[I will use “Effective Altruists” or “EAs” to refer to the people who self-identify as members of the community, and “effective altruists” (without capitalization) for people to whom effectiveness matters a lot in altruism, regardless of whether they self-identify as EAs.]
I think this post makes some important and valuable points. Even if not novel, the concise summary here could make for a good WikiHow article on how to be a more effective fundraiser. However, I believe that this post falls short by failing to mention, let alone wrestle with, the tradeoffs involved with these strategies.
I don’t believe there is a clear and obvious answer to the many tradeoffs involved with adopting various sales tactics that compromise epistemic value. I believe, however, that not even acknowledging these tradeoffs can lead to potentially worse decisions.
My points below overlap somewhat.
First, effective altruists in general, and EAs in particular, are a niche segment in the philanthropic community. The rules for selling to this niche can differ from the rules of selling to the general public. So much so that sales tactics that are considered good for the general public are actively considered bad when selling to this niche. Putting an identifiable victim may help with, say, 30% of potential donors in the general public, but alienate 80% of potential donors among effective altruists, because they have (implicitly or explicitly) learned to overcome the identifiable victim effect. In general, using messaging targeted at the public for a niche that is often based, implicitly or explicitly, on rejecting various aspects of such messaging, is a bad thing. A politician does not benefit from taking positions held by the majority of people all the time; rather, whereas some politicians are majoritarian moderates, others seek specific niches where their support is strong, often with the alienation of a majority as a clear consequence (for instance, a politician in one subregion of a country may adopt rhetoric and policies that make the politician unpopular countrywide but guarantee re-election in that subregion). Similarly, not every social network benefits from adopting Facebook’s approach to partial openness and diversity of forms of expression. Snapchat, Pinterest, and Twitter have each carved a niche based on special features they have.
Second, in addition to the effect in rhetorical terms, it’s also important to consider the effect in substantive terms on how the organizations involved spend their money and resources, and make decisions. Ideally, you can imagine a wall of separation: the organization focuses on being maximally effective, and a separate sales/fundraising group optimizes the message for the general public. However, many of the strategies suggested here actually affect the organization’s core functions. Pairing donors with individual recipients significantly affects the organization’s operations on the ground, raising costs. Could this in the long run lead to e.g. organizations selecting to operate in areas where recipients have characteristics that make them more interesting to donors to communicate with (e.g., they are more familiar with the language of the donor’s country?). I don’t see a way of making overall effectiveness, in the way that many EAs care about, still the dominant evaluation criterion if fundraising success is tied heavily to other outreach strategies.
Third (building somewhat on the first), insofar as there is a tradeoff between being able to sell more to effective altruists versus appealing more to the general public, the sign of the financial effect is actually ambiguous. The number of donors in the general public is much larger, but the amount that they donate per capita tends to be smaller. One of the ingredients to EA success is that its strength lies not so much in its numbers but in the depth of convictions of many self-identified EAs, plus other effective altruists (such as GiveWell donors). People who might have previously donated a few hundred dollars a year for an identifiable victim may now be putting in tens of thousands of dollars because the large-scale statistics have touched them in a deeper way. GiveWell moved $103 million to its top charities in 2015, of which $70 million was from Good Ventures (that’s giving away money from a Facebook co-founder) and another $20 million is from individual donors who are giving amounts in excess of $100,000 each. To borrow sales jargon, these deals are highly lucrative and took a long time to close. Closing them required the donor to have high confidence in the epistemic rigor from a number of donors, many of whom were probably jaded by psychologically pitch-perfect campaigns. I’m not even saying that GiveWell’s reviews are actually rigorous, but rather, that the perception of rigor surrounding them was a key aspect to many people donating to GiveWell-recommended charities.
Fourth, if the goal is to spread better, more rational giving habits, then caving in to sales tactics that exploit known forms of irrationality hampers that goal.
None of these imply that the ideas you suggest are inapplicable in the context of EA or for effective altruists in general. Nor am I suggesting that EAs (or effective altruists in general) are bias-free and rational demigods: I think many EAs have their own sets of biases that are more sophisticated than those of the general public but still real. I also think that many of the biases, such as the identifiable victim, can actually be epistemically justified somewhat, and you could make a good epistemic case for using individual case studies as not just a sales strategy but something that actually helps provide yet another sanity check (this is sort of what GiveWell tried to do by sponsoring field trips to the areas of operation of its top charities). You could also argue that the cost of alienating some people is a cost worth bearing in order to achieve a somewhat greater level of popularity, or that a wall of separation is not that hard to achieve.
