This is post four in a series arguing for adjusting a core idea of EA: the pledge to donate 10% of one’s annual income to an effective charity. This is most well-known in the form of the Giving What We Can pledge.
In conversation with commenters, I found that I had not sufficiently defined my proposal. Here, I will define my proposed modification unambiguously, and then discuss it in relation to the idea of Schelling points, a common way of explaining why EA advocates for a 10%-income annual donation to effective charities.
Defining the 2%/8% fuzzies-utilons pledge
The donor pledges to donate 2% of annual income to a charity of one’s choice. This could be an EA charity, the opera, or any other cause the donor is passionate about.
On top of this, the donor pledges an additional 8% or 10% of annual income specifically to charities deemed to be effective. Here, one definition of “effective” is that the charity in question has been found via an unmotivated, in-depth and professional cost-benefit analysis to provide extremely high expected value to the beneficiaries. A variety of definitions of “effective” could suffice. I primarily mean “effective” in any of the ways that EAs mean when they refer to “effective charities” or “effective giving.”
The key misunderstanding that has arisen in the past is that this calls for a mandatory 2% donation to “ineffective” charities, such as one’s alma mater or the opera. This is not what I am calling for. In fact, I personally would prefer if EAs who already give 10% or more to EA charities continued to do so.
Instead, I am calling for EA to center conversations with non-EAs around the idea of a this 2%/8% or 2%/10% variant that I have defined here.
How I think about Schelling points
EAs often use the idea of Schelling points to explain why we have instituted a 10% donation to an effective charity, rather than some other amount, such as 2%, 9%, 11%, 20%, 50%, or “until your material quality of life has degraded to the point that the dysfunction this causes you decreases your ability to donate, perhaps by threatening your performance at work.”
For example, in my last post, mhendric explained the specific choice of a 10% donation thus:
EA argues for a duty of beneficence and asks members to donate 10%. 10% is an arbitrary [Sch]elling point. Why not 11%? Why not 12% (you are here)? But consider: why not 13%? (...) Why not 99%? These worries are a classic critique of duties of beneficence, at least since Singer released Famine, Affluence, and Morality.
In a previous post, mhendric expanded on the rationale for a 10% donation:
10% effective donations has brand recognition and is a nice round number, as you point out. It is used by other groups, such as religious groups, making it easy to re-funnel donations to e.g. religious communities to effective charities. This leaves 90% of your income at your disposal, part of which you may spend on fuzzy causes. It does not seem required to me to change the 10% to allow for fuzzy donations, nor do I think there’s a motivation to make donations to fuzzy causes morally required.
First, we are going to start with a note about jargon. In game theory, the notion of a Schelling point is used specifically to refer to a way that players coordinate in the absence of direct communication. A classic example is that two friends separated in New York City without a way to contact each other or a predetermined meeting spot would each go to Grand Central Station (a famous landmark in the city), assuming that the other would also assume that’s where to look.
Here, we are using Schelling point in a broader sense that ignores the need to converge on the same solution. Of course, EAs could select their own donation levels, from 0-100% of annual income, and those choices would not necessarily impact the choices of other EAs. There is no actual need for EAs to all donate the same amount, and indeed we don’t. Will MacAskill donates 50% of his income to charity, for example, and advocates that the super-rich donate 99% or more of their wealth. Undoubtedly, there are EAs who earn incomes or have obligations such that they understandably feel unable to commit to a 10% annual donation.
This means that a 10% donation standard is not a Schelling point in a game theoretic sense. It is just an arbitary standard, which is a less specific thing.
A more conventional way to refer to an arbitrary practical standard is as a “line in the sand.” One sympathetic journalist asks:
… where do we draw the line? If we’re prepared to donate one-third of our incomes to maximize happiness, then why not two-thirds? Why not live in a tent in a park so as to be able to donate 99 percent and prevent even more cases of blindness?
This journalist, Nicholas Kristof of the New York Times, is not describing Giving What We Can, but the 50% donation practice of the trader Matt Wage. Giving What We Can draws the line at 10% to answer objections such as Kristof’s.
On the one hand, what makes the movement distinct is its demand for absolute moral rigor, a willingness, as they like to put it, to “bite the philosophical bullet” and accept that their logic might precipitate extremes of thought and even behavior—to the idea, to take one example, that any dollar one spends on oneself beyond basic survival is a dollar taken away from a child who does not have enough to eat. On the other hand, effective altruists, or E.A.s, have recognized from the beginning that there are often both pragmatic and ethical reasons to defer to moral common sense. This enduring conflict—between trying to be the best possible person and trying to act like a normal good person—has put them in a strange position. If they lean too hard in the direction of doing the optimal good, their movement would be excessively demanding, and thus not only very small but potentially ruthless; if they lean too hard in the direction of just trying to be good people, their movement would not be anything special...
