If you want to disagree with effective altruism, you need to disagree one of these three claims

Effec­tive al­tru­ism is of­ten mo­ti­vated by ap­peal­ing to Singer’s pond ar­gu­ment.

This is good be­cause it’s a strong, con­crete and well-stud­ied ar­gu­ment. How­ever, two down­sides are that (i) it as­so­ci­ates effec­tive al­tru­ism with in­ter­na­tional de­vel­op­ment (ii) it makes it seem like you can re­fute the im­por­tance of effec­tive al­tru­ism by re­fut­ing the pond ar­gu­ment.

In fact, the im­por­tance of effec­tive al­tru­ism is much more ro­bust than the pond ar­gu­ment. It in­stead re­lies on what I call the “gen­eral pond ar­gu­ment”. If in pro­mo­tion of effec­tive al­tru­ism we fo­cus more on the gen­eral pond ar­gu­ment rather than origi­nal pond ar­gu­ment, then we can make the case for effec­tive al­tru­ism more strongly, and in a way that’s not so tightly as­so­ci­ated with in­ter­na­tional de­vel­op­ment. This might work bet­ter with some au­di­ences.

In the rest of the post, I sketch the gen­eral pond ar­gu­ment, ex­plain how to use it to clar­ify ob­jec­tions to effec­tive al­tru­ism, then sug­gest what this might mean for pro­mot­ing the ideas.

In­tro­duc­ing the gen­eral pond argument

The origi­nal pond ar­gu­ment goes as fol­lows:

1. If you can help oth­ers a great deal with­out sac­ri­fic­ing some­thing of similar sig­nifi­cance, you ought to do it. For in­stance, you ought to save a child drown­ing in a pond, even if it would ruin some ex­pen­sive clothes.

2. We can help the global poor with lit­tle cost to our­selves by giv­ing to effec­tive char­i­ties.

And (1) and (2) im­ply that:

3. We ought to give our wealth to effec­tive char­i­ties un­til it be­comes a sig­nifi­cant sac­ri­fice.

One prob­lem with bring­ing up this ar­gu­ment is that it of­ten leads to a de­bate about whether (2) is true i.e. whether in­ter­na­tional aid re­ally helps the global poor.

How­ever, you can deny that in­ter­na­tional aid works, but still think that effec­tive al­tru­ism is im­por­tant.

Let’s call an ac­tion that benefits oth­ers a great deal with lit­tle cost to your­self “pond-like”. Effec­tive al­tru­ism is im­por­tant so long as there are some pond-like ac­tions to be found.

Here are some ex­am­ples of pond-like ac­tions that are widely dis­cussed in the com­mu­nity:

1. Donat­ing to GiveWell-recom­mended and ACE-recom­mended char­i­ties.

2. Per­suad­ing oth­ers to make these dona­tions.

3. Giv­ing up fac­tory farmed meat, and per­suad­ing oth­ers to do the same.

4. Vot­ing in close elec­tions, where you think one can­di­date would be much bet­ter than an­other.

5. If you have a good fit for the area, do­ing a wide va­ri­ety of high-im­pact ca­reers, such as re­search, earn­ing to give and ad­vo­cacy, or work­ing at the Open Philan­thropy Pro­ject.

6. Pro­mot­ing effec­tive al­tru­ism.

And many oth­ers within each prob­lem area.

(Some of these in­volve more sac­ri­fice than oth­ers, but that’s OK so long as the ac­tions that re­quire greater sac­ri­fice also do more good.)

If any of the ac­tions above are truly pond-like, then effec­tive al­tru­ism is an im­por­tant idea.

We can lay out the case more for­mally with the “gen­eral pond ar­gu­ment”:

A. If you know there are pond-like ac­tions, and they don’t vi­o­late the or­di­nary rules of moral­ity (e.g. vi­o­late rights), you ought to do them. (This is the moral claim.)[1]

B. There are some pond-like ac­tions that are not already widely taken. (The em­piri­cal claim.)

C. We can come to know which ac­tions are pond-like, in par­tic­u­lar by us­ing ev­i­dence and rea­son (edit: “more than peo­ple nor­mally do”). (The epistemic claim.)

A, B and C im­ply that there are some pond-like ac­tions that we ought to be tak­ing that aren’t cur­rently taken.

We could see the mis­sion of effec­tive al­tru­ism to be to iden­tify these ac­tions, and help them to be­come more widely adopted, thereby do­ing a lot of good.

Why are there so many pond-like ac­tions? (and so, why is effec­tive al­tru­ism im­por­tant?)

Any of the fol­low­ing five ob­ser­va­tions im­plies there will be lots of pond-like ac­tions. We tend to only just on the first of these, but I think all five are sig­nifi­cant.

