Hide table of contents

This post aims to explain the work the community health team at the Centre for Effective Altruism does about particular kinds of community problems. 

The team does several kinds of work aimed at supporting the EA community and reducing risks to EA’s ability to have positive impact. We spend most of our time on those other kinds of work, but this post only focuses on work on interpersonal harm. We think this is likely the part of our work people have the most questions and confusion about, so we wanted to share more about it.

In short, we try to reduce risk of harm to members of the community while being fair to people who are accused of wrongdoing. That’s a tricky balance, particularly when the need for confidentiality limits our ability to speak to everyone involved, and we sometimes get parts wrong. This post describes both some general principles and a year’s worth of specific examples.

I’m not writing this now because anything in particular is going on. My goal here is to provide some transparency about how these things work in general, rather than commenting on any particular current situation.

What kind of situation is this post about?

It’s about actions people sometimes take that cause harm to others in the community, for example:

  • Someone pushes past another person’s boundaries. This ranges from accidental discomfort, to sexual harassment, to deliberate sexual assault.
  • Needlessly harsh or mean behavior.
  • Erratic behavior that causes disruption or harm for others
  • Deception / dishonesty

Internally, we call this “risky actor” work. Concrete examples are below.

Responses the community health team might make

  • no action
  • talking with the person about how to improve their behavior
  • restricting them from CEA events
  • Informing other EA groups / projects / organizations about the problem
  • (very rarely) publicly warning others about the problem

Often it’s very unclear what the best response would be, and people will disagree about whether we handled something well.

Who works on these situations?
The community health team is Nicole Ross (manager), Catherine Low, Chana Messinger, Eve McCormick, and me. I’ve done this kind of work at CEA since 2016. Currently Catherine and I are the main people on the team handling these kinds of situations. More background.

Other people including group organizers also end up handling such situations.

Difficult trade-offs

Balances where I think both sides are valuable, and I’m not sure if we’ve got the balance right:

Avoid false negatives: take action if there’s reason to think someone is causing problemsAvoid false positives: don’t unfairly harm someone’s reputation / ability to participate in EA 
Keep the community health team’s scope within what we can realistically handle; don’t take on too muchDon’t stand idly by while people do harm in the community
Encourage the sharing of research and other work, even if the people producing it have done bad stuff personallyDon’t let people use EA to gain social status that they’ll use to do more bad stuff
Take the talent bottleneck seriously; don’t hamper hiring / projects too muchTake culture seriously; don’t create a culture where people can predictably get away with bad stuff if they’re also producing impact
Try to improve the gender balance / not make it worse; take strong action on behavior that makes women uncomfortable in EA spacesDon’t crack down too much on spontaneity / dating / socializing; don’t make men feel that a slip-up or distorted accusation will ruin their life
Let people know we take action against bad behavior and we care about thisDon’t create the impression that EA spaces are fully screened and safe - that’s not the case
Give people a second or third chance; adjust when people have changed and improved*Don’t try to be a rehabilitation space - that’s not a good use of the EA community

 *Added February 2023: I wasn't very clear here. I mean situations like “someone has made inappropriate comments and gotten feedback about it,” not something like assault.

Ways of handling evidence

When we have permission to discuss the situation openly

These are cases when Person A says “I had a bad experience with Person B.” I
discuss with A how the problem might be handled, and I ask if I can talk with B about it. If A says yes, I contact B.

I talk with B, usually along the lines of “Here’s the account I’ve heard. Do you have more info you want to add, or things you think I should know?”

Typically the two people agree about the basics of what happened, but often emphasize different elements of it that cast what happened in a different light. In many cases there’s disagreement about how to interpret the situation. 

Sometimes the two people then communicate with each other, either directly, via messages relayed by me, or in a meeting with me or on their own. In some cases, discussion leads to something like a resolution to the problem (perhaps with both people resolving misunderstandings, and often with apologies on both sides). In other cases they continue to see the situation very differently. 

I might come away feeling I understand the facts pretty well, or I might remain pretty uncertain about what really happened.

When we don’t have permission to discuss the situation

But in some cases, the person who raises the concern says “No, don’t talk directly to B — I don’t want them to know I spoke to anyone about this.”

I’m going to give a composite example based on several real cases. Let’s call him Steve. Steve’s ex-partner said that during an argument he acted in a way she found intimidating. Later an EA from an area where Steve used to live said that he had a bad reputation in multiple EA groups there because of his pushy behavior. One of those local people said Steve pressured them to let him stay at their house when they wanted him to leave. None of the people I heard from wanted to be identified to Steve. 

