Much has been written about the idea of optimal philanthropy. Yet, it seems like optimal philanthropy isn't a single claim. Instead, it's a collection of related, but quite distinct, claims that have all been bundled together, much like the Singularity.

Here's the website of GiveWell, and here's the main video introduction for 80,000 Hours, two of the major optimal philanthropy sites. I'll try to break them down into their component claims (written in bold), and also give my views on each of the claims. Some of them are explicitly stated, but others are more implicit, so I definitely welcome feedback if optimal philanthropists feel they disagree with some of the claims as stated.

1. We should evaluate charities according to how efficient they are, along some common metric - for example, number of lives saved per dollar, or existential risk reduction per dollar. We should then encourage charities to be more efficient, and selectively donate to (or otherwise help) the most efficient ones.

This one I support wholeheartedly. It's the main message of GiveWell, and though I have disagreements with their methodology, the basic idea (of marginal utility evaluation) is one that must happen more often. People are way too prone, by default, to donate to the Society for Curing Rare Diseases in Cute Puppies. Much has been written about this in Purchase Fuzzies And Utilons Separately, and other Less Wrong posts.

2. In order to do the most good for our fellow humans, we should start/work for/donate to/otherwise become involved in charitable organizations.

Of course, this is widely believed outside the optimal philanthropy movement. But I think this belief is inherent in many optimal philanthropy claims, and it ought to be examined more critically. It's plausible, but if one looks at the total good done over the last thousand years, the vast majority comes from science and various businesses, not popular causes like (back then) "tithe to the church" or (now, from the 80,000 Hours video) "campaigning against climate change". (Examples: electricity, air travel, enough food, trains, air conditioning... ) However, it's also true that much more total effort has been put into for-profit organizations than non-profits. Which one is more efficient per dollar, I don't know, but it's a question worth examining, rather than just ignoring it by default.

3. We should design careers around being able to donate the largest possible amount.

This one I see as highly damaging. Human psychology is such that, in order for a movement to get long-term, voluntary participation by highly capable people, stuff needs to be fun. Less Wrong itself, and the New York Less Wrong meetup group, are two obvious examples. "The more fun we have, the more people will want to join us."

Of course, the primary purpose of a community or activity doesn't have to be fun. Eg., Google doesn't exist for its employees to have fun. But working for Google still is fun, and if it weren't, Google would soon start losing people, become less productive, and ultimately go bankrupt. (Disclosure: I am a former Google intern.)

Writing a donation check can be very useful, but it isn't fun - it violates all the principles of Fun Theory. To go through the list, it isn't novel, doesn't involve tackling new challenges, doesn't engage the senses, doesn't get better over time (if we assume things work well, the marginal utility of dollars donated should go down, not up), doesn't involve long-term personal consequences, doesn't involve freedom of action, doesn't involve personal control over politics (assuming that one isn't personally involved in the charity, which is generally assumed), etc. etc. etc. (I'm referring to the actual act of writing the check here - for earning the money in the first place, see the next claim.)

Not everything in life is fun, nor can it be, at least pre-Singularity. Taking out the garbage isn't fun, but I do it anyway. However, trying to design lives around things that are inherently un-fun will probably lead to bad outcomes.

4. People can donate the largest amount through a traditional "high-earning career", like investment banking.

This one involves, to some extent, the classic American confusion between social class and income. One might think of "lawyer" as a high-earning career, since it's an upper middle class career; you need a graduate degree and dress up in suits. However, lawyer is actually a terrible career from a money-making perspective, and a law degree usually leaves people worse off financially (details here). Investment bankers themselves don't make that much money, except at the top levels (details here, and see here for general analysis of why gross pay isn't money in the bank).

In fact, all things being equal, one would expect a negative correlation between how prestigious a career is and how much money it makes. Prestige is, to some extent, a substitute for money - a musician might happily play for nothing, because being a musician is cool. "Where there's muck, there's brass" - for info on people who made millions doing boring stuff, see the excellent books The Millionaire Next Door and How To Get Rich.

