Oumou Ly: Welcome to The Breakdown. My name is Oumou. I’m a fel­low in the Assembly Disinformation pro­gram at the Berkman Klein Center for Internet and Society at Harvard. Our top­ic of dis­cus­sion today con­tin­ues the dis­cus­sion of the elec­tion, and this par­tic­u­lar episode we want­ed to talk about domes­tic actors, some of their pat­terns of manip­u­la­tion, the meth­ods that they use, and what their objec­tives are.

I am joined today, and thrilled to be joined today, by Joan Donovan, who is the Research Director of the Shorenstein Center on Media, Politics, and Public pol­i­cy. Dr. Donovan leads the field in exam­in­ing Internet and tech­nol­o­gy stud­ies, online extrem­ism, media manip­u­la­tion, and dis­in­for­ma­tion cam­paigns. Thank you Joan for join­ing us.

Joan Donovan: I’m real­ly excit­ed to talk about this stuff today.

Ly: So our dis­cus­sion today cen­ters on domes­tic actors and their goals in pur­vey­ing dis­in­for­ma­tion, and we would be remiss if not to men­tion that at the time of this record­ing, just last night, the President fired Chris Krebs who is the head of CISA at DHS, the agency with­in the fed­er­al gov­ern­ment that takes the lead on coun­ter­ing, mit­i­gat­ing, and respond­ing to dis­in­for­ma­tion, par­tic­u­lar­ly as it relates to demo­c­ra­t­ic process­es like elec­tions. Joan, what do you sort of make of this late-night fir­ing, this last-minutes development?

Donovan: You know, if you study dis­in­for­ma­tion long enough, you feel like you’re look­ing through a crys­tal ball in some instances. So we all…we all knew it was comin’. Even Krebs had said so much. And that’s because coun­ter­ing dis­in­for­ma­tion is…you know, it’s a real­ly thank­less job. In the sense that you know, it was­n’t just the fact that Krebs had built an agency that over the course of the last few years real­ly had flown under the radar in terms of any kind of par­ti­san divides, had done a lot of work to ensure elec­tion secu­ri­ty, and cared about the ques­tion of dis­in­for­ma­tion, mis­in­for­ma­tion, as it applied to ques­tions about elec­tion integri­ty, right. So CISA and Krebs was­n’t try­ing to dis­pel all of the crazy myths and con­spir­a­cies out there, but they were doing their part with­in their remit to make sure that any kind of the­o­ry about vot­er fraud was some­thing that they took seri­ous­ly and took the time to debunk. 

And so it was­n’t nec­es­sar­i­ly just the kinds of tweets that were com­ing out of CISA but it was real­ly about this web site that they had put togeth­er that was real­ly a kind of low-budget ver­sion of Snopes in the sense that…the web site’s called Rumor Control, and the idea was very sim­ple, which was pro­vide very rapid analy­sis of any con­spir­a­cy or alle­ga­tion of elec­tion fraud that was start­ing to reach a tip­ping point—not every­thing, but things that start­ed to get cov­ered in dif­fer­ent areas, cov­ered by jour­nal­ists, and to give peo­ple an anchor that says, This is what we know to be true at this moment.”

Of course, as the President has come to refute the elec­tion results rather force­ful­ly online, Krebs’ role became much more impor­tant as a vocal crit­ic, with the truth on his side. And over the last few weeks, espe­cial­ly the last week, we’ve seen Trump move any­body out of his way that would either con­tra­dict him in pub­lic or would seri­ous­ly imper­il his desire to stay in the White House.

Ly: That makes me think of some­thing that I think about a lot recent­ly, par­tic­u­lar­ly over the last four years but espe­cial­ly in 2020, is this use of dis­in­for­ma­tion as polit­i­cal strat­e­gy by the GOP. It seems like you know, one pil­lar of that strat­e­gy is just one, to spread dis­in­for­ma­tion. The sec­ond is to sort of lever­age our insti­tu­tions to legit­imize the infor­ma­tion that they’re spread­ing. And the third is just to accel­er­ate truth decay in a man­ner that’s advan­ta­geous to the GOP’s par­tic­u­lar polit­i­cal aims. How do you respond to that, and how do you think the infor­ma­tion ecosys­tem should be orga­niz­ing around that prob­lem that we have a major polit­i­cal par­ty in the United States for whom this is a strat­e­gy for them?

