How To Respond To Code of Conduct Reports by Valerie Aurora and Mary Gardiner

License and attribution
Written by Valerie Aurora, based on a short guide written by Mary Gardiner
Edited by Annalee Flower Horne https://www.flowerhorne.com/
Copyright © 2018 Valerie Aurora https://valerieaurora.org/
Copyright © 2012 Mary Gardiner https://mary.gardiner.id.au/
Cover image © 2018 Mary Gardiner https://mary.gardiner.id.au/
CC BYSA 4.0 Valerie Aurora, Mary Gardiner, Annalee Flower Horne, DjangoCon EU, Write the
Docs EU 2016, PyGotham 2017 https://creativecommons.org/licenses/bysa/4.0/
Published by Frame Shift Consulting LLC https://frameshiftconsulting.com

Version history
20180915: Draft release for review
20181126: Prerelease version for review
20181128: First edition

Table of contents
Introduction
If you are in a hurry
How to use this guide
About the authors
Terminology
Chapter 1: Code of conduct theory
Purpose of a code of conduct
What a code of conduct should contain
How a code of conduct works
Education
Normfollowing
Attraction and repulsion
Deterrence
Boundary setting
The Paradox of Tolerance
What a code of conduct can't do
Codes of conduct govern community spaces
Violations must have meaningful consequences
Codes of conduct must apply to powerful people
Visible enforcement is required
Summary
Chapter 2: Preparing to enforce a code of conduct
Publicizing the code of conduct
Identifying community members
The code of conduct committee
Choosing code of conduct committee members
Communicating with each other and the public
Choosing a decision method
Adopting an incident response guide
Recordkeeping
Training the committee
Training reporttakers
Avoiding or mitigating higherrisk activities
Make arrangements for legal advice
Updating code of conduct materials
Summary

Chapter 3: Responding to a report
Start the response deadline clock
Check to see if everyone is safe
Write down the report if necessary
Make a preliminary announcement if appropriate
Ask for recusals
Organize a committee meeting
Do additional research
Meet as a committee
Choose a response
Take any actions necessary to implement the response
Inform the target and harasser of the response
Communicate the response to others
Respond to criticism
Summary
Chapter 4: Discussion
What does not belong in a code of conduct
List of unacceptable behaviors
Transformative justice and codes of conduct
Recusing committee members
When individual safety conflicts with community safety
Protecting the community's reputation
Safety is more important than privacy and confidentiality
Responses not to use
Do not ask for apologies or forgiveness
Do not ask the target to decide the response
Do not mediate
Do not guard the harasser or the victim
Do not ask the harasser to stay away from the target
Holding powerful people accountable
Putting legal concerns into context
Responding to incomplete or late reports
Investigating the incident
Impact is more important than intent
Distinguishing good intent from bad intent
DARVO: Deny, Attack, Reverse Victim and Offender
Judging competing claims of marginalization
Social awkwardness and harassment

Mental health and harassment
Children, caregivers, and harassment
Sexual behavior and communities
Alcohol and drugs
Choosing a proportional response
If a harasser refuses to follow the code of conduct
Responding to protest from the alleged harasser
Communicating the response to others
Responding to criticism
Dealing with attacks on the committee or community
Summary
Chapter 5: Examples of responding to reports
Wikimania 2012 sexualized presentation
DjangoCon EU 2017 transparency report
Denial of validity of code of conduct
Harmful question during a talk
Sexist comment on clothing
Photographer creates awkward situation
Write the Docs EU 2016 transparency report
Attendee uses derogatory term
Inappropriate joke in talk
PyGotham 2017 transparency report
Selfreport of an ambiguous joke
Attendee denies making offcolor joke
Volunteer overwhelmed by requests
Attendee makes unwelcome advance
Badfaith code of conduct report
Racist comments at a conference
Oppressive comments in online chat
Anonymized conference transparency report
Attendee invites women to hotel room under pretext
Inappropriate touch reported after conference ended
Unwanted sexual advance
Inappropriate touch
Inappropriate pulling on clothing
Drupal community incident
Background
Precipitating incident

Response
Analysis
Conclusion
Summary
Acknowledgements
Appendix 1: Additional resources
Appendix 2: Reporttaking form

Introduction
Enforcing a code of conduct is difficult without the right training and knowledge. Most people
enforcing a code of conduct for the first time make mistakes, and sometimes those mistakes
have major consequences for their community. Unfortunately, few communities have people
who have experience enforcing a code of conduct, and only a few communities can afford
professional code of conduct training or consultants. As code of conduct experts with practical
experience in responding to code of conduct reports, we wrote this guide to enforcing codes of
conduct so that every community can have access to the current best practices in handling code
of conduct reports.

This guide is not legal advice. If you need legal advice, please contact a qualified lawyer.
If you are in a hurry
If you are reading this guide after you have already received a code of conduct report and need
to respond quickly, we recommend reading these parts of the guide:

● Introduction
● Chapter 1: Code of conduct theory
● Chapter 3: Responding to a report

Then follow the links from within those sections to go into more depth in the areas that are
relevant to your current situation. We strongly recommend scheduling time to read through this
guide in its entirety at a later time.
How to use this guide
We recommend that everyone involved in enforcing the code of conduct read this guide from
beginning to end, including this introduction. This guide is organized into the following topics:

● Chapter 1: Code of conduct theory
● Chapter 2: Preparing to enforce a code of conduct
● Chapter 3: Responding to a report
● Chapter 4: Discussion
● Chapter 5: Examples of responding to reports

Each chapter is divided into sections, roughly in chronological order of when you will use the
information in each section. At the end of each section, we include a list of related sections in
this guide. At the end of each chapter, we include a brief summary of what you should have

learned from that chapter and a list of outside resources referenced in the chapter. All of these
outside resources are collected in Appendix 1 for easy reference.

Chapter 3: " Responding to a report " is designed to be usable as a standalone guide to
responding to reports of code of conduct violations in your community. Appendix 2 is a onepage
guide to taking code of conduct reports for sharing with people in your community who are likely
to take reports.

Everyone in community leadership should read at least this introduction and Chapter 1: " Code
of conduct theory " so that community leadership and the code of conduct committee have a
common set of terminology and principles to use when working together.

This guide assumes your community has already adopted a code of conduct and has a
governance structure that allows it to be enforced. If you want help with writing and adopting a
code of conduct, see Appendix 1 for additional resources.

If your community can afford it, we highly recommend formal code of conduct training and
engaging expert code of conduct consultants, which you can find at:

https://frameshiftconsulting.com/codeofconducttraining/
About the authors
This guide is written by Valerie Aurora, based on a short guide written by Mary Gardiner, with
editing and other contributions from Annalee Flower Horne. It includes quotes by permission
from the conference transparency reports written by DjangoCon EU 2017, Write the Docs EU
2016, and PyGotham 2017.

In 2010, Aurora and Gardiner were the lead authors of the Ada Initiative antiharassment policy ,
which became the basis of thousands of codes of conduct in use today. For more than four
years, they directly handled or gave advice on handling code of conduct reports at conferences
or in online forums.

Aurora has consulted for several organizations on codes of conduct as part of her work at
Frame Shift Consulting , and taught and wrote a code of conduct training based on more than 7
years of professional experience writing and implementing codes of conduct. Flower Horne
wrote and contributed to several codes of conduct for both volunteer and professional
communities, and offers code of conduct consulting in multiple areas, including writing and
adopting codes of conduct and responding to code of conduct reports.

This guide represents the collected knowledge and experience of these Aurora, Gardiner, and
Flower Horne and the many people who advised us along the way. Any comments using "I" or

"me" are from Aurora personally; those using "we" or "us" refer to Aurora, Gardiner, and Flower
Horne. All errors and mistakes are the responsibility of the lead author, Valerie Aurora.
Terminology
This section defines the terms we use in this guide. Our definitions are specific to this
document, and may differ from how these terms are used in other contexts.

We'll use many different terms to refer to the person who has been reported for breaking the
code of conduct, depending on what seems most appropriate to the situation: harasser ,
perpetrator , serial predator , violator , offender , or alleged harasser . We call the person or
people harmed targets , victims , or survivors . The current best practice on how to refer to a
person targeted by oppression or assault is to ask the person themselves how they would prefer
to be described . As we are most often talking about theoretical situations, we use the term that
seems most appropriate to us in each case.

We refer to the group of people governed by the code of conduct as a community .
Communities come in many different shapes, such as attendees of a conference, volunteers for
a nonprofit, members of an online community forum, or a collaborative project that mixes
volunteers and paid employees. Community spaces are spaces each community controls or
governs, such as its conferences, meetings, workplaces, buildings, mailing lists, publications,
and online forums.

For the purposes of this document, "community" only includes communities in which
membership is optional and voluntary, as this guide relies on the ability to exclude someone
from a community if they are a threat to others in the community. If excluding someone from a
community results in a direct loss of a person's legal rights as provided by the state or a
direct, severe, immediate threat to their life, then that community is not included in our
definition of community. For example, this guide won't work for most families with young
children as most young children don't have a safe alternative to living in their current family
situation.

We do include communities in which excluding someone from the community might cause
them to lose friendships, professional opportunities, business relationships, income, a forum
for their speech, the respect of their peers, access to sexual partners, emotional support,
social contact, or similar things. For example, claiming that membership in a book club is
essential to someone's mental health does not require that community to include them (see
the section on " Mental health and harassment " for more detail).

Managing communities that consist of the employees of a company depends on many legal
considerations that differ based on jurisdiction. This guide can be used to help inform decisions
in an employment context, but does not address many aspects of employmentrelated decisions

and is not legal advice in any situation. In cases of employees violating a code of conduct, we
recommend consulting with an employment lawyer and any employment experts at your
company (e.g. human resources) prior to taking any action. As a note, we encourage people
making decisions in an employment context to prioritize the welfare of the alleged target(s) of
harassment over that of the alleged harasser.


Privilege refers to an unearned or partiallyearned advantage society gives only to members of
a particular social group. Oppression is the converse of privilege: systemic, pervasive
inequality present throughout society that benefits people with more privilege and harms those
with less. Some examples of social groups which enjoy some privilege include the dominant
ethnic group in your society, those who are (currently) ablebodied, people who are the same
gender as the gender assigned to them at birth, and men, to name just a few.

This guide uses a number of social justice terms which may not be familiar to everyone.
Wikipedia often has good definitions for these terms. We also recommend referring to the
handout that accompanies the Ally Skills Workshop for a short summary of common terminology
as well as resources for more indepth reading.

Chapter 1: Code of conduct theory
In this chapter, we will explain the basic ideas underlying codes of conduct and what makes
them effective. We will describe the purpose of a code of conduct, outline what a code of
conduct should contain, and explain how a code of conduct works.

We strongly recommend that everyone in community leadership take the time to read and
understand the introduction and this chapter. A community's leadership must understand
and agree on how a code of conduct works if they want the code of conduct to be effective.
Unfortunately, how codes of conduct work is not obvious to the average person, and many
people misunderstand key aspects of them. With this background, your community is less likely
to make major, avoidable mistakes when responding to code of conduct reports.
Purpose of a code of conduct
The purpose of a code of conduct is to protect members of a community from harm in
that community's spaces. The people who need the protection of a code of conduct are
usually those with less power or privilege, as more powerful or privileged people are often
already protected from most harm. For example, a straight white cis man who is the leader of an
important computer software project is highly unlikely to be the target of sexual harassment by
anyone participating in that project. It's far more likely that a sexual harasser will target someone
with less privilege: someone who is a woman, a person of color, and/or queer, to name a few
possibilities. When working properly, codes of conduct will most often be invoked to
protect those with less power and privilege from those with more.
What a code of conduct should contain
We assume that you've already chosen a code of conduct, so this section briefly lists what a
code of conduct should contain as context for later discussion. For examples of codes of
conduct and resources for choosing and writing a code of conduct, see the links in Appendix 1 .

A code of conduct should include (in roughly this order):

● Optionally, a short statement describing the goal of the code of conduct
● A list of unacceptable behaviors
● A description of where the code of conduct applies
● A list of potential consequences for violating the code of conduct
● Detailed, specific, simple instructions for reporting a code of conduct violation
● A list of the people who will handle the code of conduct report
● A promise that anyone directly involved in a report will recuse themselves
● Optionally, contact information for emergency services

● Optionally, links to related documents

That's it! Everything else must go into a different document entirely. Including other content in
the code of conduct will weaken your code of conduct and result in more code of conduct
violations.

See also:

● Appendix 1: Additional resources
● What does not belong in a code of conduct
● List of unacceptable behaviors
How a code of conduct works
Codes of conduct create a more inclusive and welcoming community through the following
methods:

● Education: they teach people what behavior is unacceptable
● Normfollowing: most people follow group norms when they know what they are
● Attraction/repulsion: they attract people you want and drive away people you don't want
● Deterrence: they create consequences for unacceptable behavior
● Boundary setting: they keep unsafe people outside the community

We'll describe each of these methods in more detail in the rest of this section.
Education
Codes of conduct teach people what behavior is unacceptable. I was surprised the first time I
heard someone who included pornography in their slides at a computer conference say
something like, "Well, if I'd known other people didn't like that, I wouldn't have done it!" People
are often surprised that whatever they did to violate the code of conduct is unpopular with the
rest of the community. You can see why they might be surprised: for example, sharing
pornography is encouraged, rewarded and perhaps even the primary purpose of many
communities. It's not surprising that someone coming from one of those communities would
assume the same of their new community unless explicitly told otherwise.

People often overestimate the level of shared values they have with other people in their
community, which is why it is helpful to state your community's values explicitly. When a
community writes down and publicizes a specific list of unacceptable behaviors, fewer people
will engage in them, even if they already seem obviously wrong to most of the people in the
community. To make sure this information reaches everyone in the community, we highly
recommend advertising the code of conduct prominently, making it easy to find on the
community's web site and online forums, and announcing and reading it aloud at inperson

meetings. A code of conduct is much more effective when people learn about it before they
have a chance to violate it.
Normfollowing
Codes of conduct work because most people just want to fit in and get along with everyone
else—they want to follow the norms of the rest of the community. They will agree to abide by a
code of conduct, even if they don't agree with everything in the code of conduct, just to avoid
conflict. When people like this do break the code of conduct, it is often through lack of
understanding, forgetfulness, or by accident. This is another reason to advertise and talk about
a code of conduct frequently, especially when it has been recently adopted or updated, or if you
are in a community space where newcomers are common.
Attraction and repulsion
People are often looking for a community where they will fit in. Some people will read a code of
conduct and say, "What, no homophobic jokes? I don't want to hang out here!" and move on.
Others will read the same thing and say, "Hey, no homophobic jokes? That sounds great, sign
me up!" The result is a community with more people who think homophobic jokes are
unacceptable and relatively few people who think otherwise, and fewer homophobic jokes
overall.
Deterrence
A code of conduct deters people from unacceptable behavior by explicitly warning people of the
consequences for engaging in that behavior. This is for the people who know that a behavior is
unacceptable, don't care about following community norms, and join a community even when
they know that they disagree with the community norms. This kind of person may still obey the
code of conduct because they don't want to suffer the consequences for violating the code of
conduct, up to and including expulsion from the community.
Boundary setting
A code of conduct creates a boundary around a community, and pushes or keeps people
outside that boundary if they have harmed or are likely to harm the community. Banning
someone from the community is the only method to protect your community from people who
refuse to agree to the code of conduct or seem likely to violate the code of conduct in the future.
The power of a code of conduct ultimately derives from the willingness of community leadership
to eject people from the community's spaces and prevent them from returning, or refuse to allow
them to join in the first place. Without this willingness, a code of conduct relies solely on an
individual's internal motivations, such as sense of shame, sense of empathy, or desire to fit in.
Some people are not held back from harming others by any internal motivations. Unless these
people are kept out of your community, they will dominate your community and drive out more
considerate people.

The Paradox of Tolerance
When harmful people dominate your community and drive out more considerate people, that's
an illustration of the Paradox of Tolerance , a philosophical concept named by Karl Popper in
1945. The "paradox" is a more of a rule with one exception, and the rule is: a tolerant society
must be tolerant of everything—except intolerance itself . This is because if a tolerant
society allows people to express and practice intolerance, intolerant people will eventually take
over all of the society, and the tolerant society will eventually disappear. This applies to
communities too: if a community tolerates intolerance, it will eventually be taken over by the
intolerant. Enforcing a code of conduct is one important way to protect a tolerant community
from destruction.
What a code of conduct can't do
The purpose of a code of conduct is to protect members of a community from harm in that
community's spaces . Sometimes communities want to use a code of conduct for other
purposes, such as to:

● Punish wrongdoers
● Serve as some form of restorative or transformative justice process
● Reinforce existing power structures
● Protect people from harm outside community spaces
● Mediate arguments between individuals
● Force people to apologize to or forgive each other
● Rehabilitate or provide a path to redemption for offenders
● Substitute for a process of building community consensus around its culture
● Help community members improve their morality, personality, or habits

These are understandable goals and reflect varying theories of justice and punishment that
have been explored throughout history. However, the communities we are talking about in this
document don't have the power, expertise, or time to do anything other than protect members of
their community from harm in community spaces. Attempting to do more than this rarely
accomplishes the desired goal and ends up harming the community more.

As an example, social justiceoriented groups often attempt to respond to code of conduct
violations by using transformative justice concepts, without the right context or training. The
most common end result is that the abuser doesn't change their behavior and the target leaves
the community (see the section on " Transformative justice and codes of conduct " for more).

Protecting a community from harm is often compatible with other goals, such as protecting a
specific person from harm, protecting someone's privacy, or raising sponsorship money for a
conference. However, if these goals conflict with the primary goal of protecting a community

from harm, protecting the community must take precedence. We will talk more about ways to
prioritize conflicting goals in the " Discussion " chapter.

See also:

● Transformative justice and codes of conduct
● When individual safety conflicts with community safety
● Protecting the community's reputation
● Safety is more important than privacy and confidentiality
● Responses not to use
Codes of conduct govern community spaces
A community can only enforce a code of conduct within community spaces. A community does
not have the ability to enforce a code of conduct in spaces it doesn't control, such as a social
media platform open to all, the workplace of another company, or a public park. However, a
community can (and should) respond to events that occur outside of the community's spaces
and take appropriate action to protect the community within community spaces.

For example, if someone writes a racist post on their personal blog, a programming conference
can refuse to allow them to attend or speak. If someone is alleged to be physically abusing their
intimate partner, a community can ban them from attending their meetups or posting on their
mailing lists, even if they are not convicted under the formal legal system. In many jurisdictions,
an employer can fire an employee for behavior that happened outside the workplace.

Violations must have meaningful consequences
A code of conduct is only effective if violating the code of conduct has meaningful
consequences, up to and including expulsion from a community. Some people will continue to
harm the members of that community until they experience some significant consequence for
their actions; others will not stop their actions in response to any consequence the community
can impose and must be removed from the community entirely. The people enforcing the code
of conduct must have the support of community leadership and be confident they won't be
overruled unless they have made a significant mistake. The community must have the ability to
exclude people from their community spaces.
Codes of conduct must apply to powerful people
People who violate the code of conduct are likely to have more power and privilege, since they
are used to doing what they like without consequences for harming others. As a result, it is quite

common for powerful people in your community to violate the code of conduct. Your
community must not adopt a code of conduct if it will not apply to the most powerful
people in your community . If the code of conduct does not apply to the powerful, those with
less power will be held accountable to the code of conduct while the powerful get away with
violating it, and the code of conduct will just be another method of strengthening oppressive
power structures—a classic double bind for the oppressed. If the code of conduct can't be
enforced on everyone in the community, we recommend not having a code of conduct at all.

See also:

● Holding powerful people accountable
Visible enforcement is required
Many communities want to respond to violations of codes of conduct secretly, usually out of a
desire to avoid drawing unwelcome attention to the victim, the perpetrator, the community, or the
community's leadership. Unfortunately, a code of conduct is only effective if the community sees
it being enforced. Without visible enforcement, other community members can't learn from
previous mistakes. People may assume that the code of conduct isn't serious and break it more
often than if they saw it being enforced. At a minimum, everyone who knows about the violation
of the code of conduct should see the community's response to the violation. Ideally, the
response will be shared publicly, possibly in an aggregated and anonymized form.

Another reason for public enforcement is that community members may decide to leave or
never join in the first place if they don't see the code of conduct being enforced. If they can't see
any evidence of a code of conduct being enforced, then it is reasonable to believe that it isn't
being enforced.

Case study: In the chat room for a feminist software community, one member repeatedly
made sexist comments to another member in a public channel. No one said anything about
this behavior in the public channel or to the target, so the target stopped participating in the
community. Later, the target found out that several people contacted the harasser privately
and told him to stop harassing other people, which he agreed to do. Those community
members thought they had handled the problem quite well, but based on the publicly available
information the community appeared unsafe and at least one person stopped participating as
a result.

Another reason is transparency: public enforcement allows the community to keep an eye on
enforcement and course correct if the code of conduct is being enforced poorly or in a biased
manner. For example, it is quite common for marginalized people to be accused of being "too
aggressive" or "too angry" when more privileged people get away with objectively far more
aggressive or angry behavior (the " Angry Black woman " stereotype in the U.S.). If the

community is notified every time a community member is sanctioned for being too aggressive,
they can raise a fuss if marginalized people are disproportionately sanctioned. Being public
about how a code of conduct report is handled helps keep the community accountable to its
members.

See also:

● Communicating the response to others
Summary
After reading this chapter, you should understand the following:

● The purpose of a code of conduct is to protect community members from harm in
community spaces.
● A code of conduct should contain a specific set of information and nothing else; including
other topics will seriously weaken the code of conduct.
● A code of conduct achieves its goals through education, normfollowing,
attraction/repulsion, deterrence, and boundary setting.
● A community should not use its code of conduct for anything other than protecting
community members from harm in community spaces.
● A code of conduct only governs community spaces.
● Your community can and should take into account behavior that occurs outside
community spaces when attempting to protect members in community spaces.
● Violations of the code of conduct must have meaningful consequences up to and
including a permanent ban from the community, and banning someone from your
community must be possible.
● If your community can't enforce the code of conduct on the most powerful members of
your community, it reinforces existing imbalances of power and your community is better
off without a code of conduct.
● Community members must be able to see that the code of conduct is being enforced, or
else you will scare off potential community members and make it more likely that
community members will break the code of conduct.

Resources referenced in this chapter:

● Paradox of Tolerance on Wikipedia
● " Angry Black woman " stereotype on Wikipedia

Chapter 2: Preparing to enforce a code of conduct
A fast and effective response to a report of a violation of a code of conduct doesn't happen by
accident; instead, it is the result of careful preparation and planning long before a report is
made. In this section, we will describe the steps to take before you receive your first code of
conduct report.

If you have not taken the steps in this section before your first report, that's okay. Being
prepared for the first report is the exception rather than the rule. We recommend reading this
section anyway because it sets up the structure you will need to handle a report successfully.
Since you won't have time to fully think through all of your choices at this time, we suggest
you:

● Announce that your decisions about process are temporary and subject to change
● Make temporary decisions (such as appointing an interim code of conduct committee)
● Commit to a schedule for making permanent choices about process by a specific date

Publicizing the code of conduct
A code of conduct should be prominently advertised in community spaces so that it is difficult or
impossible for someone to participate in the community without reading the code of conduct.
The more people see and read the code of conduct, the less likely it is that someone will violate
it. Ideally, the code of conduct will be one of the first things new members see when they join a
community. Some common ways to publicize a code of conduct include:

● An email or personal message to every community member
● A link in the footer of every email to a mailing list
● A toplevel menu item on a website
● A link in the topic of the main chat channel (or every chat channel)
● Posters on the walls of community spaces
● In an employment contract
● In new member orientation materials
● At the registration desk for a conference
● In the printed program for a conference
● Announcements during the first (or every) plenary session at a conference
● In the registration form for a conference

Identifying community members
To enforce a code of conduct, community members need unique, persistent identifiers,
otherwise, people violating the code of conduct can't be held responsible for their actions or
banned from the community if necessary. Often people use their legal names for this purpose,
but there are many other options. In online communities, members can be identified by
persistent usernames or email addresses. In small, closeknit, inperson communities, members
may be able to recognize every other member by their faces. For large inperson gatherings, we
recommend requiring name tags with large, easytoread print to be worn in a visible location by
all attendees at all times. People who are undergoing a traumatic situation easily forget things,
like what clothing their harasser is wearing, and having a large prominently displayed name tag
makes it much easier for someone to identify the person harassing them later on.

