Listen: Software Engineer Jeff Kaufman '08 on Effective Altruism

This spring, Jeff Kaufman '08 visited campus to give a talk, "Effective Altruism in Practice." Kaufman discussed how he began to donate half of his income in 2009 and connected with other people who wanted to make altruism an important part of their lives, coalescing into a small movement called "effective altruism."

Kaufman talks about his path - from earning money as a Google programmer specifically in order to donate it, to building a mobile money system in West Africa - and shares some of the ways other people have tried to put effective altruism into practice. 

Kaufman, a member of Giving What We Can, keeps a popular blog in which he shares his ideas on issues such as how to make donations more effective. 

Kaufman graduated with Honors from Swarthmore with double majors in computer science and linguistics. He works as a software engineer at Wave.


Audio Transcript

Speaker 1:  We are very happy here to invite Jeff Kaufman who works at Wave. Jeff was an honors major in computer science, and linguistics ... a double honors major, actually. Is it?

Jeff:  Yeah.

Speaker 1:  Okay. You know what that means, right? Okay.

He graduated, and he worked for the ITS where he was a sophomore after graduating in 2008. He worked as a software engineer for many tech companies, including Google. Now Jeff left Google just this year, and works as a senior software engineer in Wave, a startup company that allows people to send money to East African communities directly through their Google app and in the Apple Store, which he will talk about today.

Jeff is also an active figure in effective altruism and community, and has been invited to talk at EA Global conference in California. His story is well covered in the Daily Mail and many books, including Peter Singers, the most you can do, which is a very good book, very good introduction to effective altruism. If you haven't read it, I highly suggest it.

So in that book his story is mentioned as a leading example of earning to give. Together, he and his wife, Julia Right has been consistently donating 30% to 50% of their annual income to effective charities. So far, I just looked it up on this website, their donations have exceeded $679,000. So more than half a million. And today, this number will add $400 because Jeff generously donated the auditorium for today's talk against malaria condition.

He also keeps a popular blog, Jeff2k.com, sharing his ideas on effective altruism and other cost effective measures. Including how to add warmers to their house actually. So it's a very fun website, and I highly suggest you to check it out.

Besides all this accomplishment, he's also a musical artist and has written tunes for hundred bands. Needless to say, Jeff is truly an interesting and inspiring person. So without further ado, Jeff Bachman.

Jeff: I'm really happy to be back here. I really like being at turf more, and I'm very happy every time I come back. It's changed remarkable little, I guess in 10 years. The thing still smells the same, it feels very similar. I'm here to talk about how effective altruism, as a way of sort of looking at the world, and looking at my place in it has effective what I decided to do with my life. So I'll be talking through that. I'm going to talk some about effective altruism in general, and also just what I've done and series of steps I've taken between doing random things trying to be helpful, and doing what I do now, which I think is quite a bit better than that.

I'm Jeff, I'm about 9 years out, Spring 2008, the engineering students made a balloon that went across filming applause at graduation. I was a CS student so I was not involved in this. The other thing about graduating in 2008 was, it was the beginning of a bigger session, which was not great, it made a lot of people have trouble finding jobs. It was a somewhat hard cohort to be graduating in. I spent that summer washing dishes, which was actually pretty fun. It was a combination of what I could find work doing, and also what I enjoy doing, and it was really nice having hands on work.

But I didn't spend all that long, I then switched to looking for actual programing jobs. And so I started at a technology company that created the internet, had the second oldest domain name ever. It's pretty cool from that prospective. I was interested in it for natural language processing. So writing software that tries to understand the meaning of text for purposes like translation, or answering questions about the text. There's a huge amount of text out there, and if we can efficiently process it with machines, there's tons of cool stuff you can do.

I was a computer science and linguistics major, so you'd think this would be a really good fit for me. It turned out it wasn't, at all. The problem is the parts of linguistics that I enjoyed doing, turned out to be the same parts that we made the computer do. The sort of solving puzzles and looking for connections, I really enjoyed that in linguistics class, but that was what we made the computer do. The remaining work was getting stuff into just the right format that the computer will be able to process from it, make progress on it, which was much less interesting.

