First of all, thank you so much for this post! I'm only just starting to understand the logic behind "optimal"/reasonable grant allocation, and I definitely don't think I'm in a position to give any advice yet. But I wanted to say two things:
This post finally made some of the things I'd read before click into place, and I feel like I'm starting to have a gears-level model of "why the things I'm observing are happening". I'm incredibly grateful for that!
Before thinking about "how I'd like things to be", I usually spend (perhaps too much) time trying to understand how things actually work, in enough detail not to accidentally tear down some Chesterton's fence. So I wanted to share a piece of reasoning that came up while reading (with some hope to get feedback on that):
I somewhat disagree with (or possibly just don't understand) the idea behind:
"Non-profits don’t have to optimize for keeping their pot of money"
I tend to think of money in investments as a proxy for some "true" utility fuction. And when a for-profit investor puts $100k into something that fails, this proxy somewhat accurately represents what's lost: $100k + potential compounding returns. Both terms can be expressed in dollars. That's because for-profits maximise money (they can have other preferences too, so it won't be pure optimisation of the amount of money, but that's not relevant right now).
But for nonprofits, this proxy fails to capture what is being maximised (or at least it is significantly less good for it), so it only represents one part of that sum. The second part, the potential impact lost, is still there, but it's much more challenging to assign a dollar value to it in a precise and consistent way.
But I think it's still present in people's intuitions. And that's why I can't fully agree with the conclusion:
"All things equal, this makes it easier to take risks with grants that you might not want to take with for-profit investments."
Because I'd assume that this internal drive of "If I support a project that fails, I'll lose all these nice outcomes I'm trying to maximise" would make people more risk-averse.
I suppose I might be underestimating the difference in risk aversion caused by monetary losses versus the potential failure to fulfil other goals. Or I might be wrong for some completely different reason I'm currently not aware of.
I tried to look up some statistics on this, specifically whether for-profit investors or nonprofit grantmakers are more willing to take high-risk bets.
I haven't found a direct comparison, and the closest evidence seems mixed. A 2015 study of foundation staff found that they were somewhat more willing to take risks in grantmaking than in their personal decisions. Meanwhile, a 2019 study found that social entrepreneurs were more risk-seeking than commercial entrepreneurs, while a 2025 study found the opposite. Neither directly compares grantmakers with for-profit investors, though.
So I'm not sure the empirical evidence settles the question. Or I'm just not good enough to agregate existing evidence.
But I still find it hard to agree with the intuition that the second term would have a negligibly small effect. Because it includes things like "having a mission / caring about achieving actual results", and I would expect this component to be stronger than average among people working at socially impactful nonprofits.
And things like "reputational costs, which mean that if I support too many failed projects, I might lose the right to decide which projects to support (or, in simpler terms, get fired)" . And I would expect this component to be present in the heads of most actual human beings.
And I suspect there are other components too. Like "I just want to win this bet". etc etc...
I don't think any of this affects the central logic of the post very much. But I found the post very useful, partly because it gave me a gears-level model of why grantmaking dynamics in my field look the way they do. So I wanted to share this piece of my reasoning in case someone else finds it interesting too (and maybe someone will tell me something new =)).
Some effective altruists expect artificial general intelligence (AGI) or transformative AI (TAI) to solve the taste and price challenges for alternative meats, and so they wonder: What's the point of work on alternative meats now?
I'd be interested to learn more about the reasoning behind this view. So far I'm only familiar with [1](https://www.lesswrong.com/posts/DuCWLYx4jAFPMXKjZ/paper-artificial-intelligence-in-life-extension-from-deep) and [2](https://forum.effectivealtruism.org/posts/TYhs8zehyybvMt5E4/).
And I have an intuition (possibly because I'm unintentionally making a strawman out of the actual position simply because I don't understand it) that the "we need AGI to solve this problem" relies on the assumption that "AGI is a thing that will solve basically all existing problems simply because it's generally much better at solving problems than we are". And the implicit conclusion is "so we don't need to think about how exactly AGI will solve this particular problem".
It seemed to me that the post was arguing against this approach.
And I actually want to argue even further:
I think we somewhat understand the problem at the gears-level where we can build models and test them against reality, given that we have narrow AI with today's SOTA capabilities at our disposal. Current models have already shown themselves to be effective in some kinds of scientific research (mostly when the task is formulated in a specific, AI-tractable way).
And overall I assign a higher probability to A than B.
