Predicting Open Phil Grants

By Charles Dillon 🔸 @ 2021-07-23T14:00 (+57)

Edit: to view some analysis of the predictions from this question series, see my follow up post here

Open Philanthropy (OP) is a the single largest donor in the EA space (with the possible exception of the Gates Foundation's global health interventions, though these are not generally considered part of EA) and so how much money they grant, and to what causes, have major implications for what the EA and longtermist landscape looks like. It has a bearing on many issues, including:

Whether an individual should pursue direct work, or earn to give
If one is earning to give, whether to give now vs later
Whether there will be sufficient funding in a specific area for individuals to aim to have a career in that area
Influencing the Overton window of what is considered an EA cause

In recent years, OP has been granting about $280m per year ($298m in 2019, and $271m in 2020). This is just over 1% of the net worth of it's main funders Dustin Moskovitz and Cari Tuna, according to Forbes (note that this number has gone up significantly in the last year, and it was likely closer to 2% in 2019 when those grants happened). OP note on their "who we are" page that Moskovitz and Tuna incubated them with the intention of giving away their fortunes during their lifetimes, so having an estimate of when this spending will ramp up and where the donations will be focused seems valuable to me.

I have created a series of questions on Metaculus attempting to forecast various facets of this topic. The questions are:

They currently have relatively low numbers of predictions, so part of my motivation for posting them is to solicit more predictions and sources of information to make better estimates so that these predictions can be as good (and therefore useful) as possible.

This question series is a project of Rethink Priorities.

It was written by Charles Dillon, a volunteer for Rethink Priorities. Thanks to Peter Wildeford for feedback on this post, and Peter, Michael Aird and Linch Zhang for discussions which led to this project. If you like our work, please consider subscribing to our newsletter. You can see all our work to date here.

WilliamKiely @ 2021-07-23T18:40 (+23)

Open Phil obviously has more information on this than members of the general public. Do we know whether Open Phil is willing to share any of its internal forecasts on this (or internal knowledge that could be used to create such forecasts) publicly?

Davidmanheim @ 2021-07-29T16:19 (+5)

If so, several people at Open Phil are on Metaculus, and can write comments - but I'm skeptical that they will, or that they have very concrete ideas about what they will be doing - I think the plan is essentially to fund everything that seems very promising, and keep looking for more very promising things.

WilliamKiely @ 2021-07-23T18:32 (+8)

This is great. Small point of feedback on the title: I'd prefer: "Forecasting Open Phil Grant Amounts". ("Donations" doesn't seem like the right word to me for large philanthropic grantmaking.)

Charles_Dillon @ 2021-07-23T23:02 (+2)

Agreed, and changed, though I preferred "grants" to "grant amounts"

Nathan Young @ 2021-07-25T11:11 (+5)

I would like a similar set of questions around GiveWell recommendations. You could ask for 10 charities they said they might look into, the forecast their final value for money score. Then, you could resolve based on what they found in any they looked into.

- No risk to them.
- Might help them search
- Only requires names of charities they are researching

WilliamKiely @ 2021-07-23T18:36 (+2)

I really like your questions! If you're open to creating more, I'd suggest creating another (seven) question(s) on the average yearly amount of money granted from 2025-2030 (overall and for each of the six focus areas), since the yearly variance is significant.

Charles_Dillon @ 2021-07-23T23:00 (+3)

This is a good idea - I considered making the original questions averages for this reason, but erred on this side of making the question simpler. As is, I think the variance around the underlying distribution outcomes is large enough to compensate for the variance in year to year grants, such that I would not expect a big difference between 2028-2032 average predictions and 2030 predictions, and I'm hesitant to ask too many questions until the current ones have received sufficient attention.

Davidmanheim @ 2021-07-29T16:25 (+9)

I strongly dislike the idea of having tons more predictions - it increases effort both for people looking up forecasts, and for forecasters.

Forecasting effort is a commons, in the economic sense - forecaster effort is limited, and overuse of forecasts is effectively diluting the available forecaster resources among more questions.

WilliamKiely @ 2021-07-24T00:56 (+2)

That's a good point I didn't consider. I think you're right that the average question would not be very helpful, and not helpful enough to be worth adding.