Update on How We’re Thinking about Openness and Information Sharing

One of our core values is sharing what we’re learning. We envision a world in which philanthropists increasingly discuss their research, reasoning, results and mistakes publicly to help each other learn more quickly and serve others more effectively.

However, we think there has been confusion – including in our own heads – between the above idea and a related one: the idea that philanthropists should share and explain their thinking near-comprehensively so that the reasoning behind every decision can be understood and critiqued.

Such near-comprehensive information sharing is an appropriate goal for GiveWell, which exists primarily to make recommendations to the public, and emphasizes the transparency of these recommendations as a key reason to follow them. (See GiveWell’s approach to transparency.)

However, we now feel it is not an appropriate goal for the Open Philanthropy Project, whose mission is to give as effectively as we can and share our findings openly so that anyone can build on our work. For our mission, it seems more appropriate to aim for extensive information sharing (well in excess of what other funders currently do) but not to aim for near-comprehensiveness.

This distinction has become more salient to us as our picture of the costs and benefits of information sharing has evolved. This post lays out that evolution, and some changes we plan to make going forward. In brief:

• For a number of reasons, we now see greater costs to high-volume information sharing, and lower benefit, than we saw previously.
  • We’ve taken on projects with increasingly complex and resource-intensive-to-explain justifications, which has both raised the costs of information sharing and lowered the benefits. Since we’re not able to make the full case for our thinking to a general audience, we see few helpful reactions and criticisms via this channel, and we rely on the communities with the most knowledge of our issues – rather than our general audience – for most critical feedback.
  • We’ve entered into some areas that are subject to controversy, where sharing information publicly can create tangible programmatic risks. (This also pertains to the previous point, since risks can include impairing the quality of feedback we’re able to get from the communities with the most knowledge of our issues.)
  • We’ve also changed our process for writeups such that our overall efficiency has improved, but costs of information sharing are now higher.
• We still see major benefits to openness, but believe we can realize similar benefits with less volume. Our main goal is to help others understand the big picture behind how we think and the reasons for our major choices. We believe we can accomplish this by publicly sharing a lot of information about our thinking rather than publicly explaining each grant and other decision we make.
• We have stopped the practice of writing in detail about every grant that we make. We plan to continue to write in detail about many of our grants. We will try to focus on those that are especially representative of our thinking and strategy, or otherwise seem like they would be interesting and helpful to discuss. We will continue to maintain a number of other information sharing practices. We believe that our information sharing will remain much more extensive than what we currently see from other funders.
• We have also reduced our use of the term “transparency,” which we think has too strong a connotation of comprehensiveness. We prefer “openness” and “information sharing,” and plan to revise some of the language on our website accordingly.

1. Changes in our picture of the costs and benefits of information sharing

Over the last couple of years, the following factors have affected our thinking about information sharing:

We’ve taken on projects with increasingly complex and resource-intensive-to-explain justifications. We’ve built knowledge about our focus areas; we’ve hired several Program Officers and Scientific Advisors who bring deep expertise and networks in particular areas; and we’ve built other relationships with domain experts (for example, technical advisors on artificial intelligence research) whose views we put strong weight on. In many cases, the result has been a rapidly growing gap between (a) the reasons we believe what we believe and (b) the reasons we’re able to share publicly and relatively efficiently.

For example, earlier this year, I sought to lay out why we believe potential risks from advanced artificial intelligence represent an outstanding philanthropic opportunity. Writing and revising two blog posts on the subject took more than 70 hours of my time, plus time from many other people. Yet these posts present only a broad outline of our thinking, which is based on many in-depth conversations with experts in a variety of domains; they don’t present a thorough justification on par with GiveWell’s top charity recommendations.

GiveWell explicitly seeks out giving opportunities where the case can be made using publicly disclosable and verifiable information. By dropping this criterion, the Open Philanthropy Project has gained the ability to consider a bigger universe of possibilities, particularly high-risk, high-reward giving opportunities. The disadvantage is that it becomes prohibitively difficult for our public content to completely explain our thinking.

As this challenge has grown, it has not only raised the costs of explaining our thinking; it has lowered the benefits. Since we’re not able to make the full case for our thinking to a general audience, we see few helpful reactions and criticisms via this channel. In general, we rely on the communities with the most knowledge of our issues – rather than our general audience – for most critical feedback.

We’ve entered into some areas that are subject to controversy, where sharing information publicly can create tangible programmatic risks. As discussed previously, we expect our work to sometimes involve controversial or adversarial situations.

