In the age of social media and disinformation, journalists, civil society groups, researchers, and media watchdogs in democracies are figuring out how to band together to create a line of defense against those who seek to sow division and doubt in advance of elections.
In the 2020 election cycle in the United States, we saw a variety of groups and coalitions emerge to counter disinformation, from the Election Integrity Partnership to the Disinfo Defense League. This week, a French coalition calling itself the Online Election Integrity Watch Group published a summary report on its activities ahead of this spring’s national election there. The group includes entities such as the Alliance for Securing Democracy, Check First, GEODE, the Institute of Complex Systems, the Institute for Strategic Dialogue, Tracking Exposed, and Reset Tech, an initiative run by the Luminate foundation.
To learn more about what the Watch Group learned in this election cycle, I spoke to the report’s lead authors:
- Iris Boyer, a senior advisor at ISD, and head of ISD France, and
- Théophile Lenoir, a PhD student at the University of Leeds
Below is a lightly edited transcript.
Justin Hendrix :
So we are about two months out from the French election, which I understand took place between the 10th and the 24th of April, where French voters went to the polls and ultimately reelected Emmanuel Macron for another five years. And you are now putting out this report a summary of the report on the French elections, looking at the issues of disinformation and the information ecosystem as it regards the election. Tell me, how did this coalition of groups come together?
We wanted to gather a group of researchers that was able to do some monitoring on disinformation during the elections. This was an initiative that was funded by Reset.tech, which is an initiative inside Luminate, the philanthropic organization from the US. And the idea was really to be able to mobilize civil society to share some methodologies and to do some tracking of content in real time, to be able to alert the media on the one hand and also governments, organizations, and platforms, if anything happened that could put at risk the integrity of the election.
What’s particularly of interest with this initiative is the fact that it’s quite a bottom up approach. And we decided to have it really civil society led. Because we basically realized, and we were not the only ones, I think we are sharing this observation with each and everyone of the members that the French ecosystem was quickly organizing itself and so the French ecosystem was organizing itself, basically professionalizing itself.
But still we were observing a very siloed and segmented approach whereby you could see new government agencies working primarily or exclusively on foreign influence. And then you had a number of academics doing great work, but not necessarily working with one another. And then you had the platforms, obviously, not being necessarily transparent about what their findings were.
So we felt it would be really interesting to have a civic society led approach where we share transparently our results. And primarily with the aim of confronting methodologies. Because it’s also something where the French ecosystem has a long way to go in terms of digital analysis, OSINT techniques or pure academic analysis. We felt it would be really interesting to confront different methodologies.
Justin Hendrix :
So I’m talking to you in a moment where all of the headlines of the United States still regard the 2020 election here, which of course was marred by disinformation, violence ultimately. For particularly my American listeners, can you give a little bit of context about the French election and what the sort of sensibility was going into it? You note in the report, the importance of the rise of anti-system narratives, which to some extent mirrors perhaps what’s happened here in the US, but may look a little different in the French context.
As you said, what was particularly striking and interesting with this campaign, I mean, the first point is that, the very campaign started in a heated context where you had really strong opposition to the sanitary government policy and a lot of heated debates about the extension of the sanitary pass and also sanitary measures.
And this was not the only institutional crisis. Obviously, its followed a series of other institutional crisis like the Yellow vest movements and other social movements about the retirement policies in France, for instance. And what’s really important to understand is that throughout this different and successive crisis, the digitally skilled activist stakeholders have mobilized themselves online and they have basically structured their movements.
So we approach this election in a context where these communities happen to be quite structured and conspiracy theories or radical ideas happen to be much more integrated to these networks. And so in a way, the fear that we had that the threats would necessarily come from abroad, we had to nuance that, and this didn’t happen exactly as such. Probably because anyway, these theories were already really well embedded and especially given the uptake they took during the COVID crisis.
For setting the scene, that’s kind of the specific context when we approached this campaign. And obviously the campaign was kind of segmented across different timings. So first there was a monopolization of the campaign debate around the sanitary policies, and then really the Identitarian stakeholders the Zemmour campaign tend to lead the debate and to impose Identitarian topics. And then we could see definitely a hijacking of the pure traditional political debates in an electoral context, by the specific context of the war in Ukraine, obviously.
