Guest Host: Christina Bellantoni

Did you hear the one about the exploding bong? Or the family of five quarantined in Texas because they all have Ebola? Stemming the tide of untrue stories circulating online is a challenge that websites like Snopes and Politifact have been tackling for years in the hope of debunking some of the biggest whoppers. And journalists and computer scientists are working on new programs and algorithms to track and analyze how rumors spread online. We explore what it takes to spot and correct falsehoods on the Web.


  • Craig Silverman Editor & Author, 'Regret the Error', Poynter.
  • Paul Resnick Professor, University of Michigan School of Information
  • Angie Holan Editor, Politifact


  • 12:06:40

    MS. CHRISTINA BELLANTONIFrom WAMU 88.5 at American University in Washington, welcome to "The Kojo Nnamdi Show," connecting your neighborhood with the world. I'm Christina Bellantoni, editor-in-chief of Roll Call, sitting in for Kojo. Coming up this hour, did you hear about the Twitter panic over brat pack actor Judd Nelson's supposed death? Or the one about the family of five quarantined in Texas because they all had Ebola? Stemming the tide of untrue stories circulating online is a challenge that websites like Politifact and Snopes have been tackling for years.

  • 12:07:18

    MS. CHRISTINA BELLANTONIAiming to debunk some of the biggest whoppers. But as social media spreads the spread of both good and bad information exponentially, keeping up with all the falsehoods is becoming an ever more Herculean task. Now, journalists and computer scientists are working on new programs to help detect and track rumors circulating on the web. This Tech Tuesday, joining us to discuss, here with me in studio, is Angie Holan. She's the editor of Politifact, Pulitzer Prize winning, independent fact checking political journalism website. Hi Angie.

  • 12:07:51

    MS. ANGIE HOLANThanks for having me.

  • 12:07:52

    BELLANTONIAnd joining us on the phone is Craig Silverman. He's a fellow at the Tow Center for Digital Journalism at Columbia University. He created, which is a tool to track rumors online. He's also the editor and author of "" with the Pointer Institute. Hi, Craig.

  • 12:08:11


  • 12:08:13

    BELLANTONIAnd joining us from Ann Arbor, Paul Resnick is a Professor of Information in the School of Information at the University of Michigan. Nice to hear from you, Paul. We're, we're working to get...

  • 12:08:26

    MR. PAUL RESNICKWell, did you not...

  • 12:08:27

    BELLANTONIHi, Paul, can you hear us?

  • 12:08:27

    RESNICKI'm hearing you fine.

  • 12:08:28


  • 12:08:29

    RESNICKGlad to be on.

  • 12:08:30

    BELLANTONISo, you know, we'll just sort of open up with -- there's so much false information on the internet. The Washington Post has actually taken to round up of these stories on a weekly basis. How are internet rumors spreading? I'll start with you, Angie. And why is this something that technology needs to address?

  • 12:08:48

    HOLANYou know, we've seen a very interesting evolution at Politifact with dealing with internet rumors, false claims, conspiracy theories. When we launched in 2007, most of those types of things that we fact check were chain emails. These are the emails -- they're often written in all capital letters. They have a lot of exclamation points. And they say, forward this to all of your friends. And then, usually, in our experience, they tend to have a lot of false information in them. Now, with the rise of social media in recent years, we've been doing more and more fact checks of Facebook memes.

  • 12:09:31

    HOLANOr information spread through Twitter. So, I think that there are a lot more channels online for people to spread information. And it's really never been easier to cut and paste something and change its meaning. And then see how far it will spread.

  • 12:09:46

    BELLANTONISo Craig, speaking of spreading, of course, the viral virus Ebola that we've seen become a viral email initiative. And also Twitter information, you know, as Angie points out. You know, Craig, how -- what kind of a case study is Ebola and all of the Ebola rumors?

  • 12:10:03

    SILVERMANWell, it's an interesting one, because it has some of the elements that really lead to the creation of rumor. You know, people are very anxious about Ebola. They may not have the best information about it. There might be a bit of a vacuum of information or some misunderstanding. And it's something that creates a lot of anxiety and uncertainty. And in those kinds of situations, human -- the way we try to deal with them is actually to kind of talk through them and to see what might be going on.

  • 12:10:29

    SILVERMANAnd that leads us to actually create and pass on rumors. So, I mean, we've seen tons of rumors. Some of them, you know, deliberately false claims that are spreading. And then other things about, you know, rumors about it being airborne, which is not true. You mentioned off the top, you know, a false report about a Texas town that had been quarantined. So, lots of really alarmist kind of rumors that scare people and that feed into our worst fears.

  • 12:10:51

    BELLANTONIAnd very different from, you know, another one that we've seen out there is whether there is a red velvet Oreo or not. Right? You know, this is something that, some of them are fun, some are a bit more sinister like this. So Paul, joining us, again, from the University of Michigan, you're on the computer science side of this, creating a computer program to track these rumors. How did you get interested in looking at them?

  • 12:11:13

    RESNICKWell, we originally started trying to think about how we could organize the crowd to respond to rumors. And found that there wasn't really a systematic way of finding, quickly, the rumors that were emerging so that we could feed them to people to respond to them. And so we started developing tools for just finding those rumors that are emerging and then being able to enable analysis of them by finding all of the tweets that might be related to a particular rumor and creating some tools that let us analyze how many people those rumors have reached.

