How AI could return trust to journalism, healing the rifts in our society
OpenWeb develops tools to foster meaningful conversations in an effort to return the Internet to its original ethos of being an open, welcoming space where people can freely exchange ideas.
It should be perfectly obvious by now to anyone with any common sense that US President Donald Trump stands for peace and prosperity, at home and abroad. Anyone who doesn’t think so clearly hates America, and by extension, Israel.
No, wait, strike that.
It should be perfectly obvious by now to anyone with any common sense that US President Donald Trump is a threat to democracy and decency, at home and abroad. Anyone who doesn’t think so clearly hates America, and by extension, Israel.
Hang on, this is ridiculous. Do we really have to be this polarized? For New York and Tel Aviv-based OpenWeb, the answer is: no.
Over the summer, Netflix released a documentary titled The Social Dilemma, detailing how the social media giants – primarily Facebook and Twitter – in their mission to maximize profits have been fueling division within our societies. Their algorithms are set to serve up content that makes people emotional, and all too often that emotion is fury. Hate. Anger. “Something must be done!” we rage as we hit the share button, perpetuating the cycle. “People need to know!” But with that anger starting to spill over into our streets, does it really have to be this way?
“I think what The Social Dilemma taught us is that these companies in a way are held hostage by their business model. They are in this loop of figuring out what makes people stay longer and be more emotional, but that doesn’t necessarily always go hand in hand with fostering great conversations,” Ido Goldberg, vice president of product at OpenWeb said.
“It’s not just the technology [i.e. the platforms]. Obviously the technology allows people to use this amazing invention of the Internet to communicate with folks all over the place all the time.
“Technology does what we want it to do, and Facebook wanted a very simple thing: they wanted people to stay longer, and so their algorithms amplified the loudest and more violent voices because that’s what brings more time spent.
“We definitely think that technology can be used in a different way. It’s definitely a decision.”
Advertisement
Since 2012, OpenWeb have been making that positive choice. They develop tools that foster meaningful conversations through online moderation in an effort to return the Internet to its original ethos of being an open, welcoming space where people can freely exchange ideas.
But more recently, OpenWeb decided to go further. Rather than simply moderating retroactively, could they use technology to nudge people proactively to moderate their own comments, making these conversations more civil overall?
The concept is simple. When someone wanting to comment on an article hits “post,” OpenWeb uses Perspective API (a coding package) by their partner Jigsaw to scan it for problematic content. If it finds it, OpenWeb’s new Real-time Feedback feature pops up a message warning the user that their comment contains content that might be moderated, and asks whether they want to review the message. They then have the option to modify their comment, or send it anyway. Commenters have only one chance to edit to stop them from gaming the system.
To find out whether this feature could make a real difference to conversations online, in May, the company set up a wide-scale experiment in a study involving nearly half a million comments and 50,000 users across sites including AOL, Salon, Newsweek, RT, and Sky Sports. The platforms were chosen to represent a diverse group of audiences, interests and engagement levels.
Two overall test groups were set up: half of the users received feedback on their comments, and half were allowed to post freely.
They found that 34% of users edited their comment upon being nudged, against 36% who posted it anyway. Some 12% of users simply abandoned their comment, while 18% decided to try and test the system, in some way, by moving their comment to a reply or refreshing the page, for example. Of the 34% of users who chose to edit, 54% changed their comment to be immediately permissible.
Overall, they found that this did have a positive effect on the quality of the conversations taking place on publishers’ platforms.
“The way we operate is not by amplifying the loudest or more violent voices, but rather by amplifying the folks who really want to have a meaningful conversation,” Goldberg said. “And so in that way I think that these tools, algorithms, machine learning technology, whatever it is, can definitely be used to drive something better. It’s just a decision of what we value as a society, as a company and so on.” But hang on, there’s a problem here. At the moment the vast majority of online conversations, including discussion of articles, take place on social media platforms, not on the publishers’ websites. So will this tool really have much of an impact?