In general, this is just an aside, my experience as a programmer, is that the correlation between how much I've enjoyed things and what domain I've been working in is just completely non existent. I've worked in natural language processing, I've worked in computational advertising, I've worked in web performance optimization, and now I'm working in international money transfer. How much I've enjoyed all of those, I would have had no good wat to predict that. They're very different domains and my actual enjoyment of the job tends to do much more with who I'm working with, do I have good tools, do I have good management. I only have a small number of data points, and maybe other things. But something to consider, initially when I was graduating I really though, oh I like this thing, and I like this thing, so I should get a job that really has both of those, and that turned out to be a very poor strategy for me.

But okay, working as a programmer. So I worked as a programmer for a while, I finished paying off my student loans. I started having this money that I didn't already have something that I needed to be doing with it, which was a very weird position for me to be in. I was on financial aid as a student. Senior year I was working 20 hours a week for ITS in the CS department. It was weird for me to have this money, that I didn't have an immediate need for, like paying back debt or whatever. Can I spend it on myself? Can I buy lots of video games, and computer equipment? What should I be doing?

So my then girlfriend, now wife said no, that's really not what you should do with your money. Other people need it more. And so here is a graph that I find pretty interesting for this. We've taken all the people in the world and we've lined them up from the people earning the least, to the people earning the most, okay? And we've given them hats, where you have a much taller hat if you are earning more money. Now there's some people in the world like kids, who don't earn any money. So we've grouped people into households, and so if you're household is earning let's say, you have two parents earning $20,000 a year and $20,000 a year, and you also have two kids with a total household income is $40,000, but you have 4 people so we have $10,000 so we would put all 4 of those people at the $10,000 mark. At $10,000 with the hat going up to here, at $10,000 a year, which would put them somewhere at 80th percentile in the world.

This is adjusting for money going farther in poorer countries, so it says PPP, that's purchasing power parity. So in general, in foreign countries there are some things that you can buy more cheaply than in richer countries. So this tries to adjust for it, it doesn't adjust for things like, it's cheaper to live in rural parts of the US than in Boston, or New York City or something. It doesn't adjust within country things, only between countries. So there's still maybe factor of 2 here. But it's about right.

And at that point, if you consider my household being me, as a just graduated software student. I was earning that when I got there, and you can see that puts me extremely far in this tail of very lucky people who have much more money than almost everyone else. This is important for two things. One is, if you're there you have much more power to change things, because you have money, and also someone who is at this tail of the distribution, down here, an additional dollar goes so, so much farther in their life than if you are at the other tail of distribution. Additional dollar does not make much difference in my life.

And we're kind of talking about privilege here, race and gender, make people more likely to see me as a potential engineer. Economic privilege, where my parents could afford a computer when I was growing up. And I have quite good education before I went to college. Nationality, I was born in a rich country, lots of opportunity related to that, including that Swarthmore had need blind admission for my category, but international students I think still don't have need blind admission. And this country has very well paying programming jobs. All of these things made it more likely that I would end up in this position. And also just luck, things that were outside of my control that were very helpful to me.

So what am I supposed to be doing with these unearned advantages? I should be trying to make things better. In 2009, I was like okay, I'm going to take this seriously, I'm going to donate half of what I earn. Then, at the end of the year I realized that I wasn't thinking about taxes and I did not have enough money to actually do that. Over the course of the year I had been trying to do it, and at the end of the year I realized I had fallen short and wasn't going to be able to make donations up to 50%. I managed to do a third, which I was sad about, but it was still quite a bit better than what I had been doing, which is not anything. And so I was trying to figure out how to do better than this.

One thing that came to mind is, well I've got this money. I am currently donating this much, and then this much does to taxes, and this much goes to me. If I spend less on myself I have more that I can give. So going to thrift stores, bicycling everywhere, trying to figure out how to make cheaper foods. All ways to try and spend less on myself. But then I talked to another software alum, Joseph Cenik, at someones wedding. Who played out, that actually there's much larger gains if you look at effectiveness, than if you look at spending less on yourself. Let's say I'm currently donating half, the complete ceiling for improvement there is to donate everything, which I'm not actually going to be able to do. At best, that's a factor of two improvement. If you can find someplace where your money can go three times as far, or ten times as far, that's a much higher ceiling. There's a lot farther you can go there. In fact, there's quite a lot of difference.