So these were the different timings, and we could also say a lot more for your American audience if you want, and Théo would be really well suited to do that about the specific sort of gatekeepers in a French context or the specific framework we have in place around elections and the differences in our media ecosystem that they all started really well.
You note in the report that you regard the French ecosystem as resilient, and I’d love to talk a little bit about what makes up that resilience from your point of view. Here, we’re trying to figure out how perhaps how to manufacture that resilience again.
So what we mean by resilience is the fact that during the election, we did see a lot of problematic content. We did see a lot of growing narratives around the polls, around the legitimacy of the election process, around questioning institutions. But these have remained relatively closed and french communities. And so these narratives haven’t entered very highly visible spaces.
We have a few hypotheses trying to explain why this is the case. The first one, we’ve briefly mentioned it, is the war in Ukraine, which has diverted attention from the campaign. And also as a consequence of the war in Ukraine, you had a ban on Russia Today and Sputnik, which may have had an impact on the ability of foreign actors to interfere with the election. And maybe we can talk about that later.
Then the second hypothesis is really about institutional pressure. France was President of the Council of the European Union at the time of the election on the one hand. And also the government had put in place an agency visioning to track foreign interference during the campaign. And for five years, they passed legislation to better regulate the internet. And so we had a major regulator that was more powerful in 2022 than it was in 2017. And we think that might have, all these combination of institutional factors might have an impact on how platforms have implemented their response to this information.
Then we have a third hypothesis, which is really to do with the electoral system. And maybe that’s something that’s very important to add in the comparison between the US and France. First of all, it’s a multi party election, right? So you don’t have one party opposing one another, you have multiple candidates participating in the campaign. Which I think has obviously some effects on polarization and such things.
But also you don’t have distance voting in France except for some rare occasions. So you don’t have distance voting. And you have a lot of polling places all around France, around 70,000 places, which means a lot of citizens are involved in the process of the election and participate in the places and participating in counting the votes and things like this. So I think that narratives around voter fraud are probably less successful also because of this.
And then one of the last hypotheses is the centrality of the media system, which is what Iris was just saying, in France, the traditional media still have a strong gatekeeping role. I think it’s being challenged, but it’s still there. Which is good news because it prevents disinformation from entering traditional media and highly visible spaces. And it’s also bad news and that’s what worries us. Because this is the criticism of French politics. We have a media ecosystem that’s very institutionalized and very attached to institutional values.
And the issue is that you have a part of the population that doesn’t feel represented by that media ecosystem. So you have a strong center that keeps the conversation going, but at the same time, you have a lot of people outside of that media environment that don’t feel represented
In many ways, stepping back from the report, I suppose we could say that the system stayed on the rails. That’s kind of the bottom line of this report ultimately. But you do talk about, of course, the rise of these anti-system narratives, and then some specific ones, the importation of the quote unquote dominion theory to France, which Americans will recognize as the idea that perhaps the voting machines are rigged, various other rigged election narratives. You’ve already talked about the relationship between COVID and fraud narratives.
And then I was interested as well to see some evidence of the idea of the use of citizen election monitors, to not only kind of collect possible artifacts of voter fraud, but also to then sow doubt using those artifacts. All of that sounds very familiar, in fact, to the playbook that was used here in 2020, the fact that it, of course, didn’t tip over into having a sort of definitive impact in France, I assume that’s good for now, but is there danger for the future?
Definitely. That’s what we believe and that’s what we fear. By no means this report should read as self congratulating. Because if anything, it would just be a reminder we need to be careful for probably a very near future and just to chime in on what Théo was saying earlier about still relatively resilient and relatively centralized media ecosystem in France, it’s being challenged, as you said.
And the problem is, what we can see now is definitely a distortion of the information ecosystem online and a sort of parallel track of information. For instance, in the report, we have a study that’s that shows that when we compare the re information website, so the disinformation website that’s labeled themself re information, the 100 biggest French language conspiracy theory website in 2021 had over 60 million monthly visits, which is as many as the mainstream outlets, Libération and France Bleu. So it just gives you an idea of the scale and the apparatus of this machinery to disseminate highly problematic content and conspiracy theories.
