Howdy,
This is a difficult essay to write. I know people will disagree with me. But, it is an important and timely topic. Issues falling under the “ethics” umbrella seem to be riling computer science more than at any time I can remember. Many of these end up in my inbox.
I don’t like the word “ethics.” It is too broad an umbrella and captures everything from plagiarism, to the potential negative uses of our research, to creating hostile workplaces. So, when someone says “there is an ethical issue,” I have not a clue what they mean. My own model has three “buckets”: publications, harassment and discrimination, and negative impacts of our research. Before I describe these in detail, let me talk about how ethics are handled by SIGGRAPH and ACM.
One of the biggest benefits of being a SIG of ACM is that ACM handles almost all of these issues. They don’t always handle them the way I would like, but they do hire the lawyers, pay for the insurance, setup the review committees, and hire the investigators. It is unimaginable that SIGGRAPH could take these things on. I do wonder if the lack of a parent organization is why some of our neighboring communities are struggling more than SIGGRAPH to deal with some of these issues. At a high level, when a complaint is filed, ACM hires investigators, volunteer committees review the investigation, and potentially a sanction (e.g. you cannot participate in ACM activities for X years) is imposed. This approach borrows some elements from the US criminal justice system: the investigators are analogous to prosecutors and the committees are like juries. But it does leave out an impartial judge or referee to guide the process and to interpret the policies in place at the time of the alleged infraction, which is an important part of the US criminal justice system.
Also of note, thus far ACM has worked hard to keep all allegations and sanctions confidential, though they don’t always succeed. Beyond that, because nothing is public, subcommunities create lists and whisper networks. Clearly this is not an ideal situation. Many professions make allegations and/or sanctions public. This creates the perverse incentive to use the complaint process to avenge some other grievance. Veterinarians live in constant fear that if someone’s beloved pet dies the owner will file a false complaint to the board and, because that complaint becomes public, it may hamper their career. Another analogy would be only showing the one-star reviews for restaurants—only the complaints. Yet another analogy that might hit closer to home is “ratemyprofessor” or reddit. Clearly, this alternative is also not ideal. The ACM council will soon be grappling with these tradeoffs. I do not envy them.
Now let me discuss my buckets:
Publications: This bucket includes academic dishonesty—acts of plagiarism or falsifying results—which humans have been dealing with for hundreds, if not thousands, of years. (The word “plagiarism” dates to the early 17th century, but surely there were earlier cases.) Over the centuries, we have gotten reasonably good at handling most forms of academic dishonesty. ACM handles these complaints for us and they do a good job.
Of course challenges remain; just before drafting this essay, I read this piece about collusion rings. I have heard stories of such behavior within SIGGRAPH before my time, stories of powerful people corrupting the review process for personal gain. But, I think the SIGGRAPH community established norms that made such practices uncommon before my involvement and I have never witnessed any major abuses. However, I am sure there are unconscious biases when a reviewer knows the authors’ names. One strategy we are using to combat this problem is switching almost all of our reviewing to double blind. It is not perfect, but it is a big step forward. I’ll give a shout-out to Olga Sorkine-Hornung for implementing fully double blind review for our main SIGGRAPH conferences; that was a pretty heavy lift.
A more difficult issue in the publications bucket is inappropriate content or examples. Such standards vary across time, culture, and personal taste. Currently, I would not use the Lena image, a decade ago I would have used it without a second thought. SIGGRAPH is constantly modifying its policies to keep up with the times; our governance committee, which reviews our policies, meets every other week. As an example, five years ago our policies allowed exhibitors to wear only body paint; clearly that was a policy from a different time, and it has been updated.
Harassment and Discrimination: I think of harassment as being overt, usually intentional, actions that make someone else uncomfortable. I think of discrimination as being subtler than outright harassment. It is less overt behavior, but still unacceptable. Sometimes it is intentional, sometimes it is the result of implicit bias, and often it is executed through micro-aggressions.
Historically, I do not think SIGGRAPH or ACM did a great job with harassment, but we have made strides in recent years. Official policies are now in place and we introduced SIGGRAPH Cares as a first point of contact for victims (ACM also introduced a Cares committee). Both SIGGRAPH and ACM have created committees to support diversity, equity, and inclusion (though ACM leaves out equity). I am also pleased to say that ACM will now allow anonymous harassment complaints, sparing victims the burden of dealing with an investigation. Though this step does have drawbacks: it is very difficult to investigate an anonymous complaint and it is very easy to make a false anonymous complaint as revenge for some other grievance. I am not sure how they plan to handle anonymous complaints.
Spurred by recent events, ACM and SIGGRAPH are doing some deep thinking about discrimination and I do expect some solutions moving forward. SIGGRAPH has an outstanding Diversity, Equity, and Inclusion standing committee, led by Tony Baylis, that is helping all aspects of our organization. One reason that discrimination is tricky is that discrimination, or making decisions, encompasses a large part of what we do as humans: we make judgements on everything from which papers to accept, to whom we choose to hire, to what music we like. You cannot look at a restaurant menu (remember those) without discriminating. It is the basis of that discrimination that is the issue. One small step that we have already taken is that, going forward, all our calls for submissions will include our anti-discrimination policy. Our double-blind review should also help.
Negative Impacts of Our Research: Thus far, the community has left the discussion of this bucket largely to the press, which is probably a mistake. I have heard stories of researchers ambushed or taken out of context by the media and made to seem like evil-doers. SIGGRAPH has run a few workshops on “Truth in Graphics” discussing the implications of some of the work our community does. There was even an effort to codify the acceptable uses of our image and video manipulation technology. Unfortunately, that proved too difficult a task. There have been suggestions for an “ethics” section in papers or on review forms to discuss the potential negative impacts of research, but that idea has not yet gained traction.
I have a few final thoughts.
I think it is extremely dangerous to assess past behavior by current standards. Like the use of the Lena image, policies and rules and culture evolve over time. For most of my life the rule was: don’t date your students, but any legal behavior outside of that relationship is okay. Of course, that approach ignored the incredible power disparities that exist outside a direct teacher-student relationship, and ACM and SIGGRAPH have updated our policies to recognize that behavior in any professional relationship can be inappropriate. Our current rules will likely be modified as time goes on and culture shifts. It is not fair to punish someone who was playing by the rules of the time. Contrariwise, I believe it is right to sanction individuals for violating standards that were in place at the time of an infraction—if I intentionally falsified the results in my thesis, Berkeley should revoke my PhD. Of course, they should not take such action simply because my thesis is obsolete, that is how science evolves. We also should not ignore prior art that includes images or examples that are unacceptable today. The Zimbardo prison experiment could never happen today; it was one of the impetuses for IRBs. But ignoring the results does not make us more ethical. I don’t mean to make light of these issues, but as another analogy, I do hope no one judges me based on my fashion choices in the 80s.
Finally, I want to end on a positive note on an unpleasant topic. Though there are no easy answers, ACM SIGGRAPH is committed to creating a diverse, equitable, and inclusive environment. From how we choose our leaders, to our speakers, to our program committees, we strive to always be welcoming to everyone. And we are making progress. Our DEI committee is doing great work, we created SIGGRAPH cares (we are looking for a new chair, email me if interested), we have explicit policies regarding harassment and discrimination, the community is beginning to think about the potential impacts of our research, and we have had a few test cases regarding inappropriate content. While there has been some inappropriate public shaming, thankfully our community has not devolved into social media flame wars or creating long lists of names that resemble McCarthyism.
Adam Bargteil
ACM SIGGRAPH Chair