Thanks for the nitpick, fixed!

I mean something like "the average person in the EA community" or "the average person who is interested enough in EA's ideas and methods to take action" — that seems like the relevant audience for the question of what the community should focus on.

(This is complicated by giving being an easier pitch/reaching more people than career choice — but that's beyond the scope of this comment.)

Some thoughts:

1. I'm not sure what it means for a candidate to be "more cost-effective", but I assume it equates to "better at their job" (assuming a fixed salary).
2. My impression, from spending a lot of time in EA orgs and working on several hiring rounds, is that the difference between "best candidate" and "second-best candidate" is often well over 10% (though, to be fair, orgs may be wrong about which candidate is best).
3. And even a candidate who is only 10% "better" might have additional impact worth far more than 10% of their salary. Let's say an average org employee generates $2 million in value for the organization (seems realistic based on this). Getting the "best" candidate for a job would then be worth $200,000 to the org. If that employee earned a $100,000 salary (and donated $10,000/year), they'd generate more value by working than giving even if their donation target was 10x as cost-effective as their employer.
   1. The linked figures are old and could be flawed in many ways, but I'm not aware of anything better.
   2. The implicit assumption I'm reading from your model is that someone's impact within a job is worth exactly what they are paid. But I may be misunderstanding something — you have much more experience with this kind of modeling than I do!
4. Your model seems to assume that a donor will support the most cost-effective intervention in an area. But if less cost-effective interventions have a lot more jobs, doesn't that imply they also have more funding, and thus that the average donor is more likely to support them?
   1. This is hard to support empirically without some shared views on which interventions are best. But I will say that donors, relative to job-seekers, have historically been much more interested in global health than animal welfare or X-risk — if you think the most effective interventions are in the latter categories, that's some evidence in favor of prioritizing career choice over giving.
   2. If new donors did reliably support the most cost-effective interventions, that seems like it would open up many more jobs in those interventions and boost the impact of an expected career change (though perhaps not above the impact of donating).
5. People who decide to change their careers don't just add to existing interventions — they sometimes make it possible to support new interventions (e.g. by founding an org, or bringing a new kind of expertise into an existing org).
   1. Donors can also do this, but it's hard — I'd guess the average EA-interested person is more likely to be capable of starting a new org than providing sufficient seed funding to get one running.
6. In my experience, EA orgs tend to describe themselves as much more limited by talent than funding. This doesn't disprove your model — it could be that orgs are broadly wrong about this, or that the most effective orgs are more likely to be primarily funding-bottlenecked, or that the average person can do more to resolve a funding bottleneck than a talent bottleneck, etc. — but it seems like evidence that the model is incomplete.
7. In theory, the best outcome is for someone to change their career and start donating. And I think someone who changes their career is more likely to start donating than vice-versa. A career change is a big change to someone's environment/what they think about all day, and likely connects them to lots of other people who donate, while donating is a smaller change with little day-to-day impact on how the donor thinks about their career.
   1. If you think most people make far more impact by donating than working, and that it's much easier to convince people to donate than change careers, this factor won't change much; you'd still prefer lots of new donors to a small number of new workers who also donate. But I still thought it was worth noting.

I probably won't have time to respond to more comments at length (if at all), but I appreciate the impetus to think about the question!

Thanks! I didn't add Richard initially because he's said he doesn't think of himself as part of EA (didn't want to saddle him with associations he might not want). But his donation writeups are great, so I adjusted the language of my list to be more inclusive of anyone who shares an interest in EA-linked causes.

Maybe "naive" isn't the right language -- I mean it mostly in the sense of "it's a bad idea to commit crimes in the service of charity" rather than "the expected value was negative".

If Mario cared sufficiently little about being imprisoned, damaging a masterpiece, or generating opposition to famine relief writ large, I could see the theft as a positive-EV move from his perspective. But on the "benefit" side of the tradeoff, I'm skeptical that there was even a remote possibility of the Belgian government putting up ~$17 million to ransom the painting, especially on the deadline he set. (Claude notes that governments have a strong incentive not to set a precedent by making public ransom payments.)

That said, when I did some more reading on the case, I saw this:

"According to the Dutch art historian and Vermeer expert Albert Blankert, Roymans' demands were "alarmingly well received" by the Dutch public. Petitions were circulated urging authorities to rescind the warrant for his arrest and there were spontaneous initiatives in favor of the refugees from East Pakistan. A wave of sympathy spread across the country, and slogans in favor of Royman's cause were seen chalked on bridges and walls."

So it may have been a surprisingly effective publicity stunt, if the public's reactions were really so positive! (That's not something I'd expect in the modern world.) 

But I continue to think it's generally misguided to steal money so you can give it away,* for reasons including "I wouldn't want someone to steal my money for their own favorite charity" and "if your cause is known for support from thieves, you should expect people to turn against it".

*But if you can steal bread to feed your starving child, why not someone else's children? As the guy who played Javert in my high school's production of Les Mis, I can't help thinking about Jean Valjean here. But I'm not inclined to spend the time I'd need to work through the relevant arguments and counterarguments.

TIL: In 1971, Mario Pierre Roymans stole a Vermeer painting and tried to ransom it for a donation to starving Bengali refugees. It's an interesting example of naive altruistic utilitarianism before EA — inspired by the same famine that led Peter Singer to write "Famine, Affluence, and Morality".

(Roymans was apprehended and spent six months in prison; no ransom was paid. )

Excellent post! 

I appreciated the steelmanning, the clear action item, and the discussion of how EA's strengths could overlap with the need in this area. A lot of posts asking "why doesn't EA do X?" conclude that EA must be defective in some way for not making X a primary cause area; this post shows a refreshing understanding of community dynamics and makes a modest ask (making something a "normal problem" — good phrasing).

I was glad to hear about the Democracy Unconference, and I hope that people/groups with fewer legal concerns than CEA will continue to work on strategy in this space. It's tough for a movement built around elite persuasion and donations to work with nigh-unpersuadable elites in an environment where the wrong donation could disqualify you from a civil service job. But EA is capable of being flexible; we've found many different ways to create change over the last 15 years. Just on priors, I think there's room for creative action by the smart, motivated people in our tent.

As the person who led the development of that policy (for whatever that's worth), I think the Forum team should be willing to make an exception in this case and allow looser restrictions around political discussion, at least as a test. As Nick noted, the current era isn't so far from qualifying under the kind of exception already mentioned in that post.

(The "Destroy Human Civilization Party" may not exist, but if the world's leading aid funder and AI powerhouse is led by a group whose goals include drastically curtailing global aid and accelerating AI progress with explicit disregard for safety, that's getting into natural EA territory -- even without taking democratic backsliding into account.)

The core point of the post is, in fact, true! (And all the other points, to the best of my knowledge.)

I'll make the same offer Joseph did; I'd be happy to talk to you in depth about this post, review a resume, and try to be otherwise helpful. I've run some hiring processes, failed some applications, and had a lot of conversations with people in similar positions (thanks to this post).

Do you have any updates on whether the Los Angeles summit is happening?

I didn't see any mention of Loretta Mayer's work here. She is testing what seems to be a viable product in several major cities (here's some NYT coverage). Do you see this work as having a different purpose/target market?

(I only skimmed the post — sorry if I missed an obvious reference!)