But, even supposing that a "high-earning career" actually pays a lot (eg., partner at a Big Law firm), standard "career tracks" have serious disadvantages, like working insane hours doing unpleasant stuff. They sap what I call human capital and social capital - human capital is your skills, capabilities, and the value you can provide to an organization, while social capital is your network of friends and people who want to work with you. Human capital and social capital are the two critical things one needs to do anything, including world saving; they shouldn't be spent lightly.

5. People are morally responsible for the opportunity costs of their actions.

This is somewhat tricky/ambiguous, so I've deliberately made the wording vague, but the best example I've found is Peter Singer's argument (analyzed here). Singer compares philanthropy to a Trolley Problem. There's a set of train tracks, which a child is lying on, and a train is fast approaching. You're driving a luxury car, and if you drive the car on the tracks, the train will run into the car and save the child. What should you do?

In standard morality, the right thing to do is save the child, even if it means destroying your really expensive car. Indeed, we might socially shame someone who didn't. According to Singer, this means that we should be willing to donate any amount of money less than the price of an expensive car to charity, if it meant saving a life. Not donating to charity would be the same as letting the train run over the kid - murder through inaction.

I haven't figured out in detail what the real moral framework should be, but this argument doesn't work. For one thing, it produces atrocious incentives. Suppose you have a nice, cushy software job, and donate 10% of your income to charity, even though you could easily afford 20%. You work really hard, and a year later, you get another job for twice as much money. If not donating surplus money is morally equivalent to causing whatever bad outcome the donation would prevent, you are now twice as guilty, since the amount you aren't donating (20% vs. 10%) is twice as large. This is despite the fact that the total amount of good done is also twice as large. Why punish an improvement?

Another huge problem is the creation of unbounded obligations. I suspect a lot of thinking is inherently binary - you've either graduated college or you haven't, either paid back the loan or you haven't, either obeyed the rules or you haven't. With this line of argument, there's literally no point at which one can sit back and say, "I've fulfilled my duty to charity - there's nothing more to do". There's (short of FAI) always another child to save. One can never say, "I've met the goal", or even "I've gotten a third of the way to the goal", since the goal of solving all the world's problems is so huge. But if all states of the world - whether they be donating 0%, 10%, or 20% of income - result in 0% total goal fulfillment, then they're all equivalent, at least in some sense. A moral framework should make the good outcome and bad outcome as distinct as possible, not the same.

6. More people interested in doing good should become professional philanthropists.

This one I totally agree with, which might seem odd, given how closely related it is to #3 and #4. However, I think there are two important differences. A professional philanthropist is, typically, someone whose full-time job it is to figure out how to give away their money. But almost always, it's someone who already has lots of money. Historically, there isn't much precedent for people taking high-paying jobs and donating most of their salary... but there's lots of precedent for getting rich first, in whatever field, and then working full-time on donating.

The other difference is that professional philanthropists don't optimize for donating the maximum amount. They see donating as good, but they also see it as a good to be traded off against other goods, like having lots of nice stuff and social respect. Optimizing for more than one thing allows one to have a lot more Fun, as I suspect Bill Gates and Warren Buffett do.

This really does seem to be better than conventional routes of do-gooding. When I was in college, a huge number of people did stuff like fly to Africa to dig wells. This isn't just inefficient - it actually does net harm, since the cost of utilizing unskilled labor usually outweighs the benefits of such labor. Surely we can do better.

(Note to LWers: This post was written for a general audience at my blog, but I think it's particularly applicable to Less Wrong, as many here are already interested in programming. Programming is also an important route into two of the main paths to get rich, entrepreneurship and angel investing. Many of the leading donors to the Singularity Institute are professional programmers.)

You’ve already graduated college and found a job, but aren’t making the money you’d like. Or you live in the middle of nowhere. (Or, your job just sucks.) You’re pretty smart, and want to do something about this. What should you do?

One option is working as a computer programmer. Programming has a lot going for it: people tend to enjoy it, software companies have great perks, the work is usually in a laid-back atmosphere, and of course there’s no manual labor. Programming salaries generally range from high five figures (just out of college) to mid six figures (for senior people, and quants at Wall Street banks). This assumes you live in a major city, so be sure to factor that into cost-of-living math. (If you aren’t in a major city, seriously consider moving – most of the best jobs are there.)

