Contra Emile Torres On Unethical Jobs

Thinking about replacability is hard

I.

I’m working on a longer response to Emile Torres’s response to my critique of their arguments against so-called TESCREALism. But I wanted to pull out one specific argument, because I think it makes a particularly subtle point about career decision-making and is a good example of the difference between moral philosophy and real-life moral reasoning. Also because I would like to publish something.

II.

I claimed that effective altruist career-advising nonprofit 80,000 Hours has consistently advised against taking harmful jobs. But, Torres reasonably objected, that doesn’t mean that effective altruism consistently opposed taking evil jobs. They linked an article from Nathan J. Robinson, which in turn linked the 2013 paper Replaceability, Career Choice, and Making a Difference by effective altruist philosopher Will MacAskill.

Replaceability, Career Choice, and Making A Difference argues:

• [Weak Claim] It is usually better to do good by pursuing a high-paying but morally innocuous career than to do good by working for a nonprofit.

• [Strong Claim] It is often better to do good by pursuing a high-paying but morally controversial career than to do good by pursuing a low-paying but morally innocuous career.

MacAskill distinguishes “morally controversial” from “reprehensible.” Reprehensible careers (like hitman or concentration camp guard) violate commonly held non-consequentialist beliefs, like that you shouldn’t murder or torture people for money. Morally controversial careers, such as working for a petrochemical company or in the arms industry, are plausibly net-harmful. But common-sense morality doesn’t treat arms-industry employees the same way it treats murderers.

MacAskill provides two kinds of arguments in favor of his claims. The first are practical arguments about career choice. Sometimes people earn so much money that their donations can pay for more than one nonprofit worker. If you donate, you can donate to the absolute best cause, but if you work at a nonprofit, you’re likely going to be working for a cause that’s just okay. And if you’re wrong about what cause is best, it’s easier to switch donation targets than careers.

The second kind of argument MacAskill makes is about replaceability. It’s easy to assume that the positive effect of you working a particular job is the good you did. But, in reality, someone else would have probably worked that job if you didn’t. So your actual effect is the difference between this world and the world where someone else worked the same job. If you didn’t take the morally controversial job, MacAskill argues, someone else would, and they probably wouldn’t donate half their income to charity.

Before I start arguing with Replaceability, Career Choice, and Making A Difference, I want to make something clear. Everyone—me, you, Will MacAskill, Nathan J. Robinson, Emile Torres—agrees that people sometimes ought to work morally controversial careers. I can tell, because no one is #cancelling Oskar Schindler. “Sure, you saved thousands of Jews from the Nazis, but you made enameled cookware for the Nazi army! Collaborator! You should have quit your job and basked in your moral purity and left your employees to die!” No. Come on.

I could have got fewer out. I don't know. If I'd just... I could have got fewer. I didn’t do enough. If I’d quit my job… the Nazis could have had somewhat worse enameled cookware.

The question here isn’t whether people should work morally controversial jobs. It’s in what circumstances people should work morally controversial jobs.

Unfortunately, the kind of thought experiments brought up in Replaceability, Career Choice, and Making a Difference aren’t helpful for making the real-world decision of which job to work.

While they might have been true when Replaceability, Career Choice, and Making a Difference was written, many of MacAskill’s factual claims are presently false. For example, a number of career advice organizations have sprung up to help people choose the absolute best careers. There is no reason to suppose that a dedicated, talented graduate will wind up working a nonprofit job that’s just okay.

To simplify his thought experiments, MacAskill assumes a very limited array of available jobs. For example, he says that nonprofit workers are unable to switch jobs if they discover a new cause is more important. But a nonprofit employee who is uncertain which cause is best can invest in skills that are useful in a wide variety of causes: fundraising, accounting, communications, political lobbying, investigative journalism, and many more.

MacAskill’s unrealistically limited array of available jobs creates the most serious problems for the Strong Claim. Realistically, a trader engaged in wheat speculation that impoverishes the global poor could easily get any number of equally high-paying finance jobs. Wheat futures aren’t this incredibly specialized and uniquely high-paying area that leaves you unable to transfer to the nickel futures desk. Similarly, the CEO of a petrochemical company could become the CEO of a different chemical company, a defense-industry engineer could engineer cars instead, and a gacha game programmer could program literally anything else.

With a few exceptions like petroleum engineers, people who work high-paying morally controversial jobs have valuable skills that would enable them to get a different, less morally controversial job that pays almost as much. The people who are trapped in morally controversial jobs, such as factory farming, are usually relatively poor. MacAskill’s Strong Claim just doesn’t accurately reflect the world.

To similarly simplify his thought experiments, MacAskill supposes a person who has a choice between a job at a nonprofit and a job that pays more than twice as much.¹ Most people don’t face that decision. I mean, the most common jobs in the U.S. are home health care aide, retail worker, and fast food employee: a lot of people don’t have either option as a live possibility.

