March 2017 Open Thread

This post aims is to give blog readers and followers of the Open Philanthropy Project an opportunity to publicly raise comments or questions about the Open Philanthropy Project or related topics (in the comments section below). As always, you’re also welcome to email us at if there’s feedback or questions you’d prefer to discuss privately. We’ll try to respond promptly to questions or comments.

You can see our previous open thread here.


I don’t particularly feel that OpenAI is uniquely positioned for this. Paul can collaborate with people working anywhere. I think OpenAI is uniquely positioned for work where it is helpful to be densely surrounded by top deep learning researchers and infrastructure.

Can you comment on the criteria for grant pages to have an explicitly listed grant investigator? I notice that the recently published animal welfare and criminal justice reform grants have a grant investigator, but the AI risk grants don’t. Is the criterion cause area-based?

Hi Vipul, we are now planning to put the grant investigator on grant pages as a general rule, but we are only implementing this going forward (not retroactively). Pages that were already partway through the publication process when the change was made may not reflect the change.


I’m pleased to find this post! I have recently been quite interested in how Animal Charity Evaluators recommends charities, allocates funds, has constructed the animal movement more broadly, and the way that necessarily influences recommendations.

My question would be the extent to which the Open Philanthropy Project accounts for this issue. So how are different perspectives taken into account when donating in a social movement space, and what are the checks and balances that OPP employ to ensure these issues are reasonably accounted for.

I’m hoping that doesn’t sound too vague, i’m happy to add more detail,

With regards,


Hi Kevin, I don’t think I understand your question - could you clarify?

Hi Holden,

I can add more detail. So my question is centred on how we might recognise and address bias.

If for example, there were forty groups (twenty with a similar outlook and twenty with different outlooks) and we allocated funding to the twenty groups all with a similar outlook, what would be the impact on the twenty that we didn’t fund?

It wouldn’t be neutral, so how is this factored in, and how are these other groups accounted for in terms of how Open Phil decides to allocate funding?

I think in terms of how we understand how the animal movement works, we tend to support different systems of thinking. In my personal view, and from what I have seen, Open Phil funding is largely centred on groups with a fairly mainstream and narrow ideological outlook, which seems to me that on its own, it might not be a very effective way of allocating funds. A different or additional form could be to take account of the movement as a whole and allocate funding in terms of how it could improve functioning. This approach could be added to the current one employed by Open Phil, essentially a form of hedging potential negative impact.

That might help explain my point, so I would like to know the steps Open Phil currently take to guard against generating bias through its system of giving. Another way to frame the question might be to ask how potential concerns raised in “hits-based mentality vs. arrogance” are accounted for in the work Open Phil does in the animal movement?

With regards,


Hi Kevin,

Thanks for the clarification. 

We’re trying to fund the farm animal welfare work we find most promising and believe will have the best per-dollar impact. It isn’t clear to me that this has more negative than positive effects on other parts of the field. And even if it is having major negative effects, this could well be good for the cause overall. It doesn’t seem healthy to me for funders to support work they think is not promising; I think a norm of doing this would create some substantial problems.

I believe our work in this area is following the principles outlined in the piece you mention: communicating uncertainty, trying hard to be well-informed, respecting those we interact with, and avoiding deception, coercion, and other behavior that violates common-sense ethics.

Hi Holden,

Thank you for the response. In the sense that it suits the Open Philanthropy Project organisationally, then it would be less worthwhile to consider the various issues, but it partly depends on the extent to which principles and values are going to be applied. So I don’t think it is enough to believe certain things happen, I think it is reasonably more important to know that certain things happen. I think in this regard we can also factor in uncertainty, however, i think we need to make a reasonable effort to diminish uncertainty, and so more thoroughly address various issues including potential biases. For instance, the types of things that you are excited about in relation to animal welfare, are not the types of things i am excited about. So what is the difference? Is it subsequently my lack of knowledge or understanding? (It could be) Or is it something else?

For example, I reasonably believe that Open Phil is a donor of last resort for GFI, so I would suggest that people fund other groups aside from GFI (also an ACE top charity). If the premise of GFI is that good then Open Phil will fund it, on top of the fact likely donors to Open Phil will benefit from GFI through their own investments in the field, for instance through New Crop Capital.

If indeed, Open Phil are fairly on board with the arguments put forward by GFI it consequently appears that welfare is a low value area of giving compared to other opportunities outside the animal movement; whilst it also suggests that hits-based giving could exclude mainstream animal groups that aren’t doing additional high value work in relation to social change and development. For instance, the ‘mainstream’ groups largely avoid this area out of concern for putting people off considering other animals, so these would instead represent low value options compared to others at the margin who do that work. Factor in how mainstream groups can access state funding, and are already well known in their field to attract donations as a matter of course, and value diminishes yet further.

