The Washington Post - USA (2022-05-15)

G4 EZ EE THE WASHINGTON POST.SUNDAY, MAY 15 , 2022

not to mention we the users — can’t see inside their black box for ourselves. “We the users deserve transparency,” says Haugen. “We deserve to have the same level of nutritional labeling for our infor- mational products as we have for our nutritional products. We deserve to see what goes into the algorithms. We deserve to see what the consequences of those things are. And right now, we’re forced to just trust Facebook.” It’s more than an academic is- sue. “Platforms get to experi- ment on their users all the time without letting them know ex- periments are going on,” says Laura Edelson, a researcher at New York University whose Face- book account was cut off by the company for studying political advertisements and misinformation. “Consumers deserve no- tice.” In the U.S., at least five bills have been introduced in Con- gress that focus on responsibility for algorithms. One of them is the bipartisan Platform Account- ability and Transparency Act (PATA), which would force companies to open up their algorithms by turning over information about how they work — and their consequences — to researchers and the public. “We agree people should have control over what they see on our apps and we’ll continue working on new ways to make them more transparent, while also supporting regulation that sets clear standards for our industry in this area,” said Otway, the spokeswoman for Instagram. Now it’s time to hold them to it.

dations. Instead, they should focus on the signals where we ex- plicitly say we’re interested, such as pressing like or following an account. Apps also need to provide us better ways to give negative feedback on their algorithmic choic- es. Right now it’s too hard to tell Instagram or Facebook you don’t want something. It could move a “no thank you” button out from behind the menu screen, and to right next to the Like button. Instagram is starting down this path. It tells me it is at the early stages of exploring a control that would allow people to select keywords to filter from their recommendations. To use their example, if you asked that the word “bread” be removed from your recommendations, In- stagram wouldn’t show posts containing the word “bread.” I’m also intrigued by a bolder idea: Let us choose between competing algorithms to order the information on our feeds. Al- gorithms can be programmed to show or bury content. Some people might want to see Donald Trump, while others might want feeds that are completely politics-free. It could work kind of like the app store on your phone. Different algorithm developers could compete to organize your Instagram, Facebook or Twitter feed, and you settle on the one you like the best. Or maybe you switch from time to time, de- pending on your mood. These are all product fixes, but bigger solutions have to address another problem: We actually know very little about how these algorithms work. Right now, researchers and governments —

information.
On my son’s account, I wit-
nessed another unintended con-
sequence: what Haugen calls
“engagement hackers.” They’re a
kind of spammer who has
learned how to hijack Insta-
gram’s logic, which encourages
them to post shocking images to
elicit reactions from viewers and
thus build their credibility with
the algorithm.
Several of the accounts behind
the images Instagram recom-
mended to my son’s appear not
to be parents of the children fea-
tured in the images. One image
I’ve seen repeatedly, of a baby
with what appear to be severe lip
blisters, was shared by accounts
called kids_past (with 117,000
followers) and another called
cutes.babiesz (with 32,000 fol-
lowers). The captions on the
photos don’t make sense with
image, and don’t appear to be re-
lated to the other children fea-
tured on the account. Both also
suggest in their biographies that
they’re available for paid promo-
tions. Neither account replied to
messages asking where it had
gotten the blister image.
Instagram doesn’t completely
throw caution to the wind. It has
community standards for con-
tent, including guidelines on
what kinds of themes can be in-
cluded in posts that its algo-
rithms recommend. It says con-
tent that’s either “clickbait” or
“engagement bait” is not al-
lowed. In April the company an-
nounced a new effort to down-
rank content that is not “origi-
nal.”
Haugen says Facebook doesn’t

FOWLER FROM G3

that category. An even better idea: Give us an algorithmic reset button. I understand many people really enjoy social media recommendations, especially on TikTok. So give us the power to clear what the algorithm thinks about us without deleting the whole account and losing our friends, just like you can clear your history and cookies in a Web browser. To give users more control, apps could also stop using un- conscious actions — like dwell time while you’re doomscrolling — as signals to feed recommen-

Instagram declined to let me speak with Mosseri for this col- umn. Let’s hope he’s open to feedback. Here’s a start: Let us just turn off algorithms. In March, Insta- gram announced it would bring back a version of its main feed that sorts posts in reverse chron- ological order. That’s good. But to completely shut off Insta- gram’s recommended posts from accounts you don’t follow — and make at least your main feed a friends-only experience — you have to select the Favorites-only view, and put all your friends in

have leadership that can ask hard questions about its impact — and accept hard answers. “When you acknowledge power, you also then acknowledge responsibility. And Facebook doesn’t want to allocate any more time to features that don’t cause it to grow.”