But let’s limit the discussion to relatively privileged people. Usually, people have a much higher personal fit for some jobs than others: that is, one job suits their skills and preferences much better than the other. If the only thing you love more than puzzles is betting on puzzles, you’d be ill-suited for vaccine development. Conversely, if you read papers about gene expression for fun, quant trading is not for you. Personal fit matters a lot: a crappy quant trader gets fired and doesn’t make much money, while a crappy vaccine researcher contributes nothing to scientific knowledge. If we narrow the claim to “if you are somehow equally suited for quant trading and vaccine research, you might make the world better as a quant trader donating two-thirds of your income than as a vaccine researcher,” it seems a lot more plausible.

Finally, I think MacAskill made a mistake in how he thought about replaceability. In general, if someone hires you for a high-skill job, it’s because they think you’re the best person for the job. For altruistic jobs, the difference between the best person and the almost-best person often matters a lot. The best person can get a bill to barely scrape its way through the legislature when it otherwise would have failed. The best person can motivate people who would otherwise have burned out and left the movement. The best person can come up with an innovative idea no one else would have thought of.

Further, if you take a job, the person who would have worked that job takes some other job. The vaccine researcher you displace will probably do some other kind of valuable medical research. In some cases—from technical AI safety researchers to charity founders to public defenders to people who perform late-term abortions—there’s such a shortage of people able and willing to do the work that if you don’t take the job it just won’t get done.

When deciding whether to take a morally controversial career, replaceability is something to think carefully about. Sometimes you can do less harm than your replacement: a voice for peace in the Department of Defense; a chicken company CEO who pushes for higher-welfare farms and meat alternatives; and, of course, Oskar Schindler. But sometimes you cause more harm than your replacement would: you program more addictive gacha games than they would; you invent a new weapon they wouldn’t; you commit one of the largest frauds in human history. If the latter is true, steer away from morally controversial jobs; you can probably make almost as much money at more ethical forms of employment.

III.

Emile Torres calls concern about harmful jobs “a relatively recent development.” I think this depends on what you mean by “recent”. 80,000 Hours wrote its first article advising people not to take harmful jobs in 2017. This is a “recent development” if by “recent development” you mean “more than half of 80,000 Hours’s existence.” And many effective altruists had advised people not to take sufficiently harmful jobs long before.

But Torres is correct that, as effective altruists have descended from the ivory tower and come into contact with reality, our thought about harmful jobs has grown more nuanced. Replaceability, Career Choice, and Making A Difference is from 2013, which was more than a decade ago. Compare it with Should you work at a frontier AI company?, an article which in my opinion represents the best of present-day effective altruist thought about morally controversial careers. The 80,000 Hours team thoughtfully goes through all the issues I discuss in this post, as well as others I didn’t have space to mention, such as the effect of a morally controversial job on one’s character. Their carefulness and nuance, and the clarity they provide on this complicated issue, is a model for ethical thinkers.

Academic moral philosophy incentivizes drawing counterintuitive and interesting conclusions: “did you know that, in some situations, it’s better to be an unusually generous petrochemical company executive than a social worker?” I’m not criticizing this endeavor in general: effective altruism itself comes from accepting certain counterintuitive and interesting conclusions, such as that you ought to help sick children even if they are located very far away from you. But “what is the weirdest edge case I can come up with?” is a different question than “what do I need to know to make the world better right now?”

The latter question, unavoidably, involves engaging with facts. You have to ask questions like “are you simultaneously independent enough not to give in to social pressure and charismatic enough to convince people to do what you want?” and “is this job something like ‘information security’ that robustly reduces the harm of your job or something like ‘fundraising’ that doesn’t?” and “is your employer going to use nondisparagement agreements to coerce you into not being a whistleblower?” This kind of careful engagement with the particulars of a situation doesn’t make for good philosophy papers—but it does make for good decisions.

IV.

The elephant in this discussion is that Nathan J. Robinson is a socialist and Emile Torres wants to return everyone to a hunter-gatherer lifestyle. When they talk about morally controversial jobs, they’re not really talking about defense engineers or petrochemical company CEOs: effective altruists don’t work those jobs. They’re talking about the kinds of jobs that earning-to-give effective altruists actually work: tech startup founders, quant traders, management consultants, cryptocurrency moguls.

I think the primary disagreement between modern effective altruists and Torres/Robinson is not at all about whether you should work morally controversial jobs. On this subject I think the two groups would share a remarkable degree of overlap. The disagreement is about whether tech startup founder, quant trader, and management consultant are morally controversial jobs.²

It’s true that some finance, tech, and consulting jobs are net-harmful, and a lot of them are just dumb. But finance and tech also make the world better. Like, for one thing, I can hail a self-driving taxi while chatting with my friends halfway across the world and selecting a book from the largest bookstore in human history.