But acknowledging these tradeoffs openly is a first step to letting others (including the orgs and fundraisers you are targeting) make a careful, informed decision. It can also help people figure out new, creative compromises. Perhaps, for instance, showing an identifiable victim and, after people are sort-of-sold, then pivoting to the statistics, provides the advantages of mass appeal and epistemic rigor. Perhaps there are ways to use charities’ own survey data to create composite profiles of typical beneficiaries that can help inform potential donors as well as appeal to their desire for an identifiable victim. Perhaps, at the end of the day, raising money matters more than spreading ideas, and getting ten million people to donate a few hundred dollars a year is better than the current EA donor profile or the current GiveWell donor profile.
Thank you for your comprehensive and well thought out comment. Sorry for my delay in response—I have been sick and busy.
The aim of this article, at least for me, was to provide some sort of a wiki-how for how to generally be effective at fundraising. I think it does a reasonable job in that regard and I am not sure if a discussion of tradeoffs would fit well with the format and aims.
With that said, I think that your insights are very valuable for us to consider. You have made several points that I realize I need to think more about (for example the differences in operational costs for different approaches, and the need to consider the amount that EAs give as an argument against their small numbers). Thanks for that. In particular I now recognise to a greate extent the potential tradeoff issues for arguments 2 & 4 as these points related to approaches that might work better for the general public but be less effective at persuading EAs. For 1 I think that there is no serious challenge, nor tradeoff, in combining statistics and emotional persuasion—most of the givewell charities do this to differing extents.
I will definitely consider writing a follow up article which discussed potential tradeoffs in more detail. Even if I don’t directly build on the above article I will keep the comments in mind when writing/advising about similar topics.
Would it be possible to combine individual stories and general statistics? If I tell you there are X number of children dying from malaria, you may want to know what it is like to die from malaria. If I show you a video of one child dying from malaria, you may want to know how many other children are in a similar situation. If individual stories do have some usefulness, then would it be a good use of money for an EA organization to make some short films about the stories of individual children dying from malaria? It could bring people into the movement the same way that undercover videos of factory farms bring people into the animal movement. If people who are already in the movement are more likely to respond positively to a different type of appeal, could we target different types of appeals to different audiences? I did a quick search on YouTube for videos about malaria, and I found way fewer videos than I know are out there about factory farming.
Excellent point, upvoted. Thanks for constructive criticism of the points made here, it’s really helpful! This is exactly the kind of dialogue we were hoping for :-)
Peter and I wrestled with a number of issues you brought up, and ended up deciding to write an essay that went against the grain in the EA movement itself of focusing only on statistics, since there is so much literature on this topic already. We didn’t want to deal with the ground that was previously covered, but I recognize it would have been helpful to make more of an effort to describe the tradeoffs involved.
Several of the things discussed in this comment were touched on in the essay. For instance, the difference between communicating to people more oriented toward effectiveness, the small minority of EA participants, versus those more oriented toward emotions, the large majority of those who give, was covered in these sections: “while fact and statistics are often an effective way to convince potential donors, it is important to recognize that different people are persuaded by different things. While some individuals are best persuaded to do good deeds through statistics and facts, others are most influenced by the closeness and vividness of the suffering” and also “experiments and other forms of prospect research to better understand their donors’ reasons for giving.” I acknowledge this could have been made more explicit.
Regarding messaging to different groups, we intended this essay to show folks fundraising for effective charities how to speak to non-EA participants. At the same time, this essay has the meta-function of helping EA participants not be put off by emotionally engaging messages conveyed by effective charities. As EA members, we believe we should optimize for money flowing to effective causes, and be happy about effective charities using such strategies in their fundraising, while ourselves relying on GiveWell and other well-reputed charity evaluators. In general, charities should use segmented messaging, with different messages for different audiences.
To Peter and I, this is a matter of the inference gap. If we can encourage people to give to effective charities and use GiveWell and similar EA-themed charity evaluators such as Animal Charity Evaluators, folks will likely grow in their epistemic rigor in assessing charities, slowly crossing the inference gap. However, this is something that deserves a much larger treatment to uncover similar issues.
Good point about the challenges involved in optimizing for individual victims! Notice that Peter and I did not suggest that charities should change their programming in any way, but simply adjust their messaging. I do see the danger you point to, and it’s something to watch out for.
Thanks again for your helpful comments, and Peter, please jump in if you have further thoughts.
What I would be interested in is a follow-up post where you discuss the various trade-offs, how one might make them, what the implications of each are, etc. As it stands now, your post just doesn’t seem very useful or actionable. But it may be a nice starting point that you can build on in the future.
I also think that some of the stylistic simplifications that you engage in might be well-optimized for publishing in mass media like Time Magazine or Huffington Post, but potentially hurt your credibility when posted in a narrower context like the Effective Altruism Forum (or LessWrong, where you cross-posted this). I would recommend setting higher standards both for epistemic rigor and for the level of detail of your actual content when posting here, and preempting objections better (something that may not matter when you are writing for an audience that is totally new to your ideas and have limited space).
