In our 2017 and 2018 surveys of EA organisation leaders, the median respondent reported being willing to forego very large amounts of additional donations to hold on to their recent hires - several hundred thousand dollars for a junior employee, and millions for a senior one.
However, commenters on this forum and elsewhere have pointed to two observations that seem to be in tension with these figures:
- These organisations typically aren’t hiring a large number of people, opting instead for a more gradual approach (though it’s something like 30-50 over the last year, so more than it might seem).
- Candidates who on their face seem qualified are sometimes turned away after they apply.
How can this make sense?
One possibility is that the survey responses are wrong. We list weaknesses with the methodology in the article about the survey. But there are other possibilities.
Below we offer a potential explanation of what is going on. In the process we hope to clarify that the answers to the above survey question are far from a decisive consideration in favour of working at those organisations.
Most importantly, we asked the organisations to value typical recent hires. So this is a an ex post estimate of the value of recent hires, rather than an ex ante estimate of the value of future hires.
Future hires need to be searched for, selected, trained and come to be trusted. These are large additional costs, since they consume senior staff time. Even if you do everything right, there’s a good chance that new hires won’t work out. The organisations also have a limited number of management slots in which to add new staff. Read more in this great post by GiveWell.
These factors mean both that existing staff are very valuable, but the expected returns of hiring new staff may not be high - and in particular, may not beat the returns of other activities.
What does this mean for whether you should try to work at one of these organisations? If you’ve already got an offer (or could easily get one because they’re already familiar with your work) then working there is potentially very impactful. It also means that it’s valuable to find out if you could get one of these roles, so long as it’s not too expensive to learn. But it doesn’t necessarily imply very much at all about the likelihood of additional jobs opening up at these organisations, or the competitiveness of opportunities.
We propose that these jobs be seen as great roles to aim for, but with a high risk of not working out. If you think you might be a good fit, then look for cheap ways to test them out, but do this with a solid back-up plan, so you can easily switch to something else if it doesn’t work out. If you do get an offer, and you’re motivated to take it, then it’s likely a top option.
More broadly, we recommend using the process here to generate the best options for you. Ultimately we should all be aiming for the best thing we can personally do.
Below, we go into some more depth on the reasons for the above.
--
If an organisation’s previous hire is really worth hundreds of thousands of dollars, let alone millions, shouldn’t they be snapping up more staff members by the pound? Here are some possible reasons that they aren’t:
- The standard advice most new projects get is to ‘hire slowly’ and only when you’re confident it’s the right decision. That’s what 80,000 Hours was told during Y Combinator, and we have stuck to it. Hiring quickly can create many problems, such as greater difficulty coordinating a team, confusion about how the organisation works and what it’s trying to do, plus factorial growth in possible sources of interpersonal conflict.
- To get around these growing pains, organisations must invest in training, relationship building, and so on - but they can only do this so fast. The groups surveyed are all fairly small (under 30 people). If an organisation has set itself a maximum percentage growth rate, we can imagine it as having a limited number of ‘hiring slots’ each year. In that circumstance, i) a *better* hire within that given number of slots is very valuable, ii) a hire who has already made it past these barriers is even more valuable still, but iii) many reasonable applicants who aren’t one of the top few candidates will have to be turned away, simply because they can’t on-board so many people simultaneously.
- A single ‘bad hire’ who becomes a disgruntled employee, or ex-employee, can demoralise a team, and temporarily undo the work of several good hires. This rightly makes managers concerned about downside risk. Even in well-run projects, a substantial fraction of hires usually don’t stay for the long term, and end up being a net drain to a project once the costs of finding and managing them is accounted for. A degree of risk aversion can mean that even if the expected value of new hires is high, you might still want to hire slowly.
- “Trust bottlenecks” can be significant and take a long time to resolve. Many projects don’t trust new staff members to do high-stakes tasks without a substantial period of training and mentorship. Even someone who usually performs well could seriously screw up from time to time due to lack of experience. An organisation may highly value someone who knows enough that they trust them to ‘hit the ground running’ without much oversight, but place much less value on someone who can’t.