  • 12:11:55

    BELLANTONIAnd it's amazing watching how quickly they can spread. You can join our conversation and tell us what's the biggest online whopper you've seen on social media lately? Join us by giving us a call at 800-433-8850. Send a tweet to @kojoshow. Or email us at So, obviously, when I'm not sitting in Kojo's seat here, my normal job is pretty obsessed with politics and campaigns. And I'd love to ask Angie, you know, obviously Politifact's a well respected site. You know, what are some of the biggest political rumors that you've seen this campaign season? We're a week out from Election Day, mid-term elections, and a lot is at stake.

  • 12:12:33

    HOLANYou know, one of the biggest rumors that we've seen was repeated by a sitting Congressman, Representative Duncan Hunter of California. And I should point out, he's the younger Duncan Hunter. Some people may remember his father. He said that 10 fighters with ISIS, that's the Islamic State, some people call it, the Islamic State in Syria. He said that 10 of these fighters were apprehended at the Texas/Mexico border. Well, we looked into this, especially our journalists at Politifact Texas. That's our partnership with the Austin American Statesmen.

  • 12:13:06

    HOLANAnd there was really just nothing to this. The evidence that they offered were anonymous news reports in some very partisan media outlets. And public officials at the Department of Homeland Security and the Texas Department of Public Safety all said that there was nothing to this. So, that's something that -- our rating was pants on fire. And that's something that was repeated by a member of Congress, which is kind of interesting to see that sometimes people at very high levels of the government will repeat things that don't have much evidence to them at all.

  • 12:13:39

    BELLANTONIWhat are some of the other political rumors that you've seen this season?

  • 12:13:42

    HOLANOkay, the other types of things we see are kind of more under the radar. Now, one check I wanted to talk about is a video that goes around the internet. And supposedly, this video shows President Barack Obama giving a speech, and he says that ordinary men and women are too small minded to govern their own affairs. And that these people should surrender their rights to an all-powerful sovereign. It's kind of a whacky claim. It's like, if President Obama had actually said we should all submit to monarchy, that might have been in the news.

  • 12:14:15

    HOLANBut it's this claim that's promoted by social media, and when you look at the video, it looks very realistic. Well, we researched the video, and what we found was it was constructed by doing a very artful edit. President Obama was giving a speech where he was kind of giving this history of Democratic thought, and he said some people used to think this. And these, you know, whoever it was, and we don't know who it was, took his speech and they, at just the right moment, they cut away to the audience and then came back to him.

  • 12:14:45

    HOLANSo it looked like he was saying something he wasn't. And this is something that I haven't read much about in, you know, mainstream media, but it's been an incredibly popular check on our site. It's one of our most popular checks.

  • 12:14:56

    BELLANTONIAnd then, obviously, little specific things like, you know, whether it's about women's health and maybe a legislative bill that passed or was signed or wasn't signed. You guys sort of get at the truth of those, but they spread very quickly, it seems like.

  • 12:15:09

    HOLANYes. They do spread very quickly. And our approach at Politifact is we just apply old fashioned journalism. We look into it. We look for the original source documents. If you are used to this kind of fact checking and this kind of method, it's not that difficult. The conspiracy theories, they just tend to fall apart when you give them any sort of really close scrutiny. And then, in our reports, we lay out the evidence and we have links so people can see for themselves.

  • 12:15:35

    BELLANTONISo Craig Silverman with the Tow Center for Digital Journalism at Columbia University is joining us on the phone from Montreal. And I'm really fascinated with and how it works and I definitely want to ask you about politics, but I'll start with what most people are going to be familiar with that happened recently. Is Judd Nelson did not, in fact, die. How did that start and how did you debunk it?

  • 12:15:57

    SILVERMANYeah, so, that started in a way that I'm starting to kind of encounter more and more. So there was a website that had the domain name foxnews.ef. And they had a report up that he was dead. And so this is, this is a site that has absolutely no affiliation with him or with Fox News, rather, whatsoever. And, in fact, it was a site that just had this one report up there. And that started to get out there. People started sharing it on Twitter and a lot, in particular, on Facebook.

  • 12:16:27

    SILVERMANAnd so it got to the point where journalists, sort of, saw this and said, okay, let's see if this is real or not because it's being repeated by so many people. And so a lot of folks, you know, at the Hollywood Reporter, at Vanity Fair, reached out to his rep. and they heard -- and he actually ended up tweeting a picture of Judd Nelson holding up that day's Los Angeles Times to show no, no, I am alive and well and this is all fake.

  • 12:16:53

    BELLANTONIWow. Well, you can definitely join our conversation and give us a call at 800-433-8850. Tell us if you've read misinformation about the upcoming election or about Ebola or ISIS, any of these rumors that we've been discussing. You can send a tweet to @kojoshow. Or send an email to Rachel from Silver Spring, Maryland joins us now. Hi, Rachel, thanks for joining us.

  • 12:17:17

    RACHELThank you for taking my call. It's always interesting to me to hear that, I don't know, people are still able to believe just about anything they see. Back in the late 90s, I was working at a museum. And the security guard, I guess, cause he didn't have a lot to do because it was a safe place, started sending out emails to the whole staff, to everyone he knew, with the same urban legends that, you know, we used to hear about years ago, how gangs are putting LSD on payphones. We still had payphones back then.

  • 12:17:49

    RACHELBut the worst one was that all veterans had to report to their local VA or they would lose all benefits because President Clinton was signing a secret bill that very day. And, of course, it had a fake phone number at the end. And I called the VA until I found someone who I could talk to, and he said oh please, please send out an email that says this isn't true. I mean, you know, we are getting calls from people. We don't know how this started. So I did and I sent it to his entire list. And then I was called out for it. And they said, why are you trying to make this person look foolish?