Name tags or usernames do not have to display a person's legal name; they can also be
pseudonyms, as long as the people enforcing the code of conduct can use them to identify a
person uniquely over the long term (occasional changes of pseudonyms or legal names are fine
as long as the community leadership can keep track of them). Pseudonyms are especially
useful when some members of the community are worried about being stalked or persecuted by
more powerful organizations and don't want to share information that can be used to harm them.
Women and/or activists are especially likely to need this protection.

Case study: At an invitationonly event, one attendee was concerned about physical attacks
from people at the event, due to previous credible death threats. She agreed to attend the
event on the condition that the organizers screen all registered attendees and require
attendees to wear their badges at all times during the event to show that they had been
screened. Another attendee had been stalked repeatedly during and after events, and did not
want to wear her name tag to protect herself from being stalked by other attendees. The
compromise was for the attendee worried about being stalked to wear her name tag, but use
a pseudonym on her name tag that could not be used to stalk her.

The code of conduct committee
A code of conduct committee is a small group of people who are responsible for enforcing the
code of conduct: taking reports, investigating reports, deciding on responses, executing those
responses, and informing the community about their decisions. The committee is what turns a
code of conduct from a written document into meaningful action. The committee must have the
full support of community leadership and have confidence they will not be overruled by
leadership unless they have made a significant mistake.

Naming a specific code of conduct committee, rather than just hoping someone will step up, is
necessary because enforcing a code of conduct is difficult, expert, emotionally intense work.
Most people don't want to be involved in highstress, high emotional labor like listening to
complaints of harassment or assault, and don't want to take actions that may result in them
being personally criticized or attacked. Like most emotional labor, if no one is assigned to do this
work, it often falls to the more marginalized members of your community to do it without support
or reward; they will then be more likely to leave under the stress. Even those who are willing to
do this difficult work are unlikely to feel empowered to act unless they are specifically authorized
to do it. As a result, if enforcing the code of conduct is left up to a nebulous "community" with no
specific people responsible, the code of conduct will not be enforced.

Case study: In 2011, a member of the Geek Feminism community repeatedly physically
touched other people without permission at communityrelated events, even after being
explicitly told to ask permission before touching others. They also made unwelcome sexual
advances to others. Due to a lack of formal governance structure in the community, it took
about 18 months to eject the harasser from the community after members became aware of
the first reports, and more than 3 years to create a formal code of conduct and code of
conduct committee. Until that time, not only did the Geek Feminism community not have a
code of conduct, no one felt authorized to adopt or enforce a code of conduct. This was true
despite the community including multiple code of conduct advocates (including the authors
and editor of this guide, who led the work to eject the harasser informally and adopt a formal
code of conduct). A formal, defined governance structure is a prerequisite for adopting and
enforcing a code of conduct, regardless of the expertise or character of the individual
community members.

Since serving on the code of conduct committee is usually a parttime and/or volunteer
responsibility in most communities, the committee should be large enough that if about one third
of the members are unavailable it can still operate quickly and effectively. Most committees are
between 3 and 6 people in size.

The members of the code of conduct committee must be publicly listed in the instructions for
reporting code of conduct violations. Many people are skeptical that their reports will be taken
seriously, but are more likely to report if they know exactly who will be handling their report.
Naming committee members is also essential for transparency and accountability. The code of
conduct must also specify that members of the code of conduct committee who have a major
conflict of interest will recuse themselves from handling that report (more on this on the section
on " Recusing committee members ").

See also:

● Recusing committee members

Choosing code of conduct committee members
Choosing code of conduct committee members can be done several different ways. This section
will outline the most common method.

Often the initial committee is appointed by existing leadership. Whenever possible, the code of
conduct committee should be separate from leadership. However, in a small organization, the
code of conduct committee may be the leadership itself, in which case being clear and
conscientious about recusal is even more important than usual, since leadership is often the
source of harassment. Even in this situation, the committee should strive to include people on
the commitee who aren't in leadership.

Once appointed, the committee chooses its own successors. An option for larger organizations
is to have a separate nominating committee. The committee must choose a chair, who is
responsible for moving the decisionmaking forward and assigning responsibilities. If the chair is
temporarily unavailable, the committee must choose an interim chair to fill their duties.

Whatever method you use for choosing committee members, it must not be popular vote by the
community. This is because many people are uncomfortable with enforcing a code of conduct
even in the most egregious cases, and will vote for people who are less likely to enforce the
code of conduct. In general, committee membership and operations should not be up for vote in
any way due to this effect.

The members of the committee should have fixedlength terms to prevent burnout. To preserve
working knowledge and best practices, committee members should serve overlapping terms
such that only some of the committee's terms expire at the same time. For example, when
starting a new committee with 6 people on it, appoint 3 people to a 6 month term, and 3 people
to a 1 year term, and after that appoint people for 1 year terms whenever an existing term is up.

Serving multiple terms in a row is reasonable for people who are paid to do this work as part of
their jobs, who have lots of energy for this work, or for organizations where this role is relatively
lowstress. Committee members should never feel irreplaceable or continue serving because
they feel guilty about letting the organization down. Prevent burnout by checking in regularly
with committee members about how they are feeling.

Committee members should include several people with significant knowledge about and
experience with being the target of oppression, ideally from several different marginalized
groups. Without this experience, a committee might make bad decisions due to not
understanding the pressures of systemic oppression or a failure to recognize patterns of abusive
behaviour that are more obvious to people with personal experience as the target of abuse. The
easiest way to get this knowledge is to include people who are the primary targets of a system

of oppression themselves and who have done the work to recognize and understand that
oppression.

While it is impossible to represent every axis of oppression on a committee, it is helpful to find
people who are aware that more than one kind of oppression exists. In particular, committee
members should be familiar with the concept of intersectionality , which was named and
popularized by the legal scholar Kimberlé Crenshaw. Intersectionality is the idea that people can
be subject to multiple overlapping forms of oppression, which interact and intersect with each
other in unique and specific ways. A committee should avoid recruiting a member who believes
all oppression can be reduced to class oppression, or that sexism is vastly more important than
any other kind of oppression.

The committee should include people with a public reputation in the community for fighting
abuse and oppression, so that people will be more likely to make reports. The committee should
also include people who are well respected within the community, to make community members
more likely to respect the decisions of the committee.

One kind of person must never be on the code of conduct committee: people who have doubts
about the effectiveness of the code of conduct, reservations about enforcing it, or more empathy
for the people breaking the code of conduct than the people who are being harmed. Likewise,
someone known to have violated the code of conduct should rarely be on the committee, and
only after they've made thorough amends for their actions, at minimum. Even people who have
made amends for their harmful actions should wait until they've established a pattern of better
behavior over several years before being considered for the committee. Often people who have
been caught harming others seek to immediately be welcomed into the group of people
protecting others from harm to cement their narrative of rehabilitation. Do not allow their desire
for redemption to take precedence over the safety of your community.

Committee members must commit to upholding the confidentiality of the committee publicly
even if they disagree with its decisions. If a committee member feels that they can't support the
decision of the committee, they should resign from the committee but not speak publicly about
the process unless absolutely necessary. In general, it is harmful for a committee member to
publicly disagree with any decision they were privy to. Public disagreement or breaking the
confidentiality of committee deliberations should be reserved for major breaches of ethics of the
committee. (An example of a situation in which breaking committee confidentiality was justified
was the Jim Frenkel at WisCon 38 incident , which was badly mishandled.) Anyone who thinks it
is reasonable to break committee confidentiality for small disagreements (e.g., because they
disagreed on minor wording decisions in the public statement) should not be on the committee.

Mandated reporters—people who, due to their job, are required by law to report sexual assault
or other forms of lawbreaking to authorities or superiors—should have explicit discussions about
their responsibilities with the rest of the committee before agreeing to serve on the code of
conduct committee. Mandated reporters should share what incidents they are required to report,

how likely those kinds of incidents are in their community, and talk about what they would have
to do if they learned about such an incident.

Mandated reporting responsibilities should be viewed as a potential conflict of interest,
particularly when it comes to any commitment to keep reports confidential. Mandated reporting
in the absence of meaningful protection for victims may do more harm than good. Whenever
possible, code of conduct committees should let victims choose whether they wish to involve
law enforcement. To protect victims’ right to privacy, mandated reporters should carefully
consider whether they should participate in a code of conduct committee.

Overall, members of the committee should have reputations for fairness, for standing up for
what is right, for speaking truth to power, and for knowledge about oppression.
Communicating with each other and the public
The committee must have a fast and efficient way of getting in touch with each other, such as a
private group mailing list, a group Signal chat, or a private chat channel. For inperson events
where decisions often need to be made within hours, committee members should share phone
numbers and start a group text message chat before the event begins (such as a Signal group).
The committee should also have every other committee member's direct contact info, such as in
a shared private document listing every member's phone number and email address.

The committee should also already know how to (and be able to) make public statements,
representing the committee in an official capacity. Some questions the committee should be
able to answer:

● What communication methods might you use to make announcements: community blog,
community social media account, email to an announcementonly mailing list?
● Does at least one person on the committee have the permissions to post to these
forums?
● If not, who will they contact to do this and will they agree that the committee is
authorized to post to them?
● How will the statements be signed?
● Who will review the statements before they are published?
Choosing a decision method
In the majority of cases we know of, the code of conduct committee is unanimous in its
decisions. Occasionally, some committee members will disagree. The code of conduct
committee should decide in advance how to make decisions if it does not have consensus,
including plans for what to do if there is a tie or if not everyone is present (e.g., deciding that if 3
out of 5 people are able to meet, then they can make a decision). We recommend a decision
method that errs in the direction of taking action against the alleged perpetrator, since the

majority of code of conduct enforcement errors we see stem from taking too long to make a
decision about what to do, or taking no action on a valid complaint.
Adopting an incident response guide
If possible, adopt a written incident response guide to guide your work, such as the one in the
" Responding to a report " chapter. When first receiving a report, often the committee members
are stressed and not capable of thinking well. Even when calm, people often forget important
steps of a process. A written guide provides a framework and a set of reminders to work through
during a stressful time.

An incident response guide is only a guide; it is intended to help a committee through the
decisionmaking process but not necessarily be followed to the letter. Don't treat the incident
response guide like a legal code to which you must adhere or you cannot take action. Legal
codes are followed so carefully because governments have the power to take away people's
property, imprison people, and kill people. A formal, detailed, timeconsuming legal process is
one method of protecting innocent people from the enormous consequences wielded by
powerful organizations like governments. If your community has enough power to inflict these
sorts of major consequences, the code of conduct model is not right for it.

The kinds of communities addressed in this guide can't take anyone's property or liberty, have
limited resources for conducting investigations, and cannot engage in the expensive,
timeconsuming processes which are part of what many people loosely refer to as "due
process." For these communities, a loose guide, focused on process, with lots of room for
interpretation is the right tool.

One part of the guide that must be strictly adhered to is the commitment to make a decision and
take action by a particular deadline, spelled out as a specific number of days or hours since the
report was received. Making a decision on how to respond to a report can be frightening and
unpleasant. In many cases, unless the committee has publicly committed to respond by a
certain time, they will put off any decision or action for so long that it harms the community. To
prevent this, make a public commitment to respond in a certain amount of time and ask the
community to hold you accountable.

If a committee adopts a written incident response guide, it must not make the guide public in a
way that suggests the guide is a hard and fast legalstyle code which must be adhered to in
every detail by the committee. The problem with people treating the guide this way is that many
people are not comfortable enforcing a code of conduct, and will use the guide to oppose the
committee's efforts to enforce it by arguing over whether the committee followed the guide
exactly. Even if this opposition is a good faith effort to help the community, it greatly reduces the
ability and willingness of the committee to enforce the code of conduct if every enforcement
action risks a long legalistic argument over whether the committee followed the response guide.

If your committee does choose to publish a guide, we recommend publishing a publicfacing
guide to what to expect when making a report, and explicitly state that the committee has private
internal guidelines that will be updated frequently. Both the code of conduct itself and the
incident response guide should be updated whenever the community finds that it could be
improved, without concern for abiding by the version in effect at the time of an incident. You can
use the " Responding to a report " chapter of this guide as a basis for your incident response
guide.

Obstructionist behavior around policing the committee's process is most often a concern in
looselyknit volunteer communities, such as online forums, where members interact daily and
some members have much more free time than others. Do not give into the demands of
community members if they require large investments of time from the code of conduct
committee before the code of conduct can be enforced.

Case study: An online community for the purpose of supporting women in a particular field
had a member that repeatedly posted comments that were right on the edge of violating the
code of conduct. The moderators would spend hours or days arguing about whether the
comment violated the code of conduct before making their decision. The member in question
would then use any public information about how the moderators made their decision to
create another borderline allowable comment that wasted several more hours of the
moderators' time in argument. Finally, the moderators recognized that the commenter was a
net drain on the community and banned him simply because he was distracting from the
community's mission.

Recordkeeping
The committee must keep records of previous reports and past decisions to aid in enforcement
and to detect patterns of behavior requiring action by the committee. For example, one offcolor
joke followed by a sincere apology is different from a pattern of offcolor jokes and unwanted
touching followed by "sincere" apologies. Decisions to make:

● Who will have access to the records (including ways to prevent committee members
from accidentally or intentionally seeing records about reports they are recused from)
● When will committee members consult the records (during deliberations of new reports,
at registration time, at checkin time, when people request commit access to the source
code repository, etc.)
● How long the committee will retain records
● What system the committee will use to keep the records
● How to keep the records secure and private

Training the committee
Committee members should be formally trained in some manner. This can be as simple
requiring new committee members to schedule a phone call with an existing committee member
to read over the incident response guide, or as complex as a multihour class with an external
trainer . They might talk over some theoretical code of conduct incident responses together, or
do a postmortem on previous incidents the committee has handled. The more practice the
committee has discussing the response plan and coming to agreement on theoretical cases, the
easier it will be to act quickly and effectively when the committee receives a real report.

The committee should practice responding to reports as a group on a regular basis (e.g. every
year in January, or at the beginning of any new member's term, or a month before a big event).
The more lifelike the practice, the better. See the " Examples of responding to reports " chapter
for a selection of realworld code of conduct reports that can be used for practice.

Some useful reading for committee members:

● Timeline of Incidents on Geek Feminism Wiki
● Conference code of conduct resources on Geek Feminism Wiki
● Community code of conduct resources on Geek Feminism Wiki
● " ‘Why Didn’t You Report It?’ " by s.e. smith
● " 'Why don’t you just hit him?' " by Mary Gardiner
● " Women, Race, & Class " by Angela Y. Davis
Training reporttakers
Ideally, all reports of code of conduct violations would be made directly to a member of the code
of conduct committee. However, in many situations, the people taking reports will not be on the
committee: conference volunteers, managers, or event staff. Your community will need to teach
everyone who might be approached by someone with a report one of two things: how to take
reports, or how to refer them to someone who has been trained to take a report.

Reporttakers should have a "cheat sheet," a short form, or some other guide to remind them
what to do (see an example reporttaking form in Appendix 2 ). Some questions the reporttaker
should ask (but not pressure the reporter to answer if they don't want to share):

● Identifying information for the alleged harasser
● Reporter's name and contact information
● Time and date of incident
● Place of incident
● What happened
● Any other people who were involved or witnessed the incident

When taking a report in person, the reporttaker should find a quiet place where others can’t
easily overhear the report. The reporter may need some time and support to make a report: time
to breathe deeply, the presence of a trusted friend or colleague, food, water, medication, tissues,
etc. The reporttaker should pay attention to the reporter's emotional and physical state and take
reasonable steps to meet the reporter's needs.

The reporttaker should also write down the report as it is being given, or take notes and write a
written report as soon as possible. If possible, they should read back the written report to the
reporter to check for accuracy. Writing down the report is crucial: with verbal reports, the story
gets slightly changed each time it is repeated, which can result in major mistakes in the
committee’s response. The reporttaker should not share what they have learned with anyone
not on the code of conduct committee.

Early on, the reporttaker should find out whether the reporter wants this report to go to the code
of conduct committee, with the knowledge that they will make their own decision on how to
respond, or if the reporter is just looking for a sympathetic ear and does not want to make a
formal report. If the reporter is looking for a sympathetic ear only, the reporttaker should feel
free to decline listening if they don't have the energy to listen to and keep a secret, especially if
they are a member of a marginalized group that is disproportionately burdened with emotional
labor.

The reporttaker cannot promise unconditional confidentiality because they may learn
something that threatens the safety of other community members and need to communicate it to
others. Even seemingly minor incidents can be a warning sign if they are part of pattern of
behavior. More than once, we’ve seen a person commit several apparently minor infractions,
and go on to commit a serious assault later in the event. To avoid this danger, we recommend
reporttakers pass on even minor complaints to the committee, even if the reporter wants to stay
anonymous and/or doesn't think it is worth a formal complaint.

The reporttaker shouldn't try to pressure the reporter in any way, and should not make any
promises about how the committee will respond to the report. In particular, people reporting
incidents will sometimes ask for a promise that no action be taken against the harasser. But the
reporttaker cannot make promises on behalf of the rest of the code of conduct committee. The
most they can do is say that the committee will try its best to protect the reporter from retaliation.
The reporttaker can't promise confidentiality either, just promise to try their best to keep it
confidential. Protecting the community may, in some cases, require risking others finding out
who the reporter was. See the section " When individual safety conflicts with community safety "
for more details on how to handle this situation.

The reporttaker should also not ask the reporter for any solutions or suggestions on what action
to take. It's fine to ask clarifying questions like, "Do you feel safe staying at the conference right
now?" or "Do you feel safe around this person?" The reporttaker should not pressure the

reporter to contact security or law enforcement but they should offer their support in doing so if
the reporter wants it and it is safe for the reporttaker to support them (for more on this, see
" 'Why Didn't You Report It?' " by s.e. smith). If the reporttaker thinks there is immediate physical
danger to anyone, they should follow the community's security plan or find someone who can do
so safely.

Once the reporttaker has taken the report and taken care of any timesensitive tasks, they
should spend some time checking on the reporter's immediate needs. The reporter may be
feeling fine and be happy to go back to what they were doing, or they may be in need of help. If
it seems appropriate, the reporttaker should offer the reporter a private safe place to sit, ask if
them if they want anyone to be with them and send someone else to fetch that person if so, and
ask "Is there anything that will make you feel safer that I can help with?"

Do not assume that the reporter will trust any specific person, including the reporttaker or any
member of staff. For example, don't offer a random member of staff to walk with them alone
back to their hotel room, because there is no reason they should feel safe alone with a person
they don't know. Instead ask them if there is someone they would feel safer to have with them
and send for that person. There's no reason the reporter should trust the reporttaker or any
other community member automatically—plenty of abusive people deliberately seek out
positions where they have access to people at their most vulnerable, such as volunteer
conference security or the human resources department at their company. The community
should do its best to prevent this from happening, but should not ask anyone to automatically
trust another person.

Once the immediate concerns have been taken care of, the reporttaker should send the written
report to the code of conduct committee and make sure a committee member acknowledges its
receipt.
Avoiding or mitigating higherrisk activities
The best way to handle code of conduct reports is to prevent incidents in the first place. One
way to do that is to avoid or mitigate activities that carry a higher risk of people violating a code
of conduct. Some higher risk activities include:

● Serving alcohol, especially with an open bar (unlimited free alcoholic drinks)
● Hot tubs, swimming pools, saunas, or other partially clothed or nude activities
● Events at nightclubs, bars, and dance halls
● Comedy and especially improvisational comedy (e.g. slide karaoke, hired comedians)
● Activities involving fake or real guns or weapons (e.g. laser tag, target shooting)
● Short unreviewed talks (e.g. lightning talks, Ignite talks)
● Coercive bonding exercises (e.g. trust falls, icebreakers)
● Activities involving costumes
● Sexualized staff (e.g. servers wearing sexy costumes, models wearing logos)

Any outside staff or contractors hired to work in your community should be educated on your
community code of conduct and asked to agree to it, ideally in a contractually binding manner.

If your organization decides to include a highrisk activity, you can take steps to mitigate the
risks. For example, if you have a party with a swimming pool, you should explicitly remind
people not to touch other people without their explicit permission, announce that the code of
conduct applies in the pool area, and remind people not to comment on other people's bodies.
Serving alcohol is less of a problem when the organizers limit drinks per person or don't include
hard alcohol (see " Alcohol and Inclusivity: Planning Tech Events with NonAlcoholic Options " by
Kara Sowles for more ideas). People who want to give lightning talks can fill out a form asking if
their presentation touches on any topics likely to result in a code of conduct violation, and the
organizers can ask them further questions if they answer yes to any of the questions. For more
details, see " Higher Risk Activities " by Mary Gardiner et al.

" Inclusive Offsites " by Sara Smollett et al. is a guide to choosing group activities that not only
lower the risk of code of conduct violations, but are more accessible and more fun for more
members of your community.
Make arrangements for legal advice

Most code of conduct reports and responses will not require legal advice, but you should spend
some time identifying any areas where your committee may need legal advice, and what you will
do to get that legal advice. Here are some examples of areas where a committee may consider
asking for legal advice:

● Publicly accusing someone by name of a crime or unprofessional behavior
● Anything involving employment
● Physical or sexual abuse of children or minors
● The community is associated with, part of, or funded by the government (e.g. public
education, government contractors, public hospitals)
● Anything that a mandatory reporter in your community might be required to report

Often, nonprofit or charitable organizations can get legal advice for free through pro bono
programs of forprofit firms, or nonprofit sources of legal advice.

See also:

● Putting legal concerns into context

Updating code of conduct materials
As the community and the code of conduct committee gain practical experience, it makes sense
to update the code of conduct and any supporting materials every year or two. For example,
your community's understanding of oppression may have changed, or your reporting
instructions turned out to be poorly worded. Most often the updates are minor changes, but
occasionally a full rewrite is necessary, especially if the community's understanding of the
purpose of a code of conduct has changed. For example, some communities have rewritten
their code of conduct to explicitly acknowledge that its primary purpose is to protect those with
less power and privilege from abuse from those with more power and privilege. This eliminates,
for example, false claims of racism against the dominant racial or ethnic group, or false claims of
sexism against men. The Geek Feminism published an example community code of conduct
constructed in this way.

We recommend dating every edition of the code of conduct, and keeping a publicly available
history of previous versions. If the code of conduct changes substantially, we recommend
making a strong effort to notify everyone in the community of the changes.
Summary

● A code of conduct must be publicized widely and new community members should be
told about it as soon as possible.
● Your community must have a way to identify members uniquely and persistently via
names or persistent pseudonyms.
● Your community must have a committee tasked with enforcing the code of conduct
whose names are publicly available to people making reports.
● Committee members must recuse themselves from handling a report when they have a
real or perceived conflict of interest.
● Code of conduct committee members must be chosen carefully according to a range of
criteria via a nomination process, not by popular vote.
● The committee should agree in advance on the broad outlines of how to respond to a
code of conduct report, and ideally have a written response guide.
● The committee should keep records of code of conduct reports and adopt rules for
keeping some reports private from future committee members if necessary.
● Both the committee and potential reporttakers should be trained in some manner.
● Avoiding or mitigating highrisk activities, such as those involving less clothing or alcohol,
will reduce the likelihood of code of conduct violations.
● The committee should make plans in advance about when and how to seek legal advice.
● The code of conduct should be updated when necessary.

● A committee can take action on something that harms their community even if it is not
banned by the current code of conduct.


