The Global Priorities Institute has published a new research agenda, which aims to reflect our current research focus accurately.

GPI's mission is to conduct and promote what we call ‘global priorities research’: research into issues that arise in response to the question, ‘What should we do with a given amount of limited resources if our aim is to do the most good?’ This question naturally draws upon central themes in the fields of economics and philosophy.

Thus defined, global priorities research is in principle a broad umbrella. Within that umbrella, this research agenda sets out the more specific research themes that GPI is particularly interested in at the present time.

The research agenda is structured as follows:

  • Section 1 outlines what we call the longtermism paradigm. This paradigm centres around the idea that because of the potential vastness of the future portion of the history of sentient life, it may well be that the primary determinant of which actions are best is the effects of those actions on the very long-run future, rather than on more immediate considerations. Because these ideas seem plausible, seem likely to have fairly radically revisionary implications if correct, and are currently quite neglected, this is the main focus of GPI’s own research (at the time of writing and, we predict, for at least the next two years). We are particularly keen to hear from other researchers who share this interest.
  • Section 2 concerns general issues in cause prioritisation. This covers issues that are not specific to a longtermist point of view, but that arise for agents engaged in an exercise of global prioritisation.

The intended audience for this document is academics (especially, but not only, in economics and philosophy) who are potentially interested in working with GPI, whether as GPI researchers or as external collaborators, or who are otherwise interested in the same mission.

We invite you to read the new research agenda here.

Comments11


Sorted by Click to highlight new comments since:

This is really really impressive. An amazing collection of really important questions.

POSITIVES. I like the fact that you intend to research:
* Institutional actors (2.8). Significant changes to the world are likely to come through institutional actors and the EA community has largely ignored them to date. The existing research has focused so much on the benefits of marginal donations (or marginal research) that our views on cause prioritisation cannot be easily applied to states. As someone into EA in the business of influencing states this is a really problematic oversight of the community to date, that we should be looking to fix as soon as possible.
* Decision-theoretic issues (2.1)
* The use of discount rates. This is practically useful for decision makers.

OMISSIONS. I did however note a few things that I would have expected to be included, to not be mentioned in this research agenda in particular there was no discussion on
* Useful models for thinking about and talking about cause prioritisation. In particular the scale neglectedness and tractability framework is often used and often criticised. What other models can or should be used by the EA community.
* Social change. Within section 1 there is some discussion of broad verses narrow future focused interventions, and so I would have expected a similar discussion in section 2 on social change interventions verses targeted interventions in general. This was not mentioned.
* (which risks to the future are most concerning. Although I assume this is because those topics are being covered by others such as FHI.)

CONCERN
Like I said above I think the questions within 2.8 are really importation for EA to focus on. I hope that the fact it is low on the list does not mean it is not priorotised.
I also note that there is a sub-question in 2.8 on "what is the best feasible voting system". I think this issue comes up too much and is often a distraction. It feels like a minor sub part of the question on "what is the optimal institution design" which people gravitate too because it is the most visible part of many political systems, but is really unlikely to be thing on the margin that most needs improving.

I hope that helps, Sam

What are these other questions about optimal institution design, which you consider more important than voting systems?

There are maybe 100+ other steps to policy as important as voting system design. In rough chronological order I started listing some of them below (I got bored part way through and stopped at what looks like 40 points).

I have aimed to have all of these issues at a roughly similar order of magnitude of importance. The scale of these issues will depend on country to country and the tractability of trying to change these issues will vary with time and depend on individual to individual.

Overall I would say that voting reform is not obviously more or less important than the other 100+ things that could be on this list (although I guess it is often likely to somewhere in the top 50% of issues). There is a lot more uncertainty about what the best voting mechanisms look like than many of the other issues on the list. It is also an issue that may be hard to change compared to some of the others.

Either way voting reform is a tiny part of an incredibly long process, a process with some huge areas for improvements in other parts.


