Exploiting digital (il)literacy: extreme-right manipulations of the logics of search engines, search practices, and data voids
by Ov Cristian Norocel and Dirk Lewandowski
The ubiquity of digital services (search engines in particular), their quasi-invisibility, and the seeming easiness of using them obfuscates the lack of serious digital literacy among most ordinary users. It opens the door for their manipulation to spread propaganda and disinformation. In several previous entries, we discussed the problems of a single actor (Google, that is) dominating the search engine market, we warned of the negative effects of the poor understanding of Google search for research purposes, and last but not least we mapped out the level of interest in search engines in the Swedish society. In this Medium contribution, we take a closer look at the potential of exploiting the ordinary user’s digital (il)literacy in their search practices. We explore how the logic of search engines, and missing or low-quality data are vulnerable to extreme-right manipulations.
This vulnerability is described in research as a “data void” (Golebiewski & boyd 2018), which is made possible when search engines return low quality content to a specific query because there is hardly any high-quality content corresponding for that specific search. Of interest here are two types of data voids: problematic queries, which concern search results for highly-contested or fraught terms that yield hits only from extreme-right outlets; and strategic new terms, which denote new terminologies specially created at the center of extreme-right information ecosystems before being amplified to reach wider audiences, with the aim of introducing (unaware) ordinary users to problematic content and polarizing frames.
To assess this, and aware that data voids are difficult to identify, we concentrate on the topic of migration in Germany and Sweden. The two countries have been the most generous in the context of the 2015 humanitarian crisis that witnessed hundreds of thousands of people seeking refuge in Europe from war-torn countries. In addition, the extreme-right environments have witnessed an explosive growth in both Germany and Sweden in the context of this crisis.
With this in mind, we drafted a catalogue of queries, on a continuum from mainstream to extreme right vocabularies. We ranked them from innocuous queries containing words/concepts commonly occurring in a casual conversation on the topic at hand, informed by ongoing debates in mainstream media, to queries containing extreme-right “red pills” — words or combinations of words which have been developed as part of a specific extreme-right vocabulary used to describe this topic. To develop this catalogue we turned to the debates in mainstream media on the issue of migration in both Germany and Sweden, as well as the expert reports of Amadeu Antonio Foundation in Germany and Expo in Sweden. Our catalogue sorts queries on three discreetly distinct categories:
Level A queries — contain keywords used in the mainstream media to frame the events. An ordinary (moderate) user in either Germany or Sweden with an interest in the issue of migration, may search for these terms. For example, in German, we chose both the more neutral “Flüchtlingskrise” (refugee crisis) as well as “Masseneinwanderung” (mass migration) denoting a more social-conservative framing; in Swedish, we chose both “flyktingskris” (refugee crisis) and “flyktingsvåg” (refugee wave).
Level B queries — contain keywords that move beyond the mainstream political debate, which an ordinary user may identify as specific to a radical right populist manner of describing the issue of migration and therefore use in their searches as a means to become acquainted with the radical right populist political agenda. For example, in German we chose “Asylschmarotzer” (asylum parasite) and “Asylmissbrauch” (asylum abuse); in Swedish, we chose “asylparasit” (asylum parasite) and “massinvandring” (mass migration).
Level C queries — contain keywords specifically developed within the extreme right environment that aim at polarizing and radicalizing further the attitudes on migration, which are rarely known to an ordinary user. For example, in German, we chose “Rapefugees” (puts together rape and refugees) and “Krimigranten” (fuses crime and migrants); in Swedish, we chose “invandrarvåldtäkt” (merges migrant and rape) and “kulturberikare” (juxtaposes culture and enricher, with derogatory meaning).
We suggest these categories (see Table 1), while being aware that the distinctions between them are rather fluid. This is due to the continuous process of mainstreaming the radical right populist points of view on these matters (Ekman & Krzyżanowski 2021; Norocel, Hellström & Jørgensen 2020), and to the continuous efforts of the extreme-right environment to manipulate the mainstream debate towards polarized and extremist views in general (Askanius 2021; Åkerlund 2021).
To assess the existence of data voids, we designed a series of Google searches, which combine the selected keywords from this catalogue of queries with the names of municipalities in the two countries, such as “Asylschmarotzer Schwetzingen” in German, or “asylparasit Västerås” in Swedish. Our choice is motivated by the fact that search engines, and Google in particular, are trained to prioritize mainstream news content, authoritative sources, and Wikipedia entries. At the same time, search engines continuously fine-tune their algorithms to demote extreme-right content; consequently, ordinary users may not be exposed to extreme-right attempts to manipulate them (Torres & Rogers 2020). Despite this, it seems that searches with a local focus, such as those in which the name of a specific municipality is added, open up for data voids. Indeed, the situation may change, since quite often there is not enough mainstream content to fulfil the parameters of the query.
