-- KamilaKoronska - 17 Apr 2023

Mapping Facebook post-truth spaces in Eastern European countries - Moldova

1. Introduction

The project seeks to develop a method to map regional social media information spaces concerning the Russo-Ukrainian war. It is a multiple country study focusing on Eastern Europe.

In this phase of the project, we are taking up Moldova. Russia’s interests in Moldova are exercised through a large minority of Russian speakers, and the unrecognised, breakaway state of Transnistria in the east of Moldova which March, 2022 was defined as a territory under military occupation by Russia [Globsec: 2021]. Like other post-Soviet states, Moldova may be targeted by Russian information operations, spreading Russian point of view about the war.

The project searches for evidence of Russian influence, through mapping the network and relevance of Facebook entities that spread pro-Russian or pro-Kremlin propaganda.

In this project, we define the post-truth space as an area where epistemic failures are no longer seen as problematic. The post-truth environment is an environment of "alternative knowledge", that becomes an option of choice in opposition to the "mainstream" opinion and realist thinking. We define post-truth to be a space where disinformation is often injected and from where it spreads.

Empirically, we define the post-truth space as one where concerning practices occur, such as doxxing of Ukrainian civilians and soldiers who support Ukraine in the conflict, as well as the use of hate speech, including discriminatory and derogatory language directed towards Ukrainian nationals. Moreover, this space contains content creators who have previously been flagged by reputable, independent fact-checking organisations

2. Research Questions

• Who are the dominant pro-Kremlin voices in Moldova?
• Can we see evidence of pro-Russian narratives entering mainstream-media?

3. Methodology and initial datasets

The method put forward here has multiple goals, beginning with mapping social media information spaces in the countries or languages in question that include terminology framing the war global as well as in Russian terms. (These ‘Russian’ terms may be employed in information operations or in countering them.) In these spaces the methods subsequently seek to locate specific Russian or pro-Russian sub-spaces, including the extent to which it is integrated in other war discourses or isolated.

To map the social media information spaces, we project a bipartite network graph where domain nodes are linked when a page points to both of them. Domain nodes are web domains, Facebook entities (such as Pages or Events), Telegram channels and groups as well as YouTube channels and videos. Contained in the method is a fine-grained URL resolution procedure which takes varieties of the same social media URL and resolves each of them back to its original entity (e.g., a YouTube video resolved from a shortened URL of a YouTube video).

Once mapped in a network graph, the social media information spaces are optimised for clustering using force directed layout algorithms, such as Force Atlas 2 in Gephi, together with community detection. Once the clusters have been identified, each is researched for its main nodes, whether web domains or social media entities.

As a part of the overall analysis of the clusters’ significance, we note the following: (1) cluster distance from the core and (2) the provenance of the ‘bridge’ source nodes (and the extent to which they could be considered to be laundering the information from the problematic cluster). We also examine the leaning of the contents.

In order to determine the extent to which a cluster may be considered a post-truth space, we apply OSINT verification techniques to the URLs in order to determine whether they are routinely identifiable, their content has been flagged or removed (on social media) and/or the sources are ‘verification-evasive’ by which is meant that a further phase of identification does not turn up source provenance.

Corpus keywords: Bucea, Azovstal, Harkov, Uzina Azovstal, Podul Crimeei, Podul lui Putin, Mariupol , Herson Donețk Luhansk, Republica Populară Lugansk, Zaporijjea, razboi + zelenski, razboi + putin Grupul Wagner, refugiati ucraineni, Donbas, Prigojin + Wagner, Shoigu + rusiei, refugiati ucraineni, Bandera, banderovistă, Rusofobie, Anti-Rusismul, Moscoviţi, Basarabeni, Volânia, ucrainizare, ucrainizare + forţată, imperiul rus, naţionalismul ucrainean, Lumea Rusă, denazificare, Spețnaz eroi ruși, laboratoare biologice + ucraina, arme biologice americane, operațiunea militară, actrițe de criză, actori de criză, miliția nazistă ucraineană, Bakhmut, razboi + Ucraina, razboi + rusiei

4. Findings

Qualitative analysis:

For Moldova we have identified a clear case of a pro-Russian cluster (marked in red). Its distance from the core, which is a mainstream media space, is considerable. The bridge actors, or those in the community that are between the core and the problematic cluster, do not appear to be sustaining the narratives. Balanced, mainstream sources (such as articles from Republical.ro) are often seeded in the conversation of pro-Kremlin actors as a "bite", and a way to anger the followers, who most likely will disagree with the proposed mainstream perspective on the events of the war. Nonetheless, post-truth spaces appear on the periphery of the graph, further away from the mainstream conversation about the war, that is led by online version of the Romanian literary magazine “Timpul” , a television channel of TVR in the Republic of Moldova - “TVR Moldova”. Other big media sources such as - “Pro TV” and online news portal “ZUGO” don’t have any modularity class and are outside the graph, due to its self referencing practices on Facebook, however they are still significant sources of information.

