The one-on-one team at 80,000 Hours has changed quite a bit. We thought the EA community might be interested to hear our current plans, and we’d be keen for feedback.
This post is written by Niel and Michelle. The team also includes Jenna and Habiba. The team is an amalgamation of 80,000 Hours' previous advising and headhunting teams.
Over the next year, we plan to grow our team in order to be able to have more conversations with people coming through our website. A big worry we have with scaling this up is that there will be more people filling out a form on our website who don’t get the opportunity to speak with us. We’re taking a number of steps to try to make doing this as helpful for people as we can.
Since 80,000 Hours was founded, we’ve been doing one-on-one calls with people about their plans to have impact through their careers.
Our main focus is talking to people who have done quite a bit of reading and thinking about how to have an impact in the world, but who haven’t yet interacted much with the EA community and have some questions about how to apply our research to their career. From the tracking we’ve done over the last few years, these types of conversations seem to be an effective way of helping people increase the impact of their career.
Over the last few years we’ve experimented with various alternatives to using applications to determine who to speak with. For example, we tried proactively reaching out to people who might be particularly good fits for certain career paths. We also tried 'role-based' headhunting, which involved working with hiring managers to search for, approach, and speak with people who could be a good fit for key vacancies in the EA community. Based on the data we gathered, our best guess is that none of these alternative models were as effective for us as talking to people who applied via our website, though we are fairly uncertain.
One key reason the conversations we have with people coming through the website seem to be particularly useful is because people fill out an application form. That means we have quite a bit of information about the person before we chat, and therefore in the chat we can focus on the areas where we can be most helpful to them. It also means we talk to the people we have the most useful things to say to.
We’re a small team, so we’re keen to focus on our best guess version of the programme and then add more variants of the model when we have more capacity in the future and/or are hitting diminishing returns.
Since we’ve now gathered several years of data on the above model and have tried out a few alternatives without any seeming clearly better, we’re keen to scale this intervention up.
We would like to be speaking with 1,000 people per year in 2023, and we hope to have hired at least two more advisors to the team by the end of 2022.
To do this we want to speak with more people who are:
Newly coming across the ideas of effective altruism
Interested in longtermism
Excited to figure out how to use their career to help others as much as they can
For whom we’ll have useful things to say
We have a few ideas of how to find more people like this. We could make the option to speak with us much more prominent on the website, and are experimenting with doing this and monitoring the kinds of applications we receive. We will also be thinking about where to find the kinds of people who are most likely to find our content and conversations useful. (For example, we found people that we thought we could help via the Wait But Why article on career choice, so we’ll be thinking about other venues that might be relevantly similar.)
Although we know the rough contours of the model we’re growing, we’ll be continuing to test out ways of improving it. Some of this testing will involve trying out ways of improving the conversations themselves, which is often dependent on what individual team members find hard. For example, I (Michelle) am not very good at making small talk, which makes it feel hard to put people at ease at the start of a conversation. That’s crucial if someone is going to feel happy openly discussing their life plans with me. So I want to try out some different ways of improving at smoothly starting and getting into conversations. Our hope is that as different team members think through and test out improvements, we’ll build up a collection of ways to improve at having these sessions such that it’s much easier for a new team member to slot in and skill up in whatever they find hardest.
Another type of experiment we’ll be running is the processes around the calls. For example, over the last couple of years we’ve largely only had one conversation per person we were working with. That’s largely due to having a waiting list, and also from a sense that the first conversation with someone has often been the most useful for them. But it seems possible that knowing someone’s situation in more depth will allow us to be more useful, and therefore that returns don’t diminish much over the first couple of conversations. We'd like to do more experiments with varying the length of calls and the number of calls we do with each person. We hope to run one experiment of this type each month or two.
Worries we have
One of the biggest worries we have with our plans is that it’s annoying to fill out a form if you then don’t actually end up getting the opportunity to speak with us. Unfortunately, growing in the way we described above means decidedly more people being in that position. To mitigate this downside, we plan to take the three steps described below.
We want to try to make filling out the form as useful as possible for those who don’t end up getting the opportunity to speak with us. We’ve had feedback in the past that people find thinking about the questions we ask to be useful in clarifying their views on their own careers. We plan to continue to try to encourage people to fill out the form in a way that will be useful for their career thinking. We will also alter the form to fit in more closely with our career planning process and to make it easy for people to share their thinking with their friends and mentors to get advice.
