Engineer by education, I discovered the EA world through the 80,000 Hours book in 2023 and went all in: Pledge, volunteering for EA France, where I manage the translation of the 80,000 Hours guide and organize an event with Peter Singer, and now by shifting my career from industrial engineer to founder of an effective giving initiative.
Romain Barbešø
Karma: 304
1. Money Spent: Basically nothing. We just spent ā¬200 trying to boost visibility on LinkedIn offers and the platform Jobs that make sense. For LinkedIn, the quality was too low, and for the second one, it didnāt lead to significantly more applicants.
2. Data Outreach/āreferrals: We donāt easily have this data, and itās difficult to attribute someone who saw the offers on multiple channels and applied a few days after. I think our LinkedIn posts (not the LinkedIn job offer) did the majority of our outreach in terms of quantity.
3. Focus on one channel: At our stage, our priority was to find the really best people and we were ready to invest the time. Even though there is quality variation between channels and between sources, you can still find your superstar on any channel. I think I will keep all the channels, and I will just reconsider LinkedIn job offers depending on whether I find a linear way for us to evaluate the first step.
4. Reuse our pool: The majority of the people who made it to the interview didnāt pass the bar to be hired at Mieux Donner, meaning that even if there was no one else applying, we would probably not have hired them. Therefore, there are only a few people from the finalists that we might re-interview. Maybe I can send an email saying that if I re-open a job, I would send an email to the top 5% saying that if the situation has significantly evolved and they have developed new skills in the past year, I would encourage them to apply again (inspired by the āShould I reapplyā from CE).
5. Value back: I think if two out of the three people stay one year, it will have been a great success.
6. Simiar role: Itās difficult to say. Some have the generalist background, some were in the sector of the role (not the exact role). Almost all of the top candidates were actively looking for a job. We still had plenty of people who were employed when they applied.
From job postĀing to hire: temĀplates, sourcĀing camĀpaign, and LLM-reĀsisĀtant tasks
Some tryouts in different languages: low to no effect?
Most effective altruism content is in English, so I wondered whether asking the questions in French or Spanish would change the results. I wasnāt exhaustive: I just tried ChatGPT and Gemini without history, and the answers didnāt seem to differ much from the English versions. I was also surprised by how much the answers varied between attempts, with some being noticeably more EA-leaning than others. If anyone wants to run a larger experiment, testing the effect of language could be worthwhile.
RoĀmain Barbeās Quick takes
Hello Richard, thanks for raising this point.
I was thinking that this process changes little for applicants, and weāve actually tried to design the process with their experience in mind.
First, a general note on the process itself (happy to have feedback on this too): the full process takes around 6 hours across 4 steps: a written application (30 min), practical exercises (1h30), a coworking session (1h30), and a final interview (1h). Only 15% of applicants pass step 2, so most people invest no more than 2 hours before getting a clear answer. Our rejection emails include specific feedback on each exercise, which we hope makes the time feel less wasted regardless of outcome. We also tried to design the exercises to be intellectually engaging, connected to real effective giving work, and useful for candidates to assess their own fit for this kind of role.
On the specific concern about opening multiple roles: from an applicantās perspective, we donāt think this changes much, and having four broad roles rather than two narrow ones may offer more people the opportunity to find a role that fits them, and could reduce exclusion based on specific backgrounds.
When we open any hiring process, even for a single role, there are two reasons we reject someone: either no one passed the bar and no one was hired, or someone else was judged better suited to our mission.
The role a candidate applies for should not change how they are judged. We try to run the process this way. If you are not hired because someone outperformed you, it shouldnāt matter whether that person applied for the exact same role or a different one. And in our case, the role titles are not even fully fixed: they will be adapted to the person we hire. Someone better than you might end up in a role that is a blend of two or three of the ones we listed.
For candidates who reach the final round and are ultimately not selected, I will tell them specifically where they underperformed: on the task exercises, on conscientiousness, on value alignment, or on the interview, depending on what the key determining factors were.
The asymmetry of effort in hiring is real, but we donāt think opening four roles makes it meaningfully worse. Happy to push back!
Mieux DonĀner is hiring: why we opened 4 roles to fill 2 positions
Thanks for the post! Even though your career is probably your highest leverage for impact, Iād add that these positions are well paid (or even extremely well paid compared to standard non-profit jobs), which allows you to donate significantly to effective charities. Even if youāre assigned to lower-impact units in the first few years, you can still have a substantial impact through your donations. Additionally, and worth noting, this career path is also accessible to people in well-paid careers who are not ready to take a significant salary cut.
Thank you for taking the time to write this up. This kind of post is incredibly valuable, and people should feel encouraged to share their operational experiences. Wishing both Hive and Sentient Futures to thrive!