Donovan: They’re real­ly just lever­ag­ing the com­mu­ni­ca­tion oppor­tu­ni­ties in our cur­rent media ecosys­tem to get their mes­sag­ing across. And in this instance, when we know where the pro­pa­gan­da’s com­ing from—that is it’s com­ing from the White House, it’s com­ing from Giuliani, it’s com­ing from Bannon, it’s com­ing from Roger Stone…how then do we reck­on with it, because we actu­al­ly know what it is. So the con­cept of white pro­pa­gan­da’s real­ly impor­tant here because when we know what the source is, we can treat it differently. 

However, the dif­fer­ence between some­thing like what went down in 2016 and what hap­pened in 2020 is an evo­lu­tion of these strate­gies to use some auto­mat­ed tech­niques in order to increase engage­ment on cer­tain posts so that more peo­ple see them, cou­pled with seri­ous, seri­ous mon­ey and influ­ence in order to make dis­in­for­ma­tion trav­el fur­ther and faster.

The third thing about this com­mu­ni­ca­tion strat­e­gy in this moment is that the prob­lem real­ly tran­scends social media at this point, where we do have our more legit­i­mate insti­tu­tions start­ing to bow out and say, You know what, we’re not even going to try to tack­le this, where like for us it’s not even an issue. Because we’re not gonna play into alle­ga­tions that there’s vot­er fraud. We’re not gonna play into any of these pet the­o­ries that’ve emerged about Hammer and Scorecard and Dominion,” and if you’ve heard any of those key­words then you’ve you’ve encoun­tered disinformation.

But, it does go to show that we are immersed in a hyper­par­ti­san media ecosys­tem where the future of jour­nal­ism is at stake, the future of social media is at stake. And right now I’m real­ly wor­ried that the US democ­ra­cy might not sur­vive this moment.

Ly: I com­plete­ly agree with you. And that is a real­ly scary thing to think. Can you talk a lit­tle bit about sites like Parler, Discord, Telegram, Gab. Just recent­ly after the elec­tion, Facebook dis­band­ed a group called Stop the Steal,” and then many of those fol­low­ers found a new home on Parler. Why are sites like this so attrac­tive to peo­ple who have a his­to­ry of cre­at­ing affin­i­ty around con­spir­a­cy theories?

Donovan: So I think about Gab, for instance…me and Brian Friedberg and Becca Lewis wrote about Gab post- the Unite the Right ral­ly. Because Gab real­ly put a lot of ener­gy into recruit­ing white suprema­cists who were being removed from plat­forms for terms of ser­vice vio­la­tions. And they were basi­cal­ly say­ing, We’re the free speech plat­form, and we don’t care what you say.” And for Gab, that…you know, went ass over head pret­ty fast, where they did have to start ban­ning white suprema­cists because unfor­tu­nate­ly what you get when you make a plat­form that empha­sizes lack of mod­er­a­tion, is you get some of the worst kind of pornog­ra­phy you can ever imag­ine. No style, no grace, noth­in’ sexy about it, just…—

Ly: The worst.

Donovan: —here’s a bun­cha peo­ple in dia­pers, right. Like it’s just not good. And so right now, these minor apps that’re say­ing, We’re unmod­er­at­ed, come one come all,” are actu­al­ly fac­ing a pret­ty strong con­tent mod­er­a­tion prob­lem where trolls are now show­ing up pre­tend­ing to be celebri­ties. There’s lots and lots of screen­shots out there where peo­ple think they heard from some celebri­ty on one of these apps and it’s real­ly just a troll with a fake account. 

