Thanks for the great post, Matthew!

This means that each dollar given to Givewell prevents about a trillion years of insect life! And these estimates have been pretty conservative. It means that you personally might prevent 5 quadrillion insects from coming into existence throughout the history of Earth.

I think you mean each life saved, not $ donated, prevents a 1 T arthropod-years (= 10^(22 - 10)). Likewise, I think you mean each person might prevent 1 T arthropod-years.

It’s not clear exactly what impact this has on how we should give.

I plan to publish a post about this on Tuesday.

Great post, Neel! I can see myself going back to it.

Thanks, Joel! The giving multiplier is now 10 instead of 13 as mentioned in the post, which you may want to update.

Hi Joel,

Just in case you are not aware, the link below is no longer publicly accessible.

For all our calculations and sources, refer to our spreadsheet (link).

Thanks, Joel! Do you also think your estimate that donating to Giving What We Can (GWWC) this year is 13 times as cost-effective as GiveWell's top charities is also 3 times as high as it should be, such that your best guess is that it is 4.33 (= 13/3) times as cost-effective as GiveWell's top charities (although there is large uncertainty)? Or is the adjustment only supposed to be applicable to CEARCH's CEAs listed here?

Thanks for clarifying, Joel! I plan to recommend people donate to CEARCH's High Impact Philanthropy Fund (HIPF) in a post I am writing which I will share in this thread once it is published. Is HIPF trying to avert as many DALYs as possible in a risk neutral way? If so, I do not have to recommend restricted donations. Do you have a guess for HIPF's marginal cost-effectiveness as a fraction of that of GiveWell's top charities? I would guess 55 as implied by CEARCH's CEA of advocating for taxing sugar-sweetened beverages (SSBs).

Impact, nuclear, and volcanic winters would decrease the number of wild animals a lot, but replacing forested area with cropland to produce more food would decrease them further.

Thanks, Joel! I guess you would recommend donating to Resolve to Save Lives (RTSL) in order to increase human-years as cost-effectively as possible.

Nuclear/volcanic winter famine mitigation is another candidate (CEA in the spreadsheet), though obviously there's a strong self-defeating element from a WAW perspective.

Expanding cropland is a great way to increase food production in nuclear and volcanic winters.

Thanks for the great context, Joel!

Which organisation would you recommend to someone wanting to maximise human-years in a fully causal and risk neutral way (the organisation does not have to work on non-communicable diseases (NCDs))? What is your best guess for its cost-effectiveness in terms of additional human-years per aditional $ spent as a fraction of that of GiveWell's top charities, which I estimate to be 0.0128 human-year/$? It would be great if you could briefly explain why, such as by linking to any supporting cost-effectiveness analyses (CEAs). I am asking because I suspect increasing human-years as cost-effectively as possible is the most cost-effective way to decrease negative animal-years of wild animals, via increasing cropland supporting food consumption.

I thought GiveWell was quite risk neutral considering their extensive funding of deworming. I also wonder why Open Philanthropy's (OP's) Global Public Health Policy (GPHP) team is not focussing on diabetes and hypertension.

Thanks for sharing, Mo! Very interesting. That makes me more pessimistic about finding the relationship between biological indicators and self-reported human welfare. I still think tracking more objective metrics would be helpful such that is is harder to game the system. If welfare surveys became widespread, and consistently used to make decisions, people could try to give answers which benefit them the most instead of reporting their welfare as accurately as possible. I like the assumption that welfare per human-year is proportional to the logarithm of annual consumption because this is hard to game.

• Your numbers imply silkworms have 2.50 (= 1.0*10^-5/(4*10^-6)) times as many neurons as black soldier flies. I expected adults to have more neurons than larvae, and ChatGPT guessed black soldier flies have 10 times as many neurons as silkworms.

@Laura Duffy, you assume the number of neurons of silkworms is 10^-5 that of humans citing Chiba et al. (2010), but these may have meant to imply a ratio of 10^-6 (= 10^(5 - 11)) for silkmoths. They say the "number of neurons of silkmoth brain is very few (about 10^5) compared with human brain (about 10^10)", which implies the number of neurons of silkmoths is 10^-5 that of humans. However, humans have 86 billion neurons, which is around 10^11.