Hi Larks! A lot to unpack here but in general, we only decided to start a family planning organization after we concluded that it also has a positive near-term effect on human well-being, and when it comes to long-term, very uncertain flow-through effects, we don’t strongly take them into account for any of the interventions we evaluate because they are too uncertain, but we gladly will when more evidence is available. I will address your specific points below.
In 2018 you published a report (now deleted from the internet) that advocated for reducing human populations in the third world in order to reduce meat consumption.
[...]
I see you have continued to do work on charities that would reduce human populations, though without making as explicit that the original motivation was not so much to help people directly but rather to reduce their number.
First of all, just a clarification here, we are not focusing on reducing human populations to start with, but on reducing suffering and increasing well-being. Also, we did not assume that it would be an advocacy for reducing human populations in the third world specifically. As the summary at the top says: “The intervention ended up looking surprisingly impactful for animals, particularly if conducted in countries with a high need for contraceptives and high fish and poultry consumption.” We took into account all possible countries and applied those criteria. Usually, countries that have lower GDPs also have higher unmet needs, including for contraceptives, so they scored high on our priority country research, but some high-income countries were also at the top, for example Israel, so we did not have any agenda targeted at developing countries.
When the CE team writes about family planning, it does includes all the effects, including animal welfare. For example, in the post you linked, in the table of effects, you can see the impact on animals. Welfare Points are the unit we use to evaluate animal welfare organizations. So I don’t know where the concern about hiding that reason is coming from.
Yes, this report just focused on animal welfare effects. We were aware of previous work analysing family planning for its human health effects, but none that analysed the animal effects, so we decided to start our analysis from that perspective. Note that we did not start any family planning charities until we looked into these effects deeper and focused on, the near-term effects on humans as well. The timeline was:
2018 - We analysed the animal effects of family planning interventions
2019 - mid No family planning charities were incubated.
Later 2019/ early 2020 - we decided to look deeper into family planning, focusing on the effects that it had on human health, education, income etc. We published this post explaining why we are researching family planning and outlining those effects. After looking into those positive effects on people’s well-being, we were convinced to do a round of research on family planning interventions.
Mid/end 2020 - We incubated our first family planning charity.
So we only started a family planning organization after looking into the near-term human well-being effects and concluded that they are positive.
Additionally, as Charles mentioned, we have incubated many other charities that directly aim at improving people’s health (e.g., LEEP, Fortify Health) and well as those that save people’s life (e.g., Suvita) in low-middle income countries.
Some subsequent reports do not mention the impact on animals at all, even though this initially seemed to be the primary motivation. Perhaps this is because you planned to obscure this fact
Note that this post is not written or published by the CE team. It was written by Sarah and Ben who are co-founding a post-partum family planning organization. This post was written by them to outline their motivations and the effects that they are counting.
This was criticised by several people, both for ignoring flow-through effects (like existential risks, wild animal suffering, or long run growth, or population ethics) and for seeming dishonest about your true motivations / resembling eugenics. [...]
Honesty:
Promoting a intervention based on analysis X (e.g. neonatal health) when the actual reason you believe in it is Y (animal welfare), and being intentionally misleading about this fact.
[...]
in a contrived, remote way, these could lead to people slipping in implausible, extremely negative associations
It’s not a crux for me so I don’t want to go to deeply into this point, but I don’t think it is that implausible that normal people might connect ‘we want to reduce populations in the third world’ with ‘eugenics’.
Those two points you are making are contradictory to each other, and we see that trade-off as well, that’s why we previously said that we “prefer to discuss it in conversation rather than in writing” because that enables more open and honest discussion without having to worry about potential PR risks to the EA movement.
But in balancing those two, I don’t see evidence that the CE team has been intentionally misleading about it. Our first post was explicit that is was an analysis of the animal welfare effects of family planning. We took it down from the EA forum and the website because we have been advised that it may be potentially a PR risk for the EA movement. We agreed and took that post down, however, we continued to express that those effects are also taken into account when analysing the promise of the intervention, as you can see in the post you linked and the above screenshot, to be honest about the effects we take into account, even though it was recommended as an organization after we looked into nearterm human wellbeing effects as well.
Flow-through:
Analyzing in depth one positive flow-through effect (animal welfare) but not other flow through effects that plausibly are very large and negative (e.g. existential risk, wild animal welfare, growth, population ethics). (Unless there is unpublished research on this subject, hence my asking).
I think the question of how to treat flow-through effects is really hard, and people have different perspectives on it (Graves, St Jules, Wildeford). In general:
For the effects that we have evidence for, such as human health, animal welfare, and income effects on the household, we model them explicitly.
