FTX/CEA - show us your numbers!

I'm obviously not speaking for Jessica here, but I think the reason the comparison is relevant is that the high spend by Goldman ect suggests that spending a lot on recruitment at unis is effective. 

If this is the case, which I think is also supported by the success of well funded groups with full or part time organisers, and that EA is in an adversarial relationship to  with these large firms, which I think is large true, then it makes sense for EA to spend similar amounts of money trying to attract students. 

The relvent comparison is then comparing the value of the marginal student recurited with malaria nets ect. 

Curious if you disagree with Jessica's key claim, which is "McKinsey << EA for impact"? I agree Jessica is overstating the case for "McKinsey <= 0", but seems like best-case for McKinsey is still order(s) of magnitude less impact than EA.

Subpoints:

• Current market incentives don't address large risk-externalities well, or appropriately weight the well-being of very poor people, animals, or the entire future.
• McKinsey for earn-to-learn/give could theoretically be justified, but that doesn't contradict Jessica's point of spending money to get EAs
• Most students require a justification for anyone charitable spending significant amounts of money on movement building & competing with McKinsey reads favorably

Agree we should usually avoid saying poorly-justified things when it's not a necessary feature of the argument, as it could turn off smart people who would otherwise agree.

Sorry, I was trying to get a quick response to this post and I made a stronger claim than I intended. I was trying to say that I think that EA careers are doing much more good than the ones mentioned on average and so spending money is a good bet here. I wasn’t intending to make a definitive judgment about the overall social impact of those other careers, though I know my wording suggests that. I also generally want to note that this element was a personal claim and not necessarily a CEA endorsed one. 

Thirdly, each of these are (broadly) free market firms, who exist only because they are able to persuade people to continue using their services. It's always possible that they are systematically mistaken, and that CEA really does understand social network advertising, management consulting, trading and banking better than these customers... but I think our prior should be a little more modest than this. Usually when people want to buy something it is because they want that thing and think it will be useful for them.

I consider this to be a pretty weak argument, so it doesn't contribute much to my priors, which although weak (and so the particulars of a company matter much more), are probably centered near neutral on net welfare effects (in the short to medium term). I think a large share of goods people buy and things they do are harmful to themselves or others before even considering the loss of income/time as a result, or worse for them than the things they compete with. It's enough that I wouldn't have a prior strongly in favour of what profitable companies are doing being good for us. Here are reasons pushing towards neutral or negative impacts:

1. A lot of goods are mostly for signaling, especially signaling wealth, which often has negative externalities and I'd guess little positive value for the individual. Brand name versions of things, clothing, jewelry, cars.
2. Many modern ways people spend their time (enabled by profitable companies) have probably made us less active, more indoor-bound, less close with others, and less pursuant of meaning and meaningful goals, which may conflict with people's reflective preferences, as well as generally be bad for health, mental health and other measures of wellbeing. Basically a lot of the things we do on our computers and phones.
3. Many things are stimulating and addictive, and companies are optimizing for want, not welfare. Want and welfare can come apart when we optimize for want. So we get cigarettes, addictive video games, junk food, algorithms optimizing for clicks when we'd be better off stepping away from the internet or doing more substantial things online, and lots of salt, sugar and calories in our foods.
4. Media companies may optimize for revenue over accurate reporting. This includes outrage, playing to our fears, demonizing and polarization.
5. Some companies make us want their stuff for fear of missing out or social pressure, so it can be closer to coercion than providing a valuable opportunity.
6. I'd guess relatively little is spent on advertisement for things that we have good evidence for improving our welfare, because most of those things are hard to profit from: basic healthy foods, exercise (although there are certainly exercise products and programs that get advertised, but less so just gym memberships, joining sports leagues, running outside), just spending more time with your friends and family (in cheap ways, although travel and amusement parks are advertised), pursuing meaning or meaningful goals, helping others (even charity ads are relatively rare). So, advertisement seems to push us towards things that are worse for us than the alternatives we'd have gone with. To capitalize on the things that do make us substantially better off, companies may sell us more expensive versions that aren't (much) better or things to go with them that don't substantially help.
7. I'd expect a lot of hedonic adaptation for many goods and services, but not mental health (almost by definition), physical pain and to a lesser extent general health and mobility, which are worsened by a lot of the things companies provide, directly or indirectly by competing with the things that are better for health.
8. Company valuations don't usually substantially reflect their externalities, and shorting companies is riskier and more costly than buying and holding shares, so this biases markets towards positively valuing companies even if their overall value for the world is negative.
9. There are often negative externalities on nonhuman animals in particular, although the overall effects on nonhuman animals may be complicated when you also consider the effects on wild animals.

