It’s a stunningly blunt and emotion-laden headline: “Facebook’s Secret Censorship Rules Protect White Men From Hate Speech But [Do] Not [Protect] Black Children.” And the full ProPublica article posted online in June 2017 appears beneath an equally blunt subhead: “A trove of internal documents sheds light on the algorithms that Facebook’s censors use to differentiate between hate speech and legitimate political expression.” The discussion extends into the ProPublica Facebook account (irony, anyone?), which includes a series of slides summarizing how “Facebook has used these rules to train its ‘content reviewers’ [aka “censors”] to decide whether to delete or allow posts.” And there is an additional thought-provoking follow-up in an article (“What Does Facebook Consider Hate Speech? Teach Our Quiz”) published in The New York Times this month.
The articles and that post provide a highly-nuanced, very thoughtful examination of the difficulties we face in establishing universally acceptable standards in a world where universal standards appear impossible to establish—and raise questions for at least a few of my colleagues as to whether we should even be attempting to establish those standards.
Let me be blunt: when I read the six statements included in The New York Times article to see how our own conclusions might differ from the conclusions resulting from those Facebook guidelines for its content reviewers, I don’t see a single comment there that I’m comfortable expressing or defending. I’m not going to tell anyone that they can’t say any of those things, but I’m also not going to remain silent face-to-face or online rather than expressing my firm opposition to those words and other thoughts that are so patently and disgustingly uncivil, incendiary, and destructive; terribly hurtful to friends, colleagues, and other members of our extended onsite and online communities; and in opposition to so much of what I hold to be foundational beliefs as to how we should be treating each other. I want us, collectively and collaboratively, to be seeking ways to make America (and our social media environments) a bit more civil again, and I believe that starts with us doing our best to find some acceptable minimum standards to which we can comfortably adhere.
Let’s start with the six “true-false” statements cited by Times staff members Audrey Carlsen and Fahima Haque, including their up-front statement and question (“Most readers will find them offensive. But can you tell which ones would run afoul of Facebook’s rules on hate speech?”) and the same question (“Would this statement meet Facebook’s criteria for hate speech?”) posted after each of the six statements:
- “Why do Indians always smell like curry?!They stink!”
- “Poor black people should still sit at the back of the bus.”
- “White men are assholes.”
- “Keep ‘trans’ men out of girls (sic) bathrooms!”
- “Female sports reporters need to be hit in the head with hockey pucks.”
- “I never trust a Muslim immigrant…they’re all thieves and robbers.”
The final entry on the ProPublica slide deck provides answers purportedly taken from the training Facebook has provided to its content reviewers, and a follow-up article in The Times provides additional information on that topic. It’s not pretty; as ProPublica suggests in its in-depth article, some of those comments make it past the Facebook guidelines, as users of Facebook must know from reading some of what comes into their feeds. But that doesn’t make them defensible, acceptable, or right—at least to many friends, colleagues, and acquaintances, from a variety of political spectrums, with whom I communicate via Facebook and other social media platforms.
I have the same reaction to those six statements that I had last week to some extremely crude and derogatory comments an acquaintance made the mistake of making to me, face to face, in front of my wife and another woman—in a way that suggested he thought he was being clever and funny: I want to—and in this case did—ask him what made him think that what he was saying was acceptable discourse among friends or acquaintances (although my wording was much less civil and much more crude than the paraphrase I’m offering here). Being tone-deaf to the question, he continued in a similar vein for a few more seconds until I explicitly told him—again, in much cruder and far less clever language—that he could take his trash to a different garbage can.
As I carry that thought back into that ProPublica article written by Julia Angwin and Hannes Grassegger, and the responses Facebook representatives provide in that article, I’m not left feeling that the people at Facebook are completely tone-deaf, despicable, or out of touch with the world around them. One important conclusion reached by reading and re-reading that article is that they—and we—are struggling with some very wicked problems here. I also acknowledge the truth behind one of the many thoughtful observations included in the story:
“‘The policies do not always lead to perfect outcomes,’ said Monika Bickert, head of global policy management at Facebook. ‘That is the reality of having policies that apply to a global community where people around the world are going to have very different ideas about what is OK to share.’”
That article is also very good about citing positive steps Facebook employees have taken when they create their own do-not-cross lines (“graphic violence, child abuse, revenge porn and self-mutilation”) and how they have apologized when some of their decisions and actions (including deleting comments and temporarily locking users out of their Facebook accounts).
But what we’re left with is a classic example of a wicked problem: how to establish minimum community standards when significant numbers of people within a community are far from being in agreement. Which, of course, doesn’t mean we shouldn’t try.
N.B. — Paul is currently writing Change the World Using Social Media, scheduled for publication by Rowman & Littlefield in 2020. This is the first in a series of excerpts from the manuscript in progress.