2 weeks out from the new GiveWell/GiveDirectly analysis, I was wondering how GHD charities are evaluating the impact of these results.

For Kaya Guides, this has got us thinking much more explicitly about what we’re comparing to. GiveWell and GiveDirectly have a lot more resources, so they can do things like go out to communities and measure second order and spillover effects.

On the one hand, this has got us thinking about other impacts we can incorporate into our analyses. Like GiveDirectly, we probably also have community spillover effects, we probably also avert deaths, and we probably also increase our beneficiaries’ incomes by improving productivity. I suspect this is true for many GHD charities!

On the other, it doesn’t seem fair to compare our analysis on individual subjective wellbeing to GiveDirectly’s analysis that incorporates many more things. Unless we believed that GiveDirectly is likely to be systematically better, it’s not the case that many GHD charities got 3–4× less cost-effective relative to cash transfers overnight, they may just count 3–4× less things! So I wonder if the standard cash transfers benchmark might have to include more nuance in the near-term. Kaya Guides already only makes claims about cost-effectiveness ‘at improving subjective wellbeing’ to try and cover for this.

Are other GHD charities starting to think the same way? Do people have other angles on this?

I think this is good. I think it would be valuable for most people here to read a criticism from an uncommon angle, even if you disagree with the author’s argument. Thank you for sharing it here!

So nice to see & thank you for sharing! If you’re ever interested in picking up habit tracking again, I’ve been through the ringer with apps and settled on Bearable; they have a lot of the more correlative stuff you were looking for (i.e. you can set up unlimited inputs & outcomes and it will correlate between them in a simple way).

But, also, I’ve been using habit/factor trackers for years and I’ve never gotten anything useful out of their analysis tools because there are too many confounders. I mostly find them useful as a habit that causes me to be more mindful about those things, which is valuable in and of itself. (And also I have chronic pain so can specifically keep an eye on my flare-ups for my rheumatologist).

I also think many EAs are still allergic to direct political advocacy, and that this tendency is stronger in more rationalist-ish cause areas such as AI. We shouldn’t forget Yudkowsky’s “politics is the mind-killer”!

Hmm, no. I wouldn’t want Vida Plena to update without evidence that they have those secondary effects.

But I think it would also be misleading to compare direct effects + household spillovers (in the case of Vida Plena) to direct effects + household spillovers + community spillovers + mortality reduction + consumption increases (GiveDirectly), unless you had good reason to believe that Vida Plena’s secondary effects are much worse than GiveDirectly’s. So I suppose I would be wary of saying that GiveDirectly now have 3–4x the WELLBY impact relative to Vida Plena—or even to say that GiveDirectly have any more WELLBY impact relative to Vida Plena—without having a good sense of how Vida Plena performs on those secondary outcomes. (But I feel like maybe I’m misunderstanding what you meant by applying a discount?)

The cost-effectiveness of GiveDirectly has gone up by 3-4x (GW blog, GD blog), though this was recent news and does not necessarily imply that WELLBYs will also go up by 3-4x (most of this increase is attributable to increased consumption) - but should constitute a discount at least.

I’m not sure about this; the HLI’s analysis of GiveDirectly only looks at direct individual effects and household spillovers. Whereas GiveWell’s update seemingly only found additional effects in terms of non-household spillovers, mortality, and consumption (based on a 5 minute check, so I might be wrong here).

I think it’s reasonable to argue that depression prevention would also have effects on mortality, consumption (via productivity increases—my guesses here peg this quite high, especially in LMICs and UMICs), and non-household spillovers (via increased income being reinvested into communities, using the same mechanism as GD). Unless there’s reason to believe that the non-accounted-for impacts on WELLBYs systematically favour GiveDirectly I’d be cautious about applying a discount—but curious for your take on that :)

AIM have already tried to do this for research, and they aren’t sure whether to continue their research fellowship in 2025.. I imagine they’d have some very good learnings on this topic if you got in touch!

Interesting to see the same from the EA Animal Welfare Fund, who only gave ~1.2% of funds to explicit alternative protein work. I suspect this is emblematic of a broader shift within EA toward getting easy, quick wins in neglected countries (?)

I am surprised to see such a gap between Europe and North America, given that both are at least economically similar! Would love to hear more about this—in my mental model there is probably more regulatory capture in the U.S., compounded by generally less ideological willingness to help animals. Is this correct?

Military applications of AI are not an idle concern. AI systems are already being used to increase military capacity by generating and analysing targets faster than humans can (and in this case, seemingly without much oversight). Palantir’s own technology likely also allows police organisations to defer responsibility for racist policing to AI systems.

Sure, for the most part, Claude will probably just be used for common requests, but Anthropic have no way of guaranteeing this. You cannot do this by policy, especially if it’s on Amazon hardware that you don’t control and can’t inspect. Ranking agencies by ‘cooperativeness’ should also be taken as lip service until they have a proven mechanism for doing so.

So they are revealing that, to them, AI safety doesn’t mean that they try to prevent AI from doing harm, just that they try to prevent it from doing unintended harm. This is a significant moment for them and I fear what it portends for the whole industry.