pieterh wrote on 04 Dec 2015 16:50
Whenever people gather together, there's a risk of misbehavior. In tech conferences we are trying to solve this by declaring and enforcing a "Code of Conduct." I've had and seen harassment multiple times, over the years. It seems to me that the Codes of Conduct are not working. In this article I'll try to explain the problems, and propose a fix.
Update
I'm going to link to Stephanie Zvan's critique of this article.
Identifying the Problem
Unwelcome sexual attention, aggressive conversations that won't end, intrusive photo-taking, offensive language and behavior… most of us have seen or experienced this. The mainstream solution is obvious: tell people to stop it, or else.
Here is a typical statement by a large tech conference:
All delegates, speakers, sponsors and volunteers… are required to agree with the following code of conduct. Organizers will enforce this code throughout the event.
Many participants do not like such statements, and the discussions over codes of conduct can become heated, emotional, and ugly. In my work in group psychology, there's a rule of thumb I have developed: When two sides argue emotionally it is because both are working from false assumptions.
And indeed, I think the mainstream Code of Conduct model is based on false assumptions. The theory of harassment (let's call it "Model A") has these assumptions:
- Anyone can be the harasser.
- Harassment is a motiveless act.
- Outlawing harassment will stop it.
All three of these are provably false. Bad actors do not respect rules. They habitually avoid sanctions, and reflect them onto others. You can read my article, "Ten Myths About Harassment" to see other related myths.
By accepting and then acting on false assumptions, we get divisive policy and argument. And I think the outcome, with division and ill-feeling, is worse than having no code of conduct at all.
Here is a more accurate theory of harassment (Model B):
- A specific subset of participants (the "bad actors") are systematic harassers.
- Bad actors harass as a persistent strategy for getting power, sex, amusement, etc.
- They treat rules and laws as challenges to work around, or misuse.
The term "bad actors" is a convenient euphemism for sub-clinical psychopaths. When you understand model B, the picture shifts.
Let's take a typical scenario, and see how it works with both models. The scenario is: at an event, participants are drinking. Mallory makes an unwelcome sexual advance towards Alice. Alice feels very uncomfortable and moves away. Mallory insists, and follows her. Alice explicitly asks him to stop, and moves. Once again, Mallory approaches and tries again.
Model A explains it thus: Mallory was drinking and didn't realize that his behavior was causing distress to Alice. By explaining what it means to be decent, and enforcing that, he will refrain from harassing people.
Model B explains it thus: Alice was drinking, and became more open and vulnerable. Mallory saw this, and decided to exploit it. Her display of fear and moving away drove him to be even more predatory. He was trying to isolate her, so she would give in.
It's not that alcohol turns good actors into bad actors. Rather, it makes the good actors more vulnerable. This distinction is critical. This is the key problem: people in a foreign country, who may not be with friends, are easy targets. Instead of solving Mallory's behavior, we must solve Alice's vulnerability.
Mallory is always going to be there. In a group of 100 people, we will have a handful of Mallorys. In a conference with 1,000 attendees, we'll have dozens of people capable of harassing others, if the conditions permit it.
In my view, events must be radically inclusive. To exclude people on any basis except real danger to others is a poor strategy. It introduces real, major risks such as false positives. To be accused of a Code of Conduct violation, and then excluded from an event can be life damaging. Organizers wield massive power, and bad actors can turn this power on people they see as a threat.
Let me be explicit about this. If Alice complains to the organizers, she puts herself at even more risk. Mallory now appears to be calm and sincere. He accuses Alice of stalking him. He explains how she would not leave him alone. He looks vulnerable, gentle, and seriously distressed. Alice is angry, outraged, and emotional. Who do the organizers believe?
And Mallory comes in all genders. Harassment is not a gender issue. So while sanctions must be possible, the policy goal must be to make them unnecessary.
The second problem with exclusion of supposed bad actors is that it is prejudicial. Even bad actors add to the diversity of a crowd. This is hard to admit. When I cross a bad actor, my instinct is to push them out, reject them, and tell everyone what happened. Yet I believe that is the wrong reaction.
Rather, we must understand, and then negate and switch off their predatory behavior.
Bad actors are pragmatic, and driven by clear calculations of opportunity versus risk. So the solution is to remove the incentives and opportunities for misbehavior, and raise the risks of exposure. Then, bad actors will adapt and self-select. Either they will behave, or they will go elsewhere.
This removes the need for enforcement, and the risk of false positives.
Designing a Solution
As a parent, I cannot stop my children from being exposed to bad actors. They will inevitably, regularly, come across people who see them as potential targets. It will happen on-line, in the streets, in schools, at friends' homes.
One option is to keep them at home, or never let them roam alone. Yet that seems counter-productive in many ways. So instead, I teach them how bad actors think and operate. I explain what kinds of conversation and interaction are dangerous. And then I teach them what to do if such a thing happens. Find an adult you can trust, I say. Tell them what happened, call me, and wait.
I explain to my children how bad actors stalk the Internet looking for victims. I explain what grooming looks like. I teach them to never give their real name, age, or location to strangers on-line. Never to send pictures of themselves. Never to accept a statement from a stranger as true, no matter how convincing it is.
This is the model I believe we need for conferences and similar gatherings. Instead of telling people to be good, we teach people to stay safe.
Here are my specific recommendations:
- We teach awareness of the behaviors one should consider "bad." We speak to potential victims and bystanders, not to bad actors. Bad behaviors are those that bad actors use to profile, stalk, isolate, and pursue their targets.
- We teach organizers how to accept and process complaints. These cannot be anonymous. They must be discreet, with protection of privacy for both parties. There must be fair follow-up, with sanctions as needed.
- We teach organizers how to protect their guests from predatory behaviors. For example, in any space where the organizers offer alcohol, they should offer food and water, and a host who does not drink, and who patrols discretely.
- We formulate these as a written protocol that events can adopt. Rather than ask every event to re-invent the wheel, we can build standards that events can reuse, and improve over time.
Such a protocol will, I'm arguing, create safe spaces by design. It will speak to bad actors and say, clearly: you are welcome as long as you act decent. Otherwise, we are watching, and we will catch you.
A Code of Conduct Protocol
This is a raw proposal. To be expanded and refined.
- The "event" is any gathering of people under auspices of an organizer. This includes formal events and informal gatherings before, after, and during a main event.
- The "organizer" is the legally responsible entity or person who hosts an event.
- The "participant" is an adult person attending or helping with an event. Minors at events are assumed to be under protection of their parents or guardians.
- Organizers and participants have rights and obligations under this protocol, which they understand and accept.
- Participants accept that the arbitrators in any event dispute are the organizers.
- Organizers who respect and enforce this protocol may say so on their conference web site.
- In an event, every participant has the right to the sanctity of their person, their time, and their space. We define a "violation" as any act that interferes with this sanctity, without consent of the participant. Silence is not consent.
- Violations include, though are not limited to, doing any of the following without consent: talking to or following a person who has expressed their desire to end a conversation; taking close photographs of a person; touching a person in an intimate or controlling fashion; use of sexualized language; display of sexual imagery; offering unsafe quantities of alcohol; offering illegal drugs of any sort; making unjustified complaints; tampering with a person's possessions.
- In all circumstances where the organizers offer alcohol, they shall also offer drinking water, and sufficient food to keep participants safe. Additionally, they shall have one or more "hosts" who do not drink alcohol, and who are trained to intervene in case of conflict or visible distress.
- The organizers shall provide a secure, private means for reporting violations. Such a report shall be considered circumstantial unless or until backed up. A report should state the time, place, name of the person accused of committing the violation, nature of the violation, and names of witnesses if any. The accused person shall have the right of reply. This material shall be kept private.
- If a participant experiences what they consider a violation, or sees what they consider a violation happening to someone else, they should report this to the organizers.
- if the organizers feel a particular individual is the cause of more than one complaint, and does not self-correct, they may exclude that person from the event. This should be done with respect for the privacy of all parties.
- The organizers shall be aware of the risk of "date rape" drugs. If they receive information about a participant who claims to have suffered from blackouts, or who behaves in uncharacteristic ways causing hurt to themselves or others, they shall assist the participant with a medical blood test for drugs, within 24 hours.
- Organizers should provide participants with an emergency contact phone number during events, for reporting urgent incidents and asking for help.
- If the organizers receive information about a possible criminal offense by or towards any participant, they should report this to local law enforcement and provide suitable support.
Comments
You have probably heard of the Stanford prison experiment. How does it fit with your first Model B assumption?
Most people (as in that experiment) will switch off their empathy and hurt others, in specific circumstances. It's not systematic.
Portfolio
Great article, thank you.
Quick question, did you mean "discrete" or "discreet" in your article?
I can't post a link, but just google for the difference if it's unclear.
Discreet, and thanks for the fix :)
Portfolio
I have browsed over to Zwan's post, without knowing either of you, and was shocked by the way she opened her argument with a scathing and abusive ad hominem attack directed at you. This seemed to disqualify her as an expert for good conduct from the start.
I also disagree with your too charitable observation that her points are "spot on". While a lot of them are accurate, they also elaborately repeat the three 'false assumptions' you point out, i.e. absence of focus on individual bad actors, assuming that harassment is motiveless, and belief that CoCs will usually prevent it.
There is a major element missing in the discourse on all sides: the clash of cultures, communication styles and personality structures between "tech people" and "socializers".
Consequently, you have opposite ideas of what constitutes a harasser. It seems that in your view, harassers are a few socially highly competent, but ill-intentioned individuals, and we need to educate well-intentioned but vulnerable participants how do be resilient. In Zwan's world, a harasser is usually well-intentioned, but does not conform to the tacit social rule-book, and therefore needs to be regulated or sanctioned to enforce the desired rules and boundaries.
Simply put, your CoC intends to protect nerds against Zwans, and her CoC intends to protect Zwans against nerds.
The abusive language (from someone who claims to be fighting abuse) is interesting data.
I'm not sure whether the author is protecting her business, playing to an audience, or seriously claiming to own harassment and morality.
(Edited my post to remove the "Spot on" and other wrapping.)
Portfolio
I haven't read Zwan's CoC, yet, but according to pieter, Zwan's feminist CoC is about protecting women from men, and most men are not nerds.
Pieter wants to protect people from every bad actor. Pieter's CoC is not necessarily in conflict with Zwan's CoC except that her CoC has feminist assumptions that he doesn't agree with.
Although I agreed with the points in her article, her replies to your comments were painful to read. I frowned a lot.
She attacked you with such provocative language as
1) "Bullshit. Nice try, but still bullshit."
2) "You’re not going to do the work to learn. You just want to be recognized as an expert."
3) "Pieter, I suppose that if you learn nothing else from this discussion except that your pretensions to expertise where you have none and seek none are annoying".
4) "They forget about GIGO. They come to think of themselves as experts without having done any of the work." (Somehow, the work you put in thinking about the problem with the information you have and writing the book is not work although I agree that thinking alone without peer review could lead to many errors.)
At this point, I started thinking, "wait, is she a harasser?" If she is, then it is a female-to-male harassment, and it doesn't seem to be a good idea to frame it as a female-to-male harassment.
She insulted without actually telling what was bullshit. Even, people with doctoral degrees often insult without justification. This makes me strongly suspect that she couldn't explain what was bullshit in your comments and that she is a low quality person.
What did you learn from the discussion with her?
I learned a few interesting things.
She was clearly not interested in discussing what I wrote, only taking offense and then insulting me. When I asked her for a concrete critique, she told me to pay her, and then insulted me again.
I like to be polite and assume I'm the asshole in any discussion. If someone says I offended them, I'll apologize. A good actor will then step back, and either go away or start a dialogue. A bad actor will attack again. It's a classic thing: bad actors cannot resist the smell of blood. Offer them a wound and they will bite.
So, the author is behaving like a bad actor.
This was kind of obvious from the start, because anyone sincerely trying to create better spaces for our communities is always looking for issues to solve. It's how we work when we look to improve things.
Next, I learned that this person is instrumental (or influential) in defining CoC policy for many events. She pointed me to a wiki which looks fine, until you scratch deeper. Then you see a culture that claims to own harassment.
This is a familiar theme. It's the "only whites can be racist" oxymoron, turned into "only straight men can be abusers." All too familiar and if there was one idea I wanted to kill with my book, it was this one.
So to be blunt: policy for large communities is being defined by a fringe that is essentially a cult of hate. It is using all the techniques of psychopathy to construct and sell a Narrative that is based on a truth (that sexual harassment is a real problem) yet is mostly false. This narrative is sold through manipulation of guilt and empathy, and those who raise questions about it are attacked as ignorant, sexist, and worse.
It doesn't take a conspiracy. I'm not accusing Zvan of being a psychopath. It just takes a few individuals over years to construct such cultures, and they are infectious. Once you buy into such a narrative, you invest so much that it's always cheaper to defend it than to fight it.
Oh, and if you like, take screen shots of that thread because it may well disappear. :)
Portfolio
You are basing this whole piece on the fact that other people (mainstream Code of Conduct proponents) are working from false assumptions. Yet, here you make assumptions about what these people assume, pretty much out of thin air, it seems.
Where did you get these three assumptions from? How did you come to the conclusion that these three are the premises "mainstream code of conduct theory" is based on? Sources?
Thanks for this feedback.
The piece isn't based on these assumptions, though. It's based on model B, which I'm gratified to learn is understood more widely than I thought it was.
I'll rewrite this piece later, when I've had more feedback and time to think of it. It's all grist for discussion.
Model A is how many people perceive COCs, and harassment. I don't have references. It's an observation of the (to me) insanely emotional and argumentative discussions that surround COCs. People who refuse to use COCs for their events are, I believe, assuming model A. (It's a hypothesis… if you have other explanations, so much the better.)
Portfolio
Model A doesn't resemble any Code of Conduct proposer's assumptions I've seen, though maybe some of the opponents of CoCs think it does. I've primarily seen it discussed in the science fiction convention and technical conference environments.
What I've seen has been much more like:
1 - There are bad actors out there, but the organizers don't necessarily know who they all are - sometimes they're even speakers / Guests Of Honor / etc., and you've got to be able to deal with them, whether they're randoms or well-known bad actors.
2 - We don't really care why harassers do it, but if they've done it once they'll probably do it again, and even if they don't, the people they've harassed in the past shouldn't have to put up with the risk that they will. (This includes the "X is only a bad actor when they're drunk" case.)
3 - Outlawing harassment in advance gives the conference organizers the responsibility and permission to do something about the problem - otherwise, yes, some bad actors will treat the rules as a challenge to work around, and they'll show up next time, or at some other conference, and harass more people.
I don't really care if somebody's a bad actor because they're a borderline psychopath or just an asshole. Yes, there's sometimes a loss of diversity by not having Some People at your conference; there's also a loss of diversity if other people don't come back to your conference next year because of how badly they were treated, especially if they tell all their friends and random strangers.
At least a large part of the mainstream COC framework seems to stem from an absolute, and determined, view that calls itself feminist. The problem is phrased in terms of male-on-female harassment and aggression from micro to macro.
Which is valid data, yes. Yet it's incomplete. And, it leads directly to model A by implying that all men are potential threats. Whereas a broader data set shows that bad actors are equally spread among men and women, and will attack anyone they see as vulnerable, a competitor, or a fun toy to play with.
The research that backs mainstream COCs appears to disregard data it considers inconvenient, in order to push a political agenda. This is not how science works. If we actually care about equal rights (and I do) then this should disturb you even more. What we see is magical thinking (model A) being used as the core narrative for what can become a cult of hate. The science means looking at all data, and falsifying theories that can be disproven.
I think we need to care about the why because it gives us tools to model and predict how this works. We know that harassment is a tactic that bad actors use to manipulate, divide, attack. Yet there are also tactics that put people in danger, and which are common at events, and which we ignore. I mentioned offering "unsafe quantities" of alcohol. There are others. The "danger" is often slow and creeping, and stealthy.
I'm glad the article has provoked discussion: this was the point. I don't have real answers, only questions and guesses. What I am starting to see is that:
As for diversity and the right to attend events… my own take is to give people multiple chances to correct their behavior and if they still cause pain to others, to eject them. It does take time to decide, and you need material evidence.
Portfolio
I a mostly agree with that blog post.
But still is a "code" really needed? Why should a code of conduct or a protocol should be defined when some legal rules already exist to garantee the safety in the organisation of an event? Why not simply re-calling these rules instead of trying to enforce custom rules?
A public information helping to guarantee the safety of the participant is more than welcome. Ie: displaying what to do, who to call when someone put you or another person in unsafe condition should be really available to anyone.
But the definition of a good conduct is very hard and imposing a view of what is a good conduct to a group can be dangerous. Some groups, with powerful people (or rather people with a power) inside an event can take it has a way to impose their view. We all know about these behaviour policies in the past enforced by some people that have made other groups of people in danger or made their life uncomfortable.
Limiting to safety is a good step but is still ambiguous.
I think that if a protocol or an RFC should be define, it should rely on "commons" standards and always redirect to the local or international laws/rules instead of the organisers' rules. The organisers shouldn't substitute themselves to the law in any case.
You should learn more about polycentric law. It has been with us since the beginning of humanity. Read athousandnations.com /2010/06/23/polycentric-law-a-k-a-anarcho-capitalism-as-no-big-deal/ for a start.
You're basically saying it's a bad idea for universities, companies, home-owner's association and so on to have their own rules in addition to the local state's legal system.
As you said, some people can have improper views. The problem with the situation you said above is that there is no competition among authorities. With multiple authorities, if people are not satisfied with an event organizer's decision, they will turn to other event organizers of the same event. Competition among authorities provides strong incentives for them to behave.
Competition among authorities exist among open source projects, too. Open source projects maintained by bad authorities will be forked by better authorities, and people will move to better authority figures.
Increasing competition among authorities is not a panacea but helps.
No this is not what I am saying. What I am saying is that no code of conduct should go over the law. I am tired of these "codes" that violate laws consistently by judging others outside of any competence based on some principles like in the old farwest.
In short going to an event shouldn't be like beeing a member of a club. Not every time at least. If I am going to an event this is not to hear from someone, a random group what I should believe or how I should behave. Bad or good are subjectives. This is why i am cautious about it. Not saying it shouldn't be improved. Safety like I hear it is the base of the UHDR. I don't think we need more but a framework that tell to the organisers how they should make sure that attendant can find some help if thy need it.
False complaints are themselves a violation. We know from experience that predators will try to use laws and rules to persecute others. Often they will act injured and vulnerable to achieve this. They may try to shift what constitutes a "violation" to entrap others. They will exploit existing emotional arguments and amplify them, for their own benefit.
Then, there are the obligations of the organizers, which we often forget. For example, providing security and support in case of emergency.
Yet what I've started to notice is that most harassment is smooth, stealthy, and unless you know the game, almost invisible. That means victims often simply won't call for help.
The protocol has to understand these dynamics and prevent it. The text I have now isn't ready, it's missing a lot.
Portfolio
I have increasingly become aware of you as an expert on psychopaths, trolls, and harassment. In my opinion, for the code of conduct to work properly, situations should be engineered so that virtually every participant reads code of conduct.
I have some questions.
1) Are the vast majority of trolls subclinical psychopaths? How do you categorize trolls?
2) How should I deal with trolls and/or psychopaths on IRC channels and other suboptimally regulated communities?
3) In what kinds of online communities, do you think people are safe from predators?
Yes, I think deliberate trolls are psychopaths. There is a growing body of research in this direction. Once you model trolls as predators, there is no need to categorize except from "smaller" to "larger" perhaps.
I will write an article on dealing with online harassment, based on this model. I think the main thing is to never take it personally, and to fight back in the right way. That means: don't argue, instead aim to block, report, and expose. Treat trolls like annoying mosquitoes.
No community is entirely safe from trolls and predators, yet those with clear rules (as we use in our open source projects) seem most robust. Psychopaths need chaos and confusion. If they can't create that, they tend to go away.
Portfolio
In my experiences, unclear rules led to a lot of confusion around which questions should be closed on stackexchange. Besides, stackexchange was designed such that trolls could gain administrative power over time by giving the same answer to a specific type of questions.
On stackexchange, random questions are closed for being off-topic and being subjective and for not sticking to the proper question format. However, stackexchange communities don't define what is off-topic clearly. And, the decision of what is subjetive is subjective. Stackexchange prepared lots of excuses trolls could use to close random questions. Those rules made people more vulnerable to trolls.
When I design a new organization, I will definitely think about making people less vulnerable.