I do think it's plausible McKinsey and Goldman have done and do more good than harm for humans in the short term, based on the arguments you give, but I don't have a strong view either way. It could depend largely on whether raising people's consumption levels makes them better off overall (and how much) in the places where people are most affected by these companies. Measures of well-being do seem to positively correlate with income/wealth/consumption at the individual level, and I'd guess also at the aggregate level for developing countries, but I'd guess not for developed countries, or at best weakly so. There are negative externalities for increasing an individual's income on others' life satisfaction, although it's possible a large share is due to rescaling, not actually thinking your life is worse absolutely than otherwise. See:

1. Haushofer, J., Reisinger, J., & Shapiro, J. (2019). Is your gain my pain? Effects of relative income and inequality on psychological well-being.
   1. Based on GiveDirectly in Kenya. They had multiple measures of wellbeing, but negative effects were only observed for life satisfaction for non-recipient households of cash transfers in the same village. See Table A5.
2. This table from Veenhoven, R. (2019). The Origins of Happiness: The Science of Well-Being over the Life Course., reproduced in this post.
3. This graph, reproduced in this post.
4. Other writing on the Easterlin Paradox.

Some companies may also contribute to relative inequality or even counterfactually make the median or poor person absolutely poorer through their political activities.

The categories of things I'm optimistic about for human welfare in the short to medium term are:

1. Things that save us time, so we can spend more time on things that actually make us better off.
2. Things that improve or protect our health (including mental health).
3. Things that make us (feel) safer/more secure (physically, financially, etc.).
4. Things that make us more confident, but without substantially net negative externalities (negative externalities may come from positional goods, costly signaling, peer pressure).
5. Things that help us make better decisions, without important negative effects.

I'm neutral to optimistic about these (possibly neutral because they just replace cheaper versions of themselves that would be just as good):

1. In-person activities with friends/family.
2. Things for hobbies or projects.
3. Restaurants.

I'm about neutral and pretty uncertain about screen-based entertainment (TV, movies, video games), and recreational substances that aren't extremely addictive or harmful (alcohol, marijuana).

I'm pessimistic about:

1. Social media.
2. Status-signaling goods/positional goods/luxuries.
3. Processed foods.
4. Cigarettes.

Ok. Lark’s response seems correct.

But surely, the spirit of the original comment is correct too.

No matter which worldview you have, the value of a top leader moving into EA is overwhelmingly larger than the the social value of the same leader “rowing” in these companies.

Also, at the risk of getting into politics (and really your standard internet argument) gesturing at “free market” is really complicated. You don’t need to take the view of Matt Stoller or something to notice that the benefits of these companies can be provided by other actors. The success of these companies and their resources that allow recruitment with 7 figure campus centres probably has a root source different than pure social value.

The implication that this statement requires CEA to have a strong model of these companies seems unfair. Several senior EAs, who we won’t consider activists or ideological, have deep experiences in these or similar companies. They have opinions that are consistent with the parent comment’s statement. (Being too explicit here has downsides.)

I think the main crux here is that even if Jessica/CEA agrees that the sign of the impact is positive, it still falls in the neutral bracket because on the CEA worldview the impact is roughly negligible relative to the programs that they are excited about. 

If you disagree with this maybe you agree with the weaker claim of the impact being comparatively negligible weighted by the resources these companies consume? (there's some kind of nuance to 'consuming resources' in profitable companies, but I guess this is more gesturing at a leaving value on the table framing as opposed to just is the organisation locally net negative or positive.