On its own, the post-truth cluster comprises roughly 150 entities, such as websites, Facebook profiles, pages, and groups, as well as Telegram accounts. Furthermore, there are numerous links to images and original video content shared as Facebook native media. This diverse array of content, which often expresses a similar ideological agenda, albeit in various languages - primarily Russian (but also Romanian, Spanish, English, and Italian) - enabled us to track down these conversations in Facebook's peripherals.

There are links to 10 open facebook groups, of which 6 actively share pro-Kremlin content in Russian, English and Moldovan (“РУСИЈА У СРЦУ”, “Единомышленники Кагул”,” Help Donbass People”, “ВКС России”, “WorldRusМир”, “Russia 🇷🇺 News 🗞 in English”). 2 ceased to exist and 2 were sharing political rumours (Moldova grăiește!, Societatea Civila) with some traces of pro-Kremlin content shared among the users. The biggest group in this sample - РУСИЈА У СРЦУ (“Russia in the heart”) counts over 90,000 members. Second biggest is WorldRusМир (“WorldRusWorld”) with over 85,000 Facebook users. The type of the content these two groups share are mostly photos (averaging 90% of the time) and videos (appx.7%). Statuses and link posts are less common.

The post-truth cluster includes 17 Facebook pages, among them are: a Facebook page of the Ministry of Defense of the Russian Federation, "Sputnik Moldova," "Russian News Official" in Italian, CGTN, and a page belonging to pro-Russian military analyst Vasilescu Valentin, who contributes to Sputnik.

The most striking of these pages may be that of a Spanish-Russian, self-proclaimed journalist, who allegedly works from Donbass to dispute Western propaganda. "Liu Sivaya" has over 170,000 page followers (on Facebook alone) and covers a range of topics from the reconstruction of Mauropol to attempts at disputing news reports about beheading videos shared by Wagner mercenaries in Ukraine, as well as promoting false narratives about NATO troops fighting in Ukraine. As a result of parroting unsubstantiated claims, such as the one about sabotaging Nord Stream - Liu’s content was flagged in the past by independent fact checkers, such as the Italian’s Open.

Predominantly, the cluster includes individual Facebook profiles (71) that belong to public, personal and private people who have openly expressed their support for Putin’s war, Kremlin or the Russian Orthodox Church. Most of them appear to have affiliations with the Russian state, either ideological or through blood-ties. There are also controversial public figures such as Adrian Severin - a Romanian politician sentenced for bribery and corruption.

In addition to individual pro-Kremlin accounts, the cluster also features numerous Facebook photo posts depicting the Russian president as a powerful, yet compassionate leader with a clear "vision." Interestingly, many of these pictures are old and show a younger Putin. They contribute to normalising his image by portraying him as a religious and occasionally even jovial person. Each picture is accompanied by a commentary, which predominantly reinforces pro-Kremlin viewpoint, such as the one from a woman who writes in both Russian and Romanian about the situation in the Zaporozhian power plant: “Ukrainians are the ones shooting the nuclear power plant. The supply of weapons, transfer of money and intelligence to Zelensky should be stopped. More sanctions? This joker is really trying to drag the whole world down with him. The nuclear plant is under Russian control and yet the Russians are hitting it with missiles. Are we supposed to believe that the Russians are shooting at themselves? Earlier, Ukrainian leader Volodymyr Zelensky accused Russia of bombing the Zaporozhye nuclear power plant.”

The cluster also includes two instances of Telegram accounts, perhaps the most striking is the case of (@TrackAMerc), created for “tracking mercenaries and Nazis” in Ukraine that posts leaked social-media pictures of Ukrainian and foreign soldiers (but also ordinary Ukrainian citizens), in an aim to disclose their identities and location. This is done in “revenge” for their military and other support for Ukraine. Personal information related to people’s families, medical history and profession, whether true or made up, are also leaked in the channel.