“I think platforms are recognizing that they’ll have to change if society forces them to,” Goldberg said.
And then he pitched a game-changer.
“But do we really want to have huge corporations that manage all of our conversations online?” he asked.
The most obvious problem with the social media giants is that which The Social Dilemma set out: that amplifying belligerent voices is fueling division within society. What is less often discussed is the catastrophic effect that channeling all online discussion through a handful of platforms has had on journalism, and by extension, on public discourse.
IN JANUARY 2017, in his first press conference following the 2016 election, then president-elect Donald Trump refused to take a question from CNN’s Jim Acosta. “You are fake news,” Trump boomed, as some in the room clapped.
“Fake news,” “propaganda,” “lamestream media,” these terms are all bandied about with increasing regularity as public trust in the media continues to flag heavily. A June Gallup poll ranked the media last when it came to Americans’ trust in various people and institutions; only the media got an overall negative rating, 10 points behind President Trump himself.
In September, another Gallup poll placed trust in the media at 40%, down significantly from the seven in 10 Americans who trusted the mass media at the time of Nixon’s impeachment – although not as bad as the all-time low of 32% recorded by Gallup in 2016.
But the headline figures, as worrying as they are, hide an even more concerning picture. The September poll, like others before it, showed a stark divergence between Republicans and Democrats on the issue, recording an all-time high in the gap between the two parties of 63 percentage points. That is, whereas trust in the media among Democrats has actually risen to 73% over the last two decades, among Republicans that trust has plummeted to just 10 percent saying the same this year.
It seems, then, that growing distrust of media goes hand in hand with the growing division within society.
The reasons are, of course, complex, but one aspect has to do with the fact that the social media giants have not only invaded the media ecosystem, but decimated it.
Here’s the issue: before the rise of the Internet, people would have their trusted newspaper delivered to their door. Over breakfast they would read a range of views delivered by journalists who had the time to write thoughtful, researched articles, curated by editors who had the budget to pay their staff commensurately.
Because they used to attract large audiences, newspapers used to command large advertising revenues, but social media changed all of that. Now people don’t wake up to a morning paper delivery, but to their Facebook or Twitter apps. In their drive to keep us hooked, what started out as platforms for staying in touch with friends have evolved into platforms for feeding us content we want to consume.
And so we read an article – and then head straight back to Facebook to discuss it with our friends, or to find another to read, and to argue over. The upshot is that Facebook, not the media outlets, now commands the large advertising revenues that come with large audiences. In 2019, Facebook’s revenue totaled a whopping $70.7 billion, up from $55.8 billion in the previous fiscal year. Some 99% of that revenue was from advertising.
Media companies, unable to compete, face problems on two fronts: their budgets are ever shrinking, leading to fewer journalists on the payroll, and the articles must have ever more clickbaity headlines in a desperate bid to pull readers out of the social media miasma and onto their sites.
The outcome is poorer quality journalism, plummeting trust in the media itself, and nations growing increasingly divided.
“What journalists bring to the table is this amazing commitment to truth and to journalistic values,” Goldberg said. “We feel that if we can have conversations distributed across all of these different press outlets, this can give a voice to each group of society. Journalists can really be part of the solution here if they are willing to participate, recognizing their mistakes, maybe recognize that other people have other opinions and so on. We definitely feel that this can positively impact the trust that people have in these outlets.”Clearly, what OpenWeb are proposing goes a lot further than the comments section underneath online articles. What Goldberg is suggesting is a new way of doing things entirely. The vision being put forward combines the old – a plurality of media outlets all offering quality content to loyal audiences – with the benefits of direct communication that the modern era brings, through allowing journalists to interact with their audiences in real time.
“What we want to help media outlets do is really develop an audience that keeps coming back directly to the publisher’s site as a destination. Then hopefully what publications can do is really go back to where they were 20 years ago where the folks who subscribed to your newspaper got it each morning to their doorstep.