We would like to increase the number of cases in which we introduce people to someone else who may be able to help them (in cases where we don’t speak to them ourselves).
We also want to be as clear as possible about what types of people we’re likely to be most helpful for. We plan on putting up a separate post on the EA Forum on this topic. While people who are already pretty connected to others in effective altruism aren’t our main target, in some instances we’ve been able to significantly help people in this category. We want it to be as easy as possible for people to get an idea of whether they’re likely to be someone we can be useful to.
A second significant worry we have with our plan is that by doubling down on a particular strategy we’re missing out on creating value through some of the other strategies we mentioned, such as headhunting. While we didn’t get evidence that these other projects were more effective than our current work, we still expect them to be valuable. In the longer run (perhaps as soon as a year from now) we hope to expand to cover a wider range of projects. But over the coming year we think it would be better to focus on doing one thing well.
Things we’d like to see more of
Given how significant career decisions are, it’s worthwhile to discuss them with many different people—ideally people with a variety of viewpoints. It therefore seems useful to have a number of groups and individuals providing career decision advice as well as ongoing career mentoring for people interested in effective altruism.
Because of the value of information, we're most excited about groups trying out different processes. (For example, the Student EA Network does a multi-session program.) We would also like to see more groups specialising in particular areas (such as Animal Advocacy Careers).
A third gap left by our type of conversations is talking to people already highly involved in the in-person EA community. There are many different things that could look like, such as helping people set out a career plan and checking in on it periodically with them or talking through specific professional challenges people face. There are a couple of people testing out specific types of such mentoring, for example Lynette Bye who provides productivity coaching, and Daniel Kestenholz doing personal growth coaching. A nice thing about this is the person doing it can test out their fit for providing this kind of coaching alongside another job (as Daniel is!), particularly if the two are complementary (say, project management coaching alongside working as a project manager). Michelle has written a few more thoughts about mentoring in EA in this post.
We're also really happy to see groups doing something pretty similar to us, since we think it’s sensible for people to talk to a range of people about their careers. For example, we’re very much looking forward to the launch of Probably Good.
In addition to people whose focus is on talking through career decisions, we’d be keen to see more EA networks develop along the lines of EAs in consulting, WANBAM, or the London directory. If you’re on a career path you think might be effective, a great way to increase your impact is to talk to others about how to think about whether it’s right for them, how impactful it might be, and how to get onto it. It can also be really fun to connect with other EAs with the purpose of helping them answer these kinds of questions.
We mentioned above that in the immediate future we don’t expect to be doing ‘role-first’ headhunting (i.e. working with hiring managers to search, approach, and speak with strong candidates for open jobs in the community), or actively reaching out to people who haven’t contacted us. We do think that these are potentially really valuable activities, and we hope to come back to them within the next few years. In the meantime, if you have a large network in a relevant area of EA, it could be worth considering doing headhunting within that area.
If you have ideas for projects along any of these lines, a first step might be to write them up in order to gauge interest and get feedback on the Forum. If you’re at the stage of wanting to try it out, the EA Infrastructure Fund might be a great place to get seed funding.
This is a rough sketch of our current plans. We’d be interested to hear people’s thoughts on them, and we’re happy to field questions!
Unfortunately we haven’t published these results publicly, though you can see some results from our 2019 evaluation. ↩︎
Note that our website has a wide reach, and so I could imagine other groups finding other interventions, such as city visits, more effective. ↩︎
Note that we found the 80,000 Hours’ brand and website to be particularly valuable when proactively reaching out to people to discuss their career plans and opportunities for impact, so this may be less effective for organisations without a strong existing brand and web presence. ↩︎
Meta: below is a very non-generous view of 80k one-on-one career advising (kinda bitter to be honest). I will probably be raising points that the 80k team thought about over the years and decided against for a good reason but I have not seen them publicly discussed. I will be very happy to be wrong about this.