When Steve later applied to attend EA Global, I told him he could not come to CEA events because of multiple concerns I’d heard about him. Understandably, he asked for more information. I told him I was sorry I couldn’t share more, because I didn’t have permission from the people who had spoken up.

There are conflicting requests here: the people who have raised concerns about Steve don’t trust him, don’t want to be retaliated against, and don’t want to get involved in an ugly public debate about him. But Steve understandably wants to hear what the concerns about him are. And if Steve tells people wasn’t allowed at EA Global, they will naturally wonder what’s happening and whether the process was fair (especially if he says that he’s done nothing wrong and that he has no idea what this could be about).

A court of law wouldn’t make any kind of judgment without openly discussing the evidence and allowing Steve a chance to argue his case. I’ve heard people say that similarly, a community shouldn’t bar someone from spaces or events without discussing the accusations with the accused. Ideally, I would also like that to be the process.

But what should we do when we don’t have permission to discuss the evidence? Communities have always needed ways to handle these in-between situations where there hasn’t been an open discussion of the facts.

Sometimes this in-between system will be wrong. If I’ve only heard a concern about someone from one person, that one account could be mistaken, distorted, or outright false. Even in situations with multiple reports against the same person, the picture still might not be clear or accurate. If the accused had a chance to respond to the things I’ve heard, maybe I’d come away convinced they’re right and the problems didn’t happen or weren’t that serious. In some cases, my understanding of what happened changes a lot after hearing from the person accused of causing harm.

But when I hear an accusation and don’t have permission to talk to the accused, I have to make some call about whether to do anything. I’m weighing the possible harm to the accused if the accusation is inaccurate against the possible harm to other people in the community if the concern is accurate.

Often the best compromise I can come up with looks like:

  • Keep them out of CEA spaces like EA Global and EAGx conferences
  • In some cases, talk to their local group organizer about the type of concern I’ve heard
  • Recommend against them for some types of roles, like community organizing (though usually not for less people-facing roles like writing or research.)

These responses will be unfair if the concerns are false or overblown. Any kind of restriction causes harm to someone’s ability to network, and a private recommendation against someone for a role may still damage their reputation.  But making these restrictions as privately as possible seems to avoid the worst harms to their reputation if the concerns are unfounded, while also avoiding the worst harms to the community if the concerns are valid.

Why so little action in some cases?

In some cases, the community health team doesn’t take action because it’s not very serious, there doesn’t seem to be any good step to take, or it’s not within an area where CEA should get involved. There are also cases where I find a report to be alarming and would like to take action, but the person reporting does not want it known that they spoke up. In these cases, sometimes there’s very little that can be done without breaking their confidentiality.

In some institutions like universities, there's a policy that confidentiality can be broken for the good of the community in cases like this. That’s not my policy. Breaking the confidentiality of someone who reports a problem may allow you to take action on a problem now. But it means that person, and others they talk to, will be much less willing to tell you about problems in the future. Overall I expect that breaking confidentiality in those situations is worse for the community in the long run. I never want someone to regret talking to me about a problem. (I take confidentiality really seriously, but I have made occasional mistakes. I’ve described those here.)

I think it can be helpful for the community when people are willing to openly state problems, or let me be open about a problem I’ve learned about. But I understand it can be costly for those people and that it may not feel safe or worthwhile.

In these cases, I try to provide supportive listening, discuss options with the person reporting the problem, and let them know that I am ready to consider more steps if they choose to let me speak more openly about the problem. If there are any steps I’m planning to take such as keeping the person out of EA Global, we discuss those. Later, if I learn more about the situation I might circle back to the person who told me about, and we can discuss if the plan should change.

Once I know about a problem, I can keep an ear out for other similar problems. Sometimes incidents that seem minor at first are more concerning if the person repeats them, and the people reporting the problems might be ok with more action at that point. So if you’re on the fence about whether to tell me about a minor problem, I encourage you to go ahead and let me know.

How does EA compare to other communities?

As far as I can tell, not that different. The number of problems I’ve heard about isn’t surprising to me, given the thousands of people involved with EA. Anecdotally, I’ve heard from people who feel EA is better or worse than other communities they’re familiar with.

CEA has opted to put staff time toward addressing problems that come up, because we think it’s better to try to address them than to leave them unaddressed or to leave problems for whoever encounters them to handle without help. 

Some other parallel efforts can be found in religious denominations, LARPing, and open source programming communities.

Scope of this work beyond the community health team

In a growing community with thousands of people, one person or one organization can’t handle all the situations that come up. There are other options:

Group organizers also handle these situations, both in taking steps to prevent problems and in reacting once a problem has happened. If you notice a problem in your group, a group organizer may be a good first person to turn to.

