Fake News is Solvable

misinformation online, especially on social media. For the most solvable, I think we need increasing awareness, increasing cooperations, helping algorithms make better decisions, recognizing that recommendation engines are not functioning as they should, and that we should be taking tangible steps to think about ways in which algorithm curation serves

information to people. In late twenty sixteen, Oxford Dictionaries selected post truth as their word of the year, defining it as relating to or denoting circumstances in which objective facts are less influential in shaping public opinion than appeals to emotion and personal belief. It's like, I want to believe that nachos are the ideal balanced nutritional snack that appeals to my emotional and personal belief system, because God, I

love nattos. So I'll go and I'll find some vague chitchat online that tells me, you know, something melted cheese is totally full of calcium that is good for your bones, and it's important for you, as an immigrant to the US, to assimilate by eating their national dish of natos. So I'll convince myself of that, and I'll maybe even eat

myself into a delicious early grave. The age we live in, the digital age, affects every narrative we see and absorb, and that can be news based, or cultural or artistic. We have always had an instinct to find information that sinks with our perspective, and now a host of new platforms are only too happy to oblige that part of us.

Pew reports that an analysis of almost four hundred million Facebook users interactions with over nine hundred news outlets found that people tend to seek information that aligns with their views. That makes many of us vulnerable to accepting and acting on misinformation. Social media firms are under pressure to halt the spread of fake contents on their platforms, and we know that the problem has both human and technical side,

and so too does any potential solution. Reneed Arresta is the director of research at New Knowledge and a Mozilla Fellow in Media Misinformation and Trust. She investigates the spread of malign narratives across social networks and helps policymakers to understand and respond to the problem. Renee has advised Congress and the State Department, and she studies some fascinating areas of disinformation in contexts like pseudoscience, conspiracies, terrorism, and state

sponsored information warfare, all that spooky stuff. I'm so glad she is scouting ahead and sending us back the best ways to deal with this. Let's take a listen and I'll speak to you after. So, Renee, you're one of the few people who identified the problem of online anti vax disinformation very early on. How did you first come into contact with the problem? How did you know it was a problem at all. I started working on a law in California called SP two seventy seven, and it

was a law to eliminate vaccine opt outs. And I was a parent at a new baby, and I wanted, as a mom, just to volunteer to help get this law passed. So I am a data analyst, and I offered to do some analysis into things like the social media conversation around the law. And I was really surprised because the legislators, there are a number of legislators on

both parties who were supporting the law. They were saying that their constituents were polling at around eighty five percent in favor, but the social media conversation was almost one hundred percent negative, and that was on Facebook and Twitter.

So I started working with another data scientist named Glad Latan to look at the conversation on Twitter, to look at the different distinct groups, how they were evolving their messages, how they were connecting with other activists outside of California,

how sometimes activists outside of California. It turned out we're pretending to be Californians, bunches of new accounts that had been created, and we were really looking at the idea of what had no name then but kind of came to be called manufactured consensus, the idea that the conversation online was really being driven by a relatively small number of people who were using things like tools to be always on, constantly being in the hashtag Facebook groups and

ads to amplify their message, and then the way that the algorithm was amplifying the message. In addition to that, so ways in which I, as a parent who had just gotten involved in the conversation, had just demonstrated an interest in vaccine policy, was all of a sudden getting pushed tons of anti vaccine content on Facebook. It was recommending groups to me, it was recommending pages to me.

And the realization that what was really not a very large number of people was actually having an extreme disproportionate amount of a share of voice in the conversation, and did you have to create tools in order to begin identifying who the people were who was being pushed. They

were actually not very quiet about that. There was a page called tweet for Vaccine Freedom, and it was actually you know, when out of state activists were asking how can we help because the entire anti vaccine movement across the entire United States decided to fight this battle. They would say like, oh, you should just create an account and say or from California. So it was actually really transparent. It wasn't that hard to figure out that there were

people pretending to be from California. There were also Twitter accounts that all of a sudden had a vested interest in California politics. But if you read their past material, which again is also public it was really right out there that that's not where they were actually from. Kind of a very interesting because it was extremely small, local

and niche we you know, we thought in California. But as the law began to get more press coverage and things or would actually be like comments section battles, you know, the same kinds of stuff that we saw later with you know, entities that go and are like almost incentivized to leave comments on news articles to shape a perception about the topic. And actually we on the provac side thought, oh boy, I guess we're gonna you know, we need to do this too. Are we really engaging in Okay?