But a lot of the ways that finance and tech improve the world are invisible. Some boring business-to-business Data Enterprise Logistics Analytics Cloud Marketing SAAS company shaves two pennies off the cost of your banana and three off the cost of your avocado, but if you multiply that by all the bananas and avocados in the world, it adds up. Better drug discovery doesn’t look like anything until you’re the one with the rare disease. Commodities futures stabilize the prices of everything from iron to rice to silicon, which means you aren’t suddenly paying five times as much for a toaster. The stock and bond markets give good companies the money they need to expand, so you’re not stuck on a waiting list for an electric car or a solar panel. And loans mean you can buy a house or a college education. I think consulting has similar invisible good effects but I’m still not sure what a consultant actually is.

And, yeah, a lot of finance or tech or presumably consulting is stupid. People used to pay programmers a lot of money to engage in an elaborate cash transfer from Saudi Arabian royalty to middle-class Americans. And nothing will explode if a trade in a particular security takes several entire minutes, so the ongoing effort to make them take fractions of a second instead is mostly dumb. This is what MacAskill called a morally innocuous job: a job that might not improve the world, but that doesn’t make it worse either. (It’s not like anything will explode if the trades are fast.)

Torres is intellectually aware that effective altruists are, on the whole, not anticapitalist. They repeatedly (and inaccurately³) characterize so-called TESCREALists as libertarian. They go so far as to say:

Historically speaking, utilitarianism (which consists of two components, one of which is the Total View1) emerged around the same time as capitalism, and hence I don’t think it’s surprising that they’re very similar: both want to maximize something without any limit—profit in capitalism, and “intrinsic value” in utilitarianism. Longtermism thus essentially reduces ethics to a branch of economics, and it’s this economic conception of ethics that leads to conclusions far, far beyond “we should care about future people”—that is, as most people, like myself, would understand that phrase.

But this awareness seems to desert them when they’re talking about whether effective altruists work evil jobs. It’s like they’re completely incapable of comprehending that a logical implication of being pro-capitalism is thinking that, in the usual case, the market rewards people with money for creating value for other people.

Robinson doesn’t even manage to reach this prerequisite, writing:

Effective Altruism’s focus has long been on philanthropy, and its leading intellectuals don’t seem to understand or think much about building mass participation movements. The Most Good You Can Do and Doing Good Better, the two leading manifestos of the movement, focus heavily on how highly-educated Westerners with decent amounts of cash to spare might decide on particular career paths and allocate their charitable donations. Organizing efforts like Fight For 15 and Justice For Janitors do not get mentioned…

I have to say, my own instinct is that all of this sounds pretty damned in-effective in terms of how much it is likely to solve large-scale social problems, and both MacAskill and Singer strike me as being at best incredibly naive about politics and social action, and at worst utterly unwilling to entertain possible solutions that would require radical changes to the economic and political status quo.

I am astonished at this failure to realize that someone might want to make the world better without being a very confused socialist. Effective altruists have supported both mass social movements (animal advocacy, Pause AI) and radical changes to the status quo. They just aren’t the ones Robinson likes.

Torres thinks that consulting, tech and finance are morally controversial occupations, because Torres is morally opposed to self-driving cars, chatting with friends across the world, incomprehensibly huge bookstores, bananas and avocados unless you live in an area where those naturally grow, drug discovery, toasters, electric cars, solar panels, modern housing technology, and college educations. Robinson thinks that consulting and finance are morally controversial occupations,⁴ because as a socialist he believes that they misallocate resources and cause immiserating poverty. Effective altruists (as a whole) don’t think that finance, tech, and consulting are morally controversial occupations because (as a whole) they are pro-capitalist.

There are a lot of debates to be had about whether finance, tech, and consulting are harmful to the world. I certainly don’t consider my viewpoint on the subject settled (especially since I still don’t know what consultants do all day). But let’s argue about the thing that we actually disagree about, instead of being caught up in moral philosophy utterly irrelevant to practical career decision-making.

1. If the person lives on the same amount of money either way and is paying the salaries of more than one nonprofit worker.

2. I will give them cryptocurrency moguls. If there’s one thing we’ve learned from the FTX Incident, it is that you should not be a cryptocurrency mogul in the hopes of donating most of your wealth.

3. Only six percent of effective altruists identify as libertarian, and three-quarters identify as left or center-left. A third of rationalists identify as liberal, and a quarter as libertarian. I’m not aware of a poll, but Marc Andreessen types tend more libertarian than rationalists. In general, I think even liberal-identified TESCREALists have notable libertarian sympathies. But as a group we’re much more Matt Yglesias than Ron Paul.

4. Presumably he’s fine with tech.