Why not give 90%?

(Cross-posted from my own blog. I also gave a talk on this at EAGxAustralia 2019.)

Agape donates 10% of her income each year to effective charities. This donation brings far greater happiness (or welfare, or whatever you think is valuable) to the recipients than it’d bring to her. And 10% is no small thing. (You could say it’s a good attenth.)

But Agape has a comfortable life. She could donate up to 90% before it really got tough. Before each dollar would do more good for her than it would for those in poverty, or those on factory farms, etc, whom she might benefit with it.

By only giving 10%, is Agape doing the wrong thing?

Agape and her dilemma might sound familiar. She’s me. She might be you too.

Personally, I give a lot less than 90%. So I worry about this a lot– am I doing something deeply immoral, by not sacrificing much more than I do? I suspect that many effective altruists have the same worry, and perhaps even feel guilty about not doing absolutely everything they can.

And this doesn’t just apply to donations. It’s similar for careers – you might, for example, go into a less high-impact career, maybe in academia, doing research which is a bit less beneficial to the world than some of the things you could do, but you really enjoy that research. Or, you could go do a job that you don’t enjoy at all. Maybe earning to give in finance could be an example of this, if that’s something you wouldn’t be able to bear. Not that that’s the most impactful option for many people, but imagine your own hypothetical high-impact career path that you’d find unbearable. The question is: by choosing a career which is more enjoyable but a bit less impactful, am I doing the wrong thing?

Or, even more importantly, am I not being a ‘Proper Effective Altruist’™?

Here’s my preferred, and pretty standard, definition of effective altruism.

• Effective altruism: the practice of 1) using some amount of your resources to improve the world, and 2) spending those resources in whichever way will have the greatest positive impact, according to the best available evidence.

Note that there’s no moral claim here, in the sense that it doesn’t say anything about what you should do. Effective altruism is simply something you do, rather a belief about what you’re obligated to do. Regardless of someone’s moral beliefs, if they use some amount of their time, money, etc to help others as effectively as they can, then they’re ‘doing effective altruism’.

In Agape’s case, giving less than she could, there’s no direct conflict between that and effective altruism.

But a lot of effective altruists do think we have moral obligations to engage in effective altruism – e.g., Peter Singer. I do too. I believe the following, which I suspect Peter Singer would agree with.

• Obligatory, demanding effective altruism: Whenever our resources would produce greater moral value if spent to benefit others than if spent on ourselves, we are obligated to spend those resources on others.

And anyone who believes this should conclude that we should use a large portion of our resources to help others. That would probably involve giving 90% of our income to charity each year, or to use the 80,000 hours in our careers to do whatever helps others most, no matter how unpleasant the job is.

Of course, this wouldn’t imply that we should give away 99% of our income, that we should sell every piece of clothing and wear potato sacks to work, that we should literally bring ourselves down to the poverty line, that perhaps we shouldn’t even spend money on water to shower. That would probably be counterproductive. If you want to keep earning money, or keep having an impact in other ways, you probably need to smell okay, and not show up to work in a potato sack. Or you’ll get fired. (Maybe not in academia though…) Plus, it’s probably worth paying to have a place to live, a good night’s sleep, a decent diet and so on, so you can keep your productivity up.

But basic hygiene and so on often won’t cost the majority of your income. Maybe 10%, leaving you 90% to give away. Or maybe it takes you 50% to satisfy those basic needs. It’ll depend on the person.

But whether it’s 90% or 50%, Obligatory, demanding effective altruism seems to entail that almost all of us should do a lot more than we currently do. So, are most of us doing something deeply wrong? Should Peter Singer be disappointed in us?

No, I don't think he should. I think we're doing okay. In fact, I think we may be required to give a lot less than 90%. Peter Singer should still be happy with us.

To make the case for this, I’ll first have to introduce you to Professor Procrastinate - a classic example from philosophy.

Professor Procrastinate

A student is applying for a grad job at the last minute and needs a reference. She emails Professor Procrastinate on Monday, asking him to write her a reference letter. If he says yes, she will send him more info on Tuesday. The reference is due on Friday.

Procrastinate is the best person to do the reference and he can do it on time. However, he knows that he (almost) certainly will not do it on time  (P<5%). He is a habitual procrastinator and will (almost) certainly choose not to finish it. This failure would not be due to outside factors (e.g., a natural disaster). What’s more, Procrastinate’s failure to deliver would have very bad consequences. The student would not have time to seek another reference, so would not get the job.

If Procrastinate says no on Monday, the student will ask Dr Reliable, who would write the reference on time. It wouldn’t be quite as good as Procrastinate’s – perhaps good enough to get the job at a lower salary - but this would be much better than no reference.