  • 12:18:29

    RACHELAnd even back then, I thought, oh boy. If they think it's bad that I respond now, what's going to respond in the future when everyone has email? And now we know.

  • 12:18:41

    BELLANTONIThank you for calling, Rachel. Rachel brings up such an interesting point, because we've all gotten those emails from, you know, our great aunt or someone that might not be as internet savvy as we are. And one quick Google search can bring up a Snopes or any one of your wonderful sites. So Paul Resnick, I'm curious, how do you combat this issue when it's coming from friends and loved ones?

  • 12:19:01

    RESNICKIt's tricky, both because of the social niceties. You don't want to be seen as, you know, a know-it-all who's telling people that they're foolish, and also because there's some research questioning the effectiveness of corrections. And that, not so much with -- the studies haven't been with friends and loved ones, but you know, in some public forum, somebody says something and you correct it, the people who are listening may be more likely to believe the falsehood than if it had gone unmentioned.

  • 12:19:39

    BELLANTONIHmm and what about you, Craig Silverman, is that a method of debunking rumors?

  • 12:19:45

    SILVERMANYou know, it's -- the caller raises a totally valid and documented aspect and Paul talks about it a little bit. That, you know, it -- they say it's been documented that people who actually step in and say hey, that's actually not true. Rather than getting any gratitude, people often say well, you know, it was a nice story. Why did you have to ruin it? Or, you know, as Paul said, why do you have to be such a know-it-all?

  • 12:20:07

    SILVERMANAnd so this is one of the biggest challenges, I think, when it comes to debunking, is that people in general often don't want to hear it. And, you know, one of the reasons is that, you know, sure it may make them feel a bit embarrassed that they had put something out there, what have you. But the other reason is that when it's information connected to, you know, strong beliefs we have, when it's information that's already kind of implanted in our minds, our reaction is to reject things that contradict that.

  • 12:20:32

    SILVERMANAnd that's, you know, a very well documented fact. It's called the backfire effect, where when you have your beliefs and views challenged, you actually kind of double down on those beliefs. And that's one of the reasons why, you know, I'm doing the research project (unintelligible) is to figure out, well, what debunking strategies actually work? Which ones, you know, get the shares moving in the direction of truth and which ones will people actually pay attention to? And the truth is we don't have great answers about that yet.

  • 12:20:57

    BELLANTONISo we got a tweet from someone to @kojoshow on Twitter, saying, "If the issue warrants I try to alert as many in the chain of posting that it's an urban legend." We're talking about new tools to track falsehoods online. And we will be right back after a short break. This is Tech Tuesday.

  • 12:21:37

    BELLANTONIWelcome back to Tech Tuesday. I'm Christina Bellantoni, editor in chief of Roll Call, sitting in for Kojo Nnamdi. This is a conversation about tracking rumors online and I'm joined here in studio by Angie Holan of Politifact, on the phones from Craig Silverman, at the Tow Center for Digital Journalism at Columbia University, joining us from Montreal. He's also the creator of And Paul Resnick, a professor of information in the School of Information at the University of Michigan, joining us from Ann Arbor.

  • 12:22:05

    BELLANTONIA really interesting conversation today, I'm going to back to Craig because something I noticed Emergent is an interesting question about how the news media can pick up some of these rumors. One of the lists -- issues that you have listed is whether the man who was the Sergeant-at-Arms in the Canadian Parliament, whether or not he shot and sort of was the hero in this shooting situation earlier this month.

  • 12:22:28

    BELLANTONIAnd you had this listed as unverified. It's been widely reported -- and in fact Roll Call even wrote a story asking if our own Sergeant-at-Arms in the U.S. Congress have guns and weapons. So how does this become unverified and what do you look for to verify it?

  • 12:22:42

    SILVERMANYeah, I mean, that's -- it's a great example because, as you know, the immediate reports at this point -- anything that's mentioning him as, you know, as we look and see on Google News and other places, they pretty much reported it as fact, that, yes, he's the one who took down the gunman on Parliament Hill. So the first thing we look -- and it goes back to sort of what Angie said about their approach at Politifact is, you know, so what's the evidence?

  • 12:23:04

    SILVERMANAnd in this case, right now, the RCMP in Canada, sort of, the equivalent of the FBI, are looking into the incident. And they're recreating it and they're looking at what exactly happened. So the only evidence that's actually come out to say definitively that he was the one who sort of fired the kill shot is attributed to anonymous sources, you know, who have spoken to the Canadian press and other certainly reputable news organizations, but they're anonymous sources.

  • 12:23:28

    SILVERMANAnd the other is that, you know, in the wake of the shooter being taken down, some of the MPs, who were kind of under lockdown on Parliament Hill, tweeted out that Kevin Vickers, the Sergeant-at-Arms, had taken out the shooter. So what we have are some people who are on the scene, but didn't actually see it with their own eyes, relaying second hand information. We have anonymous sources telling this news organization.

  • 12:23:50

    SILVERMANAnd then the other piece is that people in one of the other sort of party caucuses that were locked down, were actually told that there may actually been another person in the key takedown. And so at this point we leave it unverified because in a few weeks' time or maybe it might even take a few months' time, the RCMP is going to come out with a report and they'll actually say, well, here's what the evidence tells us what happened. And at that point we'll be ready to change the rating.

