Tech Talk: Post-Election Audits, Disinfo — Talking Tech w/ Kate Starbird, Will Adler, Aliya Bhatia
And we're back! For our first episode of 2023, we're excited to have invited Kate Starbird, Associate Professor in the Department of Human Centered Design & Engineering at the University of Washington, to join CDT's own Will Adler, Senior Technologist for the Elections & Democracy project and Aliya Bhatia, Policy Analyst for the Free Expression project to sit down together. Jamal leads us through a conversation delving into the ways in which misinformation and disinformation can have disastrous effects on election integrity and trust, both domestically and abroad.<br><br>The discussion explores the misconceptions about what disinformation is and whether the term "disinformation" is even a useful term anymore ("rumoring"?), as well as important work going on by coalitions of researchers like the Election Integrity Partnership.<br><br>More on our host, Jamal: bit.ly/cdtjamal<br>More on Kate: https://twitter.com/katestarbird<br>More on Will: https://twitter.com/wtadler<br>More on Aliya: https://twitter.com/AliyaBhatia<br><br>More on the Election Integrity Partnership: https://www.eipartnership.net<br><br>(CDT relies on the generosity of donors like you. If you enjoyed this episode of Tech Talk, you can support it and our work at CDT by going to cdt.org/techtalk. Thank you for putting democracy and individual rights at the center of the digital revolution.)<br><br>Attribution: sounds used from Psykophobia, Taira Komori, BenKoning, Zabuhailo, bloomypetal, guitarguy1985, bmusic92, and offthesky of freesound.org.
Top Keywords
- disinformation 0.021
- election 0.015
- false 0.012
- claims 0.010
- rumors 0.010
- rumoring 0.009
- false claims 0.009
- kate 0.008
- influencers 0.006
- different 0.006
- brazil 0.005
- election disinformation 0.005
Transcript
Welcome to Tech Talk. Bye.
CT. Tea. Welcome to CDT's Tech Talk, where we dish on tech and Internet policy while also explaining what these policies mean to our daily lives. I'm Jamal Magby, and it's time to talk tech. On this episode, we will talk with Kate Starbird, associate professor in the department of human centered design and engineering at the University of Washington, Will Adler, senior technologist for the Election and Democracy Project for CDT, and Alya Bhatia, policy analyst for the Free Expression Project here at CDT, about how mis and disinformation can have disastrous effects on elections, both domestically and abroad. Kate, Will, and Alia, thank you all so much for being here today. So to get us started, Kate, one thing one of the things you research is how election disinformation spreads online. And I want us to get into what that looks like for you. But before we go there, I just like to talk about the term disinformation. People typically say that disinformation must be false information or that it must be spread intentionally by malicious actors. But you said that may be an overly simplistic way to think about disinformation. Can you tell us more about why that is?
Yeah. I mean, even if we look to the definitions of disinformation and when I do that, I look to, the work of Ladislav or Lawrence Bittman, who changed his name when he defected to The United States after being a practitioner of disinformation in, Czechoslovakia back in the day. He wrote some of the only literature that we have on, like, sort of early understandings of particularly Soviet style disinformation. And even as he describes it, he'll he'll he'll describe it as, you know, there's a campaign, and there is some coordination and orchestration, but often that leverages unwitting participants or unwitting agents in its spread. And as we see this sort of modern manifestations of disinformation campaigns, yes, we often have, you know, some intentional shaping of the discourse, but a lot of the people that are spreading false claims and even in some cases generating false claims don't know that they are playing a role in these campaigns. And then though they're motivated, they're often unwitting participants in the generation or spread of of false or misleading claims. Another piece of disinformation is it's not just false claims, it's often misleading, and it's often built around a true or plausible core, but then layered with with distortions and exaggerations to create a a misperception or a a deception.
So should we even call it disinformation anymore? Is is that the correct term? Is it useful?
This is this is a hard one. I I there there's a reason to use disinformation when you can really see the pieces of the intentionality. So, certainly 2016, the Russian disinformation campaign. I've looked at disinformation in the context of Syria, Russian invasion of Ukraine, those kinds of things. With something like election what we're saying election rumors now, election rumoring in the in the in The US right now, 2020, you could really see some of the evidence of the campaign, Donald Trump, and some of his supporters intentionally seeding the space with false and misleading claims. But now it's so the ground is so cultivated. It's the the the the soil is fertile, that it's it's been seeded, and and these rumors grow on their own from everyday people who have just come to see everything about elections as being potentially fraudulent. So they misinterpret what they see and generate these false claims. And it's hard to talk about that anymore as a as a disinformation campaign. So as we moved into 2022 and look further into the election space, we're really thinking about returning to the concept of rumors and rumoring.
That's interesting. And and you touched on a on something I wanted to to to dig a little deeper on. And as you call it, election rumors, can you can you can you just give us a little more information on on what exactly that is and how is it damaging? What what does it do and what effects, does it have on a voter or a likely voter or or someone who's, you know, hesitant about voting?
