I like your analysis, but the numbers here feel made up at worst and unexplained at best. It would have been better to look at real numbers rather than speculating.
The most concrete number I think we can consider are donation figures from the 2013 EA Survey. We tracked $5.2M in giving, with $2M (38% of donations) coming from the top one person (0.1% of the sample), and the top 10% of donors in our sample accounted for 90% of the donations (being in the top 10% of our sample required donating $8.7K/yr).
This would suggest that landing one big fish (e.g., Dustin Moskovitz and Good Ventures, not in our sample, but worth ~$5B) is on par with the entire rest of the EA movement from the limited donations-only perspective.
However, this does not mean that it is right to focus more on landing “big fish” as you’re right that:
(a) moving a typical $1k/yr donor from ineffective charity to AMF charity is likely higher ROI (if you believe the 1000x effectiveness multiplier) than moving a $1k/yr donor from AMF to $10k/yr to AMF (10x multiplier).
(b) the incentive effects and PR effects on the overall movement may still be negative enough.
(c) work on drawing in a large movement also helps draw in the big fish because they’re attracted to the movement size.
I’d be curious if the model could be refined or adjusted. You’re right that this contributors follow a Pareto distribution, but it’s not obvious to me what this would mean in the model. It could mean that it’s worth recruiting lots of new people because one may be a ‘big fish’, or it could mean that it’s more worth taking people who are doing very well and making sure they do better. It depends on other factors.
I guess this would help us to at least look at the model through the lens of the most high-impact (or people with ‘potential to be high impact’, whatever that may mean) people.
That said, the really big question is what the purpose of the model is. What is the decision being made? If it is being used by ‘average ea-forum visitors’ to do local work, it’s quite different from a model aimed at a particular EA org. For instance, many people may not really be able to persuade or help those in the top 3%, but they may be able to make great gains with the others.
I hear you about the numbers. As I stated above, I specifically used numbers that came from highly-upvoted comments not made by myself, to minimize the bias of my own personal take on the matter.
I actually considered using the numbers on the surveys, but I found they didn’t capture volunteer efforts. A large proportion of people who identified as EA members reported not giving any money at all. Yet since they considered themselves EA movement participants, I felt it important to avoid leaving them out. Moreover, I personally know a number of EA members who give their time and skills as opposed to money. This aligns with the recent great piece by Ben Todd about talent being more important than money.
Moreover, in alignment with Robert Wiblin’s comments, I think that a number of EA projects may end up doing less good than it may appear right now. So I put personally a higher level of value on simply being involved with the EA movement, since I have a strong belief that over time, the EA movement will figure out more optimal ways of estimating the impact of various interventions, and then we as a movement can update and shift our giving. Similarly, it’s important for more non-EAs to behave like EAs and give effectively—not as part of the movement, but influences by the memes of the movement to update their giving choices based on shifting evidence.
I expect time would be equally disproportionate, with the people working at EA orgs generating a huge % of the value of “EA time”. I would expect if you surveyed EA orgs there would be broad agreement about staff vs volunteer value.
I like your analysis, but the numbers here feel made up at worst and unexplained at best. It would have been better to look at real numbers rather than speculating.
The most concrete number I think we can consider are donation figures from the 2013 EA Survey. We tracked $5.2M in giving, with $2M (38% of donations) coming from the top one person (0.1% of the sample), and the top 10% of donors in our sample accounted for 90% of the donations (being in the top 10% of our sample required donating $8.7K/yr).
This would suggest that landing one big fish (e.g., Dustin Moskovitz and Good Ventures, not in our sample, but worth ~$5B) is on par with the entire rest of the EA movement from the limited donations-only perspective.
However, this does not mean that it is right to focus more on landing “big fish” as you’re right that:
(a) moving a typical $1k/yr donor from ineffective charity to AMF charity is likely higher ROI (if you believe the 1000x effectiveness multiplier) than moving a $1k/yr donor from AMF to $10k/yr to AMF (10x multiplier).
(b) the incentive effects and PR effects on the overall movement may still be negative enough.
(c) work on drawing in a large movement also helps draw in the big fish because they’re attracted to the movement size.
I’d be curious if the model could be refined or adjusted. You’re right that this contributors follow a Pareto distribution, but it’s not obvious to me what this would mean in the model. It could mean that it’s worth recruiting lots of new people because one may be a ‘big fish’, or it could mean that it’s more worth taking people who are doing very well and making sure they do better. It depends on other factors.
I guess this would help us to at least look at the model through the lens of the most high-impact (or people with ‘potential to be high impact’, whatever that may mean) people.
That said, the really big question is what the purpose of the model is. What is the decision being made? If it is being used by ‘average ea-forum visitors’ to do local work, it’s quite different from a model aimed at a particular EA org. For instance, many people may not really be able to persuade or help those in the top 3%, but they may be able to make great gains with the others.
I hear you about the numbers. As I stated above, I specifically used numbers that came from highly-upvoted comments not made by myself, to minimize the bias of my own personal take on the matter.
I actually considered using the numbers on the surveys, but I found they didn’t capture volunteer efforts. A large proportion of people who identified as EA members reported not giving any money at all. Yet since they considered themselves EA movement participants, I felt it important to avoid leaving them out. Moreover, I personally know a number of EA members who give their time and skills as opposed to money. This aligns with the recent great piece by Ben Todd about talent being more important than money.
Moreover, in alignment with Robert Wiblin’s comments, I think that a number of EA projects may end up doing less good than it may appear right now. So I put personally a higher level of value on simply being involved with the EA movement, since I have a strong belief that over time, the EA movement will figure out more optimal ways of estimating the impact of various interventions, and then we as a movement can update and shift our giving. Similarly, it’s important for more non-EAs to behave like EAs and give effectively—not as part of the movement, but influences by the memes of the movement to update their giving choices based on shifting evidence.
Agreed on points a, b, and c.
I expect time would be equally disproportionate, with the people working at EA orgs generating a huge % of the value of “EA time”. I would expect if you surveyed EA orgs there would be broad agreement about staff vs volunteer value.
Just to clarify, I’m talking about those who donate their resources of time and talents, not those who get paid for it.