Thanks for this Mauricio. It’s good to have an alternative perspective added to this, which was written by quite convinced advocates for one way of thinking!
I think you make a good point that this is a theory that seems to align very closely with the reality of EA, rather than an absolutely established phenomenon. So, for example, we don’t have data in the EA survey that says ‘people say they would likely drop out if they weren’t donating’ or ‘we see higher rates of drop out amongst people who don’t donate versus those who do’. That’s not to say those statements aren’t plausibly true—it’s just the survey isn’t set up to capture them.
It seems unlikely, though, that it’s coincidental that the foremost and most longstanding members of EA have given throughout their engagement and often seem to increase their giving over time (cf. Julia, Will, Toby, everyone at Longview, ~everyone at GiveWell). This also aligns with our experience of talking to the EA community. Obviously anecdotal evidence is weaker than some sort of systematic evidence but if you have a theory that is plausibly true, aligns with common sense and then is supported by a lot individual cases, that seems enough to think this is ‘signal’ rather than coincidence.
To address some specific points:
-Careers advice may be more popular than programming about giving—it makes sense, as both parties want the thing on offer. It’s the opposite of asking for some sacrifice—you can receive careers advice purely out of self-interest. Equally, though, lots of students are passionate about social justice, making a difference etc. and can be attracted to EA precisely by talking about giving. Career change isn’t for everyone, especially when EA careers advice can focus on careers that need significant technical expertise, like biorisk or AI safety. Careers advice also has some hazier routes to impact in its theory of change than a lot of effective giving. - I’d challenge the idea that the majority of students are charity sceptics. A quick Google suggests exactly the opposite Gen Z gives more and more widely than older generations. Gen Z and Millennials are seen as activist generations, so I’d be really surprised if the median Gen Z-er is a donation sceptic, and the data seems to undermine this idea reasonably firmly. - I’m surprised a) that you haven’t seen the result of any donations and b) that you’re sceptical that is has shorted feedback loops than a career change. If you’re at university and alter your career plans, I’d guess you’d have to wait at least 2-3 years to see any impact from that? And plausibly way, way longer? If you donate $10 to AMF today, you’ll be able to see the bednet distribution you funded in a much shorter timescale. I can see a donation I made in November ’21 has already funded nets that are ready in the factory for distribution in the Congo. Maybe this changes depending on what you donate to? - costly signalling is a widely-referenced theory (the Wikipedia pages on it are instructive), although in fairness it’s more broadly cited in relation to signalling to others rather than necessarily deepening your personal commitment (a costly signal is seen as more honest and therefore more powerful) - Candidly, I think opportunity costs are frequently overstated. We do acknowledge this above and give examples of how giving can be incorporated into existing programming. However, we also think there’s a frequent fallacy in EA, where we make all decisions as if they are zero sum (e.g., to pick a particularly odd example, ‘we shouldn’t give blood because we could spend that time earning $x and giving it to an effective charity’, when of course almost everyone in EA can do both simultaneously). Often this choice isn’t real. Of course EA groups need to make some decisions about prioritisation; but are most EA groups genuinely so maxed out that they couldn’t weave giving into their existing programming or even run an extra session?
Overall, I think you do a good job of laying out possible drawbacks of this approach. I’m not convinced they add up to a really robust argument to neglect effective giving entirely, though. And I’d challenge you in return that maybe you’re understating the opportunity costs of only focussing on careers advice, while overstating some of these drawbacks.
Equally, though, lots of students are passionate about social justice, making a difference etc. and can be attracted to EA precisely by talking about giving.
Yup, agree that giving can be attracting. I’m not sure it’s equally attracting at universities though, based on what I mentioned earlier.
Obviously anecdotal evidence is weaker than some sort of systematic evidence but if you have a theory that is plausibly true, aligns with common sense and then is supported by a lot individual cases, that seems enough to think this is ‘signal’ rather than coincidence.
I agree it’s totally plausible that giving deepens involvement! Where I might be more hesitant is in whether this plausibility should make us confident. I don’t know if I can point to a single source on where my intuition here is coming from, but my impression is there’s many cases across social sciences (including RCTs on effective giving) where using plausibility, common sense, and anecdotes to make causal inferences will lead us to mistaken conclusions.
Maybe we can agree that groups should test multiple kinds of programming and then choose how much of various kinds of programming to have based on the results?
I’d challenge the idea that the majority of students are charity sceptics.
Agreed—I’m not sure I made claims about the majority of students. [Edit: looking back at my earlier comment, I see how that could have been inferred—I should have been clearer.] I also agree that this significantly limits the extent of the downside.
I’m surprised a) that you haven’t seen the result of any donations and b) that you’re sceptical that is has shorted feedback loops than a career change. [...] If you donate $10 to AMF today, you’ll be able to see the bednet distribution you funded in a much shorter timescale. [...] Maybe this changes depending on what you donate to?
I think that’s it—the charities I donated to didn’t have that nice feature. Good to learn that some do!
Candidly, I think opportunity costs are frequently overstated.
I’m sympathetic to this—I suspect it’s often overstated but still significant. (I think the attentional costs are especially significant: taking your example about a presentation on AI safety, if the call to action switches from just careers to careers and donations, that’s a ~50% dilution of how much attention is being directed to careers.)
are most EA groups genuinely so maxed out that they couldn’t weave giving into their existing programming or even run an extra session?
Maybe just the especially intense ones :)
I’m not convinced they add up to a really robust argument to neglect effective giving entirely, though.
Also agree here—my sense is they add up to an argument that should make us hesitate about giving being “a core part of pretty much any strategy for EA movement building.”
