Querying the Internet as a mnemonic practice: how search engines mediate four types of past events in Russia

Abstract

In the digital memories literature, the practice of searching for information – one of the most frequent online activities worldwide – has received comparatively little attention. To fill the gap, this exploratory study asks how search engines affect the representations of the past that they produce in query results. Designed as a single revelatory case study, with a focus on Russia, this article delineates a typology of four types of memory events based on four types of websites dominating search results. For each type of event, we discuss recurring locations and mechanisms of power struggles over competing memory narratives. We conclude that within Russia’s authoritarian context, the mnemonic practice of Internet searching tends to reproduce and reinforce the dominant narratives supported by the ruling elites. Search engine companies are thus only one of several powerful institutions that constitute the social framework within which querying the Internet is pursued as a mnemonic practice. Others include mass media, non-governmental organisations (NGOs), and online encyclopaedias.

Keywords

digital memories Google memory politics mnemonic practice Russia search engines Yandex

Introduction

Digital memory scholars have analysed, over the past decade, multiple aspects of the relationship between memory and technology. Among other things, they have scrutinised software programmes created to store personal memories (Van Dijck, 2007) and studied how social networks such as Twitter, Facebook and YouTube are used as memory-sharing platforms and digital archives (Garde-Hansen, 2011; Garde-Hansen et al., 2009; Horsti, 2017; Rutten et al., 2013). Yet, the relationship to memory of the practice of searching for information – one of the most popular activities pursued online by users worldwide (Hargittai, 2007; Jiang, 2014) – has received comparatively little attention (for studies that touch on this question in passing, consider Jakubowicz, 2009; Reading, 2011; Van Dijck, 2007). Given this lack of previous writing on the topic, the overarching exploratory research question of this study can be formulated as follows: how do search engines affect the representations of the past they produce in query results and can any systematic patterns be identified with regard to that?

To answer this question, we adopt a single ‘revelatory case study’ approach (Yin, 2014: 51). The selected case is that of searching for past events of national political relevance in the socio-political context of Russia. This was chosen for three main reasons. First, Russia is one of the few countries where two search engines, Google and Yandex, continuously compete for market leadership. In April 2016, Google had a monthly audience of 20.5 million people and Yandex 20.4 million, in a country of roughly 146.5 million population (Boletskaya, 2016). Second, one of these search engines (Google) is owned by a foreign company, while the other (Yandex) is local. Within Russia’s authoritarian regime (Toepfl, 2016), this implies that the country’s political leadership may put pressure on Yandex to a degree that is uncommon in democratic contexts (Parfentyeva et al., 2016). A comparison between the functioning of Google and Yandex may thus serve to reveal potential channels and consequences of governmental influence. Third, Russia is a major player on the international stage. Thus, identifying how search engines relate to the memory politics in this country can be regarded as of high political relevance.

To scrutinise our case, we adopted a qualitative approach broadly following the logic of ‘grounded theory’ (Lindlof and Taylor, 2011: 250). For data collection, we relied on comparing query results across search engines (‘information retrieval research’, Jiang, 2014: 215). Specifically, we entered a variety of ‘purposefully sampled’ (Lindlof and Taylor, 2011: 110) search terms related to past events (for details, consider section ‘Methods of data collection and analysis’, below) into the search bars of Google and Yandex and systematically analysed the first 10 websites obtained as results. Pursuing this approach, in the process of grounded theory development, four types of memory events emerged, with regard to which the two search engines, operating within the Russian authoritarian context, systematically produced different results. We termed these events: (1) non-actualised memory events, (2) actualised memory events, (3) counterpublic memory events and (4) emerging memory events. For each of these four event types, search results showed a particular combination of four types of websites: (1) pro-regime mass media websites, (2) encyclopaedic websites, (3) personal websites and (4) independent media and non-governmental organisation (NGO) websites. For instance, we found that in the case of actualised memory events (i.e. events that are highly present in the dominant media discourse and thus important to the country’s memory politics), major pro-regime mass media and, to a lesser extent, large encyclopaedias dominated the search results of both Yandex and Google. This implies that with regard to actualised memory events, the power struggle over memory narratives obtained through search engines is located primarily in controlling the country’s media landscape.