Giving What We Can’s 10% donation standard can be seen as an arbitary line in the sand, which we draw in order to create a movement that is neither too lax nor too demanding of its members.
I would argue that it’s more subtle. In fact, mhendric (who I greatly respect and appreciate for their contributions to the comments on my previous posts) across multiple comments serves as an illustration of the inconsistencies in viewing the 10% standard as ‘arbitrary’.
10% effective donations has brand recognition and is a nice round number, as you point out. It is used by other groups, such as religious groups, making it easy to re-funnel donations to e.g. religious communities to effective charities.
So the 10% figure is not exactly arbitrary. It is chosen for a specific set of practical reasons—having brand recognition, being a nice round number, and being used by religious groups.
Let’s briefly integrate the argument so far as a set of premises. These premises will serve as the starting point for the remainder of the discussion.
Premise 1: The Giving What We Can 10% pledge is best understood not a Schelling point, but as a line in the sand or conventional community standard.
Premise 2: The Giving What We Can 10% pledge is not arbitary, but selected for specific practical reasons.
Premise 3: It is the specific practical problem we are trying to solve with a community standard that determines its optimal structure.
Note: if you do not agree with one or more of these premises, please let me know in the comments! I am genuinely unsure of whether this will be an important objection or not. If this is your belief, you are unlikely to find the remainder of this discussion valid.
Since the Giving What We Can pledge is not functioning as a way for EAs to coordinate in the absence of the ability to communicate, but as a practical community standard, we need to understand what problems we might be trying to solve with such a standard.
mhendric and the aforementioned journalists have already articulated some of them:
Encouraging a level of giving that is effective, sustainable, and motivating.
Choosing a number that orients discussion in a productive way.
Aligning our standards with those of major existing altruistic traditions.
Maintaining consistency in the EA movement.
I agree with mhendric that these are the three main problems we are trying to solve with our community pledge standard.
With both a 2%/8% fuzzies/utilons pledge and our current Giving What We Can 10% pledge, we achieve an equally sustainable level of giving, while the 2%/10% fuzzies/utilons pledge is equally effective.
So our disagreement is over the other problems.
The 2%/10% choice is slightly less sustainable, since the total donation level is higher, while the 2%/8% choice is slightly less effective, since the effective donation component is lower. These differences are real. But a simple comparison assumes that the choice of pledge does not have an effect on the total amount of donations received. In other words, two people donating $2,000 to the opera and $8,000 to GiveWell contribute more to fighting malaria than one person donating $10,000 to GiveWell. So our more important question is how the choice of standard would motivate donations.
Regarding motivation of donations, let’s note that “orienting discussion” and “aligning standards” are things we care about primarily for their instrumental effects in motivating donations. Maintaining consistency is something we care about to both motivate and sustain donations. How our choice of community standard affects the inflow and outflow of committed donors is what’s truly at issue.
My arguments for a 2%/8% fuzzies/utilons standard have made similar appeals to a practical benefit for sustainability and motivation.
First, my perception is that the “10% to effective causes” standard orients the discussion in a way that encourages many to interpret us as follows:
EA wants me to donate 10% of my income
EA wants me to give 100% of my donations to charities they deem effective
Nicholas Kristof’s article above doesn’t engage with a 10% standard—more like a 30-50% standard. But Kristof explicitly articulates the second idea—that EA wants him to make 100% of his donations to our preferred charities.
In fact, a sophisticated EA might point out that he could donate 10% to EA charities, and then extra on top of that to whatever causes he prefers (in line with my 2%/10% pledge). But that is not what Kristof took away from his discussion with Matt Wage.
I believe it’s well known that EA has a reputation among critics for being absolutist. I covered some of those criticisms in my last post. I believe that an “X%-to-effective charities” standard reliably contributes to this perception, not because it is an accurate takeaway, but because it is what outsiders understand when they hear this in the context of our discussions with them. If we had the chance to explain in more detail, we could enlighten them. But we won’t get that chance. For example, Nicholas Kristof is unlikely to read an email from me explaining this point and then publish an updated version of his original column.
By opening the explanation with a 2%/8% fuzzies/utilons pledge, it gives the person we are talking to a chance to see that we approve of their multiple loyalties and have an explicit community standard that lets them see how they can satisfy the competing demands of their moral parliament.