Some people put together, some sort of development economist, put together a list of estimates for a whole lot of different things you can do to try and make people healthier. These are things like, disturbing anti malaria nets, or AIDS antiviral drugs. You have lots and lots of things that people do in the field of local health and nutrition. Pretty much everything on this chart is worth doing. This is another chart like the previous one where we've shown everything starting from the ones where you get the fewest numbers of years of healthy life per dollar. Then the ones where you get the most, and you can see there's an enormous different here. Some of that tall tail is probably just error, is like when we go and measure things, some things we're going to have to get measures where we make it look really good, and these are just mistakes. So probably some of those are just mistakes. But still even if we chopped it in half or whatever, if we can try and pull from over here, instead of over here, it's enormously better to do that. So we need to prioritize.

There's an organization that is in a position to do much better for an organization than I am, like in my free time. It's called Give Well. It's a bunch of people who look at the research on different way of improving peoples lives, what they call interventions. Things like, you could give someone a chicken. Or you could try and from a prospective it could not be peoples lives, but it could be animals lives. You could try and persuade people to be vegetarian. Or you could do research on climate change. These are all the sort of things you might do if you're trying to make the world better. And it's really hard to tell which of these are better. And so, Give Well is trying to work on that, and they're making some progress. Better progress than you do by not trying, but there's still lots of things you cannot evaluate or had not evaluated very well. I think they are farther along in this direction than other public resources that I know about. So I learned about them from Joseph Cenik, and he's really positive on them. They were just getting started at that point.

And in 2011 I switch from ... we had been getting into Oxfam, spent pretty much an hour or two of looking up online at different charity website. But charity websites don't tell you very much because they're all just like we have smiling happy people, you should give to us. And that's not how you would make this decision. Because what you really care, is if I give them money, what would they do with it. And charities are going to say all sorts of things and not be super reliable sources. So I switched to following their recommendations.

So now I'm trying to spend less on myself, but with the money I'm donating I'm trying to give better by following some sort of expert recommendations. And then, maybe I could be earning more. So I was talking to another friend who said well, again maybe you could work somewhere like Google, they pay pretty well. So I tend to be sort of satisfy a certain way. I tend to be like eh things are fine, why do I need to do anything different, which is not really what I should be applying here, but is sort of my natural approach. Many people will be like, wait you could be earning much more and you weren't doing anything about that, you just kept doing ... well I was happy, I liked the people, I liked the work, what did I need to do?

But anyways, I got this push from Carl and a few other people. I brushed up on algorithms, I applied, I joined Google in 2012. And yes, they do pay more. So this on the left is what I was earning before Google, and then each year Google tended to pay me more, I'm not entirely sure why. But they paid me a lot, and I was able to donate a lot. In fact, by 2016 I was each year donating twice what I was earning in 2011. So trying to find a job that would pay me more was pretty worthwhile. This general approach of trying to find a job that will pay you a bunch so that you can have more money to donate, is often called earning to give, and is sort of the approach that I was taking at this time.

But still, even though I'm naturally a satisfy person, I should be an optimizer in terms of trying to do good things. Can I be doing better than this? And let's take a detour. In Kenya, there is a system called uncasa, and on your phone, which could be a little Nokia, or a smartphone, anything. You can send money to other people just using their phone number. And it's run through the tele code there, and you enter someones number you say how much you want to transfer to them and it's deducted from your balance and put in theirs. It's kind of like Venmo or something, but you can do it with technology that is not nearly as powerful as smartphones. And also it doesn't need a data connection, I think uncasa relies on SMS. So it works even in places with very poor telecommunications. And uncasa has been a really big success in Kenya, there's a bunch of research on this, in addition you can talk to people who are quite excited about it who are there. Lots of reasons to think it's very valuable. These are two papers that I really like, by the same people. I think they're at Jaypal. Anyway ...

