Ben_West🔸

Member of Technical Staff @ METR

Bio

Non-EA interests include chess and TikTok (@benthamite). We are probably hiring: https://metr.org/hiring 

How others can help me

Feedback always appreciated; feel free to email/DM me or use this link if you prefer to be anonymous.

Posts
86

Sorted by New

Sequences
3

AI Pause Debate Week
EA Hiring
EA Retention

Comments
1020

Topic contributions
6

Thanks for writing this! I do feel like base rates/priors are something where many of us (including me) could have done better pre-ftx crash, and I am optimistic that we have learned from this and will do better the next time.

Thanks for sharing this Zach! I know it must have been a ton of work.

Thanks for writing this! One small comment:

EA organizations often underrate experience relative to “intelligence” and “value alignment”

I believed this and wanted EV to hire more outside experts. To support my case, I made a spreadsheet of all the major issues EV had run into that I was aware of and whether having non-EA experts helped. To my dismay, the result was pretty equivocal: there were certainly instances where non-EA experts outperformed EAs, but ~as many instances to the contrary.

I don't think EA is unique here; I have half a foot in the startup world and pg's recent Founder Mode post has ignited a bunch of discussion about how startup founders with ~0 experience often outperform the seasoned experts that they hire.

Unfortunately I don't have a good solution here - hiring/contracting good people is just actually a very hard problem. But at least from the issues I am aware of at EV I don't think the correct update was in favor of experience and away from value alignment.[1]

  1. ^

     If I had to come up with advice, I think it would be to note the scare quotes around "value alignment". Someone sincerely trying to do well at the thing they are hired for is very valuable; someone professing to care about the organization's mission but not actually doing anything is not very valuable. And sometimes people confuse the two. [This is a general comment, not specific to EV.]

Thanks for being willing to share criticism!

I’d be excited to see more celebration or appreciation for people’s work.

I would be excited about this and have wondered for a while if we should have EA awards. This Washington post article brought the idea to my mind again:

Civil servants who screwed up were dragged before Congress and into the news. Civil servants who did something great, no one said a word about. There was thus little incentive to do something great, and a lot of incentive to hide. The awards were meant to correct that problem. “There’s no culture of recognition in government,” said Max Stier, whom Heyman hired to run the Partnership. “We wanted to create a culture of recognition.”

Interesting, I had not thought about things this way before.

It seems uncontroversial that we should value assets which can be used over the long term more highly then those which can't, all else being equal, but I mostly see people modeling this by amortizing them with some constant discount rate. I am vaguely aware of that accountants instead classify expenses as opex vs. capex but honestly couldn't explain to you why that's better.

I guess you are saying that it is useful to do so because, heuristically, you should prefer to cut investment in opex before cutting investment in capex?

Thanks for doing this investigation, and also thanks to CAIS for making a website which lets people easily do this kind of investigation/replication!

Thanks! For what it's worth, the thing you are describing seems consistent with describing EAs as "teammates" (I also think that sports teams are successful ~entirely because of the work of their constituent team members) but I concede that the term is vague.

Thanks! See my response to Michael for some thoughts on diminishing returns.

10x increase in labor leading to 3x increase in impact feels surprising to me. At least in the regime of ~2xing supply I doubt returns diminish that quickly. But I haven't thought about this deeply and I agree that there is some rate of diminishing marginal returns which would make marginal growth net negative.

Load more