They commented over here, so you know, we have to go comment over here. They have bots that are on twenty four to seven? Do we need bots that are on twenty four seven? Just became this this interesting firsthand experience of what it was going to be like to try to run any kind of influence or policy campaign in the future. I found it really troubling, especially when the algorithms just began recommending anti vaccine content to me constantly.

And how did the Facebook and Twitter and other algorithms work. Were they affected by this campaign? The where the search engines affected by it. I don't think the search engines as much because it was you know, the Google is a little bit more sophisticated about this stuff than the social platforms. Social platforms the number one signal that they're using as popularity, and so you either if you have

real popularity or if you can feign popularity. The number of likes and engagements and comments and things is what decides, you know, whether this is how Facebook was deciding what gets pushed into your feed. Instagram is like that too. Google has a framework now it has a proper name.

It's called Your Money or Your Life, and it says that on topics related to health issues and financial issues they have to have a higher standard of care to make sure that it isn't just what's popular that's rising

to the top. But even with that policy, one of the things that we consistently see is anti vaccine activists producing content at a higher rate and also candidly more engaging content, you know, a much more emotionally resonant versus more authoritative medical quote unquote establishment doctors, the CDC, the National Instituites for Health, their contents not as emotionally resonant.

It doesn't get as much engagement, and so the search engines and the algorithm aren't amplifying the more factual, reality based content, and instead what we're getting is this conspiratorial stuff. Walk me through what it means to be emotionally resonant online. Is this something that's being done deliberately to the people who are creating it understand that that's what it is? Or is it that the human brain is just tuned

to conspiracies and prefers them. Some of it is platform culturism, of it is the way that the algorithm understands engagement. So there's the human element which gets kind of the initial signal shows that there's a lot of people who are watching this, and the algorithm recognizes that a lot of people are watching it and then begins the amplification process.

But the first step is actually the content, of course, and in that particular area, it's usually a first person, you know, looking directly at a camera, speaking about a personal experience they've had, recounting a narrative or an interesting story. So a lot of times with the anti vaccine movement, that's a person claiming that their child has autism and telling a story, you know, usually very sad story about

their child's health, and so it is engaging. It is much more resonant versus seeing kind of infomercial about how vaccines don't cause autism because thousands and thousands and thousands of studies have said that they do not. I know that you were part of the Senate commission that looked through material that we knew that Facebook handed over to Congress which was originally created by the IRA, the Russian

Internet Agency, in order to influence the US elections. When you looked over that material, did it seem to use those same tactics? Can you see a relationship between the way the Russian influence campaign worked and the anti vax campaigns? The Russian content was distinct, and that this was a foreign intelligent service of a foreign entity that was trying to pretend to be American. So it was far more duplicitous than anything that we've seen related to domestic activists

pushing for a cause. Really, but what was happening there was again they were taking these extremely big topics things like who is America for? What does it mean to be an American? How do we feel about immigration? How do we feel about gay rights? How do we feel

about police brutality? They were creating these pages, and each page was designed for a very particular type of person, So they were really creating these tribes, again relying on the sort of first person experience, first person concerns and fears, and putting out content that was again very much focused

on achieving an emotional response. So for the black community, the content took the form of constant references to police violence mixed in with narratives of pride, and so it was really very much designed to evoke cultural pride and then also a sense of deep harm. And on the right leaning pages, it was really concerned about what America is and who it's for, and so a lot of

photos of things like homeless veterans. This is a very real problem that we have in this country, and they were using the images of homeless veterans to say, why are we allowing in all of these outsiders when we can't take care of our own. This is how propaganda is most effective. It's when it has some degree of truth to it, and it spins it just enough that it doesn't necessarily trigger the part of the brain that says, hey,

this is false. Instead it the person relies on the emotional reaction to it, and that's how they begin to develop a sustained relationship with the page and sustained engagement with that type of content. You know, as I'm listening to you, I'm wondering whether different kinds of propagandists they understand now that they need to tailor messages to particular audiences.

