Center for Applied Rationality — General Support

The Open Philanthropy Project recommended a grant of $1,035,000 over two years to the Center for Applied Rationality (CFAR).

CFAR is an organization that runs workshops aimed at helping participants avoid known cognitive biases, form better mental habits, and thereby make better decisions. Our primary interest in these workshops is that we believe they introduce people to and/or strengthen their connections with the effective altruism (EA) community. CFAR is particularly interested in growing the talent pipeline for work on potential risks from advanced artificial intelligence (AI). More on our interest in supporting work along these lines is here.

$915,000 of this grant will support CFAR workshops and organizational improvements. $120,000 of this grant will fund a pilot version of EuroSPARC, an eight-day summer program in Europe run by CFAR for mathematically gifted high school students, modeled on the San Francisco-based Summer Program in Applied Rationality and Cognition (SPARC), which CFAR has helped run for the past three years. The Open Philanthropy Project previously made a separate grant to support SPARC’s operations in the U.S.

Based on discussions with CFAR staff and others connected to the organization, our impression is that CFAR’s management and operations are below the standards that we normally expect from our grantees. Despite that, we decided to make this grant primarily on the following grounds:

• We see CFAR’s values and goals and broadly aligned with our own.
• We believe CFAR has already had some success attracting and cultivating talented people to work on causes prioritized by the EA community, particularly potential risks from advanced artificial intelligence, which is one of our focus areas within the category of global catastrophic risks.
• Creating more of this kind of successful pipeline-building seems to us to be a valuable goal.
• Our understanding is also that funding is a substantial constraint for the organization at present, and that it believes it could use additional funding to address its organizational weaknesses.
• We therefore hope that this grant will allow CFAR to become a better-run and more professional organization that can more effectively carry out its mission.

1. Background

This grant is one of a small number of grants we are making in 2016 to support organizations focused on expanding and supporting the effective altruism (EA) community. More on our interest in this space here.

2. About the grant

2.1 Proposed activities

The bulk of this grant is intended to allow CFAR to fund a variety of things related to organizational development, specifically:

• Increasing salaries for existing staff, including implementing a variable pay scale
• Hiring an executive assistant
• Hiring a staff member dedicated to operations
• Freeing up senior staff time by reducing the amount of time they spend on sales
• Outsourcing accounting and other organizational services

CFAR also plans to use part of this grant to fund workshops aimed at introducing participants to EA, CFAR’s approaches to making better decisions, and the associated communities, with the goal that some participants will make career plan changes in order to increase the chance that they can have a large positive impact on the world. This could include deciding to work at organizations that focus on global catastrophic risk reduction, work for EA organizations, or work on other projects, for example as scientists or entrepreneurs.

We believe that this kind of pipeline-building work has the potential to be very beneficial, and we believe it is plausible that CFAR’s workshops are effective at it. Some examples of cases which we believe plausibly demonstrate this:

• Conversations with CFAR and comments by Max Tegmark¹ suggest to us that CFAR’s workshops played a significant role in the founding of the Future of Life Institute (which we have supported in 2015 and 2016).
• CFAR’s MIRI Summer Fellows Program² last summer appears to have been successful in recruiting researchers for the Machine Intelligence Research Institute (MIRI). At the time this grant decision was made, we found it difficult to evaluate the value of MIRI’s being able to recruit additional researchers, and guessed that it might be quite valuable. Having now completed an evaluation of MIRI, we now have a substantially lower estimate of the value of this activity, but are still quite unsure.
• The founder of Wave³ (a startup that reduces the fees paid on remittances) describes CFAR as influential in the company’s success.⁴

The EuroSPARC portion of this grant will be used to cover program and travel expenses for 10 instructors and roughly 20 students for the eight-day EuroSPARC program this summer. The program will be free for participants and provide need-based travel support. Funds will also be used to compensate instructors for their time (including planning and preparation), and a portion of the budget is set aside to cover unanticipated expenses or otherwise go towards general support for CFAR.

2.2 Budget and room for more funding

CFAR plans to spend our two-year grant roughly as follows:

First year

• $360,000 for organizational improvements
• $100,000 for scholarships for CFAR workshops
• $120,000 for EuroSPARC 2016
• $47,500 for half the salary and benefits of a new staff member, whom CFAR expects to work half-time on support for CFAR’s operations and half-time on logistics for SPARC (the other half of the funding for this staff member will come from our separate, previous grant to SPARC)

Second year

• $360,000 for organizational improvements
• $47,500 for half the salary and benefits of a new staff member (see above)

Our understanding is that CFAR is unlikely to receive funding at this scale to implement organizational improvements from other donors, and that EuroSPARC would be unlikely to go ahead this year without our support.

Our impression is that CFAR has been relatively poorly funded in comparison with many other organizations in the EA community; for example, its 2015 winter fundraiser went relatively poorly and left it with a very small amount of reserves.

2.3 Risks and reservations

Our primary reservations about this grant are around CFAR’s strength as an organization. We have some doubts about CFAR’s management and operations, and we see CFAR as having made only limited improvements over the last two years, with the possible exception of running the MIRI Summer Fellows Program⁵ in 2015, which we understand to have been relatively successful at recruiting staff for MIRI.

We see a need for more people with experience in AI or machine learning to become involved with work on potential risks from advanced AI. However, we believe there is some risk that CFAR might end up mainly attracting people without such a background, and we are highly uncertain about the value of this.

2.4 Goals for the grant

By the end of the first year of this grant, we hope to observe:

• Substantial improvements in CFAR’s organizational effectiveness.
• CFAR alumni (who attribute some aspects of their decision-making, planning, or success to CFAR’s influence) engaged in highly impactful work.

In deciding whether to renew our grant at that point, we plan to assess whether CFAR has achieved a number of milestones that it set out,⁶ including:

1. Publishing metrics of its impact (especially on significant plan changes by CFAR alumni).
2. Improving its accounting and financial reports.
3. Issuing quarterly updates to its supporters, quarterly financial reports to its board, and annual reviews and plans on its website.
4. Improving its fundraising by, e.g., improving its donor database, cultivating relationships with top donors, and holding open house events for donors.

If CFAR has improved in these ways, we are likely to renew our grant for additional years and would consider supporting CFAR to scale up its operations.

We may also decide to continue our support if, alternatively, CFAR has other unanticipated achievements that we see as highly impactful.

2.5 Key questions for follow-up

• Did CFAR effectively document plan changes attributable to CFAR among its alumni?
• How many plan changes were there? Which were the most significant?
• How impressive do we find the recipients of CFAR’s workshop scholarships? How likely would they have been to attend CFAR without the scholarships?
• Did EuroSPARC attract a good group of students? Did the students enjoy the program? Do SPARC instructors believe that EuroSPARC was successful overall?
• Did the MIRI Summer Fellows program and other efforts focused on potential risks from advanced AI lead to new hires at MIRI?
• Do groups working on potential risks from advanced AI report hiring any new staff or finding new donors that were influenced by CFAR’s programs?

3. Our process

While investigating this grant, we had several conversations with Anna Salamon, as well as with various other contacts of ours in the EA community. Nick Beckstead was the primary investigator for this grant.

We are currently considering grants to some other organizations focused on effective altruism (more). The order and amounts of grants we make in this area will depend on a number of factors, including the order in which grants are considered and the timelines by which potential grantees need funding decisions. So we would encourage readers not to use our decision to make this grant (in isolation) to draw strong conclusions about the Open Philanthropy Project’s views and priorities regarding other organizations working in the area.

4. Sources