What should Procrastinate do on Monday? Say yes or no?

Of course, the best thing that Professor Procrastinate can do that week is to say yes on Monday and then write the reference on Tuesday-Friday. But we're interested in what he should do on Monday.

He can't control his future actions; he can't ensure that his future self will carry through. In fact, he's nearly certain that he won't. Perhaps past experience has shown him this over and over. And yes, that may make him a bad person. But does it mean that he should say yes this time?

Suppose I and Procrastinate were a two-person academic team. And it was up to me whether to answer emails - to say yes or no to the student - and then up to him to actually write the reference. Surely I shouldn't say yes unless I actually think he'll do it. If I'm almost certain he won't, I should say no. I think this situation is the same in the relevant ways as above - Present!Procrastinate and Future!Procrastinate are effectively separate agents. And Present!Procrastinate should say no - doing so saves the student from disappointment. (I talk more about the case for saying no in this post.)

Back to Agape now. Here's the case from above, but with some extra details. You might notice that it's starting to sound similar to the case of Professor Procrastinate.

Agape donates 10% of her income each year to effective charities. This donation brings far greater value to the recipients than it’d bring to her.

But Agape has a comfortable life. She could donate up to 90% before it really got tough. Before each dollar would do more good for her than it would for those in poverty, or those on factory farms, etc, whom she might benefit with it.

Agape really likes a bit of luxury, e.g., her sports car. Without her luxury comforts, she'd be 'demotorvated', you might say.

By her best estimate, if she had to subsist on 10% of her income, there’s a 50% chance each year that her future self would give up on donating altogether and keep everything for herself. She has 40 years left in her career.

By only giving 10% this year, is Agape doing the wrong thing?

Why would she give up on her plan to donate? It might be due to burnout, which is discussed a lot in the effective altruism community. Or it might just be due to changing her mind later on. I think that too is a risk worth predicting and mitigating. Just as it is for Professor Procrastinate.

If I was donating 90% every year, I think my probability of giving up permanently would be even higher than 50% each year. If I had zero time and money left to enjoy myself, my future self would almost certainly get demotivated and give up on this whole thing. Maybe I’d come back and donate a bit less but, for simplicity, let’s just assume that if Agape gives up, she stays given up.

And Agape has 40 years left in her career. Suppose she tries to give 90% each year. Then, over 40 years, the expected amount she donates is:

𝔼(amount donated) = 0.9 + 0.9×0.5 + 0.9×0.5^2+...+0.9×0.5^39 = 1.8

That's in units of "years' worth of income". And 1.8 years' worth of income isn't a huge amount.

What if she tried to donate just 10% each year? And she has no risk of giving up. Then we have:

𝔼(amount donated) = 0.1+ 0.1 + 0.1 + ... + 0.1 = 4

That's more than twice as much - more than twice the positive impact (assuming constant marginal impact per dollar). And, if we suppose that her income increases over time, then the difference would be even greater.

Now, what if she donated 20%? And that brought her up to 5% annual risk of giving up?

𝔼(amount donated) = 0.2+ 0.2×0.95 + 0.2×0.95^2 + ... + 0.2×0.95^39 = 3

Again, that's less than if she donated 10% with no risk! And that seems surprising. She's giving twice as much as she was at 10%; at 90% she was giving 9 times as much! But, with that additional risk, the expected total shrinks down to even less. This is because the probability of failure is compounding over 40 years, and ends up awfully high. For instance, with a 5% per year chance of giving up, that ends up being an 86% chance of giving up by the end.

So Agape’s total impact is more sensitive to changes in that annual probability here than it is to how much she’s actually donating each year. A similar result holds in general. Whenever she could donate twice as much per year and that would incur an additional 5% annual risk of giving up, it's actually not worth it! Any additional compounding 5% probability ends up cutting your impact in half over 40 years. Which is pretty counterintuitive, but that's probability for you.

Of course, I’ve made up all of Agape's probabilities. I don't know of any good data on the probability of effective altruists giving up based on their level of donation or career demandingness, so I’ve picked numbers out of thin air. If you, dear reader, want to figure out what to do in your own situation, you’ll have to figure out how much more likely you are to give up if you had a certain amount less spending money. You might be unaffected by that, or you might really struggle (as I probably would).

But, in general, I think we can justify donating less than 90%, since that'd be enough to make any of us very likely to give up. In fact, we might be able to justify donating a lot less, depending on how sensitive our motivation is to being deprived of nice things. Assuming that we’re at least a bit sensitive, we should probably donate quite a lot less.

In fact, donating a full 90% would then be reckless. You’d end up doing a lot less good. According to Obligatory, demanding effective altruism - the same view that initially recommended giving 90% - you’re actually morally obligated not to!