If you’re a group organizer, Catherine Low specializes in advising EA groups about difficult interpersonal situations. The EA Hub also has written resources. CEA doesn’t dictate your choices as an organizer: we can swap information with you, and Catherine or I will give our thoughts on what we’d do in that situation, but you may well decide on a different approach.

Sometimes resources outside EA are more appropriate. If a crime has happened, we think the victim should consider going to the police (although we understand why victims sometimes don’t find it worthwhile to go through the legal system). If they want to try a restorative justice approach, we don’t have the training or capacity to run that, but there are other groups that may be able to help.

If you're a group organizer or work at an EA organization or project and have a situation where you think services like a professional mediator, HR advisor, or lawyer would be helpful, CEA might be able to help pay for that. Feel free to ask.

Last thoughts

I’ve been heartened to see people looking for ways to help each other. I’ve seen people do their best to seek the truth in messy situations with no clear answers. I’ve seen men refuse to be part of a “boys’ club” (for example, talking to another man about how his behavior was unpleasant for women in the local EA group). I’ve seen people receive feedback about how their actions were causing problems for others, and do their best to remedy the harm and change how they act. 

I also see an important role for those of us who have been around EA a long time, who work at EA organizations, who are major donors, or who otherwise hold some position of status or power. We can keep an eye out for people who are newer, less established, or vulnerable in some way. Sometimes we can be their best supports. (See also this post about power dynamics in EA.)

In some cases, problems still go unrecognized or unaddressed. I don’t want people to assume they can trust everyone in EA without reservation, or that people have been thoroughly pre-screened before participating in EA groups or events. Please still use your own judgment and pay attention to your instincts if something seems wrong.


If you want to talk about a problem you’ve experienced or noticed in the community, you can reach me at julia.wise@centreforeffectivealtruism.org or schedule a call.
If you want to talk about a mistake I’ve made or a problem with my work, you can talk to me, my manager Nicole Ross, or CEA’s executive director Max Dalton. You can also leave CEA a message anonymously at our contact form, which goes to operations staff. 

Appendix: 12 months of cases

In order to give a picture of what kind of situations come up and how we handle them, here are anonymized versions of all 19 of the cases of this kind the community health team handled in a twelve-month period (roughly 2021). I’m not including other types of work the team does like mental health support, trying to improve EA culture in other ways, etc.

I’m providing this partly so that people who are concerned about people being banned from events, etc, can get a sense of how often this is happening. I think this time period is pretty typical. They’re ordered roughly from “cases where stronger action was taken” to “cases where little or no action was taken.”

It’s hard to write summaries that both give key information and are anonymized enough, and I can imagine that some of these have elements that may be puzzling or worrying to readers. Feel free to write to me or my manager if you want to ask for more details about how something was handled and why, and we’ll share more info with you if we can.

It might be helpful to skim this or read a few entries, but I doubt most people will find it useful to read the whole list. 