I have written to Open Phil a few times on this issue, and I’m disappointed by the lack of response. From my own perspective, it seemingly isn’t possible to have things both ways (appreciate plurality / work with a system that undermines plurality). Whilst if this issue isn’t sufficiently contextualised then how is it people would know what they were doing is either effective or worthwhile? Instead it appears as if Open Phil engage in predictable giving within a narrow ideological framework, which has the effect of perpetuating failures in the animal movement. The argument it could be worthwhile dismissing those failures over addressing them at little cost seems unsatisfactory. I hope at some point you will choose to look further into the issue, where groups such as ACE and Open Phil choose not to engage with these matters, it makes working in the animal movement frustrating and less worthwhile, with a notable exception for people who benefit from choosing to accept and conform. It also seems to matter little how rational or good arguments could be, if people aren’t willing to engage, and instead prefer to follow perceived authorities.



Hi Kevin, I’m having trouble following you again. I think the work we are funding is high-impact and has been successful, so I don’t see it as perpetuating failures.

Perpetuating failure in the sense the movement is modelled as either ‘welfarist’ or ‘abolitionist’. So i would argue that basing our calculations on this paradigm generates movement failure because it inherently excludes groups that are generally working outside welfare, yet are largely working within a social justice framework.

The framework that is used by most people in EA reflects the welfare / abolition dichotomy. See for instance Jon Bockman, Bruce Friedrich and Paul Shapiro. As i view the situation this paradigm has largely privileged ‘welfarism’ or the approach generally adopted by the ‘mainstream’ animal movement.

Where we consider the impact of funding a certain set of ‘welfare’ groups, i think we need to go beyond where a donation might provide a certain result, and give consideration to the broader movement impact. In this way bearing some responsibility to reduce potential harms to different groups outside the traditional welfare / abolition paradigm.…

Hi Kevin, I think my first response to you covers my thoughts on these points.

Seeing, I was curious why there was no relationship disclosure provided? My understanding is that Nick Beckstead was, and still is, a “trustee of CEA”, which could indicate a significant conflict of interest. I also imagine that there are many other relationships between CEA staff and OpenPhil staff.

While I applaud OpenPhil’s transparency and still think that, despite this, OpenPhil remains more transparent than nearly every other foundation, this seems like a large break in OpenPhil’s policy toward disclosing relationships and other conflicts of interest.

Seems like they discontinued relationship disclosures. See for more.

Hi Peter, Vipul’s correct - we’ve changed our policy to no longer include relationship disclosures in our public writeups by default. Nick remains a trustee of CEA and that relationship was disclosed to decisionmakers here as part of the grant approval process, but consistent with our evolving thinking on what information is important to share publicly, we no longer plan on putting things like this in public grant writeups. We aren’t aiming for our grant writeups to be construed as arm’s-length charity reviews.

In these late years and especially in late months, mass-media of the whole world are rousing panic in society regarding AI and activities of its creation. They publish interviews with popular scientists, cybernetists and engineers where the main stage is not given to real predictions of AI capabilities, not to real predictions of AI and Human relationships, but to future dangers, associated with AI. Such famous people, as entrepreneur and engineer Elon Musk, physicists Max Tegmark and Steven Hawking claim that not only the AI will turn out to be hostile to human, but also that it can destroy whole human civilization. Alarm moods in society do not benefit popularization of AI creation works that are conducted throughout the world, moreover, such moods oppose the natural course of technical and scientific progress. In fact, today in the Western culture there is no positive AI models and future Humanity-AI coexistence models that are widely known in society. Meanwhile both things compose an essential part of evolution of humanity and technology.
We should also note that alarm scenarios of coexistence of humanity and AI cause in society distrust to AI creators, doubt in their competence, especially when they talk about such unlikely, even in future, things as immortality, omniscience and omnipotence.
Thus, I can offer you an alternative of real research of supposed positive scenarios of mutual and friendly evolution of human and AI. These scenarios are more realistic than alarm ones, which give to humanity nothing but harm. These scenarios take into account culture traditions, historical images and other (including physiological) individualities of different ethnos, that are being mentioned today by authors-alarmists. In order to include positive scenarios to the general model of a human-friendly AI, we need to conduct a thorough research, that takes into account all peculiarities of reception of AI ideas by different cultures and ethnos.
Scenarios that I can present to your attention are based on personal researches I conduct for many years as well, as ideas of different cultures of the East.

Dr. Pesakh Amnuel, astrophysicist, science-fiction author, journalist, TSIP* master

* - TSIP – a Theory of Solving of Invention Problem, developed by Soviet inventor Genrich Altshuller. Today TSIP is one of the most popular methods of creative problem solving in the world.

Have open threads been discontinued? I didn’t see one for September.

I’m following up on this. Is the discontinuation of open threads part of the “evolving” approach to openness?

This topic is on my to do list and I expect to get to it this week.

Hi Vipul, it looks like we just lost track of this task, and have revised our reminder procedures. I expect to post the next open thread in the morning on Monday or Tuesday.


Leave a comment