Suggestions for developing national-level effective altruism organizations

By Jan_Kulveit @ 2018-10-17T23:35 (+31)

Suggestions how to develop national-level effective altruism organizations so that they are robustly beneficial. If you are unconvinced that creation of such organizations is a positive development, this is disussed in he first part.

Status: This document was created mostly during my summer internship at CEA. It is a result of many discussions with people both from CEA, other central organizations, and EA community builders working in national and city level effective altruism organizations. The opinions expressed are often an attempt to find something which can be broadly supported - but should not be understood as an official position of CEA (or any other organization).

Sometimes, recommendations are based on experiences of several organizations, sometimes on my personal experience working in Czech EA Association, sometimes they are guesses based on research in relevant fields.

Part 1: Why develop national-level effective altruism organizations?

Roles of national-level organizations

In the global EA landscape, regional EA organizations in non-English-speaking countries (such as EA Geneva and CZEA) can fulfill various roles:

There is significant uncertainty about the desirability of these roles - we reflect on this by describing, in each case, what seems to be a promising approach to the role, to what extent this seems desirable, risky, or resource-intensive, how likely it is that this will get support, and what could be suitable sources of funding.

Localization of effective altruism

Localization of effective altruism to a new language or culture should not be understood as a mere translation of some texts, but more as the “transmission of a tradition of knowledge”. The primary task can be understood as answering the question: “how can we use our resources to help others the most?” in the context of a specific location. The answer will be somewhat different between places: the further away from Oxford or Berkeley, generally the more different from the standards established in those places.

We will use a three-component model for the task, heavily inspired by the post On the Loss and Preservation of Knowledge by Samo Burja on LessWrong - it is highly recommended to read the original post.

Three-component model of transmission of knowledge

The claim is: a successfully transmitted tradition of knowledge (e.g., research field, political philosophy) will cover these aspects:

We create similar model for translation and localization and translation of effective altruism

Three-component model of localization of effective altruism

The generators include Effective Altruism’s core question (“how do we help others in the most effective way?”), philosophical background, prioritization research and frameworks, and object-level knowledge of some of the cause areas.

Generators also include people with sufficient knowledge of the above that they would be able to essentially re-derive effective altruism, taking into account local differences such as cultural circumstances and local comparative advantages.

The knowledge itself is the actual state-of-the-art knowledge about effective altruism, represented for example by the Effective Altruism Handbook. Knowledge is likely the easiest part to “translate.” Surprisingly, from the three components, knowledge should receive the smallest amount of effort. This is because at least at present, new ideas in effective altruism are communicated in English, and it does not seem possible to keep up with the latest ideas in the movement without the ability to read in English.

So the purpose of direct translation of materials is not to translate the entire body of knowledge into another language, but only to translate some key texts to lower the barrier for entry. Human behavior is surprisingly influenced by small barriers and “bumps” in the incentive landscape - the point is to make the barrier to noticing the ideas smaller, and to direct the interested individuals to higher fidelity channels of an idea spreading, and to the whole body of EA knowledge in English. (If you would like to translate CEA's EA Handbook, please email content@effectivealtruism.org for permission)

Our current understanding of how to spread knowledge effectively part is covered in CEA’s fidelity model of spreading ideas.

Verification in practice rests on bidirectional, personal communication with people in the core and in one’s peer group (community builders in similar groups), due to the lack of formal verification mechanisms to check one’s understanding of effective altruism. This is important to ensure that what is being recreated is effective altruism and not just some superficial version.

Evaluation: Localizing effective altruism is a potentially high-impact endeavour with long-term consequences. It seems very important to do it right, as due to the founder effects and organizational inertia, larger organizations are less likely to change trajectory significantly after inception. Localizing effective altruism to a new country is not a small project and needs significant resources (outlined later). If a group has the outlined resources, it is likely such effort can get funding via Community building grants, and various other forms of support.

Routing talent

One of the main ways the effective altruism community creates value is through the development of people who are highly dedicated to, have a sophisticated understanding of, and possess skills relevant to working on the world's most pressing problems. (See CEA’s three-factor model of community building.)

National organizations can fill several important roles in creating talent pipelines. Apart from the usual community-building considerations common to all local groups, these are:

However, note there is an important trade-off between routing talent and preserving the talent pipeline: if all talent is routed elsewhere, and even the people creating the pipeline move, it may hinder routing talent in the future.