To sum up: 80k one-on-one career advising has a small negative effect on the world
I have an impression that 80k accepted a long time ago that that wait time will just have to be pretty long. Here is a bunch of ideas to shorten the wait time that I don't think were attempted historically:
The advising page also says that
It is not clear to me that the price of having in-house career advisors is justified. I think there are a lot of people (like a hundred) in the community who could gladly volunteer a couple of hours per month to do career advising and would be super excited about the opportunity to help out and share their knowledge and connections with the newcomers.
I believe a structure that has a small experienced 80k career advisor team (2-4 people) managing a community of vetted experienced EA volunteers would be a much more promising way to go. Or alternatively have the community fully self-organise for this project.
I don’t quite agree here. I was counting ‘additional resources’ like guidance and introductions as ‘things to say’. So focusing on people for whom we have useful things to say should increase rather than decrease the extent to which we talk to people who need these resources to increase their impact.
I agree we’re not always good at figuring out which people could most benefit from our providing resources / introductions. We try to keep calibrating on this from our conversations. That’s clearly easier in the case of noticing people we talk to for whom we couldn’t be that useful than the opposite. To counter that asymmetry, we try to do experiments with tweaking which people we speak to in order to get a sense of how useful we can be to different groups.
With respect to your concrete examples:
The descriptions we’ve given of people on that page is actually from where they’re at a year or two after we speak to them. That’s because it takes a while for us to figure out if the conversation was actually useful to them. For example, I think Cullen wasn’t President of HL EA when we spoke to them.
That aside, on the question of whether we should generally speak to people with these types of profiles:
Being a PhD student in Machine Learning doesn’t seem like an indication of how much someone knows about / has interacted with the effective altruism community. So it doesn’t seem to me like it should count against us talking to them. (Though of course the person might in fact already be well connected to the EA community and not stand to benefit much from talking to us.)
It seems like a hard decision to me whether someone running an EA student group should count in favour of or against our speaking to them. On the one hand, they might well be steeped enough in effective altruism they won’t benefit that much from us recommending specific resources to them. They’re also in a better position to reach out to other EAs to ask for their advice than people new to the community would be. On the other hand, it’s a strong signal that they want to spend their energies improving the world as much as possible, and so our research will definitely be applicable for them. It’s also not a foregone conclusion that someone running a student group has had much opportunity to sound board their career with others who feel equally strongly about helping the world, let alone those with similar values but more experience. So I could imagine us being really useful for EA group leaders, despite the caveats above.
I'd like to second the opinion that it is a bit of a turn off that the resources go toward people who already have resources. I understand that "justice" and "equality of opportunity" aren't core EA concerns, and I also realize that giving an hour of time to a person at an elite university who has received lots of educational benefits in life very well may have a higher ROI than giving an hour of time to a "normal" person (I'm using normal here to indicate a person who grew up in a family with a more median income, and who went to a less outlier school).
Unfortunately, I don't have a solution for this. The current practice is very much in line with the career advice on 80,000 Hours, which seems to be primarily applicable to people who are able to get jobs at McKinsey, get into PhD Programs about Artificial Intelligence, and able to earn well-above a median income. Elitism isn't inherently a bad thing; it can sometimes simply be a way of having high standards.
For context, I'm writing this as a person who grew up in a lower-middle class family, who didn't live in a big city with lots of opportunities, who went to a university that is not famous, and who has never earned more than the average income. I'm privileged in lots of ways in my life, but because the paths that are highlighted on your website aren't realistic options (unless I were to spend large amounts of money on re-schooling), it sends a message of "if you aren't in this particular privileged class of people who have received lots of education at elite institutions, then you probably aren't the right fit for our club."
I am so sad that we are causing this. It is really tough to make yourself vulnerable to strangers and reach out for help, only to have your request rebuffed. That’s particularly hard when it feels like a judgement on someone’s worth, and more particularly on their ability to help others. And I think there are additional reasons for these rejections being particularly tough:
I do want to highlight that our not talking to someone isn’t a sign we don’t think they will have an (extremely) impactful career; rather it is simply a sign that we don't think we’ll be as helpful to them as we could be to some other people. So while I deeply empathise with the feelings I describe above and I expect I would feel the same way in a similar situation, I don’t think people are actually right to feel like they “are not good enough”.