When it comes to population ethics, this is particularly tricky. We started to work on a project that would take it into account more explicitly using a moral weights framework. Some inputs we were thinking about including were pretty standard: “Averting the death of an individual under-5 is equivalent to doubling consumption for one individual for how many years?” or “Averting the death of an individual under-5 is equivalent to how many DALYs?” Other inputs were more unusual: “Preventing an individual from being born who would live in a population scoring 3⁄10 on life satisfaction is equivalent to how many deaths of an individual under-5 averted?”, “Preventing an individual from being born who would live in a population scoring 4⁄10 on life satisfaction is equivalent to how many deaths of an individual under-5 averted?” or “Averting the death of an individual under-5 is worth preventing how many factory farmed chickens from being born?” But we decided that it is too complicated and that we are not going to “solve population ethics.” We also learned that Rethink Priorities is undertaking research into moral weights and tradeoffs against humans and animals, and as an org solely focused on research, they will be in a better spot to address those sorts of questions and we have been tracking the progress of that work. Given that population ethics seems to be one of the least tractable problems in practical ethics and that smart aligned people widely disagree about it, we are leaning more toward a “balanced portfolio” approach (that has been described by Michael St Jules). For example, we can aim to balance life-saving interventions (example of CE charity—Suvita) with family planning interventions (example of CE charity—Family Empowerment Media) and aim at minimizing total externalities on population size in our charity portfolio. And even though it is hard to do it precisely, we are aiming at that balance. In the future, as more progress will be made on moral weights and how to include population effects in evaluations of interventions, we will be updating our approach, as long as it will be consistent with how we in general want to treat uncertain effects.
On wild animals- I would like to include those effects. Unfortunately, there is currently very little evidence on it. I’m not aware of anyone who evaluates global health interventions that would include wild animal effects, or any nuclear security interventions that take into account farmed or wild animal welfare effects. The reason is that it is hard and the effects are uncertain to the level that sign of the effect can flip very quickly. So until more research is available on that, we are not able to include those effects in our evaluations. I’m excited about any efforts that go into that, and in the future, as more evidence comes out, we will gladly include that in our analysis as well.
Until all of those effects are studied, we plan to approach it from a worldview diversification approach and aim at balancing the portfolio. This is similar to how it is being treated in EA more broadly e.g., GiveWell doesn’t model long-term populations effects, AI safety work rarely, if at all, takes into account effects on farmed animals (Peter Singer& Yip Fai Tse, 2022), farmed animal evaluations don’t take into account effects on wild animals, and nuclear risk analysis doesn’t take into account potential positive effects in the reduction in the number of factory farmed animals, etc. Ideally, we all would do it, but for now, it is too impractical to do.
All in all, I recognize this is a tricky issue and that in our communication, we have to balance being transparent about what effects we take into account (human health, animal welfare etc.) whilst also managing potential PR risks for the EA community, and in our approach to flow-through effects we have to take into account the amount of evidence that is available for those effects and at the same time not completely neglect uncertain effects.
Hi Larks! A lot to unpack here but in general, we only decided to start a family planning organization after we concluded that it also has a positive near-term effect on human well-being, and when it comes to long-term, very uncertain flow-through effects, we don’t strongly take them into account for any of the interventions we evaluate because they are too uncertain, but we gladly will when more evidence is available. I will address your specific points below.
First of all, just a clarification here, we are not focusing on reducing human populations to start with, but on reducing suffering and increasing well-being. Also, we did not assume that it would be an advocacy for reducing human populations in the third world specifically. As the summary at the top says: “The intervention ended up looking surprisingly impactful for animals, particularly if conducted in countries with a high need for contraceptives and high fish and poultry consumption.” We took into account all possible countries and applied those criteria. Usually, countries that have lower GDPs also have higher unmet needs, including for contraceptives, so they scored high on our priority country research, but some high-income countries were also at the top, for example Israel, so we did not have any agenda targeted at developing countries.
When the CE team writes about family planning, it does includes all the effects, including animal welfare. For example, in the post you linked, in the table of effects, you can see the impact on animals. Welfare Points are the unit we use to evaluate animal welfare organizations. So I don’t know where the concern about hiding that reason is coming from.
Yes, this report just focused on animal welfare effects. We were aware of previous work analysing family planning for its human health effects, but none that analysed the animal effects, so we decided to start our analysis from that perspective. Note that we did not start any family planning charities until we looked into these effects deeper and focused on, the near-term effects on humans as well. The timeline was:
2018 - We analysed the animal effects of family planning interventions
2019 - mid No family planning charities were incubated.
