As It Happened: An Analysis of TikTok Activity in the Wake of the Ban on Calin Georgescu’s 2025 Presidential Run and the Protests that Followed It

by Dan Mercea

‘calingeorgescupresedinte revolutie democratie poporulroman ccr’

There are many ways in which 2024 will be remembered. To political observers, the year stood out for the extraordinary number of elections that took place around the world. In more than 50 countries, more than half of the world’s population got a chance to elect their leaders[1]. Romania was one of those countries. Parliamentary and presidential elections were held there, in November 2024. The presidential elections threw up the biggest surprise. Calin Georgescu, an obscure, independent, far-right candidate—whose best polling numbers in the week of the vote were around 10 percent[2] but mostly stayed well below that figure—zipped past all the other contenders. Following the first round of voting Georgescu ended up the top contender for the highest office of the realm with a share of 22.94% of the votes cast[3].

In the wake of his shock victory, in a frantic scramble to make sense of the results, public attention concentrated on his support on TikTok, the short video platform. TikTok appeared to have been instrumental to his ability to gain momentum among the Romanian electorate[4]. Following intense scrutiny by, among others, Romania intelligence services, allegations were made that TikTok was used in a covert influence operation orchestrated with Russian backing and designed to change the outcome of those elections[5]. As a NATO member and neighbour, Romania supported Ukraine in its fight to stave off the Russian invasion of the country.  In short order, those allegations led to the annulment of the Presidential elections by the Romanian Constitutional Court, before the presidential run-off elections, in early December 2024. The Presidential elections were later rescheduled for May 2025, by the Romanian government[6].

In early 2025, as the country started to prepare for the rescheduled elections, candidates for President again registered their intention to stand. Calin Georgescu was one of them. He registered his candidacy in early March 2025. It was considered by the Romanian Electoral Bureau (Biroul Electoral Central), which decided to reject it on Sunday, 9 March 2025. The decision was followed by violent demonstrations outside the offices of the Electoral Bureau, that same night, and calls to mobilisation in the days that followed[7].

As the TakePart project examines how social actors leverage personal networks on social media to attain collective outcomes, we were interested in whether and how digital actions and interactions on TikTok mapped onto those unfolding events. After a number of test runs, we settled on the term ‘georgescu’ (lower case) as a seed search word we then used to retrieve the communication on the platform following the decision of the Romanian Electoral Bureau, via the TikTok Research API. In late March, we thus gathered data on 2297 videos posted on the platform from Sunday 9 March to Monday 17 March 2025.  Below, we present several initial results through a number of plots that summarize the distribution of videos over that period, and which show how the videos circulated within Romanian and abroad. We do not provide an analysis of the video content itself but illustrate it with excerpts from the descriptions of the videos provided by the authors, in the video titles.

The 2,297 short videos accrued a total of 33,786,931views. They were posted from devices registered in 35 different countries. The first 10 countries were Romania (1,298 videos), Italy (216), the UK (180), Germany (161), France (119), Spain (84), Austria (39), Moldova (32), the Netherlands (26) and Poland (23). The most popular video—a mashup with several candidates for Romanian president, including Calin Georgescu, dancing in a public setting—was viewed 1594836 times (see Figure 1). On average, a video was viewed 14,825 times (median number of views was 707). The videos were accompanied by a total of 22,643 hashtags. The largest number of hashtags for a video was 108, the average was 10 (the median was 7)[8]. The largest number of views in a single day, 7014488, was recorded on 9 March 2025, the day of the Electoral Bureau’s decision.

Figure 1. The number of video views broken down by date.

Next, we used locational, and hashtag data made available through TikTok’s research API to visualise posting patterns aggregated at country level. Following that, we explored the way in which hashtags connected users from different countries. Location is a key variable embedded in TikTok’s proprietary algorithm that is designed to customise the user experience (Boeker & Urman 2022). To determine location, TikTok relies on network information (e.g., one’s IP address, the SIM card region of a user’s phone), as well as locational information users may share voluntarily (e.g., when registering their account or by turning on locational services on their phone)[9]. In our analysis, we used the country code where users registered their accounts.

Figure 2 shows the distribution of the videos by country from Sunday the 9^th to Monday the 17^th of March. Users registered in the top six countries in our dataset—Romania, Italy, the UK, Germany, France and Spain—posted videos throughout this period. At the opposite end, all videos from users registered in Brazil or Norway were posted in two different days; from users in Israel, Guadeloupe, Nigeria or Argentina, in a single day. In most countries, there was a peak in posting activity in the first three days following the decision by the electoral bureau, a time when street protests were in full swing, in Romania[10]. Lastly, some of the locations for registered accounts—e.g., Belarus, Guadeloupe or the Faroe Islands—seemed improbable. We can only surmise that users chose them so as to dissemble their actual location. We cannot say when they made that decision, i.e., how long before they engaged in the posting activity that we recorded.

Figure 2. Distribution of videos per country and day. 

