To be clear (a genuine question, not a criticism, although I do strongly disagree), have you counted the cost-effectiveness of all donations to global health and development as zero because of possible harm to terrestrial arthropods?
If you really have that much uncertainty along that line of thinking, I'm not sure there's too much benefit in an analysis like this when of course most billionaires give most of their money to those initiatives. Happy to be pushed back on this though!
Hi Nick,
Yes, something along those lines. As of now, I am pretty clueless about the effects of global health and development interventions on terrestrial arthropods in the near term, and I am also quite unsure about their longterm effects.
This is a little hard for me too. Obviously, I feel a strong intuitive pull towards preventing deaths from malaria and malnutrition. In the past, I donated to GiveWell's top charities, and wrote articles in the online newspaper of my university applauding them (here and here; you can right click, and translate to English).
If you really have that much uncertainty along that line of thinking, I'm not sure there's too much benefit in an analysis like this
My hope was that the data about the donations could still be useful.
Just wondering how it is possible to be so unsure about the impact of global health interventions but still have „enough“ certainty regarding the positive impact of orgs like FLI? I mean there is still lots of stuff that can go wrong based on FLI interventions. Maybe that’s just the work that tipps us into an astronomical suffering scenario?
It seems rather arbitrary how you make those decisions. Imo, for this to have any value beyond being personal speculation, you should at least start to make explicit your reasoning process in more detail and also express the range of uncertainty you see. Maybe using conditionals as well to cover different scenarios.
Valuing Bill Gates philanthropy at 0 value outright without justification does not seem to be plausible or rigorous to me.
Hi Alexander,
I mean there is still lots of stuff that can go wrong based on FLI interventions. Maybe that’s just the work that tipps us into an astronomical suffering scenario?
I agree longtermist interventions are quite uncertain too. Moreover, I actually think they have wider confidence intervals for reasons like the one you pointed to. However, since they explicitly try to ensure the longterm effects are positive, and I believe most of the expected effects of interventions tend to be in the future, I guess the expected value of longtermist interventions is more likely to be positive than that of neartermist ones.
Imo, for this to have any value beyond being personal speculation, you should at least start to make explicit your reasoning process in more detail and also express the range of uncertainty you see.
I explained my process:
I only spent about 5 s setting the cost-effectiveness of each donation, guessing it based solely on the name of the recipient.
I agree it is not rigorous. This was supposed to be represented by elements like the title including "very shallow analysis" and the point in the summary saying (emphasis added only here, not in the summary):
This analysis is very (too?) shallow.
I have collected 65 donations made by the 10 richest people in the world, on 25 December 2022 according to this page from Forbes. I used the following sources, which I last checked on 25 December 2022:
I have included donations made together with partners, but excluded pledges. For the Wikipedia pages, I disregarded donations whose specific year or amount was unclear.
I set the cost-effectiveness of the donations in terms of increasing the expected value of the future. I only defined non-null values for those to:
I assumed null cost-effectiveness for the other donations due to very tentatively thinking they are neither robustly beneficial nor harmful. Feel free to check this post to get a sense of why I think this also applies to donations in the area of global health and development.
The table below contains the amount, social impact, and cost-effectiveness of the donations I collected, listed by descending social impact[1]. The calculations and data are in this Sheet.
Donor | Donations (G$) | Social impact (bp) | Cost-effectiveness (bp/G$) |
|---|---|---|---|
| Jeff Bezos | 10.3 | 0.640 | 0.0620 |
| Elon Musk | 5.82 | 0.0395 | 0.00679 |
| Bill Gates | 20.4 | 0 | 0 |
| Carlos Slim Helu & family | 8.00 | 0 | 0 |
| Warren Buffett | 1.08 | 0 | 0 |
| Steve Ballmer | 0.728 | 0 | 0 |
| Larry Ellison | 0.0316 | 0 | 0 |
| Gautam Adani & family | 0.0138 | 0 | 0 |
| Bernard Arnault & family | 0 | 0 | ND |
| Mukesh Ambani | 0 | 0 | ND |
Jeff Bezos and Elon Musk top the list. This is in agreement with my prior expectations. However, the present analysis is very (too?) shallow:
The social impact of 0.680 bp I estimated for the donations of the 10 richest people in the world is equivalent to donating 172 M$ (= 0.680/3.95*10^3) to the Longtermism Fund, supposing its cost-effectiveness is 3.95 bp/G$ as I estimated here for the spending of the effective altruism community on longtermism and catastrophic risk prevention.
Another shallow analysis was published here by Nuño Sempere, and a proposal for a more detailed one was presented here by Elliot Olds.
In case of null social impact, I listed the names by descending amount donated. If this was also null, I listed the names alphabetically. ND stands for not defined.
I think Bill Gates' donations are probably very high impact, not zero.
Example source:
Against Against Billionaire Philanthropy / Scott Alexander
In your sheet, you seem to count it as zero (without any formula, it's hard coded zero)
Hi Yonatan,
Thanks for commeting. I agree the Gates Foundation has saved many lives, but I am very unsure about the sign of global health and development interventions for the reasons I point to here.