On Wednesday, an 85-page document leaked to Breitbart News. Titled “The Good Censor,” it’s a presentation that traces the evolution of content moderation on tech platforms to the present day. The document is more descriptive than it is prescriptive, aiming to capture the current debate rather than influence Google’s leadership directly. Still, in speaking frankly about Google’s role as a censor, the document — which you can read in its entirety here — could fuel new calls to rein in platforms’ power.
First, some context. The presentation was put together in March of this year by something called Brand Studio, which describes itself as “Google’s internal think tank that uses creativity, media, and technology to create experiences that connect Google products to the people who use them.” It also has a team that develops programs for “crisis response and sustainability.” Generally speaking, Brand Studio talks to experts and puts together white papers and does various marketing stunts around them.
In other words, the presentation that leaked to Breitbart News is not a memo from the head of Google search, or the CEO of YouTube. Still, it’s worth taking look, primarily for the way it frames the debate around content moderation for companies like Google, Facebook, and Twitter.
The presentation is divided into five sections: describing the importance of free speech; outlining the dimensions of bad behavior and censorship online; examining how companies are responding; looking at the balance between free speech and censorship around the world; and finally, asking how people want Google to respond.
Everywhere the authors look, they find people “behaving badly,” to use their chosen expression. Individuals harass and abuse other users. Governments employ bots, troll farms, and hackers to wage influence campaigns. Tech companies promote fake news, underinvest in moderation, and profit from the spread of misinformation.
”Shares, likes, and clickbait headlines — monetized online conversations aren’t great news for rational debate,” the authors write. “And when tech firms have an eye on their shareholders as well as their free-speech and censorship values, the priorities can get a little muddled.” They add: “In responding to public pressure, tech firms haven’t managed the situation particularly well, either.”
What have tech companies mismanaged? According to the report: inconsistent application of content moderation guidelines; opaque explanations around their policies; underplaying the scope of the problem; slow response times; and a reactionary posture that can seem more attuned to public perception than addressing root-level problems.
So what should Google do about it?
”The answer is not to ‘find the right amount of censorship’ and stick to it,” the authors write. Instead: “Google might continue to shift with the times — changing its stance on how much or how little it censors (due to public, governmental, or commercial pressures). If it does, acknowledgement of what this shift in position means for users and or Google is essential. Shifting blindly or silently in one direction or another right incites users’ fury.”
Other recommendations from the report: remain neutral; “police tone instead of content”; clearly enforce policies; offer justifications for global policies around censorship; explain the underlying technology that the platforms run on; and do a better job talking about all of it.
The company’s official position on content moderation remains political neutrality, a spokeswoman told me in an email:
Google is committed to free expression — supporting the free flow of ideas is core to our mission. Where we have developed our own content policies, we enforce them in a politically neutral way. Giving preference to content of one political ideology over another would fundamentally conflict with our goal of providing services that work for everyone.
Of course, it’s impossible to read the report or Google’s statement without considering Project Dragonfly. According to Ryan Gallagher’s ongoing reporting at The Intercept, Google’s planned Chinese search engine will enable anything but the free flow of ideas. Even in an environment where American users are calling for tech platforms to limit users’ freedoms in exchange for more safety and security, many still recoil at the idea of a search engine that bans search terms in support of an authoritarian regime.
And that’s the unresolvable tension at the heart of this report. Almost all of us would agree that some restrictions on free speech are necessary. But few of us would agree on what those restrictions should be. Being a good censor — or at least, a more consistent censor — is within Google’s grasp. But being a politically neutral one is probably impossible.
Sen. Richard Blumenthal (D-CT) sent a letter to the Federal Trade Commission calling for an investigation into the data exposure that resulted in Google+ being shut down.
Richard Pinedo, 28, made up to $95,000 selling stolen bank accounts to Russians, which were then used to buy internet ads during the 2016 election. Today he received six months in prison and six months of home confinement after pleading guilty to a felony identity fraud charge.
Alexandra Stevenson checks in with Rappler, a Filipino news outlet, which is Facebook’s fact checker in the Philippines:
“It’s frustrating,” said Marguerite de Leon, 32, a Rappler employee who receives dozens of tips each day about false stories from readers. “We’re cleaning up Facebook’s mess.”
On the front lines in the war over misinformation, Rappler is overmatched and outgunned — and that could be a worrying indicator of Facebook’s effort to curb the global problem by tapping fact-checking organizations around the world. Civil society groups have complained that Facebook’s support is weak. Others have said the company doesn’t offer enough transparency to tell what works and what doesn’t.
Twitter signed a deal with local broadcasters to stream seven debates that may have national interest:
The seven debates that will be livestreamed throughout October include the debate between Sen. Cruz and Rep. O’Rourke and the Oregon governor’s race. Zuckerman said Twitter is actively looking to add more to the list.
