EA Forum Podcast (All audio)

This report is a “shallow” investigation, as described here, and was commissioned by Open Philanthropy and originally produced by Rethink Priorities from September to October 2024. Open Philanthropy does not necessarily endorse our conclusions, nor do the organizations represented by those who were interviewed.

Our report focuses on exploring health systems strengthening (HSS) as a potential new cause area for Open Philanthropy. We examined a range of interventions to improve health system performance, analyzed case studies of successful systems in low- and middle-income countries, and highlighted promising opportunities for philanthropic funding. We reviewed the scientific and gray literature and spoke with six experts to describe the state of the field as of late 2024. In early 2025, we revised the report for publication; we also added a short addendum based on a second round of interviews with a different group of experts to incorporate updates related [...]

---

Outline:

(00:12) Executive summary

(00:16) Scope

(00:40) Research process

(01:34) Key findings

(07:02) Key uncertainties

(08:17) Health systems strengthening as a cause area

(08:21) Shortcomings in LMIC health systems are large and comparatively neglected

(14:44) What is health systems strengthening?

(16:57) A taxonomy of typical HSS interventions

(22:05) Challenges in evaluating HSS interventions and their cost-effectiveness

(24:30) Country case studies: Lessons from high-performing health systems

(27:39) Bangladesh

(31:33) Thailand

(35:32) Ethiopia

(41:09) Ghana

(44:33) Kerala

(47:31) Valuable case studies recommended by experts

(48:14) Rwanda

(49:27) Sri Lanka and Cuba

(49:53) South Africa

(50:35) Vietnam

(50:55) Highlighted HSS interventions

(50:59) Our research and prioritization process

(55:29) Integrated Management of Childhood Illness (IMCI)

(55:34) What is it?

(56:56) Reasons in favor

(59:22) Reasons against

(01:01:06) Crucial considerations

(01:02:13) Potential grants

(01:02:40) Our take

(01:04:03) Supervision and mentorship for IMCI providers

(01:04:09) What is it?

(01:04:47) Reasons in favor

(01:07:44) Reasons against

(01:08:39) Potential grants

(01:10:07) Our take

(01:10:59) Community health workers (CHW) / service extensions and health posts

(01:11:06) What is it?

(01:12:17) Reasons in favor

(01:15:26) Reasons against

(01:17:30) Crucial Considerations

(01:19:43) Potential grants

(01:21:03) Our take

(01:21:38) Social accountability interventions / Community score cards

(01:21:44) What is it?

(01:22:23) Reasons in favor

(01:25:36) Reasons against

(01:26:57) Crucial Considerations

(01:27:40) Potential grants

(01:28:20) Our take

(01:29:14) Continuous Quality Improvement (CQI)

(01:29:19) What is it?

(01:30:46) Reasons in favor

(01:32:23) Reasons against

(01:33:13) Crucial Considerations

(01:34:15) Potential grants

(01:34:28) Our take

(01:35:05) Drug delivery system reform

(01:35:09) What is it?

(01:36:11) Reasons in favor

(01:38:13) Reasons against

(01:40:28) Crucial considerations

(01:41:31) Potential grants

(01:42:47) Our take

(01:43:19) Interventions discussion

(01:43:23) Our impression from the studied interventions

(01:45:22) Leadership & governance

(01:47:01) Health workforce

(01:48:25) Financing

(01:49:47) Health information

(01:51:35) Supply chain strengthening

(01:52:20) Service delivery

(01:53:51) Other recommendations from experts

(01:56:04) Spotlight on two implementing organizations

(01:57:48) mPharma

(01:57:51) What is it and what does it do?

(02:00:01) How well does it work?

(02:01:14) Our takes

(02:02:29) Last Mile Health

(02:02:33) What is it and what does it do?

(02:04:22) How well does it work?

(02:05:50) Our takes

(02:06:24) Recommendations for further exploration

(02:08:51) Higher-level takes on health systems strengthening

(02:10:43) Reasons HSS interventions might be more effective than vertical programs

(02:10:50) Broader, longer-term, and leveraged impact

(02:11:45) Resilience to respond to crises

(02:13:12) Reduction of duplication and administrative overhead

(02:14:02) Addendum: Reflections on USAID withdrawal and emerging gaps in HSS

The original text contained 44 footnotes which were omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

James Özden and Sam Glover at Social Change Lab wrote a literature review on protest outcomes[1] as part of a broader investigation[2] on protest effectiveness. The report covers multiple lines of evidence and addresses many relevant questions, but does not say much about the methodological quality of the research. So that's what I'm going to do today.