  • 12:24:14

    BELLANTONIVery interesting, I'm going to go to Paul Resnick, with the University of Michigan, to talk a little bit about the technology of this. You know, this seems like a 24/7 job. And how do you study and track a rumor, how fast does it spread and what are you -- what kind of technology tools are you actually using to be able to see if something is going viral?

  • 12:24:37

    RESNICKGreat, the first test that we set for ourselves is can we detect new rumors as they happen and how quickly? And our basic technique there is to look for certain words that people use when they are skeptical of something that they've seen. And it turns out that for -- that things that spread, somebody is usually skeptical about it very soon. Even though the rumor may spread a lot, there's somebody who's saying, is this really true, or what? Or, you know, one of these phrases that is expressing skepticism. And we've found that the median time for three people to have expressed skepticism like that is 10 minutes.

  • 12:25:26


  • 12:25:27

    RESNICKSo even though most people may not be seeing the correction, somebody's expressing it. And we can use that as a way to automatically run algorithm that look for these special phrases. And that's an indicator of new rumors that may be on the rise.

  • 12:25:45

    BELLANTONIWell, you can try and stump our experts. Let us know if you've seen any of the -- something you might think is a false rumor. Give us a call at 800-433-8850. Join our conversation by sending email to, or send a tweet to @kojoshow. Mona, here in Washington, D.C. has one for us. Thanks for calling us, Mona. You're on the air.

  • 12:26:05

    MONAYes, hello. Yes. What about the pictures of Brzezinski with the Taliban and more recently McCain with armed ISIS soldiers?

  • 12:26:14

    BELLANTONIGood question. I'm going to throw it open to the panel. Anybody have that one?

  • 12:26:20

    HOLANYou know a number of outlets have looked at the question about McCain and ISIS and there doesn't seem to be much to that. I know that we have not looked at this at Politifact. I think I'm going to go back to the office and put this on our list. I will say that the photos -- because of Photo Shop, we see a lot of manipulated photos. So it's not so unusual to see not only political photos, but also like nature photos, like down in Florida.

  • 12:26:52

    HOLANPolitifact is run by the Tampa Bay Times Newspaper. And when Bay Shore Boulevard in Florida flooded last year -- or a few years ago, there was a photo going around that was supposed to show a giant alligator on Bay Shore Boulevard, swimming along. Well, you know, when you drill down into some of these photos it's very easy to manipulate them. So we're always a little bit extra skeptical about photos.

  • 12:27:15

    BELLANTONISome of those creepy giant spiders, that you see, that sort of a thing, although some of those are real. Mike, in Oxen Hill, Md., has a thought on this. Thank you for joining us, Mike.

  • 12:27:26

    MIKEYeah, hi there, Theodore Sturgeon, the great science fiction writer, said probably 50 years ago now, that 90 percent of everything is crap. And I think he replied recursively because 90 percent of what's left is suspect. And I think that sense of this conversation is inverted, in that the question is not how do you tell something is wrong. The question is how do you tell something is right. Because, frankly, 90 percent of everything you see is crap. So that's how I get by.

  • 12:27:58

    BELLANTONIWell, I think that's a good question for Craig Silverman, with, because you're verifying things as true just as often as you are false, right, Craig?

  • 12:28:07

    SILVERMANYeah, I mean, certainly -- so the first thing is that I agree with the caller's orientation. When you're practicing verification of this kind of stuff, the best way to come at it is to assume it's not true and then to force it to prove its trueness to you. And part of that is, you know, just an approach. But the other part is, you know, going back to the way we process information, if we start by believing something to be true, then everything that -- all the evidence we take in, all of the information we gather from that point forward, it's seen through the prism of, oh, it's probably true.

  • 12:28:38

    SILVERMANAnd so starting by believing things are false is definitely one piece, one thing that I adhere to. And in terms of stuff that ends up true or false on the site, you know, I haven't done a sort of tally and a breakdown. One of the things that's actually kind of interesting and probably relates to this as well is I have been surprised by the number of items that end up staying in that grace space of being unverified. Stuff that we just haven't seen the evidence to kind of sway it one way or another.

  • 12:29:03

    SILVERMANAnd what's kind of interesting is I feel like there's a lot of things out there, claims that are made, information that goes around, that you may never really get a resolution to. And then this stuff launches in peoples' minds. And in many cases over time, it starts to take on element of trueness. And that's one thing that kind of really concerns me. There's so much, you know, emerging and unknown and information in that gray area, and we often don't get answers to it.

  • 12:29:27

    BELLANTONIWell, that is a good segue for me to ask about the Affordable Care Act, sometimes referred to as Obama Care. And emails, you know, there are many years old emails about rationing of health care, chain emails that even sometimes point to specific pages in the Affordable Care Act legislation. But somehow these falsehoods are still spreading, Angie Holan, with Politifact.

  • 12:29:48

    HOLANYeah, we've been fact-checking the Affordable Care Act for years and years. I mean, we started when it was just the platform of the Democratic candidates. And have followed it through the lawmaking process to its final form. We fact-checked more about health care than any other topic on Politifact, and, you know, there's stuff -- there are claims that you see in political ads that -- where people are giving their spin, but then there's this whole other genre, I would call it, of emails that make wild claims about the health care law.

  • 12:30:20

    HOLANLike it gives free health care to illegal immigrants or Congress gets free health care for life or one that is of enduring popularity -- and I really cannot explain why it's so popular. This is a chain email that says the word "dhimmitude" is on page 107 of the health care law and it means Muslims are exempted from the government mandate to purchase insurance. We have rated that Pants on Fire. If you haven't heard of the word "dhimmitude," the email claims that it's supposed to claim the status of non-Muslims living under some sort of Muslim authoritarian state.