It's an interesting question. I think, you know, rumors and and rumoring actually can serve a a positive purpose. Rumors are often defined just as as unofficial unofficial accounts or or stories or things that are contested or informal. And sometimes rumors can turn out to be true. And so rumoring actually can can generate a lot of, good content especially in in in at times where we can't trust, authoritative sources. The problem is is that in especially in the election context, as rumoring, you know, becomes persistent and it gets picked up and strategically used as part of disinformation campaigns, it it can undermine trust in in the process and especially when we get false rumors. Right? True rumors are actually kind of helpful. They can help us address actual problems in the rumoring sorry, in the election infrastructure. But, but but false rumors and especially persistent and just pervasive false rumoring about, about election processes and procedures is what I'm thinking about, just about how voting works, how the votes are counted, those kinds of things, that can undermine trust in the in the in the system. And, if people don't trust the system, we we can see, interruptions in the peaceful transfer of power. We can see things like 01/06/2021 here in The US Capitol. We can see things like what just happened in Brazil as well, where people don't trust the results anymore because there's just been so much persistent, false claims about, about the the processes.
Kate, I'm so fascinated by this discussion here and the the question of which terms we use to try to describe the, I guess, like, what we see as the most damaging kind of discourse. And, you know, in doing so, it seems like disinformation as a term has this problem where you're sort of making a judgment about intentionality. And I wonder if you can kind of map this onto the kinds of narratives we saw in 2016 versus 2020, where in 2016, there was so much narrative that was driven by, you know, what was an intentional, campaign by the Russians to kind of manipulate discourse in The US. But, of course, that disinformation kind of develops more into misinformation or rumoring as it kind of gets picked up by, as it got picked up by, Americans. And maybe, like, 2020 was kind of, like, a little bit more on the the rumoring side. I wonder how you kind of see that and how you might compare the two years. And do you think that that's, like, kind of a, a fair way to start distinguishing between the different way these narratives sort of develop?
Yeah. I mean, we could think of the terminology as being descriptive of the best way to to characterize things and and that may cause us to move across terms. But also, it's a it's a way of operationalizing. And for us, rumoring there's a reason to use rumoring as opposed to mis- or disinformation, not just on the descriptive, but also on the operationalizing. And I can explain that a little bit later. But it doesn't quite answer your question. Your your your question is more about, you know, were these phenomenon different phenomena different over the the course of the across these different elections. And certainly, 2016, we see this, like, orchestrated top down foreign disinformation campaign. In in 2020, we see what I would call participatory disinformation where you have this campaign, but it really becomes, you know, integrated with everyday people who aren't just repeating and and amplifying false narratives, they begin to generate themselves. Where the disinformation campaign has really sunk into the ground and begins to change how people experience their own their own voting experiences, right, how they interpret their own voting experiences. And in this case, often misinterpreting their election experiences. For instance, you know, people who thought they were cheated because their Sharpies bled through their ballots and and it wasn't true, but it it was they but they thought they would be cheated. They saw that happen and they and they and they misinterpreted it that way. And so so '20 so 2020 had that sort of like you could still see remnants of the campaign, but it was beginning to take root. 2022, you know, the hands are taken off the wheels. The folks running the the campaigns, except in Arizona, and and you can see that Arizona gets the outsized attention from some of these, false claims about about elections and, and rumoring. Actually, some of it turned out to be true. There was problems with the the infrastructure, but the the problems weren't intentional. So there were pieces of truth and and pieces of of false claims there. But in in 2022, a lot of a lot of that content is just, you know, the ground is the ground is already already fertile as I said. And so, I've described this in other places as, like, the long term goal of a disinformation campaign is to is to is to seed that ground and and and and create that fertile ground so that the the the folks running those campaigns can take their hands off the wheel and it can keep running. We can see it still see remnants of old Soviet disinformation campaigns in some of the content we see about Syria or about Ukraine. Right? Like, this is, you know, I think we make a mistake to think disinformation campaigns are about that time or changing votes in that election. No. Disinformation campaigns are about changing the perceptions of reality for the long for the long term. And so, it's all the same thing. It's just a different state of it. So I think what we see in 2022 is a different state of a disinformation campaign where the the folks that were pushing it have mostly take their taken their hands off the wheel, and it sustains itself with with people that have become unwitting agents in its in its in that in that campaign.
Right. Wow. So, yeah, I mean, these successful campaigns sort of create an idea that just has a life of its own and can self perpetuate. It's it's really fascinating.
Indeed. And and, Kate, I I I wanna I wanna eventually talk talk about your work on the election integrity partnership, but I but I have one question before we move on. What can we do now, now that seeing that we've kind of almost seeded so much ground to to this, to this disinformation? How do we course correct?
