What's the best structure for optimal allocation of EA capital?

By Milan_Griffes @ 2019-06-04T17:00 (+10)

In 2017 (the last year for which we have complete data), Open Phil accounted for 78.7% of all EA grantmaking. 2018 & 2019 will probably follow a similar pattern, though the data isn't in yet.

(Derived from Rethink Priorities' recent analysis; here's a spreadsheet showing my work.)

So EA is currently in a regime wherein the large majority of capital flows from a single source, and capital allocation is set by a small number of decision-makers.

Rough estimate: if ~60% of Open Phil grantmaking decisioning is attributable to Holden, then 47.2% of all EA capital allocation, or $157.4M, was decided by one individual in 2017. 2018 & 2019 will probably have similar proportions.

It seems like EA entered into this regime largely due to historically contingent reasons (Cari & Dustin developing a close relationship with Holden, then outsourcing a lot of their philanthropic decision-making to him & the Open Phil staff).

It's not clear that this structure will lead to optimal capital allocation. It's straightforward to imagine alternative structures:

I'm not claiming that any of the above will necessarily lead to better capital allocation than the current structure, but it seems plausible that they might.

I'm not aware of much discussion of alternative capital structures in the EA community, so exploring this seems valuable.





Denise_Melchin @ 2019-06-05T13:04 (+31)

I find some of the statements in your post a bit jarring, and this is not the first time I feel like this when reading your writing. The founders of Good Ventures are multi-billionaires who have been influenced by EA ideas which stem to some extent from the EA community. This is excellent. But the EA community does not have ownership over this money. Your writing makes it sound like it does, which I find presumptuous and off-putting.

For the future, I would recommend that you should try to better understand the relationships between different individuals and institutions that are associated with the Effective Altruism community before asking questions like this one.

(writing in personal capacity here, not as a mod)

Milan_Griffes @ 2019-07-04T20:05 (+8)
The founders of Good Ventures are multi-billionaires who have been influenced by EA ideas which stem to some extent from the EA community. This is excellent. But the EA community does not have ownership over this money.

cf. this Quora answer (a) by Dustin Moskovitz:

I'm very fond of this quote from Louis C.K. (comedian) below and generally view the world through this lens:
"I never viewed money as being 'my money' I always saw it as 'The money' It's a resource. if it pools up around me then it needs to be flushed back out into the system."
In other words, Cari and I are stewards of this capital. It's pooled up around us right now, but it belongs to the world. We are not perfect in applying this attitude, but we try very hard.
All that said, it turns out to be quite difficult to flush such a large sum back into the world in a way you can feel confident about, which is why we started Good Ventures (http://www.goodventures.org/) and work so closely with GiveWell. But we're learning more and more every day and accelerating our pace as we do. We intend not to have much left when we die (i.e. we have a "burn down" foundation).

fwiw, a printout of that quote was posted on the wall at GiveWell when I worked there. I think it's a great sentiment & it made me happy to be regularly reminded of it.

Evan_Gaensbauer @ 2019-06-04T17:09 (+14)
Rough estimate: if ~60% of Open Phil grantmaking decisioning is attributable to Holden, then 47.2% of all EA capital allocation, or $157.4M, was decided by one individual in 2017. 2018 & 2019 will probably have similar proportions.
It seems like EA entered into this regime largely due to historically contingent reasons (Cari & Dustin developing a close relationship with Holden, then outsourcing a lot of their philanthropic decision-making to him & the Open Phil staff).

This estimate seems to have a lot of problems with it, from attributing so much credit to Holden that way not only seeming in practice treating him as too unique an actor without substantiation, if not in principle playing fast and loose with how causation works, while being a wild overestimate from nowhere. This is a pretty jarring turn to what up until that point I had been reading as a reasonable question post.

Milan_Griffes @ 2019-06-04T17:13 (+4)

Thanks. I think my main point is pretty orthogonal to this estimate. [edit: my main point is that we arrived at this structure for historically contingent reasons, and it's not clear that the present structure is the best one for optimal altruistic allocation of capital, so perhaps we should consider alternatives.]

---

e.g. if instead we attribute 30% of Open Phil decisioning to Holden, the numbers are then:

  • 23.6% of EA 2017 capital allocation is attributable to Holden, or ~$62.0M in 2017

I think attributing a sizable portion of Open Phil decision-making to Holden is appropriate, as I believe he has final sign-off on all major grants.

Evan_Gaensbauer @ 2019-06-04T17:26 (+1)

Right, but if all he is doing is signing off, then you're attributing to him only the final part of the decision, and treating that as if it's the whole decision.