I reviewed the evidence on protest outcomes, focusing only on the highest-quality research, to answer two questions:

Here's what I found:

Do protests work? Highly likely (credence: 90%) in certain contexts, although it's unclear how well the results generalize. [More]

Are Social Change Lab's conclusions consistent with the highest-quality evidence? Yes—the report's core claims are well-supported, although it overstates the strength of some of the evidence. [More]

Introduction

This [...]

---

Outline:

(01:16) Introduction

(04:07) Studies on real-world protest outcomes

(09:12) Madestam et al. (2013) on Tea Party protests

(12:09) Placebo tests

(13:42) Wasow (2020) on 1960s civil rights protests

(15:12) Klein Teeselink & Melios (2021) on 2020 Black Lives Matter protests

(17:42) Failed placebo tests

(19:54) Larreboure & González (2021) on the Womens March

(22:18) Hungerman & Moorthy (2023) on Earth Day

(25:13) Meta-analysis

(33:05) Potential problems with the research

(33:09) Spatial autocorrelation

(34:22) Publication bias

(38:27) Data fabrication

(40:17) Data errors

(41:25) Will the results generalize?

(42:18) Meta-concerns with this meta-analysis

(43:01) Are Social Change Labs claims justified?

(43:06) Broad claims

(48:36) Claims about individual studies

(52:02) Conclusion

(52:51) Source code

(52:59) Appendix A: Additional tables

(57:18) Appendix B: Methodological revisions

(59:38) Appendix C: Comparing the strength of evidence to saturated fat research

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Need help planning your career? Probably Good's 1-1 advising service is back! After refining our approach and expanding our capacity, we’re excited to once again offer personal advising sessions to help people figure out how to build careers that are good for them and for the world.

Our advising is open to people at all career stages who want to have a positive impact across a range of cause areas—whether you're early in your career, looking to make a transition, or facing uncertainty about your next steps. Some applicants come in with specific plans they want feedback on, while others are just beginning to explore what impactful careers could look like for them. Either way, we aim to provide useful guidance tailored to your situation.

Learn more about our advising program and apply here.

Also, if you know someone who might benefit from an advising call, we’d really appreciate [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

My PhD is in evolutionary biology, and I spent much of my time learning and thinking about genetic governance mechanisms like meiosis. I find myself wanting a link to share on these concepts, so I'm publishing some old writing I did in 2022 that makes the comparison.

Gradient descent and natural selection are analogous

Gradient descent can be compared to natural selection. Both are optimization algorithms. But while gradient descent is only decades old as part of the field of machine learning (ML), natural selection is the reason complex living beings like human readers exist. Are there any lessons we can learn about alignment by looking at the products of natural selection?

Oly Sourbut has written up a set of conditions under which natural selection and gradient descent are equivalent, i.e., where the same inputs will lead to the same outputs under both algorithms [...]

---

Outline:

(00:32) Gradient descent and natural selection are analogous

(01:38) Gradient descent can be hacked

(03:09) There is something like gradient hacking in biology: meiotic drive

(05:34) How meiotic drive works

(08:30) Recombination is the major genetic alignment technology

(09:46) Examples of meiotic drive

(17:46) Losing the ability to do meiosis is a death sentence for a species

(19:45) Acknowledgments

---

First published:

Source:

Linkpost URL:
https://hollyelmore.substack.com/p/genes-did-misalignment-first-comparing

---

Narrated by TYPE III AUDIO.

EA-aligned animal advocacy is often criticized for measurability bias. In this post, I argue that the usual epistemic safeguards of leftist advocacy are unavailable in animal advocacy, and as a result, measurability is doubly important in our domain.

What do pro-measurement people say?

Intuition and abstract reasoning are very poor predictors of charity performance. Furthermore, charities differ 100x in effectiveness, therefore measuring the cost-effectiveness of charities enables us to do much more good.

We need careful scientific studies and preferably RCTs to figure out which interventions work. Through measurement, we will have a transparent accountability mechanism which will help us identify cases of failure and stop funding things that don't work.

What are the usual criticisms of measurability bias?

• Not all important outcomes are measurable. How do you quantify the value of a social movement, a shift in public discourse, or groundbreaking legal precedent?
• External measurement misses [...]

---

Outline:

(00:27) What do pro-measurement people say?

(01:00) What are the usual criticisms of measurability bias?

(01:52) What is the main leftist alternative to measurable outcomes?

(02:35) You cant do standpoint epistemology with non-human animals

(03:27) Balancing measurement with movement building

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

Saw this post by @Henri Thunberg 🔸 on Twitter:

If true, these numbers are suprisingly low to me. Eyestalk ablation gets quite a bit of attention on the forum. Shrimp Welfare Project focuses on it and it forms part of their Shrimp Welfare Index. It gets attention from various animal welfare groups (Mercy For Animals, Animals Australia). A post on Reddit about it from a couple of days ago has 35 k upvotes and 2000 comments. An Australian politician recently posted about it. A lot of this attention, I suspect, is based on the false impression that this is happening to most or all of the shrimp.