Some specific risks of sharing too much information in controversial or adversarial areas:

• In some areas – particularly those related to policy – sharing too much information about our grantees’ plans and reasoning can undermine their work by helping people with directly opposing goals anticipate their actions. This creates an additional challenge for public writeups that we navigate along with grantees.
• In some areas, we risk being subject to harassment from people who oppose our goals.
• In some areas, we run the risk that our thinking will be widely misinterpreted, affecting our reputation and our attractiveness as an employer, funder, partner, etc. This could have major implications for our long-term ability to have as much positive impact as possible, which is our most important goal. Note that this point also pertains to the previous section, since problems on this front might impair the quality of feedback we’re able to get from the communities with the most knowledge of our issues.Potential risks from advanced artificial intelligence is one such area: we know that working on this cause creates a risk of being perceived as anti-technology and alarmist by some, as well as being perceived as technology-obsessed and overconfident by others. So far, we think we’ve largely been able to manage these risks, but it has come at the cost of putting significant time and care into our public communications.

We’ve changed our process for write ups. Previously, our Program Officers created a single writeup for two purposes: recommending and explaining a grant internally, and sharing information about the grant publicly. Now, we have a separate process – undertaken by other staff – for producing external writeups. This frees up our Program Officers to spend more time on their core role. But it also means that public writeups involve an extra layer of staffing to produce, and they no longer serve a helpful dual purpose. (Previously, they were useful internally for documenting and tracking our thinking, but now internal writeups serve that purpose.)

2. Realizing the benefits of openness we most value

We still see major benefits to openness. At this point, we think the most significant benefit is helping others learn about the “big picture” behind our thinking. We want people to understand how we think and the reasons for our major choices, so they can decide whether our basic style of reasoning resonates, and we want to give other philanthropists food for thought.

We believe we can realize essentially this full set of benefits by publicly sharing a lot of information about our thinking rather than publicly explaining each grant and other decision we make. We think the existing content on our website – particularly our Notable Lessons page, our annual self-evaluations, and our description of our process for choosing focus areas – already goes a long way toward explaining our basic approach to philanthropy and our most important choices (and represents a higher degree of openness than any other funder we’re aware of).

3. Our approach going forward

This post is largely intended to mark a conceptual change in how we think about information sharing (though the change has occurred gradually over the last few years). The specific actions that accompany this change, for the time being, are relatively minor.

First, we have been generally reducing our use of the term “transparency,” which we think has too strong a connotation of comprehensiveness. We prefer “openness” and “information sharing,” and plan to revise some of the language on our website accordingly.

Second, we have stopped the practice of writing in detail about every grant that we make. We plan to continue to write in detail about many of them. We will try to focus on those that are especially representative of our thinking and strategy, or otherwise seem like they would be interesting and helpful to discuss.

In most cases, when we do not produce a detailed writeup on a grant, we will still include the grant in our database with basic information about the date, amount and recipient, and a link to the relevant focus area so people can understand what broader goals the grant supports.

In some cases, even basic information about a grant (date, amount, recipient) could constitute sensitive information, for reasons including:

• It might reveal our tactical approach to a contested issue, helping people with directly opposing goals anticipate our and our grantees’ actions.
• It might pose a serious risk of misinterpretation, and explaining our reasoning might be difficult and time-consuming.
• It might pose a risk of exposing us to active harassment from people who oppose our goals.

In these cases, we will generally delay publishing basic information about the grant until and unless we have settled on a good strategy for communicating about it. We expect these cases to be rare.¹

In addition to maintaining our grants database, we will also:

• Maintain focus area pages with links that explain the basic justifications behind and strategies within our focus areas.
• Blog about important big-picture changes in our thinking, and add key posts to our Notable Lessons page.
• Put out annual reviews of what we’ve done over the last year and are planning to do over the next year.
• Post conversation notes when we think a conversation would be interesting and/or helpful in understanding our thinking on a particular topic, and when the interviewee is comfortable with our doing so. (Note that our Program Officers tend not to publish notes from conversations with close contacts; these conversations are sometimes sensitive and typically have deep context behind them. We are more likely to publish notes when we investigate a particular question of interest via a small number of conversations.)
• Write in detail about a number of our grants, when we think this would be particularly interesting or helpful, as discussed above.
• Share additional information with specific parties, when we have a specific reason to do so and trust the parties in question to respect confidentiality.

We share information with the public primarily so we can help others learn more quickly and serve others more effectively, thereby amplifying our impact. We expect to continue revising our practices to balance this long-term goal with the more direct goals of our grantmaking.