Strong agree with the third point, weak agree with the second point (the difference between a house and the pledge is that a house costs a fixed amount of money every month, whereas the pledge is a percentage of income), weak disagree with the fourth point (I think the pledge isn’t that weird, and that of those who give to GiveWell as a default, that would not be any different if GWWC were not around), and strong disagree with the first point (see below).
TL;DR of the rest: I disagree with the first criticism. I agree that the benefits to cost ratio changes, but the actual cost doesn’t, and for each individual person I think it makes sense to focus on an “acceptable cost”, and the easiest way to do that is a flat percentage. I don’t think the pledge encourages most donors to give too much, and I don’t think there are any professions where the value of their time outweighs the benefits of donation. Fermi estimates for Congressmen below.
“Different professions also have dramatically different ratios of direct impact to money generated. Eg., an American congressman’s salary is $174,000, but their votes in Congress are so important that even an 0.1% improvement in voting skill outweighs a $17,400 donation; hence, spending even a tiny amount of effort donating the 10% is likely not worth it.”
I doubt there are any professions where the “tiny amount of effort” to donate 10% of income is actually not worth it because of the direct impact of the job. For example, for Congressmen:
I’m going to operationalize voting skill in the following way—Suppose that a Congressman votes on N propositions, and currently X of them are “correct” (best for the world). Then, a 0.1% increase in voting skill means that 1.001X of the votes are now “correct”. If we assume that currently X = N/2, then a 0.1% increase in voting skill means an increase of 0.05 percentage points in the percentage of “correct” votes. (Choosing X = N/2 is the most generous possible choice for the Congressman in terms of the amount of marginal impact (s)he has.)
I’m not sure that a 0.1% improvement in voting skill would be worth more than a $17,400 donation.
Here’s a quick, stupid Fermi estimate—the budget each year is $4 trillion to the US people. Most of that is likely committed and can’t be changed, so let’s say Congress controls $1 trillion. Each Congressman is one of ~500, so each one should expect to control ~$2 billion (this will be higher for those in the Senate and lower for those in the House). A 0.1% improvement in voting skill would mean that 0.05% of the budget is better allocated (see above), which is an improvement of ~$1 million. But this is to the US people, whereas GWWC donations usually go to the global poor. I’d say that $50 to the US people is about the same as $1 to the global poor (the ratio of the poverty lines in the US vs. the world is 20:1, but most of the money in the budget goes to the average American who makes 100 times more than the global poor, so let’s call it 50:1 total), so this is about the same as controlling $20,000 to the global poor. However, counterfactually if you didn’t vote, the money would still have been used for the American people, so let’s say it did half the good in that case (note that this means good voting doubles the good on average, which I think is generous). So really it’s more like giving $10,000 to the global poor.
This is so close to $17,400 that I’m basically split 50-50 on whether a 0.1% improvement in voting skill is actually worth more than a $17,400 donation to the global poor. On the one hand, Congress controls other things besides the budget (such as Presidential appointments), though I would guess that the effects are smaller than that of things in the budget, on the other hand, they don’t control the entire budget, and I think I was generous in several assumptions (such as what the counterfactual is, and how much funding Congress actually controls). Perhaps Congress can affect the global poor immensely (whether or not to go to war, what to do with foreign aid spending), but I suspect that’s still too small a part of the budget to significantly change the calculation.
I really doubt that a “tiny amount of effort” would lead to a 0.1% improvement in voting skill.
I’m interpreting “tiny amount of effort” as approximately two hours, which is about how long my parents take to make their annual donations. Then, if you do this 700 times, you’ll have double the voting skill, which comes out to about 35 work weeks. And this is only counting the Congressman’s time—if you include all the aides and others, you’d probably get that much time in under a month. (Of course, you probably shouldn’t count their time just as much as the Congressman’s time, but it should be counted somewhat.)
This seems really implausible—based on how I operationalized voting skill, this would suggest that in a single year, Congressmen would be able to double the number of votes they make that are “correct”. That seems very unlikely.
In addition, an increase of 0.1% voting skill seems like a very high-value thing that Congressmen can do—it doesn’t seem like the marginal use of their time. I’m pretty sure that there would be less important uses of their time (even within their job, not including personal time) that they could cut out in order to make time to donate.
My guess is that for most EAs who have an income, the target donation rate should be higher than 10% (certainly not <1%), though I think a flat 10% is the right number because it’s a good Schelling point. The exceptions would be people with a lot of student debt, people who don’t make a large enough income to afford the loss of 10% of their income, and people who plan to use their money for good in a way that is not “donating” (for example, saving up to start a nonprofit or something like that). I think the GWWC pledge was explicitly not aimed at the first two groups. The last group is more of a problem and I think that is a legitimate criticism of the pledge, but I think it’s a fairly small group and it’s reasonable for the pledge to sacrifice that group in order to have a simpler, more viral pledge.
Reading the estimate in a different thread, it turns out I forgot one factor—the laws passed by Congressmen can last decades into the future, and so have an impact there. This will in general tend to cancel out the effect where “most of the budget is already committed and can’t be changed”, so I should undo that, which was a multiplier of 4x, getting you to an estimate of $40,000, which is still at the point where my uncertainty includes $17,400. So like maybe 70% belief that 0.1% improvement in voting skill is better than $17,400 donated to the global poor.
I still stand by my second claim that a “tiny amount of effort” would definitely not increase voting skill by 0.1%.
