By now, you’re probably sick of hearing about the Facebook data science experiment on emotion manipulation. Depending on whom you ask, it’s either a serious breach of ethics or the harbinger of a golden age for research.
But as a web scientist with some public visibility, I feel compelled to enter into the fray. Especially since I just gave a keynote about how web science has brought forth a new age of experimentation.
I haven’t read all of the reactions, but what I have read suggests that many people are missing the point. The only person I see asking the right question is danah boyd, and it’s this: “what was the intended positive contribution of this study?”
Experimentation is Good. Mostly.
Let’s be clear: everyone is experimenting on you. We are in a golden age of randomized controlled experimentation (aka A/B testing), and that’s a good thing. People and organizations are making decisions based on data, rather than blindly following the Highest Paid Person’s Opinion (HIPPO).
So if experimentation is good, what’s wrong with what Facebook did?
Was it a lack of informed consent on the part of the users subjected to the experiment? That’s certainly an issue. Even if users signed a blanket consent form as part of their membership agreement, this kind of experiment is probably not what they had in mind.
The Kahneman Colonoscopy Experiment
But I think there’s a better explanation. As context, I’d like to talk about one of the most important psychology experiments of our generation. And yes, it involved colonoscopies.
In 2003, Nobel laureate Daniel Kahneman and colleagues conducted an experiment to assess how colonoscopy patients experienced and remembered the pain associated with the procedure. The control group underwent the usual procedure. But the experimental group had a slightly longer procedure: the colonoscope wasn’t removed for an additional three minutes.
Let’s put the concerns with invasive experimentation in perspective: this experiment was literally invasive. The experimenters stuck tubes up the subjects’ butts. And I’m skeptical that any of the patients realized they were signing up for an unnecessarily longer colonoscopy.
The result of the experiment: patients who underwent the extended procedure rated the entire experience as less unpleasant than the normal one and showed higher rates of returning for a repeat colonoscopy.
Cui Bono?
So why do we condemn Facebook’s experiment on mildly manipulating people’s emotions and celebrate Kahneman’s experiment of causing pain by sticking tubes up people’s butts?
Because of cui bono, Latin for “to whose benefit?”. It’s not clear how Facebook’s users benefit from the experiment in manipulating their emotions. As lead author Adam Kramer conceded, the authors didn’t clearly state our motivations in the paper.
In contract, Kahneman and colleagues achieved a breakthrough not only in understanding the psychology of how we remember pain, but also in how to improve patients’ willingness to undergo future unpleasant medical procedures. Patients benefit, society benefits, science benefits.
Lessons for the Rest of Us
I feel for the data scientists at Facebook, who may feel they’re only being singled out because they had the temerity to publish what everyone else is doing without talking about it. I certainly hope people don’t learn from this experience that it’s better to not talk about experimentation. Duncan Watts is right that we need more and better science — especially in the social sciences.
But ethics compel us to perform experiments that benefit the subjects of those experiments. At LinkedIn, we have a core value of putting members first. It’s a value that informs all the work we do, and I hope we’re setting an example that other web scientists can learn from.
Head of Query Understanding at LinkedIn