So another thing that's out there is remittances. This is people sending money home to their families. So someone lets say grew up in Kenya, but came to the US and has a job in the US and is sending money back home. Taking advantage of the same chart I showed earlier where people in the US, even a relatively low amount of money in the US can be quite a lot, relative to somewhere else. So sending money back home, and over the entire world, we're talking about half a trillion dollars a year of people sending money home.

So right now, this tends to go through Western Union. And if I have $100, I want to send it to my family in Kenya, I could go to Western Union's website, and I could fill it out, and if I am going to spend $100 I can get about 9,000 Kenyan chillings. So I can do that transfer. Now if I go into Google and I search for the exchange rate between the dollar and the Kenyan chilling, it tells me that one dollar is 103 Kenyan chillings. If you look back on the previous page, the exchange rate their giving here is one dollar is 98 Kenyan chillings. So not only are they charging me a $7 fee, but they're also offering me a pretty bad exchange rate. They are making money in two ways, and the result of this is quite a bit of my transaction is going to Western Union instead of to my family at home.

The effective fee here is nearly 12%, it's very sad. This goes to debit card fees, cause they have to take the money off your card here. The cost of getting it there, they are taking some foreign exchange risk, so they have over head. But maybe we could do better, somewhat, by delivering to uncasa, instead of two Western Union agents who you have to pay to sit there in case someone shows up and wants money. So here, we've used that cash at Western Union location, and we switch it to mobile wallet, but the amount they get doesn't change, they are doing this way more efficient thing. But they are not passing any of those savings on to anyone, that's kind of depressing. So okay, we could do better than this.

Here is another way you could transfer money. This is called Wave. And with this, if you send $100, they will receive a little over 10,000 Kenyan chillings. All of the rates I'm giving are based on this morning, I just pulled it this morning. This effective fee is 2.7%, which is really quite a lot better. This is only a little bit higher than US debit card fees. You can't get much better than that and still pull from debit cards. So this seems really important, we have a huge amount of money moving, we have existing players who are taking lot of that money from themselves, instead of helping it actually go to the people who need it. I want to work for them, I kind of do. I joined there two months ago.

But before I worked there, I needed to consider, okay, they are doing this really cool thing, but if they get a new employee, what will they be doing? Because sometimes you have a case, like Wikipedia. Wikipedia does something incredible valuable, providing this free encyclopedia. But if they gain another employee, they're not going to provide ... let's say they have 50 employees, and they gain a 51st employee, they're not going to provide 2% more of Wikipedia. What they're going to do instead is they have other programs that are mostly less valuable than Wikipedia that they're going to work on instead. In general, if you're going to consider doing something you want to consider not just average impact. You take the total things they're doing and you divide by the people. Instead, what's the marginal impact, what will they do as a result of my contribution that they wouldn't do otherwise.

So in this case, the whole process of waste process of transferring money from debit cards to mobile wallet accounts requires a thriving mobile money system. And only a few countries are like that. But maybe we could ... Kenya definitely, and then Uganda and Tanzania, kind of, there aren't that many places you can send money to this way. But maybe we could build something. We did. And so if I get on my phone and I am in one of the countries that we've watched in, I could dial a number to connect to Wave, and it could say, hey welcome, I put in my name, it wants a PIN, because this is like banking and you want to be secure. So I put in the most secure PIN I know, and then it says, hey you are signed up. And then you would go to an agent, give them some money, and then your phone would talk to their phone, while you're pressing a whole bunch of buttons, and you would end up with a balance in your account, which you could then transfer to someone else.

And this is useful both as a place that you can receive transfer, so overseas like I was talking about. Also, as a place where you could transfer money within the country. Either as just instead of cash, or more commonly ... actually the same situation we had in the international case, but within a country. Like I am in a big city, and I want to transfer to my family in the countryside, instead of putting cash on a bus and having relatively hight transaction cost in terms of the risk that it would get stolen, the hassle, the needing to pay someone to do it. I can just transfer the money from phones and they can withdraw it from an agent there. Now the cash physically has to move, but now that's our problem, and not their problem.