● Intersectionality on Wikipedia
● Jim Frenkel at WisCon 38 incident on Geek Feminism Wiki
● Code of conduct training resources from Frame Shift Consulting
● Conference code of conduct resources on Geek Feminism Wiki
● Community code of conduct resources on Geek Feminism Wiki
● " ‘Why Didn’t You Report It?’ " by s.e. smith
● " 'Why don’t you just hit him?' " by Mary Gardiner
● " Women, Race, & Class " by Angela Y. Davis
● " Alcohol and Inclusivity: Planning Tech Events with NonAlcoholic Options " by Kara
Sowles
● " Higher Risk Activities " by Mary Gardiner et al.
● " Inclusive Offsites " by Sara Smollett
● Example community code of conduct on Geek Feminism Wiki

Chapter 3: Responding to a report
This chapter summarizes the steps a code of conduct committee should take when responding
to a code of conduct report in quick reference form. It includes a short description of each step
to take in the process of handling a report:

● Receiving the report
● Investigating the report
● Making a decision
● Announcing the decision
● Responding to any criticism

This chapter is written so that it can be copied and used directly as a guide for handling code of
conduct reports. Indepth discussion of each step is reserved for the preceding and following
chapters, and links to relevant sections of this guide are included at the end of each step.

The instructions in this chapter start at the time that the code of conduct committee receives the
report. The committee chair (or interim chair) is responsible for moving the committee forward
through each step in this chapter, following up on tasks assigned to other committee members,
and meeting the deadlines the committee has chosen.
Start the response deadline clock
Whatever deadline you committed to for a response (e.g. 24 hours to a public acknowledgement
and 10 business days to a decision), the deadline clock starts ticking when the committee
receives the report. Write down and share the deadline with the entire committee: make several
calendar appointments, put it in the notes for the incident, email everyone, etc.

See also:

● Responding to incomplete or late reports
Check to see if everyone is safe
If unclear, the committee should first find out whether any immediate safety concerns have been
taken care of. The original reporttaker may not have realized that anyone was in danger, or
have been so focused on getting the report to the committee that they forgot to take steps to
protect the reporter or the community. Follow your organization's security response plan if
necessary.

Write down the report if necessary
If the report reached the committee in nonwritten form, it must be written down by a committee
member as soon as possible and shared with the rest of the committee.
Make a preliminary announcement if appropriate
For incidents that are widely known (e.g., an event that happened in the plenary keynote or on
the community mailing list), consider an immediate announcement that the code of conduct
committee has received a report about the event and is working on it. Give the committee's
contact information for people who might want to add relevant information. Be ready to redirect
harmful discussion about the incident to a venue where it is less harmful.

See also:

Ask for recusals
The chair should explicitly ask every committee member if they have a strong conflict of interest
that would bias their decision away from protecting the community from harm, or that would
produce an appearance of bias. If anyone needs to be recused, assign a specific committee
member to take the steps necessary to carry out the recusal: removing them from documents,
setting up a new mailing list without them, creating a new chat channel, etc. Recusal begins as
soon as someone realizes they should recuse themselves. For example, when reading a
new report, committee members must stop reading as soon as they realize they need to recuse
themselves.

See also:

● Recusing committee members
Organize a committee meeting
Arrange for the committee to meet (in person or online) as soon as is practical. The bigger the
reported violation, the sooner the committee should meet. If the report happens during an event,
meet as soon as possible, since the event will be over soon. It's okay if some committee
members can't make the meeting, as long as enough people can attend to do a good job (we
recommend 23 minimum).

Do additional research
Assign committee members to find more information on the incident as appropriate, either
before or after the meeting. This might include:

● Interviewing the alleged harasser (if safe)
● Interviewing other witnesses
● Checking with their social networks for other people who have had bad experiences with
the harasser
● Checking the committee's records
● Searching community records
● Searching the Internet for relevant information
● Attempting to identify any unidentified people involved

See also:

● Investigating the incident
Meet as a committee
Only members of the code of conduct committee who have not recused themselves should be
present at this meeting. In particular, neither the alleged harasser or the target(s) should be
present at this meeting.

At this meeting, discuss:

● What happened?
● What are we going to do about it?
● Who will do it?
● When will they do it?

Have the written incident response guide on hand and check with it frequently to make sure the
committee isn't forgetting steps.

Some useful principles to keep in mind:

● The committee's primary purpose is to protect the community from further harm
● The people most likely to be the targets have less privilege and power
● You are more likely to disbelieve the words of people with less privilege and power
● You are likely to have more practice empathizing with people with more privilege and
power
● Whenever possible, protect the reporter and any targets of harassment

● Consider the longterm effects of taking action (or not taking action)
● Remember your deadline for response

For complicated cases, the committee may need to meet more than once. If this is the case,
schedule the next meeting before leaving the current meeting.

See also:

● Impact is more important than intent
● Distinguishing good intent from bad intent
● DARVO: Deny, Attack, Reverse Victim and Offender
● Judging competing claims of marginalization
● Social awkwardness and harassment
● Mental health and harassment
● Children, caregivers, and harassment
● Sexual behavior and communities
● Alcohol and drugs
Choose a response
The committee should choose the response that will best protect their community from future
harm. Some common responses include:

● Nothing (if you think no action is necessary to protect your community from harm)
● Have a chat with the person(s) who violated the code of conduct
● Issue a formal warning
● End a talk early
● Do not publish videos of the offending talk
● Edit out offending portions of videos of talks
● Do not let the person speak in the future
● Remove responsibilities, privilege, or access
● Temporary ban from community spaces
● Permanent ban from community spaces
● Recommend that they be fired or demoted
● If you can't identify or find the harasser(s), state publicly what you would have done if
you could have

In addition to any direct action, the committee might also consider:

● Contacting the harasser's employer or other associated communities
● Warning other communities that may also be at risk
● Warning the public in general

In some cases, individuals or organizations have knowingly and voluntarily enabled abuse over
a long period of time. This is especially common with longterm serial abuse by highly placed
members of organizations or famous people. To protect the community, the committee may
decide to sanction people or organizations that enabled the abuse, to discourage them from
enabling abuse by the original abuser or by new abusers. This might look like:

● Not partnering or doing business with the organization
● Not promoting the organization's work or the organization's leadership
● Not allowing talks about the organization at community events
● Not allowing the organization to sponsor community events or organizations
● Not allowing the leadership of the organization to attend events or speak

If you are considering a response that is not listed in this section, it may be on the list of
responses not to use . Please read that section in detail before using a response not listed here.

Remember, when choosing a response, the highest priority of the committee must be protecting
the community from further harm.

See also:

● When individual safety conflicts with community safety
● Protecting the community's reputation
● Safety is more important than privacy and confidentiality
● Choosing a proportional response
● Responses not to use
Take any actions necessary to implement the response
This step may be done before or after communicating the response to the target(s), harasser(s),
or the rest of the community, depending on what you think the likely risks are. For example, if
you are worried that the harasser will vandalize the community website, you might remove their
website editing permissions before you tell them about your decision. It's important to have a
specific person or persons responsible for taking these actions, along with a deadline to do them
and clear communication about what order to do things in and whether they have been
completed. Sometimes it is helpful to have two or more people doing the work together, so they
can support each other or check each other's work. Creating a shared spreadsheet to track the
tasks is a reasonable thing to do if there are several steps or if getting them out of order would
be a big problem.
Inform the target and harasser of the response
If feasible, privately inform the target(s) of your proposed response before announcing it
publicly. This is not to get their approval, but to find if there is information the committee has

missed or misunderstood. Often an important detail or new information will come out if the
committee proposes a response that doesn't solve the whole problem or seems disproportionate
to the target. While it is nice when the target agrees with your response, it is not necessary for
them to agree with your response.

Privately inform the harasser(s) of the response if and when that is feasible. Again, the harasser
does not need to agree with the response, they just need to indicate whether or not they are
going to cooperate with the committee's decision. Do not argue with the harasser. They may
provide new information or ask for a change, but the committee's decision should take effect
immediately and continue while the committee decides whether to reconsider their decision.

See also:

● If a harasser refuses to follow the code of conduct
● Responding to protest from the alleged harasser
Communicate the response to others
A useful guideline is that everyone who is aware of the violation of the code of conduct should
also know about the committee's response. You should also err in the direction of more people
knowing than fewer. The announcement should be brief and factual, with no extraneous
material. Avoiding ambiguity or statements with several different possible interpretations is
especially important, as any misinterpretations will require a followup statement to clarify. For
best results, have someone who is not a member of the committee review the announcement
before sharing it.

The statement should contain:

● Usually, no names, either of alleged harasser or target (some exceptions exist)
● A brief description of what happened
● A brief description of the response
● A brief, neutral description of any apology from the harasser
● Instructions on how to send feedback to the committee (usually an email address)
● Optionally, instructions on where to discuss the committee's decision
● Optionally, a brief restatement of the community's commitment to safety

Some common options for publicizing the statement are:

● Announcement mailing list
● Community blog
● Announcement at a plenary session or allhands meeting
● Social media

For smaller incidents, it is reasonable to periodically publish a "transparency report"
summarizing all the reports the committee handled in the time since the last report. For
conferences, a few days after the event is a good time to publish this. For ongoing communities,
every month, quarter, or year might make sense.

See also:

Respond to criticism
We recommend responding as little as possible to criticism. This approach is often successful:

● Wait to see how people react to the initial announcement
● If necessary, post one followup to clarify any genuine misunderstandings
● Do not try to persuade people who strongly disagree with you
● Refuse to provide more details about the incident or its handling
● Refuse to engage in oneonone arguments, online or in person
● Redirect any communitywide discussions into smaller venues so that victims and
survivors don’t have to see their community debating their right to safety

See also:

● Responding to criticism
● Dealing with attacks on the committee or community
Summary

● When the committee receives a report, it should start the response deadline clock, check
to make sure everyone is safe, write down the report if necessary, and make a
preliminary announcement if necessary.
● When necessary, investigate the report and the people involved.
● Ask for recusals of committee members.
● If another committee meeting is necessary, schedule the next meeting before ending the
current meeting.
● If your committee wants to use a response that is not on the list of common responses to
code of conduct violations, it should check if the response is on the list of responses not
to use.
● If feasible, tell the target and the harasser about your decision before announcing it
publicly.

● When possible, communicate the committee's decision as widely as possible, using a
short, concise, unambiguous statement.
● Respond to criticism sparingly, and do not provide further details unless absolutely
necessary.
● Set up a communication channel for criticism of the committee's decision, and refuse to
have oneonone conversations about the committee's decision.
● If you need more detail on any topic in this chapter, you can follow the links at the end of
each section.

Chapter 4: Discussion
In the previous chapters, we outlined the steps involved in enforcing a code of conduct at a high
level. In this chapter, we go into depth on various aspects of enforcing a code of conduct. The
sections are arranged roughly in order of when they are useful during the code of conduct
process, starting with sections relevant to writing and adopting a code of conduct, followed by
sections relevant to preparing to enforce a code of conduct and responding to a report.
What does not belong in a code of conduct
Often, communities try to use a code of conduct for purposes other than protecting a community
from harm, such as describing best practices for meetings. However, cramming these other
documents into the code of conduct weakens the code of conduct because it's unclear which
parts of the code of conduct are requirements, and which parts are helpful suggestions or
recommendations which can be ignored without consequence. To differentiate between
behavior that is not permitted, and behavior that is simply not recommended, codes of conduct,
mission statements, and rules for conducting meetings, and similar items belong in separate,
distinct documents.

A code of conduct must only list unacceptable behaviors, not acceptable or "positive" behaviors.
People often object that a code of conduct is offputting because it's a list of "don'ts" and they
would rather see a list of "dos." However, there is a reason that most legal systems in the history
of the world consist of a list of "don'ts": there are usually far more acceptable behaviors than
unacceptable behaviors, and the reader has to put in extra mental energy to deduce what isn't
allowed by figuring out what isn't listed. For example, if you wanted to ban racial slurs and
insults, would you list all of the nonracial slurs and insults that it is okay for people to use, and
let them puzzle out which ones aren't listed? No; you'd just tell them not to use racial slurs or
insults.

In our experience, when people object to a list of "don'ts," they aren't objecting because it is an
ineffective format, but because they are feeling uncomfortable with the idea of having to tell
people how to behave and enforcing those standards of behavior. Someone who objects to a list
of "don'ts" should not be involved in writing or enforcing your code of conduct.
List of unacceptable behaviors
The list of unacceptable behaviors should be short, direct, and to the point. It should only
include items that meet all three of the following criteria:

● Behaviors that are somewhat likely to happen
● Behaviors that some people think are acceptable within the community

● Behaviors that will result in action if someone reports them

A code of conduct should not bother to list behaviors that are unlikely to occur, such as "no
riding horses through the office hallways," because otherwise the code of conduct becomes
unmanageably long and people are less likely to read or remember it. However, the authors
should listen when community members say that a behavior does occur, even if they have not
personally experienced it. For example, we know of more than one large tech company where
people literally rode motorcycles and electric scooters (not mobility scooters) through the
hallways—not something you'd ordinarily think you would have to ban.

A list of unacceptable behaviors should also not include behaviors that are universally known to
be unacceptable, such as murder. Again, listen to your community members about behaviours
that they have observed or that others think are acceptable.

Note that something does not have to be explicitly listed as unacceptable in the code of conduct
for the code of conduct committee to take action. For example, if a community member did
commit murder, you could still ban them even if your code of conduct did not specifically list it.
For serious crimes outlawed by the legal system, such as spousal abuse, a code of conduct
committee should take action even if the perpetrator is not convicted by a court of law if they
believe the allegations are likely to be true. In general, feel free to ban someone for any harmful
behavior regardless of whether it is in the code of conduct, and then update the code of conduct
afterwards if appropriate. You can add "Obey instructions of staff" to allow a staff member to
establish any behavior as against the code of conduct without delay.

Everything in the code of conduct should be serious enough that the community would take
action if someone reported it. That is, the committee would at least have a chat with the person
who did it to explain why they should stop, and would take more serious action if they did not
stop. Guidelines for more fruitful discussion or better problemsolving don't belong in a code of
conduct if disobeying them would not result in action from the code of conduct committee.
Imagine reporting "He didn't keep an open mind and explore many possible solutions" to the
committee and you'll see why including these kinds of guidelines in a code of conduct
undermines the code of conduct.

Case study: A code of conduct listed "interrupting" as unwanted behavior, as part of
guidelines on having better meetings. Unfortunately, interrupting was extremely common and
continued to happen in the community without consequence. This encouraged community
members to disregard the rest of the code of conduct. Interrupting was also necessary to
create a more inclusive environment—e.g., interrupting someone who is dominating the
conversation so other people have a chance to speak. What the authors of this code of
conduct wanted to stop was interrupting that harmed productivity, or silenced marginalized
people. I recommended listing "sustained disruption or heckling" as an unacceptable behavior,
and moving the guidelines for better meetings into a separate document.

Guidelines for better discussion or problemsolving are useful as long as they are in a separate
document from the code of conduct, especially if they take a systemic form that is easier to
follow than to depart from. For example, you can print out meeting role cards , leave them in
meeting spaces, and encourage meeting participants to assign specific roles to each other.

Any guidelines you do create should be carefully researched and vetted, especially by
marginalized members of the community, to see if they have a counterproductive effect. For
example, adopting a guideline of "Assume positive intent" will have the counterintuitive effect of
centering the feelings of the privileged person and will end up silencing those with less power.
For more on this, see " How 'Good Intent' Undermines Diversity and Inclusion " by Annalee
Flower Horne.

Similarly, you must not include descriptions of the ideal behavior of a community member or the
ideal form of a community culture in the code of conduct. Often these statements describe an
idealized version of a community as though it was already true. For example, one code of
conduct includes the statement, "We gain strength from diversity, and actively seek participation
from those who enhance it." Aspirational statements make sense when describing principles or
goals ("We believe... We strive to... We aim to..."), but it's unlikely that any community
universally lives up to these principles and goals.

Stating these goals as accomplished facts has a silencing effect on people who have differing
experiences and makes them less likely to report a violation of the code of conduct—after all,
the people who wrote the code of conduct have already said that doesn't happen in our
community, so why would they believe me when I say it happened? It also weakens the rest of
the code of conduct, by mixing vague aspirational descriptions with specific lists of unacceptable
behavior that have important consequences.

A code of conduct must not contain:

● Jokes
● Arguments for freedom of speech
● Statements that everyone should already know these things
● Expressions of certainty that no one will break the code of conduct
● "Be excellent to each other," “don’t be a dick,” etc.
● Anything else that trivializes or reduces the seriousness of the code of conduct

For more details on what does and doesn't belong in a code of conduct, read " Code of Conduct
Enforcement Warning Signs " by Sage Sharp.

Transformative justice and codes of conduct
Transformative justice is a system for responding to harmful actions by finding solutions that
transform the relationships between victim, offender, and the entire community for a more
positive outcome in the longer term. It is based on restorative justice, which aims to change the
formal legal justice system so that it prefers restorative solutions (repairing the harm done to the
target) over retributive solutions (punishing the offender). One of the explicit goals of
transformative justice is to keep offenders integrated into the community they live in, as a way
for communities to protect themselves from losing too many members to prison, criminal
organizations, or death. (In this context, "community" means people living nearby, family
members, schools, coreligionists, coworkers, employers—the entire social fabric that supports
a person in a loving, fulfilling, and meaningful life.) Another goal of transformative justice is to
center the victim and their needs.

Transformative justice is often successful in communities where community bonds are strong
and people leaving the community is extremely harmful to both the community and the person
leaving it. It is particularly useful for communities that are unjustly targeted by the formal legal
system. For example, it can prevent shoplifting a few dollars worth of food from escalating into
years of imprisonment, joining a criminal organization, or death. It tries to repair the harm done
to the community instead of increasing the harm.

Unfortunately, transformative justice is rarely successful in a more casual or looseknit
community of the sort this document is aimed at, such as hobby groups, open source software
projects, or attendees of a conference. Factors that prevent transformative justice from being
effective in looseknit communities include:

● Transformative justice seeks alternatives to state violence against marginalized people,
but harassers in looseknit communities tend to have significant privilege in the
community and in society at large, and are not at risk of facing disproportionately harsh
consequences even if they behave egregiously.
● A foundational tenet of transformative justice—that the community is better off keeping
the offender in the community than ejecting them—is not true for these communities.
● The process is enormously timeconsuming and harassers often have more free time
than their targets; many targets simply leave rather than continue the process.
● Often offenders would rather leave the community than repair the harm and change their
behavior—they don't have enough incentive to change.
● Offenders who are talented manipulators can hijack the transformative justice process
and use it to harm the targets even more.
● These communities rarely have experts in transformative justice, and when amateurs
attempt to implement it, they often revictimize the targets unintentionally.
● The transformative justice process is affected by the implicit bias of the people
implementing it, often giving an unfair advantage to the more privileged people involved.

Too often, people approach transformative justice as a neat idea they'd like to experiment with in
their community, and end up harming the most marginalized members of their community
instead of helping them. If you want to use transformative justice in your community, we
recommend you only do so if you have several people who have formal training in
transformative justice available to work the process, and you only attempt the process with
members of your community who have a strong incentive to stay inside the community and to
treat other members of the community well.

Community members often worry that if their community kicks out the offender, they'll just find a
new community and hurt people there. They then conclude that it must be their current
community's responsibility to rehabilitate the offender. This is somewhat analogous to advising
someone with an abusive spouse to stay married because otherwise the abusive spouse will
simply find someone else to marry. All we can do is use the power we have to protect people in
the communities we control. You don't have a responsibility to rehabilitate an abuser just
because they joined your community, and you can't force an abuser to rehabilitate themselves.
Your community can best care for other communities by warning other communities the abuser
is likely to join.

When enforcing a code of conduct, we recommend people focus on protecting the safety of their
community, not on retaining members of their community who are offenders. When offenders
sincerely accept responsibility, want to repair the harm they did, want to prevent future harm,
and prioritize community safety over their own desires and needs, that's a great outcome and it
is reasonable to work with the offender. Otherwise, we recommend using deterrence and
prevention: showing the community that you're serious about the code of conduct by ejecting
anyone who seems likely to violate the code of conduct in the future, as judged by their actions
and statements both within and outside the community.

Our experience with looseknit, casual communities is that unless the offender fairly quickly
agrees to repair the harm and change their behavior, they don't have enough incentive to
change and are likely to continue harming the community until they are forced to leave. Putting
significant effort into changing the offender's beliefs is a waste of the time and energy of the
community, which would be better spent on supporting the targets.
Recusing committee members
To recuse means to remove someone from a position as a judge in a case because a conflict of
interest may cause them to make (or appear to make) a biased decision. Committee members
must recuse themselves from handling any report if they have any significant conflict of interest
which might get in the way of making a decision in the best interests of the community, or if
others would believe that to be the case. Other committee members must insist on the recusal
of committee members who do not recuse themselves but have or appear to have conflicts of
interest.

Some examples of when a committee member must recuse themselves include if they are:

● The specific target of harassment
● The alleged harasser
● Close friends with either party
● Business partners with either party
● Romantic partners with either party
● In a family relationship with either party
● In some hierarchical academic or business relationship with either party
● Engaged in a significant and personal conflict with either party
● In some other significant power relationship with either party

For example, if a student was required to be part of the committee handling an incident in which
their academic supervisor was the alleged harasser, the student would reasonably be afraid of
retaliation if they were part of a decision that sanctioned their academic supervisor. As a result,
they might make a more lenient decision than they would if they were solely concerned with the
safety of the community.

If most or all members of the committee have a similar relationship to the harasser or target,
they probably should not recuse themselves. It's possible that a harasser could, for example,
harass every single member of the committee, in which case no one needs to recuse
themselves. Also, sometimes every member of the committee will happen to be a target of the
harassment, such as if they were all in the audience when a pornographic slide was shown in a
keynote talk at a conference. In this case, no one needs to recuse themselves.

Be thoughtful about the level of friendship that results in recusal. Plenty of people might be
friends with every member of the committee (especially popular people), but only members who,
for example, considered that person their best friend would reasonably be in a position to recuse
themselves. Likewise for personal conflict; it's entirely possible for a target to have had a fight
with every member of the committee, or for every member of the committee to think poorly of
that person based on their actions. In these situations, no one should recuse themselves unless
they had some truly unique and exceptional interaction with that person.

For example, in a speculative fiction community, if one author writes a series of blog posts
expressing their racist and sexist opinions on other authors' work, every single member of the
committee should strongly dislike the author and none of them should recuse themselves.
However, if one of the committee members had a sexual relationship with the author (previously
or currently), they probably should recuse themselves. As another example, if a member of the
committee has previously responded to a code of conduct report about this person, and the
person did not like the response, the committee member does not have to recuse themselves
on that basis.

The appearance of a conflict of interest is important as well. For example, if someone on the
committee is employed by the alleged harasser but is already planning to quit their job and
doesn't need a reference from that employer, they must still recuse themselves because the rest
of the community is likely to think their decision was biased because they assume an employee
needs their employer's good opinion.

When someone has recused themselves, that means they must not influence the decision in
any way, or even be in a situation in which they might influence the decision, even accidentally.
This means that as soon as they recognize they have a conflict of interest, they must:

● Stop reading any documents or records
● Remove their access from any documents or records
● Stop talking about the report with other committee members
● Leave the room or call if people are discussing it
● Not go to meetings about it
● Leave a meeting if people start talking about it
● Not read, send, or receive emails about it
● Ask the rest of the committee to stop sending them emails about it
● Delete any emails they receive about it
● Not participate in online chats about it
● Leave any online chats if the topic changes to it
● Delete any personal logs of committee chats about it

In particular, when committee members are first reading a new report, they must stop reading as
soon as they suspect they may need to recuse themselves and inform other committees that
they may need to be recused. The other committee members can finish reading the report and
figure out if the first committee member needs to recuse themselves.

When it comes to enforcing the recusal, err on the side of safety. If someone doesn't know what
the rest of the committee is saying, isn't present for any discussions, and doesn't have access to
any documents, it's much easier for them to avoid influencing the response to the report.

When possible, announce any recusals publicly. In some cases, the reasons for recusal or the
existence of recusal may need to be kept private to protect committee members or targets. For
example, if a committee member reports sexual assault, it may be necessary to keep their
recusal private to protect their identity.
When individual safety conflicts with community safety
The goal of protecting the community from further harm will sometimes conflict with other
community goals or values. For example, when someone is the target of sexual harassment or
assault, the target often rightly fears retaliation by the harasser, or further harassment and

assault from other people (e.g., an Internet troll mob doxxing and harassing them). Blaming and
silencing the victim is part of rape culture.