But this moment is an oppor­tu­ni­ty for these apps to grow. And they will say and do any­thing in order to cap­ture that mar­ket seg­ment. If we think about infra­struc­ture as all three things: the tech­nol­o­gy; the peo­ple that bring the tech­nol­o­gy togeth­er, includ­ing the audi­ences; and the poli­cies, right now we’re hav­ing a cri­sis of sta­bil­i­ty in terms of con­tent mod­er­a­tion poli­cies. And so peo­ple are seek­ing out oth­er plat­forms that increase that kind of sta­bil­i­ty in their mes­sag­ing. Because they want to know why they’re see­ing what they’re see­ing, and they want for those rules to be real­ly clear.

Ly: Picking up on that con­tent mod­er­a­tion thread to talk about larg­er and sort of more lega­cy tech plat­forms more broad­ly, what is your sense of how well con­tent mod­er­a­tion and maybe even more specif­i­cal­ly label­ing efforts work? We saw Twitter and some of the oth­er plat­forms too do a pretty…I think com­par­a­tive­ly good job when you com­pare it with the past, of slap­ping labels on the President’s tweets. But that’s because there was such an expec­ta­tion that there would be pre­ma­ture claims of vic­to­ry. What’s your sense of how well it min­i­mizes virality?

Donovan: Um…so, we don’t real­ly know or have any data to con­clude that the label­ing is real­ly doing any­thing oth­er than aggra­vat­ing people. 

Ly: Yeah.

Donovan: Which is to say that you know, we thought that the label­ing was gonna result in mas­sive reduc­tion in viral­i­ty. In some instances you see influ­encers tak­ing pho­tos or just screen­shots of the labels on their tweets on Twitter— [some dropped audio] —say­ing like, Look, it’s hap­pen­ing to me,” as a kind of badge of honor. 

But, at the same time they do…when done well they con­vey the right kind of mes­sage. Unfortunately I don’t think any of us antic­i­pat­ed the amount of labels that were gonna be need­ed on key pub­lic fig­ures, right. And I imag­ine that you know, okay they’re going to do these labels for folks that have over 100,000 fol­low­ers on Twitter. Or you know, they’re gonna show up on YouTube in ways that deal with both the claims of vot­er fraud but also the viral­i­ty. But it’s hard to say if any­body’s click­ing through on these labels. I’ve clicked through some of them and the infor­ma­tion on the oth­er side of the label is total­ly irrel­e­vant. That is, it’s just not about the tweet or any—it’s not spe­cif­ic enough? 

Ly: Yeah.

Donovan: Which is to say that you know, in watch­ing the tech hear­ing this week, Dorsey seemed to not real­ly be com­mit­ted to a con­tent mod­er­a­tion pol­i­cy that deals with mis­in­for­ma­tion at scale. And as a result, what you get is these half mea­sures that…we don’t real­ly know what their effect is going to be. And for the part­ners in the fact-checking world that part­nered with Facebook, they’re now under a del­uge of alle­ga­tions that they’re some­how par­ti­san and they’ve been weaponized in a bunch of dif­fer­ent ways. And so I don’t even know what the broad pay­out is to risk your rep­u­ta­tion as a news orga­ni­za­tion to do that kind of fact-checking on Facebook, where Facebook isn’t real­ly com­mit­ted to remov­ing cer­tain kinds of misinformation.

Ly: Joan, why is med­ical mis- and dis­in­for­ma­tion dif­fer­ent than oth­er types of infor­ma­tion we see cir­cu­lat­ing maybe relat­ed to elec­tions or oth­er demo­c­ra­t­ic processes?

Donovan: So, when we think about med­ical mis­in­for­ma­tion we’re real­ly think­ing about well how quick­ly are peo­ple going to change their behav­ior, right. If you hear that coro­n­avirus is in the water, you’re gonna stop drink­ing water, right. If you hear that it’s in the air, you’re gonna put a mask on.

Ly: Some of us.

Donovan: And so the way in which peo­ple receive med­ical advice, real­ly it can stop them on a dime and move them in a dif­fer­ent direc­tion. And unfor­tu­nate­ly we’ve entered into this sit­u­a­tion where med­ical advice has been polar­ized in our hyper­par­ti­san media envi­ron­ment. And there’s been some recent stud­ies that can even show the degree to which that polar­iza­tion is hap­pen­ing that is real­ly lead­ing peo­ple to down­play the risks of COVID-19, and this has a lot to do with them encoun­ter­ing mis­in­for­ma­tion from what they might con­sid­er even trust­ed sources. 

And so, when we think about the design of social media in this moment we actu­al­ly have to think about a cura­tion strat­e­gy for the truth. We need access to infor­ma­tion that is time­ly, local, rel­e­vant, and accu­rate. And if we don’t get that kind of infor­ma­tion today, peo­ple are going to con­tin­ue to die because they don’t under­stand what the real risk is, they don’t under­stand how they can pro­tect them­selves, and espe­cial­ly as we enter into this hol­i­day sea­son where a lot of peo­ple are start­ing to relax their vig­i­lance and are hop­ing that it won’t hap­pen to them, that’s the exact moment where we need to crank up the health mes­sag­ing and make sure that peo­ple under­stand the risks and have seen some form of true and cor­rect infor­ma­tion about COVID-19. Because I’ll tell you right now if you go on social media and you start pokin’ around, sure there’s a cou­ple of inter­sti­tials or there’s a cou­ple of ban­ners here and there, but we can do a lot bet­ter to make sure that peo­ple know what COVID-19 is, what the symp­toms are, how to get test­ed, how to keep your­self safe, and how to keep your loved ones safe as well.

Ly: I’m just curi­ous what are the sorts of data points you’ve seen that would explain why some peo­ple don’t nec­es­sar­i­ly sub­scribe to…you know, believe infor­ma­tion from author­i­ta­tive sources sources about the spread of COVID-19, mit­i­ga­tions you can take, not hang­ing out with fam­i­ly mem­bers, and such and such and this. Why do peo­ple… Why are some peo­ple inclined not to believe that author­i­ta­tive information?

Donovan: It’s a good ques­tion. And you know, part of it has to do with the echo cham­bers that they’ve been get­ting infor­ma­tion in for years. We’ve start­ed to see cer­tain Facebook groups that maybe it’s a local Facebook group, and you’ve been in it a long time, and it is about exchanging…like the free list, you know, exchang­ing things in your neighborhood.

And then peo­ple slow­ly start to talk about these real­ly impor­tant issues, and mis­in­for­ma­tion is intro­duced through a blog post, or an arti­cle. Or you know, I saw this on the news’ ” and you find out that they’ve been watch­ing one of these hyper­par­ti­san news sources that is down­play­ing what’s hap­pen­ing. And so you kind of see it in the ephemera. But in our jour­nal the Harvard Kennedy Misinfo Review we’ve pub­lished research around, even with­in the right-wing media ecosys­tem, the degree to which some­one watch­es a lot of let’s say Hannity ver­sus Tucker, they’re gonna have dif­fer­ent asso­ci­a­tions with the risk of COVID-19 because it’s cov­ered dif­fer­ent­ly by these folks that are at the same outlet. 

And so it’s real­ly impor­tant to under­stand that this has to do with the com­mu­ni­ca­tion envi­ron­ment that is designed, and the fact that peo­ple are real­ly try­ing when they’re shar­ing things that are nov­el, or out­ra­geous, or things that might be med­ical­ly incor­rect, they’re doing it in some cas­es out of love. They’re doing it just in case. Maybe you did­n’t see this. And it’s an unfor­tu­nate sit­u­a­tion that we’ve got­ten our­selves into where the more out­ra­geous the con­spir­a­cy the­o­ry, the more out­landish the claim, the more viral it tends to be. And that’s an unfor­tu­nate con­se­quence of the design of these systems.

Ly: Yeah. Thank you so much for join­ing me today Joan. I real­ly enjoyed our conversation.

Donovan: Great. Thank you so much. I real­ly appre­ci­ate you doing these series.

Further Reference

Medium post for this episode, with intro­duc­tion and edit­ed text