Where A is: "We can build a high-level model of what interventions are needed to achieve large-scale replacement of animal products, ensure a smooth and comfortable transition, and make this new dietary pattern stable. And then decompose the problem in such a way that the applied parts of this research would benefit significantly from using SOTA AI. And eventually succeed."
Where B is: "We manage to build an AI that will solve a whole range of very different problems, including this one, without us having to put effort into investigating this particular problem now."
And I assign a substantially higher probability to A than B. (And I'm not even talking about existential risks here, but never mind.)
Is it really true that my view is far from consensus? I thought pretty much everyone thought roughly along these lines, but after reading this post I'm somehow not so sure anymore.
First of all, thank you so much for this post! I'm only just starting to understand the logic behind "optimal"/reasonable grant allocation, and I definitely don't think I'm in a position to give any advice yet. But I wanted to say two things:
I somewhat disagree with (or possibly just don't understand) the idea behind:
I tend to think of money in investments as a proxy for some "true" utility fuction. And when a for-profit investor puts $100k into something that fails, this proxy somewhat accurately represents what's lost: $100k + potential compounding returns. Both terms can be expressed in dollars. That's because for-profits maximise money (they can have other preferences too, so it won't be pure optimisation of the amount of money, but that's not relevant right now).
But for nonprofits, this proxy fails to capture what is being maximised (or at least it is significantly less good for it), so it only represents one part of that sum. The second part, the potential impact lost, is still there, but it's much more challenging to assign a dollar value to it in a precise and consistent way.
But I think it's still present in people's intuitions. And that's why I can't fully agree with the conclusion:
Because I'd assume that this internal drive of "If I support a project that fails, I'll lose all these nice outcomes I'm trying to maximise" would make people more risk-averse.
I suppose I might be underestimating the difference in risk aversion caused by monetary losses versus the potential failure to fulfil other goals. Or I might be wrong for some completely different reason I'm currently not aware of.
I tried to look up some statistics on this, specifically whether for-profit investors or nonprofit grantmakers are more willing to take high-risk bets.
I haven't found a direct comparison, and the closest evidence seems mixed. A 2015 study of foundation staff found that they were somewhat more willing to take risks in grantmaking than in their personal decisions. Meanwhile, a 2019 study found that social entrepreneurs were more risk-seeking than commercial entrepreneurs, while a 2025 study found the opposite. Neither directly compares grantmakers with for-profit investors, though.
So I'm not sure the empirical evidence settles the question. Or I'm just not good enough to agregate existing evidence.
But I still find it hard to agree with the intuition that the second term would have a negligibly small effect. Because it includes things like "having a mission / caring about achieving actual results", and I would expect this component to be stronger than average among people working at socially impactful nonprofits.
And things like "reputational costs, which mean that if I support too many failed projects, I might lose the right to decide which projects to support (or, in simpler terms, get fired)" . And I would expect this component to be present in the heads of most actual human beings.
And I suspect there are other components too. Like "I just want to win this bet". etc etc...
I don't think any of this affects the central logic of the post very much. But I found the post very useful, partly because it gave me a gears-level model of why grantmaking dynamics in my field look the way they do. So I wanted to share this piece of my reasoning in case someone else finds it interesting too (and maybe someone will tell me something new =)).
There's a quote in the post:
I'd be interested to learn more about the reasoning behind this view. So far I'm only familiar with [1](https://www.lesswrong.com/posts/DuCWLYx4jAFPMXKjZ/paper-artificial-intelligence-in-life-extension-from-deep) and [2](https://forum.effectivealtruism.org/posts/TYhs8zehyybvMt5E4/).
And I have an intuition (possibly because I'm unintentionally making a strawman out of the actual position simply because I don't understand it) that the "we need AGI to solve this problem" relies on the assumption that "AGI is a thing that will solve basically all existing problems simply because it's generally much better at solving problems than we are". And the implicit conclusion is "so we don't need to think about how exactly AGI will solve this particular problem".
It seemed to me that the post was arguing against this approach.
And I actually want to argue even further:
I think we somewhat understand the problem at the gears-level where we can build models and test them against reality, given that we have narrow AI with today's SOTA capabilities at our disposal. Current models have already shown themselves to be effective in some kinds of scientific research (mostly when the task is formulated in a specific, AI-tractable way).
And overall I assign a higher probability to A than B.
And I assign a substantially higher probability to A than B. (And I'm not even talking about existential risks here, but never mind.)
Is it really true that my view is far from consensus? I thought pretty much everyone thought roughly along these lines, but after reading this post I'm somehow not so sure anymore.