.

-Maggie

• Hi Maggie,

Thanks for your response. I have another question: About 18% of GiveWell’s money moved went to GiveDirectly:
https://blog.givewell.org/2021/11/12/givewells-money-moved-in-2020/

Why is that, considering other high-impact projects with 5-8x GiveDirectly impact aren’t getting funded? How much of that was directed by GiveWell, vs directed by donors but distributed through or attributed to GiveWell? It may be useful to make a distinction between those 2 cases.

Another question: I made a donation through a DAF for the Maximum Impact Fund, but the letter from the DAF says that it’s for Maximum Impact Fund but it’s also an unrestricted gift. Does it still go to Maximum Impact Fund?

Thanks,
Phil

• Maggie (GiveWell) on January 7, 2022 at 3:29 pm said:

Hi Phil,

We haven’t allocated funds from the Maximum Impact Fund to GiveDirectly since 2015. The current funding directed to GiveDirectly is from donors who donate direct-to-charity and report those donations to us due to our influence of their donation, or donors who donate to us with an allocation for GiveDirectly. Additionally, we recommend a $500,000 grant from Open Philanthropy each year; this is granted to each of our top charities annually as an incentive to participate in our assessment, which can be quite intensive. Your suggestion to make a distinction between donor-directed gifts and GiveWell-directed donations is an interesting one. When we receive gifts without an allocation directive, they are designated as unrestricted. Those gifts are used for our operating expenses, subject to our excess assets policy , and any remaining unrestricted funds are granted to charities. I wasn’t able to find your particular donation via the email associated with your comment, so I’m going to follow up with you directly. It’s important to us that donations are allocated according to the intentions of our donors. Thanks for your support, and for letting us know about this. -Maggie • Austen on January 22, 2022 at 9:07 pm said: You should make Centre for Pesticide Suicide Prevention a recommended charity. I don’t see how GiveDirectly, at 1/50 cost effectiveness is recommended and CPSP is not. • Maggie (GiveWell) on February 2, 2022 at 12:08 pm said: Hi Austen, Thanks for writing in regarding CPSP. We supported the Centre for Pesticide Suicide Prevention as part of our work on public health regulation . We have funded them through incubation grants in August 2017 and January 2021 and have written a bit more about their work here . Our incubation grants are made to organizations that may become top charities in the future, once we can be confident that they meet our high bar for recommendation. -Maggie • Ethan Kennerly on March 5, 2022 at 3:39 pm said: Do you know a rough estimate of the expected life-days added to an infant per one marginal US dollar donated to Against Malaria Foundation? GiveWell estimated nets in the DRC would have the highest cost per life saved:$7,426.

Out of 13 lives saved, about 10 would be an infant (under age 5).

For a decade, GiveWell researched thousands of estimated factors in cost-effectiveness spreadsheets. The sound estimates established my confidence in the selection of top charities. I would like to imagine an estimated health result per marginal dollar donated to the Maximum Impact Fund.

• Miranda Kaplan on March 9, 2022 at 1:59 pm said:

Hi, Ethan!

Re: expected days added to an infant’s life per dollar donated to AMF, we’ve done our own quick calculation here, and arrived at roughly the same estimate as you for DRC. (Note that such an estimate doesn’t represent the full benefit of AMF’s program, since it excludes deaths averted in people over five, as well as development benefits—that is, increases in consumption due to reduced illness from malaria.)

We caution that, while our overall cost-effectiveness analyses give some idea of cost per life saved for each program, we now assess cost-effectiveness at the level of individual funding opportunities in specific places and times. Within AMF, for example, the most recent version of our analysis gives a range of costs per life saved (among all ages) from $4,113 to$7,426, depending on location. We publish backward-looking impact estimates for grants we’ve made from the Maximum Impact Fund or recommended to Open Philanthropy (see here), but we don’t necessarily recommend that donors attempt to calculate the impact of a generic individual donation to AMF, or any other of our recommended charities.

Re: the impact of a marginal dollar donated to the Maximum Impact Fund, this is also difficult to estimate in a forward-looking way, since the MIF may be allocated to any funding gap among our top charities that meets our cost-effectiveness threshold. That includes non-life-saving programs, such as deworming. We’ve calculated the average cost per life saved for all grants to life-saving top charities through the MIF in 2020 at around $4,500, but again, this is just a backward-looking average. Our estimated cost per life saved through MIF grants to life-saving charities in 2020 ranged from$1,188 to \$9,207 (see this column). Our best guess is that in the short term, future MIF grants will be similarly cost-effective to past grants, but that they’ll eventually become less cost-effective over time.