In turn, this aligns with what Lewis-Kraus calls “traditional liberal concerns.” Of that portion of society that believes in charitable giving at all, the vast majority of it believes that there is hidden virtue in giving to the charities they currently support. Ideas of obligation, loyalty, insider knowledge, cosmopolitanism, liberalism, risk-taking, democracy, systemic change, beauty, religion, and many other values guide people’s choices about charitable giving. If a 10% standard aligns with religious tithing, a 2%/8% split aligns better with traditional notions of liberalism and moderation.
I argue that EA is primarily trying to appeal not to religious donors, but to secular donors. As of 2019, EA was 86% agnostic/atheist/non-religious. Furthermore, a 10% religious tithe is typically supposed to be made to one’s church. EA is not a church, and aligning with a 10% tithing obligation seems unlikely to be useful, because most of those who actually consider themselves under obligation to tithe will not consider donating to EA in the first place.
Thus, I believe that appealing to the cultural norms of the communities EA is trying to reach as donors is most important. And our critics most often criticize our donation standards as preventing them from donating to secular causes, such as alma maters, political movements, the arts, and so on. Showing them that there is a way to include these interests in their giving, while still saving lives in a way that can be demonstrated with cost-benefit analysis in the manner of Effective Altruism seems to be a promising strategy to me.
This leaves the issue of maintaining consistency. Clearly, a 2%/8% fuzzies/utilons pledge maintains some of the consistency. In fact, since the 2% portion is optional, all those currently donating 10% to EA charities could continue doing so with no disruption at all. If the 10% community standard is all that’s preventing a subset of current pledgers from redirecting 20% of their annual giving to the opera instead of GiveWell, are we really pleased that the 10% community standard is having that effect?
I think the main concern is that shifting to a 2%/8% pledge might alienate more potential donors than it attracts—that there is some motivating and sustaining power in the way we currently structure discourse primarily around the idea of a 10% pledge to effective charities. As mhendric put it:
When I encountered EA, a pitch of “Donate X% to the most effective ways of improving lives, then spend an additional 2% on whatever you feel like” would have created more rather than less confusion in me.
I have no reason to doubt mhendric is describing their experience accurately. However, I think this then becomes a messaging problem to solve in cases like theirs. And for others, the 2%/8% pledge would be solving a messaging problem. Figuring out the right language and context to bring it up would be key, and one benefit of the 10% standard is that we have worked some of that out already. There would be switching costs to a new standard, perhaps substantial. This is one of the true downsides to changing our messaging and community standards. These costs include changing descriptions on our various websites, figuring out new ways to describe the 2%/8% pledge, and evaluating whether it is in fact as useful as I believe it would be in driving and sustaining donations.
The question is one of costs and benefits, as always. I cannot complete the cost-benefit analysis here. We need empirical data on how a broad segment of EAs and non-EAs would react to the idea of making such a change, when presented well. My goal for now is to continue addressing the conceptual questions and critiques raised by commenters here. At some point, I hope either to have them change my mind, or to change theirs and move from the realm of discussion and into the realm of a more substantial empirical investigation.
I thank all readers for their continued engagement and thoughts.
Functions of a community standard in the 2%/8% fuzzies/utilons debate
This is post four in a series arguing for adjusting a core idea of EA: the pledge to donate 10% of one’s annual income to an effective charity. This is most well-known in the form of the Giving What We Can pledge.
In conversation with commenters, I found that I had not sufficiently defined my proposal. Here, I will define my proposed modification unambiguously, and then discuss it in relation to the idea of Schelling points, a common way of explaining why EA advocates for a 10%-income annual donation to effective charities.
Defining the 2%/8% fuzzies-utilons pledge
The donor pledges to donate 2% of annual income to a charity of one’s choice. This could be an EA charity, the opera, or any other cause the donor is passionate about.
On top of this, the donor pledges an additional 8% or 10% of annual income specifically to charities deemed to be effective. Here, one definition of “effective” is that the charity in question has been found via an unmotivated, in-depth and professional cost-benefit analysis to provide extremely high expected value to the beneficiaries. A variety of definitions of “effective” could suffice. I primarily mean “effective” in any of the ways that EAs mean when they refer to “effective charities” or “effective giving.”
The key misunderstanding that has arisen in the past is that this calls for a mandatory 2% donation to “ineffective” charities, such as one’s alma mater or the opera. This is not what I am calling for. In fact, I personally would prefer if EAs who already give 10% or more to EA charities continued to do so.