I mean, it's really difficult. It it's it becomes similar for to me with, sort of the vaccine context where, you know, initially there were some bad studies, some people pushing something, but now it's so sunk in to to people's mentality to be distrustful of vaccines that they're interpreting their own experiences and their election context, I will say, like, some of this might burn out where people get, you know, bored with seeing these same kinds of claims and they might just there's a second there's a second perspective where, you know, actually, a lot of the right wing influencers who are really involved in 2020 in pushing these weren't so involved in twenty twenty two. Two. So though we see a proliferation of claims coming from the ground up, they don't get as much visibility, they don't get as much pickup from the influencers, and that may actually, depress some of the motivation for sharing those. Right? And except in Arizona. Arizona was a different case. It was connected to a candidate who made that part of her platform, and there it really proliferated. So where we where we continue to see candidates pick up that voter fraud frame, we will continue to see that kind of rumoring and that kind of participation. If the candidates themselves decide that that's not the way they're gonna go, that could that could, you know, help to sort of alleviate the the conversation. And one of the things that we're beginning to see in our data and we think could be really persuasive and we actually see some conservative influencers begin to tune into this is that the conspiracy theorizing and rumoring, especially in the state of Arizona, may have cost them close elections. They encouraged people to vote they encouraged their voters to not vote until election day, to not use mail in ballot balloting or early voting, due to some conspiracy theory about how they were gonna be cheated. And so they were more GOP voters, because of that, were more affected when there were problems on election day. And then they encouraged their voters with another set of rumoring and conspiracy theories not to use the one remedy that would have been easiest for them to use, which was to put their their ballot into a box where it could be counted later by hand, at a different facility. And because they wouldn't do that, they all waited around trying to figure out another remedy and some of them may have been deterred from voting just because it took too long to figure out another remedy. And so you can actually see the rumors and conspiracy theories there causing them, potentially causing them votes in the close election that may have actually mattered. And you begin to see conservative influencers tune into that, that they may be suppressing their own votes with these conspiracy theories and rumors.
Wow. That is a a interesting cycle, you know, to to see kind of the the misinformation and the disinformation kinda come back to to hunt, to hunt the people that that were intentionally spreading it.
Well, sometimes the the influencers might have a motive that's not actually the same as the politicians that they seem to support. Right? Some of these influencers who are spreading these false narratives, you know, they get a lot of eyeballs. They get a lot of marketing revenue, you know, and so on. And, you know, if the the politicians who they, you know, claim to support lose, they might not be the actually the ones, like, feeling the the negative effects of that. So that's a a pretty interesting dynamic potentially.
Indeed. Yeah. I think, you know, helping to highlight where, you know, we've I've said this before, like, undermining trust in the democratic process is gonna backfire on anyone who cares about democracy in the long term. It may help you win something in the short term, but it's gonna backfire in the in the long term. Some people don't care because they have different plans for for the long term and for maintaining their political power, financial status, or whatever. But a lot of folks don't, especially at the local level. They wanna participate in the process. And, if they do begin to catch on that this is undermining their own, what they care about in their in their own participation, that could be a deterrent. I I still think they're more likely to get bored than than to be persuaded, but there's always there's always hope.
So so, Kate, you helped lead the election integrity partnership, a coalition of researchers, election officials, fact checkers, and representatives from social media platforms that track and respond to election disinformation. Can you tell us more about how this partnership works, and what did you all find in the twenty twenty two midterms?
Yeah. It's a it's a good question. I think, it's gonna be hard for me to completely answer here because how it worked in 2020 is different than how it worked in 2022, and it'll be different than how it works next. It's sort of it's it's an academic partnership in some ways. We're trying to learn about how to do rapid response to to rumors, and to miss and disinformation. And so, as we do that, we're we're learning and we're iterating and we're figuring out what we do. In 2020, we were four organizations, four core organizations. It was Stanford, Internet Observatory who actually founded the project. They brought on our team, University of Washington, Center for Informed Public, and we also worked with Graphica, which is a a sort of private company and, the DFR Lab, which is associated with, a nonprofit. And we, those four organizations all have research teams that are just really good at doing analysis of, false claims, disinformation campaigns, and similar in a multitude of con context. And we came together in in 2020 to try to address, what I think we caught talked at the time as misinformation about election processes and procedures. So we were really finely scoped on claims about the how elections work and what the outcomes were. So we left everything else out of scope despite people saying, why didn't you cover this? Why didn't you cover that? Were you involved in this or that? Everything else was out of scope, just just election processes and procedures. And, in that, we had four different categories. One of and this is and this is maintained throughout our our collaboration. The first is just like false claims about when and where to vote, which are actually illegal. Right? You can't mislead people about when or where to vote and then it it, creates, you know, it can suppress it can suppress votes. And so that was our first category. In fact, I think the project originated with that category. We also had false information that, about sort of threats and intimidation, or even true information with threats and intimidation that might suppress votes was was a piece of it. Our our third category was actually content that encouraged fraud. So those those three categories were were actually the original ones. And I think when you dubbed during the project, we really pushed for this fourth category, which was, misleading false misleading or unsubstantiated claims that delegitimize election processes and procedures, things that try to sow trust in that process and procedures, for instance, around mail in ballots or around voting machines or or whatever. And so we had those four categories, and really kept, you know, and what we would try to do is in real time with a huge group of undergraduate students mostly, was to collect, it was trying to find, identify social media posts across a range of platforms that had, that had some sort of false, misleading, unsubstantiated claim that wasn't within our scope, then to to analyze how those were spreading, to get a sense of whether they were going viral. And for a subset of those claims, if they were going viral on a specific platform, in 2020, we would tag those platforms that which would send them a message that that they could see this piece of content if, if we thought it violated one of their policies, and then they could choose to do with with it what they may. In 2022, we actually did not do a lot of sending things to platforms. I think we sent a handful of of tickets, but mostly we focused not on that aspect, but on what I always thought was the more important part of the project was just, like, getting this content out in the world. So in both both the 2020 and 2022, we worked with, we wrote our own blogs, tweet threads. We had rapid, we worked with, journalists. You know, we tried to help them report on stories. We tried to get this content as fast as we could out into the world to help them see what these false claims were, to help correct them, and to help show how they were spreading and who was spreading them. And and in both, in both elections, we, we kept up that work, and it became our primary focus in 2022. It was just sort of this real time reporting on false and misleading claims.