- Our question asked about retaining someone who is already working for you. The costs of finding those folks and convincing them to work with you have already been paid and can’t be recovered. But when an organisation looks forward, they know that hiring someone else will be a difficult and time-consuming process that will have to involve senior staff. While an organisation may place great financial value on good hires, it may place similarly great financial value on other things its managers alone can do - like making grants, writing up reports, meeting government officials, and so on. If they decide to hire, then it’s very valuable for them to get better applicants. But they might reasonably decide it’s not worth the opportunity cost of doing so.
- An organisation may not know what to do with extra donations, and so not value them very highly. This can blow out the number of donations they’d be willing to give up to retain someone, even if they don’t value that extra hire very much in absolute terms. (This is avoided if the survey respondent considers the option of regranting the money, but we expect they often neglected this possibility.)
- An applicant may in fact have the ability to do a good job, but not yet be able to demonstrate it. Organisations can only hire based on their perception of how likely someone is to work out.
- Someone could have the particular abilities you’re looking for, but be ruled out for some other reason - for example, a poor ability to resolve interpersonal conflicts.
- Some of the organisations surveyed take pretty unconventional moral or epistemic stances. This raises the importance of value alignment, because they can’t count on common-sense to keep someone in sync with the rest of the team. This can unfortunately force them to turn away people who look suitable on paper.
- Read more in this article by GiveWell, Why We Can’t (Simply) Buy Capacity.
The upshot of the above is that:
- An organisation reporting being ‘talent constrained’ doesn’t necessarily indicate that they are about to hire a large number of people.
- An organisation can be held back by hiring and still have a high bar to hire anyone.
- The fact that a given project places high value on their previous hire, doesn’t imply that most readers should reorient their careers around trying to work there. Indeed in one respect that information points in the other direction: the reason they value that person so much may in fact be that there are very few possible substitutes, because what they know or can do is especially rare. Rather people should aim for those positions if doing so might be their comparative advantage within the community.
We think our articles about the survey and talent gaps haven’t highlighted the above as much as they should.
The numbers in the survey are most relevant to someone who is offered a job at one of these projects, and is deciding whether to do that, or earn to give, or do direct work elsewhere. Added: Though even then, insofar as they are only part way through the discovery and training process, the numbers will still be inflated.
One way around the bottleneck described above is suggesting that our readers go work for larger established organisations which have the capacity to absorb many more people simultaneously, thanks to more existing staff or scaleable training processes. Examples of this could be government agencies, existing foundations, or some forms of earning to give. That is something we will be trying to do more of over the next year.
My hunch is (as implied elsewhere) 'talent-constraint' with 'talent' not further specified is apt to mislead. My impression for longtermist orgs (I understand from Peter and others this may apply less to orgs without this as the predominant focus) is there are two broad classes, which imperfectly line up with 'senior' versus 'junior'.
The 'senior' class probably does fit (commensensically understood) 'talent-constraint', in that orgs or the wider ecosystem want to take everyone who clears a given bar. Yet these bars are high even when conditioned on the already able cohort of (longtermist/)EAs. It might be things like 'ready to run a research group', 'can manage operations for an org' (cf. Tara's and Tanya's podcasts), 'subject matter expertise/ability/track record'.
One common feature is that these people add little further load on current (limited) management capacity, either because they are managing others or are already 'up to speed' to contribute themselves without extensive training or supervision. (Aside: I suspect this is a under-emphasised bonus of 'value-aligned operations staff' - their tacit knowledge of the community/mission/wider ecosystem may permit looser management than bringing on able professionals 'from outside'.) From the perspective of the archetypal 'pluripotent EA' a few years out from undergrad, these are skills which are hard to develop and harder to demonstrate.
More 'junior' roles are those where the criteria are broader (at least in terms of legible ones: 'what it takes' to be a good generalist researcher may be similarly rare to 'what it takes' to be a good technical AI safety researcher, but more can easily 'rule themselves out' of the latter than the former), where 'upskilling' is a major objective, or where there's expectation of extensive 'hands-on' management.