The numbers of animals involved seems like pertinent information but I can't see the number mentioned on the Shrimp Welfare Project website, nor on any of their forum posts. Rethink Priorities wrote a report here that on page 13 seems to estimate [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Three Epoch employees – Matthew Barnett, Tamay Besiroglu, and Ege Erdil – have left to launch Mechanize, an AI startup aiming for broad automation of ordinary labour:

Today we’re announcing Mechanize, a startup focused on developing virtual work environments, benchmarks, and training data that will enable the full automation of the economy.

We will achieve this by creating simulated environments and evaluations that capture the full scope of what people do at their jobs. ...

Currently, AI models have serious shortcomings that render most of this enormous value out of reach. They are unreliable, lack robust long-context capabilities, struggle with agency and multimodality, and can’t execute long-term plans without going off the rails.

To overcome these limitations, Mechanize will produce the data and evals necessary for comprehensively automating work. Our digital environments will act as practical simulations of real-world work scenarios, enabling agents to learn useful abilities through RL. [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

If you’re interested in having a meaningful EA career but your experience doesn’t match the types of jobs that the typical white collar, intellectual EA community leans towards, then you’re just like me.

I have been earning to give as a nuclear power plant operator in Southern Maryland for the past few years, and I think it's a great opportunity for other EA's who want to make a difference but don’t have a PhD in philosophy or public policy.

Additionally, I have personal sway with Constellation Energy's Calvert Cliffs plant, so I can influence the hiring process to help any interested applicants.

Here are a few reasons that I think this is such an ideal Earn to Give career:

• A high income job in a low cost of living area means you will be able to donate a significant portion of your paychecks and still live comfortably. [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

Exactly one year after receiving our seed funding upon completion of the Charity Entrepreneurship program, we (Miri and Evan) look back on our first year of operations, discuss our plans for the future, and launch our fundraising for our Year 2 budget.

Family Planning could be one of the most cost-effective public health interventions available. Reducing unintended pregnancies lowers maternal mortality, decreases rates of unsafe abortions, and reduces maternal morbidity. Increasing the interval between births lowers under-five mortality. Allowing women to control their reproductive health leads to improved education and a significant increase in their income. Many excellent organisations have laid out the case for Family Planning, most recently GiveWell.[1]

In many low and middle income countries, many women who want to delay or prevent their next pregnancy can not access contraceptives due to poor supply chains and high costs. Access to Medicines Initiative (AMI) was incubated by [...]

---

Outline:

(00:15) TL;DR

(01:28) The Problem

(03:30) Year 1 Retrospective

(03:34) Pilot

(07:19) Preliminary Pilot Results

(07:23) Increasing the Total Amount of Contraceptives

(09:57) Investigating the Distribution of Contraceptives

(11:37) Year 1 Spending

(12:44) Updated Cost-Effectiveness Analysis at Scale

(15:59) Our Strategy for Year Two (And Beyond)

(17:00) Expanding Our Procurement of Contraceptives

(18:21) Providing Technical Assistance for State-Funded Procurement of Contraceptives

(20:12) Improving and Expanding our Monitoring & Evaluation

(21:53) Projected Cost-Effectiveness of Our Year 2 Work

(22:38) Budget and Funding Gap

(24:00) Why Support Access to Medicines Initiative?

(26:50) How You Can Help

(27:50) Acknowledgments

The original text contained 7 footnotes which were omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

We’re holding a DIY debate week on the Forum, from April 28–May 2nd.

It's been great seeing and hearing people responding positively to debate weeks (1,2,3), writing posts, getting into discussions, and voting in large numbers.

I especially like that voting on the banner prompts people to leave a comment, which then starts discussions. I’d love to know if that vote→conversation pipeline holds if we let everyone make polls.

So… we are releasing a poll-making feature for everyone to use. We’re holding this DIY debate week to celebrate that release, during which we’ll ask users to make and discuss polls.

Throughout the week, the Forum team will promote polls we think are particularly valuable. This could be by pinning them on the Frontpage, linking them on social media, putting them in the Forum Digest, etc...

The polls will look like our debate week [...]

---

Outline:

(00:18) What's a DIY debate week?

(01:27) How can I take part?

(02:04) How does the poll feature work?

(02:48) Putting a poll in your post

(03:33) Interacting with the poll as a reader

The original text contained 1 footnote which was omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.