I realise it’s probably no consolation, but, on a personal note, needing to turn down people who are asking for my help is unquestionably the worst part of my job. We spent a significant part of last year trying to find an alternative model we believe would be as impactful as our current process but wouldn’t involve soliciting and then rejecting so many applications. Unfortunately, we didn’t find one. I think it’s my responsibility to implement the model we think is best, but it’s hard to feel like I’m doing the right thing when I know I’m disappointing so many people. I often only get through reviewing applications by reminding myself of our mission and trying to bring to mind the huge numbers of people in the future who may never get to exist and are entirely voiceless, and for whose sake it is that I have to refuse to help people in front of me today that I care about.
Spitballing here, but have you considered putting some thoughts to this effect on your website? Currently, the relevant part of the 80k website reads as follows.
This is ok as far as it goes, but to me does feel a little like a fake-positive 'I'm sure you'll do just fine, whoever-you-are!'. Pointing out things like the fact that you have very little information to go on, and that you're optimising for people you can help most rather than making some kind of pure 'how valuable is this person' call, seems like it could help soften the blow the margin, though I appreciate it'll never make that large a difference given the other things you mentioned.
Thanks for this feedback. I had a go at rewriting that our 'why wasn't I accepted' FAQ. It now reads:
Why wasn’t I accepted?
We sincerely regret that we can’t advise everyone who applies. We read every application individually and are thankful that you took the time to apply. It’s really touching reading about people who have come across 80,000 Hours and are excited about using their careers to help others.
We aim to talk to the people we think we can help most. Our not speaking with you does not mean we think you won’t have a highly impactful career. Whether we can be helpful to you sometimes depends on contingent factors like whether one of our advisers happens to know of a role or introduction right now that might be a good fit for you. We also have far less information about you than you do, so we aren’t even necessarily making the right calls about who we can help most.
You’re very welcome to reapply, particularly if your situation changes. If you’re thinking of doing so, it might be worth reading our key ideas series and trying out our career planning process, which we developed to help people think through their career decisions. You can also get involved in our community to get help from other people trying to do good with their careers.
I agree this is too long, and I’m sad that it was actually longer than this at times. Right now I’m mostly managing to review them within a week, and almost always within 2 weeks. I wouldn’t want to promise to always be able to do this, but it’s much easier now we have a team of people working on advising.
I'm actually really keen to avoid us having long wait times. Career decisions are often pretty time sensitive due to application and decision deadlines. Thinking about your overall career also seems pretty aversive to me, so I think it's important to capitalise on people's enthusiasm and energy for doing those occur. Right now we're aiming to have slots available in the next couple of weeks after we've accepted an application, though it might take a few weeks before there are slots that work for a person, particularly if they're in a very different time zone than us.
Agreed on this prioritization, also I think both in principle (and as you noted) in practice, long wait times are highly avoidable. In principle you don't need that much resources to turn a particularly long wait time to one that is pleasantly short, certain exigencies (eg correlated staff vacation times) aside. The mathematics of queueing theory also might be helpful here.
Thanks for sharing your view. It’s useful for us to get an overall sense of whether others think our work is useful in order to sense check our views and continue figuring out whether this is the right thing for us to focus our time on. It's also important to hear detail about what the problems with it are so that we can try to address them. I’ll respond to your points in separate comments so that they’re easier to parse and engage with.
Thank you for your thoughtful replies, Michelle.
I can second feeling pretty heavy-hearted after my rejection, and really like the idea of vetting a crowd of volunteers. A similar idea would be to offer rejected people to share the info from their form, plus maybe their most important questions, with people who agreed to maybe take a look, e.g. via the EA Hub, where you could also filter relevant background. Or alternatively into a private group like „AI Safety Career Discussion“. I’m one of the shy people who would probably never do something like that themselves, but if it were an „official“ and recommended thing from 80,000Hours it would feel somehow much less scary.
Thanks for this feedback! It's really useful to know that this would make it easier to put yourself out there. We're in the process of changing the application form to connect better with our career planning process, to hopefully make filling it out a commitment mechanism for getting started on making a career plan (since doing so is often aversive). As part of that, we aim to send people a google doc of the relevant answers in a readily shareable format and encourage people to send it to friends and others whose judgement they trust.