  • 12:31:01

    HOLANBut the word isn't in the health care law. There's no exemption for Muslims. There's a religious conscious exemption, but all the evidence points toward Christian groups who would be taking that exemption, not Muslims. So overall the email is just wrong on so many points, Pants On Fire. And yet, it's been floating around the internet for years and I expect it will continue to do so.

  • 12:31:24

    BELLANTONIAnd sticking with Angie Holan, from Politifact, for a moment, obviously during the 2008 campaign, questions about the president's heritage or his own faith background were very prevalent. And people would assert this, you know, as fact, both on the news or whether they were pundits. It came up many times. And how far do you go to prove something incorrect?

  • 12:31:44

    HOLANWell, came to the point that President Obama released his full form birth certificate at a press conference. I mean, I think that shows how they do consider some of these internet rumors -- and, although, I think that one, like, started as an internet rumor, but went well beyond. So that elected officials were being asked about it and expressing doubt. And I do think, as a politics fact-checker, we focus on politics at Politifact.

  • 12:32:11

    HOLANA website like Snopes does a lot of pop culture fact-checking, but we focus on the politics. And it sure does seem like we've seen elected officials, who see these reports and don't check it out and then just repeat it themselves. So the line between viral internet and what elected officials are saying is -- sometimes it's very porous.

  • 12:32:31

    BELLANTONISo, Craig Silverman, who is a fellow at the Tow Center for Digital Journalism, at Columbia University, creator of, is also the author of "" with the Poynter Institute. And what have you found about when something is spread that is wrong, and then the corrections -- how often do people actually look at those, Craig?

  • 12:32:50

    SILVERMANWell, you know, there's two pieces to the correction puzzle. So the first being, are corrections actually offered by news organizations who report something that turns out to be incorrect? And then the second piece, of course, is well, you know, do people see them? On the first, the only study that's been done about kind of the number of corrections we get was done a few years ago, related to newspapers in the United States and it found that roughly about 2 percent of articles that had verified factual errors in them were ever corrected.

  • 12:33:17

    SILVERMANSo there's a problem in the sense that we arguably need many more corrections coming from the press. And this is something we're documenting on Emergent. Where we see, you know, a lot of news organizations jump on a kind of viral type story early on, but then when it's proven later on, they don't necessarily come back and update or just, you know, do another article. So there's that piece where I don't think they're being offered enough.

  • 12:33:40

    SILVERMANAnd then the second part is often corrections are kind of added to an existing article or maybe a second article, but oftentimes there isn't as much effort put into promoting that. And, frankly, the other piece is that corrections aren't as interesting as the initial claim. So if we take another recent kind of viral story that turned out to be a fake, the woman who claimed to have had a third breast added. And this went everywhere online.

  • 12:34:03

    SILVERMANLots of people looked at it. And we saw some early reports treating it like it's true or early reports treating it like it could be true. And then when it was actually debunked that this was a hoax claim by this woman, frankly, there really weren't a lot of articles written getting that out there. And I wouldn't be surprised if there were people out there in the public who think that she, you know, may have actually been telling the truth about that. So it's definitely a real problem that the misinformation in the space get lodged in peoples' minds.

  • 12:34:30

    BELLANTONIAnd sometimes, you know, the Google analytics tell us something. So, Paul Resnick, with the University of Michigan, tell us what you see from a technology standpoint about how the corrections get circulated.

  • 12:34:40

    RESNICKYeah, I was just going to jump in here because Craig was talking about the journalists publishing the corrections. But we've been tracking what corrections get tweeted about. And for several rumors that we've looked at we've found that there are a lot of people tweeting the rumor, a lot of people tweeting a correction. Typically, the correction is reaching fewer people than the rumor. But the really astonishing thing is that we've found that they're reaching almost disjoint audiences.

  • 12:35:16

    RESNICKThe people who are -- who have been exposed to a tweet spreading a rumor are not the same people who are following the people who send out the corrections. So we have this, you know, double problem that the correction may not be getting to that many people, but it's also getting to people who weren't the people who needed to see it in the first place.

  • 12:35:39

    BELLANTONIAngie Holan with Politifact?

  • 12:35:40

    HOLANYou know, the issue of correcting misinformation is so interesting. One thing that I've been following is that the American Press Institute has been trying to organize research around fact-checking and to see what kind of fact-checking can correct misinformation. It's, you know, I think they're still gathering evidence. But one point that I see for optimism is sometimes it seems like fact-checks are more effective if they're presented from someone who is either a friend or within the group.

  • 12:36:08

    HOLANSo if there are listeners out there who are, you know, wondering, if you offer better information to someone you're on good terms with, someone you're friendly with, that can make an impact, whereas, if it's the uncle at Thanksgiving dinner who you always fight with anyway, maybe that won't help so much.

  • 12:36:25

    BELLANTONISo tell us if you've ever shared an article that turned out to be fake. Or have you corrected a relative, if they've done the same? Give us a call at 800-433-8850. Send a tweet to @kojoshow. Or email, like Chet, in Bethesda did, saying, "I am fed up with lies in political ads, but I understand why the tactic is used because the lies are so much more interesting than the truth."

  • 12:36:48

    BELLANTONIAnd I think we've addressed this about how false information circulates often more widely or faster than the truth in fact. Crystal, from Washington, D.C., has some thoughts here. Thanks for joining us, Crystal.