Do you think people are better off overall than otherwise because of Facebook (and social media generally)? You may have made important connections on Facebook, but many people probably invest less in each connection and have shallower relationships because of social media, and my guess is that mental health is generally worse because of social media (I think there was an RCT on getting people to quit social media, and I wouldn't be surprised if there were multiple studies. I don't have them offhand). I'd guess social media is basically addictive for a lot of people, so people often aren't making well-informed decisions about how much to use, and it's easy for it to be net negative despite widespread use. People joining social media pressures others to join, too, making it more costly to not be on it, so FB creates a problem (induces fear of missing out) and offers a solution to it. Cancel culture, bubbles/echo chambers, the spread of misinformation, and polarization may also be aggravated by social media.

That being said, maybe FB was really important for the growth of the EA community. I mostly got into EA through FB initially, although it's not where I was first exposed to EA. If we think the EA community is important enough, then this plausibly dominates. And, of course, it's where Open Phil's funding came from, but that seems to be historical luck, not really anything special about Facebook, except the growth of its market cap.

On the other hand, FB accelerated the development of AI capabilities, e.g. PyTorch was primarily built by FB. But maybe we should also consider this to be only weakly related to FB's role in social media, and more related to the fact that it's just a large tech company.

There are also multiple counterfactuals we could consider: no Facebook + people spend less time on social media, and no Facebook + people spend about as much time on social media (possibly on one similar to FB, or whatever other options there are now). In the first case, I think it's hard to make a balanced argument for FB being robustly net positive. In the second case, the impact is closer to 0, from either direction, and it's harder to evaluate its sign. Then there's the counterfactual impact of FB getting a more productive hire, or one who is otherwise more valued by FB.

I think McKinsey and Goldman would have other firms step into their spaces if they weren't around.

I don't think this is persuasive. I think most actions people take either increase or decrease x-risk, and you should start with a ~50% prior for which side of neutrality a specific action is on (though not clearly true; see discussion here). I agree there's some commonsensical notions that economic growth is good, including for the LT future, but I personally find arguments in the opposite direction to be slightly stronger. Your own comment to an earlier post is one interesting item on the list of arguments I'd muster in that direction.

• Some non-CEA people have made estimates that we sometimes refer to. I'm not sure I have permission to share them, but they suggest significant value. Based in part on these figures, I think that the value of a counterfactual high-performing EA is in the tens of millions of dollars.
  • I think we should also expect higher willingness to pay than private firms because of the general money/people balance in the community, and because we care about their whole career (whereas BCG  will in expectation only get about 4 years of their career (number made up)).
• I'll let Jessica answer with more specifics if she wants to, but we're currently spending much less than $1m/school.
• Yes, it's obviously important that figures are counterfactually discounted. But groups seem have historically been counterfactually important to people (see OP's survey), and we think it's likely that they will be in the future too. Given the high value of additional top people, I think spending like this still looks pretty good.

I think the key point here is that it is unsually easy to recuirt EAs at uni compared to when they're at McKinsey. I think it's unclear if a) among the the best things for a student to do is go to McKinsey and b) how much less likely it is that an EA student goes to McKinsey. I think it's pretty unlikely going to McKinsey is the best thing to do, but I also think that EA student groups have a realtively small effect on how often students go into elite coporate jobs (a bad thing from my perspective) at least in software engineering.  

I also couldn't find much information on campus recruitment expenses for top firms. However, according to the US National Association of Colleges and Employers (NACE), in 2018 average cost-per-hire from US universities was $6,110. 

FAANG and other top tier employers are likely to spend much more than the average.

For each of the companies, if you look at publicly available websites for the campus recruiting centre for one of the HYPS schools for these companies, and just look at the roster of public facing “ambassadors”, who have significant skills and earning counterfactual (so fully burdened cost may be over 200K per head) it’s clear it’s a 7 figure budget for them once you include operations, physical offices, management and other oversight (which won’t appear on the PL per se).

1 mil is the low end.

I can’t immediately pull up a link here as I am on mobile.