Before you apply, you’ll need to learn how to program. To get started, there are lots of books on introductory programming – just search for “Introduction to C”, “Introduction to Python”, “Introduction to Haskell” and stuff like that. It’s good to know at least one language well, and also have experience with a few others, preferably ones that differ in important ways. Once you’ve learned the basics, there are lots of problems online to practice on. If you’re into math, Project Euler has a big, well-curated collection of them. You’ll also want to know your way around Linux, since it’s the main operating system of web servers; try installing it, and using it as your main OS for a few months.

To actually get a programming job, you’ll mainly need to demonstrate a) programming experience, and b) knowledge of computer science. For the first one, the most important thing is to build lots of real software that people use, since this teaches some vital practical skills you won’t learn in books or college classes. Working on open source projects is an excellent way to build experience – check out the list of Debian packages for stuff people are currently doing. You should also get an account on Github, browse some of the projects, and pay attention to the issues; notice how programmers talk about bugs and ideas for fixing them. It’s sort of like being an artist – you want to have a portfolio of past work to show employers.

As you learn to program, you’ll discover that a majority of programming is debugging – discovering errors, and figuring out how to fix them. When you get an error (and you will get errors constantly), Google, Google, Google. Especially when working with open source software, just about any concrete problem you encounter has been dealt with before. If there’s anything you want to learn to do, or an error message you don’t understand, Googling should be a reflex.

For the second part, a lot of interview questions ask about computer science concepts like linked lists, sorting, standard search algorithms, and so on. To learn these, you can usually read a standard freshman comp sci textbook all the way through, and do a few of each chapter’s problems. It might be helpful to write out code on paper, as practice for job interviews where you might not have a computer.

For most smart people, learning programming won’t be terribly difficult… but it does require sitting down at a desk every day, and putting in the hours. If you aren’t disciplined enough to work without a boss or a formal structure, you’ll have to figure out how to solve that problem first. Try working on problems you care about, ones that inspire you. It can be really helpful to work with a more experienced programmer, especially in person – they’ll review your work, correct your mistakes, and (more importantly) it’ll make you feel like you have to get things done on time.

Practicing programming isn’t like studying for a test – anything that requires flash cards or lots of memorization is likely a waste of time. “The best way to learn to program is by doing it.” Don’t memorize; build. If you have any task you find boring, annoying, or repetitive, see if you can write a program to do it instead. (This may require using other people’s software. Eg. if a website makes you type in the same number ten thousand times, Selenium is your best friend.)

A college degree isn’t necessary, but it can be very useful, since a lot of larger companies will only hire people with degrees (especially if you lack previous experience). If you don’t have a degree, try applying to smaller companies and startups, which have more flexible hiring procedures. Majoring in computer science and having a high GPA may help, but it won’t get you a job by itself, and a lot of places don’t care much. A lot of job postings will say things like “X years experience in languages A, B and C required” – ignore these, apply anyway. (Famously, one HR department once posted “ten years of experience required” for a language invented eight years ago.)

Some words of caution: Don’t work for anyone who offers you a full-time job without a decent salary (generally, at least $50,000) because you’ll be “paid in equity”. This is almost certainly a scam. If you want to get experience or learn skills, and someone is working on a cool project but doesn’t have money to pay you, do it on a volunteer basis so you won’t be tempted to see it as a job.

Also, don’t quit your current job to be a programmer, unless you either a) have lots of professional programming experience, or b) have some other way to pay rent. Generally, finding programming jobs in a major city is fast and easy (1-2 months), but a lot of people overestimate their skills, and you don’t want to run out of cash while discovering you aren’t as good as you hoped. It isn’t an overnight project; getting basic competence will take months, and true skill takes years.

Lastly, like most fields, don’t be afraid to apply to a zillion different jobs, and network, network, network. If you’ve been working with any other programmers for a while, ask (once, politely) if they know where you can get a job; they’ll likely have some ideas. Also, you might want to check out the Who’s Hiring thread on Hacker News. Go forth, and build the future!

(Disclaimer: This is based on my and my friends’ personal experiences. It isn’t applicable to everyone, and the reader must be responsible for the consequences of any decisions they make.)

This post was co-authored with Mike Blume, who went from having little programming knowledge as a physics grad student to being a software engineer at Loggly.

Further resources: How To Become A Hacker, Get That Job At Google