Structurally, the basic logic of effective altruism is quite demanding, bordering on totalizing. Yes, most EAs are naturally very high integrity people in their hearts (in some ways the opposite of ruthless consequentialists); this sociological fact, plus a dash of esoteric decision theory and a sprinkle of pragmatic instrumental reasoning, means that the EA memeplex ends up very strongly discouraging violating norms for the greater good. But by default, there isn’t much of a limit on what the ideology can ask you to give of yourself.
But a real world community consisting of a large diverse group of people, almost all of whom will ultimately place a lot of weight on their own comfort and happiness, cannot practically coordinate around maximal sacrifice as the norm.
When it comes to donations, the EA community has borrowed from centuries-old social technology and settled on tithing. When it comes to work hours, most EAs basically default to the prevailing norms in their surrounding culture, which where I’m from is somewhere between the standard 9-to-5 and an intense hedge fund.1 Diet is the most heterogeneous, but big-tent EA spaces seem to have settled on a norm of offering beef and fish but not chicken, eggs, and pork, which cause substantially more animal suffering per calorie.
Percent of income donated, number of hours worked, and to some extent diet have two properties that make it easy to build norms around them: it’s relatively clearly observable what exactly everyone is sacrificing,2 and the total feasible sacrifice anyone could make is bounded (you can’t give more than 100%, you can’t work more than 100ish hours per week, you can’t eat less than no animal products). This makes it a lot easier for a group of humans to settle on and enforce some ultimately arbitrary middle ground without slippery-sloping down to zero or purity-spiraling up to infinity.
But there are a lot of other forms of sacrifice that are a lot harder to govern with the social technology we have.
When I was burned out and took time off to find myself a couple years ago, I talked to several fellow EAs about how they approached their life. One friend told me that back when the Thing to Do was earn to give at a hedge fund, he decided he wouldn’t do it even if it was arguably higher impact because it would eat into his soul too much. Now, by this point I was already many years into this EA gig and very much thought of myself as an annoying ecumenical elder rather than a hot-headed youth — but still, I was a little scandalized. Like, are you allowed to do that? When it would be the highest impact job?!
In the abstract, it doesn’t make much sense to be scandalized by that but not scandalized by people taking any days off or spending any money on themselves. I think a big part of what was going on for me was something like: “Well if we allow this, where will it end?” You can’t have an EA community that just lets you take whatever job you think is best for your soul; that kinda defeats the point. I can’t easily draw a line on how much you’re allowed to factor in your soul’s integrity into the career selection process, so I want to protect the norm that it just doesn’t count.3
There are many types of sacrifices you can make in a career. You can work long hours or work for little pay, yes — but also, you can be on call to jump into action at unpredictable hours. You can work on deeply confidential projects you can’t commiserate or celebrate with your loved ones. You can work alone, on something no one else believes in. You can do work that requires you to hide your true beliefs. You can be on the hook for making a lot of politically-loaded choices that will never make anyone happy and always make you feel dirty.4 You can spend your time dealing with Kafka-esque bureaucracies. You can take swings that could blow up in your face publicly. You can work with people you don’t like or respect.
The equilibrium right now seems to be that people mostly leave each other alone about career choices and emit some sort of bland message like “Impact is important but personal fit matters too.”5 Ideally, the community would develop some case law about this. Sometimes you really do need to take one for the team and do a really unpleasant job that is really important, and sometimes you need to unclench and let your freak flag fly.6 I wish we had scripts and case law about these kinds of choices; they are the most important ones we make.
More than one intense young EA has told me that when they first got sucked up into the professional scene, they were surprised and disappointed at how little most people worked.
At least on a basic level — of course, some people will have better returns to consumption than others, some people will experience more pain per hour worked, some people care more about the pleasure of eating good food or have health issues that make it more painful to be vegan.
To some extent, pushing to keep it out will drive it underground — people will be more inclined to find motivated reasons why the job that best feeds their soul is actually secretly their highest impact option. That said, this dynamic probably doesn’t claw back all the value.
On another one of my sabbatical conversations, another friend told me: “One thing I’m grateful for is that I get to feel thoroughly proud of my job. Not all jobs worth doing have that property.”
Our difficulty coming up with norms about this is very tied up in our difficulty coming to consensus about what’s high impact, beyond very crude cause-area-level heuristics like “AI risk is bigger than bio risk.” One of my deepest wishes for the community is that we had made better intellectual progress on these questions (sometimes called “macrostrategy” though I hate that word a lot), and now I think we have very little time to do it well.
One important dynamic here is that people are more likely to do well at their jobs if they like them, but it’s easy to take this valid heuristic too far — the job where you’ll have the most impact is probably not the one you’ll enjoy most.
A general category of sacrifice that comes to mind is: for some jobs, your impact consists in making your team or organization's output go from "good" to "great". For other jobs, much of the impact comes from turning "dumpster fire" into "merely bad".
I think some government jobs are like this, as well as working for understaffed safety teams at AI companies that don't value safety, or urgently joining to firefight a very messed up situation.
The latter jobs often feel worse on the day to day. My impression is that they are less socially rewarded, because people will have a tendency to assess the team/organization's overall output rather than how much the job holder improved it. But they might be where the most of the impact lies.
Current norms seem good. It would be bad to impose costs on someone just because they were doing less than they could, no matter how fair the sense of "what they could do" was.
Once someone is in the 99.7th percentile of acting on the principles you endorse, it's unwise to gatekeep them. Better to fully endorse those people and simply admire and celebrate people who go further.