Is it the case that some of the solutions to this they're also going to involve thinking differently about different audiences or offering different kinds of counter messaging or counter strategies to different audiences. Yeah. Absolutely, And this is something that you know. A third area I worked in was countering violent extremism. Briefly was Isis. The idea that we would kick Isis off the platforms was sort of a

stretch at the time. There were a lot of people who were very concerned at the idea that we would delete terrorists accounts, and so a lot of the focus instead was on counter messaging. How do we reach these audiences that are receptive to ISIS propaganda and present counter narratives to them. Who is the authentic voice for the

counter narrative. It's definitely not the United States State department, So who is it and what are the ways in which we can come together to think about ways to counter message to try to present people with it an alternate, also emotionally resonant narrative instead of just saying it's a bad idea to be a terrorist because you're going to go to jail or you're going to die. A lot of the tribal deep affinity ties is what is my

place in society? This is something that comes up with conspiracy theorists also, they're looking for answers, they're looking for an explanation. What you get hooked into oftentimes is what is most visible to you, what's most prevalent in your

space at that moment. Now that we're spending so much more of our time online, things like ad targeting and participation in Facebook groups where you're kind of declaring a particular alignment mean that bad actors who want to target you with certain types of propaganda can find you very easily. And can we reapply some of that thinking back, for example, to the anti vax problem. Can we think about counter messaging there? Can we think about how to reach people

using counter emotional stories? Yes, absolutely, that is That's something that our groups like Voices for Vaccines are trying to work on that. The group that was formerly called every child by two, it now goes by vaccinate your family is trying to do that. We have to get out of statistics and get into storytelling. That's the one of the key takeaways of how the information ecosystem has evolved.

If you look at even just from a design perspective, one of the things I always get at is the the subject of the narrative is interesting when you're thinking about how to counter message to a particular group of people. But when you think about this as a problem written large, a lot of it comes down to the algorithms and the design. And so memes in particular are getting more

and more important in our lives. And that's because the design of the platform itself is privileging this large you know, this large square image or this piece of video, this short video clip. So what can you convey in the construct of that design. As people are scrolling by, they

see your message in it immediately sticks. The fact that the algorithm will continue to serve up types of content that you've engaged within the past means that if you do engage with anti vaccine content, you're likely to see more of it. The challenge of algorithms that don't know what they're pushing because they have no actual awareness of

what the underlying content is. So they treat something that's potentially radical, they treat something that's potentially blatantly false the exact same way that they would treat something that's accurate or uplifting. They don't actually know. They just know that this content drives engagement, and so they continue to show it to people. We see all this disinformation online. We you know, we hear about it. You know, we can sometimes see it in our Google searches. But doesn't really matter.

I mean, for example, in the anti vax campaign, has this really affected anything, Does it make any difference? Or is this just stuff that exists somewhere in the ether and if we ignore it will go away. Let me give you two quick examples on that. First of all, with the anti vaccine movement, Yes, it absolutely has an impact. It really creates a lot of fear and hesitancy, and that translates very directly into vaccination rates declining in the communities that are that are seeing it. And so this

is something that in California. The reason I started looking at it was because immunization rates in California communities had declined, and when I was trying to find a preschool for my son, I was actually looking at these rates, and there are certain schools in California with thirty percent immunization rates,