And in the very months of the presidential election, in fact, in April, 2022, the top five websites that were analyzed, managed to rack up over 12 million views. So that’s not nothing. And it’s something we were able to also compare to the uptake in telegram channels or in Facebook groups that showed a pretty similar development.
Perhaps coming back to your very question on the importation of US inspired electoral fraud narratives, what’s really interesting is that we have definitely seen coordinated efforts to amplify the hashtag dominion. When you compare the influence of this campaign to any other hashtag operation in a political context, its impact was relatively small.
And this could partly be explained by what Théo mentioned earlier, which is the fact that in France, we don’t have a remote voting system. So in a way, this theory about the voting machines is not so relevant. So these kinds of observations can tell us a lot about successful and unsuccessful attempts to import foreign narratives. We’ve seen it in the context of the QAnon theory’s importation. Some were complete fails when they were not anchoring themselves into a very French specific context.
But others that really made the effort of adapting the content to French specific contexts were super successful. And in fact, spread really, really quickly during the pandemic. Just to talk about the theories around the idea that the elections would be rigged, this is something that is also relatively minor, definitely very minor when you compare it to the Stop the Steal movements in the US.
But that being said in a French context, the developments are worrying and recent polling showed that 14% of people a few weeks before the vote definitely believed that the election will be rigged. It’s not even conditional. Other polling results actually showed that also 14% of people surveyed believed that if Macron was reelected, the elections would’ve been rigged for sure.
So we are definitely seeing these developments. They are a worry. And we have also observed the overlap between these beliefs, anti-system rhetoric. Some of the communities are leaning more towards an anti-alight kind of rhetoric. And when they do so, they do tend to spread across the ideological and political spectrum. They are not necessarily far right or far left, but they were spreading across. But we also saw a very strong overlap with the anti COVID measures, movements.
And those also had some overlap both with far left anti-globalist narratives and with the opposite of the spectrum with far right Identitarian narratives. So this was quite interesting. And perhaps just to illustrate this worrying trends, even if it’s a very self-contained movement so far, we have seen the development of citizens check or citizens watch movements that basically consisted in motivating people who were already convinced that the elections would be rigged to go and watch and observe the elections and report on potential frauds.
So naturally you could think this is a good idea for the health of the political system. But at the same time, when we did develop our analysis using open source investigation techniques, we realized some very problematic links between antisemitic actors or conspiratorial actors, including people who had been previously convicted or threatening police forces.
The good point is that, and we have to recognize that, in the end, even this small hardcore, so to speak movements, didn’t end up calling out a so-called fraud of the election, and they all accepted the results. Which in fact is a direct mural of what French politicians across the political spectrum have done. Because even the ones who across the campaign have flirted with the idea that the election might be rigged, or the results would be stolen, because they were criticizing, for instance, the “parrainage” system, even those have all accepted the results and the political community was somehow very responsible in the end. (editor’s note: in France, each candidate in the presidential election must receive at least 500 endorsements, or “parrainages, from French elected officials. It would therefore be possible that a candidate ranking very high in the polls, but not backed by other elected officials because of extreme views, could be refused a candidacy. This almost happened to Marine Le Pen.)
This is really a key point in understanding the difference between France and the US. Because French politicians have played the game of the election. They’ve respected the democratic process, and obviously the worry is that they might not do that in the future. And that’s where this is really related to the discussion on the centrality of the media space and of institutional politics. You have politicians in France. But are very respectful of institutions so far, but you see that in other information spaces, you don’t have that respect for institutions and that these narratives are growing. And they’re growing in very structured mobilized communities. And so the risk is that someday, it might become legitimate for a politician actor to say that we should disobey institutions and that the whole process is rigged. And that would be much closer to what happened in the US.
Of course, here in the United States, the presidential candidate and former President, Donald Trump, raised a huge volunteer network of individuals that were meant to fan out and used their phones to collect what they believed was evidence of voter fraud. And that did in fact, filter into media reports and into a kind of broader narrative that was seized on by the President and his advisors. So a kind of very dangerous sensibility if it’s not done in an earnest way.