Instead, I am calling for EA to center conversations with non-EAs around the idea of a this 2%/8% or 2%/10% variant that I have defined here.
How I think about Schelling points
EAs often use the idea of Schelling points to explain why we have instituted a 10% donation to an effective charity, rather than some other amount, such as 2%, 9%, 11%, 20%, 50%, or “until your material quality of life has degraded to the point that the dysfunction this causes you decreases your ability to donate, perhaps by threatening your performance at work.”
For example, in my last post, mhendric explained the specific choice of a 10% donation thus:
In a previous post, mhendric expanded on the rationale for a 10% donation:
First, we are going to start with a note about jargon. In game theory, the notion of a Schelling point is used specifically to refer to a way that players coordinate in the absence of direct communication. A classic example is that two friends separated in New York City without a way to contact each other or a predetermined meeting spot would each go to Grand Central Station (a famous landmark in the city), assuming that the other would also assume that’s where to look.
Here, we are using Schelling point in a broader sense that ignores the need to converge on the same solution. Of course, EAs could select their own donation levels, from 0-100% of annual income, and those choices would not necessarily impact the choices of other EAs. There is no actual need for EAs to all donate the same amount, and indeed we don’t. Will MacAskill donates 50% of his income to charity, for example, and advocates that the super-rich donate 99% or more of their wealth. Undoubtedly, there are EAs who earn incomes or have obligations such that they understandably feel unable to commit to a 10% annual donation.
This means that a 10% donation standard is not a Schelling point in a game theoretic sense. It is just an arbitary standard, which is a less specific thing.
A more conventional way to refer to an arbitrary practical standard is as a “line in the sand.” One sympathetic journalist asks:
This journalist, Nicholas Kristof of the New York Times, is not describing Giving What We Can, but the 50% donation practice of the trader Matt Wage. Giving What We Can draws the line at 10% to answer objections such as Kristof’s.
As another journalist, Gideon Lewis-Kraus writing for the New Yorker puts it:
Giving What We Can’s 10% donation standard can be seen as an arbitary line in the sand, which we draw in order to create a movement that is neither too lax nor too demanding of its members.
I would argue that it’s more subtle. In fact, mhendric (who I greatly respect and appreciate for their contributions to the comments on my previous posts) across multiple comments serves as an illustration of the inconsistencies in viewing the 10% standard as ‘arbitrary’.
In my last post, they commented:
Yet in an earlier post, they argued:
So the 10% figure is not exactly arbitrary. It is chosen for a specific set of practical reasons—having brand recognition, being a nice round number, and being used by religious groups.
Let’s briefly integrate the argument so far as a set of premises. These premises will serve as the starting point for the remainder of the discussion.
Premise 1: The Giving What We Can 10% pledge is best understood not a Schelling point, but as a line in the sand or conventional community standard.
Premise 2: The Giving What We Can 10% pledge is not arbitary, but selected for specific practical reasons.
Premise 3: It is the specific practical problem we are trying to solve with a community standard that determines its optimal structure.
Note: if you do not agree with one or more of these premises, please let me know in the comments! I am genuinely unsure of whether this will be an important objection or not. If this is your belief, you are unlikely to find the remainder of this discussion valid.
Since the Giving What We Can pledge is not functioning as a way for EAs to coordinate in the absence of the ability to communicate, but as a practical community standard, we need to understand what problems we might be trying to solve with such a standard.
mhendric and the aforementioned journalists have already articulated some of them:
Encouraging a level of giving that is effective, sustainable, and motivating.
Choosing a number that orients discussion in a productive way.
Aligning our standards with those of major existing altruistic traditions.
Maintaining consistency in the EA movement.
I agree with mhendric that these are the three main problems we are trying to solve with our community pledge standard.
With both a 2%/8% fuzzies/utilons pledge and our current Giving What We Can 10% pledge, we achieve an equally sustainable level of giving, while the 2%/10% fuzzies/utilons pledge is equally effective.
So our disagreement is over the other problems.
The 2%/10% choice is slightly less sustainable, since the total donation level is higher, while the 2%/8% choice is slightly less effective, since the effective donation component is lower. These differences are real. But a simple comparison assumes that the choice of pledge does not have an effect on the total amount of donations received. In other words, two people donating $2,000 to the opera and $8,000 to GiveWell contribute more to fighting malaria than one person donating $10,000 to GiveWell. So our more important question is how the choice of standard would motivate donations.
Regarding motivation of donations, let’s note that “orienting discussion” and “aligning standards” are things we care about primarily for their instrumental effects in motivating donations. Maintaining consistency is something we care about to both motivate and sustain donations. How our choice of community standard affects the inflow and outflow of committed donors is what’s truly at issue.