1. Global in­equal­ity. Col­lege grad­u­ates in de­vel­oped coun­tries are about 100 times richer than the global poor. That means these peo­ple can do about 100 times as much good if they take ac­tions to help the global poor rather than them­selves. If the benefits are 100 times larger than the sac­ri­fice in­volved, then the ac­tion is pond-like. Most sim­ple, you could do this by trans­fer­ring your in­come to the global poor. How­ever, you can prob­a­bly find *even more* effec­tive ways to help the global poor, such as health in­ter­ven­tions or sup­port­ing greater in­ter­na­tional mi­gra­tion, which would raise the ra­tio of benefits to costs over 100.

2. Mo­ral con­cern for an­i­mals. If you be­lieve the wellbe­ing of an­i­mals is morally im­por­tant, then there will prob­a­bly be pond-like ac­tions. This is be­cause an­i­mals have no eco­nomic or poli­ti­cal power, so are un­able to pro­tect their own in­ter­ests. This will mean we should ex­pect there to be ways to benefit lots of an­i­mals with small costs. One sim­ple ex­am­ple is go­ing veg­e­tar­ian. The av­er­age Amer­i­can eats about 100 an­i­mals per year, al­most all of which live in fac­tory farms, so by go­ing veg­e­tar­ian you pre­vent 100 an­i­mals from liv­ing in ter­rible suffer­ing each year for a small cost to your­self. And again, you can prob­a­bly find even more effec­tive ac­tions than this.

3. The abil­ity to af­fect the fu­ture. There will be many more peo­ple liv­ing in the fu­ture than are al­ive to­day. If you be­lieve we should have moral con­cern for fu­ture gen­er­a­tions and some of our ac­tions can af­fect them, then it raises the pos­si­bil­ity of pond-like ac­tions. One sim­ple way we can af­fect all fu­ture gen­er­a­tions is to cause hu­man ex­tinc­tion, so if there’s any­thing we can do to­day to make hu­man ex­tinc­tion less likely, then there’s a good chance those ac­tions will be pond-like.

4. The pos­si­bil­ity of lev­er­age. If you fo­cus on find­ing the best ways to help oth­ers, you can of­ten find ways of do­ing good that are higher lev­er­age than just do­ing good things your­self. By “higher lev­er­age”, I mean some­thing like “make use of more re­sources than just your own”. For in­stance, if you think some ac­tion, A, is good, you can prob­a­bly find a way to get 10 peo­ple to do A. This ac­tion is 10 times higher im­pact than A it­self. So, even if A isn’t pond-like, there’s a good chance that 10xA is pond-like. If you think there are many op­por­tu­ni­ties for lev­er­age, then there will be many pond-like ac­tions. I think many op­por­tu­ni­ties for lev­er­age ex­ist be­cause few peo­ple aim to have a large so­cial im­pact.

5. Poor ex­ist­ing meth­ods. Many cur­rent at­tempts to do good aren’t very strate­gic or ev­i­dence-based. Given this, if you ap­ply more ev­i­dence-driven meth­ods, you might think you can find ways to do­ing good that are 10 or 100 times bet­ter than what peo­ple nor­mally fo­cus on. Maybe nor­mal ways of do­ing good aren’t pond-like, but some­thing 10 times more effec­tive is pond-like. So tak­ing an ev­i­dence-driven, strate­gic ap­proach could mean you find lots of pond-like ac­tions.

I think all five of these are prob­a­bly true. In brief, we live in an un­in­tu­itive world, where there are mas­sive in­equal­ities, and our ac­tions have diffuse, but sig­nifi­cant effects on oth­ers. This means our moral in­tu­itions reg­u­larly mis­fire. Although it doesn’t feel like it, we’re sur­rounded by chil­dren drown­ing in ponds. And this is why effec­tive al­tru­ism turns out to be a novel and im­por­tant idea.

How not to re­fute the im­por­tance of effec­tive altruism

To dis­agree with effec­tive al­tru­ism, you need to dis­agree with one of the three claims in the gen­eral pond ar­gu­ment.

Most cri­tiques of effec­tive al­tru­ism fail to hit the mark. Some com­mon mis­fires in­clude:

1. Equat­ing effec­tive al­tru­ism with util­i­tar­i­anism, then rais­ing clas­sic ob­jec­tions to util­i­tar­i­anism. How­ever, effec­tive al­tru­ism ac­tu­ally rests on the much weaker moral claim (A), that you ought to do ac­tions that are a great benefit to oth­ers with lit­tle cost to your­self (or even A’ just that these ac­tions are very good, but not obli­ga­tory). In con­trast, util­i­tar­i­anism would say you ought to do an ac­tion that’s a ma­jor sac­ri­fice, so long as it does slightly more good to any­one else. Utili­tar­i­anism also de­nies that any­thing mat­ters ex­cept welfare and that it’s per­mis­si­ble to vi­o­late rights in pur­suit of the greater good. Effec­tive al­tru­ism doesn’t claim ei­ther of these things. For more on these ob­jec­tions, see Prof. Jeff McMa­han’s Philo­soph­i­cal Cri­tiques of Effec­tive Altru­ism (down­load link).