Evaluation: Talent development and routing is seen by many the most proven way by which the EA community creates value, so this is certainly encouraged. Moving people to effective careers and priority areas is also one of the best measurable outputs of effective altruism groups, so groups are and likely will be evaluated on this as a metric. Funding for groups aiming to develop talent are available via community grants, and generic support for local groups is often geared toward talent-development focused groups.

Support for local groups

Local groups in non-English speaking countries have a natural incentive to share and spread localized material, including creating structures for easier distribution. Whichever groups are largest, typically ones in major cities, will serve as natural examples for smaller groups, and people in areas with fewer EAs will have incentive to figure out what works in more EA-dense areas and learn from that. Since it’s likely that the largest group will have the most flow of EA people and ideas, prominent members of that group are likely to be involved with national efforts as well, and are in a good position to support smaller local groups directly.

This should not be understood as a strictly hierarchical relationship - local groups in non-English countries should probably get support from CEA and LEAN as well as their own national-level organizations.

Evaluation: Partial support of local groups seems to be a natural role of national organizations. However, supporting local groups shares some of the same risks as localization efforts - including spreading problematic versions of effective altruism and causing coordination problems - and efforts should be taken to mitigate these risks. Like localization, significant resources applied competently could be useful. Unlike localization, however, scaled-down versions are easy to implement - for example, a national organization could just provide narrow technical support to local groups, which would not entail much risk.

Currently, it is not clear how much funding would be useful for local group support.

Public relations and interfacing with media

National groups are the natural first-contact point for media in the national language. This is better understood as an opportunity to do some form of “damage management” rather than an opportunity to do broad outreach, as media coverage seems net negative in most cases.

Based on the experience of several national groups, the counterfactual to not having a media contact point is often not “the media not publishing about effective altruism”, but rather the media spreading obsolete and inaccurate portrayals of effective altruism based on old English-language materials, or interviewing just any effective altruist willing to speak with them.

Consequently, it seems plausible that some minimal communication with media or answering interview requests can be beneficial.

Evaluation: In general, media exposure has a negative expected value. So not communicating with media at all seems a reasonable option. On the other hand, there are cases when media-savvy communication can improve over the counterfactual where media outlets spread obsolete or inaccurate information.

Overall, this does not seem high priority and is not likely to attract funding by itself. On the other hand, there is probably value in exchanging experiences and know-how with EA organizations more experienced at managing PR.

Utilizing local comparative advantages

Effective altruism should be understood as a global collaborative effort, so part of the localization effort outlined earlier should be “given the local circumstances, what may be our comparative advantage?” (It is essential to truly grasp the concept. It is often the case that the absolute advantage in producing some effective altruist good is in a central place like Oxford.)

It seems many successful regional and national organizations’ efforts are based on careful analysis of local comparative advantage. A list of a few examples may provide a better intuitive understanding:

Note: The idea of comparative advantage should be applied mainly to functions, rather than an affinity for a particular cause area, as we think geographically-based organizations should be mostly cause-neutral. (E.g., the difference between “we have cheap access to big donors, so we can work more on fundraising” vs. “we are all transhumanists, so our national organization will focus on immortality”.)

Evaluation: The ability to do successful local prioritization is likely a prerequisite for such efforts, so an even higher level of competence and resources are probably needed. As there is less experience in this area and the outcomes are more uncertain, the utility of institutional support is less certain. Also, if comparatively advantaged activities are more “direct” than community building, it seems sources of funding such as EA Grants could be more appropriate than community-building funding.

Field-building in priority areas and project incubation

National organizations are probably not the best place to run large direct-work or research projects, but they should work as facilitators and incubators. This can be achieved by a range of techniques:

Furthermore, organizations can support very early-stage projects, in the form of mini-grants, working space, knowledge of global funding sources, etc. Once they are no longer very early-stage, the projects can become more independent and apply for direct funding.

Evaluation: While this area seems promising, there is little experience with projects successfully incubated in this way.

Tax-free routing of donations

While tax-free routing of donations to effective charities is possibly not directly the highest value project that organizations can work on, given the name of the movement and popular expectations, it is useful to do it. This may also provide some additional value for effective altruists who are not considering career changes - it can help them be involved with the community. At the same time, such members can contribute in ways which are different from students or young professionals.

