I wouldn't quite say that. In any sort of multi-period optimization you must model the consequences of actions and are (at least implicitly) predicting something about the future.
Regardless I was mostly gesturing at the intuition, I agree this doesn't solve the problem.
In my field (operations research), which is literally all about using optimization to make 'optimal' decisions, one way in which we account for issues like these is with robust optimization (RO).
In RO, you account for uncertainty by assuming unknown parameters (e.g. the weights between different possible objectives) lie within some predefined uncertainty set. You then maximize your objective over the worst case values of the uncertainty (ie, maximin optimization). In this way, you protect yourself against being wrong in a bounded worst-case way. Of course, this punts the problem to choosing a good uncertainty set, but I still think the heuristic "prefer to take actions which are good even under some mistaken assumptions" should be used more often.
I have been toying with the idea of starting a similar org (with more of a focus on OR) so excited to see this.
One suggestion I have (that I might be able to help with) is to offer a larger "free tier" that taps into the talent pool within academia. Three reasons why this is good:
I imagine most orgs have a long tail of data science projects which aren't important enough to go through the hassle of hiring a consultant, but that would still add some value. Meanwhile, students are in constant search of important real world problems to work on for their research or clubs (I was in Cornell Data Science) but generally don't have a good idea of what would actually be useful. Having a place where orgs can just write down such problems and students/academics can find them seems like it would potentially unlock a lot of value.
Based on feedback of pitching a similar idea at EAG, most of the value isn't actually in the object level work, but in identifying altruistic technical talent and getting them more engaged in high impact cause areas (and eventually into the hiring pipeline). Having lots of undergrads and PhD students working on EA style data problems seems like a good way of doing this.
Normalize X-risk and other more niche topics within academia.
I mostly expected people not to know what OR stood for, and then hover over the link to find out more. I also don't think "Operations Research" is actually much more elucidating, especially given how overloaded the word "operations" is within EA, but it seems this may have been mistaken given this feedback so I have updated.
Thank you for creating the tag! I was also a CS major at Cornell who actually only took the intro OR class (which was great but my impression is that undergrad ORIE at Cornell is not taught efficiently otherwise).
As for your suggestions
Emily Tucker has nice list of broadly useful resources for academic OR but this could probably be improved for an EA audience
This is a great idea!
The open-source solvers are way behind the commercial ones in terms of performance so I am skeptical this has high EV. AFAICT, much of Gurobi's licensing is customer specific, so I suspect non-profit orgs could get a steep discount if not a free license.
I think people with an OR background would be especially qualified for this type of role!
I wouldn't quite say that. In any sort of multi-period optimization you must model the consequences of actions and are (at least implicitly) predicting something about the future.
Regardless I was mostly gesturing at the intuition, I agree this doesn't solve the problem.
In my field (operations research), which is literally all about using optimization to make 'optimal' decisions, one way in which we account for issues like these is with robust optimization (RO).
In RO, you account for uncertainty by assuming unknown parameters (e.g. the weights between different possible objectives) lie within some predefined uncertainty set. You then maximize your objective over the worst case values of the uncertainty (ie, maximin optimization). In this way, you protect yourself against being wrong in a bounded worst-case way. Of course, this punts the problem to choosing a good uncertainty set, but I still think the heuristic "prefer to take actions which are good even under some mistaken assumptions" should be used more often.
I have been toying with the idea of starting a similar org (with more of a focus on OR) so excited to see this.
One suggestion I have (that I might be able to help with) is to offer a larger "free tier" that taps into the talent pool within academia. Three reasons why this is good:
Ironic the title of his column is "Inside View"
I mostly expected people not to know what OR stood for, and then hover over the link to find out more. I also don't think "Operations Research" is actually much more elucidating, especially given how overloaded the word "operations" is within EA, but it seems this may have been mistaken given this feedback so I have updated.
Thank you for creating the tag! I was also a CS major at Cornell who actually only took the intro OR class (which was great but my impression is that undergrad ORIE at Cornell is not taught efficiently otherwise).
As for your suggestions
Thanks for the feedback! I hope the revised title is an improvement.
Yes, definitely!
I am no expert but by far the biggest org is the UN's World Food Program.
I don't see much reporting on them from Givewell but they get 4/4 from charity navigator.