(Crossposted from LW.)
I worry a lot about the inefficiencies created by preventing competent researchers from allocating their time according to their own judgment of what is optimal. I think researchers can be several times more effective with financial freedom. There’s a steep cost paid by forcing researchers to compromise between optimising for what they care about and what their money cares about. Value of information is relative to the brain computing it, so competent people will usually best be able to estimate it on their own, even if the results are illegible to everybody else. They’ll find their path much faster if they can search through and abandon projects speedily, and don’t have to have their attention toted about by anyone or anything other than themselves—as if the harness isn’t going to chafe, as if anyone knows their own use better than they do.
But enough of my motivation. Do you feel like you are working on the most important thing you could work on? And if not, are you bottlenecked by being paid to do a different thing? Even if you are allocated optimally atm, how easy would it be for you to switch paths right now if that’s what you wanted? How costly would it be for you personally to just abandon your current project and hop to a new one if you thought it was a better use of your time?
I’m trying to get a sense of how much self-judged inefficiency is caused by a lack of financial slack among LW/EA-reading researchers. Feel free to answer here or admonymously (you don’t need to share your name, but if you do, I promise not to share it on).
Well, I work as a researcher in animal welfare, but I think that longtermist stuff is orders of magnitude more important, so if I was left to do whatever I want, I’d start looking into longtermism and try to find my place there. And I might quit my job and do that one day, but I’m not fully sure if researchers would have more impact if left to do what they want.
In terms of which cause to work on, OpenPhil thought more about which cause should receive how much money than I or probably any other researchers did. So I’m unsure if allocation between causes would be better if everyone did what they wanted. Maybe more people would work on interesting, obvious, or high-status problems. This applies not just to major cause-areas, but also to problems within causes.
In terms of what concrete projects to work on, I think that managers often know better what research would be more impactful because they are often more senior. And maybe researchers need less coordination and are more motivated if they work on what they want, but I think that the impact the project will have depends much more on the topic. And it’s great to be able to abandon projects when they no longer seem impactful, but needing to justify abandoning to your manager seems like a good safeguard against abandoning projects too much. And that is all I’d need to do to abandon my current project if I had a good reason to do that.
In practice, I sometimes was allowed to work on what I wanted, and sometimes I was given projects, and I haven’t noticed a clear correlation in which projects end up seeming more impactful in retrospect. I’m thinking about my two projects that seem most impactful per hour spent. In one case, I was passionately opposed to doing the project, argued against it in person, and wrote a longish document about why it’s a bad idea. I was told to do it anyway and I’m happy I did. In another case, everyone I talked to told me that the project I wanted to do was a very bad idea. I did it anyway during a free week we have at Rethink Priorities where we can do whatever we want, and later people who opposed the project agreed that it was a good idea. So ¯\_(ツ)_/¯
This is excellent, thank you. But if you could do whatever you wanted, and you think deferring to your manager/boss is optimal as a rule, wouldn’t you just generally do that?
Maybe it’d be systemically worse if everyone did whatever they wanted, because they wouldn’t be willing or able to coordinate to allocate themselves optimally. Instead, maybe they’d be selfishly prioritising what’s interesting/high-status rather than what’s important. Fine, maybe, idk. But I still strongly support EA tenure for the EAs you trust to be able to prioritise based on importance rather than shiny.
I don’t think that deferring to manager is always optimal, and I’d support EA tenure for some EAs too (I even suggested a mechanism of how that could work for less trusted researchers). Sorry that I didn’t make it clear in my comment, I just thought that you did a good job at presenting the pros of researchers doing what they want, so I wanted to give arguments for the other side to paint a fuller picture :)
Absolutely. I was hoping someone would provide good arguments against it too, since I know I was being overly rosy about the case in favour. Mainly this post wasn’t about convincing people that EA tenure is optimal, it’s mainly about eliciting anecdotes and perspectives.
I am not being as effective as I could be—AI x-risk is real enough that it’s the only thing I ought to be doing. Funding is not a serious obstacle in my life path.
I have skills and interest in ecology and believe it is an area I can have outsized impact by 1) updating philosophy across the field 2) improving longterm outcomes via gearing up highly effective projects and 3) leading by example. (I might be foolish) Sometimes it seems this is a crowded endeavor and sometimes it seems extremely neglected. In either case it makes me very happy so perhaps this is simply my selfish life path.
I would not be working on my current ecology-related job, but somehow researching and experimenting with ecological outcomes if I had infinite money and no reputational risk. However I am learning important things in my current role and researching other people’s ecological experiments in my free time so it’s close to optimal. I can pursue my altruistic goals more directly in a few years time when I have more experience, alliances, and specific concrete steps ready.
In my case, yes. I’m sure there’s other people doing more useful things than I am, but I think I’m doing the most useful thing I can be doing with my current set of skills/experience/connections.
It would probably be quite difficult to change paths if I thought that was no longer the case though.
Is there a true optimum? How would know its been reached?