Evaluation: Tax-free routing of donations may be worth doing, but plausibly mostly for indirect reasons like PR and community engagement.

Strategy for creating national organizations

Based on the previously described roles that national organizations may want to play, and the experience of existing organizations, a broad outline of a strategy to create a national organization may be:

  1. Gather the necessary resources, get support
  2. Do localization of effective altruism in the above described sense
  3. Create a medium-term strategy
  4. Create the organization

Further steps, such as how to successfully grow the organization, are important, but not covered in this document. In expectation, the initial strategy seems especially important.

Recommended minimum resources

The following list should be interpreted as a rough guideline, based on experience with existing national and regional organizations.

Personal resources

A dedicated group of approximately 5 people seems sufficient for creating a national-level organization. It is possible to found something with an even smaller team of 2 or 3 people, but in that case, the initial focus should probably be on attracting more co-founders (usually via a network of personal connections).

The current sample is small, but it seems that smaller teams suffer from more significant variance in energy over time; for instance, if people are exhausted after an event and catching up with their other work, it may be hard to utilize newly generated interest or engage newcomers.

Although the group will often be created from members of an existing local or student group, there are important differences, and it is crucial to have some members of the founders team who want to work on this long-term, and want to transition to professional roles, either full-time or significantly part-time. Also, it is important that some people commit to having movement building as a high priority for at least a year or two.

Skills and knowledge in the core group

Core necessary skills probably include several levels of “meta.”

On the higher levels, the founding group must have a deep understanding of effective altruism, and, essentially, the ability to go through the whole effective altruism prioritization framework, taking into account local specifics to reach conclusions valid at their region. This often requires a lot of both formal and intuitive quantitative thinking.

The second necessary part is the ability to plan, create strategies, and set goals.

The third necessary component is operations skills, follow-through, and practical problem solving ability. Experience shows that organizations with poor operations will struggle. Additionally, organizations which are unprofessional or ineffective are likely to damage the credibility of effective altruism as a brand.

Connections toward the EA movement

Connections to the “network core” are crucial: that is, well-connected people working in the central organizations aligned with effective altruism (such as CEA, LEAN, FHI, Open Phil, CFAR, GiveWell, Rethink Charity etc.). In virtually all cases of existing national organizations, this resource, or its lack, made a difference in the trajectory of the organization. Good opportunities to create such connections on a personal level include internships and shared work on common projects.  

Also important are connections to more developed groups of a similar kind. Often such groups have more “hands-on” experience with the necessary tasks inherent in group formation.

"Network of local connections"

As a large part of effective altruism is in some way related to research, it seems useful for a national effective altruism organization to have a good understanding of and some access to local academia. The most useful areas tend to be (in no particular order) economics, philosophy, mathematics, physics, computer science, public policy, cognitive science, and international development.

Other important resources are connections to the groups broadly aligned with EA priorities. This may include the local rationality/LessWrong community if one exists. Some other groups that may be helpful are animal advocacy and vegan groups, transhumanists and other far-futurists, Machine learning and AI experts, and local hackerspaces.

Localization of effective altruism

A general model for localizing effective altruism was described previously, in the section of this document about roles for national level orgs. At the localization stage, the group should already have strong links to central organizations, and should be able to get support from them, both in terms of funding and consultations.

Creating a strategy

Founding effective altruism organizations is complex, and creating a strategy up front should be a high priority (if less than ~20% of initial efforts are spent on this, it is probably too little). Ideally, the strategy should receive feedback both from central organizations and organizations which are both successful and similar to the one being created.

Creating the organization

Founding a legal body is sometimes the easiest part of the process, depending on legal system. At present, there are not enough data to really understand what internal structures of such organizations work the best, so the following are mostly personal best guesses.

Legal structure

The typical legal structure of a national organization would be a non-profit society and association, with members and elected representatives. It is highly recommended to get a lawyer to check the bylaws.

A reasonable internal structure seems to be an elected board with 4-8 members, with two directors in the lead (chair and deputy chair, etc.). As a default voting system, we recommend approval voting.

Setting up internal processes

Generally, there is much know-how about organizing small non-profits outside the EA movement, and effective altruism does not have a comparative advantage in creating such materials. It is advisable that the leaders of a prospective organization educate themselves in this area and/or take some training to avoid reinventing the wheel.