With this in mind, in Germany, we used a list of all 2,055 German municipalities from the German Federal Statistical Office (Destatis). We cleaned the data by removing identifiers in the official municipality names to obtain a form that a “common person” would use (such as “Frankfurt” instead of “Frankfurt an der Oder”, or “Bernau” instead of “Bernau bei Berlin”). In Sweden, we collected data about all 290 municipalities from the Statistics Sweden (SCB). To query Google using the selected keywords + municipality combinations, we employed software that automatically sends queries to Google and scrapes the results pages (Lewandowski, Sünkler & Yagci 2021). It then stores the found URLs and the position they were shown in a database. This database can then be queried for results to individual queries, but also for domain distributions for particular keywords.
For the purpose of this Medium contribution, we focus only on a handful of these results — our ambition is nonetheless to develop a more nuanced analysis in a forthcoming article. Taking a closer look at these, in the German context, when we query the database for the domain distribution for the queries with the keyword “Flüchtlingskrise” (Level A) vs. “Krimigranten” (Level C) (see Table 2), we notice a different distribution across the combinations keyword + municipality. The top two sources for the “Flüchtlingskrise” + municipality queries are Wikipedia and a mainstream news source (Frankfurter Allgemeine Zeitung). In turn, the top two sources for the “Krimigranten” + municipality query are an extreme right blog and Twitter.
By the same measure, in the Swedish context, when examining the domain distribution for the queries with the keywords “flyktingsvåg” (Level A) vs. “invandrarvåldtäkt” (Level C) (see Table 3), we notice a similarly sharp contrast. The top two sources for the “flyktingsvåg” + municipality query are the Swedish Association of Local Authorities and Regions (Sveriges Kommuner och Regioner) and a small commune in northern Sweden (Malå). In contrast, the top two sources for the “invandrarvåldtäkt” + municipality query are an extreme right news platform (Nya tider) and an RSS (Really Simple Syndication) web feed to the same extreme right news platform.
Zooming in at municipality level, the results become more nuanced. For the German level A query “Flüchtlingskrise” combined with the name of a municipality, we mostly get a mix of mainstream news, local initiatives that focus on helping migrants, and results from (local) government and charitable foundations. For instance, the query “Flüchtlingskrise Schwäbisch Gmünd” returns four results from mainstream news media and one from the Bertelsmann Foundation in the top 5. When one queries Google with a level C keyword and municipality combination, the top results differ considerably. Especially for smaller municipalities, there are often only few results which are dominated by the extreme right. For instance, the query “Krimigranten Schwäbisch Gmünd” returns only five results, which are from a dedicated radical right website (quod-erat-demonstrandum.net), an extreme right blog, a blog aggregator, and the well-known radical right news website “PI News”. However, even when querying Google combining “Krimigranten” with the names of the largest German cities, we find a mixture of result from social media platforms like Twitter, extreme right websites/blogs, mainstream news, and reports from the offices of the protection of the constitution.
In Sweden, the top 5 queries combining Level A keyword “flyktingsvåg” with the name of a smaller municipality (such as Vara), return national mainstream media sources (Svenska dagbladet), as well as web newspapers (Landets fria tidning; Europaportalen), and Swedish institutional repository for research publications and student theses (Diva). In turn, the top 5 queries combining Level C keyword “invandrarvåldtäkt” with the same municipality (Vara) return social media sources (Facebook), extreme right media sources (Nya tider), as well as the Swedish internet forum (Flashback) and question-and-answer site in Swedish (Quora). Like the German case, when querying Google with combination of Level C keywords and larger cities, such as Göteborg, we find a majority of extreme right media sources (Nya tider; Fria tider), as well as a Reddit discussion thread in Swedish.
To conclude, the results of a Google search in Germany and in Sweden depend heavily on the query, and that the query output can vary considerably depending on the local intent of the query. The first aspect is not at all surprising, as results are a response to a query formulated by the user. However, as it could be seen from the source-distribution examples, the way one formulates the query influences heavily the probability of coming across extreme right content. Considering the local intent, our results reveal that a search engine might predominately show extreme right content in cases where mainstream sources have not (yet) produced content related to the query in question. Furthermore, even when results from mainstream sources are available, result lists often consist of a mix of mainstream and extreme right content. Search engines like Google give more weight to source credibility (aka source popularity, see Lewandowski, 2012), leading, among others, to prioritize results from social media platforms like Facebook and Twitter, where content is user-generated and not in any way has to conform to journalistic standards. In our view this opens the possibility for extreme right content in search results via social media platforms. The strong weighting of source popularity in conditions of data void can even lead to radical and extreme right content posted on popular websites rising to the top positions in the ranking.
