Yep, I agree with this! (Perhaps I should have included this caveat.) 

Though I think often the junior ops person might be better off trying to take a month or two off to skill up more seriously, then begin applying for engineer jobs, or something? (Though obviously this is hard to do.)

Re section 4: should someone be printing off a bunch of AI safety papers and archiving them somewhere safe? (Probably a dumb idea.)

My understanding is that Coefficient remains excited about recommending us to donors, but recently confirmed with Good Ventures that we're not a good fit for Good Ventures' specific preferences at the moment. 

I'm afraid that I can't speak to Good Ventures' reasons apart from noting that they evidently didn't change Coefficient's decision to recommend us to their other donors.

Thanks for writing this up Lizka! I agree with most of it, including most of the ways I'd like us to be different.

Flagging some disagreements:

 - I feel like "smooth working together on projects" is pretty good on average: I agree with the example you give as a mostly-failed collab, though I think it was ultimately OK in that it resulted in your ITN post. But I think that there have been a lot of productive co-authored papers (I think most of our pieces are co-authored, and most of those collabs gave gone well.)

 - I also feel like "give each other lots of feedback" is generally pretty good: weekly pretty detailed feedback between managers, weekly commenting on work-in-progress drafts, 360 feedback every retreat. I think we should add in a monthly 360 though, I've been meaning to do that for a while. In fact, I'd maybe emphasize that if people don't want a fair amount of feedback, it might not be a great fit.

+1 to your other criticisms though, and I hope that this hiring round can help us bring in even more perspectives!

Other things I'd emphasize:

Impact focus: We have a culture of discussing research prioritization, and we all care about our work having impact.  I think this is contestable and could be bad for some people:  if you think you do your best work in a purely curiosity-driven way, Forethought might not be ideal.

Also, Stefan Torges recently joined to help make sure that research papers lead to action, e.g. by engaging with AI companies to make sure they introduce mitigations, and developing policy ideas. I hope that this makes it easier for researchers to focus on research, while also making sure that our research leads to some real difference. (To be clear, we're early-days figuring this out, and we might not be able to find a way to get this to work.)

+1 to just really enjoying working with this set of people, too!

Also, Fin and Mia just put out a podcast about Mia's take on what it's like to work at Forethought, in case people want another personal perspective!

Thanks for writing the post and this comment, Lizka!
 

~deferring to a fairly narrow cluster of AI worldviews/paradigms (maybe roughly in the direction of what Joe Carlsmith/Buck/Ryan have written about)

I agree that most of Forethought (apart from you!) have views that are somewhat similar to Joe/Buck/Ryan's, but I think that's mostly not via deferral?

+1 to wanting people who can explore other perspectives, like Gradual Disempowerment, coalitional agency, AI personas, etc. And the stuff that you've been exploring!

I also agree that there's some default more welfarist / consequentialist frame, though I think often we don't actually endorse this on reflection. Also agree that there's some shared thinking styles, though I think there's a bit more diversity in training (we have people who majored in history, CS, have done empirical ML work, etc).

Also maybe general note, that on many of the axes you're describing you are adding some of the diversity that you want, so Forethought-as-a-whole is a bit more diverse on these axes than Forethought-minus-Lizka.

We meant the Open Philanthropy one: apparently it's been merged into their GCR Cause Prio research team, but it was where Joe Carlsmith, Tom Davidson, Lukas Finnveden, and others wrote a bunch of foundational reports on AI timelines etc.

Also, a tougher question: how over-determined was this hiring?

[Just speaking for myself based on being a member of the hiring committee, without running this take past anyone else.]

I do think that Zach was in our top 5-10 most promising people at the start of the process. So I think that directionally the update is that we spent too much time/energy on this process, since the outcome wasn't that surprising.

However, I'm not sure if we should have spent that much less time/energy:

• In general I think that this is a really crucial hire, and finding someone marginally better or avoiding making a hiring mistake is really valuable, and worth significant delays.
• Some of our other top candidates were unknown to the hiring committee at the point where we started the process. So I think that there's a nearby-ish world where the broader/more-in-depth search led to a different outcome.
• I think that the more in-depth process can help the board, staff, and community to be more confident in the decision, and that's useful for the new CEO's future in the role. If we had appointed Zach with no process at all then in some sense that would be the same outcome, but I think it would leave Zach in a weaker position to lead CEA.
• Even if we'd appointed Zach sooner, I think that he might have only been able to start in mid-February anyway, because of his commitment to EV. Making the appointment sooner would still have been valuable in that it would have resolved some uncertainty sooner, but not as valuable as if Zach could have started several months sooner.
• I think that some aspects of the process could have gone more quickly, as I noted in my last post on this topic. But there were some important aspects that the hiring committee couldn't have altered much, and some things that mean that the actual hiring process was shorter than it seems (e.g. it took us 3 weeks or so after Zach said yes to get this post together, partly because CEA staff were on team retreat).
• I don't want to overupdate on one datapoint.
• Outside view, I think that this is a fairly standard length of time for an exec search process.

So yeah, overall I think you're right that we spent too much time on this, and I'm still confused how much we should have compressed the process.

I’m delighted that Zach has agreed to join CEA, and I’m excited for CEA’s future under his leadership.

I think that Zach is an extremely strong leader and manager, who thrives under pressure and cares deeply about building a better world. We dug deep into his strengths and weaknesses, through strategy discussions, work-history interviews, and reference calls. He has many outstanding references from people who have worked closely with him at Open Philanthropy and Effective Ventures US.

Thank you to everyone on the search committee, advisors to the search committee, staff involved in the process (especially Caitlin Elizondo and Oscar Howie) and of course to our candidates for engaging with a broad and intensive search process. The depth and detail of the search makes me confident that Zach is the right person to lead CEA going forward.

I’m also incredibly grateful to Ben West, CEA’s leadership team, and all the staff for leading CEA to one of its best years ever in 2023. It’s been an honour and a joy to work with them.

I will be available to support and advise Zach and the CEA leadership team as needed, but after 7 years at CEA I’ll now be taking a break (with a new baby!) before exploring my next career steps. 

I look forward to seeing Zach, CEA, and EA flourish in the coming years!

Thank you for all of your hard work over many years, Will. I've really valued your ability to slice through strategic movement-buliding questions, your care and clear communication, your positivity, and your ability to simply inspire massive projects off the ground. I think you've done a lot of good. I'm excited for you to look after yourself, reflect on what's next, and keep working towards a better world.

Makes sense! I was thinking of Effectiv Spenden, but I see that that's an ambiguous example. Another public attendee who is doing on-the-ground community building is Kuhan Jeyapragasan (and I think that there were 1-2 others who were invited but can't make it, or aren't public).