It's been really interesting to see the Election Integrity Partnership sort of act as infrastructure for both the you know, for both, like, disinformation studies, like, if I were were to call it that, but also for sort of how research can be done, how other sort of multi stakeholder partners can work together to do research on social media platforms. So I see, you know, EIP's work has been, like, great to read as, like, someone interested in the research access to data piece, but also in the disinformation piece specifically. I'm curious if you, you know, you mentioned, like, obviously, from 2020 to 2022, there have been changes. And so much of your work sort of has a very strong, like, reflexivity component to it. I'm curious to see, like, what was one example of, like, something that you and your team, like, reflected on. You were like, we have to look at this differently because of the sort of changing nature of narrative or the changing nature of participation.
We this is such a great question. And and it I love you how you set this up, but we you talk about, you know, there was this real time component that was really important. That's why it was set up. But for us at University of Washington, what was more important was to create this really amazing dataset that we could then use for post talk research. So it's hard to in after the fact, figure out what was a rumor, when, and how. But if you can record rumors in the process and record that sort of uncertainty and and and and do this real time collection before the data, sort of disappears in these ephemeral worlds, we got this really not complete, but pretty comprehensive set of all these different false claims that spread in 2020 and again in 2022. And so we've, we've got this great dataset and we've mined it over and over again. We got a lot of the papers are just coming up now. And then we've also made that dataset public and are happy to work with people who wanna analyze the dataset in 2020. The twenty twenty two one is on its way. In between those two elections at at the University of Washington, we spent a lot of time thinking about, how messy the 2020 data was and how much time it took to to really get it to where we could use it for research and tried to redesign the process in 2022 to just do a better job conceptualizing what our unit of analysis was, how to hone in on that unit of analysis, and then craft datasets on social media that kind of represented each rumor or what we call incidents or stories. We call them a lot of different things. And so we've we've spent a lot of time trying to isolate what a claim is, what we call an incident, which is like a discourse around one or or a set of claims that's sort of a cohesive discourse like Sharpiegate. You know, it's got several different claims, but they're all part of one cohesive narrative. This is the claim that Sharpies were bleeding through ballots in Arizona, and that was disenfranchising a certain set of users in 2020. So to think about an incident like Sharpiegate, a set of claims within that, so there's multiple different ways that they conceived of that being some kind of cheating. So kind of like isolating the claims within it. And then there's different posts within a claim. So, you know, there's there's different, like, social media posts. And in in real time, we're capturing the posts, but we've gotta quickly figure out how to isolate them into different claims and then within different incidents in order to kind of put together a dataset that that we can get then go analyze. And so we just spent we spent months, in in the in the 2022. We just spent months trying to figure out what our new units of analysis were and how we were gonna train students to capture the right things, and and not capture redundantly, you know, keep going to get the same things. Okay. We have that claim. Let's move on to the next one. And then we also spent a lot of time, figuring out how to systemize systematize our process in such a way as to, like, as far as we possibly could eliminate bias. And so making sure that we were just gathering everything from the same group of students. So the first in 2020, we used a lot of crowdsourcing. Anyone could send us tips. In 2022, we've we we refocused just on our team and training our team. We had some external partners, but didn't take a lot of external tips. We we really wanted to figure out a way to just use a certain set of keyword terms related to ballots and voting and train students to recognize where something was sort of picking up with a claim that was unsubstantiated, And so that we weren't starting with lists of influencers or certain communities and really trying to to get away from that so we could have some portion of our dataset to say, you know, we collected this with as systematically as we could, and look, there's still, you know, a difference in sort of the political identification of who's participating in this.
Yeah. That's super interesting and and also that tension of, like, doing comprehensive research but also doing quick research.
Yeah. And setting it up. Well, we'd wanna set it up so we can both have insights that can respond rapidly and get things out within, some cases, hours, in some cases, like, a couple of days, for that rapid response phase, but also set up a really nice dataset that we could use later for, you know, much slower research where we spend, you know, two and a half years trying to get out of paper, on some of these different, specific incidences or cases within our within our datasets.
So I wanna jump in here. And, Will, I I think I I think you would know here at CDT, I believe you've been working on a project that looks at groups that monitor and mitigate election disinformation, much like the Election Integrity Partnership. Can you tell us a little bit about that word?
