An update on the Global Priorities Project

The purpose of this post is to let people know a bit more about what we’ve been doing and what we hope to do. Please let us know what questions you have that we don’t answer below!

Activities

The Global Priorities Project (GPP) was created in order to pursue a wide range of low-hanging fruit. This has meant that our work has covered a fairly broad range across two areas—prioritisation and policy work.

Our prioritisation work has been about identifying considerations and methods for choosing between different ways of helping the world. Our policy work has been exploring for high-value ways for the effective altruism community to interact with policy processes. A little over one-third of our work has been policy, and the rest prioritisation.

In the next sections I’ll give an overview of what we’ve been doing in these areas. I’ll cover most of our activities, but a few are excluded because (i) they were a small time investment and we do not regard them as that important; (ii) we are still exploring them and may not complete them; or (iii) the projects involve other people, and we are being cautious about not jeopardising relationships.

Prioritisation

We have three areas of prioritisation work:

  1. Cause prioritisation

  2. Prioritisation within catastrophic risk reduction

  3. Academic prioritisation methodology

Cause prioritisation

This work is aimed principally at people who are interested in making strategic choices about what areas to focus on in helping the world—in particular the effective altruism movement. We focus on facilitating comparisons between diverse areas, rather than on making choices within areas. We have produced a flowchart as an overview and communication tool, and also have some posts looking in depth at more specific issues, for example trying to compare movement growth to more direct work. We have drawn on some of these tools (and the more methodological ones discussed below) to advise Effective Altruism Ventures on how to make comparisons between projects in very different areas. We are interested in developing the flowchart further into more detailed and interactive versions.

Prioritisation within catastrophic risk reduction

This is a strand of work which grew out of cause prioritisation. We wanted to be able to provide principled comparisons between work on catastrophic risk reduction and other areas. We still regard this as a very high-value question we have not properly answered. However in investigating it we developed methods for comparing the value of different areas of work within catastrophic risk reduction. This helped us to secure funding from the Future of Life Institute to develop and extend the methodology for prioritisation within AI safety. Other work in this category looks at the timing of work on existential risk, and at how we can produce crude estimates of the value of safety work in AI.

Academic prioritisation methodology

This work aims to identify important considerations in prioritisation, and write them up rigorously so that they can be embedded in the academic discourse. We see this as an important long-term route to impact. Our largest work in this space is a paper we are just readying for journal submission on estimating the ex ante value of marginal research. This develops a project (within cause prioritisation) from last year in which we looked at estimating the returns from working on projects of unknown difficulty. Also in the category of academic prioritisation, Owen has prepared a book chapter on discounting for uncertainty in health, and contributed to another on integrating the valuation of health and consumption effects.

Impact of prioritisation work

It is hard to estimate exactly what the impact of different bits of this research is. Particularly the academic work has quite a long feedback loop, and for some of the other research we see the main value in its capacity to inform future more developed research. Nonetheless we are interested in tracking which parts of the work seem to be particularly valuable, and are worth pursuing further or looking for opportunities to disseminate more widely. To date, we have mostly been evaluating this by (i) asking people whose judgement we trust which work is important, (ii) seeing which work is cited by others, and (iii) looking at which work gets most views on our website.

Policy

Our work in policy so far has largely been driven by high-value opportunities. We have seen two kinds of opportunity: where we have a particularly good connection to specific people, or where we have an idea that seems important.

Catastrophic risk and policy

Much of our policy work has been helping to steer policy conversations around catastrophic and existential risk. We coordinated and wrote a policy report on Unprecedented Technological Risks, which was shared within the UK government, and led to two of the authors being invited to write a chapter in the Government Chief Scientific Advisor’s annual report. More recently, we have been exploring more detailed categorisations of sources of risk (further work forthcoming). Based on this existing work, we have secured funding from the Finnish Foreign Ministry to write a report on existential risk and governance, and we have begun preparation work for this.

The UK runs a national assessment of possible catastrophic risks, assessing them on likelihood and impact. We noticed that parts of the methodology they used for assessing impact was not well-grounded. We sent a short document explaining how we thought this might be improved to the team working on it, had a meeting with them, and sent a second document. They are in the process of concluding a review of their methodology, and our understanding is that they have made some changes in the direction of our recommendations, and that we represented a good fraction of the voices pushing for these changes (although at least some were also requested by others). We think getting the methodology correct on this assessment could lead to switching the order of concern about some risks and lead to more appropriate research allocation. We hope that it may be taken up in other large-scale risk assessments.

A specific area we are interested in is biosafety policy. This seems an important and timely area. We published a preliminary policy idea last year which was well-received by some people in the biosafety community. Since then we have continued conversations, and looked at the broader academic literature on incentives and risks. More recently, this summer we had a postdoc in life sciences join us as an intern and prepare neutral overviews of the risks, benefits and possible responses to “gain-of-function” work in influenza.

Other high-value policy opportunities

We’re interested in other places where an effective altruist perspective may be useful in informing policy. We think this is important, because most people in policy are trying to help, and identifying good policy can be difficult. So far much of our work here has been strategic. We are developing one policy suggestion on improving the clarity of communication of statistics by governments.

People

The Global Priorities Project was started in January 2014. At that time it had 75% of Owen Cotton-Barratt’s time focusing on research, and a small amount of time from other individuals at the Centre for Effective Altruism and Future of Humanity Institute, notably Niel Bowerman.

In January 2015, Seb Farquhar joined full time in a research and project management capacity. In July, Seb took on a role as Executive Director of the Centre for Effective Altruism, and reduced his time on GPP to 50%.

We have also been fortunate enough to have some very talented interns join us over the summers, typically for 6-8 weeks. In 2014, Max Dalton worked with us. In 2015, Hong-Sheng Lim, Nick Phillips, Daniel Kokotajlo, and Nicolas Moes worked with us.

The total time (full time equivalent) that has been devoted to GPP to date is therefore roughly:

Owen Cotton-Barratt: 16 months

Seb Farquhar: 7 months

Niel Bowerman: 2 months

Interns: 8 months

Others: 1 month

Total: 34 months (14 months in 2014, 20 months in 2015)

Future directions

Historically we have pursued both prioritisation and policy work. We see them both as high-value activities, and the opportunities were good enough that we wanted to take them. However, it is not necessary that they are combined in one organisation, and we are looking at separating out these workstreams more explicitly.

We’re also interested in increasing our attention to areas other than existential risk. A year ago, we thought this was significantly neglected for both policy and prioritisation. However it is getting increasing amounts of attention, and this means that going forward it may not be the most neglected area. We are heavily engaged in the area, and our current funding is focused on this, but we are interested in pushing more on strategy/​prioritisation more broadly, and on other areas of policy.

Also note that we are currently hiring. We are looking to bring on more researchers and to build our policy team.

[Cross-posted from our website. Thanks to Michael Dickens for feedback which prompted this post.]