How to make algorithms accountable So how can we the users take back power over algorithms? From researchers and lawmakers alike, there’s a growing collec- tion of good ideas.

JABIN BOTSFORD/THE WASHINGTON POST Facebook whistleblower Frances Haugen, seen during President Biden’s State of the Union address this year, says that people deserve transparency about the algorithms in our social media platforms.

ballots are fraudulent — the first action by a technology company to punish Trump for spreading misinformation. Days later, the company acted again, covering up a Trump tweet about protests over the death of George Floyd that warned “when the looting starts, the shooting starts.” More such actions followed. Later that year, Gadde was in- volved in a decision that drew widespread criticism. In October 2020, the New York Post published an exclusive story based on material found on a laptop allegedly belonging to Biden’s son Hunter. Gadde and other trust and safety executives suspected the story was based on material obtained through hacking and therefore vi- olated the company’s rules against publishing such material. Anxious to avoid a repeat of Russia leaking hacked material during the 2016 election, Twitter executives took the unusual step of temporarily locking the news- paper’s Twitter account and block- ing Twitter users from sharing a link to the story. Even within liberal Twitter, the decision was controversial, two of the people said. It was not entirely clear the materials had been hacked, nor that the New York Post had participated in any hacking. A Post investigation later con- firmed that thousands of emails taken from the laptop were au- thentic. Amid mounting outrage among conservatives, Gadde conferred with Dorsey and announced an 11th-hour change to the hacked- m aterials policy: The company would remove only content posted by the hackers themselves or others acting in concert with them. It also would label more question- able tweets. Dorsey later tweeted that the decision to block mention of the New York Post story was a mistake. Recently, Musk tweeted that “suspending the Twitter account of a major news organization for publishing a truthful story was obviously incredibly inappropri- ate.” Now employees are worried Musk will undo much of the trust and safety team’s work. Many people silenced by policies adopted under Gadde are clamoring for Musk to avenge them. Johnson, for example, said he has appealed via text to Jared Birchall, head of Musk’s family office, asking when his account might be restored. Birchall did not immediately respond to a request for comment. Though Johnson does not plan to tweet, he said, he wants his account back on principle. Ac- cording to text messages first reported by the Wall Street Journal and subsequently viewed by The Post, Birchall replied: “Hopefully soon.” Birchall also shed light on one of the biggest questions looming over the Musk takeover: Will Musk undo Gadde’s decision to ban Trump? At a recent TED con- ference, Musk said he supports temporary bans over permanent ones. Musk “vehemently disagrees with censoring,” Birchall texted to Johnson. “Especially for a sitting president. Insane.”

A year later, nearly every other major platform banned Jones. Twitter initially declined to do so, saying Jones hadn’t broken any of its rules. Within a month, however, Gadde reversed course, banish- ing Jones for “abusive behavior.” In a 2019 appearance on the “Joe Rogan Experience” podcast, Gad- de explained that Jones had earned “three strikes” by posting videos that did violate Twitter’s rules, including one she deemed an incitement to violence against the news media. Jones did not respond to a request for comment. At the time, he called Infowars “a rallying cry for free speech in America,” adding that he was “very honored to be under attack.” Gadde and her team later esca- lated the company’s efforts to fight disinformation — along with spam and fake accounts — after news broke that Twitter, Facebook and other platforms had been ex- ploited by Russian operatives during the 2016 campaign. The company began removing a million accounts a day in a broad effort to crack down on abuse. In a move described as signa- ture Gadde, Twitter also launched an initiative called “Healthy Con- versations” that sought feedback from hundreds of experts about how to foster more civil dialogue. That effort led to updated hate speech policies that banned “de- humanizing speech” — such as racial slurs and negative stereo- types based on religion, caste or sexual orientation — because it could have the effect of “normaliz- ing serious violence,” according to a company blog post. In subsequent years, Dorsey be- came increasingly absent and would effectively outsource a growing number of decisions to Gadde, including those around content moderation, three of the people said. Gadde also was key to a 2019 decision to ban political advertis- ing on the platform, according to four people familiar with the decision, arguing that politicians should reach broad audiences on the merits of their statements rather than by paying for them. Other companies copied the move, enacting temporary pauses during the 2020 election. Throughout Trump’s presiden- cy, at the company’s monthly town halls, Twitter employees regularly called on Gadde to ban Trump, accusing him of bullying and pro- moting misinformation. Gadde argued that the public had a right to hear what public figures such as Trump have to say — especially when they say horrible things, the people said. Meanwhile, Gadde and her team were working with engi- neers to develop a warning label to cover up tweets — even from world leaders such as Trump — if they broke the company’s rules. Users would see the tweet only if they chose to click on it. They saw it as a middle ground between banning accounts and removing content and leaving it up. In May 2020, as Trump’s reelec- tion campaign got underway, Twitter decided to slap a fact- checking label on a Trump tweet that falsely claimed that mail-in

