Hi Matthew. Nice to know how you are thinking about this.

TLDR: my ranking for donations: Longtermist political donations>growing EA>other Longtermist donations>animal welfare>global health.

My ranking for careers: Growing EA=Longtermist careers>animal welfare>global health.

By growing EA, you mean growing longtermist EA? What makes a donation or career longtermist? If you think the longterm benefits of decreasing the risk of human extinction over the next few 10 years are much larger than its nearterm benefits, you should also think the longterm benefits of animal welfare and global health interventions are much larger than their nearterm benefits? If so, how do you compare the longterm benefits in a principled way? You estimated "each dollar [donated to longtermist interventions] increases the number of well-off future people in expectation by 10^26", and GiveWell's top charities save around 2*10^-4 lives per $ (= 1/(5*10^3)). However, I assume your best guess for the reduction in existential risks would not have to be less than 2*10^-30 (2*10^-4/10^26) times as high, i.e. less than 2*10^-32 pp (instead of your assumption of 0.01 pp), for you to prioritise global health over longtermist interventions.

1.1 The case for Longtermist interventions being best

You assumed decreasing "existential risks" by 0.01 pp (you said ".01%", which is different) would increase the value of the future (10^40 human lives) by 0.01 %, resulting in 10^36 human lives of expected benefits. How do you justify that assumption? I do not think it is valid.

Experts consistently think there’s a several percent chance that misaligned AI kills everyone.

Are you aware of any quantitative model suggesting this? I am only aware of almost purely subjective guesses, and products of these.

These tend to be super effective: dollars given directly to Giving What We Can [GWWC's] return about 6 dollars to effective charities and dollars given to Effektiv Spenden return about 13 dollars to effective charities.

GWWC estimated their giving multiplier in 2025 was 7. You linked to an analysis I did of their giving multiplier in 2023-2024. GWWC estimated this was 6, and I got values of 8 to 9.

3.1 The basic case for animal welfare

I believe you are underestimating the uncertainty involved in welfare comparisons across species. I would say animal welfare interventions may increase the welfare of their target beneficiaries much more or less cost-effectively than global health interventions. Here are my estimates assuming sentience-adjusted welfare ranges proportional to "individual number of neurons"^"exponent", and "exponent" from 0 to 2, which covers the best guesses than I consider reasonable. For an exponent of 2 (which means the sentience-adjusted welfare ranges decrease 2 % for a 1 % reduction in the individual number of neurons), I estimate GiveWell's top charities increase the welfare of humans 135 times as cost-effectively as cage-free egg corporate campaigns increase the welfare of chickens.

Nice post!

Hi Carlos. Great post.

If a value‑aligned candidate is as capable as a non‑aligned candidate, I don’t see why we wouldn’t want them in the running

Because they may have a greater (counterfactual) impact in a role where being value-aligned is more needed? In any case, I think it would still make sense for the organisation to make them an offer, and then be transparent about how they compare with the 2nd best candidate. This information could then be used by the value-aligned candidate to make a better choice about whether to accept the offer or not.

I’d argue that many donors are surprisingly diversified given direct application of philosophical theory, yes.

I agree.

I’d say the ultimate decision makers are donors, not eg evaluators.

I agree in the sense the money influenced by grantmakers and evaluators ultimately comes from donors. However, are you suggesting grantmakers and evaluators diversify their grants and recommendations in significant part to appeal to donors? I agree to some extent, and it makes sense for grantmakers and evaluators to do it up to a point such that they can influence more funds. Them having a very narrow portfolio would tend to attract less funds.

Other than argue about it, probably not much, assuming functionalism and materialism/physicalism of some kind that's compatible with artificial sentience.

You may be interested in the article A science of chimeras? The implications of illusionism for non-human consciousness research. The abstract is below.

Illusionism states that phenomenal consciousness does not exist, even though it seems to exist. While illusionism is controversial, it is a serious contender among theories of consciousness. We argue that it has substantial and non-trivial implications for non-human consciousness research (NHCR), particularly for the study of the distribution of phenomenal consciousness across beings. If illusionism is true, NHCR can be pursued if conceptualized as investigating the distribution of quasi-phenomenal consciousness, i.e. the states which are misrepresented as phenomenally conscious in humans. However, we argue that knowing the distribution of quasi-phenomenal consciousness is not highly informative. For this reason, illusionism suggests that some approaches to NHCR should be preferred over others. Approaches which focus on features that provide valuable information about non-human cognition independently of their supposed relation to consciousness retain much of their value if illusionism is true. We propose a “zombie test” and five specific heuristics to help identifying such features. Consequently, empirical researchers who take illusionism seriously gain a reason to prioritize some methodological approaches over others.

Hi Simon. In that case, how do you explain that impact-focussed grantmakers support many interventions, even within a single area (see, for example, the Animal Welfare Fund)? If the cost-effectiveness of each did not meaningfully decrease with spending, one would expect them to focus on much fewer grants?

Hi Ollie. I can no longer remove tags in my draft or published posts.

I remain sceptical of computational functionalism (CF). It implies some sets of AND, OR, and NOT operations lead to consciousness even if run at the rate of one operation every billion years, which seems very implausible to me.

Thanks, Jan. Yes, I was looking into an older version of the form.

Here is a post elaborating on why I think expanding agricultural land may increase or decrease the welfare/suffering/happiness of soil invertebrates.