When the target does not want to be identified, consider creative methods of protecting both the
community and the target's identity. One option is to find other actions that the harasser has
taken that could be used to justify the action necessary to protect your community. When
someone is reported for something as serious as sexual assault or harassment, they often have
a long record of other offences. Asking around in their social and professional community,
especially by people who have a lot of connections with marginalized people, will often result in
a long list of other violations of your code of conduct. Keep in mind that offenses committed by
abusers may come in a wide variety of forms and may not even be part of your code of conduct,
such as plagiarizing other people's work, falsifying expense reports, or embezzling money. For
more on this phenomenon, see " The Al Capone Theory of Sexual Harassment " by Leigh
Honeywell and Valerie Aurora.

Another option is to tell the harasser that you've received multiple reports of unacceptable
behavior over several months, whether or not the committee has received multiple reports—that
is, you can lie to protect the reporter. This is an option that can be used when the committee
receives a report that makes them fear for the safety of their community immediately, and does
not give them time to do the research on the harasser’s other behavior. This may not entirely
fool the harasser if the incident occurred just prior to the announcement, but it can give some
level of protection and plausible deniability to the target. Obviously, lying should be done
sparingly and only when there is no other option to protect both the community and the reporter.
It may seems shocking to advocate for lying in some cases, but we often prioritize one ethical
principle over another, and in this case, the safety of the reporter and the community takes
priority over being truthful.

Be cautious about lying in a public statement; this is another place that consulting legal advice
may make sense. In most cases, we recommend not naming the harasser in any public
statement.

Protecting the community's reputation
Many people worry that publicly acknowledging a code of conduct violation will harm the
community's reputation, especially with marginalized people. In reality, marginalized people
already know that harassment happens in all communities, and they are looking for a
community that handles harassment openly and in a forthright manner, instead of ignoring,
hiding, or downplaying them. Publicizing a situation in which the community swiftly and
effectively handled a code of conduct violation is usually a net positive for a community's

reputation. Trying to keep reports secret or refusing to comment on them publicly is usually a
negative mark or a danger sign for a community.

Here are a few examples of communities publishing "transparency reports" summarizing their
code of conduct violation reports and responses:

● PyGotham 2017 conference transparency report
● Djangocon.eu 2017 conference transparency report
● Write the Docs 2016 Prague conference transparency report

See the " Examples of responding to reports " chapter for commentary on their responses.
Safety is more important than privacy and confidentiality
Many communities value confidentiality and privacy, which when deployed properly are
important elements in creating a safer space. However, some communities take this principle
too far and prioritize privacy and confidentiality above the safety of the community. This might
look like:

● Keeping embarrassing information secret to protect the perpetrator, target, or community
leadership when the information is necessary to protect the safety of others
● Refusing to keep records of past violations to protect the privacy of harasers or targets
● Refusing to share information about previous relevant incidents with people responding
to a new incident

With rare exceptions, the safety of the community takes precedence over protecting the privacy
of an alleged harasser or the confidentiality of previous reports. When balancing the safety and
privacy of a target with the safety of the community, see " When individual safety conflicts with
community safety ."
Responses not to use
Some responses fail to prevent further harm to the community, and sometimes even directly
harm the target or the community. Some examples of these commonly proposed responses:

● Requiring the harasser to make an apology
● Asking the target to accept the harasser's apology or forgive the harasser
● Asking the target to help decide or agree with the consequences for the harasser
● Allowing the target to control the committee’s response
● Mediating between the harasser and the target
● Assigning people to watch or guard the harasser
● Allowing the harasser to remain in the community as long as they stay away from the
target

● Allowing someone who has not agreed to stop violating the code of conduct to continue
to interact with the community
● Doing nothing because the harasser is gone or unidentifiable

The next few sections go into more detail on why some of these responses must not be used.
Do not ask for apologies or forgiveness
A code of conduct committee must never ask for or compel an apology. An apology is only
relevant to a committee's actions when it is freely given. If the committee asks for an apology,
they lose important information: whether this person would have apologized on their own
initiative. When the committee is trying to judge how best to protect their community from harm
in the future, the knowledge that the alleged harasser has voluntarily made a sincere and
meaningful apology is important information. The knowledge that they apologized after being
required to do so gives them much less information.

Someone who voluntarily makes an apology that includes all of the following elements is much
less likely to harm your community in the future:

● Expression of remorse
● Explanation of their mistake
● Expression of compassion for the victims
● Acceptance of responsibility for the harm
● Explanation of how they will prevent a recurrence in the future
● Attempt to make amends for the harm they did

For more detail, see this article on the elements of a good apology by Amy Morin, summarizing
research published by Roy J. Lewicki, Beth Polin, and Robert B. Lount Jr.

A committee must never request that the target of harassment listen to or accept an apology
from the harasser. Often, people who have just been targeted by a harasser want nothing to do
with the harasser, even if the harassment was the result of a genuine mistake. Worse, many
harassers excuse and cover up further harassment under the guise of insisting on offering an
apology to the victim despite the victim’s obvious desire to avoid them. Refusing to respect a
target's desire to avoid contact with their harasser is a form of harassment in itself. Harassers
often continue threatening and coercing the target whenever they are in contact, even if they
claim they want to apologize. They may also use their own apology as a bargaining chip: “I
apologized, now you have to give me what I want."

The committee should consider the target's willingness to forgive the harasser as irrelevant to
the response they choose. The target's willingness or unwillingness to forgive is the product of
their own personality and experiences; it's not a good indication of how egregious the code of
conduct violation was or the likelihood of the harasser harming again. Many people forgive serial

sexual predators who repeatedly assaulted them and who are still assaulting others; others still
hold grudges for minor social slights that happened decades ago. Neither reaction is relevant to
deciding whether to allow the harasser to be part of the community. Also, systemic oppression
means that most people are conditioned to empathize more with the harasser than the target,
including the targets of harassment, which makes the target more likely to forgive the harasser.
An individual target's willingness to forgive must not be be a factor in the committee's response.
For more on this topic, see " Take Responsibility For Handling Abuse " by Annalee Flower Horne.

If the harasser does make a meaningful apology, it is reasonable for the committee to ask the
target if they would like to hear the apology without advocating for the target to hear it. If the
target does want to hear the apology, the committee can pass it on to them without expressing
any opinions of their own. The committee must never advocate for the harasser or pressure the
target to hear or accept any apology. We recommend delivering any requested apology in
written form, to reduce the stress and risk for the victim.
Do not ask the target to decide the response
It's not the target's job to come up with a solution to their harasser's behavior, or judge the
appropriateness of the committee's response. Resolving code of conduct violations is entirely
the responsibility of the committee. Simply being the target of harassment does not make the
target an expert in how to respond to that harassment. The target is likely already coping with
stress from experiencing the harassment (as well as possibly from reporting the harassment). If
the committee allows the target to help decide the response, others may blame the target for
whatever consequences the harasser experiences. In the worst case, this may result in the
target experiencing a second, more intense wave of harassment in retaliation for the original
report.

While the target should not be responsible for the committee's choice of response, it is often
reasonable to find out what the target thinks of the proposed response before formally
announcing the response. This is because a proposed response may reveal that the committee
is missing details, has an important misunderstanding of the problem, or isn't aware of the
potential for retaliation against the target and/or reporter. For more on this topic, see " What
Reporting Sexual Harassment Taught Me " by Simine Vazire.
Do not mediate
The committee should not attempt to mediate between the harasser and their target.
Approaching a code of conduct violation as something to be mediated changes the framing of
the problem from "What do we do about a harmful person in our midst?" to "Those people don't
like each other, oh well." By reframing a code of conduct violation as interpersonal conflict, the
committee is absolved from responsibility to take difficult action. Occasionally, a committee will
also demand some sort of feelgood conclusion in which the target forgives the harasser and no
one has to feel discomfort about the incident going forward—except for the target, who has
been not only victimized by the harasser but revictimized by the committee.

The committee's job is to protect the community as a whole from further harm, not to resolve
any dislike or conflict between any individual people. The committee's concern is, "Is this person
harming others?" and "What can we do to keep our community safe?" It is not their concern
whether the harasser and the target like each other or get along. Plenty of people in a
community dislike each other and have conflict; from the committee's point of view this is
irrelevant unless someone is breaking the code of conduct.
Do not guard the harasser or the victim
Don't ask other community members to follow and watch either the harasser or the victim to
prevent future harassment from a known harasser. Usually this solution is proposed when
powerful people are accused of harassment, or a committee is afraid of being attacked or
criticized for banning a person. This respone harms a community in several ways.

First, guard duty takes away the ability of at least one and usually several other community
members to participate fully in the community. Instead of doing what they'd normally be
doing—attending talks, doing their regular job, volunteering, networking with others—they are
following someone else around. This is a net loss to the community and a huge loss to the
community members tasked with guard duties.

Second, it makes other community members responsible for the behavior of the harasser. Their
job is to carefully watch the harasser or the victim and intervene (how?) to stop or prevent
further violations of the code of conduct. The guard may have to make the decision about
whether to risk their own personal safety to prevent someone else from violating the code of
conduct. Even in the ideal situation, in which the very next violation results in banning the
harasser from the community, the level of suspicion that justifies spending other community
members' time to guard someone is a high enough level of suspicion to justify at least a
temporary ban while the committee gets more information.

Third, it doesn't work. Read any serious spy novel and you'll realize that a 24hour tail operation
is difficult and expensive for even nationstates to successfully mount. Even if the guards do
manage to follow the harasser or the victim everywhere, what happens when they go into
private spaces, such as hotel rooms or closed offices? Many sexual assaults occur in hotel
rooms at conferences or during work trips; showing up at the door to a target's hotel room is a
common tactic for serial sexual predators. What do the guards do when the harasser is about to
enter their own or someone else's hotel room?

Guarding a community member to prevent harassment by someone suspected of being a safety
threat to your community is a waste of your community's time and energy. Instead ban the
suspected harasser, temporarily or permanently.

Do not ask the harasser to stay away from the target
Allowing the harasser to remain in the community on the conditionat that they stay away from
the target is rooted in the idea that a code of conduct violation is just a personal conflict between
two or more individuals, when it is actually a threat to the safety of the whole community. It also
presumes one of two things: (1) that the harasser is not a threat to the community, but the target
still feels frightened of the harasser and wants to avoid them, or (2) the target is not safe from
the harasser if they are physically nearby, and therefore the community is not safe (since the
target is part of the community).

In case (1), where the committee feels confident that the harasser is no longer a threat to the
community, the harasser should already be sensitive enough to the feelings of the target to
voluntarily avoid their presence until the target says differently. If the harasser doesn't realize
that one of the effects of their action is to make the target afraid of them, the committee should
be seriously concerned about whether they will do further harm to the community.

In case (2), where the committee does not trust the harasser not to harm the target if they are
nearby, this is a clear mistake by the committee. The goal of the committee is to protect their
community; the target is part of their community; the harasser is likely to harm the target again.
The harasser may harm other members of the community, or inspire others to harm when they
see how the committee responded. The committee should almost certainly ban the harasser
permanently.

If a harasser is not voluntarily staying away from the target and is then ordered to stay away, the
harasser may engage in a range of boundarytesting behavior, itself another form of
harassment. Efforts to impose a physical separation between a harasser and victim create
absurd situations: If both people are at an event, how far away is far away enough? 15 feet?
What if the harasser stands in the target's line of view but 20 feet away? What if they follow
them at exactly the required distance? If the harasser can't be in the same room as the target,
can they be in the doorway of the room? What about hallways, do they count as one room? For
example, a harasser in a speculative fiction community was told to stay away from a particular
author, but decided that showing up at the author's booksigning events qualified as "staying
away" if they did not ask the author to sign a book. A harasser might also use a separation
policy as an excuse to constantly seek out and demand information on where the target is at
any time, which is stalking and harassing behavior in itself.

Another problem with this approach is the heavy burden it places on the target. The target has
to continue being alert for the presence of the harasser, and if the harasser is already present in
a space where the target wants to be, the target either has to leave or ask the harasser to leave.
As a result the target is forced to mentally track the physical location of their harasser. The
target then has to go through the stress of reporting again if the harasser violates the
restrictions.

Holding powerful people accountable
Sometimes a community has one or more "rock stars" who are perceived as too important for
the community to lose. Often abusive people deliberately seek out and create situations in which
the community literally can't function without them. For example, only one person has the
password to administer the website, or only a few people have access to the training materials,
or the community is based on a myth that only one person has the "taste" or "vision" or "editorial
judgment" to keep it alive.

Because the abuser’s participation is perceived as crucial to the existence of the community, the
code of conduct can't be enforced on them, since the community is unwilling to enforce
consequences that they believe might destroy the entire community. If a community has a "rock
star" who likes abusing people, the community will have great difficulty creating an inclusive and
safe space of any sort. You can read more about identifying, preventing, and ending "rock star"
culture in " No More Rock Stars " by Leigh Honeywell, Valerie Aurora, and Mary Gardiner.

If a community cannot hold the most powerful people in the community accountable to the code
of conduct, it is best not to adopt a code of conduct at all. The next step in this situation is to try
to create a system that can hold the powerful accountable. This might look like:

● Replacing the board of directors
● Lobbying sponsors to withhold funding until a governance system is in place
● Pressuring powerful people to make public commitments to obey a code of conduct
● Creating petitions
● Founding or joining competing communities
● Mass resignations
● Forming unions
● Organizing strikes or boycotts
● Talking to the press
● Going public with stories of abuse

For more information on creating change within a corporation when you are part of an influential
group of employees with leverage, see " How to Organize Tech Workers to Change Company
Policy " by Liz FongJones.
Putting legal concerns into context

We do not give legal advice in this guide, but we can give some helpful guidelines on how to
integrate legal advice into designing and enforcing a code of conduct.

First, you should not take legal concerns seriously when they are raised by random people
without relevant legal training. They do not know the law and usually cannot offer you useful
advice; often they are freeassociating between what action they want you to take and anything
they have read or seen on TV that seems to support their goal. Trained lawyers usually do not
offer legal advice to people who are not specifically asking for legal advice.

Be wary of lawyers offering unsolicited legal advice; they may not have your community's best
interests at heart. Definitely don't listen to lawyers working for your opponents, or for people who
wish your community ill. If you do want legal advice, you should consult a lawyer with expertise
in the relevant field of law in your jurisdiction who has a legal duty of care towards you (e.g., you
are their client).

If you do consult a lawyer, remember that a lawyer's job is to tell you every single possible way
in which you could be exposed to legal liability. It is not the lawyer's job to balance legal
concerns with the rest of the goals of your community, which may include accomplishing your
mission, making money, or encouraging collaboration. A lawyer is concerned with protecting
their clients in a narrow legal sense and usually doesn't have expertise in all the other nonlegal
risks to a community, such as harms to your public reputation or to your members. Legal
concerns are only one of many aspects that a community leaders need to consider when
making decisions.

A common mistake is to only consider legal risks on one side of an issue, and not the other side.
For example, say you have a strong reason to believe that someone is likely to sexually assault
people at your event. Various people in your organization may be deeply opposed to banning
this person from the event, so they raise concerns about defamation or libel, and propose
instead having several conference staff guard the person. If you were to consult a lawyer, they
might inform you of the risks that come from not banning the person, which might include
increased liability if that person does assault someone, with your assignment of a guard used as
evidence that you were aware that the person was dangerous. (For more on why guarding
someone is a bad idea, see the section " Do not guard the harasser or the victim .")
Responding to incomplete or late reports
Sometimes code of conduct reports can't identify the person who violated the code of conduct.
Other reports might come in after an event is already over or after the harasser has left the
community. A community can still respond to these reports in a meaningful way.

For reports without an identified harasser, the committee can publish a description of the
incident and say what sanction they would have imposed if they had identified the harasser. This
approach has several positive effects: it educates the community on what is acceptable, it
encourages other victims to come forward in the future, it will discourage the harasser from
repeating the behavior, and it validates the reporter and victims. It may also cause other people

to come forward with reports of similar behavior, or other people may be able to identify the
harasser based on the description. To lower the chance of an unidentifiable harasser, follow the
recommendations in the section " Identifying community members ."

When the event is already over or the harasser has already left the community, the committee
can ban the harasser from future events, tell the organizers of related communities about the
incident so they can take action, and publish a description of the incident saying what they
would have done if they had received the report sooner. They can also apologize to the target
and make changes to prevent a repeat of the incident.

See also:

● Identifying community members
Investigating the incident
A code of conduct committee rarely has the expertise or resources to conduct a full background
check or act as private investigators, and must not hold itself to a standard of certainty
comparable to that required by the formal legal system. In general, the committee should make
decisions based on the preponderance of evidence—what explanation seems most likely to be
true.

Most incidents will not be complicated. In cases of severe violations of a code of conduct, often
the harasser has committed many other violations that have not been reported—sometimes at
the same event. For example, someone who has attempted to sexually assault someone at an
event may have also made multiple unwanted sexual advances and sexual jokes to other
people at the same event. Similarly, someone who made a racist joke at a conference probably
has made racist comments on publicly available social media, requiring only a few simple
internet searches to find. We've seen instances where someone reported for violating the code
of conduct had already been extremely rude or threatening to multiple organizational staff. For
more on this phenomenon, see " The Al Capone Theory of Sexual Harassment " by Leigh
Honeywell and Valerie Aurora.
Impact is more important than intent
Committees can easily spend too much time and energy arguing about the intent of the
harasser, asking questions such as:

● Did the harasser realize what they were doing was harmful?
● Was the harasser intending to be helpful?
● Should the harasser have known better?
● What life experiences or conditions led the harasser to act this way?

The following questions are far more important:

● What impact did their behavior have on the target and the community?
● How can the committee prevent that harm from happening again?
● What will help the target and the community feel safer going forward?
● What harmful life experiences is the harasser creating for people in your community?

When the committee focuses too much on the intent of the harasser, it recenters the discussion
on the feelings and reputation of the harasser, rather than centering it on the feelings of the
target or the rest of the community. For more on this subject, see " How 'Good Intent'
Undermines Diversity and Inclusion " by Annalee Flower Horne.

Keeping that caution in mind, the intent of the harasser is one of several factors for judging
whether the harasser poses a threat to the community in the future. If the harasser intended to
cause harm, they are more likely to harm the community in the future. If the harasser had
positive intent, and they recognized their mistake, took responsibility, and promised to take
steps not to repeat the mistake, then they are less likely to harm the community in the future.
Conversely, if the harasser had positive intent, but does not admit they made a mistake, does
not take responsibility for the harm their action caused, and refuses to change their behavior,
they are also more likely to harm the community in the future.

It's also possible for a harasser to intend to cause shortterm harm for some longterm imagined
good ("toughening them up," "everyone likes teasing," "I was just complimenting them"). They
might change their mind after someone explains the harm of their actions to them, and become
less likely to harm the community in the future. As you can see, intent is less important than
impact.
Distinguishing good intent from bad intent
When trying to figure out a person's intent, you should rely more on concrete actions and
longterm patterns of behavior than on their words or the action they take when they are facing
serious consequences. Serial predators are, by necessity, good at manipulating people in
power. Many serial predators have figured out a plausible story to excuse their behavior, but
their actions stay the same. They are likely to:

● Groom people with less power to accept their abuse
● Put themselves into situations where it would be easy to abuse someone
● Push back on any attempt to restrict their access to targets
● Say they accept responsibility for their actions, but object to any real consequences
● Appear to accept the consequences but then ignore them
● Continue to test and push new or existing boundaries

When the abuser is telling a plausible story to excuse their behavior, they will often minimize the
intensity or frequency of their harmful behavior, or suggest that everyone does what they did, or
wants to do it. By contrast, a person who accidentally inflicted harm and wants to stop harming
people may or may not be articulate or convincing in what they say to the committee, but their
actions will show they are more concerned about the people they harmed than about their
reputation or continued access to targets. If someone genuinely wants to change, they are likely
to:

● Have a plausible explanation for their behavior (even if it's just "I wasn't thinking")
● Recognize the harm they caused
● Not seek to minimize or normalize the harm they caused
● Express concern about the targets without prompting
● Eagerly seek out the root cause of the problem and identify ways to fix it
● Suggest meaningful and effective sanctions for themselves
● Carefully adhere to any sanctions in the future
● Avoid anything that looks like boundary pushing
● Offer a meaningful apology—without insisting that the target hear it
● Seek ways to make amends without further harming the target
● Proactively respect the wishes of the target about future contact
● Avoid situations that would make repeating their mistake more likely

When you look at their response to mistakes they made in the past, you will see this same
pattern.

Case study: A man volunteered for a program teaching women to code. He frequently
contacted students outside of class without their consent, making many of them deeply
uncomfortable. If they pushed back on his behavior, he would blame his inability to read social
cues (see the section on " Social awkwardness and harassment " for more on this topic). When
called out on this behavior, he apologized, but refused to change his behavior or step down
from his position of power teaching women to code. We believe that someone who actually
supports women in tech would, after realizing he frequently makes women uncomfortable,
avoid direct contact with women in tech and find different ways to support women until he has
learned how not to make women uncomfortable. This pattern of behavior indicates a serial
predator, not someone who supports women in tech but has poor social skills.

Many predators are good at imitating victims and manipulating people in power, and will make
the most of any source of power or privilege they have. A predator may even have read this
guide and be using it to help them lie to a code of conduct committee or make up false code of
conduct reports! We recommend relying more on longterm trends of behavior and relative
positions of power than behavior while the committee is actively investigating to help avoid this
kind of deception. A predator can successfully act like a good person for a few days or weeks,
or while speaking to people who have the power to harm them. But what is the point of being a

predator if they act like a good person all the time, to all people? Often it's how someone treats
the least powerful, especially in private, that shows the truth about a predator.
DARVO: Deny, Attack, Reverse Victim and Offender
Serial predators will often use a technique called DARVO, which stands for "Deny, Attack,
Reverse Victim and Offender ." Here is the short definition from the originator of the term, Dr.
Jennifer J. Freyd:

DARVO refers to a reaction perpetrators of wrong doing, particularly sexual offenders,
may display in response to being held accountable for their behavior. DARVO stands for
"Deny, Attack, and Reverse Victim and Offender." The perpetrator or offender may Deny
the behavior, Attack the individual doing the confronting, and Reverse the roles of Victim
and Offender such that the perpetrator assumes the victim role and turns the true victim
or the whistle blower into an alleged offender. This occurs, for instance, when an
actually guilty perpetrator assumes the role of "falsely accused" and attacks the
accuser's credibility and blames the accuser of being the perpetrator of a false
accusation.

DARVO is so powerful in part because the predator is presenting themselves as the victim, but
often from a position of more power and privilege than the true victim. As a result, most people
automatically focus more on the predator's feelings, are more likely to believe what they say,
and are more afraid of angering them. Fortunately, learning about DARVO makes people less
likely to believe predators using DARVO.

"Deny" is the first step of DARVO. Many people can't imagine anyone lying as brazenly as
predators do, and assume that there must be some truth in the predator's denials, or at least
that any disagreements about what happened are the result of a genuine misunderstanding. If
you are on a code of conduct committee, you must be open to the idea that someone is
deliberately lying to you in a convincing manner, and that you are not as good at spotting lies as
you believe you are.

Both true victims and predators may deny accusations against them, but they tend to do so in
different ways. If someone claims that the accusations against them are false, but has taken
steps to defend themselves against them before they were informed of the accusations, that is
evidence in favor of the accusations being true. This is because the most likely explanation for
the perpetrator knowing what to prepare for is their own memory of the abuse they committed.

For example, if someone is accused of sexually assaulting a woman 35 years ago, and within
hours produces a letter signed by 65 other women from that period of their life vouching for the
accused's respectful treatment of women in general, the letter is strong evidence that the
accusation is true. This is both because the rapid response makes it likely they prepared it
before the accusations were made, and because the defense relies on an element of rape

culture, which (wrongly) asserts that people who commit sexual assault against women treat all
women badly.