Milan_Griffes @ 2019-06-04T17:28 (+3)

Sure. "How much of this is attributable to Holden specifically?" is mostly orthogonal to my main point, though I think the degree of centralization is interesting.

(I edited my above comment for clarity.)

Evan_Gaensbauer @ 2019-06-04T17:43 (+1)

Like I said in my above comment, asking interesting questions to avoid stating inconvenient if valuable opinions doesn't go far in EA. If you think so much centralization of decision-making in Open Phil in the person of Holden Karnofsky is suboptimal, and there are better alternatives, why not just say so?

Milan_Griffes @ 2019-06-04T17:49 (+8)

From the OP:

I'm not claiming that any of the above will necessarily lead to better capital allocation than the current structure, but it seems plausible that they might.

I don't know enough yet to say that an alternative structure would necessarily be better.

I intend to think about this more, and it seems good to do some of that thinking in public fora so that other people can contribute when they're interested + have comparative advantage.

Evan_Gaensbauer @ 2019-06-04T19:28 (+4)

Strongly upvoted. I don't have anything else to add right now then I now understand why you're asking this question as you have, and that I agree it makes sense as a first step with the background assumptions you're coming in with.

Ozzie Gooen @ 2019-06-04T23:37 (+10)

I think that there's a pretty compelling case for Cari & Dustin to give everything to Open Phil. The main reason is that Open Phil is already evaluating and funding the other options, and will continue to do so. They are funding many moonshots (AI risk), and are basically funding many smaller grants (through things like regranting programs with BERI and CEA).

If they were to give directly to moonshots or other projects, that would just mean moving the decision from Open Phil staff to Cari & Dustin individually, which seems weird to me. They could hire other people to do this, but at that point they are basically making a competitor to Open Phil.

Should they sponsor an Open Phil equivalent/competitor? My guess is that this would most make sense if this other organization were sufficiently different and/or specialized; though in these cases, I'd expect Open Phil would be happy with them and consider funding them directly; they would be less of an equivalent and more of a collaborating group. If they were too similar they would compete for talent; I think much of the best talent is at Open Phil, so this would be a struggle for them.

From what I understand, Holden really doesn't have complete power. If his influence were too high, that could be modified while keeping the main Open Phil relationship the same.

The general structure of "I'm rich, so I'll make one, and only one, group to decide, on a high level, what to do with my money" seems pretty reasonable and common to me.

Separately, I'd of course like to see other mega-billionaires also come in, but that's somewhat of a separate issue.

Milan_Griffes @ 2019-07-05T00:07 (+2)
Should they sponsor an Open Phil equivalent/competitor? My guess is that this would most make sense if this other organization were sufficiently different and/or specialized

The concrete proposal that currently seems most compelling is for either Open Phil or Good Ventures to commit to each year granting X amount to each of the EA Funds for regranting (size of X to be determined).

I think this would diversify grantmaker decision-making. I'm currently quite bullish on diversifying grantmaker decision-making.

I don't think it matters very much if Good Ventures does this directly or if it happens through Open Phil.

Ozzie Gooen @ 2019-07-05T21:57 (+2)

Right now there is quite a bit of this happening through Open Phil. My guess would be that they would want to seem more happen, but don't feel like the existing groups could efficiently take much more money from them at the moment.

This isn't to discourage other donors; extra donor diversity seems pretty useful as well.

Milan_Griffes @ 2019-07-07T15:28 (+4)
Right now there is quite a bit of this happening through Open Phil.

Open Phil makes grants to the EA Funds for regranting?

Ozzie Gooen @ 2019-07-08T18:05 (+5)

I've heard different things, it's been harder to pin this down.

"Approximately half of the money we’re receiving from Open Philanthropy, we expect to regrant to promising work in the community, either directly through EA Grants or via separate grants to local groups." https://www.centreforeffectivealtruism.org/blog/announcing-grant-from-the-open-philanthropy-project/

My impression is that this money is being used for EA Grants and Community Grants (I believe it's called), but not EA Funds. I don't fully understand the details here, but imagine they probably have decent reasons for this.

Carl Shulman has a $5m fund for regranting. https://www.openphilanthropy.org/giving/grants/centre-for-effective-altruism-new-discretionary-fund

I believe BERI also got money to regrant, but it's possible it's not from Open Phil. You could see their list of grants here: https://existence.org/grants/

Arepo @ 2019-06-05T18:58 (+9)

A less drastic option would be for OpenPhil to just hire more research staff. I think there's some argument for this given that they're apparently struggling to find ways to distribute their money:

1) a new researcher doesn't need to be as valuable as Holden to have positive EV against the counterfactual of the money sitting around waiting for Holden to find somewhere to donate it to in 5 years

2) the more researchers are hired, even (/especially) when they're ones who Holden doesn't agree with, the more they guard against the risk of any blind spots/particular passions etc of Holden's coming to dominate and causing missed opportunities, since ultimately as far as I can tell there aren't really other strong feedback mechanisms on the grants he ends up making than internal peer review.