Later 2019/ early 2020 - we decided to look deeper into family planning, focusing on the effects that it had on human health, education, income etc. We published this post explaining why we are researching family planning and outlining those effects. After looking into those positive effects on people’s well-being, we were convinced to do a round of research on family planning interventions.
Mid/end 2020 - We incubated our first family planning charity.
So we only started a family planning organization after looking into the near-term human well-being effects and concluded that they are positive.
Additionally, as Charles mentioned, we have incubated many other charities that directly aim at improving people’s health (e.g., LEEP, Fortify Health) and well as those that save people’s life (e.g., Suvita) in low-middle income countries.
Note that this post is not written or published by the CE team. It was written by Sarah and Ben who are co-founding a post-partum family planning organization. This post was written by them to outline their motivations and the effects that they are counting.
Those two points you are making are contradictory to each other, and we see that trade-off as well, that’s why we previously said that we “prefer to discuss it in conversation rather than in writing” because that enables more open and honest discussion without having to worry about potential PR risks to the EA movement.
But in balancing those two, I don’t see evidence that the CE team has been intentionally misleading about it. Our first post was explicit that is was an analysis of the animal welfare effects of family planning. We took it down from the EA forum and the website because we have been advised that it may be potentially a PR risk for the EA movement. We agreed and took that post down, however, we continued to express that those effects are also taken into account when analysing the promise of the intervention, as you can see in the post you linked and the above screenshot, to be honest about the effects we take into account, even though it was recommended as an organization after we looked into nearterm human wellbeing effects as well.
I think the question of how to treat flow-through effects is really hard, and people have different perspectives on it (Graves, St Jules, Wildeford). In general:
For the effects that we have evidence for, such as human health, animal welfare, and income effects on the household, we model them explicitly.
When it comes to population ethics, this is particularly tricky. We started to work on a project that would take it into account more explicitly using a moral weights framework.
Some inputs we were thinking about including were pretty standard: “Averting the death of an individual under-5 is equivalent to doubling consumption for one individual for how many years?” or “Averting the death of an individual under-5 is equivalent to how many DALYs?”
Other inputs were more unusual: “Preventing an individual from being born who would live in a population scoring 3⁄10 on life satisfaction is equivalent to how many deaths of an individual under-5 averted?”, “Preventing an individual from being born who would live in a population scoring 4⁄10 on life satisfaction is equivalent to how many deaths of an individual under-5 averted?” or “Averting the death of an individual under-5 is worth preventing how many factory farmed chickens from being born?”
But we decided that it is too complicated and that we are not going to “solve population ethics.” We also learned that Rethink Priorities is undertaking research into moral weights and tradeoffs against humans and animals, and as an org solely focused on research, they will be in a better spot to address those sorts of questions and we have been tracking the progress of that work.
Given that population ethics seems to be one of the least tractable problems in practical ethics and that smart aligned people widely disagree about it, we are leaning more toward a “balanced portfolio” approach (that has been described by Michael St Jules). For example, we can aim to balance life-saving interventions (example of CE charity—Suvita) with family planning interventions (example of CE charity—Family Empowerment Media) and aim at minimizing total externalities on population size in our charity portfolio. And even though it is hard to do it precisely, we are aiming at that balance. In the future, as more progress will be made on moral weights and how to include population effects in evaluations of interventions, we will be updating our approach, as long as it will be consistent with how we in general want to treat uncertain effects.
On wild animals- I would like to include those effects. Unfortunately, there is currently very little evidence on it. I’m not aware of anyone who evaluates global health interventions that would include wild animal effects, or any nuclear security interventions that take into account farmed or wild animal welfare effects. The reason is that it is hard and the effects are uncertain to the level that sign of the effect can flip very quickly. So until more research is available on that, we are not able to include those effects in our evaluations. I’m excited about any efforts that go into that, and in the future, as more evidence comes out, we will gladly include that in our analysis as well.
Until all of those effects are studied, we plan to approach it from a worldview diversification approach and aim at balancing the portfolio. This is similar to how it is being treated in EA more broadly e.g., GiveWell doesn’t model long-term populations effects, AI safety work rarely, if at all, takes into account effects on farmed animals (Peter Singer& Yip Fai Tse, 2022), farmed animal evaluations don’t take into account effects on wild animals, and nuclear risk analysis doesn’t take into account potential positive effects in the reduction in the number of factory farmed animals, etc. Ideally, we all would do it, but for now, it is too impractical to do.
All in all, I recognize this is a tricky issue and that in our communication, we have to balance being transparent about what effects we take into account (human health, animal welfare etc.) whilst also managing potential PR risks for the EA community, and in our approach to flow-through effects we have to take into account the amount of evidence that is available for those effects and at the same time not completely neglect uncertain effects.