So this is a good business choice for the company, but is it beneficial? With the remittance thing, by charging people lower fees on remittances, is very valuable, I can try and get a sense of how valuable that is, but since that seems not to be where I would be working as marshal employee, the actual question is, how good is it to have mobile money?

So again back to these two papers. These papers actually do a reasonably good job of estimating how good it is to have mobile money. What they did was notice when uncasa was rolling out in Kenya, the initial distribution of agents was kind of random. People would say, hey I want to be an agent, or they would go out and find somebody to be an agent and they would just end up sort of scattered around Kenya. What they found was people who lived near an agent did quite a bit better, well not quite a bit better, like one or two percent better than the people who happened not to live near an agent. And over the size of the whole country, this is really very valuable. If access to mobile money, combined with lower transaction cost. And they talked some about how it works.

The paper on the left, the risk sharing one shows that a big effect here is if I get sick, and then my relative wants to send me money to help me, if we have low transaction cost it can make sense to do that on the margin. In a situation where if we have high transaction cost, like having to physically move cash, it would not make sense for them to send me the money. So people in our panel sample adoption of mobile money increasing four to seven percent and normally, in a shock here being that something bad happens to you, normally that would reduce your consumption by 7%. Like you would decrease how much you eat, how much you spent on clothing, et cetera, by 7%. For households that were using uncasa, consumption was unaffected. Basically, they were able to be an informal insurance system for each other. Like relatives and community members were able to transfer money around, so that when something bad happened to one person, they didn't bear the full brunt of that, it could be spread over their whole network. Which is really good because one of the ways people stay in poverty is they are making some progress, and then something bad happens to them, which really pushes them back, and this has some promise for avoiding that. So a few reasons to think it's really good.

This is the thing I was saying, this is a big table of breaking down different ways of dividing a sample, with different ways of evaluating whether it's positive, and they mostly show that it's consistently positive. Depending on how you break it down. It's especially positive for houses that are female headed. Houses that are run by a woman, but it's valuable overall as well.

But switching to Wave does not pay as well. And I was earning to give, I was trying to end up with money so I could donate it, but the work has a direct value so that should could in the other direction. And so I made an actual and explicit value estimate trying to put numbers on things. Not that this is a wonderful way to do things but in general it's easier to full yourself into thinking you're doing the right thing if you don't try and put numbers on things than if you do. So it should be like trying to estimate out how valuable is this, forces you to say, okay I am going to have this chain that goes from my action to my effect, and I'm going to try and put numbers on this and if you can't come up with that chain, that's pretty worrying, and if you can't get the numbers maybe even get better numbers, maybe you have to use guesses. It can be useful for clarifying your thinking. Anyway ...

So I did this, and it looked a lot better than earning to give, but also there were the other reasons to think mobile money was really positive. And again, earning to give for me. I'm not saying no one should earn to give, but in my case it made sense for me to move away from that somewhat. I'm still planning to donate with the money that I earn now, but it's just less money because I'm in an early stage start up. Very interesting for sure.

Okay, summary of the stuff I've been saying. After graduating, I decided I should probably be more altruistic, I'm in this position where I'm really pretty lucky, I have more money than I actually need for myself. And then I should be focusing on how I should use this money in a more effective way, instead of picking a charity pretty much as random based on a pretty website, I should try and pick it based on what evidence there is for the stuff it does. Maybe I can earn more so I can donate more. Then in my case, switching to a job that had more of a direct effect on the world. That's been what I've been doing.

But there are quite a lot of other ways to try and put effective altruism to practice. So I have effective altruist friends who are working active, while in other places researching global health interventions, either doing the kind of summary that Give Well does, or as development economist, running new trials, I know some people that do that. I have some other friends who decided to start ... so vaccine writers over SMS is something that is people have run a few experiments on it and it looks really promising but there don't seem to be any organizations trying it. So I have some friends who are starting an organization in India to send automated vaccine writers. And I have other friends who are still earning to give in big companies like I was, some of them are at Google and other places. And all of these are quite reasonable paths. Quite possibly the best thing for that person to be doing, at least with what they know and what they can find out. And also people doing all sorts of other things.

And okay, thank you for coming.