SETTING BOUNDARIES

  • constitution and human rights and setting remits of political powers to change fundamental structures of country
  • devolution and setting remits of central political powers verses local political bodies
  • term limits

CHOOSING POLITICIANS

  • electoral commission body setting or adjusting borders of voting areas / constituencies
  • initial policy research by potential candidates (often with very limited resources)
  • manifesto writing (this is hugely important to set the agenda and hard to change once )
  • public / parties choosing candidates (often a lot of internal party squabbling behind the scenes)
  • campaign fundraising (maybe undue influences)
  • campaigning and information spreading (maybe issues with false information)
  • tackling voter apathy / engagement
  • Voting mechanism
  • coalition forming (often very untransparent)
  • government/leader assigns topic areas to ministers / seniors (very political, evidence that understanding a topic is inversely proportional to how long a minister will work on that topic)

CIVIL SERVICE STAFFING

  • hiring staff into government (hiring processes, lack of expertise, diversity issues)
  • how staff in government are managed (values, team building, rewards, progression, diversity)
  • how staff in government are trained (feedback mechanisms, training)

ACCOUNTABILITY

  • splitting out areas where political leadership is needed and areas where technocratic leadership is needed
  • designing clear mechanisms of accountability to topics so that politicians and civil servants are aware of what their responsibilities are and can be held to account for their actions (this is super important)
  • ensuring political representation so each individual has direct access to a politician who is accountable for their concerns
  • putting in place systems that allow changes to the system if an accountability mechanisms is not working
  • ensuring accountability for unknown unknown issues that may arise
  • how poor performance of political and civil staff is addressed (poor performance procedures, whistleblowing)
  • how corruption is rooted out and addressed (yes there is corruption in developed countries)
  • mechanisms to allow parties / populations to kick out bad leaders if needed
  • Ensuring mechanisms for cross party dialogue and that partisan-ism of politics does not lead to distortions of truth

AGENDA SETTING AND INITIAL RESEARCH

  • carrying out research to understand what the policy problems are (often unclear how to do this)
  • understanding what the population wants (public often ignored, need good procedures for information gathering, public consultation, etc)

POLICY DEVELOPMENT

  • Development of policy options to address problems
  • Mechanisms for Cost Benefit Analysis and Impact Assessments to decide best policy options
  • access to expertise advice and best practice (lack of communication between academia and policy)
  • measuring impact of a policy proposal once in place (ensuring that mechanisms to measure impact are initiated at the very start of the policy implementation)
  • actually using information on
  • how politicians are allowed to change their mind given new evidence (updating is often seen as weakness)
  • mechanisms to ensure issues that are not politically immediately necessary are tackled (lack of long term thinking)

LEGISLATIVE PROCESS

POLICY IMPLEMENTATION

POLICY REVIEW

PUBLIC COMMUNICATION

RISK MANAGEMENT

GENERAL

  • flexibility to deal with shocks of every step of the above process (often lacking)
  • transparency of every step of the above process (often lacking)

Very nice.

Is there a quick way to use the agenda to see GPI's research prioritization? (e.g. perhaps the table of contents is ordered from high-to-low priority?)

Thank you for asking! The ordering of research areas in the Table of Contents does loosely track our research prioritisation, but only within the constraints of making the Agenda a reasonably coherent document, with similar subject matter grouped together.

We hope to post a web-friendly version of the agenda in the next month or two. This version will let users sort the research topics in various ways, including, potentially, directly on the basis of research priority.

Can you put up a plain text version of this? PDFs aren't absorbed nicely by other software (e.g. Facebook for sharing, Instapaper for saving to read later, etc.)

Wonderful, thank you! :)

Suppose I want to give a counter argument to one of GPI's research papers. Where can I post such a response?

More precisely, I want to argue that the reasoning in " Moral Uncertainty About Population Axiology " is not compatible with the most plausible ways of normalizing different axiologies, such as variance normalization.

You are more than welcome to post a response on any medium you choose. That said, as GPI's mission is in part to bring EA-relevant research topics into the academic mainstream, responses in the form of academic publications are particularly encouraged.

This looks great, and I'm happy to see all of these questions in one place.

In case anyone finds it useful, I visualised an alternative breakdown of the questions here, in which each category of questioning builds on the previous one.