(I wouldn't argue strongly for this, but I haven't seen a counterpoint to these arguments that I find compelling)

Peter_Hurford @ 2019-06-04T19:56 (+7)

As a typical EA nitpicker, I will avoid answering your substantive question and instead point out that your data may be missing some areas - e.g., independent EA donors (probably a few tens of millions per year) and Founders Pledge / Effective Giving (unknown, but potentially tens to hundreds of millions per year toward EA stuff). Still, I do expect OpenPhil to be a large percentage of EA capital allocation.

Milan_Griffes @ 2019-06-04T20:03 (+2)

Thanks for the nitpick :-)


... point out that your data may be missing some areas

I believe you mean "our data", given I derived this from a Rethink Priorities analysis.

Peter_Hurford @ 2019-06-04T20:05 (+8)

Haha yeah. My point was that the data I gave was not intended to represent the entirety of EA-related giving, some of which is much harder to track.

Milan_Griffes @ 2019-06-04T21:11 (+2)

Makes sense. Do you have a ballpark estimate of total EA-related giving for 2017? (even if it's a rough one?)

Peter_Hurford @ 2019-06-04T23:40 (+3)

It will vary wildly by what you count as "EA-related". Also, I think it will be more meaningful to break it down by cause area.

In Global Poverty and Economic Development, I think EA (e.g., GiveWell) is noticeable but small percentage of all plausibly EA-related activities in the space (e.g., GAVI does quite cost-effective things with billions of dollars).

In Animal Welfare, I think EA (e.g., OpenPhil + ACE + Animal EA Fund) is roughly ~1/3rd of total EA-related spending (though, again, varies a lot by what counts as "EA-related").

For the various x-risks / long-term stuff, I know the least about what the donors in the space are like, but OpenPhil feels significant. Also, Ben Delo joining the scene really complicates the estimate.

Evan_Gaensbauer @ 2019-06-04T17:10 (+2)

Why are you asking this question? I'm asking because it seems more like an academic exercise in what would be a better capital allocation structure if one were to be had, when in practice it doesn't seem like will get there.

Milan_Griffes @ 2019-06-04T17:14 (+2)

From the OP:

I'm not aware of much discussion of alternative capital structures in the EA community, so exploring this seems valuable.

Why do you think we won't get to a better structure, in practice?

Evan_Gaensbauer @ 2019-06-04T17:26 (+2)

Right, I guess I was asking why you're exploring it.

I don't think will get a better structure through the route you're going, which is just asking questions about Open Phil. I figure at the least one would try figuring out what structure they consider best, and then explaining why you think it's the case Good Ventures should switch to that structure.

Milan_Griffes @ 2019-06-04T17:33 (+2)

Pointing out that we arrived at where we are for contingent reasons & that there's plausibly unrealized upside in alternative structures seems like a good first step for figuring this out.

I'm not clear on whether you think pointing this out is just unimportant, or if you think it's actually unhelpful.

Evan_Gaensbauer @ 2019-06-04T17:42 (+2)

I think it's unimportant. I would hope everyone is already aware we've arrived where we're at for contingent reasons. I think it's more than plausible we could have an alternative structure for capital allocation than the one we have now. I think this first step should have been combined with the next couple steps to just be its own first step.

Michael Dickens took the opposite route and said Open Phil should prioritize wild animal welfare. I also remember last year there were lots of people just asking questions about whether the EA Funds should be managed differently, and nothing happened, and then I made a statement more than a question, and then the EA Funds changed a lot.

Milan_Griffes @ 2019-06-04T17:51 (+3)

Got it, thanks.


I think it's more than plausible we could have an alternative structure for capital allocation than the one we have now.

What alternative do you have in mind?

Evan_Gaensbauer @ 2019-06-04T19:27 (+2)

I think your suggestion of Good Ventures making more grants to the EA Funds would be a better alternative, though before that I'd like to be confident the kinks have been worked out of the EA Funds management system. I was speaking more generally, though, that all kinds of generic structures that merely decentralized grantmaking in EA more would be better. That it could almost be any structure that had that feature was my point. I'm aware there are reasons people might behave as though so much decision-making being concentrated in Open Phil is optimal. If you have knowledge there is a significant portion of the EA community who indeed sincerely believes the current structure for capital allocation being so concentrated as it is, please let me know. I would act on that, as I would see that as a ludicrous and dangerous notion for all of EA I wouldn't think even Open Phil or Good Ventures would condone.