[Comment cross-posted to LessWrong]
[I will use “Effective Altruists” or “EAs” to refer to the people who self-identify as members of the community, and “effective altruists” (without capitalization) for people to whom effectiveness matters a lot in altruism, regardless of whether they self-identify as EAs.]
I think this post makes some important and valuable points. Even if not novel, the concise summary here could make for a good WikiHow article on how to be a more effective fundraiser. However, I believe that this post falls short by failing to mention, let alone wrestle with, the tradeoffs involved with these strategies.
I don’t believe there is a clear and obvious answer to the many tradeoffs involved with adopting various sales tactics that compromise epistemic value. I believe, however, that not even acknowledging these tradeoffs can lead to potentially worse decisions.
My points below overlap somewhat.
First, effective altruists in general, and EAs in particular, are a niche segment in the philanthropic community. The rules for selling to this niche can differ from the rules of selling to the general public. So much so that sales tactics that are considered good for the general public are actively considered bad when selling to this niche. Putting an identifiable victim may help with, say, 30% of potential donors in the general public, but alienate 80% of potential donors among effective altruists, because they have (implicitly or explicitly) learned to overcome the identifiable victim effect. In general, using messaging targeted at the public for a niche that is often based, implicitly or explicitly, on rejecting various aspects of such messaging, is a bad thing. A politician does not benefit from taking positions held by the majority of people all the time; rather, whereas some politicians are majoritarian moderates, others seek specific niches where their support is strong, often with the alienation of a majority as a clear consequence (for instance, a politician in one subregion of a country may adopt rhetoric and policies that make the politician unpopular countrywide but guarantee re-election in that subregion). Similarly, not every social network benefits from adopting Facebook’s approach to partial openness and diversity of forms of expression. Snapchat, Pinterest, and Twitter have each carved a niche based on special features they have.
Second, in addition to the effect in rhetorical terms, it’s also important to consider the effect in substantive terms on how the organizations involved spend their money and resources, and make decisions. Ideally, you can imagine a wall of separation: the organization focuses on being maximally effective, and a separate sales/fundraising group optimizes the message for the general public. However, many of the strategies suggested here actually affect the organization’s core functions. Pairing donors with individual recipients significantly affects the organization’s operations on the ground, raising costs. Could this in the long run lead to e.g. organizations selecting to operate in areas where recipients have characteristics that make them more interesting to donors to communicate with (e.g., they are more familiar with the language of the donor’s country?). I don’t see a way of making overall effectiveness, in the way that many EAs care about, still the dominant evaluation criterion if fundraising success is tied heavily to other outreach strategies.
Third (building somewhat on the first), insofar as there is a tradeoff between being able to sell more to effective altruists versus appealing more to the general public, the sign of the financial effect is actually ambiguous. The number of donors in the general public is much larger, but the amount that they donate per capita tends to be smaller. One of the ingredients to EA success is that its strength lies not so much in its numbers but in the depth of convictions of many self-identified EAs, plus other effective altruists (such as GiveWell donors). People who might have previously donated a few hundred dollars a year for an identifiable victim may now be putting in tens of thousands of dollars because the large-scale statistics have touched them in a deeper way. GiveWell moved $103 million to its top charities in 2015, of which $70 million was from Good Ventures (that’s giving away money from a Facebook co-founder) and another $20 million is from individual donors who are giving amounts in excess of $100,000 each. To borrow sales jargon, these deals are highly lucrative and took a long time to close. Closing them required the donor to have high confidence in the epistemic rigor from a number of donors, many of whom were probably jaded by psychologically pitch-perfect campaigns. I’m not even saying that GiveWell’s reviews are actually rigorous, but rather, that the perception of rigor surrounding them was a key aspect to many people donating to GiveWell-recommended charities.
Fourth, if the goal is to spread better, more rational giving habits, then caving in to sales tactics that exploit known forms of irrationality hampers that goal.
None of these imply that the ideas you suggest are inapplicable in the context of EA or for effective altruists in general. Nor am I suggesting that EAs (or effective altruists in general) are bias-free and rational demigods: I think many EAs have their own sets of biases that are more sophisticated than those of the general public but still real. I also think that many of the biases, such as the identifiable victim, can actually be epistemically justified somewhat, and you could make a good epistemic case for using individual case studies as not just a sales strategy but something that actually helps provide yet another sanity check (this is sort of what GiveWell tried to do by sponsoring field trips to the areas of operation of its top charities). You could also argue that the cost of alienating some people is a cost worth bearing in order to achieve a somewhat greater level of popularity, or that a wall of separation is not that hard to achieve.