To add nuance to this and develop related lines of argument, the remainder of this article is structured as follows. First, we review the existing writing on the mutual influence of memory and technology, with a focus on search engines. Then we provide some context on the functioning of search engines and the politicisation of the past within Russia’s regime and specify three research questions. The subsequent section details methods of data collection and analysis. Next, we present our typology, which relates four types of memory events to four types of websites. In the concluding section, we discuss our findings against the backdrop of the literature on the power of algorithms and digital memories. We argue that the mnemonic practice of Internet searching tends to reproduce and reinforce narratives about the past supported by the ruling elites. Companies such as Google and Yandex, and the search algorithms they control, must be analysed as only one of powerful institutions that form the social framework within which the digital mnemonic practice of Internet searching is pursued. Our analysis shows that the representations of the past accessed through searching the Internet are also affected by other institutions, most importantly news media, NGOs and social movement organisations, and online encyclopaedias.

Digital memory studies: searching the Internet as a novel mnemonic practice

By the 2000s, it had become apparent that digitisation and the advent of new media were significantly changing our interaction with the past. As early as the late 1990s, Olick and Robbins (1998) observed that ‘electronic means of recording and transmitting information … not only change the way we remember, but provide new ways of conceptualizing memory’ (p. 115). These and other scholars’ interest in the topic signified the formation of a new research field, now often called ‘digital memory studies’ (Garde-Hansen et al., 2009). While the view that new media have brought about a wholesale transformation of the way we remember and recollect has been challenged (Keightley and Pickering, 2014; Keightley and Schlesinger, 2014), researchers broadly agree that technology has changed some of the old ‘mnemonic practices’ and created new ones (Keightley and Pickering, 2014; Olick, 2008; Olick and Robbins, 1998). Studying mnemonic practices implies, as Olick (2008) has emphasised, taking on a process perspective, that is, analysing collective memory as ‘something – or rather many things – we do’, and not as something ‘we have’ (p. 159, emphasis in original). Following this logic, we broadly understand mnemonic practices as any activities performed by humans for and while engaging with the past. Events of the past mediated by online sources and actualised by an individual through the mnemonic practice of Internet searching are here called ‘memory events’. It is important to emphasise that whereas our usage of the term ‘memory event’ is rooted in the concept developed by Etkind and his colleagues (Blacker et al., 2013), who define memory events as ‘acts of revisiting the past that create ruptures with its established cultural meanings’ (Etkind and Uffelmann, 2014), we nevertheless broadly follow Fredheim (2013) in ‘shifting the focus from the event itself to its representation in media discourse’ (p. 10).

Scholars have analysed as mnemonic practices an array of activities, including television viewing (Bourdon and Kligler-Vilenchik, 2011) and the taking of analogue and digital photographs (Keightley and Pickering, 2014). However, extant literature touches only in passing on the mnemonic practice of Internet searching, with scholars typically not grounding their claims in systematically collected empirical evidence. Jakubowicz (2009), for instance, notes in his case study of Shanghai’s Jewish history that major search engines privilege websites of higher quality, ‘as qualified by government, education or research criteria’ (p. 99), and strive to balance this approach with commercial incentives. Similarly, Reading (2011: 250) supplements a broader line of argument in her analysis with an illustrative example, referring to the outcome of a single search query she conducted in 2009 using the term ‘Neda’ (referring to Neda Agha Soltan, an Iranian woman who was shot dead on 22 June 2009, during street protests in Tehran). As Reading (2011) reports, the search results privileged Wikipedia, Facebook and mainstream news organisations, rather than citizen media. In similar cursory ways, other scholars have pointed to the highly consequential – and potentially dangerous – archival power of search engines (Garde-Hansen, 2011: 72; Garde-Hansen et al., 2009: 188). However, while undeniably illustrating the existence of a relationship between search engines and memory, these studies mention the topic only briefly, conducting no systematic empirical analysis of, or theoretical reflection on, how memories of social groups may be affected by the practice of Internet searching. Here is precisely the gap in extant research that this study seeks to fill.

Search engines as media: the power of algorithms

While digital memory scholars have paid relatively little attention to the mnemonic practice of Internet searching, a vast body of research on search engines has emerged in the discipline of communication (for overviews, consider Jiang, 2014; Zimmer, 2010). This literature has not investigated how search engines affect collective recollection, but it has scrutinised how they function as media and contribute to the Internet’s ‘political life in a large sense, both as a means of finding information and as a space for expression and deliberation’ (Rieder and Sire, 2014: 196). Often at the centre of researchers’ attention have been issues of algorithmic power, ideology and bias (Amoore and Piotukh, 2016; Beer, 2017; Introna and Nissenbaum, 2000; Pasquale, 2015).