Following on, in Figure 3, we give an overview of the distance in time between the videos posted in each of the 35 countries. Our interest was to see how quickly or otherwise the videos were posted. We converted all times to UTC and calculated the time difference in minutes per day, in relation to the earliest post in a day, in each country. The mean distance in time between posts was 5 minutes (median was 2 and a half minutes, standard deviation was 9 minutes) indicating that half of the posts were made within a relatively short period of time, each day, in all countries (75 percent within approximately 6 minutes from the first post in the day). The longest distance in time between two videos was 158 minutes (over 2.5 hours). These figures may not necessarily point to a coordinated effort, as part of an information operation (cf. Giglietto et al., 2023, who suggest that such operations take place in close to real time), namely, to post videos labelled with the presidential candidate’s name. Nonetheless, they allude to a notable degree of rapid user mobilisation across all countries from which the posts originated. Figure 4 additionally illustrates how hashtags were used in each country. Again, Romania stands out for the largest number of hashtags (108) linked to a post, closely followed by France and then Italy. As the colour coding reveals, the vast majority of the posts were deployed with the first 20 minutes from the earliest post in the day.

Figure 3. Distance in time (minutes) between posts in each country.

Figure 4 The number of hashtags associated with posted videos in each country. Colour gradient represents distance in time. 

Figure 5 displays the connections hashtags helped to create among users registered in the 35 countries. Hashtags have been described as an instrument for grouping ideas or posts together, on social media (Ince et al., 2017). To be clear, by connection, in this case, we mean a relation between two or more countries where users attached the same sets of hashtags to their videos. Figure 5 reveals that the largest collection of hashtags was used in Romania. The country occupied the most central position in the network. Equally, there were countries in the network where only one hashtag was used and solely in that country. Israel is one but not the only such example. Starting with the hashtag ‘georgescu’ and the bigrams [‘calin’, ‘georgescu’], [‘romania, georgescu’], [‘duet, georgescu’][11],  lists of hashtags associated with several videos helped to connect videos posted in Romania with videos posted from devices registered in some of the other most active countries, including Germany, Italy, France or Spain. For example, the hashtag list [‘protest’, ‘romania’, ‘digi’, ‘diaspora’, ‘bec’, ‘9’, ‘03’, ‘stiri’, ‘2025’, ‘masina’, ‘calin’, ‘georgescu’, ‘rasturnata’, ‘galeriile’] connected users in Italy and Romania; the list [‘democratie’, ‘democracy’, ‘georgescu’, ‘romania] connected users in Romania and Sweden.  In that way, those hashtagged videos formed a transnational ecosystem of content linked to the recently disqualified candidate.

Figure 5. Network of videos posted in the 35 countries and the links between them established with hashtags. 

Lastly, we used Bertopic[12], a machine learning technique, to identify patterns in the descriptions of the videos that are technically called topics. With this technique, we selected representative video descriptions from the top five topics found among the 2297 videos. In Tabel 1, below, we list those five representative descriptions[13]. First, they point to the use of very common terms for the platform such as the acronym ‘fyp’ (short for ‘For You Page’). The acronym namechecks the main ‘For You’ feed on TikTok as a means to influence its recommender algorithms and get it to include posts featuring it in users’ personal feeds (Klug et al., 2021). Klug and colleagues have shown that this assumption whereby the acronym and other platform-specific terms and hashtags help push videos up in TikTok’s trending section was unfounded. Nonetheless, the acronym was used in the description of 529 (or 23 percent) of the ‘georgescu’ videos and 585 times as a hashtag (i.e., it was attached to every fourth video). Second, references to popular uprisings—‘protest’ (241 occurrences), ‘revolution’  (23) likewise featured prominently in the video descriptions. Third, there were key campaign terms such as ‘curateniegenerală’ or ‘CG11’ that were previously identified by Romanian media organisations as polysemous phrases used to state one’s support for Calin Georgescu that averted recently introduced penalties by the Romanian government for undeclared campaign advertising[14].  ‘curateniegenerală’ (loosely translated as ‘spring clear out’ in English) was likewise used in relation to earlier protests by supporters of Calin Georgescu, in January 2025. At that point, Georgescu’s supporters rallied against the annulment of the first round of presidential elections, in December 2024. Finally, the video descriptions included references to other Romanian political actors, both allies and rivals of Calin Georgescu’s.

 

Table 1. Representative TikTok Video Descriptions (search term: ‘georgescu’).

‘romania stire romania Calin Georgescu presedinte calingeorgescupresedinte georgesimion parlament Calingeorgescu calingeorgescu2024 cg11 cg cg125 cg160 roman revolutie democratie poporulroman ccr’

‘curateniegenerală Calin georgescu pyシ viral_video cg Calin cg tik_tok anamarie’

‘unicul_lidersingur diană iovanovicișoșoacă diană președinte unicul_lider lider unic singur sos diană dianapresedinte dianaiovanovicisosoaca șoșocar sosromanie sosromanie presedinte presedinteleromanie romania România România româniatiktok Romania Romania Romanian romaniatiktok fyp politic politicaltiktok political politică politic ccr bec bes bucuresti Bucurestiromania Bucurestiromania București alegere alegeriprezidential guvern guvernulromanie parlament senator senat deputat 2025 vot vote votetiktok cg Cgtiktok Calin Georgescu calingeorgescu2024 calingeorgescupresedinte George simion Georgesimion partidulsosromanie partidulaur popor libertate revolutie‘

