Spread the word in the comments, please!

2024-09-30

Péter Magyar often makes public appeals to his followers to spread his messages on unfriendly Facebook pages. How successful was this strategy in the election campaign? Have other messages spread in a similar way through repetitive comments? Were the disseminators real people or pseudo-profiles posing as activists? These are some of the questions we have been trying to answer in our current research.

Summary

  • Peter Magyar regularly asks (and has asked in the period under review) his followers to share his messages in the form of comments on non-friends' Facebook pages. This is a novelty because we did not previously know whether the repetitive comment messages we examined were coordinated and, if so, who might be behind them. The machinery behind most of the repetitive comments remains unclear, but Magyar volunteered to do this publicly, so we were able to investigate the extent to which his operation was successful.
    • At least 400 users have responded to the call that generated the most comments, spreading the message more than 2,000 times on 132 Facebook pages. Among the profiles involved in the distribution, there were few suspicious or obvious fake profiles, their role is not significant and their number is negligible compared to other activists. The spread was therefore largely controlled organically.
    • The second most successful government-critic repeated comment has had a much weaker showing. The 83 identified users commented on the message 603 times, but it still made its way to 95 Facebook pages, where it was used as an argument in the political debates between the commenters. Our analysis suggests that the spread of this text was largely spontaneous and organic.
  • During the campaign period before the June elections, there were no repeated comments in favour of Fidesz that were circulated several times by several users. This type of narrative dissemination did not appear on the government side, neither directed nor spontaneous. However, one enthusiastic and persistent activist achieved a result far superior to the repeated comments critical of the government. His self-penned messages encouraging Fidesz voters on pro-government Facebook pages alone have attracted more than 6,000 comments.
  • The fact that narratives and messages favourable to the government side did not spread en masse indicates that this type of activity is not typical of Fidesz voters active on social media, and does not develop spontaneously. There is no central call or direction for this, so it is not part of the campaign strategy to involve supporters in this way. Fidesz did not really need such a campaign in the spring campaign, as our earlier research revealed that the government side, including its proxies, spent HUF 2.1 billion on Meta and Google ads. This provided a level of reach on social media that would have dwarfed the impact of repetitive comments.

Methodology

Since the 2021 opposition primary, Political Capital has been investigating political campaigns disseminated through so-called repetitive comments. Immediately after the invasion of Ukraine in February 2022, narratives aimed at legitimising Russian aggression began to circulate, followed by narratives containing disinformation.

In this analysis, we zoom in on the campaign activity that is embodied in Facebook comments ahead of the June 2024 elections and is barely visible to the naked eye. Using the SentiOne online monitoring platform, we have collected relevant posts published on Facebook between February and July 2024. All comments that contained at least one of the politically relevant keywords (e.g. Orbán, Gyurcsány, opposition, government, etc.) were included in the initial database. From the resulting large database of more than one million comments, we used programmatic text mining tools to identify repetitive comments, i.e. text of at least four words and repeated at least 200 times. Subsequently, the repetitive comments were subjected to qualitative and quantitative analysis. This was done in order to examine the narratives that were disseminated, the profiles of those involved in the dissemination process, and the strategies that were employed in the dissemination process.

We consider repetitive comments to be problematic when the messages spread organically in a controlled or spontaneous way, contain disinformation or hostile narratives, or when fake profiles play a dominant role in the dissemination.

Opposition messages

1. 5 questions from Péter Magyar to the Prime Minister

The most significant domestic political event of the period was undoubtedly the clemency scandal, the main cause of which was the unforeseen rise of Péter Magyar and the TISZA Party. The new politician has a rapidly expanding fan base, with 374,000 followers on his Facebook page as of 20 September and 175,000 in his official public group. The latter represent a potential activist base that can help get the party's message out into social media bubbles far from politics.

Péter Magyar also brought something new in the field of repetitive comments, the subject of our study. Whereas in our previous analyses there were typically only indirect indications of coordinated distribution, Magyar repeatedly and publicly encouraged his community to distribute his messages by commenting on posts on unfriendly sites. In our analysis, we have processed the most successful of these.

Magyar posted on 15 April 2024 in their official Facebook public group five questions addressed to the Prime Minister and asked his followers to copy the post to the various interfaces of the government website.

As you can see from the graph below, the post immediately turned into a comment and spread like wildfire: it was viewed 1,279 times in the first day. The following day, we also recorded an extremely high figure, 317 repetitive posts, only to see it disappear as usual, except for minor spikes. In total, this message was shared 2029 times.

Daily occurrence of text in Facebook comments 

The repetitive post was widely circulated: it was found as a comment on 132 Facebook pages. However, their distribution was not even, with most comments (314, 15%) were reactions to posts by the official government Facebook page. Another four pages circulated the message more than 100 times: Péter Szijjártó (149), Tamás Menczer (146), Borsonline (136) and Hír TV (101). These five pages altogether accounted for nearly 42% of all comments. It is striking how disciplined users were in following Péter Magyar's call, as almost all the comments appeared under a Facebook page linked to the government side (politicians, influencers or government-controlled media). A negligible proportion of the message was copied under the pages of independent media and opposition figures. The type of comment was predominantly a direct one on a post, only occasionally appearing as a reply to another post (Magyar did not even encourage its use in a discussion with other posters).