The "Attack" portion of DARVO usually looks different from a true accusation. A predator using
DARVO is more likely to try to discredit the accuser with sweeping personal attacks and
emotionally loaded words, such as, "She's deluded! Only a crazy person could think I did that!"
or "His accusations are disgusting and cruel. I would never say something racist!" Their goal is
to not only to deny the accusation, but dehumanize and diminish the accuser across the board.
True victims may also engage in personal attacks, but are more likely to focus on factual
statements and descriptions of the abuser's actions, such as, "He called me into the other room
and pushed me against the wall and kissed me without my consent," or "While we were sitting
next to each at dinner on Tuesday, she leaned over and whispered a racist comment in my ear."

Be careful to distinguish between harsh but accurate statements of facts and personal attacks.
For example, "He's a racist," is not a personal attack, it's a factual description summarizing a
person's behavior, as are "He's a creep," and "She's a habitual liar." A personal attack from an
abuser will be more sweeping and dehumanizing, such as, "She's completely hysterical and no
one should listen to her about anything," or "He's always bringing up racism because he's
obsessed with being a victim."

"Reverse victim and offender" often involves a false accusation against the victim. False
accusations are more likely when the more powerful person is claiming to be the victim, since
there is little or no risk for the powerful person to accuse a less powerful person, and enormous
risk with little gain for a less powerful person to accuse a more powerful person. Sometimes the
accusation will be ludicrous outside of the context of rape culture or other forms of oppression,
such as claiming that a young sexual assault victim seduced an adult abuser. Be extremely wary
when the more powerful person presents themselves as the true victim. (For more detail on the
differences between true and false rape accusations, see " What kind of person makes false
rape accusations? " by Sandra Newman.)

DARVO involves a lot of lying and deception. When this happens, we recommend paying
particular attention to the opinions of marginalized people in your community. They usually have
more experience with predators trying to harm them and are familiar with the patterns of
manipulation predators use. Marginalized people are more likely to pick up subtle
inconsistencies in the predator's story or demeanor; even if they can't consciously identify the
inconsistencies, they may have a subconscious sense of something wrong.

Case study: As I child I had a friend who had long shiny hair that I openly admired. My friend
would occasionally criticize her hair and look sad; in response I would spend several minutes
trying to convince her how beautiful her hair was. One day as I was doing this, she briefly
smiled and then quickly changed her expression back to sadness. I instantly realized she was
lying to me to get compliments, and had been doing so for years. I stopped being friends with

her soon after. After that experience, I was more alert to small signs of good moods in people
when they claimed to be sad.

We don't recommend using "deception detection" techniques unless you are fullytrained expert
aware of their many limitations and potential pitfalls. Many popular ideas about the signs that
someone is lying have been debunked . Other techniques, such as reading microexpressions,
only reveal underlying emotions, not the reason for the emotions . Even a fullytrained expert in
these fields is wrong a significant percentage of the time. Instead, we encourage you to pay
attention when someone's intuition is warning them about a person. Intuition can be wrong and
it can be affected by bias, but as Gavin de Becker puts it in The Gift of Fear :

Intuition is always right in at least two important ways.
1. It is always in response to something.
2. It always has your best interest at heart.

Combined with efforts to mitigate bias, knowledge of people's past behavior, and awareness of
the systemic oppression at play, the intuition of marginalized people is valuable to your work as
a committee.
Judging competing claims of marginalization
Sometimes a harasser will claim that they can't have harassed someone because the harasser
is themselves a member of a marginalized group (sometimes with absurd definitions of a
marginalized group, such as "people who think white people are genetically superior to people
of color"). They may be part of a marginalized group that is stereotyped as incapable of
harassment, such as a lesbian woman claiming she can't sexually harass a man (since sexual
harassment is about power, it isn't necessary for someone to be sexually attracted to someone
to sexually harass them). They may claim that their behavior would be acceptable from a more
privileged person, such as someone claiming that their behavior is only being treated as sexual
harassment because the harasser is considered to be old, fat, ugly, or otherwise sexually
unattractive. They may even claim to be the true victim (see the section on DARVO ). The
harasser may also deny the committee's right to enforce the code of conduct because they are
more marginalized than one or more of the committee members or community leadership.

These claims are complicated because the committee needs to avoid two different mistakes,
each pulling in a different direction. The first mistake (and the more common in everyday
society) is to have more empathy for the more privileged person, to be more ready to believe
them, to be more accepting of their claims, and to be more lenient towards them. This goes
handinhand with a tendency to punish members of marginalized groups more harshly for the
same behavior. For example, in the Linux kernel programming community, women were
criticized for occasionally using mild profanity at the same time that male leaders in the project
were praised for heavy use of profanity and personal insults. The second mistake is to

overcorrect for the first mistake by allowing more abusive behavior from a person because they
are a member of a marginalized group.

The best technique here is to stay in touch with objective facts about the situation and focus
your empathy and concern on the entire community, while keeping the needs of the most
marginalized members of your community uppermost. A person may claim that, as a member of
the supposed marginalized group, "men who think women are genetically inferior," if you enforce
the community’s code of conduct you are discriminating against them. Ask yourself, "What effect
does this person have on the safety of my community?" and "What does the research say on
whether this group is marginalized?" Quickly you'll realize that this person harms the safety of
pretty much every marginalized group in your community (see the " Paradox of Tolerance "), and
that men who think women are inferior quite often hold positions of great power in our society.
The answer in this case that this person is not a member of a marginalized group, and you do
not have to take their claims of discrimination and silencing seriously.

Another person might say to you that, as a woman of color, she can't be held accountable for
making homophobic comments. Ask yourself, "Are we enforcing this rule equally for white
people and men of all races?" and "What effect does this person have on the safety of my
community?" You may realize you need to step up enforcement on more privileged people if you
haven't been enforcing this rule for them, and you will also realize that if someone believes the
code of conduct doesn't apply to them, they will harm your community. The answer is that no
one can make homophobic comments without being held accountable, whether they are a
woman of color or not.

Another complicated situation is when a member of a marginalized group claims that they
cannot be doing things that are harmful to their own group, such as Latinx person engaging in
antiLatinx racism, or a gay person making homophobic comments. It is important to protect
" gallows humor "—dark humor employed by members of a marginalized group to lighten the
mental burden of their oppression—but that is separate from and different than reinforcing
oppression against a person's own marginalized group. Again, we recommend first ensuring
that more privileged members of your community are being sanctioned for this behavior and that
marginalized people are not subject to a higher standard of behavior.

Competing claims for who is truly the victim in a situation should never be settled by engaging in
“ Oppression Olympics ”—arguing over which person is more marginalized or has it worse, with
the idea that the more marginalized person "wins" the status of victim.

Someone's membership in a marginalized group can be taken into account in the following
ways:

● Reminding yourself to examine and correct any bias you or your system may have
against that group
● Judging what kind of risk the person is taking by reporting harassment

● Taking into account the prevalence of oppression against that group when trying to
decide what really happened
● Comparing how similar actions by more privileged people are handled
Social awkwardness and harassment
Many communities include people who are socially awkward: they might have difficulty
perceiving social cues, identifying emotions in other people, or following social rituals. The
causes of social awkwardness are many, and may include anxiety, lack of opportunity to
socialize, socialization in a different culture, being the target of abuse, and some kinds of
neurodivergence, including autism. We do not include uncontrollable violent behavior in our
definition of social awkwardness.

We recommend that committees treat incidents with socially awkward harassers
substantially the same as incidents with harassers who have average or higher social
skills. If the harasser is genuinely socially awkward, the process might take more time and the
solution might need to be explained or simplified. The committee may need to explain the
problem in more detail, the harasser may ask more clarifying questions to understand what they
did wrong, or the recommended change in social behavior might be need to be broken down
into smaller pieces so it is easier for the harasser to implement. For example, a committee might
tell a person with high social skills, "Don't follow people around," but tell a socially awkward
person, "Do not spend more than 30 minutes a day in the company of another attendee unless
they explicitly invite you individually to join them." Otherwise, the process should be the same.

Some people argue that socially awkward harassers should not be held accountable to the code
of conduct in the same way as people with average or higher social skills. We'll explain those
arguments and why we disagree with them in the rest of this section.

The argument for treating socially awkward harassers differently looks something like this: the
harassment is caused by the harasser misunderstanding social cues due to their social
awkwardness, they can't change their actions to be less harmful, and it isn't fair to hold the
harasser accountable to the code of conduct for something they can't control. In other words,
the person doing the harassment is the marginalized person, their comfort and happiness
should be prioritized over their that of their victims, and the harasser cannot change their
behavior, so other people should just accept harassment as the cost of including socially
awkward people in their community.

To start with, this argument ignores the possibility that the target is also socially awkward.
Socially awkward targets are at a disadvantage in defending themselves from harassment, and
are sometimes targeted precisely for that reason, including by other people who claim to be
socially awkward. Understanding that socially awkward people are more likely to be targets
increases the importance of enforcing the code of conduct equally against harassers who use

social awkwardness as an excuse. Exempting people who claim to be socially awkward from the
code of conduct will harm socially awkward people more than it helps them.

Another false assumption in this argument is that socially awkward people are more likely to be
harassers than people with average or higher social skills, stereotyping socially awkward people
as abusive and harmful. In our experience, socially awkward people are far more likely to be the
target of harassment than the perpetrator, as is the case for most marginalized groups. It is
possible for someone who is socially awkward to harass others but harassment is not the
inevitable result of their social awkwardness.

A second false assumption is that a socially awkward person doesn't have control over their
harassment of others because they can’t perceive subtle social cues at the same speed and
detail as someone with average social skills. This cannot be true, since the majority of socially
awkward people are generally successful in not harassing or assaulting others in their
communities.

While difficulty perceiving social cues can slow down or complicate social interaction, it does not
inevitably produce harassment. Socially awkward people who do not want to harass others often
reduce the chance of accidentally harassing someone by consciously following selfimposed
guidelines, such as:

● Asking explicitly about the other person's feelings
● Asking for clarification of ambiguous statements
● Periodically checking in to see if they missed a social cue
● Always asking for permission before touching other people

One clue about whether a person genuinely doesn't realize their behavior is harmful is if they
engage in the same behavior in the presence of people who are more powerful than they are. If
they only act inappropriately around less powerful people, then they are abusive people
pretending not to understand the impact of their actions and choosing to act inappropriately
when they believe they will not be punished for it.

A third false assumption implied by this argument is that socially awkward people cannot learn
new behaviors—which should seem ridiculous and patronizing when stated outright. If a social
skill is simple enough that most sevenyearold children can learn and follow it most of the time,
then most socially awkward adolescents and adults can also learn and follow it. Like many
people with average or higher social skills, many socially awkward people who want to avoid
harming others can accept feedback, apologize, make amends, and work hard to learn new
behaviours when they do harm. Difficulty perceiving social cues does not imply inability to learn
new social behaviors.

A final false assumption is that a socially awkward person cannot deliberately abuse others, and
any abuse they perpetrate must be accidental. While the vast majority of socially awkward

people do not want to be abusive, some socially awkward people do intentionally engage in
abusive behavior, just like people with average or higher social skills.

Some people will use this argument to claim that all neurodivergent people and especially
Autistic people are more prone to harassing behavior and can't be held responsible for
harassing others, since neurodivergence and autism can include difficulty perceiving social
cues. Some people even make the argument that most harassment is perpetrated by Autistic
people, and that enforcing a code of conduct is therefore ableist and antiAutistic. These
arguments are themselves ableist; they simultaneously infantilize and demonize Autistic people.
For more on this issue, including several articles by Autistic authors, see this summary .

Sometimes, people who have average or higher social skills will selfidentify or be described as
socially awkward to excuse their harassing behavior. Social awkwardness is varied in its
presentation and cause, and in many cases, no one can definitively prove or disprove whether
someone is socially awkward or has a specific cause of social awkwardness. Questioning
someone's selfidentified neurodivergence or mental disorder can be a form of ableism; at the
same time, falsely claiming to be neurodivergent or have a mental disorder for the purpose of
escaping consequences for abusive behavior is coopting a marginalized identity for personal
gain. Behavior isn't a foolproof guide: many people are capable of intentionally acting socially
awkward when they actually have no trouble perceiving social cues. Some people do this for a
living, including some actors or comedians.

Some people take advantage of this ambiguity and the simultaneous infantilization and
demonization of neurodivergent people to give themselves a convenient excuse for abusing
others. These abusers falsely claim or suggest they might be neurodivergent in some way,
perhaps as part of the "Reverse Victim and Offender" element of the DARVO technique .
However, it is unnecessary to determine whether a harasser is genuinely socially awkward or
only pretending to be, as the committee should treat socially awkward harassers and harassers
with average or higher social skills the same, with the exception that a socially awkward
harasser may benefit from slightly longer or more detailed discussions.

Case study: The leader of the Linux kernel software project, Linus Torvalds, personally
attacked other members of the Linux community on a regular basis, going so far as to publicly
ask why another developer had not been "retroactively aborted" for stupidity. After more
media attention to this behavior than usual, Torvalds stepped down temporarily from his
position as project leader in 2018. His email announcing the change explained his abusive
behavior as the result of a "lifetime of not understanding emotions" and said he was taking
time off to "get some assistance on how to understand people’s emotions and respond
appropriately." Reviewing Torvalds' abusive emails makes it clear that he understands his
targets' emotions quite well, he just enjoys making other people feel humiliated and shamed
and did not have any incentive to stop doing so until recently. Claiming that he didn't
understand the emotions he was creating is an attempt to avoid accountability for his actions.

Note that any harasser may pretend not to understand the committee to delay or prevent being
held accountable. To avoid this, pay attention to how well the harasser understands similar
discussions in other contexts. If they can understand the same concepts when it is to their
advantage to understand them, then they are likely faking incomprehension. Another clue is
whether the harasser acts more appropriately in some circumstances, such as when other
people are present, or when the potential target is powerful. In other words, are they socially
awkward when they might face serious consequences for their actions, such as losing a job or
losing money?

In our experience, claims of social awkwardness or misunderstanding are most often mobilized
to excuse the behavior of white men who have sexually harassed or assaulted others. It is a
version of the "Male Bumbler" myth —the idea that otherwise competent and successful men are
inexplicably incompetent in one specific area (judging the sexual receptiveness of others) and
therefore can't be held accountable for sexually harassing other people. When it comes to race,
the same behavior might be viewed as "socially awkward" in a person of one race, and "violent"
or "dangerous" in a person of another race. Overall, the excuse that someone can't help
harassing others because they are socially awkward is more available to people with more
privilege and less available to people with less privilege. As a committee, you must not accept
this excuse.
Mental health and harassment
Mental health conditions, such as anxiety disorders, personality disorders, and bipolar disorder,
are not excuses for violating the code of conduct. Much like social awkwardness, some people
will argue that someone with a mental health condition is incapable of participating in a
community without violating a code of conduct, and therefore requiring members to follow a
code of conduct is ableist.

As in the case of social awkwardness, enforcing a code of conduct on all community members
will protect more people with mental health conditions than it harms. In the majority of cases, a
person with a mental health condition is just as capable of following the code of conduct as
anyone else. In the relatively rare case that a person's mental health condition makes it
impossible for them to follow the code of conduct, then protecting the community as a whole
takes precedence over including that person in the community.

No community of the type addressed in this guide has a duty to include or care for every person
in existence. Some people require trained professionals and specially designed supportive
environments before they can participate in a community without significantly harming others.
Your community does not have to include people who regularly harm other members of the
community, regardless of the state of their mental health.

It is likely that the harasser and/or other community members will argue that your community
does have an obligation to help manage the mental health of a harasser by allowing them to
stay in your community and harm others (you don't have an obligation to do this). To help you
identify when this is happening, here are some examples of what that looks like:

● Hints that the harasser may selfharm in some way if forced out of the community
● Claims that the harasser doesn't have a support network outside of the community
● Crediting participation in the community as the only thing preventing harm to self or
others by the harasser
● Sharing the harasser's fantasies about harming self or others followed by gratitude to the
community for preventing them
● Open claims of the form "The harasser has XYZ condition and therefore you cannot
enforce the code of conduct against them or make them leave"

Your community should not take on sole or partial responsibility for treating someone's mental
health condition for a simple reason: mental health conditions should be treated by mental
health professionals. If the harasser does not have access to mental health professionals, you
should not conscript your community members into service as untrained caregivers.

People who aren't mental health professionals often don't know what to do and can harm the
patient, themselves, and others. "Allow this person to stay in our online chat channel without
consequences for harming others" is not mental health treatment, it may hurt the person
needing treatment, and it will harm many others, including possibly other community members
with mental health conditions. It is unethical to attempt to treat someone without appropriate
training and the informed consent of the person being treated. Resist the argument that caring
for someone with mental health means allowing them to violate the code of conduct without
consequences.
Children, caregivers, and harassment
If a community wants to include children and their caregivers, it should enforce an
ageappropriate code of conduct on all children. It should also hold caregivers responsible for
supervising their children and for getting their children to follow the code of conduct. If children
will be cared for by people other than their primary caregivers, a system should be arranged in
advance so that community members don't end up caring for children against their will or
without preparation, which is more likely to result in harm to the child and to others. For
example, if a caregiver brings a child to a community space, does not supervise them properly,
and other members step in to supervise in an ad hoc manner, then the space is not safe for that
child, or for anyone who could be harmed by an unsupervised child.

Enforcing an ageappropriate code of conduct on children will make the community more
inclusive of children and caregivers for children. For example, if a community encourages
bringing children to the space to make it easier for caregivers to participate, and one of the

children physically attacks the other children on a regular basis, then other caregivers will not be
able to safely bring their children to the space. The net result is that caregivers and children are
less able to participate.

If the child and the caregiver can't be separated, then the consequences for violating the code of
conduct will often affect both the child and the caregiver. For example, if a child won't stop hitting
other children at an event, and the committee tells the child to leave, and the child can't leave
without the caregiver, then both the caregiver and the child will need to leave.

These principles also apply to dependent adults or elders who need supervision at a level
similar to a child. Caregivers for adults who are not dependent are not responsible for the
conduct of the person they are caring for, and vice versa, but if the adult and the caregiver
cannot be separated, they may still be affected as a pair by code of conduct enforcement. For
example, if an independent adult's caregiver violates the code of conduct, and the penalty is a
permanent ban, and the caregiver and the person being cared for cannot be separated, then
caregiver will also have to leave the community space.
Sexual behavior and communities
Many difficult issues around community management come up when sexual behavior is part of
community interaction, as is often the case. It may be difficult to distinguish between sexual
behavior that should be tolerated and included, and sexual behavior that is abusive and should
not be tolerated. This topic could fill an entire guide by itself, but we'll briefly discuss some of the
issues that come up around sexual behavior when enforcing a code of conduct.

First, some background. Some communities are set up for the purpose of engaging in sexual
behavior, some communities are places where people normally engage in sexual behavior with
other community members, some communities strongly discourage members from engaging in
sexual behavior with each other (such as in many workplaces), and some communities are for
people who do not want to engage in sexual behavior at all. Sexpositivity is an attitude of
acceptance and support for people fulfilling their own unique, consensual sexual desires (or lack
thereof). Sex work is work and sex workers should be treated with respect and care; while some
sex work is coerced, that does not change this basic principle (after all, many forms of work are
coerced). Kinkshaming is shaming someone for having a sexual preference that is viewed as
abnormal or taboo, such as bondage or roleplaying as animals. For the purposes of this guide, a
sexual preference is only a kink if it involves only consenting adults and does not dehumanize or
sexualize anyone without their consent. Sexual preferences that rely upon assault, harassment,
invasion of privacy, racism, sexism, or other forms of exploitation are not "kinks," they are
abuse. For more on this topic, see " Acceptance, kink shaming, and calling out bad behavior " by
Anabelle Bernard F.

A common misconception about sexpositivity is that any sexpositive adult should be willing to
discuss sex, observe sexual behavior, or receive a sexual proposition at any time, in any

context, with any person. In actuality, sexpositivity includes respecting the consent of all
participants for any sexual activity, including activities such as talking about sex or observing
sexual behavior (which is participating in sexually related acts, just without physical contact).
True sexpositivity means that if sexual behavior or discussion is going to happen in your
community, everyone who might be present should be aware of it in advance and able to make
a decision about whether or not they want to participate, free of coercion, pressure, or
mindaltering drugs (including alcohol). Sexual activity on mindaltering drugs is risky but may
be consensual if several precautions are taken, such as obtaining consent when the participants
are sober, making sure that participants are able to withdraw consent, and respecting the
withdrawal of consent while on the drugs.

For example, if an event will have a room devoted to BDSM (bondage/discipline,
dominance/submission, and sadism/masochism) activities, we recommend the following
precautions:

● The event's primary purpose must be directly related to the BDSM activities
● Attendees must be notified in advance of the BDSM activities
● The room for the BDSM activities must be clearly labeled
● People must be able to participate in the rest of the event without entering or passing
through the room
● People must not be able to accidentally see or hear the BDSM activities in the normal
course of attending the event
● There must be rules governing the activity in the room that strongly ensures everyone is
participating freely and safely (establishing safe words, rules on intoxication, limitations
on types of play, etc.)

Some people believe that a sexual proposition is always acceptable in any context, as long as a
direct "no" is respected. However, this belief ignores the reality of systemic oppression, power
dynamics, and abuse. Systemic oppression means that marginalized people are often punished
for saying a direct "no" to sexual propositions and other requests. Rejecting a sexual proposition
is stressful due to this implicit threat of punishment. As a result, marginalized people are more
likely to say yes to a proposition when, if acting in complete freedom, they would say no.
Marginalized people often use a " soft no " instead of a direct no to avoid punishment for refusing
the request, which some people deliberately ignore and continue pressuring the person to
accept. Sexual propositions are also frequently used as a form of harassment rather than as a
genuine proposition for sexual interaction, as can be seen in street harassment of women
around the world.

Power dynamics mean that some people have more ability to punish or retaliate against
someone for saying no to a sexual proposition, again leading to less powerful people being
more likely to be coerced into sex they don't want. Abuse means that some people have been
trained in various ways to fear saying no to sex and no in general, are triggered by sexual
propositions, or are currently in an abusive sexual relationship where true consent is impossible

because they will be punished for saying no. A more meaningful definition of consent than "they
said 'yes'" is:

1. They feel entirely comfortable and safe in saying "no," and
2. They have the information and mental capacity they need to make the decision that is
best for them

Overall, allowing sexual propositions in all situations will put marginalized members of the
community at a huge disadvantage, as they are forced to spend enormous amounts of energy in
avoiding or deflecting propositions, escaping sexual assault, or dealing with the effects of sexual
assault.

To account for systemic oppression and power dynamics, all communities must have rules
around the time, place, and frequency of sexual propositions. Some communities will have a
rule of no propositions ever; others may have a rule of once and then do not repeat; others may
limit propositions to certain times and spaces; others may have explicit signals saying whether a
person is open to propositions of a particular type.

When deciding on limits to sexual propositions, the community must place the purpose of the
community first and limit propositions in such a way that they don’t harm or detract from the
purpose of the community. For example, a martial arts club run by a man who propositions all
the women who attempt to join the community will create a community that is unfriendly and
unwelcoming to women and likely many other marginalized groups. Even communities which
are formed for the purpose of engaging in sexual activity will harm marginalized members if the
sexual desires of the more powerful members are prioritized above those of the less powerful.

Case study: S ex parties organized by Silicon Valley venture capitalists are carefully designed
to give powerful, older, richer, male venture capitalists a structural advantage over less
powerful, younger, poorer, women who work in the same industry. The men organizing these
parties do this by inviting more women than men, only inviting rich men, only inviting younger
and less powerful women, not allowing sexual activity between men, pressuring women to
take drugs that lower inhibitions, inviting women who work in the same industry as the men,
and suggesting that women's careers will be helped by attending and hurt if they don't attend.
In reality, the women attendees' careers are hurt whether or not they attend, and the sex
parties reinforce the existing oppression of women in venture capital.

We don't recommend using examples of existing oppressive environments as a guide for how to
deal with sexual behavior in your community. For example, many bars and clubs aimed at adults
looking for heterosexual relationships are also places where people are sexually harassed,
assaulted, or drugged. Your community must not use that environment as a standard when
deciding how and when sexual propositions are allowed.

