I like this post. I'm coming from an American perspective, where we have four branches of Quakerism and I'm in the one most similar to the British Quakers (liberal FGC), but we at least (and maybe British Quakers?) have something called worship sharing that might be a little more useful for EA stuff than true Meeting for Worship. Usually how it works is that people read some quotations or something like that and a couple of prompts, and then they do something Meeting-for-Worship-like for a bit of time (I think it shoul be ~15 minutes but some people seem to like shorter ones), and then everybody goes around in a circle sharing as they feel lead, if you have enough time people can share multiple times. Afterwards people can discuss stuff if they want but at least among Quakers I'm familiar with it operates on something kinda like Chatham House Rules, where you should keep the exact substance of people's comments private.
On a side note, I'm a bit surprised the reception to EA among Quakers is so positive? At least in my experience in the US Quakers tend to be against the quantification/measurement side of EA, although they definitely agree among altruism. Some of that is probably just my personal experience though.
Also, on the topic of econ papers with much bigger effects than I was expecting: a very simple savings exercise gives maize farmers in Zambia ~1 month more of expenses to use during the hungry season, and this lets them then have a better harvest the next year.
Eyal Frank (the guy behind the vultures paper) also has a paper finding pretty large effects from decreasing bat populations on infant mortality in the same vein- the channel here is that the bats eat insects that otherwise farmers use insecticide to treat, and the insecticide causes infant mortality (he doesn't say how but I think it's birth defects?). He also has one on killing sparrows leading to the Great Famine in China, but the solution to that problem is "don't put Mao in charge of your economy" which doesn't seem very relevant in a modern context.
I took a gander at looking into the vultures paper and my first impression is that their methodology isn't unreasonable, but I wouldn't expect the effects to be as big in other contexts. For one, a lot of the carrion in India apparently comes from cows, which are treated differently because of their status in Hindu religion and so are more likely to die naturally and less likely to be slaughtered. Plus, the mechanism they hypothesize is that dead animals end up in the water instead of being eaten by vultures, and so the effects would only hold up a) where water contamination is a major portion of the disease burden and b) where those animals can easily end up in the water instead of being eaten by vultures.
If people are interested, I might look into this further and write something up, but I'm not an expert on this and so far everything I saw looked like defensible choices.
This is too tangential from the forecasting discussion to justify being a comment there so I'm putting it here:
Forecasting makes no sense as a cause area, because cause areas are problems, something like "people lack resources/basic healthcare/etc.", "we might be building superintelligent AI and we have no idea what we're doing". Forecasting is more like a tool. People use forecasting to address AI, global poverty, and all sorts of more general problems, including ones that aren't major EA focuses.
For instance, we could treat vaccines as a cause area. All the funding to some AI-x-biosecurity people, GAVI campaigns for existing vaccines, and people working on bird flu vaccines could be treated like they're doing the same thing. And then we could argue about whether vaccines meet the funding bar. But that would be a pretty pointless argument, when really all those projects are trying to do different things with similar tools.
So I'd rather judge the AI forecasting by AI standards, the general-purpose forecasting by metascience standards, and the global development forecasting by global development standards, rather than trying to lump them in as a single entity. That being said, I do side with the view that there's too much money and enthusiasm being spent on forecasting, but it's a weakly held view, and that doesn't mean that every forecasting project isn't worth being funded, or even that they're all equally inflated.
Not an expert on this/haven't read all the prior discourse, but it seems like AIs acting like they're humans is a major cause of "LLM psychosis" now and is implicated in a lot of future concerns about AI takeovers due to making it seem like people should trust the AIs more than they really should, plus anything about being worried about AIs turning evil due to the representation of scheming AIs in their training text (as well as concerns about handing over the future to AIs that aren't actually moral patients). This work might make AIs act more human, or at least be useful for people who want to do that.
This post also criticizes AI 2027 (https://forum.effectivealtruism.org/posts/KgejNns3ojrvCfFbi/a-deep-critique-of-ai-2027-s-bad-timeline-models) and its critiques seem much more concerning? Including a bunch of links to papers that don't really back up points is not great practice or anything but also we don't have AI papers from 2027 yet, so I'd just presume that they were clumsily trying to go for something like "we'll have a better version of this paper in 2027" from what you've said.