  • 12:37:00

    CRYSTALHi. I was just wondering about the authenticity of an article going around I think on various websites about adult legal immigrants posing as children to enroll in high school. I read this and it seemed quite absurd that -- I remember one aspect of the article where it said that the principal was forced to enroll people who had more gray hairs than she did.

  • 12:37:27

    BELLANTONIOh, my goodness.

  • 12:37:27

    CRYSTALAnd so it just seemed to be like feeding on the whole moocher, like people, immigrants coming in and taking advantage of all the resources. It didn't make sense.

  • 12:37:35

    BELLANTONIThanks for the question, Crystal. Are panel of experts, anyone want to tackle that one?

  • 12:37:40

    HOLANThat's a new one to me.

  • 12:37:41

    SILVERMANI have to say I haven't seen that one, no.

  • 12:37:44

    BELLANTONIOkay. Well, we've got another one from Diana, in Washington, D.C., noticed something when you were living abroad. Hi, Diana, thanks for joining us.

  • 12:37:50

    DIANAHi. A long time ago, before the internet, in the '80s I lived in Costa Rica when the Smurfs were popular. And there was this rumor flying around the country that if you bought Smurf-motif sheets or bedclothes or pajamas for your kids, kids were waking up scratched. Now, it's laughable. We were laughing at it. But it took the country by storm. And anybody who had imported these sheets to sell couldn't sell them.

  • 12:38:19

    DIANAAnd normal people were believing this stuff, and they had testimonials of kids and pictures of scratches on their faces and little bodies. And, you know, I mean, it was normal people. It was relatives of mine who were educated and they were saying, well, if this is true you shouldn't buy this stuff. And I think the moral of the story is, no matter how many fact-checkers are out there, no matter how implausible something seems, people just want to believe crazy stuff sometimes. Maybe they're bored, maybe it's a lack of imagination, or maybe we just still love storytelling. But, you know, once the story gets out there, there will always be somebody to buy it.

  • 12:38:58

    BELLANTONIThank you for giving us a call, Diana. Angie Holan, you were laughing at the Smurf reference.

  • 12:39:03

    HOLANYou know, I just -- I think that's -- there's something to that. I think people do like stories and they do like wild stories. I mean, you know, when we're looking at these chain emails that our office is getting ready to fact-check them, when we have a really good one we forward it to the whole team. And we're like, oh, can you believe this? Oh, this is so crazy.

  • 12:39:21

    HOLANSo I do think there is that element of enjoying a wild story. I think the problem happens when people start to believe them or become confused as to whether, you know, we should take public policy action on things that are not based in fact. That's when it becomes a real problem.

  • 12:39:40

    BELLANTONIOne of our callers is pointing out that the John McCain ISIS photo rumor was debunked by the New York Times back in September. That's something we can take a look at and maybe post at We're going to be right back to continue our Tech Tuesday conversation about internet rumors right after this break. I'm Christina Bellantoni.

  • 12:40:53

    BELLANTONIWelcome back to Tech Tuesday. I'm Christina Bellantoni, the editor-in-chief of Roll Call, sitting in for Kojo Nnamdi. And we're talking about how Internet rumors spread online and the technology to help prevent some of this. In studio with me is Angie Holan, the editor of Politifact. And on the phones, we have both Craig Silverman -- a fellow at the Tow Center for Digital Journalism at Columbia University, works with the Poynter Institute and also created, joining us from Montreal -- and Paul Resnick, who's a professor of information in the School of Information at the University of Michigan, joining us from the lovely Ann Arbor.

  • 12:41:25

    BELLANTONII'd love for each of you -- we have a lot of phone calls we're going to get to, a lot of great questions about Internet rumors -- but I'd love for each of you to talk about the technology, sort of how that works. I'm going to start with Paul. Both the technology part we've talked a little bit about and then when do you need human intervention or a little bit of that shoe-leather work?

  • 12:41:43

    RESNICKSure. I'm going to focus on, how do we know what we know about rumors? And it's tempting and most of what we've talked about here is to focus on the few things that have spread far. But most things that are not true that somebody says actually don't get forwarded. And we don't have a very good way of more systematically finding the things that have only spread a little bit and understanding those. And that's where I think the technology can really help, is to allow us to do more systematic investigations. And then when we do find something that has, you know, spread some, to understand, how far has it spread?

  • 12:42:22

    RESNICKAnd the technology can then -- can again help us to, you know, find all of the tweets that are mentioning this rumor -- or maybe almost all, because we'll never be perfect with our technologies for doing this. But I think that's the place where we can use machine learning and classification algorithms and things like that to try to automatically and systematically find things. And, you know, the truth is that these technologies are very good. But they don't work completely independently. They depend on people to give them training data and, that is, training data being people looking at particular tweets and saying -- yes, this rumor -- this tweet is spreading the rumor, that tweet is correcting it, and this third tweet is completely unrelated to the rumor.

  • 12:43:14

    RESNICKSo we need to have, in the technologies that my students and I have been developing, we need to have some inputs from people. But then we can amplify that with a lot of computing in the background that takes the human input and lets us then systematically find all the rumors -- or all the tweets related to particular rumors.

  • 12:43:38

    BELLANTONICraig Silverman, has a similar process?

  • 12:43:43

    SILVERMANSo one of the things for us is we're very focused on how news organizations are reporting rumors and whether or not they come back and ever correct them or update them. So like Paul said, you know, there's an interplay between human and machine with us. So on one part of it, we're pretty human oriented in terms of spotting rumors. We use things like, you know, searches with Google Alerts and some things using Twitter Searches, just surface things early on. But we're not as systematic as some of the things Paul has built.