Sure. Yeah. So, you know, in response to the foreign election disinformation that spread online in 2016, platforms really geared up to respond to election disinformation in 2020. And they responded in a bunch of ways. Right? They took down misleading and harmful content. They added labels to other content. They disabled the accounts of some repeat disinformation spreaders or repeat rumors, if we wanna call it that. But there are all these big questions. Right? Like, what are the most impactful narratives? Are the efforts to minimize harmful narratives really working? What does it mean to effectively respond to this kind of disinformation without needlessly affecting the rights to free expression of all users. You know, this is stuff that the platforms are in a position to have real insight to, but they aren't always so forthcoming about it, and they aren't necessarily incentivized to be that forthcoming about it. And that leaves this huge space for groups like, the Election Integrity Partnership to have eyes on the problem and to get it out there to journalists in the way that, that Kate was describing. You know, they can they can try to identify what are the most harmful election related narratives and rumors that seem to be taking hold. How are they spreading? What kind of harm are they causing? What's the right way to respond? I think it's fair to say that the platforms really aren't giving us enough insight into this, which makes it hard to address. It's a little bit like, you know, the early days of COVID where we didn't have enough tests, so we didn't know where the where COVID was spreading, didn't know how to treat it. So, you know, for this research product, we've conducted interviews with folks like Kate who run groups that monitor and mitigate election disinformation of which EIP is one of the most prominent ones. But there are other groups, like, there's a group run by people at Common Cause. There's the disinformation defense league. You know, election officials even have their own little channels of communication about these sorts of narratives. So some of the questions that we're trying to answer with this research are, what are some of the factors that make these groups successful? Do they have sufficient visibility into what's happening online to accomplish their missions? How involved should the government be in these partnerships? That's a question that's kind of flared up a lot, this year and last year. So, you know, there's a lot of open questions like that that we're looking into, and we hope that that'll lead to recommendations that could make these groups work more effectively and build up support for effective groups in the long term. And, you know, the the need for these kinds of groups appears to only be growing. So if you believe various news reports, it looks like Meta, for instance, feels that they've been burned by their efforts, trying to fight election disinformation, and they've been dialing down some of the resources that they've been investing in the space. And Twitter, of course, is another story. So for all its faults, Twitter in 2020 seemed to be making some pretty serious efforts to enforce their, their civic integrity policies in a consistent way. But, you know, last year with Elon Musk taking the helm during the run up to the midterms, things went a little haywire. They started making really ad hoc moderation decisions, basically, at Musk's whim. They gutted the trust and safety team. They disbanded the trust and safety council, which was a way for civil society groups like CDT to give Twitter feedback on their policies. And we, by the way, condemned that move in pretty strong terms last month. And, you know, all that is to say that we appear to be entering a time when the platform seem to be dedicating fewer and fewer resources towards making the platforms a a safe place for all users. And when it comes to making sure that this doesn't affect people's right to vote or affecting trust in democracy, I think that creates an even greater need for outside groups to have better insight into what's happening online. But, you know, it's, of course, not easy, and, researchers like Kate are now becoming characters in this partisan fight about speech online, which is, a really troublesome development. So, you know, Kate, I wonder if you can kind of fill us in on all that, you know, as if it weren't already enough of a a challenge to do effective work in the space. You're now, you know, facing various forms of of harassment and attack, which are themselves driven by disinformation in a way.
Yeah. It's fascinating. After all these years of studying conspiracy theories and disinformation campaigns, I I always knew I didn't know what to tell people to do to stop it once it started happening to them. And you can you can just see, like, oh, there's there's almost no reaction you can have to that's gonna, be beneficial. You kinda feel like whatever you do is gonna backfire. But now being the center of conspiracy theories and disinformation campaigns, it, it is it's a great exercise in, you know, better better understanding the phenomenon from the position of a target and, and the and even more sort of empathy now for the folks that have gone this gone through this. And we're trying to I'm trying to reframe this as a as a learning exercise because it has been a little bit painful in the sense of just the amount of time and energy, that we've been spending in the last, you know, about six months and sort of dealing with all the the false claims and some of the repercussions of those false claims against our our our team at the University of Washington, against the Election Integrity Partnership, and some that that specifically focused on me. I don't even know where to get get started, but it it's a range of things. There's some conspiracy theories about the Election Integrity Partnership, some of them based on just absolutely false claims, some of them based on misinterpretations of our research. I wish I could teach people a little bit of reading comprehension. There's some claim that we had 22,000,000 pieces of content censored. I'm like, are you kidding me? We had a data set that was 22,000,000 tweets large. That's not how many we didn't have and first of all, we have nothing censored. But, but we did report, you know, about, you know, a thousand. I think we reported 1,400 tweet or no. It's 1,400 posts to different platforms. That is not two 22,000,000 pieces of content. So just, yeah, some things are just misrepresented. And then that kind of led to a couple of things. Some some political folks, including the then attorney general of the state of Missouri. I think he's now a a congressperson. He, is is suing some platforms and things, and about, COVID misinformation and other kinds of things or what they call censorship of conservatives over COVID misinformation. I don't know how I'm involved in that because we didn't have anything to do with vaccine misinformation, but or COVID misinformation. But, he's, public records request to us, and it wasn't just from from his his office. We have now 14 public records requests. And every time I say this, I know it spawns people to send more of them. And at this point, there's just a long a long list, and we can only go so fast for them. We're a very small organization. It turns out University of Washington doesn't just have all these people on on hold waiting to help us fill out public records requests, and so, we're going slowly and surely through them all. But, it's a it's a real drain on a small organization. It's a real drain on on in in sort of a chilling factor, potentially for collaborators that wanna work with us because we're public university, and and that does mean that, you know, we, we're we're excited about transparency. We we think it's a good thing, but, we're a little bit worried about the strain on our organization and also how things are gonna be taken out of context, put out in the public sphere, and then used to generate more conspiracy theories. And so, and I don't know how many times we're talking about pizza or something else, it doesn't matter what you're talking about. Once your content's online, people think there must be something there and they start reading into it. So, we're a little bit worried about the negative impact of all of that for sure. Yeah. And do you see echoes in the these kinds of, you know, public records requests that you guys are facing that are kind of, I mean, it sounds like they're sort of hamstringing the ability of your group to do the real work that it needs to be doing.