Common challenges involve motivating volunteers, leading teams created from a mixture of volunteer and paid staff, and creating robust feedback loops. As effective altruism itself is a rapidly developing area, it generally seems better to keep as much of small team/startup/hacker culture as possible and avoid formalization when not necessary. “Agile” management methods can be more applicable than more traditional “waterfall” project management.

Setting up the technical infrastructure

It seems that most organizations use some combination of Google Suite, Slack and in some cases Asana or Trello.

Coordination

With a growing number of organizations, intentional efforts to keep the network coordinated and aligned become important. Various models of coordination of an organizational network are possible:

The hierarchical model is probably neither feasible nor desirable - central organizations such as CEA do not have the capacity to collect all relevant information, distill it, and provide detailed guidance.

The fully distributed model probably either requires too much overhead for all organizations or leads to only loose coordination.

What seems to be the most appropriate compromise is the model of a dedicated coordination facilitator. That is, someone, or some specific organization, facilitates coordination. This seem a role which can be filled by central organizations, or also, for example, by regional organizations on a rotating basis.

The most important tools for coordination of an organizational network (besides shared goals and values) are:

Based on research on organizational cooperation in social movements, personal overlaps and connection are very important tools in fostering cooperative behaviour. Hence, it is important to promote bi-directional flows of people.

Funding is a powerful factor which can lead to alignment or misalignment. In particular, if the organizations are forced to do their own fundraising, they will be pushed toward activities which are easier to present to donors and have a shorter inferential distance (for example, raising for GiveWell-style effective charities, or NGO impact evaluation), in contrast to activities which are harder to understand (x-risk mitigation, long-term future considerations). In addition, the need to fundraise incentivizes doing outreach, being more public, and optimizing the message for the growth of the supporter base.

If the organizations can rely on stable, predictable funding from central sources, they can direct fundraising effort to fundraising for effective charities or EA Funds. Such funding will also create strong incentives for cooperation. If local groups want to hire people to help them spread EA locally, or want to help others change their careers in other effective ways, they are much more likely to be successful if they can guarantee stable funding for at least a year. If they can’t, they would be exposing any potential employees or partners to substantial, and potentially irresponsible, risk. It is worth bearing in mind that centralised funding comes with its own risks. If poorly done it might reinforce a specific set of views, create groupthink, push groups in a harmful direction, reduce opportunities for learning. etc

Shared communication channels are also straightforwardly useful. Ideally, local organizations should feel like they are part of the debate on topics where they have expertise.

Further discussion

The suggestions presented here certainly should not be understood as some final answer. I hope more insights and alternative viewpoints will get expressed either in discussion here or in posts describing what different organizations do, As the number of national-level organizations grows, we will certainly have more data points, and hopefully better understraning.

I'd like to thank Harri Besceli, Kit Harris, Julia Wise, Pablo Melchor, Justis Mills, Sam Hilton, Nora Aman, Tobias Pulver, Konrad Seifert, Laura Green, Markus Anderljung, Remmelt Ellen, Sebastian Oehm and others for providing valuable feedback on this document, and participants of CEA community building retreat for discussions obout the topic. At the same time I apologize for not being able to answer every comment or incorporate every valuable idea. Justis Mills and Michael Chen helped with copyediting.

Furthermore, I'd like to thank CEA for the opportunity to work on this project, and my awesome colleagues from CZEA for general support.

Paul_Lang @ 2021-12-25T17:05 (+1)

Regarding media: to keep track of media coverage and potentially react accordingly, it seems that https://www.google.com/alerts can be helpful.

frankieaw @ 2019-03-10T00:54 (+1)

Hey Jan - thanks for this piece.

"The hierarchical model is probably neither feasible nor desirable - central organizations such as CEA do not have the capacity to collect all relevant information, distill it, and provide detailed guidance.

The fully distributed model probably either requires too much overhead for all organizations or leads to only loose coordination.

What seems to be the most appropriate compromise is the model of a dedicated coordination facilitator. That is, someone, or some specific organization, facilitates coordination. This seem a role which can be filled by central organizations, or also, for example, by regional organizations on a rotating basis."

^^could you explain what you mean by hierarchical vs distributed?

Jan_Kulveit @ 2019-03-10T11:53 (+3)

Here by hierarchy I mean strictly tree-like flow of information, where the centre collects the inputs, decides, and sends commands. By fully distributed I mean "everybody talking to everybody" (fully connected network) or a random network (random pairs of orgs talking to each other).