Kurt Wagner looks at candidates to take over Elliot Schrage’s job at Facebook. It’s going to be a difficult job: the communications team has been extremely stressed out lately. One person told me that a recent comms meeting ended in tears:
“I can’t think of another company that’s facing the [challenges] they have, or even anticipated facing those [challenges],” said Brandee Barker, co-founder of The Pramana Collective, a marketing and communication firm. Barker was head of global communication and public policy at Facebook from 2006 to 2010, when the biggest issue facing the company was explaining its technology. When Facebook launched News Feed, angry users protested outside the company’s office.
“It’s clear that from when I was there the role has evolved so significantly,” Barker said. Noting that the job has two elements, she thinks it’s predominantly a policy role right now. “They have challenges now at the governmental level internationally, in the U.S., in the EU, and it will only continue to increase.”
A day after the company promised it would address this issue by somehow applying artificial intelligence to photographs, Taylor Lorenz shines a light on Instagram bullying:
According to a recent Pew survey, 59 percent of teens have been bullied online, and according to a 2017 survey conducted by Ditch the Label, a nonprofit anti-bullying group, more than one in five 12-to-20-year-olds experience bullying specifically on Instagram. “Instagram is a good place sometimes,” said Riley, a 14-year-old who, like most kids in this story, asked to be referred to by her first name only, “but there’s a lot of drama, bullying, and gossip to go along with it.”
Teenagers have always been cruel to one another. But Instagram provides a uniquely powerful set of tools to do so. The velocity and size of the distribution mechanism allow rude comments or harassing images to go viral within hours. Like Twitter, Instagram makes it easy to set up new, anonymous profiles, which can be used specifically for trolling. Most importantly, many interactions on the app are hidden from the watchful eyes of parents and teachers, many of whom don’t understand the platform’s intricacies.
Gretchen McCulloch has a paper that gives a name to that thing where people refer to the president as “Cheeto” or whatever on social media sites:
A recent paper by researcher Emily van der Nagel puts a name to this phenomenon of hiding a word in plain sight. She calls it Voldemorting. Van der Nagel traces Voldemorting back to the Harry Potter books, where most characters are too afraid of Voldemort to say the word directly, instead replacing his name with euphemisms like You Know Who and He Who Must Not Be Named. This practice starts as a superstition, but by the final book there’s a deeper purpose: The word Voldemort is revealed as a way of locating the resistance: “Using his name breaks protective enchantments, it causes some kind of magical disturbance.”
The internet practice of Voldemorting, van der Nagel says, comes via a comment left by a user named Eugene, who made the connection as part of a discussion about deliberately starving “trash celebrities” of attention by not referring to them by name.
A leading explanation for the decline of drinking in Britain is the rise of social media, Iliana Magra reports:
Social media has made users more image-conscious, he noted, while also providing lasting documentation, in text and images, of behavior people might prefer to forget.
“There’s a trend of greater sense of health consciousness among young people,” Mr. Nicholls added. “There’s a move away from alcohol and drugs, there’s less of a culture of intoxication.”
Bijan Stephen checks in with the Google+ diehards:
“It might not be an amazing site but IT DAMN sure feels like it. So why should I move?” asked user Buruburedo Boudreaux.
I traveled to Santa Monica to interview Snap’s head of original content, Sean Mills, as Snapchat tries to reinvent MTV for a younger generation:
In an interview at the company’s offices in Santa Monica, California, Mills said Snap’s original shows are engineered to succeed in the difficult environment of mobile video where viewers are never more than a thumb-tap away from abandoning a show. Snap Originals are designed to hook viewers within seconds and keep them stimulated with flashy visuals, he said.
“I feel like I’m watching the beginning of a fundamentally new medium, where people are just waking up to how you have to take a very different creative approach,” Mills said.
Left out of the competition among social companies to announce that they had put artificial intelligence into things, LinkedIn entered the arena today to announce that its AI would go to work on diversity issues. Honestly none of this sounds like AI to me:
LinkedIn will track what happens in the hiring process with regards to gender, showing companies reports and insights about how their job postings and InMail are performing on this. In addition, LinkedIn will re-rank the top search results in LinkedIn Recruiter to be more representative.
The Magic Leap One Creator Edition mixed reality headset is now shipping across the contiguous United States, at a low, low price of $2,295. It will look great next to your Juicero!
And finally …
After the pop singer broke her legendary silence about politics, her fans are registering to vote en masse:
“We are up to 65,000 registrations in a single 24-hour period since T. Swift’s post,” said Kamari Guthrie, director of communications for Vote.org.
For context, 190,178 new voters were registered nationwide in the entire month of September, while 56,669 were registered in August.
Taylor Swift is good again, and I encourage everyone to resume playing “Shake It Off” accordingly.
Talk to me
Send me tips, comments, questions, and secret Google presentations: [email protected].