Do you see some kind of parallels in that, with what was, what election officials were facing last year? You know, election officials seem to be, getting all sorts of frivolous records requests in the run up to 2022. Again, all about the twenty twenty election, which seemed to materially affect their ability to actually run the twenty twenty two election, which, you know, then, of course, leads to potential mistakes and fueling future rumoring. Do you feel like it's part of the same phenomenon that election officials faced?
I I I think it's related. If if you look closely at our our public records requests, I I I I'd be hesitant to say they're all frivolous. There are a few that I think are just people piling on and and are are there sort of the harassment and the timing of them. But there's several of them that people are are are believe some really wild conspiracy theories, and they really think that we somehow had massive amounts of content. Their words, and we got my little quote fingers up, censored, and as part of some massive government scheme, that involved somehow involved involved the Biden administration even though it was in 2020. You could explain that to me because I still haven't been able to explain it. But there are people that are really convinced of that that's true. And so and so for me, it's not that it's frivolous. It's just that it's motivated by by, conspiracy theories and politically motivated weaponization of those conspiracy theories is what I I was that what I feel like we're experiencing.
And I and I think in the we've all seen just how dangerous these conspiracy theories, and distrust in our elections can be, here stateside with January 6 and just recently, the attack in Brazil on The US Capitol, I mean, on the on the Capitol. Will, I I wanna ask you, how do you see do you see and and what type of relationship do you see, happening between these two events?
Yeah. So, you know, many people have pointed out that the attack on Brazil's capital was eerily similar to the attack on DC two years ago, in in many ways. And, you know, a year ago, right in between the two attacks, we teamed up with Brazilian researchers to kind of compare the narratives that had spread about elections among supporters of president Trump in The US and amongst supporters of president Bolsonaro in Brazil. So in both cases, you have presidents who, for years, laid the groundwork to dispute their possible loss in reelection. So in The US, Trump kept saying that mail in ballots were gonna be the way that he was cheated out of reelection. And in Brazil, Bolsonaro kept saying that their electronic voting machines were going to be rigged, and that's how he was going to be cheated out of reelection. And then each of them lost. They didn't concede to their opponents, and all of that sends this potent message to their supporters. Right? People get really invested in in politics. And, you know, it hurts when you believe in a candidate and your candidate loses. But when your candidate doesn't concede, that sends this strong message, you're right to disbelieve the outcome, and you're maybe even right to try and reverse the will of of the people in this democracy. So in The US, that looked like this campaign to, say, force Mike Pence's hand. Or, you know, in Brazil, Bolsonaro supporters have been urging the military to intervene and take over the government, which was this, narrative that that Bolsonaro, helped fuel. And in each case, the president kind of sets the tone from the top, which sort of creates this this scaffolding, this broader narrative, and it allows everyone else in the space to kind of fill it in. So online influencers, create misleading videos that make it look like the machines aren't working. They make money off of those. There are legal challenges which may or may not be frivolous. Those end up raising a lot of money. But, you know, the top influencers, in in this case, the respective presidents, really get the wheels turning and and create this environment where all of these damaging narratives spread. And as far as what the platforms can do about it, it's really complicated. So the January 6 committee had, specific social media investigators who just wrote a piece last week about how hard it is to know what the platforms could have done because there's so little transparency into how they make the decisions, and the data that they use to make decisions. And likewise, the picture is still coming into focus in Brazil. So there's some evidence that the Brazil insurrection was organized in plain view on Facebook and Twitter, but it's still hard to say how top down versus how bottom up, the the movement was. And, you know, another factor in Brazil is that tons of online activity happens on encrypted messaging apps like WhatsApp, which makes it even harder to get a clear picture of what's going on. So, you know, as it did in The US, it might take months or years to get a full picture of how everything happened and what the key factors were. Kate, I wonder how you see the relationship between the two insurrections in terms of online activity. You know, also as a researcher, I'm curious to hear your thoughts on the different challenges of studying, Brazil versus The US.