to his 2016 presidential campaign. With nearly 90 million followers at his peak, Trump routinely lobbed tweets at political oppo- nents, journalists and even private citizens, triggering waves of online harassment. After Trump’s election, Gadde and Dorsey convened a “free speech roundtable” at the company’s San Francisco headquarters, where top Twitter executives heard from Citron, former New York Times editor Bill Keller and Tom Goldstein, former dean of the graduate journalism school at University of California at Berke- ley. During the meeting, which has not been previously reported, Cit- ron expressed concerns about online harassment, especially direct- ed at journalists.

Gadde “understood how speech could silence speech,” Citron re- called, “and could be incredibly damaging to people’s lives.” Goldstein declined to comment on the meeting. Keller said the group discussed how new standards could bring order to the “wild west” of social media. Internally, some employees faulted Gadde for ineffectiveness, as rules were unevenly applied across the massive platform. Three former workers said her trust and safety unit did not coor- dinate well with other teams that also policed the site. Even as the company took action to limit hate speech and harassment, Gadde resisted calls to police mere misinformation and falsehoods — including by the new president. “As much as we and many of the individuals might have deeply held beliefs about what is true and what is factual and what’s appro- priate, we felt that we should not as a company be in the position of verifying truth,” Gadde said on a 2018 Slate podcast, responding to a question about right-wing media host Alex Jones, who had pro- moted the falsehood on his show, Infowars, that the Sandy Hook school shooting was staged.

for the company, said software engineer Brianna Wu, one of the women targeted in GamerGate, who worked with Twitter to im- prove the site. In an op-ed published in The Post, Gadde wrote that she was “seriously troubled by the plight of some of our users who are completely overwhelmed by those who are trying to silence healthy discourse in the name of free expression.” By then, Gadde had been pro- moted to general counsel, oversee- ing all legal and trust and safety matters facing the company. In response to GamerGate, Twitter streamlined the company’s complicated nine-step proc- ess for reporting abuse and tripled the number of people on its trust

and safety team, as well as other teams that protect users, according to the op-ed and other reports at the time. But the moves to clamp down on harassment soon stirred fresh controversy. Internal emails obtained by BuzzFeed in 2017 showed Gadde and other executives engaged in messy, seemingly ad hoc deliberations over whether to shut down the accounts of alt- right provocateur Milo Yiannopo- ulos and right-wing flamethrower Chuck C. Johnson, who had tweeted that he was raising money in the hopes of “taking out” a leader of the Black Lives Matter move- ment. Johnson, who says his comment was part of a “journalistic project,” has complained that Twitter never offered a clear rea- son for the ban. He sued the company over it and lost. He has since abandoned his alliance with Trump and declared his support for President Biden, he said, lead- ing to attacks online. Because his Twitter account is still suspended, Johnson argues he is unable to defend himself. About the same time, Twitter was confronted with another co- nundrum: the candidacy of Trump, who made Twitter central