Curated and popular this week
Ozzie Gooen
 ·  · 9m read
 · 
We’re releasing Squiggle AI, a tool that generates probabilistic models using the Squiggle language. This can provide early cost-effectiveness models and other kinds of probabilistic programs. No prior Squiggle knowledge is required to use Squiggle AI. Simply ask for whatever you want to estimate, and the results should be fairly understandable. The Squiggle programming language acts as an adjustable backend, but isn’t mandatory to learn. Beyond being directly useful, we’re interested in Squiggle AI as an experiment in epistemic reasoning with LLMs. We hope it will help highlight potential strengths, weaknesses, and directions for the field. Screenshots The “Playground” view after it finishes a successful workflow. Form on the left, code in the middle, code output on the right.The “Steps” page. Shows all of the steps that the workflow went through, next to the form on the left. For each, shows a simplified view of recent messages to and from the LLM. Motivation Organizations in the effective altruism and rationalist communities regularly rely on cost-effectiveness analyses and fermi estimates to guide their decisions. QURI's mission is to make these probabilistic tools more accessible and reliable for altruistic causes. However, our experience with tools like Squiggle and Guesstimate has revealed a significant challenge: even highly skilled domain experts frequently struggle with the basic programming requirements and often make errors in their models. This suggests a need for alternative approaches. Language models seem particularly well-suited to address these difficulties. Fermi estimates typically follow straightforward patterns and rely on common assumptions, making them potentially ideal candidates for LLM assistance. Previous direct experiments with Claude and ChatGPT alone proved insufficient, but with substantial iteration, we've developed a framework that significantly improves the output quality and user experience. We're focusing specifically on
 ·  · 44m read
 · 
There's an element of preaching to the choir in posting this here, and of course dietary choice seems much less important than choices around where to donate. But I still think this FAQ, which is really more of a Q, could be of interest to EAF users. Preamble Why did you write this FAQ? I wrote this FAQ to explain why I think it is the case that animals raised for food suffer a lot, and why, as a result, people oughtn't to eat meat. Others have written similar things, but I wanted to explain my own reasons for not eating meat. Some people think eating meat is bad because it causes greenhouse gas emissions. Others avoid meat for health reasons. Personally, I would not eat meat even if it turned out that eating meat helped fight climate change.[1] (That said, as I'll discuss later, I also think actions other than going vegetarian or vegan can have even greater impact. For example, donating to effective animal charities often does more good, and eliminating factory-farmed eggs reduces more suffering than eliminating beef. I don't claim that vegetarianism is the best thing one can do for animals, only that it is very good compared to doing nothing.) If I'm right about the suffering of factory-farmed animals -- and suspend your judgment on that point for now -- then what we're doing to those animals is one of the worst moral catastrophes in history, comparable to the worst instances of slavery, war and genocide. If so, it is an ongoing moral catastrophe affecting victims that cannot advocate for themselves. Generally speaking, whenever we risk making a serious moral mistake, we should think it through carefully and honestly, just like those involved in the Atlantic Slave trade or the Great Leap Forward (or other risky endeavours that had more positive outcomes) should have thought carefully and honestly about what they were doing.[2] But a lot of the arguments against vegetarianism that I see are weak, suggesting to me that most people have not thought carefully en
 ·  · 1m read
 · 
If you: 1. Pay the higher (40%) rate of tax in the UK (earn over £50,271) 2. Don’t fill in a Self Assessment tax return (you pay tax automatically via PAYE) 3. Made donations that you claimed Gift Aid on in this or any of the previous 4 tax years You can claim back 25% of the before-Gift-Aid amount by calling HMRC[1], or writing to them by post[2]. You can also apparently[3] do it via online chat although I haven’t tried this myself. I have personally called twice to report donations, once last year to claim for the previous 3 years, and once for the current tax year. Both times it was surprisingly easy. They just asked for my national insurance number, and the amount donated (before-Gift-Aid). They didn’t even ask for details of the individual donations, although obviously you should keep records of your donations in case you get audited, and you do have to give details if you are reporting more than £10k. The 25% goes straight into your take-home pay (via adjusting your tax code), and this won’t happen automatically if you don’t contact them. Further reading This post is mainly to serve as a nudge for what I would guess is a common cause of losing money when donating, because of the manual step required. These articles are more comprehensive: 1. The official advice from HMRC 2. UK Income Tax & Donations by @harald, which covers more than this specific case 3. The Payroll Giving topic page 1. ^  On 0300 200 3300, this is only allowed for reporting donations under £5k 2. ^ Post only, no email unfortunately. The address is: Pay As You Earn and Self Assessment HM Revenue and Customs BX9 1AS United Kingdom 3. ^  According to this post
Relevant opportunities