Such a good question. It is way too quick way too soon for for this researcher to say too much about what happened in Brazil. What I can say is that, some of the members of Bolsonaro's family and and others, in in in Brazil have have been participants in twenty twenty US election false rumors about, about election processes and procedures. And some of the same influencers we saw in The US from Steve Bannon and others are active in the Brazil space. And so we can't even consider those those those events are motivated by the same discourse. Now it was adapted for Brazil versus US elections, but the same deep story about voter fraud and election of fraud was motivated both events and was was created and and stimulated by some of the same people and influencers. And so you can't even I don't even think we can separate those two, the the two campaigns that drove those separate events because it's one one campaign, but just two different parts of it, because it was so intertwined in terms of the influencers and actors. And and I and I think, again, this is early interpretations, but based on just the participation I saw prior to what's happened in the last in the last week or so. But but I I would have said this months ago is that you could see intersections between those two campaigns.
And on your second yeah. Go ahead. Yeah. Just to follow-up on that, it's it's it's sort of frightening how flexible and adaptable this narrative is because, you know, earlier, we're talking about maybe the audience for these claims are just gonna get bored. Right? Well, you know, it it seems like it just it's it might catch on somewhere else if some of the same influencers see, you know, a a a benefit in trying to adapt it and let it take hold in some other democracy?
Yes. I think that would be a great research question. We have a couple of places where we've been measuring it in other contexts. I haven't seen it kind of in a similar way, but, we definitely got a watchful eye for sort of a contagion effect across different democracies. And the the difficulty is is is, rumors can turn out to be true. Elections can have problems, and and there have been cases where where there are issues. And so like, this per persistent and pervasive story of election fraud everywhere based on on false claims might actually obscure, places where things are happening or give cover to people for, you know, interfering in elections or or or hosting elections that aren't free and fair. And so it really the false claims really do mess things up in all directions. It's just like disinformation that the end the end goal isn't necessarily to convince you of something. It's that you no longer believe anything. Well, the problem with pervasive rumors and and disinformation about elections is eventually nobody believes any of the results of democratic outcomes. And that is is is the end of democracy. Right?
That's such a compelling and scary point. The the way that this like, crying disinformation can just become cover for just absolutely corrupt and, you know, very, very motivated attacks. That's definitely something we're seeing with, like, the erosion of trust in media in countries like India and I know in The Philippines as well.
For sure.
So we're talking we're I I noticed we we've shifted to to focusing on elections in in other countries. And that prompts me to to think about, you know, a a project that CDC has been working on where we've delve deeper into looking at disinformation in different language communities. Alia, I wanna ask, how unique or similar is election disinformation shared in languages other than English? Not just in content, but how it's spread and engaged in.
Yeah. Absolutely. I think a lot of the points that Kate and Will and your questions have raised really ring true in when we're looking at election disinformation in different languages. A lot of these sort of false these pieces of false information and misconceptions definitely run parallel to, you know, false narratives in English. So here, we're talking about closely held myths set against the background of, oh, closely held myths within certain language communities and different types of communities, and and all of them, like, set against the backdrop of just, like, distrust, distrust historical distress, distrust in institutions, and real sort of just, like, declines in, you know, quality of life and, like, other factors. But I think one thing that's perhaps unique in, about disinformation or false narratives in languages other than English is that, many times, they sort of operate within data voids. And data void is a term that's coined by Dana Boyd of data and society and Michael Golubuski, where they found that the demand for accurate information or information at all, was much higher than the supply of this accurate information. These data voids exist both in the real world. For example, if you are a Bangladeshi citizen of The United States, from Bangladesh and you're seeking information about where to vote in Bangla, many times this information is not available in your county election website, depending on the proportion of Bangladeshi Bangla speakers in your county. And they also exist online where, you know, there isn't as much media coverage about, you know, voting, where to vote, why to vote, who your candidates are, etcetera, in Bangla compared to English. It's it basically, in, mainstream media. So this sort of creates an opening for a overreliance, perhaps, of information found on either peer to peer networks or in sort of networks that don't have the same high quality accurate information. This also is coupled with the fact that a lot of information or, you know, just just just messaging, is moderated at a different inconsistent level than information moderate on moderated in English. And here, I'm talking about social media platforms. A company you know, we know, and it's been a long, suspected fact that has not been vetted by, you know, transparency reports and companies themselves that these comp big social media companies, sort of employ fewer human resources as well as fewer algorithmic systems in moderating non English language speech. That means there is a sort of, you know, preponderance of perhaps not misleading information or not totally accurate information about elections and other information in languages other than English. And that results in this sort of combination of an overreliance on social and other platforms, and then there's, like, under moderation or inconsistent moderation sort of puts these non English speaking voters at smack dab in the middle of encountering not always the most accurate information. And a lot of people have sort of written extensively about this. So the Asian American disinformation table released a really important report about also the cross platform nature of election disinfo in different languages, specifically in the Asian American community where a lot of the information that is exchanged in these in let's say, on Facebook or on WhatsApp or Telegram, for example, often come from different platforms. So we're seeing that's, you know, one one thing that a lot of researchers are looking at is this cross platform nature of of of disinformation in non English language communities. And another thing we're seeing is also just like this you know, a lot of advocates are now pushing back against just, like, this vast tide of, you know, data collection, data retention that social media platforms have that have enabled, you know, either, influencers or other sort of motivated actors to target specific language mis and disinformation to these language communities they see either as particularly likely to vote or an important voting block against them or someone who lives in a with or, like, exist within a data void to target sort of misleading information. So a lot of actors and researchers in the space are now pushing against or pushing for rather, you know, federal privacy protection to ensure that we're not targeting on these, like, immutable characteristics and, like, targeting on, or limiting the ability to target sort of disinformation in different languages. So in some ways, they're similar, but I think the you know, to go to, like, some of Kate's work, some of the material affordances are, like, almost more, sort of amplified. The material affordances of these tech platforms are perhaps more, amplified in non English language contexts.