Hello Lorenzo!
The counterfactual impact of a pledge follows GWWCās recommendations for pledge partners; we havenāt evaluated it ourselves, but yes itās closer to 0.2.
The ā0.64ā figure doesnāt refer to pledges but to donations. Itās much higher than GWWCās, but lower than Ayuda Efectivaās (in Spain), which can be explained by the larger number of actors in the English-speaking ecosystem compared to the French or Spanish ones.
For donations, you could say:
ā50% /ā 64% of the money donated was the result of Mieux Donnerās active outreach and influence.ā
StrateĀgies for proĀmotĀing givĀing pledges
ReflectĀing on One Year of Mieux DonĀner: LaunchĀing the French EffecĀtive GivĀing Initiative
Has this concept been attempted in the past, especially considering that it could be a for-profit venture (e.g., a company taking a 5% commission)? If it has been tried and failed, what were the reasons for its failure? If it hasnāt been attempted, what are the barriers preventing it from being pursued?
I find this project promising and Iām quite enthusiastic about it, but a part of me wonders why it hasnāt been done already. We donāt need to specifically promote effective charities for this project to succeed, what are the reasons this hasnāt been done successfully in the past?
Youāre shifting your resources, but should you change your branding?
Focusing on new articles and research about AGI is one thing, but choosing to brand yourselves as an AI-focused career organisation is another.
Personal story (causal thinking): I first discovered the EA principles while researching how to do good in my career, where, aside from 80k, all the well-ranked websites were non-impact focused. If the website had been specifically about AI or existential risk careers, Iām quite sure I wouldāve skipped it and spent years not discovering EA principles. But by discovering those principles and diving deeper into the content, I eventually saw existential risk as a top priority. Last year, the biggest chunk of my donations went to AI. I also managed the translation of your guide into French, and now, through Mieux Donner (the French effective giving initiative I co-founded), weāll likely raise donations to fund several AI positions.Trade-off (statistical thinking): How many people might be deterred from engaging with EA and this AGI topic because of AI branding? How many people are not working in AI because your homepage, About Us and menu bar mention other cause areas? (Especially considering your next career guide will still be multi-cause, and the information shouldnāt have time to become outdated given the AGI timeline you mentioned.)
Your focus seems well thought out, but my guess regarding the branding is that you shouldnāt change it.
By shifting to a narrow AI focus, you risk reducing by 13.5% the source of effective do-gooders (including donors!) is one negative consequence. However, I can also think of other potential downsides:
Damage to EAās reputation: This could feed into TESCREAL critics and prevent presenting EA principles to many potential supporters.
Potential lost opportunities: Even for your own organisation, focusing solely on AGI could cause you to lose out on backlinks, partnerships, and references that bring a steady flow of people.
As I read through your post, Iām still uncertain about what you plan with your branding. However, staying as a nonprofit that helps people use their careers to solve the worldās most pressing problems, while focusing the majority of resources on AGI but maintaining low-hanging fruits in other areas, seems to me to have a more positive impact than shifting your branding entirely.
So please, donāt mess up the communicationāit could have a net-negative effect on all the cause areas.
1. Money Spent: As I said, I was the only one involved in the process and I have a low salary (and a high number of hours), so it cost the charity ā¬1,500. But the value of my time is way higher (5ā10x).
3. What makes the break at interview: Itās value alignment, real motivation toward effective generosity principles, and a mindset oriented toward results that make a difference. If you request access to the questions, you will have a better sense.
4. Open vs closed: My intuition is that finding the real best person has tremendous value, and when these people work even for one year, they can get you so much more done. My organisation was constrained in terms of ideas and my network pool was very limited. Right now the limiting factor is my personal time, but actually much more so the generation of new ideas and ways to fundraise more money. Itās that I canāt easily spend money to buy myself more time, so I need to hire people.
Disclaimer: rough BOTEC coming. In my case, I think the value created by the people I hire versus the best people from the direct referral pool (I donāt have the total number of applicants, but Iām thinking of the people who reached the 3rd round) will be twice as impactful. Letās say very conservatively that the value of hired peopleās time is 1ā4 of my time (therefore the other pool would be 1ā8 of mine). Two people for one year = 4,000h. The open process took me 160h (120h more than a closed one). So the cost of 120h would generate 500h without additional costs.
I think in plenty of other contexts, depending on the questions I put at the beginning, it will not be valuable to do an open round. Plus, as I said, for Mieux Donner it also has communication value. At foundation events, different people heard about Mieux Donner and effective giving through the communication of the process.
For example, if next year I want someone capable of doing detailed research on niche effective altruism subjects, I will not do an open round. It wonāt be worth my time, because the people who can do the job well are probably already in my network.