Waking Up From Electric Dreams

Two years ago I visited Dagstuhl, a research center in Germany, for a week of game AI research. I was writing Electric Dreams at the time for Rock, Paper, Shotgun; a series about games, AI and research. In the piece about Dagstuhl, I wrote about the fear I observed that academic pressures and economic shifts would stifle great, exciting games research:

Like every other part of the games industry, games researchers have a contribution to make to the future of games. If we don’t make spaces where we can do this work, Michael Mateas’ “country of possibilities” may remain undiscovered forever.

Last week I returned to Dagstuhl, and once again found myself discussing the health of game AI research. But this time, the problem wasn’t funding agencies or university administrators: the problem was us. This is a fairly introspective, Inside Baseball-esque post, but I’ve come away from Dagstuhl with a powerful urge to write it, so I hope you’ll forgive me. If you work in games research, particularly AI, and particularly if you were at Dagstuhl, I implore you to read it.

I originally proposed that we write a Code of Conduct (CoC) for Dagstuhl several weeks before the event. I proposed it because I had seen many other events do it, and read how it helps make people feel safer, happier and better able to enjoy an event. In 2017, a year of almost non-stop horror, writing a CoC feels like a positive step to take regardless of what event you are organising. In the end, we began the event without a CoC, just like previous Dagstuhls. But what made it different was that the other organisers (I was not present due to illness) explicitly announced that having a CoC had been discussed and decided against. This kicked off a chain of debates, arguments and ill-feeling that permeated through the week, and in my opinion serves as a terrible omen for this community.

These problems peaked when when several researchers gave up their workgroup time to thoughtfully (and bravely) sketch out a CoC of their own. After presenting what they had done they received pushback from many members of the community, complaints that this was unnecessary, and (I have heard – I was still ill at this point) even a bizarre discussion of why remembering people’s pronouns is too much work. Learning about this later in the week, I found it extremely troubling.

When I proposed writing a CoC, I had in mind several benefits:

It makes the implicit explicit. The phrase “Well, everyone knows…” doesn’t need to be uttered, because what “everyone knows” is here in black and white.

It offers a way to explain issues to people. If something does happen, we can use the CoC to explain specifically why it was an issue, if that person doesn’t understand.

It makes a statement to newcomers. If you’re wondering what kind of event this will be, and the first thing you see is us taking a stand to make a safe environment, that’s a great first impression.

It helps people feel confident. One complaint about CoCs is that people are afraid they might ‘unintentionally’ offend someone – but CoCs are about the opposite! They help you think through your actions and feel confident about what you’re doing.

There are loads of reasons to have a CoC besides these ones – I learned some new ones at Dagstuhl, in fact (they help set expectations for people if something bad happens, like what help they’ll receive; they help combat the Bystander Effect). They have so many positives. Here are some of the common negatives I hear in response to CoCs:

If you’re worried a CoC makes it look like something bad has happened: something bad has happened, something bad happens at literally every single tech event. Bad things happened at this Dagstuhl – maybe you didn’t see them, maybe you didn’t notice them, but they happened. Some are small, some are large, but trust me – writing a few rules down isn’t going to cast aspersions on the reputation of computer science. What it will do is begin to formalise and give shape to a vision of the future, where everyone feels safe.

If you’re worried a CoC isn’t needed because we’re okay: great! That’ll mean it doesn’t impact any of us, and still has all those other good side effects (welcoming newcomers, setting a standard for new members, being an example to other communities). But if you’re like me – a white heterosexual guy born speaking English – you’ll know that the world is full of situations that are harder for some people than others. Just because a place seems full of friends who understand each other, doesn’t mean there aren’t uncomfortable relationships and power dynamics at play.

If you’re worried a CoC might mean you have to have awkward conversations or makes you uncomfortable thinking about ways you might be messing up: well, yeah. I know, it sucks. But I can promise you this: if you try in good faith, people will help you, forgive you when you slip up, and you will feel better. A CoC is not an edict or a bounty notice. It’s a commitment we all make to getting better, together.

If you’re worried a CoC might mean you have to change, or put in extra work thinking about what you say, or not tell a joke sometimes: it absolutely will. And it’s absolutely right that it does so.

I’m writing all of this stuff because it doesn’t make you a bad person if your initial reaction to a Code of Conduct is “I’m not sure about this” – this is a learning experience for all of us. But I’m also writing this because I don’t want other people, or myself, to have to keep explaining this. At some point the explanations and justifications and proposals have to end, and we have to decide to move on. And on that note, let me pivot into the second half of this post: our community is broken and something needs to change.

The CoC thing concerns me because although we all need time to learn about new concepts, the fundamental argument of a CoC is “this community could be better for everyone”, and the immediate pushback against this fairly straightforward idea suggests that a portion of our community are not interested in changing for others in even simple ways, and are even less interested in admitting that more serious issues may exist. Instead, we expect people with issues to adapt themselves to the status quo, or simply leave.

This is not a way to build a healthy, diverse, sustainable community. A CoC is a piece of paper that says we won’t be horrible to each other – if we can’t agree that this is a good idea, what hope do we have of fixing the lack of gender and racial representation among organisers, speakers, attendees, students and researchers? What hope do we have of uniting against increasingly commercial demands on our research, our time and our students? What hope do we have of building a brighter future for the world if we can’t even envisage a mildly better today for ourselves?

I wanted to close with a call to action or a list of next steps, but I’m honestly torn over who should do what. For myself, I think I need to be bolder, to no longer ask for small concessions but to demand more meaningful action, to fight a little harder – please join me in this if you can, in trying to have uncomfortable conversations and risk things, if you have that luxury, to make these things harder to ignore.

If you objected to the code of conduct – or would have done, were you there – then certainly please read this post and reflect on it a little. I wrote this explanation because I imagine some people will not have thought about the topic much before, and many people at Dagstuhl convinced me that change is possible and we should try to help people understand.


At the same time, I want to be really clear here: this is not an invitation to accept the status quo, to nod at this blog post and then move on as usual. We have bumbled along for too long, with little improvement to the many problems our community faces. Disagreements over a simple document are just the tip of the iceberg, and a failure to deal with these problems will undoubtedly split this community in two. And increasingly, I find myself willing to accept this as a solution, because if we cannot gently change existing spaces, then founding new ones is the next best solution.

I’ve disabled comments on this post for the same reason I did not bring this up in the closing moments of Dagstuhl: because this is not up for debate. These are not discussions, for me. These are simple facts about the community, that every one of us has to reflect on and work to change. The cracks that run through our community are widening, and if they become so wide that one cannot stand with a foot on either side, I know which side I will be stepping onto.