Are there any organizations out there that would describe their niche as advising for small/medium-sized donors? I can't think of any, and I'm wondering why not. I'm not exactly sure what organizations that claim to advise large donors actually do, but it seems plausible that some things are also effective for smaller donors just because there are larger numbers of those. I'm thinking of, for instance:
I like this post. I'm coming from an American perspective, where we have four branches of Quakerism and I'm in the one most similar to the British Quakers (liberal FGC), but we at least (and maybe British Quakers?) have something called worship sharing that might be a little more useful for EA stuff than true Meeting for Worship. Usually how it works is that people read some quotations or something like that and a couple of prompts, and then they do something Meeting-for-Worship-like for a bit of time (I think it shoul be ~15 minutes but some people seem to like shorter ones), and then everybody goes around in a circle sharing as they feel lead, if you have enough time people can share multiple times. Afterwards people can discuss stuff if they want but at least among Quakers I'm familiar with it operates on something kinda like Chatham House Rules, where you should keep the exact substance of people's comments private.
On a side note, I'm a bit surprised the reception to EA among Quakers is so positive? At least in my experience in the US Quakers tend to be against the quantification/measurement side of EA, although they definitely agree among altruism. Some of that is probably just my personal experience though.
Also, on the topic of econ papers with much bigger effects than I was expecting: a very simple savings exercise gives maize farmers in Zambia ~1 month more of expenses to use during the hungry season, and this lets them then have a better harvest the next year.
Eyal Frank (the guy behind the vultures paper) also has a paper finding pretty large effects from decreasing bat populations on infant mortality in the same vein- the channel here is that the bats eat insects that otherwise farmers use insecticide to treat, and the insecticide causes infant mortality (he doesn't say how but I think it's birth defects?). He also has one on killing sparrows leading to the Great Famine in China, but the solution to that problem is "don't put Mao in charge of your economy" which doesn't seem very relevant in a modern context.
I took a gander at looking into the vultures paper and my first impression is that their methodology isn't unreasonable, but I wouldn't expect the effects to be as big in other contexts. For one, a lot of the carrion in India apparently comes from cows, which are treated differently because of their status in Hindu religion and so are more likely to die naturally and less likely to be slaughtered. Plus, the mechanism they hypothesize is that dead animals end up in the water instead of being eaten by vultures, and so the effects would only hold up a) where water contamination is a major portion of the disease burden and b) where those animals can easily end up in the water instead of being eaten by vultures.
If people are interested, I might look into this further and write something up, but I'm not an expert on this and so far everything I saw looked like defensible choices.
If you ever want to start the EA Quaker schism or whatever, I'm down to join (and know one more person in the overlap of the two)
This is too tangential from the forecasting discussion to justify being a comment there so I'm putting it here:
Forecasting makes no sense as a cause area, because cause areas are problems, something like "people lack resources/basic healthcare/etc.", "we might be building superintelligent AI and we have no idea what we're doing". Forecasting is more like a tool. People use forecasting to address AI, global poverty, and all sorts of more general problems, including ones that aren't major EA focuses.
For instance, we could treat vaccines as a cause area. All the funding to some AI-x-biosecurity people, GAVI campaigns for existing vaccines, and people working on bird flu vaccines could be treated like they're doing the same thing. And then we could argue about whether vaccines meet the funding bar. But that would be a pretty pointless argument, when really all those projects are trying to do different things with similar tools.
So I'd rather judge the AI forecasting by AI standards, the general-purpose forecasting by metascience standards, and the global development forecasting by global development standards, rather than trying to lump them in as a single entity. That being said, I do side with the view that there's too much money and enthusiasm being spent on forecasting, but it's a weakly held view, and that doesn't mean that every forecasting project isn't worth being funded, or even that they're all equally inflated.
Not an expert on this/haven't read all the prior discourse, but it seems like AIs acting like they're humans is a major cause of "LLM psychosis" now and is implicated in a lot of future concerns about AI takeovers due to making it seem like people should trust the AIs more than they really should, plus anything about being worried about AIs turning evil due to the representation of scheming AIs in their training text (as well as concerns about handing over the future to AIs that aren't actually moral patients). This work might make AIs act more human, or at least be useful for people who want to do that.
I'm not sure if I buy 7 (10% of pop. killed, but no takeover) being doom, but I can see the case for it
This post also criticizes AI 2027 (https://forum.effectivealtruism.org/posts/KgejNns3ojrvCfFbi/a-deep-critique-of-ai-2027-s-bad-timeline-models) and its critiques seem much more concerning? Including a bunch of links to papers that don't really back up points is not great practice or anything but also we don't have AI papers from 2027 yet, so I'd just presume that they were clumsily trying to go for something like "we'll have a better version of this paper in 2027" from what you've said.
I was being purposely kind of vague, but let's say people donating <100k a year? Whatever's too small for the organizations that advise large donors.
Are there any organizations out there that would describe their niche as advising for small/medium-sized donors? I can't think of any, and I'm wondering why not. I'm not exactly sure what organizations that claim to advise large donors actually do, but it seems plausible that some things are also effective for smaller donors just because there are larger numbers of those. I'm thinking of, for instance: