What a week it has been, right? At least for those of us who follow and/or use social media. There was the naming of whistleblower Frances Haugen on “60 Minutes” Sunday night, US time; the hours-long outage of all of Facebook’s products Monday; Haugen’s testimony on Capitol Hill Tuesday, a report that hackers were offering for sale 1.5 billion people’s public data they scraped from Facebook; and the reportedly massive Twitch breach. Am I missing anything?
As for the Facebook-related news, if it interested you, I’m sure you’ve seen plenty of the very large amount of coverage all over the world. So I’ll cut right to some questions I have (Einstein did say it’s better to “listen to the person who has the questions” not the answers), then propose two major solutions.
- Is it possible that policymakers would base policy on the views of one person, a data scientist who, at least in congressional testimony, doesn’t distinguish between focus groups and peer-reviewed research – or on (leaked) product research that wasn’t peer-reviewed and didn’t use a nationally representative sample (of teen Instagram users)?
- In developing regulation, will policymakers factor in what academic research has found about social media’s positive effects as well as negative ones (see, for example, research by Mizuko Ito, et al, Candice Odgers and Victoria Rideout, whose latest study found that 43% of teens said using social media made them feel better when struggling mentally while 17% made them feel worse)?
- Is there a risk now that platforms will choose not to find out the negative impacts of their products on vulnerable groups? Wouldn’t it be better if they did their (non-scientific) product research on both positive and negative effects and published those findings as well as what they’re doing about the negative effects?
- Self-critical social comparison and body image struggles have long been part of adolescence (and being human) – how do Instagram and social media in general contribute to or increase this social problem?
- Since research shows that social media use is highly contextual, will policymakers factor in research on how it fits into young people’s lives and how they use it, not just on its impacts, positive or negative? (Clearly, Sen. Richard Blumenthal didn’t know that “finstas” are not an Instagram product but rather a teens’ own adaptation of Instagram for their own purposes.) Will lawmakers learn about how much kids play, communicate, create and hang out vs just consume their media?
- How is this a Big Tobacco moment – or not? Are social media users consumers the way cigarette smokers are? And, yes, tobacco use has social elements, but wasn’t the focus of policymakers on physical effects, and aren’t we now in a very different digital world, where our data is the source of revenue, and we’re entrusting it to the corporations using it?
- What actually needs to change or be regulated – engagement and recommendation algorithms? What companies do with user data? How product research is done? Corporate transparency levels? The balance of human and algorithmic content moderation and levels tied to size of user base? How much content moderation users of different languages and cultures in distant countries receive?
- Are lawmakers willing to consider new models for regulation being applied in other countries?
- In terms of how people use Facebook all over the world, has the company outgrown its 20th-century ad-based business model?
Proposing 2 solutions
Much of the discussion of the past week has focused on regulation. So I’ll propose two solutions that haven’t gotten a lot of discussion and deserve more. I’ll list them first, then explain. Facebook should…
- Acknowledge that it’s no longer just a corporation and structure itself accordingly.
- Build out a network of help services for Internet users around the world.
In a commentary unusually focused on solutions, law professor Kate Klonick called them “opportunities for reform” and focused on solutions internal to Facebook and social media companies – changes such as greater transparency, rethinking the “user engagement” metric and adding more content moderators. Few people would argue with those. The problem is, they’re not enough.
Content moderation AND user help
Klonick wrote, “To put this all in perspective, in the United States there is roughly one law enforcement officer for every 500 people. Facebook has 2.8 billion global monthly active users; that means just 1.3 people working in safety and security for every 100,000 users.” What she doesn’t say is that police officers have or can get context for problems they’re called in to address. Content moderators can’t. Neither can algorithms.
Neither adding more human moderators nor tweaking algorithms is likely ever to catch up with harmful content and behavior in social media. That’s because what happens on platforms is highly contextual to ever-changing offline life, speech, behavior and social norms. In terms of speech, moderators who see only what’s happening on a platform online will never have enough context for that comment to make fail-proof moderation decisions – is it parody, sarcasm, “just a joke,” a cruel joke, etc., etc.? Algorithms for deleting content that violates community rules have to be “fed” tons of data to make good moderation “decisions,” and the offline-world speech and behavior reflected in that data is nuanced and keeps changing; so it’s unlikely the algorithm can ever catch up (all this besides the privacy issues related to providing it with all that data). And there are different norms and definitions all over the world to be factored in. The exceptions, such as child sexual abuse material, are much easier to moderate because almost every country in the world agrees it is criminal activity.
Context is the missing piece in the content moderation – or, better, user care – mix. Internet helplines provide the platforms with the context they need to delete highly contextual harmful context such as harassment, hate speech and bullying. And that’s only an extremely beneficial side effect of the care and support helplines provide users. There’s a whole network of Internet helplines in Europe doing this, one that the European Commission helped set up well over a decade ago. There’s also user care provided by the eSafety Commissioner’s Office in Australia and by NetSafe in New Zealand. Ideally, every country should have one, including the US. An Internet helpline should be independent of both industry and government in the US but could be partially funded by both (and individuals), as is the National Center for Missing & Exploited Children. Each country’s helpline needs to be structured and funded as appropriate for its own context.
Finally, picking up on my last question, I suggest Facebook needs to organize itself differently, in keeping with the role it has come to play. Because of its penetration into the everyday lives of people all over the world, their dependence on it as individuals and businesses, how much of their data it handles, the infrastructure it provides in many countries and so much more, Facebook has – in practice, on the ground – evolved away from being merely a corporation. Yet it’s still identifying and acting as a publicly traded company. What it has become is historically unprecedented – part company, part utility, part social institution – so it should neither be structured nor regulated based on existing models. I believe this misalignment is a big reason for growing cognitive dissonance, if not outrage, out in the public discussion. Facebook’s business is very personal, to many people in many ways (including Frances Haugen, who said she lost a friendship to misinformation). There is nothing unethical about profit, an ad-based business or publicly traded corporations. Under that model, safety is a cost center, not a profit center, and Facebook is organized to maximize user growth, tech innovation and advertising in order to maximize profit for its shareholders.
So I think it’s only logical to say that, in order to have credibility when it tells the public and policymakers it does not prioritize profit over people – Facebook needs to see itself and act as a social institution as much as a corporation and organize itself accordingly. Direct knowledge, of its impacts on vulnerable people in every culture and political system where it has a presence, needs to be folded into product development, acquisitions, every management decision. The company needs a chief safety officer in the “C suite” – an office that has real power, doesn’t fall under marketing or lobbying, supports the kind of investigative work that journalists do and has sufficient budget to contribute to helpline operations around the world.
And that’s a wrap on only the first week of a whistleblower’s work. It feels like the ground is shifting – let’s see how much.
- Prof. Kate Klonick’s October 1 commentary in the New York Times For background, Klonick “spent months embedded at Facebook” observing the company’s creation of its (now spun off) Oversight Board. Her account is in the Yale Law Journal.
- “Facebook’s own data is not as conclusive as you think about teens and mental health,” by Anya Kamenetz at NPR
- US regulation does appear more likely in the US, now, because of unusual bi-partisan agreement on Capitol Hill (in The Guardian)
- A thoughtful piece on “Parenting after the Facebook Files”
- “Could platforms design for second chances?”
- My 2020 Medium piece offering more detail on helplines and the middle layer they represent between people and helpers on the ground (in the US, 9-1-1 and specialized hotlines for people in crisis) and content moderators in the cloud – and my related post here with thoughts on regulation
Disclosure: I serve on the Trust & Safety advisories of Facebook, Snapchat, Twitter, Yubo and YouTube, and the nonprofit organization I founded and run, The Net Safety Collaborative, has received funding from some of these companies.