I want to kind of just focus a little bit on in the report you talk about the idea that institutional pressure increased stakeholder accountability, particularly on the platforms. And of course, you point to regulation, you point to the digital services act, you point to some developments in France, specifically, what did the platforms do differently and how do you relate that to outside pressure?
So this is actually a quite difficult question to answer, what we are saying in the report is that you had a lot of actions being taken from the French government or at the European level, that went in the direction of more accountability from platforms and more actions on their part. So what we are saying is that these made friends are very dangerous ground, especially in the light of the discussion on the Digital Services Act, which is the European text that we regulate platforms that was basically being voted at the time of the French election.
So what we are saying is that this had an impact on platform responsibility during the election. But it’s really a hypothesis because we are still blocked with access to data, to be able to prove to what extent what the platforms have put in place had an impact on the circulation of content.
So the examples of things that they’ve done is labeling problematic content. Making sure that official information from the government was also labeled or inviting people to go and vote with information plans on their newsfeed. And so all of these things happened. They happened elsewhere as well. The question is to what extent they’ve been applied more carefully in France and whether platforms have been more attentive to their effect in France. And we, as I said, it’s difficult to prove and to understand because we still don’t have access to the data on this.
I would definitely agree with that point, it’s a bit of a weak point of the report in a way, but I would say it would be an interesting branch out to focus more on that if we had comparative data. And if we were able to compare data between different elections and what the platforms have put in place. And how they reacted to content being flagged by stakeholders, which we were not able to analyze this time around. And because as Théo said, we had no access to specific data.
What the study shows is that in terms of the highly questionable issue of the algorithmic amplification of content, but also Astroturfing campaigns platforms were definitely not proactive on this topic. Although some of them tend to say that Astroturfing and inauthentic coordinated behavior is not allowed on their platform.
We were able to observe problematic content. So yeah, we would need to dig this further. And that’s exactly why we made this recommendation in the conclusion section. But it’s true that the platform have created new initiatives. Some of them are in partnerships with fact checking organizations and with NGOs. So there are interesting developments, but the impact of these initiatives are yet to be studied.
Just to come back on another point, perhaps, I do think that the French regulatory body, ARCOM, which was previously called CSA, and which used to focus officially only on broadcast media, the fact that this regulatory body had an official new mandate to specifically observe platforms accountability, even if this body is still structuring itself and its work and how they want to apply the information space control to the digital space. I do think this has contributed to putting more pressure on the platforms and this body tends to do some briefings with both the political actors and the platforms. So we can only hope that it might have contributed. There are other external factors like the war in Ukraine and the diversion it has created.
And of course, we’ll have the Digital Services Act coming into effect in the next cycle. So we’ll be able to see if some of those measures and the transparency and independent researcher access that it will mandate may perhaps address some of the data access concerns that you had.
Just one quick word, perhaps, on the foreign influence before we kind of move to wrap up here. In the US, I think some observers have started to sort of dismiss Russian interference in elections as a potentially decisive or very impactful kind of thing. It’s number one in the list that you gave earlier of issues, concerns that seem to have been mitigated. Is there still a sort of sense of the impact of Russian meddling? I don’t know quite how to ask this other than just to say you’ve put a lot of weight on it. You feel it kind of merits that weight, that the fact that the Russians seem to stay out mattered.
So actually what we are saying is, it’s putting the emphasis on the war in Ukraine. And the fact that this as a major event has diverted a lot of attention from discussions on the political campaign. Because there was no heated debate, what we’re saying is that it was probably harder for domestic and international actors to try to play with the narratives and spread disinformation.
Now, a consequence of the war in Ukraine is the ban of Arti and Sputnik. And that’s why we put it in this same line of thought. What the report is saying is we did notice that there were much less content related to Arti and Sputnik related to the election following the ban, but we’re trying to be careful so that it’s not understood that we are saying that Russia is not trying to interfere and to engage in information operations that target France.