2. Ar­gu­ing that a spe­cific ac­tion is not pond-like, or that effec­tive al­tru­ists fo­cus on the wrong pond-like ac­tions, for in­stance, by crit­i­cis­ing the effec­tive­ness of in­ter­na­tional aid. This crit­i­cism is just a helpful con­tri­bu­tion to effec­tive al­tru­ism’s mis­sion to iden­tify the best pond-like ac­tions. To show effec­tive al­tru­ism is a bad idea in gen­eral, you’d need *there are no* pond-like ar­gu­ments that aren’t already widely taken. We wrote more about these kinds of ob­jec­tions here.

3. Say­ing that effec­tive al­tru­ists think you should only sup­port char­i­ties that have ran­domised con­trol­led trial ev­i­dence be­hind them. In fact, we just rely on the much weaker claim (3) that there are some ways it’s pos­si­ble iden­tify pond-like ac­tions, of which ran­domised con­trol­led tri­als are just one tool. In­deed, we of­ten think it’s more effec­tive to fo­cus on ac­tions with a small prob­a­bil­ity of a large pay­off, rather than ro­bust ev­i­dence, as we wrote about here.

What types of crit­i­cism might hit the mark?

One op­tion is to re­ject the moral claim: deny that we ought to help oth­ers even if it would be lit­tle sac­ri­fice to our­selves. This is an un­pleas­ant route to go down, since it would prob­a­bly mean ac­cept­ing that there’s noth­ing wrong with let­ting a child in a pond drown in front of you.

Un­less, that is, you can show there’s an im­por­tant moral differ­ence be­tween the child drown­ing in the pond and all the other pondlike ac­tions that the effec­tive al­tru­ism com­mu­nity sup­ports. How­ever, this is much harder than just show­ing there’s a moral differ­ence be­tween spe­cific pond-like ac­tion (such as donat­ing to effec­tive in­ter­na­tional health char­i­ties) as sav­ing the child in the pond.

More­over, even if you suc­ceed in show­ing that the new pond-like ac­tions aren’t morally obli­ga­tory, they’d still be good things to do (su­pereroga­tory). You would have shown that be­ing an effec­tive al­tru­ist isn’t re­quired by moral­ity, but it’s still com­mend­able.

The sec­ond op­tion is to deny that there are any new pond-like ac­tions that we can come to know. Again, this is rel­a­tively easy to do in the case of a sin­gle pond-like ac­tion, but it’s much harder to show that *no* new know­able pond-like ac­tions ex­ist at all.

You’d have to show that:

1. *None* of the ac­tions listed above are pond-like.

2. No fur­ther pond-like ac­tions will be dis­cov­ered.

So far no critic of effec­tive al­tru­ism has shown any­thing like this.

The third, and most promis­ing op­tion in my mind, is to ac­cept that effec­tive al­tru­ism as an idea is cor­rect—ac­cept­ing the gen­eral pond ar­gu­ment—but deny that effec­tive al­tru­ism as a move­ment will suc­ceed in do­ing a lot of good. Per­haps it’s just too hard to per­suade peo­ple to do the right thing, or the cur­rent lead­ers of the move­ment will fail, or we’re bad at work­ing out which ac­tions are pond-like. Or per­haps there’s some much more im­por­tant way of do­ing good that we should do in­stead.

Con­clu­sion, and some po­ten­tial les­sons for pro­mot­ing effec­tive altruism

I don’t pro­pose we should liter­ally lead with the gen­eral pond ar­gu­ment, since it’s far too ab­stract. How­ever, it seems use­ful to have in the back of your mind while pro­mot­ing the ideas.

In par­tic­u­lar, when mo­ti­vat­ing effec­tive al­tru­ism, I sus­pect it would be use­ful to dis­cuss a wider range of pond-like ac­tions than just donat­ing to effec­tive in­ter­na­tional health char­i­ties.

If we can com­mu­ni­cate that idea that if *any* pond-like ac­tions ex­ist, then effec­tive al­tru­ism is an im­por­tant idea, we’ll be mak­ing the case in a much more ro­bust way than if we only fo­cus on a cou­ple of spe­cific ac­tions.

More­over, we’ll be mak­ing sure that crit­ics fo­cus on the core of ideas, helping us bet­ter learn and do more good.


[1] If you pre­fer to avoid mak­ing effec­tive al­tru­ism about moral obli­ga­tion, then you can re­place (A) with some­thing like (A’): “if you know there are pond-like ac­tions, it’s a very good thing if you take them” (i.e. mak­ing pond-like ac­tions su­pereroga­tory rather than obli­ga­tory).