which is terrifying. That's like South Sudan. The reason that we passed the lawn California was because we wound up with the Disneyland measles outbreak, where two hundred and something people got sick and I believe a quarter had to be hospitalized. So this was a very real outcome of that kind of misinformation becoming so pervasive to people, creating that very real fear and then leading to an outbreak in the case of Russia. Just because a lot of

people think about this is just related to the election. No, what they were doing was they were also creating real world events. So they were sponsoring protests, and one of the things that they sponsored was an incident in Texas where they had two competing protests on the same day at the same time. So from Saint Petersburg, Troll created a Facebook event saying that people with Texas Pride had to come and protest outside of an Islamic center to

defend their way of life. They also posted an event calling on members of the Islamic Center to come out and defend the Islamic faith. So they sponsored two protests on the same day at the same time, across the street from each other. And you can go on YouTube and you can see the video footage from that day of people showing up with kind of anti Islamic material on one side of the street and then people on the other side of the streets screaming back at them,

and police getting involved in breaking up altercations. So this is an example of very real world tension erupting as a result of online disinformation. When you first started looking at this problem, did people believe it was a problem. Opinion polls all showed people were in favor of vaccinations. You saw something quite different online. How did you convince people that this was something they need to take seriously.

In the California case in particular, I sent what I was seeing, you know, kind of quantifiable evidence to the legislators and said, I don't think that people are screaming at you online, they're threatening you online, You're seeing all of this anger and rage in the hashtags, I don't think that these are not your constituents, where it's pretty pretty clear that these are not all even Californians. So when you make your decision, I would lean into the

polling numbers and the communications with your actual constituents. I don't think that we can treat the online conversation as representative of the reality of the population of California. So in that particular case, it was just really kind of appealing directly to the legislators with the evidence the challenges it really does bump up against things like freedom of expression right. So you have a right to have an anti vaccine opinion. Of course you have a right to

put the content online. The challenge was at the time, the recommendation engine, the trending algorithm, the ways in which Twitter and Facebook were amplifying information was very different, far more primitive then than it is even now two and a half years later. After those of us who work on this challenge have kind of been constantly beating the drum with example after example of example of how this

is manifesting in the real world. How do we preserve freedom of expression while at the same time recognizing that the platform is pushing this point of view at people people aren't even know me. In particular, I'm not going out there typing in anti vaccine search terms. The recommendation engine is just pushing it to me because it's seeing

that I've expressed an interest in vaccines in general. As part of working on this law, I suppose there's also a question of Okay, you have a right to write something, but then do you have a right to artificially amplify it using bots and search engine optimization? So everyone has the right to freedom of expression online. The secondary piece of that, though, is do you have a right to free reach your right to algorithmic amplification. Nobody has that right.

That is not part of the First Amendment, That is not part of our cultural experience of what it means to have a right to express, or you have never had the right to free mass dissemination as well. That's the piece where as people begin to talk about how

the platform should think about these things. One of the ways that we can continue to maximize freedom of expression is to allow people to speak, but also for the algorithm to perhaps not begin to take that kind of sensationalist content and proactively broadcast it out to massive quantities of people because it checks the boxes of being sensational

and emotional. And do you think that it's going to be enough to discuss this with the platforms, for people like you who have you, who are respected on these issues, to talk about it with people at Facebook and Google, or is this something that we're going to need to regulate or have Congress step in on. I don't think you can have Congress regulate what algorithms amplify. I think that that would probably be a little bit too close

to Congress making decisions on speech. A lot of the dissemination that come about through inauthentic amplification through things like bots and stuff, can be addressed without even knowing what the narrative is actually about. So you're not looking for content related to a particular topic. You're looking for particular dissemination patterns. So you're looking at the authenticity of the accounts. Are these real accounts where they all created yesterday? Are

they bots? Are they majority automated? Are they Twitter does have now a designation of something that considers a low quality account ways in which it surfaces top tweets, as opposed to just the straight up reverse chronological order where you see every single tweet about a particular hashtag, giving the user some control. So people who do want to go see that kind of fire hose of every single tweet coming through about a topic can go and do that.