There might be similarly convex returns to getting a slightly better top candidate (e.g. 'excellent versus very good' might be 3x rather than 1.3x). Regardless, there will not be enough positions for all the talented candidates available: even if someone at an org decided to spend their time only managing and training junior staff (and haste considerations might lead them to spending more of their time doing work themselves than investing in the 'next generation'), they can't manage dozens at a time.
I think confusing these two broad classes is an easy way of burning a lot of good people (cf. Denise's remarks). If Alice the 23-year-old management consultant might reason on current messaging, "EA jobs are much better for the world than management consultancy, and they're after good people - I seem to fit the bill, so I should switch career into this". She might then forsake her promising early career for an unedifying and unsuccessful period as 'EA perennial applicant', ending up worse than she was at the start. EA has a vocational quality to it - key it does not become a siren song.
There seem a few ways to do this better, as alluded to in prior discussions here and elsewhere:
0) If I'm right, it'd be worth communicating the 'person spec' for cases where (common-sense) talent constraint applies, and where we really would absorb basically as many as we could get (e.g. "We want philosophers to contribute to GPR, and we're after people who either already have a publication record in this area, or have signals of 'superstar' ability even conditioned on philosophy academia. If this is you, please get in touch.").
1) Concurrently, it'd be worth publicising typical applicants:place or similar measures of competition for hiring rounds in more junior roles to allow applicants to be better calibrated/emphasise the importance of plan B. (e.g. "We have early-career roles for people thinking of working as GPR researchers, which serves the purpose of talent identification and development. We generally look for XYZ. Applications for this are extremely competitive (~12:1). Other good first steps for people who want to work in this field are these"). {MIRI's research fellows page does a lot of this well}.
2) It would be good for there to be further work addressed to avoiding 'EA underemployment', as I would guess growth in strong candidates for EA roles will outstrip intra-EA opportunities. Some possibilities:
2.1) There are some areas I'd want to add to the longtermist portfolio which might be broadened into useful niches for people with comparative advantage in them (macrohistory, productivity coaching and nearby versions, EA-relevant bits of psychology, etc.) I don't think these are 'easier' than the existing 'hot' areas, but they are hard in different ways, and so broaden opportunities.
2.2) Another option would be 'pre-caching human capital' into areas which are plausible candidates for becoming important as time goes on. I imagine something like international relations turning out to be crucial (or, contrariwise, relatively unimportant), but it seems better rather than waiting for this to be figured out for instead people to coordinate and invest themselves across the portfolio of plausible candidates. (Easier said than done from the first person perspective, as such a strategy potentially involves making an uncertain bet with many years of one's career, and if it turns out to be a bust ex post the good ex ante EV may not be complete consolation).
2.3) There seem a lot of stakeholders where it would be good for EAs to enter due to the second-order benefits even if their direct work is of limited direct relevance (e.g. having more EAs in tech companies looks promising to me, even if they aren't doing AI safety). (Again, not easy from the first person-perspective).
2.4) A lot of skills for more 'senior' roles can and have been attained outside of the EA community. Grad school is often a good idea for researchers, and professional/management aptitude is often a transferable skill. So some of the options above can be seen as a holding-pattern/bet hedging approach: they hopefully make one a stronger applicant for such roles, but in the meanwhile one is doing useful things (and also potentially earning to give, although I think this should be a minor consideration for longtermist EAs given the field is increasingly flush with cash).
If the framing is changed to something like, "These positions are very valuable, but very competitive - it is definitely worth you applying (as you in expectation increase the quality of the appointed candidate, and the returns of a slightly better candidate are very high), don't bet the farm (or quit the day job) on your application - and if you don't get in, here's things you could do to slant your career to have a bigger impact", I'd hope the burn risk falls dramatically: in many fields there are lots of competitive oversubscribed positions which don't impose huge costs to unsuccessful applicants.