But acknowledging these tradeoffs openly is a first step to letting others (including the orgs and fundraisers you are targeting) make a careful, informed decision. It can also help people figure out new, creative compromises. Perhaps, for instance, showing an identifiable victim and, after people are sort-of-sold, then pivoting to the statistics, provides the advantages of mass appeal and epistemic rigor. Perhaps there are ways to use charities’ own survey data to create composite profiles of typical beneficiaries that can help inform potential donors as well as appeal to their desire for an identifiable victim. Perhaps, at the end of the day, raising money matters more than spreading ideas, and getting ten million people to donate a few hundred dollars a year is better than the current EA donor profile or the current GiveWell donor profile.
Hi Vipul,
Thank you for your comprehensive and well thought out comment. Sorry for my delay in response—I have been sick and busy.
The aim of this article, at least for me, was to provide some sort of a wiki-how for how to generally be effective at fundraising. I think it does a reasonable job in that regard and I am not sure if a discussion of tradeoffs would fit well with the format and aims.
With that said, I think that your insights are very valuable for us to consider. You have made several points that I realize I need to think more about (for example the differences in operational costs for different approaches, and the need to consider the amount that EAs give as an argument against their small numbers). Thanks for that. In particular I now recognise to a greate extent the potential tradeoff issues for arguments 2 & 4 as these points related to approaches that might work better for the general public but be less effective at persuading EAs. For 1 I think that there is no serious challenge, nor tradeoff, in combining statistics and emotional persuasion—most of the givewell charities do this to differing extents.
I will definitely consider writing a follow up article which discussed potential tradeoffs in more detail. Even if I don’t directly build on the above article I will keep the comments in mind when writing/advising about similar topics.
Would it be possible to combine individual stories and general statistics? If I tell you there are X number of children dying from malaria, you may want to know what it is like to die from malaria. If I show you a video of one child dying from malaria, you may want to know how many other children are in a similar situation. If individual stories do have some usefulness, then would it be a good use of money for an EA organization to make some short films about the stories of individual children dying from malaria? It could bring people into the movement the same way that undercover videos of factory farms bring people into the animal movement. If people who are already in the movement are more likely to respond positively to a different type of appeal, could we target different types of appeals to different audiences? I did a quick search on YouTube for videos about malaria, and I found way fewer videos than I know are out there about factory farming.
Excellent point, upvoted. Thanks for constructive criticism of the points made here, it’s really helpful! This is exactly the kind of dialogue we were hoping for :-)
Peter and I wrestled with a number of issues you brought up, and ended up deciding to write an essay that went against the grain in the EA movement itself of focusing only on statistics, since there is so much literature on this topic already. We didn’t want to deal with the ground that was previously covered, but I recognize it would have been helpful to make more of an effort to describe the tradeoffs involved.
Several of the things discussed in this comment were touched on in the essay. For instance, the difference between communicating to people more oriented toward effectiveness, the small minority of EA participants, versus those more oriented toward emotions, the large majority of those who give, was covered in these sections: “while fact and statistics are often an effective way to convince potential donors, it is important to recognize that different people are persuaded by different things. While some individuals are best persuaded to do good deeds through statistics and facts, others are most influenced by the closeness and vividness of the suffering” and also “experiments and other forms of prospect research to better understand their donors’ reasons for giving.” I acknowledge this could have been made more explicit.
Regarding messaging to different groups, we intended this essay to show folks fundraising for effective charities how to speak to non-EA participants. At the same time, this essay has the meta-function of helping EA participants not be put off by emotionally engaging messages conveyed by effective charities. As EA members, we believe we should optimize for money flowing to effective causes, and be happy about effective charities using such strategies in their fundraising, while ourselves relying on GiveWell and other well-reputed charity evaluators. In general, charities should use segmented messaging, with different messages for different audiences.
To Peter and I, this is a matter of the inference gap. If we can encourage people to give to effective charities and use GiveWell and similar EA-themed charity evaluators such as Animal Charity Evaluators, folks will likely grow in their epistemic rigor in assessing charities, slowly crossing the inference gap. However, this is something that deserves a much larger treatment to uncover similar issues.
Good point about the challenges involved in optimizing for individual victims! Notice that Peter and I did not suggest that charities should change their programming in any way, but simply adjust their messaging. I do see the danger you point to, and it’s something to watch out for.
Thanks again for your helpful comments, and Peter, please jump in if you have further thoughts.
What I would be interested in is a follow-up post where you discuss the various trade-offs, how one might make them, what the implications of each are, etc. As it stands now, your post just doesn’t seem very useful or actionable. But it may be a nice starting point that you can build on in the future.
I also think that some of the stylistic simplifications that you engage in might be well-optimized for publishing in mass media like Time Magazine or Huffington Post, but potentially hurt your credibility when posted in a narrower context like the Effective Altruism Forum (or LessWrong, where you cross-posted this). I would recommend setting higher standards both for epistemic rigor and for the level of detail of your actual content when posting here, and preempting objections better (something that may not matter when you are writing for an audience that is totally new to your ideas and have limited space).
Good points, will keep these in mind!