Yeah personally speaking, I don't have very developed views on when to go with Spaghetti-wall vs RCT, so feel free to ignore the following which is more of a personal story. I'd guess there's a bunch of 'Giving Now vs Giving Later' content lying around that's much more relevant.

I think I used to be a lot more RCT because:

1. I was first motivated to take cost-effectiveness research seriously after hearing the Giving What We Can framing of "this data already exists, it's just that it's aimed at the health departments of LMICs rather than philanthropists" - that's some mad low-hanging fruit right there (OTOH I seem to remember a bunch of friends wrestling with whether to fund Animal Charity Evaluators or ACE's current best guesses - was existing cost-effectiveness research enough to go on yet?)
2. I was basically a student trying to change the world with a bunch of other students - surely the grown-ups mostly know what they're doing and I should only expect to have better heuristics if there's a ton of evidence behind them
3. My personality is very risk-averse

Over time, however:

1. I became more longtermist and there's no GiveWell for longtermism
2. We grew up, and basically the more I saw of the rest of the world the less faith I had in people generally being sensible and altruistic and having their **** together
3. I recognised how much of my aversion to Spaghetti-wall is a personality thing [edit: maybe writing my undergrad dissertation on risk aversion in ethics made me acknowledge this more fully :P]

| Out of interest, did you read the post as emotional? I was aiming for brevity and directness

Ah, that might be it. I was reading the demanding/requesting tone ("show us your numbers!", "could FTX and CEA please publish" and  "If this is too time-consuming...hire some staff" vs "Here's an idea/proposal") as emotional, but I can see how you were just going for brevity/directness, which I generally endorse (and have empathy for emotional FWIW, but generally don't feel like I should endorse as such).

I completely agree that it is far easier to suggest an analysis than to execute one! I personally won't have the capacity to do this in the next 12-18 months, but would be happy to give feedback on a proposal and/or the research as it develops if someone else is willing and able to take up the mantle. 

I do think that this analysis is more likely to be done (and in a high quality way) if it was either done by, commissioned by, or executed with significant buy-in from CEA and other key stakeholders involved in community building and running local groups. This is partly a case of helping source data etc, but also gives important incentives for someone to do this research. If I had lots of free time over the next 6 months, I would only take this on if I was fairly confident that the people in charge of making decisions would value this research. One model would be for someone to write up a short proposal for the analysis and take it to the decision makers; another would be for the decision-makers to commission it (my guess is that this demand-driven approach is more likely to result in a well-funded, high quality study). 

To be clear, I massively appreciate the work that many, many people (at CEA and many other orgs) do and have done on community building and professionalising the running of groups (sorry if the tone of my original comment was implicitly critical). I think such work is very likely very valuable. I also think the hits-based model is the correct one as we ramp up spending and that not all expenditure should be thoroughly evaluated. But in cases where it seems very likely that we'll keep doing the same type of activity for many years and spend comparatively large resources on it (e.g. support for groups), it makes sense to bake self-evaluation into the design of programmes, to help improve their design in the future.

Agreed! Note, however, that in the case of the FTX grants it will be pretty hard to do this analysis oneself without access to at the very least the list of funded projects, if not the full applications.

Well, you don’t have to be any more, because now it’s Jessica McCurdy’s reply.

Voting is biased toward comments that are easy to evaluate as correct/helpful/positive/valuable. With that in mind, I don't especially find this individual instance dismaying?

I agree with what you are saying that yes, we ideally should rank order all the possible ways to market EA and only take those that get the best (quality adjusted) EAs per $ spent, regardless of our value of EAs - that is, we should maximize return on investment.

**However, in practice, as we do not currently yet have enough EA marketing opportunities to saturate our billions of dollars in potential marketing budget, it would be an easier decision procedure to simply fund every opportunity that meets some target ROI threshold and revise that ROI threshold over time as we learn more about our opportunities and budget. ** We'd also ideally set ourselves to learn-by-doing when engaging in this outreach work.