My arguments for a 2%/8% fuzzies/utilons standard have made similar appeals to a practical benefit for sustainability and motivation.
First, my perception is that the “10% to effective causes” standard orients the discussion in a way that encourages many to interpret us as follows:
EA wants me to donate 10% of my income
EA wants me to give 100% of my donations to charities they deem effective
Nicholas Kristof’s article above doesn’t engage with a 10% standard—more like a 30-50% standard. But Kristof explicitly articulates the second idea—that EA wants him to make 100% of his donations to our preferred charities.
In fact, a sophisticated EA might point out that he could donate 10% to EA charities, and then extra on top of that to whatever causes he prefers (in line with my 2%/10% pledge). But that is not what Kristof took away from his discussion with Matt Wage.
I believe it’s well known that EA has a reputation among critics for being absolutist. I covered some of those criticisms in my last post. I believe that an “X%-to-effective charities” standard reliably contributes to this perception, not because it is an accurate takeaway, but because it is what outsiders understand when they hear this in the context of our discussions with them. If we had the chance to explain in more detail, we could enlighten them. But we won’t get that chance. For example, Nicholas Kristof is unlikely to read an email from me explaining this point and then publish an updated version of his original column.
By opening the explanation with a 2%/8% fuzzies/utilons pledge, it gives the person we are talking to a chance to see that we approve of their multiple loyalties and have an explicit community standard that lets them see how they can satisfy the competing demands of their moral parliament.
In turn, this aligns with what Lewis-Kraus calls “traditional liberal concerns.” Of that portion of society that believes in charitable giving at all, the vast majority of it believes that there is hidden virtue in giving to the charities they currently support. Ideas of obligation, loyalty, insider knowledge, cosmopolitanism, liberalism, risk-taking, democracy, systemic change, beauty, religion, and many other values guide people’s choices about charitable giving. If a 10% standard aligns with religious tithing, a 2%/8% split aligns better with traditional notions of liberalism and moderation.
I argue that EA is primarily trying to appeal not to religious donors, but to secular donors. As of 2019, EA was 86% agnostic/atheist/non-religious. Furthermore, a 10% religious tithe is typically supposed to be made to one’s church. EA is not a church, and aligning with a 10% tithing obligation seems unlikely to be useful, because most of those who actually consider themselves under obligation to tithe will not consider donating to EA in the first place.
Thus, I believe that appealing to the cultural norms of the communities EA is trying to reach as donors is most important. And our critics most often criticize our donation standards as preventing them from donating to secular causes, such as alma maters, political movements, the arts, and so on. Showing them that there is a way to include these interests in their giving, while still saving lives in a way that can be demonstrated with cost-benefit analysis in the manner of Effective Altruism seems to be a promising strategy to me.
This leaves the issue of maintaining consistency. Clearly, a 2%/8% fuzzies/utilons pledge maintains some of the consistency. In fact, since the 2% portion is optional, all those currently donating 10% to EA charities could continue doing so with no disruption at all. If the 10% community standard is all that’s preventing a subset of current pledgers from redirecting 20% of their annual giving to the opera instead of GiveWell, are we really pleased that the 10% community standard is having that effect?
I think the main concern is that shifting to a 2%/8% pledge might alienate more potential donors than it attracts—that there is some motivating and sustaining power in the way we currently structure discourse primarily around the idea of a 10% pledge to effective charities. As mhendric put it:
I have no reason to doubt mhendric is describing their experience accurately. However, I think this then becomes a messaging problem to solve in cases like theirs. And for others, the 2%/8% pledge would be solving a messaging problem. Figuring out the right language and context to bring it up would be key, and one benefit of the 10% standard is that we have worked some of that out already. There would be switching costs to a new standard, perhaps substantial. This is one of the true downsides to changing our messaging and community standards. These costs include changing descriptions on our various websites, figuring out new ways to describe the 2%/8% pledge, and evaluating whether it is in fact as useful as I believe it would be in driving and sustaining donations.
The question is one of costs and benefits, as always. I cannot complete the cost-benefit analysis here. We need empirical data on how a broad segment of EAs and non-EAs would react to the idea of making such a change, when presented well. My goal for now is to continue addressing the conceptual questions and critiques raised by commenters here. At some point, I hope either to have them change my mind, or to change theirs and move from the realm of discussion and into the realm of a more substantial empirical investigation.
I thank all readers for their continued engagement and thoughts.