  • 12:44:15

    SILVERMANWhere we bring in kind of the automation piece is, once we've identified a rumor and we go out and we look for all the news articles about it. And then once we have that, we enter them into our system. And once a rumor article is entered in our system, two things start to happen. So the first is that we go back and automatically check that same webpage to see if there's ever any new text added -- so an update or a correction or something like that. And if that happens, there's a flag in our system where myself or a research assistant can go in and see, Oh, you know, they have actually updated this rumor. They're now saying it's false.

  • 12:44:48

    SILVERMANThe other thing we do is we look for the social share counts for that particular article and we do it at the same intervals we look for updates. So one of the things that we try to learn in that re-visualizing on the site is, you know, were any articles updated after the rumor became true or false? And what was the trend for sharing? So we're able to actually show that, you know, this phenomenon that in a lot of cases most of the shares happen very early on. And when something gets debunked or confirmed later on, the share counts don't really seem to budge all that much.

  • 12:45:17

    SILVERMANAnd I think that, for now at least, the, you know, the present and the near future is kind of an interplay between human and machine. But one of the really encouraging things that's going on right now -- aside from my project and there's lots out there -- is, one, there are lots of projects of researchers and others looking at trying to, you know, correct things in real time, identify rumors, identify falsehoods. And over time, as there are more and more people that get into this space, you know, things are going to get better. And we're going to start to be able to look at these things at a scale that we just can't do it right now. And that's an encouraging thing that I'm looking forward to seeing how it evolves over the coming years.

  • 12:45:52

    BELLANTONIAngie Holan of Politifact.

  • 12:45:53

    HOLANYeah, we still are using our old-fashioned methods of getting tips from readers. We have an email account. The email is And people forward us chain emails. They make suggestions of things we want to check. We're also on Facebook and Twitter. So most of the things that turn into reports for us are things that readers sent us and said, hey I saw this. I was wondering if this is true. Now, I think the work that our other people just described is really exciting and interesting. And I think we are going to get into more automated sorts of things.

  • 12:46:28

    HOLANBut for us, you know, we're doing some pretty -- I don't know. Some of these claims can get pretty complicated. They're often nuanced. Sometimes they have a lot of different working parts. I think algorithms will help us do our reporting better and more efficiently. But I also think that the craft of journalism is still going to be an ultimate arbiter for a lot of these claims.

  • 12:46:51

    BELLANTONIWell, in a lot of ways, all three of you work together. So I would like to get to a lot of -- we've got a lot of questions about rumors that are coming up. But Cory in Baltimore, Md., has a thought. Cory, thanks for hanging on with us, and you're on the air.

  • 12:47:03

    CORYThank you very much for taking my call. I actually have two thoughts. The first of which is that I see a lot of people on my own News Feed posting articles without even reading them. And I'd like to know what your guests think about that. And the second one is, I was wondering what they think satire sites, such as The Onion, have on people believing or not believing articles that they see.

  • 12:47:30

    BELLANTONIGreat questions, Cory, I'm going to add to that one of my own. There's a suggestion in Canada that there be some sort of media literacy education program to give people these types of critical thinking skills to be able to evaluate what's true, what's false. What do you all think about Cory?

  • 12:47:47

    SILVERMANYou know, so one thing that I'll just add quickly is, he talked about people sharing things they don't read. And there is evidence to suggest that that's very common. That there is in many cases no correlation between things that are getting shared and whether people are reading them. It's kind of this almost ambient awareness of, there's these things out there, I haven't actually read them, but they've crossed my path. So I might re-share it, I might re-tweet it. And, you know, what's really important about that and that we found in our research is, a lot of news articles about rumors will often have a headline that declares them to be true. And then the body text actually kind of walks it back into the little more nuanced.

  • 12:48:22

    SILVERMANAnd so if there are a lot of people who never end up reading the particular article and all they see is the headline in that kind of ambient awareness, I think that they end up having, you know, this kind of -- this belief that it might be true. So the relationship between sharing and reading -- it's true, there's often no correlation between these two things. And just quickly, on satire sites, I think that ones like The Onion and maybe somebody like Andy Borowitz, you know, there are typically disclaimers on there. And I think they're really practicing satire and that they comment on the world around us. And people, in general -- not always, but in general -- know that this is satire.

  • 12:48:58

    SILVERMANWhat I'm particularly concerned about is the emergence of sites like and others of its ilk that have absolutely no disclaimers, that do their best to look like a real news site and that put out things such as that false claim that there was a Texas town under quarantine. And they're doing this simply to get lots of shares and lots of clicks, which they then monetize with basic ads on their site. So they're earning money off of just creating fake things that spread. And when it comes to the Ebola stories that I've seen on National Report, I mean that's just -- that's just a step beyond, where it's actually, in some case, creating an element of public panic.

  • 12:49:32

    SILVERMANSo I think that we have to pay more attention to these fake news sites and figure out how we can flag them appropriately for people on social networks and in other venues.

  • 12:49:40

    BELLANTONISo, you know, that leads me out on a lighter note. We got an email from Dan to kojoshow@wamu -- or, saying, "It seems the most successful false postings are the ones we'd like to be true. So I'm going to believe that, you know, Joe Biden likes to wash his Corvette in the driveway of the White House, which is a common Onion theme there." Angie Holan, did you want to weigh-in?