I also find it pretty scary to email people out of the blue, even if I know them, particularly to ask them for something. But my hope is that if someone already has a doc they want comments on, and it's been explicitly suggested they send that to friends, it will make it a bit easier to ask for this kind of help. Increasing the extent to which people do that seems good to me, since my impression is that although people find it hard to reach out, most people would actually be happy to give their friends comments on something like this!
Great idea. What did you think about the idea to somehow streamline a process to share that Google Doc with others who might have something to say? A process that might require relatively little effort would be asking people in those forms "Would you be interested in receiving career plans from other people that are looking for feedback?". That might make it relatively effortless for people from a particular field, e.g. Cognitive Science in my case, to be matched to other people who might have valuable feedback.
It might be a bit effortful to match people, though I suppose you have information about the general field and that might already suffice? Or you might worry that people will receive unhelpful feedback and that this might reflect badly on you? Though I suppose you could emphasize that the people who you'd share the Google Doc are not vetted at all and are only fellow 80,000Hours fans who clicked on "I'd be down to look over other people's career plans".
Thanks! I really appreciate you sharing your thinking on this.
(And suspect it would be good if more orgs did more of this on the margin.)
Fitting into the described criteria, is there any chance that people would enjoy going over an almost complete career planning worksheet? It is the one found on the 80,000 hours website and is almost complete.
Your comment makes me think that something that might be really easy and really valuable is to just have an online space where people are encouraged to share docs outlining their career ideas, plans, and uncertainties (whether that be filled-in versions of the 80k career planning worksheet, or just sections of that, or filled-in versions of other templates, or something else), in order to get feedback and advice.
Obviously people can already take the initiative to do that in ad hoc ways, either in public spaces (like you've done) or via reaching out to specific people. And that's already useful. But I expect a lot of people who could benefit from this sort of thing would be too shy to do that, or just not think to do it, or be held back by not having enough of a network of relevant people.
Four options come to mind for how to do this:
Does anyone have thoughts on which of these options might be best? Maybe as a first step someone should just make an open thread and see how it goes?
One thing to consider is that a lot of these career docs may be sensitive, so it may be best not to share them publicly. But it might be easy to mitigate that issue, e.g. via sharing a link to a google doc that people have to request access to before seeing, or just saying the doc exists and asking people to send a message in order to be sent a copy/link.
 This also reminds me of Rob Wiblin's comments about how having a line manager can be surprisingly beneficial, and how it may be possible to capture a lot of those benefits by just having pairs of junior people who want more line management act as each other's line managers.
This sounds great to me! I'd be tempted to try out things in existing infrastructure first, like trying it out in the careers discussion fbook group, or the open thread you mentioned.
Other options that come to mind:
I've been surprised how much people's preferences on how to give comments on career plans differs. For example, I find it takes me ages to read through a plan so I end up putting it off for ages. Whereas I really like talking to people, so I'm much happier to chat to someone for half an hour. By contrast a friend of mine finds answering questions on the spot really stressful, so far prefers reading over things. So it seems worth giving people an option about whether to read through something (and if so how much) or whether to chat.
In the longer run, I think it would be cool to have a facebook group or slack for EA job seekers to keep each other motivated and accountable, because it's so hard to apply for jobs and deal with the uncertainty. That might also be a good place for people sharing and commenting on career plans.
That all makes sense/sounds good to me. I've now made an open thread on the Forum for this, as an experiment. Hopefully some of the other ideas can be tried out as well later.
These suggestions are all great. Would a discord server work? This is a community platform that is easy to download and maintain. There are individual chats, group forums, and voice channels for all means of communication. With enough support, this can be set up quickly. Please upvote if this is something that sounds useful, and depending on support, there will be a link posted on this post shortly. Keep in mind this discord server could be used for all things EA, besides, connecting individuals and providing an easy place to share documents and stories. Please provide feedback!
I'm afraid I don't really know anything about discord (me and tech are not the best of friends...), but from your description it sounds good! I think there is some EA activity on discord, so maybe you could build off that. I don't know anything about the form it takes or how to find it though unfo - but I'm guessing others on this forum do.
The link from Aaron on the Open Thread was helpful. Thanks for pointing the way!
In case you haven't seen it, I also replied to this comment when you made it on the Open Thread.