The post-truth cluster were found sourcing information from these open groups:

"Всё будет хорошо, Россия Против Коррупции, Владимир Путин, ГРУППА ПОДДЕРЖКИ ПУТИНА В.В., Путин - герой России!, Despre politica, Путин Владимир Владимирович, Мы за президента России-В.В.Путина, МНОГОНАЦИОНАЛЬНАЯ МОЛДОВА, ВЛАДИМИР ПУТИН - МИРОТВОРЕЦ., ПОДДЕРЖКА ПРЕЗИДЕНТА ВЛАДИМИРА ВЛАДИМИРОВИЧА ПУТИНА ИЗ БОЛГАРИИ И МИРА, Из России| с любовью.....Din Rusia cu dragoste..., ЛУЧШИЙ ПРЕЗИДЕНТ ВЛАДИМИР ВЛАДИМИРОВИЧ ПУТИН, Группа поддержки Путина ВВ., Мир Вам, Наш выбор В.В.Путин, Наш лидер - Путин В.В., МОЛДОВА И РОССИЯ, DIASPORA., Всегда лучший, Moldoveni din Torino Italia, Maia Sandu nu e președinta mea! / Майя Санду мне не президент!, Moldovenii contra Sandu, Totul abia începe…🇲🇩, НАША РОДИНА - МОЛДОВА! /PATRIA MEA E MOLDOVA, Stiri din Moldova, Popor dezamagit, Întoarcem Harvardului cucoana! Diaspora contra Sandu !, Vanzari Romania si Moldova, Молдавия Moldova, Accent TV MOLDOVA, Друзья телеканала «ЗВЕЗДА»✌️, Россия и Молдова ВМЕСТЕ, Moldovenii si Românii din Frankfurt am Main, Moldova| Ardealul| Muntenia| (DACIA DORITĂ DE EMINESCU), Рупор правды, Барахолка Молдовы, ЗА ПУТИНА - ЗА РОССИЮ, ПУТИН - ГЕРОЙ НАШЕГО ВРЕМЕНИ! СЛАВА БОГУ| НАМ ЕСТЬ КЕМ ГОРДИТЬСЯ!, CU SAPA, Partidul ȘOR, Новости в Молдове - Ştiri în Moldova, Патриоты Молдавии. Patrioții Moldovei, Из России, с любовью.....Din Rusia cu dragoste..., MOLDOVA-PLAI NATAL, Группа поддержки Путина

OSINT analysis:

We used GATE URL domain analysis service to analyse gathered website and Facebook urls. The service makes calls within connected databases of fact-checks to see whether it gets a match with an address. In our case the following Facebook entities were flagged as having "mixed" intent and appearing in Databse of Known Fakes (DBKF): Группа поддержки Путина ВВ., Ministry of Defense of the Russian Federation, Esanu Ion, Валерий Секриеру, Iovamspus leul, Вести Русија,Руслан Раисович Валиулин, Internacionalistas 36, CGTN, Телеканал Светло Alliance duo, Ana Iordanescu, Ong Assar, Silviu Predoiu, Icaka Yoox, Evpati Kolovrat. Rumble domain was the only one labbled as requiring "caution" and flagged by GDI reports.

6. Discussions

1. Although it lies on the periphery of the discussion around the Russo-Ukrainian conflict, we have observed evidence of foreign influence in Moldova.
2. Photos are by far the most often shared medium on Facebook, followed by owned videos. Including photos and videos urls in our analysis was critical in detecting post-truth spaces.
3. We have observed significant discrepancies between the qualitative and quantitative results. Our qualitative analysis revealed many problematic entities, but GATE did not detect them. Limited availability of multilingual fact-checking databases worldwide makes annotation services an ineffective approach.
4. Through our analysis, we discovered a few sites from Latin America and content created by Spanish speakers, including Liu Sivaya. Given that there is evidence of such content being shared in Moldova, it is possible that Facebook users in Moldova respond to them equally well. The combination of Facebook's auto-translate feature and the concerted efforts made by content creators to overcome language barriers (such as subtitles in videos) helps expedite the spread of false information. Given the scale and speed of spreading disinformation, organisations fighting disinformation have it more difficult to overcome language barriers.

7. References

1. Rogers, R. (2017). “Foundations of Digital Methods: Query Design”. In M. T. Schäfer, & K. van Es (Eds.), The Datafied Society: Studying Culture through Data (pp. 75-94). Amsterdam University Press. Available: https://doi.org/10.5117/9789462981362

2. Rogers, R. et al. (2015). “Issue Mapping for an Ageing Europe”

3. Harsin, Jayson. “Post-Truth and Critical Communication Studies.” Oxford Research Encyclopedia of Communication. 20 Dec. 2018. Available: https://oxfordre.com/communication/view/10.1093/acrefore/9780190228613.001.0001/acrefore-9780190228613-e-757

4. Hannon, M., & de Ridder, J. (Eds.). (2021). The Routledge Handbook of Political Epistemology (1st ed.). Routledge. https://doi.org/10.4324/9780429326769

5. Bruno Latour, (2004) “Why Has Critique Run out of Steam? From Matters of Fact to Matters of Concern”

6. William K. Clifford, “The Ethics of Belief” https://people.brandeis.edu/~teuber/Clifford_ethics.pdf

Data sets (links)