  • A person applied to EA Global who had previously been reported for deliberately physically endangering another community member, sending them threatening messages, and more. Written correspondence between the people appears to confirm this. I discussed with both of them separately, and told the person they cannot come to CEA events including EA Global.
  • Multiple reports of concerning behavior from a community member, with multiple other community members unwilling to attend events where this person will be present. I told the person they cannot come to at least the next CEA conference, with future events TBD.
  • A person with many past accusations of bad behavior against them posted on an EA Facebook group offering to host couchsurfers. I discussed with group moderators, removed the comment, and told the person they cannot post this kind of offer on the group. Later after a group member complained about an interaction with this person, the majority of the moderators (including me) voted to remove this person from the group.
  • A participant at a program run by an EA organization reported discomfort about past romantic/sexual interactions they’d had with a staff member at the org. These interactions had happened years previously, before they were at the same org. At their request, I spoke with the staff member and with others at the org who are better placed to monitor the situation going forward. The two people discussed what happened in the past, the staff member apologized to the participant, and they both said they felt relieved to have finally dealt with it. We also discussed the way these interactions can read very differently based on power differences between people.
  • A person was hosting an event with someone who appeared to be misrepresenting their accomplishments in the subject the event was about. I wrote to the host telling them my understanding of the person’s actual level of accomplishment.
  • An attendee at a virtual conference did not want to interact with another attendee they’d had a past bad experience with, but did not want to limit the other person from attending the event. I explained the options for avoiding contact with the other person on the virtual platform. I’ll be on the lookout for any further problems from the person.
  • A grantmaking body told me they had heard reports of misconduct by someone in their field, but they knew only a very vague summary of what had happened. I tracked down the source of the concern, which turned out to be more minor than the vague summary had made it sound. With permission, I informed the grantmakers of the concrete nature of the concern. The person who originally had the concern asked me to be present on a call between them and the person the concern was about; the three of us had a call and both of them explained their perspective to each other.
  • A community member had a bad experience during a meeting with someone in their local group. They did not want any action taken against the other person, but a group organizer and I made plans for how to avoid the type of situation that led to this.
  • A grantmaking body was considering a grant to a project including a person who’d previously written a public request to be held accountable after a mental health episode that included significant lying. I spoke with the person and later with their manager about their plan for if the person encountered similar problems in the future; we were all satisfied with the plan. I told the grantmaking body that my concerns were resolved about this situation.
  • We advised a funder about a grant that may have posed a reputational risk.
  • A community member reported being sexually assaulted by another community member years in the past. They were already in a restorative justice program with the person and did not want the other person removed from the EA community or publicly identified. I met with the person who had committed the assault, and we discussed their view of the situation and what has changed since then in their views and behavior. I told them that any further boundary violations would mean I’d warn others about them. Note that I don’t have permission from the victim/survivor to do anything more here.
  • The team advised a group about a local person with many reports about them making people uncomfortable at events. The person currently isn’t attending, but the group now has steps they plan to take if the person expresses interest in attending.
  • Group organizers asked for advice on whether and how to re-include a past group member who had committed a serious crime during a period of mental illness. I gave feedback on their plan and got more information about what warning signs had been present before the incident happened.
  • I learned about a situation between two community members years before where an interaction that might have been fine in other circumstances was very uncomfortable for one of them because of the power/status difference between them, and where an attempt at mediation at the time hadn’t gone well. I met with the one who had caused this discomfort, and we discussed it and what they would do differently now. They passed on an apology to the other person via me.
  • A group organizer asked for advice after a member reported unwanted touching by another group member. The person who reported this did not want the other person to know they had reported it. I advised the organizer on the limited steps I saw as possible given the confidentiality limitation.
  • A case involving a physical struggle. The person who reported this case did not wish any more information to be published. We suggested steps to increase safety for people who might be affected.
  • Someone wrote publicly that an EA community member had been abusive to them. After I spoke with them both, it appeared the behavior in question was not abusive, and if anything the accuser was the one more in the wrong. Since the accuser was not involved in EA as far as I could tell, I took no action.
  • Someone made statements that an EA community member had been abusive. After speaking with the accuser and someone else familiar with the situation, it seemed that the behavior was bad relationship choices but not what I consider abuse. I took no action.
  • A community member told me proactively about some unpleasant (but not clearly abusive) behavior they had done to a former partner. I took no action.
  • A community member asked for help in publicly documenting various bad things people outside EA have done to them. I explained I don’t have capacity to help with this.
Sorted by Click to highlight new comments since:

I appreciate the section on tradeoffs, and I think it makes me more likely to trust the community health team.

Thank you for the work you and your team do, Julia. Many of these situations are incredibly tricky to handle, and I’m very grateful the EA community has people working on them.

I found this post in the wake of the recent Nonlinear situation and my own response to it. I think it's worth stating at least briefly in light of all of that that it seems extremely unwise to me to take serious adverse actions against anyone, or spread rumors from any sort of official position about them, if they do not have the opportunity to hear and understand the allegations against them. There is little more unpleasant than having the people around you suddenly treat you as suspect or evil without any apparent reason, particularly if every attempt to learn the reason is rebuffed.

"EA is not the legal system" isn't a good response here: the legal system has evolved due process for good reason, and allowing a norm of confidential-from-accused allegations that lead to concrete action will almost necessarily replicate every harm that led to the evolution of due process in the legal system, given enough time. I recognize the complications of victims fearing retribution, but everyone in adversarial situations fears retribution, whether they are a victim, a false accuser, or (as is perhaps most common) someone interpreting an ambiguous situation in a negative light. I think the policy you outline dramatically downplays the harm of restrictive actions taken based on partial, one-sided views of adversarial situations; if confidential-to-the-accused accusations should be acted on at all, they should be acted on with immense restraint and caution.

I want to add this regarding confidentiality. This is a quote from an email from Julia Wise:

People do ask me for my impressions of people they're considering funding, so if someone does want to give their unvarnished thoughts/complaints to someone at CEA, it's hard for that to stay separate from what funders hear. 

This is from May 2021, so this may have changed.

[edited to add: I'm second-guessing myself and have edited a bit because I don't remember ever actually doing this. I think I should develop a clearer policy here.]

I'd like to provide a picture of how this might play out. (To be clear, I'm making up pretend examples, not referring to Linda specifically.)