Wow. Thanks for that, Alya. Alya, before I I recognize we're at time and Kate, I I wanna be respectful of your time. And and before we end, say thank you again for joining us. And, I think we I speak for everyone, you know, who has been here. We have to have you back, and and and really soon. But, before before we end, I would like to hear final thoughts from from from everyone. And and starting with Kate, I'd I'd like to hear just just kinda final top line thoughts that you have, regarding mis and disinformation in election, security and integrity.
Final thoughts. Gosh. I don't I don't know if I have any final thoughts. I I think in terms of election, rumors, election disinformation, and I'm gonna focus on The United States here. I I think, you know, it really remains to be seen where we go from here. There's a future where this continues. What we saw in in 2020 comes back with a force in 2024. In 2022, is an interim year, which was a down maybe just because it was a midterm. And in 2024, we face the same thing. There's another future where, we where we get ahead of this, where where people, especially folks that maybe have been benefiting from those rumors and and conspiracy theories decide that they're not gonna use those strategies anymore because they begin to realize that it's not in their long term interest, or that it it it backfires. Or maybe it's just the influencers in between decide that that they no longer, wanna risk their reputations on those things. Or maybe some of the lawsuits, like with Dominion Voting Machines come back and and give some some feedback to people that actually they can get in some trouble, spreading those claims when they when they are, defamatory. So there there there's both possible futures. I I think, in our space, we we we think we're gonna learn about that during the primary season. And, I I think we're gonna know, whether or not the candidates that are sort of rising in The US in The US, presidential election are gonna use those or not use those. And if the candidates that are using those claims are are successful or not early on, and that's gonna give us a little bit of a signal. But I I think it could go in a lot of different directions. And I do think I mean, the a little depressing how I spelled it out earlier. There is a future where but these continue to proliferate. They're contagious. They go across the world, and they go across the world, and they really do become a threat to democracy, at at a sort of an existential threat to democracy all across the globe. And there's another future where, like, nope. We just get beyond this, and and we begin to realize that this isn't the way to go. And, I don't I don't know which future we're in we're we're in for.
Let's let's keep our fingers crossed that it it'll be the latter of the two.
Yeah. And I don't wanna take away agency. Like, we all play a role in in how that might that might well, that might play out. And especially folks that have been beneficial or sorry, have been benefiting from from these these strategies and campaigns. The choices that they make in the next couple of years are gonna be determinant for the future
Definitely. Alya, any final thoughts?
I don't have any final thoughts, but I'm happy to see I I think Kate's point about, you know, taking accountability is, like, an important part. You know, I think what we've seen with this participatory nature of sort of false information is that it's often given sort of a license to other people to share sort of, you know, false information or incomplete information they have. So I think changing the narrative and you know? Thank you, Kate, for your work. I think changing that narrative away from that sort of top down victimhood narrative and moving towards a sort of, like, what is all of our roles in this in this ecosystem is a really important point.
I couldn't agree more. Will, rounding us out.
Sure. Yeah. I mean, just to build on what Alia and what Kate said, you know, it's it's gonna take a whole of society approach. We we can't depend on the platforms alone to address this problem. For an entirely different set of reasons, we can't depend on the government alone. There's such an important role for third parties, for researchers at, you know, at the Election Integrity Partnership, and other groups to figure out, you know, what's how how these narratives spread, figure out how to respond rapidly, which is so hard. Alia raised the point of different language communities, which poses, you know, entirely different sets of problems, into trying to figure out what is going on online. And, you know, I just hope that we find ways, and this is something that CDT is is actively working on to just increase transparency, to see what's happening online in a way that still respects people's privacy and rights of free expression so we can really get our heads around the problem and make it so our democracy remains strong.
Well, to Kate, Alia, and Will, thank you all so much for joining us here today on Tech Talks. We really appreciate your time. For anyone that would like to learn more about what we're doing here at CDT, please visit us at cdt.org. And, again, I have been your host, Jamal Magby, and thank you so much for talking tech.
Hi. I'm Cody Venske. I work on the Equity and Civic Technology project here at CDT. Recently, we've been able to petition the FCC to protect student privacy in its efforts to close the homework gap. You can support this and all we do here at CDT by going to cdt.org/techtalk and donating. Every donation matters. Thank you for enhancing civil rights and civil liberties in the digital age.