crossing over into a place where you’re preventing someone else from using their voice.” Gadde is a previous donor to Kamala D. Harris and other Dem- ocrats, and in 2017 she helped lead Twitter’s $1.59 million donation to the ACLU to fight Trump’s execu- tive order banning immigration from majority Muslim countries. Among employees, Gadde is known for taking a legalistic yet pragmatic approach to content moderation. As with Trump after the Jan. 6 insurrection, she often has argued against limiting speech and has rejected colleagues who wanted to take a stronger approach to removing content, moving to do so only after careful consideration. For years, she has been the ani- mating force pushing Twitter to champion free expression abroad. In India and Turkey, for example, her team has resisted demands to remove content critical of repres- sive governments. In 2014, Gadde made Twitter the only Silicon Val- ley company to sue the U.S. gov- ernment over gag orders on what tech companies could say publicly about federal requests for user data related to national security. (Five other companies settled.) “She wasn’t a censorship warrior or a free expression warrior,” said a former colleague familiar with Gadde’s approach. “She is pragmatic but not doctrinaire.” A dedication to free speech has been part of Twitter’s DNA since its founding in San Francisco 16 years ago. Early executives were such believers that they famously referred to Twitter as “the free speech wing of the free speech party.” That approach made Twit- ter ripe for abuse in its early days, and the platform developed a rep- utation as unsafe — particularly for high-profile women, who en- dured threats of rape and other sexist attacks. Back then, Twitter’s attitude was “we don’t touch speech,” said University of Virginia law profes- sor Danielle Citron, an expert on online harassment. In 2009, Cit- ron prepared a three-page, single- spaced memo for the Twitter C -suite, explaining the legal defi- nition of criminal harassment, true threats and stalking. Gadde joined Twitter’s legal team two years later, leaving her post at the Silicon Valley firm Wil- son, Sonsini, Goodrich and Rosati. People who worked with her said her move was inspired by the Arab Spring uprising, when pro- d emocracy activists used Twitter and other social platforms to organize protests across the Middle East. The Arab Spring solidified the belief among Twitter’s leaders that their job was to protect speech, not police it. Twitter was soon engulfed in scandal, however. In 2014, online trolls launched a brutal campaign against women in the video game industry. The attacks — which came to be known as “GamerGate” — were carried out on multiple tech platforms. But they were most visible on Twitter, where women received highly graphic threats of violence, some including the woman’s address or an exact time of attack. The incident was a wake-up call

“Twitter’s left wing bias.” Musk’s
legions of followers have tweeted
calls for her firing, some of them
racist. (Gadde, 47, is Indian Ameri-
can.)
Musk on Tuesday signaled he
would undo the permanent ban
on Trump if he completes the ac-
quisition of Twitter, potentially
undoing years of work from Gad-
de and her team. “I think that was
a mistake,” he said at an event
hosted by the Financial Times.
Twitter colleagues describe
Gadde’s work as difficult but nec-
essary and unmotivated by politi-
cal ideology. Defenders say her
team, known as the trust and safe-
ty organization, has worked pain -
stakingly to rein in coronavirus
misinformation, bullying and oth-
er harmful speech on the site,
moves that necessarily limit some
forms of expression. They have
also disproportionately affected
right-leaning accounts.
But Gadde also has tried to bal-
ance the desire to protect users
with the values of a company built
on the principle of radical free
speech, they say. She pioneered
strategies for flagging harmful
content without removing it,
adopting warning labels and “in-
terstitials,” which cover up tweets
that break Twitter’s rules and give
people control over what content
they see — strategies copied by
Twitter’s much larger rival, Face-
book.
Many researchers and experts
in online harassment say Gadde’s
policies have made Twitter safer
for its roughly 229 million daily
users and say they fear Musk will
dismantle them if the sale goes
through.
“If Musk takes things in the
direction he has been signaling —
which is a rather simplistic view
that more or less anything goes in
the name of free speech — we will
certainly see the platform go back
to square one,” said Rebekah
Tromble, director of the Institute
for Data, Democracy and Politics
at George Washington University.
Whatever happens to her pol-
icies, Gadde signaled at a staff
meeting late last month that her
days at Twitter may be numbered,
telling employees that she would
work to protect their jobs as long
as she is around, according to a
person who attended the meeting.
She did not respond to requests
for comment. Twitter declined to
comment. Musk did not respond
to a request for comment.
On Monday, Musk tweeted:
“Twitter obv has a strong left wing
bias.”
This story is based on inter-
views with 10 current and former
Twitter employees, as well as oth-
ers familiar with decisions made
by Gadde and her team, who spoke
on the condition of anonymity to
describe private company discus-
sions.
“I do believe very strongly —
and our rules are based on this
framework — that free expression
is a fundamental right, that every-
one has a voice and they should be
able to use it,” said Gadde in a 2019
interview with The Washington
Post. “There is a line between do-
ing that and committing what we
call abuse or harassment, and

TWITTER FROM G1

Twitter’s top lawyer has long balanced free speech, safety

MARTINA ALBERTAZZI/BLOOMBERG NEWS Vijaya Gadde, chief legal officer of Twitter, speaks during the 2019 Wall Street Journal Tech Live c onference in Laguna Beach, Calif.

“She wasn’t a censorship warrior or a

free expression warrior. She is pragmatic

but not doctrinaire.” Vijaya Gadde’s former colleague, describing the lawyer’s approach

The Washington Post - USA (2022-05-15)

Get our desktop app

Company

Features

Documentation

Resources