  • 12:50:00

    HOLANYou know, I think on the topic of satire, we're seeing more and more things that were written satirically that are being passed off as true. And a lot of times, we're seeing stories from satire websites that are cut and pasted into an email and then passed off as true. And then I think that that a lot of people are not that sophisticated about what is satire. I've had friends send me articles from The Onion and ClickHole asking me to fact-check them. And I've, you know, had to say, well, this is a satire website. And, you know, they didn't know that.

  • 12:50:33

    HOLANI don't, you know, I'm not sure where they're -- because you're on the Internet, Facebook and Twitter, like stories come at you and you don't know they're origin. You just see that somebody you know or trust posted it and you don't know. So I think the satire issue is one we're going to see more of.

  • 12:50:49

    BELLANTONIAnd, in fact, maybe 20 people that you know and trust have posted it. So that can get to it as well. Elaine from Lanham, Md., has a question for our experts. Thanks for joining us, Elaine.

  • 12:50:58

    ELAINEHi. Yeah, I got an email yesterday that nobody seems to know about. Nobody I know has gotten it. I've called several offices of several congressmen. I've called some newspapers. Anyway, the email says, "Tell Democratic senators sitting on campaign cash to, you know, help us save the Senate.” The charge is that there are five senators, Harkin, Salazar, Tim Johnson of South Dakota, Max Baucus, and one other, I can't -- Evan Bayh of Indiana. And they said, in total, those five guys are sitting on more than $15 million in campaign cash that they're not using and they will not relinquish.

  • 12:51:47

    BELLANTONIThank you, Elaine. I appreciate that. I might tackle just a bit of that for a moment. Evan Bayh is a former senator. Max Baucus is now the ambassador to China, a former senator. But what is interesting here is there are campaign accounts -- and this is all public information that you can go search yourself with the Federal Election Commission -- where some of them still have money that they collected for their accounts. And they could certainly be sharing that with colleagues.

  • 12:52:08

    BELLANTONIBut that goes both ways. In fact, there's been a lot of great reporting. The Hill posted a story yesterday about Republicans on that site that are sitting on campaign cash that they have not shared. I don't know if that particular email is true, if our experts are familiar with this one. But that is a bit of information you can actually go research yourself and go see, you know, who has what kind of cash on hand in their Federal Election Commission reports. Angie Holan, did you want to add to that?

  • 12:52:32

    HOLANYou know, I was just thinking about the way that people get so much news, I think, especially as Election Day is in another week. When I talk to groups about the fact-checking of elections that we do, I always tell people, you know, voters need to expect to do their homework these days. You can't just sit there and be passive. You have to go out and investigate the issues for yourself. I think that there's a lot of responsibility on voters, on readers, to go and really look closely at the information they're taking in, look at the sources and do their own homework. So I would invite our readers to, you know, be active when you're out there looking for information.

  • 12:53:13

    BELLANTONIWe should also point out in our conversation about satire a moment ago, that of course Facebook is experimenting with this satire label. So Craig Silverman from the Tow Center for Digital Journalism at Columbia University,, and the Poynter Institute, I'm curious, your thoughts on if that will actually help.

  • 12:53:32

    SILVERMANI think it's a great experiment that, you know, things from The Onion or from some of these other sites that Facebook has flagged as being a satire site, when you see those links in your feed, you'll see this little satire label put on it. You know, the truth is, we don't really know much about this experiment with Facebook. We don't know how many sites that they're labeling this way. We don't know how many people they've surfaced this for. But on a most basic level, I think that it is good to flag this kind of content for people. I do see in my own feed, people sharing links from these sites and sharing them in a way where they're presenting them as truth.

  • 12:54:05

    SILVERMANAnd just having that flag automatically placed will give that extra little indication for people that's really important. And I think there's also an argument to be made, you know, Facebook tweaked its algorithm not too long ago to downgrade some of the so-called Clickbait sites so that they weren't showing up as much in people's feeds. Because it had started to really degrade the Facebook experience for some. So now, should they be doing the same for some of the really malicious fake news sites? I think that's something that should be discussed and looked at as well.

  • 12:54:33

    BELLANTONIPaul Resnick, do you have any closing tips for people -- especially a week out from Election Day -- where they can make sure to get good tips and spread good information?

  • 12:54:44

    RESNICKI think the suggestion of skepticism that you're hearing from all of us and from many of the callers as well is a good tip. And, you know, don't be part of the problem. So before you pass something on, do a little checking.

  • 12:55:05

    BELLANTONIThat's trust but verify, right? Thank you very much to Paul Resnick, who's a professor of information in the School of Information at the University of Michigan. Craig Silverman, a fellow at the Tow Center for Digital Journalism at Columbia University, who created and "" at the Poynter Institute, and Angie Holan, the editor of Politifact, which is online. I'm Christina Bellantoni sitting in for Kojo Nnamdi. We will be back in the next hour to talk about the local life and legacy of Edgar Allan Poe.

  • 12:56:01

    BELLANTONIComing up tomorrow on "The Kojo Nnamdi Show," "Station Eleven," a novel nominated for the National Book Award, imagines a deadly flu outbreak and how Shakespeare can keep hope alive. Plus, with the D.C. mayoral election a week away, a new Kojo Nnamdi Show and Washington City Paper poll examines attitudes on pot, education, and Washington's pro-football team. "The Kojo Nnamdi Show," noon till 2:00 tomorrow on WAMU 88.5 and streaming at

Related Links

Topics + Tags


comments powered by Disqus
Most Recent Shows