The 2029 repetitive comments were shared by at least 396 users. In many cases the post was no longer available, so we were not able to identify the distributor in all cases. This may have happened because the post was deleted or hidden in the meantime, or the user may have been banned. So the number of users involved in the distribution is certainly more than that. Looking at the sharing frequency and activity of the identified users, the vast majority (61%) were one-off activists who had copied the text 1-2 times. A further 31% were identified as avid activists based on our own categorisation, commenting more frequently but less than ten times. 28 avid activists were identified who shared the text between 10 and 39 times. Three users can be identified as lead activists, who copied the questions more than 40 times. We detected 69 posts from the most active user.

The distributors of the text and the Facebook pages where it appeared

Five suspicious profiles were found among the 31 leaders and hard-working activists. Four have no profile or cover photo set up at all, and one uses a profile photo that can be found on Russian social media sites and a cover photo that can be purchased from a photo store. 8% of the comments came from these profiles, which is negligible compared to other activists. The spread was therefore largely managed organically, with minimal help from pseudo-profiles.

During the election campaign, criticisms were voiced that Meta was deliberately manipulating its algorithm to twist up Peter Magyar's posts while suppressing his critics. Our investigation cannot confirm or deny this, only Meta can answer it. However, the dynamics of the spread of the call, independent of the algorithm, and the large number of distributors, rather confirm that the popularity of Peter Magyar's posts, the large number of interactions they received, could have happened organically.

2. The Fidesz balance sheet

The circulation of Péter Magyar's call had already died down when a new, repeated comment appeared in early May. This one, reversing the ruling party's "Only Fidesz" slogan, collected a bouquet of criticisms of the government's actions. Several mutations of the text were encountered, with both shorter and longer versions circulating. In addition to changes in content, there were also variations in form (bulleted, multiple lines).

The temporal progression of the text shows a completely different pattern from the 5 points already examined by Peter Magyar. It spread sporadically for a week after the start and then disappeared completely. Then, as election day approached, the frequency of posts increased spectacularly. The peak was recorded on 4 June, when 90 such posts appeared on Facebook. After the election, the spread quickly died down, but the text did not disappear completely, reappearing on a few occasions, with one user even commenting on a much longer version of the message on 20 July. In total, the message was shared 604 times.

Daily occurrence of text in Facebook comments

We have not been able to find the source of the text, the call for dissemination and the instructions, if any. Comments were mainly posted under the posts of government-related sites, with Fidesz and Tamás Menczer's Facebook page standing out, receiving a quarter of all comments. In total, 95 pages were found to contain text, which in many cases were not even written as direct comments to a post, but in response to comments from users sympathetic to Fidesz, using the list as an argument. So the main aim here may have been to discourage pro-government voters.

Compared to the five points of Peter the Hungarian, there are far fewer deleted comments or banned users; we were able to identify the author of 91% of all comments. At least 83 users were involved in the distribution, two of whom can be described as lead activists, copying the text 54 and 52 times respectively. 14 users participated as diligent activists with 10-39 comments, and most were enthusiastic (29%) or one-off (52%) activists.

The distributors of the text and the Facebook pages where it appeared

An analysis of the profiles of the 16 lead and dedicated activists shows that, with one suspicious exception, they are all real profiles. There are several active disseminators who are clearly identifiable as TISZA supporters based on the public information in their profiles, but there are also some who are drawn to DK and even share content attacking Péter Magyar. In other words, with this collection, which has been shared hundreds of times, opposition sympathisers are making a universal argument against the government side. This heterogeneous distribution camp in relation to the opposition parties also confirms that the dissemination was largely spontaneous and organic.

Overlapping networks of disseminators of the two opposition narratives

There were only four common users in the two distribution networks, so the overlap is minimal. Two of them are dedicated activists of the Fidesz balance text, but they are also enthusiastic distributors of the 5 points of Peter Magyar. One of the latter is a dedicated activist, but has shared the Fidesz balance sheet text only twice. The fourth has distributed both less than ten times. All their profiles are real.

Pro-government messages

There was no repetitive commentary in favour of Fidesz, circulated by several users and distributed many times during this campaign period. This kind of narrative propagation did not appear on the government side, neither directed nor spontaneous.

However, we did detect two pro-government texts that were repeated as comments. One of them was unfavourable towards Klara Dobrev, but of the 1867 comments in our database, none are available today. We conclude that they were all written by a single user whose profile has since been deleted.

The other recurring comment is a pep talk arguing for voting for Fidesz. All of these comments were written by a single, real user. He has already appeared in our previous analyses, and his comments have been in favour of the governing party for years. It is presented below.

The inspiring activist

The characteristic of repetitive text is that it is always "shouting", written in all capital letters, with lots of emoji. The content of the texts is not constant, he posts shorter and longer versions. 

His activism was not limited to a single period. From January until the elections, he commented continuously on a daily basis, except for a few periods. Thus, in a period of a little more than six months, he spread his messages on Facebook in a huge number of 6,417 posts. If we also look at his historical activity from previous years, the number of versions of the text published exceeds 10,000.

Daily occurrence of text in Facebook comments

The text was commented exclusively under pages linked to the governing party. By far the most prominent of these is the Fidesz page with 1749 comments. The comments were posted directly under the posts of the pages, not in response to other comments. So his aim was not to debate or argue, but to get his message across to his own camp. According to publicly available records, the activist's activities were not directed, but were motivated by individual interests and strong political convictions.

List of Facebook pages where versions of the text have been published