Sometimes, people defend sexually abusive or nonconsensual behavior by accusing people
who object with kinkshaming. Actual kinkshaming is shaming people for sexual preferences
involving only consenting adults. The consent has to include anyone who is observing the
sexual activities. If the kink involves dehumanizing anyone, then all the people being
dehumanized need to consent as well. Without this consent, the behavior is not kink as we are
using it in this document, it is abuse and/or bigotry. For example, if someone claims their "kink"
involves exposing their genitals to nonconsenting observers, it is not kinkshaming to loudly
object to this behavior.

Another example of a false kinkshaming accusation is someone claiming "You are
kinkshaming if you object to me wearing my sexually explicit costume here," where "here" is an
area filled with people not all of whom have agreed to participate in sexually explicit cosplay.
Another example is someone saying "You are kinkshaming if you object to me engaging in
BDSM dominance play with my slave," when the play is occuring in a public area that includes
people who have not agreed to observer BDSM dominance play. In both cases, the bystanders
have not agreed to be part of this person's sexual play in the role of observer.

Another example of a false claim of kinkshaming is someone whose "kink" involves racism or
sexism—anything based on the idea that some people are inferior based on their identity—and
tells others they have to accept their "kink." "My kink is that this group of people is subhuman" is
not a way to get out of rules against racism, sexism, or any other nonconsensual
dehumanization of a group of people. For more on the use of false accusations of kinkshaming
to protect bigotry or abuse, see " Acceptance, kink shaming, and calling out bad behavior " by
Anabelle Bernard F.

To avoid improper bias or kinkshaming, it is important to apply the same rules across both
mainstream, privileged, accepted sexualities (such as heterosexual serial monogamy) and less
privileged sexualities (such as asexuality, homosexuality, bisexuality, or polyamory). For
example, if your code of conduct bars a samesex couple from kissing in public, it should also
ban malefemale couples from kissing in public. If your code of conduct bars a person in a D/s
relationship wearing a collar to mark themselves as part of a longterm relationship, it should
probably also forbid wearing wedding rings. The same rules should apply across people with
more or less privilege as well. If it is against your code of conduct for a man to wear nothing but
a tiny revealing bikini bottom, it should also be against your code of conduct for woman to wear
a similarly revealing bikini in the same spaces. If someone objects to a fat person wearing a
short skirt, thin people should also not be allowed to wear short skirts.

When making comparisons, keep in mind that many "accepted" practices in mainstream
heterosexuality are abusive and must not be used to justify accepting similar practices for other
sexualities. For example, many people find it acceptable for little boys to physically assault little
girls because "he has a crush on her," but mainstream acceptance of this abuse should not be
used to justify a gay man to stalking his male partner because he is in love with him.

Some people use kink practices to cover up abusive behavior in relationships, and it can be
difficult for others to distinguish between normal kink practice and abusive behavior. The
National Coalition for Sexual Freedom published a short guide to distinguishing between abuse
and BDSM which includes questions such as "Can you choose to interact freely with others
outside of your relationship?" and "Are your needs and limits respected?" Simply claiming that a
sexual or relationship practice is part of your sexuality or your kink does not make it
nonabusive. At the same time, be careful not to kinkshame while calling out abuse involving
kink practices; see this article on how to hold people responsible for abuse when they and/or
their victims are involved in kink practices.

Some sexual predators that present their search for victims as kindness to others. One pattern
is a white person preferentially mentors people of color and frequently ends up in sexual
relationships with them that end badly for the person of color. Another pattern is offering favors
to young people, such as rides to distant events, that isolate them from parents or guardians. In
particular, be wary of people whose "kindness" or "niceness" is only aimed towards people who
are sexually attractive to them and/or marginalized in some way, especially if they object to
reasonable precautions like chaperones or meeting only during working hours.

Sometimes a sexual predator will conceal their search for victims behind plausible explanations
for unusual but not overtly abusive behavior. For example, a man might describe himself as a
"sapiosexual" (someone claiming to be attracted to intelligence regardless of physical
appearance or gender) but nevertheless only proposition significantly younger women despite
the presence of many other intelligent people of many ages and genders in the community. This
is more likely to be someone hiding their preference for younger and more vulnerable sexual
partners under the guise that he is primarily attracted to their intelligence.
Alcohol and drugs
Two common misperceptions around alcohol and codes of conduct are that the code of conduct
doesn't apply as strongly when community members are drinking, and that alcohol lowers
inhibitions and makes people do things they ordinarily would not do, such as sexual assault,
violence, or bigoted behavior. Research shows that alcohol does not cause disinhibition; cultural
beliefs do. In societies where cultural beliefs about alcohol do not include "it lowers inhibitions,"
alcohol does not have a disinhibiting effect on people who drink it. The three physiological
effects of drinking alcohol which are constant across cultures are:

● Difficulty multitasking
● Lack of coordination
● Sleepiness

For more on the research, see " Social and Cultural Aspects of Drinking " by Social Issues
Research Center.

These facts have important implications for enforcing a code of conduct if your community
includes people with harmful cultural beliefs about drinking (and it almost certainly does). First,
you should emphasize that the code of conduct still applies when community members are
drinking. Second, inform people that if they aren't sure they can control themselves while
drinking, they should not drink. Third, structure your use of alcohol within the community to
encourage and promote moderate, responsible consumption or eliminate alcohol entirely. For
practical advice on how to do this, see " Alcohol and Inclusivity: Planning Tech Events with
NonAlcoholic Options " by Kara Sowles.

The effects of other drugs on a person's behavior will vary more widely. Whatever the effect of
the drug, it is the responsibility of the person taking the drug to not take it if it makes them more
likely to violate the code of conduct. The community must also structure events to make it easy
for people to decline any drug they don't want to take, including alcohol.

A perpetrator may claim to have been drugged against their will at the time of an incident, and
blame their behavior on the drug. However, people who are nonconsensually drugged are far
more likely to be the victim than the perpetrator. Claims that behavior violating the code of
conduct was the result of nonconsensual drugging of the perpetrator must be evaluated
skeptically and include peerreviewed research on the effects of the drug, past behavior of the
alleged perpetrator, and the power dynamics of the situation. For example, there's no drug that
would cause someone to use a racist slur who did not already have a habit of using racist slurs.
Choosing a proportional response
A code of conduct is most effective when the code of conduct committee chooses an
appropriate, proportionate response. If a committee responds to every single violation of a code
of conduct by banning the perpetrator from the community for life, people would only report
violations if they seemed bad enough to ban the perpetrator. Similarly, if a committee responds
to every violation by giving a verbal warning and nothing else, people who aren't bothered much
by verbal warnings would continue to violate the code of conduct.

Most often, a committee should look for the minimum response that will protect their community.
Sometimes this is a lifetime ban, for people who are likely to continue to harm others as long as
they are part of the community. Sometimes it is no action at all, for false reports or
misunderstandings. But in many situations, the best action is somewhere between a permanent
ban and no action, such as explaining what the violator did wrong and getting sincere
agreement from them to not repeat the violation.

Another important consideration is that the code of conduct committee is in a position of power
and authority, and any action taken by it will be magnified beyond what it would be if an ordinary
person took the same action. The difference between a social media post criticizing a
community member from the code of conduct committee and one from an ordinary person is
huge. Be sure to take this into account with your response. To learn more about this effect, you

can read " Is Shame Necessary? " by Jennifer Jacquet. (Short version: Shame is indeed
necessary for social progress, but it should be deployed thoughtfully and proportionately.)

One useful technique is not to explicitly name the perpetrator in communications from the
committee, since many people feel that naming someone for engaging in oppressive behavior is
worse than the oppressive behavior itself and therefore not proportionate. For example, in the
United States, people often object more forcefully to calling a specific behavior "racist" than to
the racist behavior itself. We disagree strongly with with this belief, but until we have normalized
naming oppressive behavior directly, it is often more effective to not name the perpetrator in
your public communications. In some cases, you will want to name the perpetrator (such as
when news of a serial predator's behavior is already widespread and people need to know
specifically whether that person will be present at future events to stay safe), but most often a
concise description of what happened without any names will be sufficient to keep your
community safe.

An unfortunate paradox results from the proportionality principle and often arises during
presentations to large groups of people. When someone is talking in a group of three or four
people and says something harmful, another person can correct or confront them about it and
their response will seem proportionate, since only one or two others witness the correction.
However, when someone is giving a talk to a room of 20 people and says something harmful,
speaking up and correcting them will embarrass them in front of 18 other people, a much more
impactful response than doing so in front of one or two others. When someone is giving a talk to
a room with 1000 people in it, interrupting them is such a hugely embarrassing act that it almost
never seems proportional, even when the speakers says something outrageously harmful.
Given this, a community must choose speakers to large groups with extra care.
If a harasser refuses to follow the code of conduct
Sometimes a harasser will explicitly refuse to agree to follow the code of conduct. In this
situation, if the harasser has up till now violated the code of conduct in small ways (or not at all),
a committee may make a mistake and allow them to stay in the community out of uncertainty or
indecision. Some harassers will carefully limit their actions to stay just below the threshold for
banning, or just on the edge of violating the code of conduct.

It's acceptable (though not a great sign) for people to disagree with a community's code of
conduct—as long as they agree to follow it. Community members can think whatever they want
about a code of conduct, as long as they follow it and do not try to weaken or destroy the code
of conduct. Some common methods for weakening a code of conduct include:

● Advocating against having a code of conduct at all
● Overemphasizing the frequency and harm of false reports
● Proposing unlikely hypothetical situations in which a code of conduct would be harmful

● Repeatedly raising legal concerns about having an enforcing a code of conduct,
especially if the person is a lawyer but not your lawyer
● Making false code of conduct reports
● Making frequent jokes about the code of conduct

Refusing to agree to follow a code of conduct is an excellent sign that someone is already
harmful to a community, sometimes in subtle ways, and will continue to harm the community
going forward. Most communities are better off banning this kind of person.

Some codes of conduct ban advocating against the code of conduct itself to make this decision
easier. Note that advocating against codes of conduct is itself harmful; allowing it to go on sends
a clear message to marginalized groups that it's acceptable to disbelieve their testimony and
that their safety is not important to community leadership.
Responding to protest from the alleged harasser
When learning of the committee's response, sometimes an experienced harasser will try to
avoid consequences by protesting the committee's decisions. Such harassers will try to
convince the committee members with a variety of complaints or excuses, such as:

● Insisting the committee's decision isn't valid until the harasser agrees with it
● Asking for a second chance
● Accusing the committee of unfairness
● Claiming to be the true victim (see the section on DARVO )
● Claiming their actions are protected by some country's free speech rights
● Framing themselves as the more marginalized person
● Complaining that their rights or their privacy are being violated
● Claiming they were just too incompetent to realize they were doing wrong
● Arguing that the committee's actions will result in unfair consequences from other parties
(such as loss of professional connections, friendships, or business opportunities0

This is a good time for the committee member relaying the decision to ask if there's any new
information they should bring to the committee, and if not, to end the conversation. It's possible
that the committee will revisit their decision based on new information or recognition of a
genuine mistake, but it's important to enforce the committee's decision starting immediately and
not to allow the harasser to delay it with specious arguments and emotional manipulation.

When harassers try to invoke the empathy and compassion of the committee members, it is
important for the committee members to consciously reflect on their empathy and compassion
for the targets (past and future), and to remember that the targets are usually far more
deserving of care and concern and yet often receive far less. Focus on the safety of the
community, not the harassers. See the " Distinguishing good intent from bad intent " section for
more information on judging the intent of the alleged harasser.

Communicating the response to others
It is almost always beneficial to inform the entire community of the committee's response, to
show that the code of conduct reporting and enforcement system is active and functioning.
Being transparent about enforcement also creates accountability and allows the community to
have appropriate input on the overall approach to enforcement. Occasionally, protecting a target
from retaliation will take precedence over informing others , but in most cases the response
should be public in some form.

We suggest leaving names out of an announcement, since it helps avoid retaliation against the
target. It also helps community members feel that the response is proportional and therefore fair,
since some people view calling out and naming someone for oppressive behavior as worse than
being the target of oppressive behavior. Leaving out the harasser's name (even if wellknown)
helps prevent this reaction. The people who give you legal advice will probably like the absence
of names too. The major reason to leave a harasser's name in your announcement is when they
are a serial predator and people need to know whether they will be safe from them in your
community.


The audience for the announcement should be people who already believe that a code of
conduct is a good idea. The announcement is not a good place to try to change the minds of
people opposed to a code of conduct.

Be careful not to describe the reaction of the target in the announcement. Describing the target
as satisfied with the apology is often a way for the committee to pass responsibility for the
decision on to the target. By focusing on the target's approval, the target is now implicitly
responsible for the committee's decision, and anyone who disagrees with the decision can be
accused of not listening to the target. The committee must never imply that the target's opinions
influenced their decisions, implicitly or explicitly. For more on this topic, see " Take Responsibility
For Handling Abuse " by Annalee Flower Horne.

A community mailing list open to all for discussion is not a good place to post the announcement
because if there is any disagreement with how the committee handled the code of conduct
violation, it will happen on the list, a difficult venue for community members to ignore. Often
people will respond to an announcement in this kind of forum by stating, at length, their belief
that harmful behavior should be allowed and why. This usually results in revictimizing everyone
in the target group, and can do more damage than the original incident. Allowing public or
community comments on an announcement blog post also often results in unproductive conflict.

The committee should announce their decision in a way that directs people to either send their
criticism directly to the committee, or makes critics host their own critical comments (on their
own blog, on a social media post, on their own mailing list). If public discussion seems
necessary, it's best to create a new smaller forum for people in the community who want to
discuss the issue, or redirect the discussion to an existing smaller forum that is more
appropriate for the topic.

Making the announcement only on social media will often make it hard to find and reference,
and part of the purpose of the announcement is create an easily referenced record of the
community's standards for behavior. Social media is often viewed as impermanent, which
encourages people to view the decision as temporary and open to change. At the same time, it
often makes sense to link to the formal announcement on social media, to reach more people
with the information and avoid any appearance of trying to downplay or conceal the situation.
Social media generally doesn't have a way to disable comments or limit them to the community,
but because that is the case, people are less likely feel threatened by those comments.

If your publication platform of choice allows comments, become familiar with any tools available
to disable or moderate comments, especially on platforms where potentially harmful comments
will appear to be hosted in community spaces. For example, one effective way to moderate
comments on WordPress blogs is to state in the blog post that you will edit any comments
promoting oppression to say, e.g., "I love puppies!" instead of the original text. It’s also worth
choosing a platform that has robust moderation tools, such as WordPress, over a platform such
as GitHub or Medium, where your ability to delete or close comments is more limited.
Responding to criticism
For many committees, communicating their decision is the most difficult and worrying part of
enforcing a code of conduct. Committees are afraid of backlash from their community and,
occasionally, of becoming the focus of criticism from a wider group of people. This fear creates
pressure to ignore reports, not make a decision, keep actions secret, or tell as few people as
possible. Unfortunately, these actions also open up a community to criticism and make the
community less safe.

It's impossible to avoid criticism or controversy when governing a community. The best a
committee can do is put significant effort into getting their facts right, keeping their community
safe, thinking about proportional responses, and being open and honest about mistakes. If the
committee's guiding motivation for all their decisions is, "How do we keep our community safe?"
then even their mistakes will be viewed more positively. Remembering that the criticism would
be as bad or worse if the committee hadn't taken action can be helpful in situations like these.

We recommend following Charles' Rules of Argument while responding to criticism. They look
something like this for this situation:

● Do not try to persuade people who strongly disagree with you
● Wait to see how people react to the initial announcement
● If necessary, post one followup to clarify any genuine misunderstandings
● Refuse to provide more details about the incident or its handling
● Refuse to engage in oneonone arguments, online or in person
● Redirect any communitywide discussions into smaller venues

One of the important points about communicating in this situation is that a community gets one
chance to correct any problems with its first announcement or decision. Each successive
announcement swiftly decreases in impact and effectiveness, so after two statements it is
difficult to change any opinions or impressions. This is why it is important to wait after the first
announcement to get a full spectrum of the responses, so any followup statement can address
all the major issues in one post.

A followup statement should ideally only clarify any genuine misunderstandings created by the
first announcement. This kind of misunderstanding is usually created by unclear or vague
statements in the first statement that genuinely have more than one reasonable interpretation,
such as "We spoke to the harasser and the target, and decided to ban them"—which could
mean the harasser was banned, the target was banned, or both were banned. A genuine
misunderstanding would result in the question, "Did you ban the target?" and the response
would say, "We want to clarify that we banned only the harasser."

The followup should usually not address any bad faith misunderstandings or outrageous
accusations (except possibly to dismiss them as unworthy of response). The followup should
generally not include new information about the situation unless it is necessary to clarify a
misunderstanding. It should especially avoid any new information or additional details that are
intended to persuade people who disagree that the incident was serious. Any new information
will be subject to another round of misunderstandings and criticisms, followed by ever less
convincing responses from the committee. People will often say that if they just knew one
particular detail about the allegation, or the way the committee discussed it, or how the target
responded, then they would support the committee's response. Don't fall for this; it's almost
always a lie unless it's a response to a genuine misunderstanding of or mistake in your first
statement.

When responding to criticism, it’s important but difficult to avoid defensiveness, especially when
the critic is intentionally lying. In most cases, simply ignoring outrageous untrue claims will make
them go away the fastest, as critics sometimes make random claims to see what gets a
reaction. Often your reaction—repeating the claim in the process of refuting it—has the
counterintuitive effect of making people more likely to believe to the false accusations .
Occasionally it will be necessary to publicly deny a false accusation when it seems plausible or
becomes general belief, but in most cases it's best to ignore falsehoods. In general, the fewer
public statements your community makes about the incident, the better off your community is.

Case study: I was publicly accused by a wellknown white supremacist of spitting in another
person's face (literally, not figuratively). Because this lie was about something so out of
character for me and not an exaggeration or spin of a real incident, I ignored the accusation
completely and it went away within a few days. If I had instead written a long blog post
explaining all of the messy personal history between me and the people involved and why I
thought they were lying, the accusation would have reached more people and stayed alive
longer. The end result of a public denial would have been more people who believed I had spit
in a person's face.

Sometimes people will approach committee members or community leadership to have a
facetoface discussion or argument about the committee's response. In most cases, you should
tell that person to email their concerns to the entire code of conduct committee. It's easy for an
individual member to be swayed by a passionate inperson argument from a person they
respect and make a decision or a promise that will harm the community. It's much harder for an
email to influence the whole committee to make a decision that will harm the community.

Oneonone verbal conversations are also easy to misunderstand or intentionally misrepresent
to match the harasser's goals; normal softening language or deescalation tactics on the part of
the community member can easily be spun into "[Committee member] agreed with me that this
ruling is too harsh." Keeping discussion in written form creates a clear written record of what
each person actually said. It's also simply a waste of the committee's time to engage in
oneonone arguments. One of the principles of Charles' Rules of Argument is to play to the
audience, and there's no audience in a oneonone discussion.

Remember that not everyone in the community has to agree with the code of conduct or the
committee's decisions for them to be valid and enforced. This is tough for people who are used
to making decisions by consensus, or who are uncomfortable with open conflict. Unfortunately,
harassers often take advantage of others’ discomfort with conflict to get and keep access to
targets. In particular, decisionmaking by consensus relies on all (or nearly all) participants
acting in good faith, when by definition a harasser is not acting in good faith. The code of
conduct committee's job is to stand between the community and people who want to harm it, not
to cajole it towards consensus.

As mentioned in the section " Communicating the response to others ," if community members
start criticizing the report and/or the committee's response in a communitywide forum, the
discussion is likely to revictimize targets, create new code of conduct violations, and decrease
the safety of the community. It's best to redirect that discussion to a smaller venue, even if it
means banning threads or otherwise restricting communication in the communitywide forum.
Don't try to stop people from discussing code of conduct incidents and responses, just don't
allow them to do it in community spaces that include people who haven’t elected to be part of
the discussion. If necessary, create a new mailing list or chat channel for this discussion, or tell

them to create their own. This allows people to criticize and question the committee's decisions
without revictimizing other members of the community.

While the committee may end up changing your decision, do not allow the community to think
that means the committee's decision is up for a vote or debate. You don’t want to establish a
sense that majority opinion or sustained complaints will change the committee's decisions on
code of conduct reports. Sometimes public pressure will cause the committee to reexamine
their decision and change it based on new facts or improvements in their understanding of the
relevant principles, but this is different than allowing public pressure to dictate the decision. If
you do change your decision, reference the facts that led to the change in your announcement,
rather than saying you listened to the community and responded to their feelings.
Dealing with attacks on the committee or community
Sometimes, an existing group of bad actors will seize on an individual incident and use it or your
response to attack the community leadership, the code of conduct committee, the reporters, or
the targets (see " Why Asking What Adria Richards Could Have Done Differently Is The Wrong
Question " by Deanna Zandt for one example of this). If this happens to your community, don't
blame any of the targets of harassment. Whatever mistakes the target may have made (if any)
in responding to the harassment, they are never responsible for the harassers' actions.

The committee should not publicly criticize the target for any missteps on their part unless they
rise to the level of a code of conduct violation themselves. If the target’s actions do rise to that
level, any public comment should prominently include the context of the target's mistake, which
is that they were being harassed. A committee should especially avoid creating a false
equivalence between the initial harassment and any missteps in the target's response. For
example, don't say, "Well, that person made a sexist joke, but the person reporting them yelled
a profanity at them, so they both did wrong." It's rare for the response to harassment to be
anywhere close to as harmful as the original incident, and it is harmful to do anything that
encourages people to view them as equivalent.

A committee must never try to deflect blame for any negative response to their decision by
directly or indirectly blaming the target for the committee's decision. Most commonly, this takes
the form of a committee citing the target's request for leniency or expression of forgiveness for
the harasser as the reason they chose an ineffective or overly mild sanction. A committee may
also blame the target if they choose a sanction critics view as too harsh by suggesting they
feared the target would criticize them publicly if they chose a lesser response. For more on this
topic, see " Take Responsibility For Handling Abuse " by Annalee Flower Horne.

If the committee and/or the target become the target of a harassment campaign, there are a few
things they can do to reduce the impact of the attack. Following Charles' Rules of Arguments
and this entire document will make the attack shorter (and make one less likely to occur in the
first place). The instructions for dealing with stalkers—basically, block them and never interact

with them—in Gavin de Becker's book " The Gift of Fear " are also extremely useful (but see this
Captain Awkward advice column post for some caveats on the intimate partner violence section
of that book). Leigh Honeywell wrote a guide to staying safe online ("digital defense") for
whistleblowers and started a company that protects employees from online threats .
Summary

● Including extraneous material in a code of conduct weakens the code of conduct.
● Your community should not attempt to implement transformative justice unless it meets
specific criteria, including the type of community and training of community members.
● It is okay for the committee to lie if that is the only way to protect both individual safety
and the safety of the community.
● To attract marginalized people to your community, be open and public about responding
to code of conduct violations instead of trying to hide them.
● The safety of your community should take priority over privacy and confidentiality when
they come into conflict.
● Don't ask for apologies, make the target decide the response, mediate between parties,
guard the harasser or the victim, or tell the harasser to avoid the target.
● Powerful people must also be subject to the code of conduct or else your community is
creating a double standard that only applies to the less powerful.
● The intent of a harasser is less important than impact.
● To distinguish a true victim from a harasser using DARVO (Deny, Attack, Reverse Victim
and Offender), look at how they behave in other situations and how they defend
themselves.
● Judge competing claims of marginalization by the alleged harasser and alleged victim by
looking at the research.
● The best way to protect socially awkward people is to enforce the code of conduct on
everyone, including socially awkward people.
● People with mental health issues should get treatment from mental health professionals
and do not need to be included in a community when they are harming others.
● Caregivers and the people they care for must also follow the code of conduct.
● Enforcing rules around sexual behavior need to take into account the consent of all
parties, power dynamics, and patterns of behavior.
● Alcohol and drugs do not excuse harassment and special care must be taken when they
are used in your community.
● People who refuse to follow the code of conduct should be banned.
● By default, the committee should tell the entire community about its response to a report.
● Community leadership should respond sparingly and carefully to criticism, and usually
only change their decision when they get important new information.
● If the target, reporter, or community are attacked for enforcing the code of conduct,
community leadership should protect and defend them.