“We definitely want to do as much as we can to help publications navigate this crazy world of the Internet and this fight against the tech giants. It’s not an easy fight for media. What media knows best is how to create content, not necessarily how to create experiences that people come back for. We’re trying to do our part with our partnerships, helping them to do that better.” IF ALL of this sounds rather far-fetched and idealistic, there are signs that Facebook’s iron grip on the online ecosystem is starting to slip.
In July, more than 1,000 advertisers officially boycotted the social media platform. Many others quietly drastically cut back their spending in support of the #StopHateForProfit campaign, a joint collaboration between the Anti-Defamation League, Color of Change, Sleeping Giants, the NAACP, Free Press and Common Sense.
Jonathan Greenblatt, CEO of ADL, has warned that the boycott was merely “a warning shot, an opening salvo.”
But not everyone was in favor of the boycott given that it called for certain voices, including some conservative news sites, to be censored by Facebook.
In June, three prominent Republicans, Jim Jordan, Elise Stefanik and Nikki Haley, joined half a million new users on Parler, the new kid on the block in the world of social media, which grew its base by 50% to 1.5 million in just one week.
The move was sparked by the news that the Trump campaign was looking for alternatives to Twitter and Facebook over concerns that they might lose reach ahead of the election. This followed a decision by Facebook in June to remove a Trump ad campaign, citing its policy against hate speech.
John Matze, Parler’s 27-year-old founder and CEO, has emphasized free speech as the key selling point of his platform. “We’re a community town square with no censorship,” he told CNBC. “If you can say it on the streets of New York, you can say it on Parler.” Facebook, then, could ironically end up being the ultimate victim of the division it has fueled, as the two tribes tear the platform apart in the fight over hate speech versus free speech.
Which brings us neatly back to OpenWeb’s experiment on nudging online conversations. Surely one of the factors driving the backlash against social media is this feeling that we are being nudged, that our behavior is being manipulated to suit the people running the platforms. Isn’t this latest nudge toward civility online simply more of the same?
“We’re not saying to people that they have to change the way they behave,” Goldberg said. “We do say that there’s consequences to the way you behave.
“The companies we work with that foster these conversations on their site have their guidelines, and there’s stuff they’re willing to accept and stuff they’re not willing to accept. If you don’t agree with the community guidance of a specific community you can definitely find a different community to speak the way you want.” In effect, the changes OpenWeb are proposing to make represent a maturing of the online public square. The Internet was, in many ways, unleashed upon an unprepared world. While the early, idealistic years of online communication can be seen as a children’s sandbox, as people started to figure out what they could do with this new tool, the last 15 years have been the teenage years as we all try to figure out how to conduct ourselves in this new, loud, oftentimes bewildering space.
It seems then that we are now moving into what could be termed early adulthood, as we all learn that taking personal responsibility for our own conduct online in a respectful manner leads to a better experience all around, for others and for ourselves.
“One other result of this [experiment] is that it teaches us a lot about who a person is. Most people have their hearts in the right place and want to meet people and have a conversation with them. Some people, a small percentage of people really, want to see the world burn.
“This idea of giving folks an opportunity to rethink, when you see someone who systematically refuses to take that opportunity, that says a lot about who that person is and really what they’re looking for.”
Those people wouldn’t automatically be excluded, Goldberg said, rather they’d be disincentivized.
“Personally for us, it’s understanding who should get the reach. We want to be very clear with our users that what we value, and what our publishers value, is really a great conversation so yes, over time we want to figure out a way to amplify the voices that are looking for that type of conversation, and do the opposite for folks that are trying to divert us all into this violent conversation. It’s more about curation than moderation in that sense.” In a year which has at times felt apocalyptic, a vision of an online world characterized by peaceful coexistence is certainly one with mass appeal. Let’s hope OpenWeb can pull it off.