Strong agree with the third point, weak agree with the second point (the difference between a house and the pledge is that a house costs a fixed amount of money every month, whereas the pledge is a percentage of income), weak disagree with the fourth point (I think the pledge isn’t that weird, and that of those who give to GiveWell as a default, that would not be any different if GWWC were not around), and strong disagree with the first point (see below).
TL;DR of the rest: I disagree with the first criticism. I agree that the benefits to cost ratio changes, but the actual cost doesn’t, and for each individual person I think it makes sense to focus on an “acceptable cost”, and the easiest way to do that is a flat percentage. I don’t think the pledge encourages most donors to give too much, and I don’t think there are any professions where the value of their time outweighs the benefits of donation. Fermi estimates for Congressmen below.
“Different professions also have dramatically different ratios of direct impact to money generated. Eg., an American congressman’s salary is $174,000, but their votes in Congress are so important that even an 0.1% improvement in voting skill outweighs a $17,400 donation; hence, spending even a tiny amount of effort donating the 10% is likely not worth it.”
I doubt there are any professions where the “tiny amount of effort” to donate 10% of income is actually not worth it because of the direct impact of the job. For example, for Congressmen:
I’m going to operationalize voting skill in the following way—Suppose that a Congressman votes on N propositions, and currently X of them are “correct” (best for the world). Then, a 0.1% increase in voting skill means that 1.001X of the votes are now “correct”. If we assume that currently X = N/2, then a 0.1% increase in voting skill means an increase of 0.05 percentage points in the percentage of “correct” votes. (Choosing X = N/2 is the most generous possible choice for the Congressman in terms of the amount of marginal impact (s)he has.)
I’m not sure that a 0.1% improvement in voting skill would be worth more than a $17,400 donation.
Here’s a quick, stupid Fermi estimate—the budget each year is $4 trillion to the US people. Most of that is likely committed and can’t be changed, so let’s say Congress controls $1 trillion. Each Congressman is one of ~500, so each one should expect to control ~$2 billion (this will be higher for those in the Senate and lower for those in the House). A 0.1% improvement in voting skill would mean that 0.05% of the budget is better allocated (see above), which is an improvement of ~$1 million. But this is to the US people, whereas GWWC donations usually go to the global poor. I’d say that $50 to the US people is about the same as $1 to the global poor (the ratio of the poverty lines in the US vs. the world is 20:1, but most of the money in the budget goes to the average American who makes 100 times more than the global poor, so let’s call it 50:1 total), so this is about the same as controlling $20,000 to the global poor. However, counterfactually if you didn’t vote, the money would still have been used for the American people, so let’s say it did half the good in that case (note that this means good voting doubles the good on average, which I think is generous). So really it’s more like giving $10,000 to the global poor.
This is so close to $17,400 that I’m basically split 50-50 on whether a 0.1% improvement in voting skill is actually worth more than a $17,400 donation to the global poor. On the one hand, Congress controls other things besides the budget (such as Presidential appointments), though I would guess that the effects are smaller than that of things in the budget, on the other hand, they don’t control the entire budget, and I think I was generous in several assumptions (such as what the counterfactual is, and how much funding Congress actually controls). Perhaps Congress can affect the global poor immensely (whether or not to go to war, what to do with foreign aid spending), but I suspect that’s still too small a part of the budget to significantly change the calculation.
I really doubt that a “tiny amount of effort” would lead to a 0.1% improvement in voting skill.
I’m interpreting “tiny amount of effort” as approximately two hours, which is about how long my parents take to make their annual donations. Then, if you do this 700 times, you’ll have double the voting skill, which comes out to about 35 work weeks. And this is only counting the Congressman’s time—if you include all the aides and others, you’d probably get that much time in under a month. (Of course, you probably shouldn’t count their time just as much as the Congressman’s time, but it should be counted somewhat.)
This seems really implausible—based on how I operationalized voting skill, this would suggest that in a single year, Congressmen would be able to double the number of votes they make that are “correct”. That seems very unlikely.
In addition, an increase of 0.1% voting skill seems like a very high-value thing that Congressmen can do—it doesn’t seem like the marginal use of their time. I’m pretty sure that there would be less important uses of their time (even within their job, not including personal time) that they could cut out in order to make time to donate.
My guess is that for most EAs who have an income, the target donation rate should be higher than 10% (certainly not <1%), though I think a flat 10% is the right number because it’s a good Schelling point. The exceptions would be people with a lot of student debt, people who don’t make a large enough income to afford the loss of 10% of their income, and people who plan to use their money for good in a way that is not “donating” (for example, saving up to start a nonprofit or something like that). I think the GWWC pledge was explicitly not aimed at the first two groups. The last group is more of a problem and I think that is a legitimate criticism of the pledge, but I think it’s a fairly small group and it’s reasonable for the pledge to sacrifice that group in order to have a simpler, more viral pledge.
Reading the estimate in a different thread, it turns out I forgot one factor—the laws passed by Congressmen can last decades into the future, and so have an impact there. This will in general tend to cancel out the effect where “most of the budget is already committed and can’t be changed”, so I should undo that, which was a multiplier of 4x, getting you to an estimate of $40,000, which is still at the point where my uncertainty includes $17,400. So like maybe 70% belief that 0.1% improvement in voting skill is better than $17,400 donated to the global poor.
I still stand by my second claim that a “tiny amount of effort” would definitely not increase voting skill by 0.1%.