What it’s saying later on in the report is that it’s probably shifted the locus of the conflicts. And so an example is in Mali, where this is probably one of the topics where Russia has tried to attack the image of France and spread some narratives around this.
And so what the report is saying is that Russia is still trying to influence and engage in geopolitical actions with information operations, but because of the war in Ukraine and the ban of Russia today, and Sputnik, it’s not been so focused on the French election this time.
I think what’s really important to understand and it’s probably similar in the US to some extent, is that the sort of Kremlin agenda or the Kremlin playbook has been active for years already. And so this playbook has been fully integrated into communities that bowed into Russian state affiliated media, like RT and Sputnik, which actually did a pretty good job at covering highly polarizing events in a French context that mainstream media were not covering in a similar way.
And so it has aggregated a community of anti-government and anti-system people around them. And in a way itself suffice, if I may say, and it can continue to grow and to adapt their narratives more organically and more focused on French context. And what Théo said is very true. The next challenge now is really the very rapid and very aggressive spreading of anti French narratives in French-speaking Africa in a number of countries. And this is something to watch out for, especially in the context where only Europe banned these media outlets, and while these media outlets are pretty active in other regions of the world.
So I want to conclude just with a question about the Online Election Integrity Watch Group, which to some extent, it looks like it was perhaps inspired or informed a little bit by the election integrity partnership here in the US. Which of course was a temporary group that was stood up to kind of monitor and counter very obvious disinformation about the 2020 election.
And I’m interested to some extent in how the groups like this, which are ad hoc, whether these capabilities are going to be formalized, funded perhaps in between the cycles, what happens to them and whether these types of efforts have to become a kind of permanent feature of the defenses that democracies have related to elections? How do you think about that? Are you going to kind of continue? Are you in a hiatus at the moment, or after the publication of this report, or does the group keep going?
What we’ve tried to do here is mobilize a group of actors that is doing daily work on disinformation operations and trying to monitor content. And by the way, I really want to thank the members of the different organizations, Alliance for Securing Democracy, Check First, GEODE, Institut des systèmes complexes, the Institute for Strategic Dialogue (ISD), ISD, Reset, and Tracking Exposed, all of these organizations have really made the report possible.
I think the main challenge that was related to the election is uniting efforts to make sure that we have eyes everywhere. And I don’t know if this is necessary at all times and I think it’s important to have in very important democratic times such as elections, but it’s not necessarily something that has to be ongoing. Obviously that’s my personal opinion. And maybe Iris have a different view on this.
No, I mean, I think overall the way I see it for us it was a test case really. We really meant to demonstrate to different stakeholders in France, including the French government, including platforms, including NGOs who often feel a bit secluded and not enough funded and that they don’t have enough resources or support. We wanted to show them that it’s possible to have a bottom up approach and to confront methodologies and put together some interesting results that show a different side of the story. And by doing so, we wanted to really shed light on the struggles and hustles of this very sector. And particularly the challenges that’s needed including in the upcoming application of the Digital Services Act.
So I think to that respect, we did well, I think. Because we managed to share these results with a variety of stakeholders. I think the problem we raised, which is the sustainability of this kind of setting is an important issue in a French context where it’s really difficult to get government money and where the philanthropic sector is actually underdeveloped.
So the sustainability of this group will definitely depend on how it can sustain itself. Because we can’t maintain such an active group just by doing voluntary work. But I do hope it will hopefully inspire more of the same and that it can also inspire institutional stakeholders to think in more cross-sectoral ways about policies and also initiatives they want to put in place to secure the next elections.
I don’t know if congratulations are in order for having a relatively clean and successful election. But perhaps as an American, I can say congratulations to you all for your role in that. And thank you for speaking with me today.
Thanks a lot.
Thanks for having us.
Justin Hendrix is CEO and Editor of Tech Policy Press, a new nonprofit media venture concerned with the intersection of technology and democracy. Previously, he was Executive Director of NYC Media Lab. He spent over a decade at The Economist in roles including Vice President, Business Development & Innovation. He is an associate research scientist and adjunct professor at NYU Tandon School of Engineering. Opinions expressed here are his own.