One reason to be lower than this per recruited EA is that you might think that the people who need to be recruited are systematically less valuable on average than the people who don't need to be. Possibly not a huge adjustment in any case, but worth considering. 

assuming that there are 5000 EAs, Sam Bankman-Fried is donating $20 billion, and all other 1999 EAs

Should be 4999

assuming that there are 5000 EAs

I know this isn't your main point, but that's ~1/10 what I would have guessed. 5k is only 3x the people who attended EAG London this year.

Cool, I see where you got the figure from. But yeah, most of that work is passing to Open Philanthropy, so we don't plan to spend $50m/year.

That's fair - I'm not the earlier commenter but would suggest (as someone who's heard some of these conversations but isn't necessarily representative of others' thinking):

For dinners: Suppose offering to buy a $15 dinner for someone makes it 10% more likely than they'll go to a group dinner, and suppose that makes it 1% more likely that they'll have a very impactful career. Suppose that means counterfactually donating 10% of $100k for 40 years. Then on average the dinner costs $15 and yields $400.

For retreats: Suppose offering to subsidize a $4oo flight makes someone 40% more likely to go to a retreat and that this makes them 5% more likely to have a very impactful career. Again suppose that means counterfactually donating 10% of $100k for 40 years. Then on average the flight costs $400 and yields $8,000.

(And expected returns are 100x higher than that under bolder assumptions about how much impact people will have. Although they're negative if optics costs are high enough.)

Unsure but probably more than 20% if the person wouldn't be found through other means. I think it's reasonable to say there are 3 parties: CEA, the group organizers, and the person, and none is replaceable so they get 33% Shapley each. At 2% chance to get a career change this would be a cost of 750k per career which is still clearly good at top unis. The bigger issue is whether the career change is actually counterfactual because often it's just a speedup.

I think what I'm getting at is that burden of proof is generally an unhelpful framing, and an action that you could take that might be helpful is communicating your model that makes you sceptical of their spending.

Hiring consultancies to do this seems like it's not going to go well unless it's rethink priorities or they have lot of context and on the margin I think it's reasonable for CEA to say no, they have better things to do.

I feel confused about the following but I think that as someone that runs an EA org you could easily have reached out directly to CEA/FTX to ask this question (maybe you did, if so apologies) and this action seems kind of like outing them more than being curious. I'm not necessarily against this (in fact I think this is helpful in lots of ways) but many forum users seen to not like these kinds of adversarial actions.

Broken into a different comment so people can vote more clearly

In many ways, if the outcome is that there isn't a clear/shared/approved expected value rationale being used internally to guide a given set of spending, that seems to validate some of the concerns that were expressed at EAG.

I think that there is likely different epistemic standards between cause areas such that this is a pretty complicated question and people underpreciate how much of a challenge this is for the EA movement.

Oh, I read it as more the former too!

I read your post as:

1. Asking if FTX have done something as explicit as a BOTEC for each grant or if it's more a case of "this seems plausibly good" (where both use expected value as a heuristic)
2. If there are BOTECs, requesting they write them all up in a publicly shareable form
3. Implying that the larger the pot, the more certain you should be ("these things have a much higher than average chance of doing harm. Most mistaken grants will just fail. These grants carry reputational and epistemic risks to EA.")

I thought Sam's comments served as partial responses to each of these points. You seem to be essentially challenging FTX to be a lot more certain about the impact of their grants (tell us your reasoning so we can test your assumptions and help you be more sure you're doing the right thing, hire more staff like Open Phil so you can put a lot more work into these evaluations, reduce the risk of potential downsides because they're pretty bad) and Sam here essentially seems to be responding "I don't think we need to be that certain." I can't see where the expected value heuristic was ever called into question? Sorry if you thought that's how I was reading this.

[Edit: Maybe when you say "plausibly good" you mean "negative in expectation but a decent chance of being good", whereas I read it as "good in expectation but not as the result of an explicit BOTEC"? That might be where the confusion lies. If so, with my top-level comment I was trying to say "This is why FTX might be using heuristics that are even rougher than BOTECs and why they have a much smaller team than Open Phil and why they may not take the time to publish all their reasoning" rather than "This is why they might not be that bothered about expected value and instead are just funding things that might be good". Hope that makes sense.]