● Meeting role cards by Frame Shift Consulting
● " How 'Good Intent' Undermines Diversity and Inclusion " by Annalee Flower Horne
● " The Al Capone Theory of Sexual Harassment " by Leigh Honeywell and Valerie Aurora
● " Study Reveals The 6 Key Components Of An Effective Apology " by Amy Morin
● " Take Responsibility For Handling Abuse " by Annalee Flower Horne
● " What Reporting Sexual Harassment Taught Me " by Simine Vazire
● " No More Rock Stars " by Leigh Honeywell, Valerie Aurora, and Mary Gardiner
● " How to Organize Tech Workers to Change Company Policy " by Liz FongJones
● " How 'Good Intent' Undermines Diversity and Inclusion " by Annalee Flower Horne
● " What is DARVO ?" by Dr. Jennifer J. Freyd
● " What kind of person makes false rape accusations? " by Sandra Newman
● Paradox of Tolerance on Wikipedia
● Oppression Olympics on Wikipedia
● List of articles debunking autism as the cause of harassment on Geek Feminism Wiki
● " The Myth of the Male Bumbler " by Lili Loofbourow
● " Acceptance, kink shaming, and calling out bad behavior " by Anabelle Bernard F
● " Mythcommunication: It’s Not That They Don’t Understand, They Just Don’t Like The
Answer " by Thomas MacAulay Millar
● " BDSM vs. Abuse Policy Statement " by National Center for Sexual Freedom
● " Google Employees: Our Executives Engaged in Abuse. Don’t Let Kink and Polyamory
Be Their Scapegoats. " by Liz FongJones
● " Social and Cultural Aspects of Drinking " by Social Issues Research Center
● " Alcohol and Inclusivity: Planning Tech Events with NonAlcoholic Options " by Kara
Sowles
● " Is Shame Necessary? " by Jennifer Jacquet
● Charles' Rules of Argument on Geek Feminism Wiki
● " The power of framing: It’s not what you say, it’s how you say it " by Steve Rathje
● " Why Asking What Adria Richards Could Have Done Differently Is The Wrong Question "
by Deanna Zandt
● " The Gift of Fear " by Gavin de Becker
● " Staying Safe when you Say #MeToo " by Leigh Honeywell

Chapter 5: Examples of responding to reports
This chapter describes realworld code of conduct reports, how they were handled, and how the
community responded to them if known. The purpose of this chapter is to show that following
the recommendations in this guide results in better outcomes in the real world. The examples
are arranged roughly in order from most successful handling to least successful, and from
adhering most closely to the recommendations of this guide to least closely. These examples
may also be used by code of conduct committees to practice responding to reports.

Several conferences have published "transparency reports," anonymized summaries of all the
code of conduct reports related to the conference, along with the actions taken by the
conference code of conduct committee. We included some of their summaries in their original
words by their generous permission. We anonymized and rewrote other conference
transparency reports. We are grateful to every organization that published how they responded
to a report.

The commentary on the public transparency reports is not aimed at the original organizers, but
at the reader. The organizers who wrote the transparency reports had much more information
than what they included in the report, and they may have had good reason to take different
action than what we recommend in the commentary. They may even have done what exactly
what we recommend and simply left it out of the transparency report. Ultimately, the sole goal of
our commentary is to help the reader understand what to do, not criticize the people involved in
these incidents.

For some of the reports, we have changed names and details to preserve confidentiality. Some
of the authors were involved in handling some of the reports, but we are not identifying which
reports to help preserve confidentiality. Additional examples of code of conduct reports and their
handling can be found at the Timeline of Incidents on Geek Feminism Wiki.
Wikimania 2012 sexualized presentation
An excellent but long description of this incident is available here ; below is a shorter summary in
our words followed by our commentary.

At Wikimania 2012, a presenter included two pornographic images in a presentation as
an illustration of the kind of image that was controversial to include in Wikimedia
Commons. Two women in the audience walked out in protest and immediately contacted
the conference code of conduct team. During the Q&A period for the talk, an audience
member asked the presenter why he used those images. Before the next presentation
started, the program chair made an announcement to the room that the code of conduct

had been violated. After the following presentation was over, a listening session for
people who had witnessed the violation was announced.

At the listening session, the presenter apologized in detail and explained his rationale for
including the images and why he now believed it was the wrong thing to do after he had
talked to several people in the audience. He also requested that the video of his talk not
be published. The code of conduct committee agreed that the video should not be
published and that no further action was necessary against the presenter. The
conference leadership published a short blog post summarizing the incident and the
response , and made an announcement with similar content at the next plenary session.
There was little or no further discussion about the incident or its handling.

This is a textbook example of handling a report in the way this guide recommends, and the end
result was a safe and satisfied community. The incident was quickly reported and the organizers
responded publicly within minutes. The presenter quickly realized why what he had done was
harmful and gave a sincere, indepth apology as well as a specific proposal for reducing future
harm from his actions. The people most affected by the incident were able to attend the listening
session and get personal apologies if they wanted them. (As a note of caution, the listening
session was moderated by an expert facilitator; without this expertise, such a session can make
things worse.) The code of conduct committee quickly drafted a short, clear summary of the
incident, published it on the conference blog, and made an announcement at the next plenary
session. As a result, the response was published before the incident could be reframed of as an
issue of censorship or sexnegativity, and before any distorted or wrong versions of the incident
could gain traction. Public discussion was negligible and mostly positive.

Many committees are afraid of provoking dissent and criticism with their responses, but this is
an example of how a swift, firm, definitive announcement can prevent harmful conflict.
DjangoCon EU 2017 transparency report
The DjangoCon EU 2017 transparency report is a remarkably helpful document, as it succinctly
describes how to handle several common kinds of incidents successfully, including one in which
the reporter did not want to identify the harasser and one in which the harasser could not be
identified. The block quotes in this section are verbatim quotes from this transparency report,
used by permission.
Denial of validity of code of conduct
During a talk, a number of comments has been loudly made from people in the audience
regarding statements from one of the speakers. An argument on twitter and on the slack
team ensued, where the main point was that DjangoCon is only meant for technical
talks. The author of this comment further replied privately to one of the Code of Conduct

committee with a denial of the validity of the Code of Conduct and the value of talks on
social, diversity and inclusivity topics.

We, as member of the Django community, well before than as member of the DjangoCon
Europe 2017 Code of Conduct committee, want to clarify the fact that DjangoCon
Europe is not a low cost advanced Django training course, but it's an important event for
the whole Django community to discuss every topic that impacts our community: being
technical advancements or community life discussions. We issued a written reprimand to
the offender and transmitted the information to the DSF Code of Conduct committee.

This is a good clear statement of the purpose of conference. Sharing the information about this
person with the parent organization is smart. A written reprimand makes sense too. In general
we recommend that a community ban people denying the validity of the code of conduct from
attending the conference in the future, as they've made it clear that they do not feel constrained
by the code of conduct and are likely to violate it in the future. It is possible that once a critic
understands that continuing to object to the code of conduct in this way will result in a ban, they
will agree to abide by it and stop protesting against it. It is reasonable to give them a chance to
change their mind by issuing a written reprimand. We also recommend asking the person to
commit to abide by the code of conduct going forward.
Harmful question during a talk
At the end of one talk, an inappropriately worded question from one of the attendees was
widely understood as offensive for the speaker. The CoC committee discussed this with
the attendee who made the statement, and a private written apology has been relayed
by the CoC to the speaker, upon speaker consent. The part of the talk has been
removed from the published videos.

We want to stress that a lot of harm can be made, even in good faith, when speaking to
the other people, and even small misunderstanding can lead to people feeling not
welcomed within the community, especially on sensitive topics, like the one in this
incident.

The committee's clear focus on the harm to the other attendees, instead of on the intent of the
questioner, is excellent. We especially admire that the committee made it clear that the target
consented to receive the private written apology from the offender. Deleting the relevant portion
of the video reduces future harm to the community. Overall, a great example of how to handle
this kind of report.
Sexist comment on clothing
A speaker reported that after their talk, one attendee reported comments on their attire
during the talk and that a different dress would have been resulted in a better impression
over the audience. The speaker involved has not provided more details about the

incident to avoid making it a personal case, but it's of the uttermost importance to note
that comments regarding the appearance are sexist, toxic and dangerous, and they have
no place within the Django community: communities—or attendees—have no role in
imposing or suggesting a dress code to other people.

This is a good example of how to handle a report from someone who does not want to name the
harasser. This clear, detailed defense of the target and explanation of why the comment was
harmful sends a clear message that this behavior won't be tolerated in the future. We'd
recommend slightly different phrasing instead of "avoid making it a personal case" because it
implies that people who do name names are engaging in undesirable behavior—perhaps
"because the reporter did not wish to share it" or "because the reporter thinks the problem is
more widespread than this one person" are good alternatives.
Photographer creates awkward situation
During the conference party, an attendee taking pictures has put two attendees in an
awkward situation to take one shot; this resulted in a very uncomfortable situation for at
least one of the person involved. Given the social context, the person felt obliged to
comply with the request, but it resulted in a high stress for them. We have been unable
to identify the photographer and thus no further action has been taken.

We want, nonetheless, stress the fact that the respect required by the Code of Conduct
goes beyond the simple "do not harm the others", but it means that a proper behavior
must be ensured during all the conference events, and especially during the social
events, and that proper and unforced consent must be asked to other people without
pushing them in unwanted situations.

This is an excellent explanation of what true consent involves and that a person must not feel
coerced in any way. It's also another great example of what a committee can do to protect the
safety of the community when they don't know who the harasser was.
Write the Docs EU 2016 transparency report
The Write the Docs EU 2016 transparency report is a short but useful transparency report.
While complicated messy incidents get the most attention, many incidents require short, simple
responses and this report reflects that. The block quotes in this section are verbatim quotes
from this transparency report, used by permission.
Attendee uses derogatory term
During the Writing Day pitch round, an attendee used a generalization term when
referring to another attendee, and this generalization term is considered derogatory to

some groups. At the request of the reporters, the CoC discussed this with the attendee
who made the statement and a private apology was relayed back to the reporters.

This response is good, with the note that apologies should only be relayed to the reporters with
their consent (which probably happened here but wasn't specified in the report). The publication
of this short description made it clear to anyone who heard about it that this behavior wasn't
appropriate.
Inappropriate joke in talk
One of the talks contained a slide that included an inappropriate joke towards a certain
user group. The speaker was asked to remove the slide from the published deck.

This is a good response, similar to editing or removing a video. (We assume there was no video
recording.)
PyGotham 2017 transparency report
The PyGotham 2017 transparency report is uniquely helpful because the committee included
several incidents that were hard to verify or did not actually violate the code of conduct. The
committee demonstrated its commitment to the code of conduct by taking these reports
seriously and giving them a full investigation. The block quotes in this section are verbatim
quotes from this transparency report, used by permission.
Selfreport of an ambiguous joke
An attendee made a comment that could have been misconstrued as an offcolor joke.
Conference staff determined that the comment was innocent, not intended to be a joke,
and not in violation of our code of conduct. The attendee selfreported this after realizing
how it could have been interpreted. We applaud their attention to the issue and
encourage others to do the same if they find themselves in a future similar situation.

Selfreporting is a good sign for the health of the community and awareness of the code of
conduct. The response described seems appropriate. If the reporter felt like it, they could also
make an apology on social media or other public forum.
Attendee denies making offcolor joke
An attendee made an offcolor joke to a large group. When conference staff spoke to the
accused person, they denied making or hearing the joke. Whether or not this person was
correctly identified, staff made it clear that this type of joke was a violation of our code of
conduct and repeat offenses would result in them being asked to leave the conference
and not welcome back the following year.

Warning the suspected harasser with a promise to ban them next year if they repeat the
behavior is a great response. It seems unlikely that in a large group, everyone present
misunderstood the joke or misidentified the person making it, and the committee is (correctly)
acting as though the incident happened, regardless of the harasser's denials. We would
recommend that the code of conduct committee investigate to find out if this behavior is
common for the harasser and if so consider more serious consequences than a warning, since
some serial harassers maintain access to communities and victims using outright denial and
relying on being believed because of they have greater social capital then their victims.
Volunteer overwhelmed by requests
A volunteer became overwhelmed by requests made of them by other attendees and left
the conference. We’ve reached out to the volunteer to get more details about the
interaction but have not heard back yet. We’re working on a plan to make sure that
volunteers’ responsibilities are more clearly defined, that volunteers are better trained
before the conference, and that special requests are handled by conference staff
directly.

This is most likely the result of a mismatch between what the volunteer was comfortable doing
and a specific volunteer role. The committee’s response, making a plan to change the volunteer
structure, is excellent and should prevent repetition of this problem.
Attendee makes unwelcome advance
An attendee made an unwelcome advance toward a volunteer. By the time the report
was made, conference events were over for the day, and the attendee was not seen at
the conference again. This was a clear violation of our code of conduct, and the attendee
would have been asked to leave if they were able to be found.

We're assuming that the unwelcome advance was sexual, and that the harasser couldn't be
identified. In that context, this response makes sense. Letting the community know what they
would have done if they had identified the harasser is quite useful. If it is possible to identify the
harasser, it's good to consider banning them from next year's conference even if you can't find
them before this year's event is over. If you're not sure about a ban, asking around to find if
there have been other incidents is helpful for evaluating if the harasser will be a threat to the
future safety of the community.
Badfaith code of conduct report
This is a published incident anonymized and summarized in our own words.

At a conference, a speaker used a quote from a public blog post by an influential
conference attendee to illustrate a pattern of negative behavior in the wider community.

The attendee reported the speaker's use of his quote as a violation of the code of
conduct. The attendee had a long public history of advocating against codes of conduct
in this community. The conference organizers interviewed the speaker and concluded
that the speaker did not violate the code of conduct. The attendee later described their
report as a "test" of the code of conduct. Much later, the attendee was ejected from
another conference for bullying conference staff.

While rare, spurious or false reports do happen—but rarely in the way many opponents of codes
of conduct envision. The usual fear is that a relatively powerless person will falsely accuse a
powerful person of violating the code of conduct, who will be believed without further
investigation. However, all of the confirmed spurious reports we currently know of involve a
relatively powerful person reporting something that did not happen or did not actually violate the
code of conduct. In all cases of false reporting we are aware of, the report was either made in
bad faith to harass an advocate for codes of conduct, or was made in retaliation for a community
member attempting to hold a powerful person accountable for their actions (e.g., calling them
out for being sexist).

Often, someone making a spurious report will go on to harass others in the future. In this case,
the spurious reporter went on to harass conference staff at another conference and was banned
from that conference. Looking at the broader pattern of behavior of both people involved in the
spurious report supports the organizers' decision to do nothing about the spurious report: the
alleged victim had a long history of bad faith complaints and outright harassment, while the
alleged harasser had a long history of speaking up to defend and support marginalized people
at a significant cost to themself.

If someone does make a spurious code of conduct report with the intent of harassing the person
they reported or the community leadership, that should be treated as a violation of the code of
conduct. A goodfaith report that later turns out to have been mistaken should not be treated this
way, of course.
Racist comments at a conference
This is an unpublished incident anonymized and summarized in our own words.

An attendee at a conference reception loudly and repeatedly made statements denying
the existence of racism against people of color to other conference attendees. He also
expressed his belief that the right to free speech in that country protected his ability to
make those comments. The committee was quickly informed of the comments and
asked him to leave the conference until they had made a decision about how to respond.
After realizing that his behavior could potentially result in him not being able to socialize
with others at the conference, he apologized and promised to abide by the code of
conduct. He was allowed to return to the conference after a long discussion with a

member of conference leadership about the seriousness of the code of conduct violation.
He followed the code of conduct for the remainder of the event.

Some people believe that the code of conduct is insincere boilerplate that leadership will not
enforce, or that they can circumvent it by invoking "free speech" or including a warning slide at
the beginning of their talk. A surprising number of people who openly violate the code of conduct
will change their behavior once they realize that they face actual, realworld consequences,
such as not getting to hang out with their friends at a conference. Plenty of people will not agree
with the code of conduct but will abide by it if that is the only way they can get what they want
(usually, continued access to your community spaces). This kind of grudging acquiescence is
still a warning sign and committee members should be alert for other code of conduct violations
from the same person. If the harasser violates the code of conduct again they should be banned
permanently.
Oppressive comments in online chat
This is an unpublished incident anonymized and summarized in our own words.

A woman of color used an insult in an online chat group that was sexist, homophobic,
and sexnegative. Several other community members noticed the comment and brought
it to the attention of the code of conduct committee. When privately contacted by a
committee member to ask her to edit or delete her comment, she stated that because
she was a woman of color she could make homophobic comments if she wanted to. She
also argued that since the committee member talking to her was white, it was racist to
ask her to change her comments. Finally, she stated that white people were making
similar comments without being reported. Her husband contacted the committee
member to make the same arguments. The committee reviewed logs to see if other
people, especially white people or men, were making similar comments and concluded
that they were not. When the committee made it clear the woman needed to abide by the
code of conduct regardless of her gender or race, she left voluntarily.

When someone claims that the code of conduct doesn't apply to them, that's a convincing sign
that that person needs to be banned from the community. However, any other issues or
problems they bring up might be genuine, regardless of their bearing on an existing incident. It's
quite common for white people of all genders and men of all races to get away with behavior
that people of color of all genders and women of all races are punished for, so it was important
that the committee look at the data while consciously attempting to compensate for their biases.

If the committee had found that white and/or male community members were making similar
comments on community forums without action, the solution would have been for the committee
to thank this person for reporting this, apologize for being racist in a publicly accountable way,
respond to the other incidents involving white/male community members, and arrange some
antiracism and/or antisexism training for the committee members and leadership as a first step

to greater change. (More antioppression training is almost always a good idea.) Regardless of
the outcome of the review for racial and gender bias in enforcement, the person who made the
original comment would still be a threat to the safety of the community because of their belief
that the code of conduct does not apply to them personally. They would still need to be banned
if they continued to refuse to follow the code of conduct.
Anonymized conference transparency report
The following incidents were published, but we have anonymized and reworded them because
we want to use these incident reports as cautionary examples of what not to do, rather than
criticize a specific conference. For context, this conference was held in a resort hotel in a tourist
area, and many attendees spent time at the swimming pool in the hotel. The indented sections
are our anonymized and reworded versions of the original public reports.
Attendee invites women to hotel room under pretext
A man attending the conference invited two different women attendees on separate
occasions to come to his room so he could teach them methods of relaxation. They both
said no and he did not ask them again. When asked about what he'd done, the man
apologized and agreed not to do it again.

The committee treated this as an innocent misunderstanding, when based on the information in
the report, it sounds far more like a possible serial sexual predator. The first sign is repetition: he
tried the same invitation on two separate women. Often serial predators are simply playing a
numbers game, following the same script with different targets until they succeed. (It's not an
accident that this method is similar to what "pickup artists" teach.) While these two particular
women were able to say no, others may accept out of fear of retaliation, a sense of guilt at being
suspicious of another person, or a belief that they are safe because they are both attending a
conference with a code of conduct.

The second sign is the harasser pretending not to realize that inviting a stranger to go back to
his hotel room alone with him to learn to relax is inappropriate—of course it's inappropriate. This
looks like an instance of the "Male Bumbler" defense : the idea that adult men who are socially
adept in many other areas of life are suddenly incompetent when it comes to judging how their
actions are received by potential sexual targets.

The third sign is the gender of the targets: strange that he invited zero men to learn to relax
alone with him in his hotel room! Always question when someone's "kindness" involves physical
touch and/or isolation in a private space and/or is always directed towards women or young
people (and especially young women).

The fourth sign is what is being offered: a way of making the women let down their defenses
and make themselves vulnerable more quickly than they might otherwise. It's another version of

pressuring someone to drink more alcohol or giving them drugs that make them more relaxed or
unconscious.

A more appropriate response would be an immediate ban for life. At best, this man is
deliberately singling out women for inappropriate invitations to his hotel room in a way that
makes him appear to be a serial sexual predator, at worst he is a serial sexual predator.
Inappropriate touch reported after conference ended
A man helped a woman do some shopping far from the venue. While they were
shopping, he touched her inappropriately. The report didn't come in until after the
conference ended. Because the report happened after the conference ended, the
organizers took no further action.

The organizers act as though nothing can be done after the conference ended. Many things can
be done after the conference has ended: people can be banned from future events, other
organizations can be informed, the committee can ask other people if they've had negative
experiences with this person, they can advertise the code of conduct and reporting instructions
more widely, they can express concern and remorse to the target, and they can also make a
public statement strongly disapproving of the action (much as other conference organizers did in
previous examples). Committees sometimes spend months doing things after the conference
has ended. In this case, the organizers described the action and stated that nothing was done.
It's not enough.
Unwanted sexual advance
After conference hours and in the pool area of the hotel, a man attending the conference
made an unwanted sexual advance to a woman who was not attending herself but was
the partner of another attendee. The unwanted sexual advance included the man's
knowledge of the woman's existing monogamous relationship. The organizers asked him
to go back to his room for the rest of the night, out of concern that someone [unclear
who] might assault [unclear physically or sexually or both] someone else [unclear whom].
He went back to his room till the morning. The organizers took no further action.

This report raised many questions, but the most worrisome was the framing of asking the
harasser to return to his room as a way of preventing an assault. If anyone is seriously thinking,
"If this person remains in a public area, there will be an assault," someone, or several
someones, should be banned. We're not sure whether the fear was that the harasser would
assault someone or someone else would assault the harasser, but either way, this was a serious
issue. That the response of the organizers was merely to ask the harasser to go back to his
room is totally inadequate.

In the most optimistic case, where the harasser genuinely believed he was giving a welcome
compliment to the target, at minimum the organizers needed to have a long conversation

emphasizing how inappropriate this behavior was that ends with the harasser voluntarily offering
a clear and detailed apology and a promise to not repeat the behavior. However, the mention of
potential assault suggests that the problem was much more serious than that. Note that if the
concern was that another person might assault the man making the advance, whatever caused
that concern (verbal threats, etc.) should be treated as a second, separate code of conduct
incident.
Inappropriate touch
In the hotel swimming pool, a man touched a woman inappropriately. The organizers
heard about it by accident and spoke to the man. He stayed in his room and only came
out for meals during the rest of the conference.

This is about half right. Removing the person from the event space most of the time is better
than nothing, but there's no stated reason why the harasser needed to leave his room for meals.
If he needed to do this, then so did his target and the rest of the attendees, revictimizing the
target and providing opportunities for the harasser to harass additional people.

Some events do require sharing living space (for example, food can only be acquired from a
public cafeteria, or multiple people share a bathroom or living room). We recommend that this
kind of event specifically state up front that anyone violating the code of conduct may be
removed from that space at their own expense, and make arrangements with the venue that
allow the organizers to remove the harasser from the venue entirely. For example, conferences
on cruise ships should warn attendees that if they violate the code of conduct, they may be
confined to their cabin and dropped off at the next port of call, and they will have to cover their
own cost to return home, and include this provision in the contract with the cruise ship.

In this anonymized case, the harasser still had access to the rest of the community at meals; we
recommend bringing him his meals instead. This summary is also missing a declaration that this
behavior is wrong, and a plan to prevent a repeat of this behavior at the next conference (a
permanent ban, a long conversation with the harasser that resulted in a clear apology and
promise, researching to see if the harasser has a history of similar behavior, etc.).
Inappropriate pulling on clothing
In the hotel swimming pool, a man grabbed and pulled on parts of a woman's swimsuit
inappropriately. The organizers asked questions of the man, who apologized and agreed
not to repeat the behavior. The woman was satisfied with the result.

At this point, it sounds like conference attendees need to be told emphatically that the code of
conduct still applies in the swimming pool area, which is not too surprising as events involving
swimming are a higher risk activity for harassment. A common problem with codes of conduct is
that some community members believe that there are implicit exceptions to a code of conduct:
when people are drinking, at night, on the dance floor, in the swimming pool—the list goes on.

The code of conduct for this event should have been updated to emphasize that it applies to the
swimming pool and the organizers should have announced this update at least once a day after
that.

While the apology and promise not to repeat the behavior from the harasser are positive signs
for the safety of the community going forward, the organizers should have also found out if this
person has a pattern of unwelcome touch or other harmful actions and banned him if so.