If you share specifics with me and want me to keep them specific, I will. So if you tell me in confidence that you're struggling with addiction or  have messed up your job or whatever, I won't share those facts with anyone.

But if a funder asked for my opinion about funding your work, I'd probably give them some general indication of how excited (or unexcited) I was  about your fit for that work.

I'm not sure this has ever actually happened based on material the person themselves told me, rather than my own observations and things other people told me about them.

The fact that talking to you can affect funding decisions is bad.  

You don't seem to understand how important funding decisions are to community members, which is baffling given your role. Or you do understand and that's why this information is not public, which is deceptive, and also very bad.

Reporting interpersonal conflicts almost always makes everyone involved look bad, at least a bit. I don't feel safe talking to someone who is also an evaluator. This is rely basic! 

It might be helpful, in this or another post, to talks bout what kind of support you can / can't offer to EA-aligned or adjacent organizations dealing with interpersonal issues in the workplace, since many of the examples here are related to groups.

(Also this is very helpful and I think will be a useful resource I expect to share with group organizers and community members in the future)

Hi Julia, this is a really valuable post. I am curious about what you and the team consider the scope of your responsibilities. Do you feel responsible for all the parts of the EA universe? Or on the other extreme are you only focused on groups and events that are directly affiliated to CEA, either through people or funding? Are there any events or groups that might be considered very EA or EA-aligned where you would say, 'We are not the right community health team to work on this?'

I have no specific cases or issues that I am alluding to, I am just curious :) 

Mostly I suggest that people ask us if a situation comes up where you think we might be able to help, and we'll let you know!

To some extent it depends on capacity at the time. Sometimes we've had capacity to help with situations that were EA-adjacent but not obviously within the EA space. Other times a situation is clearly within EA but we think the services needed are better served by a professional mediator, lawyer, etc and we'd refer to someone with those skills. CEA can sometimes help EA groups or organizations pay for those services if it would help resolve a problem - that's also something you should feel free to ask about. This is a good question - I've added a bit about this to the post.

Thanks Julia; this is a really insightful post. I will make sure to use it if anyone in the EA community asks me questions related to community health/the process for complaints in the future. 

One of the things I'm curious about is how you see the balance of these trade-offs: 

Encourage the sharing of research and other work, even if the people producing it have done bad stuff personallyDon’t let people use EA to gain social status that they’ll use to do more bad stuff
Take the talent bottleneck seriously; don’t hamper hiring / projects too muchTake culture seriously; don’t create a culture where people can predictably get away with bad stuff if they’re also producing impact

It feels like CEA's default is to be overly cautious and tread lightly in situations where someone is accused of bad behaviour. (Ie. if 'cautious action' vs 'rash action' is a metric here, I would think that CEA would sit considerably more on the cautious side.) This is quite understandable, but I wonder how you think about the risk of being too slow to condemn certain behaviours? 

For example, I could imagine situations where something bad happens, and both the accuser and the accused contribute valuable work to the community. However, due to CEA's response leaning towards the side of caution, the accuser walks away feeling like their complaint hasn't been taken seriously enough/that CEA should have been quicker to act, and possibly feels less inclined to be involved in EA in the future. Do you feel like this has happened and, if so, how do you think about these types of situations? 

due to CEA's response leaning towards the side of caution, the accuser walks away feeling like their complaint hasn't been taken seriously enough/that CEA should have been quicker to act

I'm sure this has happened, and I'm sad about that.

I also know different people who would say that CEA has been too aggressive in kicking people out, too willing to take action based on limited evidence.

I want to weigh the fact that people will feel alienated by both of these perceptions/experiences. But ultimately we can't make decisions based only on whether someone will feel disillusioned with EA as a result, because it often seems any call we make will result in someone feeling we were too far in some direction.

Thank you for this. I've long been looking for a link to explain the role of the Community Health team, and sometimes I've hesitated to tell people about it because I can't easily explain (and am not entirely sure myself) what it does. This will be my go-to [1], unless you suggest something else.

[1] For people who might benefit from getting in touch about an issue, and for those who are curious about the EA ecosystem.

Sure, I'm happy for you to link people to this if you think it would be helpful. This is part of our work and we'll be writing more over time about the rest of it.

"The team advised a group about a local person with many reports about them making people uncomfortable at events. The person currently isn’t attending, but the group now has steps they plan to take if the person expresses interest in attending."

Who else recently stopped attending their local group and is now paranoid this is about them?

In general if you think CEA or your local group is worried about you, I think you should feel free to ask.
In this case, the person is well aware of past complaints about them in other settings.

[comment deleted]0
Curated and popular this week
Relevant opportunities