‘Georgescu presedinte’, ‘Georgescu presedinte’, ‘Georgescu presedinte fyp’

‘romania alegere Georgescu politică protest protest stire anular candidatură ccr bec’, ‘protest georgescu Fyp’

In sum, the intertwined political and protest communication that we observed on TikTok, in the wake of the decision to ban Calin Georgescu from running in the 2025 Romanian presidential elections, spanned 35 countries. The posting activity we witnessed showed some degree of user mobilisation if not coordination, across those countries. That was the case especially in the first seven days of the period following the Electoral Bureau’s decision, on 9 March 2025. The videos accumulated a substantial number of views. They were carefully curated with a combination of hashtags and video descriptions likely intended to influence TikTok’s recommender algorithms so as to garner renewed attention to Calin Georgescu and the protest actions organised in support of him.

Endnotes:

[1] See https://www.kcl.ac.uk/a-guide-to-who-is-voting-and-when-in-this-historic-year-for-democracy.

[2] See https://hotnews.ro/sondaj-verifield-la-comanda-usr-pentru-prezidentiale-lasconi-si-simion-tot-la-egalitate-pentru-finala-candidat-neasteptat-pe-locul-4-inaintea-lui-ciuca-si-geoana-1841578 and https://hotnews.ro/sondaj-verifield-la-comanda-usr-pentru-prezidentiale-lasconi-si-simion-tot-la-egalitate-pentru-finala-candidat-neasteptat-pe-locul-4-inaintea-lui-ciuca-si-geoana-1841578.

[3] https://prezidentiale2024.bec.ro/wp-content/uploads/2024/11/Comunicat_RF1-1.pdf.

[4] See https://www.ft.com/content/47b8b7f0-95d6-4fe0-aec9-ae28e3690d19.

[5] See https://www.ft.com/content/8c6a1900-74db-4551-a3c5-3aa670f4d638.

[6] See https://legislatie.just.ro/Public/DetaliiDocumentAfis/293733.

[7] See https://moldova.europalibera.org/a/romania-candidatura-lui-calin-georgescu-invalidata-de-bec/33341788.html and https://hotnews.ro/jandarmeria-anunta-ca-executa-misiuni-preventive-in-zona-institutiilor-din-capitala-sustinatori-ai-lui-georgescu-stransi-la-ccr-1921555.

[8] Examples of hashtag lists created by users posting the videos include, ‘2025’, ‘alegeri’, ‘politica’, ‘viral_video’, ‘romania’, ‘georgescu’, ‘4u’, ‘romania’, ‘serbia’, ‘fy’, ‘fyp’, ‘bucuresti’, ‘proteste’, ‘tradatori’, ‘coruptie’, ‘georgescu’; ‘500k, ‘viral videos’, ‘romania’, ‘georgescu’, ‘curateniegenerala’, ‘calingeorgescupresedinte💯☑️’

[9] For further information, see https://support.tiktok.com/en/account-and-privacy/account-privacy-settings/location-services-on-tiktok.

[10] Again, see https://hotnews.ro/jandarmeria-anunta-ca-executa-misiuni-preventive-in-zona-institutiilor-din-capitala-sustinatori-ai-lui-georgescu-stransi-la-ccr-1921555.

[11] Along with [‘calin’, ‘presedinte’, ‘georgescu’], the ‘georgescu’ and the other three hashtag bigrams had the highest betweenness centrality in the network, meaning that they were highly instrumental to connecting users with each other, and across different countries.

[12] See https://github.com/MaartenGr/BERTopic.

[13] Using a dictionary of protest-related terms we created to identify Romanian newspaper articles referencing protest events, we identified 285 video descriptions that included one or more protest-related terms.

[14] See https://www.digi24.ro/stiri/actualitate/politica/votam-curatenie-generala-metoda-prin-care-sustinatorii-lui-calin-georgescu-vor-sa-evite-noile-reguli-electorale-impuse-de-guvern-309219.

Academic References:

Boeker, M. and A. Urman (2022). An Empirical Investigation of Personalization Factors on TikTok. Proceedings of the ACM Web Conference 2022. Virtual Event, Lyon, France, Association for Computing Machinery: 2298–2309.

Giglietto, F., et al. (2023). “A Workflow to Detect, Monitor, and Update Lists of Coordinated Social Media Accounts Across Time: The Case of the 2022 Italian Election.” Social Media + Society 9(3): 20563051231196866.

Ince, J., et al. (2017). “The social media response to Black Lives Matter: how Twitter users interact with Black Lives Matter through hashtag use.” Ethnic and Racial Studies 40(11): 1814-1830.

Klug, D., et al. (2021). Trick and Please. A Mixed-Method Study On User Assumptions About the TikTok Algorithm. Proceedings of the 13th ACM Web Science Conference 2021. Virtual Event, United Kingdom, Association for Computing Machinery: 84–92.