The mention of the target's satisfaction is an example of the leadership passing responsibility for
their decision for the target, setting them up to say, "Well, the target was satisfied, why aren't
you?" if anyone complains. This is not what good leadership looks like.
Drupal community incident
[Content warning: sexual assault, violence against women, misogyny, racism, transphobia,
advocacy of slavery]

While long and complicated, this incident graphically demonstrates the reasoning behind many
of the recommendations in this guide, and serves a real world case study for what happens
when they are not followed. It's difficult to correctly summarize this incident and the response, as
it played out over several years and through tens of thousands of words posted online by people
with conflicting stories and clear biases. This is a best effort summary and discussion based on
public documents. It undoubtedly still contains several errors of fact, but is close enough to what
happened to serve as a learning tool. We apologize in advance for any errors in this summary.
We did not anonymize the two main people in this incident as each published thousands of
words about their roles, but we remind readers that our purpose is to educate community
managers, not criticize these individual people.

In early 2017, the Drupal Project Leader Dries Buytaert asked Drupal contributor Larry Garfield
to leave the Drupal community. The action was prompted by concerns that Garfield was sexually
exploiting a woman he brought to some Drupal events. In response, Garfield claimed he was
being discriminated against based on his sexuality. The resulting controversy continued through
several months of sarcastic articles in The Register and other news outlets, five updates to
Buytaert's first blog post about the incident, multiple updates on Garfield's blog, and a
community protest website supporting Garfield called Drupal Confessions , among other things.
After much public criticism the project leader allowed Garfield to stay in the community but
banned him from holding leadership positions. Garfield later left the Drupal community on his
own initiative. The rest of this section delves into the details of this incident.
Background
Drupal contributor and community leader Larry Garfield made dozens of harmful and bigoted
comments in public over a period from at least 2010 to 2016. His comments included rape

apology, minimization of chattel slavery, advocating violence against women, racism,
transphobia, and sexism. Many Drupal community members publicly objected to these
comments. During this time, Garfield continued to participate in the Drupal community, including
maintaining a core subsystem, speaking at Drupal events, and serving in various community
leadership positions.

In October 2016, another Drupal community member found some information about Garfield's
sexuality and beliefs that were available on web sites requiring an account to join and view. The
information was related to Garfield's membership in and promotion of the Gorean subculture , a
community that advocates for men acquiring and training women to behave as "slaves" as a
fulltime lifestyle, based on the belief women are biologically suited to be enslaved and men are
biologically suited to own slaves and be leaders. Distinct from other types of
dominance/submission sexual roleplay communities based around fantasy worlds, many
members of Gorean subculture genuinely believe that women are biologically suited to be
slaves. Garfield has made public statements indicating he is genuine in this belief and in a
leaked chatlog discussed how to best teach 15 year old girls about the Gorean subculture (that
is, how to groom underage girls for later sexual abuse).

This information was reported to the Drupal Community Working Group, which concluded that
Garfield had not violated the Drupal Code of Conduct. Acting as an individual, a Drupal
community member pressured Garfield to leave the Drupal community. Garfield referred this
matter to the Drupal Community Working Group as well, which concluded that no violation of the
Drupal Code of Conduct had occurred. The CWG referred the matter to the Drupal project
leader and recommended that Garfield and the community member talk to each other directly.
Precipitating incident
This incident started in earily 2017 when Drupal leadership learned that Garfield had brought a
woman to Drupal events who was acting as Garfield's "slave" according to the tenets of Gorean
subculture. Drupal leadership became concerned that Garfield might be sexually exploiting her,
based on multiple red flags indicating her relative lack of power compared to Garfield, including
her relative youth and the fact that she did not speak verbally to anyone at the event. Garfield's
blog post arguing that the numerous factors contributing to the power imbalance between them
did not affect her ability to consent sexually is not convincing. No one other than Garfield
appears to have access to her side of the story, so Garfield's description of her or of their
relationship cannot be verified.
Response
As a large Drupal conference was approaching where Garfield was scheduled to speak and
chair speaking tracks, the Drupal Association cancelled his sessions and removed him as track
chair as an interim step while they investigated and deliberated. They discussed their concerns
with Garfield privately for several weeks, in response to Garfield's request to protect his privacy.

During this process, Drupal Project Leader Dries Buytaert asked Garfield to leave the
community.

Three weeks into the discussion, Garfield published a blog post that claimed Drupal leadership
was discriminating against him for his participation in kink sexual practices, using language and
analogies that framed Garfield as the victim of oppression against a sexual minority who had
been unfairly outed. In response, Buytaert published a post on his personal blog explaining why
he had asked Garfield to leave the community. Garfield and Buytaert each published several
update posts, and many tech publications wrote stories about the incident. Much of the public
commentary was supportive of Garfield, including a site supporting Garfield called Drupal
Confessions .

The response ended with Drupal partially reversing its decision and allowing Garfield to attend
Drupal events and participate in the community, while removing him from all leadership
positions. Garfield left the Drupal community in protest and continues to attend and speak at
other PHPrelated conferences .
Analysis
While Drupal leadership was correct to attempt to eject Garfield from the Drupal community, the
reasons and methods they used were counterproductive and harmful. The final decision to allow
him to remain in the community was harmful to the safety of the community. A series of
mistakes, including the apparent condoning of Garfield's bigoted opinions and behavior by the
Community Working Group , encouraged Garfield to escalate his behavior. Drupal leadership did
not act when Garfield advocated for sexism and literal chattel slavery in the abstract, but waited
to act until they faced the prospect of suffering personal discomfort watching Garfield acting out
his beliefs in person at Drupal events with a specific woman they could see and identify. The
most uncharitable reading is that Drupal leadership was inspired to act by paternalism towards
an individual young woman rather than out of a general commitment to equality and justice.

Drupal leadership made many decisions that are counter to the recommendations in this guide
and the results demonstrate why we make these recommendations. We'll examine them one by
one.

Failing to sanction the perpetrator early: The best solution to this scenario is to have
prevented it from happening at all. Based solely on Garfield's prior behavior online which was
clearly biased and discriminatory towards several marginalized groups, he should have been
banned or severely sanctioned several years before the precipitating incident. Instead, when
Garfield's behavior was brought to the Community Working Group in late 2016, it emboldened
Garfield by confirming that he had not violated the code of conduct. Often community managers
"kick the can down the road"—that is, avoid confrontation over smaller problems, only to end up
with a larger, more complicated confrontation over bigger problems. If your committee is
hesitating to enforce the code of conduct on small issues, it helps to consider what larger

problems you are preventing by acting now. (This is what it looks like when a community doesn't
understand the Paradox of Tolerance : small acts of intolerance eventually lead to widespread
intolerance that takes over the entire community.)

Ignoring behavior outside community boundaries: Drupal leadership considered Garfield's
behavior on private online forums and at private events to be out of scope when deciding
whether Garfield should be allowed to participate in Drupal. However, these actions outside the
community provided relevant and useful information about the likelihood of Garfield harming the
community in the future.

"Positive" codes of conduct with extraneous material: The early 2017 versions of the
Drupal Community Code of Conduct and Event Code of Conduct did not follow our
recommendations for codes of conduct. In particular, at the time of these events, both codes of
conduct were framed almost entirely in "positive" terms (i.e., lists of acceptable behaviors rather
than lists of unacceptable behaviors ) and included a lot of extraneous material on how to
collaborate more effectively that should be part of another document. The Event Code of
Conduct had slightly more language about what not to do, but otherwise resembled the
Community Code of Conduct. As a result, as long as Garfield did not behave in a harassing
manner at a conference, he was not clearly violating the code of conduct. After this incident, the
Event Code of Conduct was updated to include a fairly standard list of unacceptable actions, but
is still framed mostly as a list of positive behaviours and contains other extraneous material. At
the time of this writing, the Community Code of Conduct has no major changes .

No formal code of conduct committee: Drupal had no communitylevel code of conduct
committee at the time of the incident. A Community Working Group took on some code of
conduct enforcement responsibilities but had limited power and responsibility. For example, the
2017 Event Code of Conduct said "You may contact the Drupal Community Working Group to
help mediate or resolve issues [...]" and a 2017 statement from the CWG says "In case of
disagreements, we guide community members through an established Conflict Resolution
Process in an effort to help defuse tense situations and keep discussions productive." At the
time of the report, the project leader, Dries Buytaert, made decisions about community
membership by himself. Some project leaders may have expertise in code of conduct
enforcement, but most will not. In any case, a small committee will generally make better
decisions than one person. The fact that another community member personally confronted
Garfield and asked him to leave suggests a lack of either trust in or knowledge of the
projectwide code of conduct process. Drupal should have formed and publicized a formal code
of conduct committee empowered to eject people from the community.

Effort to get agreement from perpetrator on sanctions: Rather than informing Garfield that
he was ejected from the community, Drupal leadership tried for several weeks to convince
Garfield to agree to and cooperate with being banned in a way that minimized public attention.
Garfield used this time to prepare a public campaign to discredit the leadership and its decision

to ban him. Drupal leadership should have informed the perpetrator of the sanction without
attempting to get agreement.

Prioritized privacy and secrecy over keeping community safe: Drupal leadership conducted
its initial discussions with Garfield privately in part because he requested privacy around his sex
life. Emails from Buytaert also express considerable concern for the reputation of the Drupal
community as a whole if the information about a major Drupal contributor being involved in the
Gorean subculture were publicized. Even after Garfield went public about the details of his sex
life, Buytaert continued to keep some of the details that prompted the ban secret, a decision that
engendered distrust given Garfield's framing of Butyaert's actions as kinkshaming. As a side
note, it is unclear how "secret" Garfield's involvement in the Gorean subculture was even before
his official announcement; for example, he engaged in Gorean mannerisms in public , and used
the same username, "crell," to contribute to Drupal projects and participate in Gorrelated
activities online for many years prior to this incident. While privacy has its place, Drupal
leadership should have prioritized keeping their community safe over protecting the perpetrator
from relevant and necessary embarrassment as the result of his abusive behavior. The outcome
demonstrates that secrecy did not serve the reputation of the Drupal community either.

No initial public announcement of decision to ban: Buytaert privately informed Garfield of
his decision to ban him from Drupal and then continued to discuss the matter with Garfield over
several weeks without a public announcement. This allowed to Garfield to make the first
announcement of his ban and control the framing of the ban as kinkshaming and involuntary
outing of an oppressed sexual minority by the powerful Drupal leadership. As a result, Garfield
convinced a large portion of the Drupal community that he was the target of unjust oppression
against a minority sexual preference, rather than a person who advocates for and engages in a
lifestyle based on the fulltime biologically determined enslavement of women. Drupal should
have made an announcement on an official project blog shortly after informing Garfield of their
decision.

Allowed perpetrator to frame himself as the victim: In his first post , Garfield framed his
action as an involuntary "selfouting," invoking the taboo against outing of queer people against
their will. Garfield then compared himself to gay people, Muslim women wearing hijab, Jewish
people saying "Shalom," and targets of religious persecution in general, and claimed he could
not be sexist because he worked with women. Using the Paradox of Tolerance and our guide on
how to judge competing claims of marginalization , it is clear that "people who believe that
women are biologically suited to be slaves" is not a marginalized group. In particular, Garfield
cannot claim that he is merely roleplaying sexual fantasies, similar to many people who engage
in BDSM, for several reasons: because he engaged in sexual role play in public with
nonconsenting observers, because he openly asked for help grooming girls too young to
consent for later sexual abuse , because Gorean subculture often advocates fulltime
submission and slavery in a manner than crosses the line from kink to abuse , and because he
advocated for literal slavery and for biological sexism in public . Garfield is attempting to relabel
abuse and bigotry as a "kink" when actual kink requires informed consent . In general, you

should be wary of a someone with a great deal of privilege (male, cis, straight, white, etc.)
equating their treatment with the systemic oppression of historically marginalized groups they
aren't personally part of. This is almost always appropriation of a marginalized identity, rather
than good faith solidarity.

Overly detailed public announcement of ban on personal blog: When Drupal leadership did
publicly comment on the ban , it was on a personal blog owned by Buytaert. The announcement
was much too long for the purpose, included many extraneous details, and was vague and
unclear on the main issue why Garfield had been banned. By talking about several issues
unrelated to Garfield’s ban from Drupal, it allowed Garfield to continue making allegations of
discrimination on the basis of his alleged status as an oppressed sexual minority and branched
the discussion in several new directions. Drupal leadership should have written a shorter, more
concise announcement of the ban and published it on an official communications channel.

Clarification update included extraneous details: Buytaert's (first) clarification update did
clarify two important misunderstandings in the original announcement, but included a response
to Garfield's allegation that the community member who tried to eject him was not acting in line
with Drupal community values. This encouraged a shift in conversation to the behavior of the
other community member, away from the original, far more serious allegations against Garfield.
Drupal leadership's first clarification update should have stuck to the original topics, and they
should have published a separate announcement about the other community member.

Multiple updates to original announcement: Buytaert made a total of five updates to his
original blog post ; four more than we recommend. Each update intensified discussion and
prompted another round of critiques and responses, rather than convincing new people that
Drupal leadership had taken the right action. Drupal leadership should have made only one
update and then focused on updating its code of conduct procedures internally.

Sanctioning perpetrator for harming the organization's reputation: One of the updates
included this statement: "Larry's subsequent blog posts harmed the community and had a
material impact on the Drupal Association, including membership cancellations from those who
believed we doxed, bullied, and discriminated against Larry as well as significant staff
disruption. Due to the harm caused, the Drupal Association is removing Larry Garfield from
leadership roles that we are responsible for, effective today." As many people correctly pointed
out, any person pointing out actual abuse in Drupal could be sanctioned for doing so under this
logic. Garfield should have been removed from leadership roles for his many comments
promoting unfair bias or discrimination against marginalized groups, not for criticizing or harming
the Drupal Association.

Delaying or refusing to engage expert advice: Drupal leadership does not appear to have
sought out the advice of experts for weeks or months in several relevant areas: code of conduct
enforcement, public relations, community management, harassment/discrimination law, or

similar areas. Free online guides in all these areas also exist. Drupal leadership should have
sought outside expert advice sooner and in more areas.

Equal punishments for violating privacy and alleged sexual assault: The person who found
Garfield's expressions of his support for women as slaves on membersonly web sites was
punished for violating Garfield's privacy and for pressuring him to leave the community with
what could have been interpreted as threats of blackmail. This whistleblower’s punishment was
identical to Garfield's ultimate punishment: they were removed from all positions of leadership
but allowed to continue participating in the community. This decision sent a highly
counterproductive message: that finding and sharing semiprivate information that shows
someone may be a threat to the community and pressuring them to leave is as harmful as
actually being a threat to the community.

While respecting privacy and confidentiality are important, they should take second place to
protecting the community from future harm . Claims of potential blackmail are more concerning
than violating privacy, but given the vagueness of the Drupal Code of Conduct and the
ineffectiveness of the Community Working Group, what Garfield characterizes as blackmail
could also have been the result of poor community governance rather than personal desire to
harm or frighten Garfield. Overall, we consider violating the privacy of an abuser by sharing
information about their abusive actions and pressuring an abuser to leave a community to be far
less harmful than Garfield's actions. For example, if a community has a rule against publishing
private correspondence, it must have an exception that allows publication of harassing private
correspondence. Garfield should have been sanctioned more strongly than the person reporting
his abusive behavior .
Conclusion
Overall, this incident is a demonstration of the magnitude of trouble possible when a
determined, powerful, and committed abuser participates in a community with a weak and
poorly enforced code of conduct. Much of the trouble could have been prevented with a
wellwritten code of conduct, a formal code of conduct committee with full powers to act, training
for the code of conduct committee, a quick and meaningful response to the first violations of the
code of conduct, and sensible limits on followup communication.
Summary
After reading this chapter, you should have confidence that following the recommendations in
this guide results in better realworld outcomes, even though some of the advice is
counterintuitive or unpleasant to follow. In particular, you should understand what benefits your
community gets when it:

● Responds quickly to control the framing of the incident.
● Publicizes its response, even for small incidents.

● Pays more attention to the impact of someone's behavior than their intent.
● Keeps records of small incidents and consults them when getting reports of new
problems to recognize patterns of suspicious behavior.
● Treats bad faith code of conduct reports as violations of the code of conduct.
● Enforces the code of conduct equally on people with more and less privilege and power.
● Is skeptical when competent adults claim they accidentally violated the code of conduct
through social ineptitude.
● Avoids activities which have a higher risk of harassment and emphasizes that the code
of conduct still applies to higher risk activities.
● Responds swiftly to small violations of the code of conduct to prevent larger problems
later.
● Takes into account the behavior of members outside the community.
● Removes extraneous material from the code of conduct.
● Has a formal, trained code of conduct committee.
● Does not try to get agreement from the harasser on the response.
● Prioritizes the safety of the community over attempting to keep embarrassing incidents
quiet.
● Makes short, clear, concise announcements of responses on official community
communication channels.
● Responds to criticism of its decision only once, and only to clarify any genuine
misunderstandings or to explain how new information changed its decision.
● Allows criticism of the organization under the code of conduct.
● Seeks out expert advice when necessary.
● Punishes violations in proportion to the incident.
● Doesn't punish people for reporting harassment improperly with the same or greater
consequneces than the people actually doing the harassment.


● " The Myth of the Male Bumbler " by Lili Loofbourow
● " Higher Risk Activities " by Mary Gardiner et al.

Acknowledgements
This guide is the product of the work of many different people. Thank you to Mary Gardiner for
writing the first version of this guide, for being my coauthor and fellow advocate for the first five
years I worked on codes of conduct, and for reviewing many parts of this guide. I am incredibly
grateful to have Annalee Flower Horne as my professional editor and collaborator. She brought
both professional writing chops and her expertise in writing and enforcing codes of conduct to
this guide. Without her insight and framing, some of the most important sections of this guide
would not exist. YVonne Hutchinson and Paloma Figueroa at ReadySet Consulting also
provided valuable professional editing and improvements in structure and presentation. August
Bournique made the language clearer and more beautiful. Russell Coker, Joseph Reagle, and
Sara Smollett reviewed this guide and/or made many useful edits and improvements. Thank you
to the many anonymous reviewers who volunteered their time and energy to make this guide
better. Thank you to the organizers of DjangoCon EU 2017, Write the Docs EU 2016, and
PyGotham 2017, for writing transparency reports and giving us gracious permission to reprint
them. Thank you to all the conference organizers who have published public summaries of their
code of conduct reports. Thank you to the various members of Geek Feminism who shared their
knowledge and expertise on this subject. Finally, a huge thank you to the many Ada Initiative
supporters who believed in me and Mary Gardiner, supported our initial work on codes of
conduct, and ultimately made this guide possible.

Appendix 1: Additional resources
This is a list of resources helpful to writing, adopting, and enforcing a code of conduct, arranged
in the order they are first referenced in the text.

Introduction:

Conference code of conduct resources on Geek Feminism Wiki
Community code of conduct resources on Geek Feminism Wiki
Code of conduct training resources from Frame Shift Consulting
" Tips for interviewing survivors " by RAINN
Ally Skills Workshop handout by Frame Shift Consulting, Dr. Sheila Addison, and Mary Gardiner

Chapter 1:

Paradox of Tolerance on Wikipedia
" Angry Black woman " stereotype on Wikipedia

Chapter 2:

Intersectionality on Wikipedia
Jim Frenkel at WisCon 38 incident on Geek Feminism Wiki
Timeline of Incidents on Geek Feminism Wiki
" ‘Why Didn’t You Report It?’ " by s.e. smith
" 'Why don’t you just hit him?' " by Mary Gardiner
" Women, Race, & Class " by Angela Y. Davis
" Alcohol and Inclusivity: Planning Tech Events with NonAlcoholic Options " by Kara Sowles
" Higher Risk Activities " by Mary Gardiner et al.
" Inclusive Offsites " by Sara Smollett
Example community code of conduct on Geek Feminism Wiki

Chapter 4:

Meeting role cards by Frame Shift Consulting
" How 'Good Intent' Undermines Diversity and Inclusion " by Annalee Flower Horne
" The Al Capone Theory of Sexual Harassment " by Leigh Honeywell and Valerie Aurora
Djangocon.eu 2017 conference transparency report
Write the Docs 2016 Prague conference transparency report
PyGotham 2017 conference transparency report
" Study Reveals The 6 Key Components Of An Effective Apology " by Amy Morin
" Take Responsibility For Handling Abuse " by Annalee Flower Horne

" What Reporting Sexual Harassment Taught Me " by Simine Vazire
" No More Rock Stars " by Leigh Honeywell, Valerie Aurora, and Mary Gardiner
" How to Organize Tech Workers to Change Company Policy " by Liz FongJones
" How 'Good Intent' Undermines Diversity and Inclusion " by Annalee Flower Horne
" What is DARVO ?" by Dr. Jennifer J. Freyd
" What kind of person makes false rape accusations? " by Sandra Newman
Oppression Olympics on Wikipedia
List of articles debunking autism as the cause of harassment on Geek Feminism Wiki
" The Myth of the Male Bumbler " by Lili Loofbourow
" Acceptance, kink shaming, and calling out bad behavior " by Anabelle Bernard F
" Mythcommunication: It’s Not That They Don’t Understand, They Just Don’t Like The Answer "
by Thomas MacAulay Millar
" BDSM vs. Abuse Policy Statement " by National Center for Sexual Freedom
" Google Employees: Our Executives Engaged in Abuse. Don’t Let Kink and Polyamory Be Their
Scapegoats. " by Liz FongJones
" Social and Cultural Aspects of Drinking " by Social Issues Research Center
" Is Shame Necessary? " by Jennifer Jacquet
Charles' Rules of Argument on Geek Feminism Wiki
" The power of framing: It’s not what you say, it’s how you say it " by Steve Rathje
" Why Asking What Adria Richards Could Have Done Differently Is The Wrong Question " by
Deanna Zandt
" The Gift of Fear " by Gavin de Becker
" Staying Safe when you Say #MeToo " by Leigh Honeywell
Tall Poppy : Tools and services to help companies protect their employees against online
harassment and abuse

Appendix 2: Reporttaking form
1. Find a quiet place where others can't overhear you if possible. Bring something to take
notes with.
2. Ask if the person wants to make a formal code of conduct report. If they ask for promises
about how it will be handled, let them know that you can't make any promises on behalf
of the code of conduct committee, but that the committee will do their utmost to protect
the reporter's confidentiality and safety.
3. If they DON'T want to make a formal code of conduct report, feel free to decline to listen
to their report, especially if you are a member of a marginalized group who is expected
to perform more emotional labor than others.
4. Ask them to describe the incident. Some useful information to request (but respect their
wishes if they don't want to share it):
○ Identifying information for the alleged harasser
○ Reporter's name and contact information
○ Time and date of incident
○ Place of incident
○ What happened
○ Any other people involved
5. If there seems to be any imminent physical danger to anyone, follow your organization's
security plan immediately and come back to taking the report when people are physically
safe. Note your action in your report.
6. If it seems appropriate, ask if there is anything you or other staff can do to help anyone
feel more safe at that moment (e.g., finding a friend of theirs to stay with them, or giving
them a private space to sit in).
7. Thank them for their report.
8. Send the report to the code of conduct committee by [ INSTRUCTIONS HERE ].

Reminders:
● Do not share any confidential information with anyone other than the code of conduct
committee.
● Do not ask the reporter or the target(s) for any solutions or suggestions on how to handle
the incident.
● Do not pressure the reporter to contact law enforcement if they don't want to, as in many
cases this can make the reporter, target(s), and/or your community less safe. If the
reporter does want to contact law enforcement, support them if you are safely able to do
so.
● Don't assume that the reporter will trust any member of staff or person they don't
personally already know and trust (including yourself).

CC BYSA 4.0 M
ary Gardiner , Valerie Aurora

How To Respond To Code of Conduct Reports by Valerie Aurora and Mary Gardiner

Uploaded by

Copyright:

Available Formats

How To Respond To Code of Conduct Reports by Valerie Aurora and Mary Gardiner

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

How To Respond To Code of Conduct Reports by Valerie Aurora and Mary Gardiner

Uploaded by

Copyright:

Available Formats

You might also like