But the majority of people who just want to get the kind of quick takeaways are seeing more kind of higher caliber content. And that sounds like you do believe algorithms could eventually identify quality content that they could encompass a notion of better or more comprehensive or more fact based.

Remember the olden days of the Internet where you had email spam, right, we did build classifiers, We did build tools to think about how to ensure that crap wasn't flooding people's inboxes, that there wasn't this mass cognitive load every time you opened your inbox of having to sift through all of the garbage to find the communications from people that you actually wanted or or find the things that were really intended for you. We need to put some things in place here to improve the system, to

improve the user experience, to improve the outcomes. There were things like recognizing that certain domains were just not reputable domains that most people wanted in their inbox, and so some of this was user filtering, you know, feedback. You remember you used to kind of mark things as spam much more regularly then. It didn't mean that there were

never false positives. There are still false positives today, But it was how can we create greatest value while at the same time recognizing that there are extremely coordinated, deliberate groups of people working to manipulate and evade that detection, in the case of spam, to wind up in your inbox and in the case of social algorithmic manipulation to wind up in your feed People who are concerned about this problem, people who worry about online disinformation, people who

worry they're getting bad information. Is there anything they can do about it? Is there something that ordinary people can do to fight back? Stopping this spread a lot of the time is something where individuals really have a lot of power. It's been for a long time, you know, kind of a cultural norm where if you see someone sharing something a little bit nutty to just kind of ignore it, just let it go by. I don't think

that that's necessarily really helped us. I've tried lately to try, like commenting gently or sending a private message saying hey, I don't think this is necessarily the most reputable source. Maybe you know, here's a fact check on that. There's a lot of evidence that says that interventions from people, you know, even in the kind of counter radicalization space, that really engagement with friends and family and people were there's a base of trust and an assumption of goodwill.

People are receptive to rethinking maybe why they chose to share something. And then when you see something that makes you feel highly emotional and you go to click the share button or the retweet button just because you know, you feel outraged and you need to tell the world, that's where I think taking the extra second to stop and do the fact check, to stop and see is this a reputable domain or a reputable account, it really

makes a difference. So friends, don't let friends share disinformation, and always check whose account you're retweeting or reposting before you do it. Yeah, I mean, I've made this mistake a couple of times. I remember I once retweeted something and a friend of mine ping me and said, hey, I think you should go read the rest of that accounts tweets, And I went and looked, and I ninety nine percent sure it was a bot, and I was like, oh,

I fell for it, you know so. But that's the kind of thing where it's far better to tell somebody. I mean, you can just unretweet, you just click the button again. And it's more challenging if you are a person with a very, very large following, and it usually helps to send a follow up or something and say, hey, I inadvertently spread some misinformation. It's come to my attention that this is not real, or here's the actual story.

It's so wild to hear about these disinformation campaigns online right now because here in the US there have been eight hundred and eighteen measles cases reported in this year's outbreak. It's already the largest since nineteen ninety four. People are in hospital here because of misinformation, and New York is

seeing the fastest spread, particularly in Orthodox Jewish communities. The thing is that in that specific case, the misinformation about vaccines was not spread online, but through physical handbooks and phone conferences. The internet amplifies what we already do, so changing algorithms and policies and our own behavior online. It's all going to take a lot of changing, and I'm really grateful to people like Renee who work towards that

every day. Solvable is a collaboration between Pushkin Industries and the Rockefella Foundation, with production by Chalk and Blade. Pushkin's executive producer is Mia LaBelle. Engineering by Jason Gambrell and the fine folks at GSI Studios. Original music composed by Pascal Wise. Special thanks to Maggie Taylor, Heather Fame, Julia Barton, Carlie Migliori, Sherif Vincent, Jacob Weisberg, and Malcolm Gladwell. You can learn more about solving today's biggest problems at Rockefella

Foundation dot org, slash solvable. I'm Mave Higgins. Now go solve it.