Thanks Mauricio—I think we are in roughly the same place here :-)
I especially like the idea of groups testing outreach and rebalancing on the results.
To be clear, I would expect most student groups to continue to prioritise non-giving outreach and I think that’s great—it’s likely impactful and it offers variety, and an entry point for low income students, which is super important.
Our concern is the number of groups doing no giving outreach at all. If every group did their existing programming, but added a giving session each semester (or a pledge drive), we’d be delighted!
Thanks for this Mauricio. It’s good to have an alternative perspective added to this, which was written by quite convinced advocates for one way of thinking!
I think you make a good point that this is a theory that seems to align very closely with the reality of EA, rather than an absolutely established phenomenon. So, for example, we don’t have data in the EA survey that says ‘people say they would likely drop out if they weren’t donating’ or ‘we see higher rates of drop out amongst people who don’t donate versus those who do’. That’s not to say those statements aren’t plausibly true—it’s just the survey isn’t set up to capture them.
It seems unlikely, though, that it’s coincidental that the foremost and most longstanding members of EA have given throughout their engagement and often seem to increase their giving over time (cf. Julia, Will, Toby, everyone at Longview, ~everyone at GiveWell). This also aligns with our experience of talking to the EA community. Obviously anecdotal evidence is weaker than some sort of systematic evidence but if you have a theory that is plausibly true, aligns with common sense and then is supported by a lot individual cases, that seems enough to think this is ‘signal’ rather than coincidence.
To address some specific points:
-Careers advice may be more popular than programming about giving—it makes sense, as both parties want the thing on offer. It’s the opposite of asking for some sacrifice—you can receive careers advice purely out of self-interest. Equally, though, lots of students are passionate about social justice, making a difference etc. and can be attracted to EA precisely by talking about giving. Career change isn’t for everyone, especially when EA careers advice can focus on careers that need significant technical expertise, like biorisk or AI safety. Careers advice also has some hazier routes to impact in its theory of change than a lot of effective giving.
- I’d challenge the idea that the majority of students are charity sceptics. A quick Google suggests exactly the opposite Gen Z gives more and more widely than older generations. Gen Z and Millennials are seen as activist generations, so I’d be really surprised if the median Gen Z-er is a donation sceptic, and the data seems to undermine this idea reasonably firmly.
- I’m surprised a) that you haven’t seen the result of any donations and b) that you’re sceptical that is has shorted feedback loops than a career change. If you’re at university and alter your career plans, I’d guess you’d have to wait at least 2-3 years to see any impact from that? And plausibly way, way longer? If you donate $10 to AMF today, you’ll be able to see the bednet distribution you funded in a much shorter timescale. I can see a donation I made in November ’21 has already funded nets that are ready in the factory for distribution in the Congo. Maybe this changes depending on what you donate to?
- costly signalling is a widely-referenced theory (the Wikipedia pages on it are instructive), although in fairness it’s more broadly cited in relation to signalling to others rather than necessarily deepening your personal commitment (a costly signal is seen as more honest and therefore more powerful)
- Candidly, I think opportunity costs are frequently overstated. We do acknowledge this above and give examples of how giving can be incorporated into existing programming. However, we also think there’s a frequent fallacy in EA, where we make all decisions as if they are zero sum (e.g., to pick a particularly odd example, ‘we shouldn’t give blood because we could spend that time earning $x and giving it to an effective charity’, when of course almost everyone in EA can do both simultaneously). Often this choice isn’t real. Of course EA groups need to make some decisions about prioritisation; but are most EA groups genuinely so maxed out that they couldn’t weave giving into their existing programming or even run an extra session?
Overall, I think you do a good job of laying out possible drawbacks of this approach. I’m not convinced they add up to a really robust argument to neglect effective giving entirely, though. And I’d challenge you in return that maybe you’re understating the opportunity costs of only focussing on careers advice, while overstating some of these drawbacks.
Thanks for this!
Yup, agree that giving can be attracting. I’m not sure it’s equally attracting at universities though, based on what I mentioned earlier.
I agree it’s totally plausible that giving deepens involvement! Where I might be more hesitant is in whether this plausibility should make us confident. I don’t know if I can point to a single source on where my intuition here is coming from, but my impression is there’s many cases across social sciences (including RCTs on effective giving) where using plausibility, common sense, and anecdotes to make causal inferences will lead us to mistaken conclusions.
Maybe we can agree that groups should test multiple kinds of programming and then choose how much of various kinds of programming to have based on the results?
Agreed—I’m not sure I made claims about the majority of students. [Edit: looking back at my earlier comment, I see how that could have been inferred—I should have been clearer.] I also agree that this significantly limits the extent of the downside.
I think that’s it—the charities I donated to didn’t have that nice feature. Good to learn that some do!
I’m sympathetic to this—I suspect it’s often overstated but still significant. (I think the attentional costs are especially significant: taking your example about a presentation on AI safety, if the call to action switches from just careers to careers and donations, that’s a ~50% dilution of how much attention is being directed to careers.)
Maybe just the especially intense ones :)
Also agree here—my sense is they add up to an argument that should make us hesitate about giving being “a core part of pretty much any strategy for EA movement building.”
Thanks Mauricio—I think we are in roughly the same place here :-)
I especially like the idea of groups testing outreach and rebalancing on the results.
To be clear, I would expect most student groups to continue to prioritise non-giving outreach and I think that’s great—it’s likely impactful and it offers variety, and an entry point for low income students, which is super important.
Our concern is the number of groups doing no giving outreach at all. If every group did their existing programming, but added a giving session each semester (or a pledge drive), we’d be delighted!