Within this literature, it is frequently lamented that while the technical side of the search process has received a lot of attention from scholars, its social and cultural dimensions are relatively understudied (Graham et al., 2013; Hargittai, 2007). There are, however, some notable examples of the latter type of research. For instance, the contributions to Spink and Zimmer (2008) explore values embedded in search engines and their influence on equality, diversity and democracy. Hellsten et al. (2006) scrutinise how search engines influence our conceptions of time: they empirically demonstrate that by allowing users to search for information for specific periods of time and, alongside this, regularly updating their indices, search engines rewrite the past and create ‘multiple presents’ (p. 901). Finally, Jiang (2014) compares results of searches on China’s two most popular search engines, Baidu and Google, for 316 popular Internet events. Through a quantitative analysis, her study identifies a ‘low overlap [of 6.8% of links appearing in the first ten search results] and little ranking similarity between Baidu’s and Google’s results, implying different search engines, different results and different social realities’ (Jiang, 2014: 212).

The functioning of search engines and the politicisation of the past in Russia

By comparison with the more closed regime of China (Jiang, 2014), political pressure on search engines to filter search results in Russia was significantly lower at the time of our research between September 2016 and February 2017. Unlike in China, for instance, search engines were not required to filter from their search results information threatening the country’s ‘unity and stability’ (Jiang, 2014: 215). Yet, Russia’s ruling elites could be assumed to have significant informal influence on Yandex: a ‘golden share’, which allows its holder to veto any sale of more than 25% of the company, was held by Russia’s Sberbank, itself controlled by the state (Bidder, 2012). Such informal influence manifested itself repeatedly. For instance, in 2014, in response to Vladimir Putin’s criticism of Yandex’s ‘digital aggression’ with regard to the war in Ukraine, the company altered its top news aggregator algorithm, which, according to the Russian business daily Vedomosti, made it possible to filter out ‘undesirable news’ (Ruvinskiy, 2017). In September 2017, on the occasion of the 20th anniversary of Yandex, Vladimir Putin visited the company’s Moscow headquarters, which was widely interpreted as a sign of Yandex finally abandoning its reputation (with the ruling elites) of ‘an agent of the West’ and becoming ‘a helper of the Kremlin’ (Ruvinskiy, 2017). Moreover, legal pressure on the company was increasing at the time of research. Legislation adopted in 2016 (an update to the federal law ‘On Information, Information Technologies and Data Security’) significantly limited the appearance in the Yandex News section of mass media with no official Russian registration. Since many news media have operated without such a registration, especially smaller independent media and foreign news giants, the law was widely expected to complicate readers’ access to government-critical information sources (Parfentyeva et al., 2016). According to Daucé (2017), the political conflict between Yandex and the Russian authorities, which manifested itself in the latter’s attempts to ‘regulate the code’ (p. 112) of the former and culminated in the adoption of the new legislation, ‘demonstrates the complicated process of the political appropriation of [Yandex’s] algorithm’ (p. 127).

Behind the Russian government’s rationale to put pressure on search engines may be, among other things, the growing politicisation of the past in the country. Since coming to power in 2000, Putin has attempted to overcome the division of Russian society over its history. To create a consolidating narrative, the Kremlin started to borrow from ‘mutually exclusive understandings of Russian history’ (Gjerde, 2015: 149). These attempts at reaching a societal consensus have had, according to Gjerde (2015), the opposite effect: they have led to the perpetuation and amplification of ‘the very division’ that the authorities wished to overcome (p. 149). Miller (2012) describes these efforts of the Kremlin as a ‘historical politics’, or memory politics, that in the context of Eastern European non-democratic states amounts to the usage of the past for political purposes, that is, as a tool to mobilise the population around the leader and strengthen the regime. In this regard, Kalinin (Blacker et al., 2013) talks about Putin’s ‘resource-centered politics of memory’ aimed at ‘naturalising the past’ (pp. 255, 260).

Developing research questions

Having familiarised ourselves with the two strands of literature reviewed above, and with how the two search engines operate in Russia, we embarked on the process of grounded theory development with three questions in mind. As Jiang (2014) argues, search engines ‘can be architecturally altered to serve political regimes’ and may be considered, to a certain degree, ‘arbitrary in rendering social realities’ (p. 212; for similar arguments in the memory studies literature, see Jakubowicz, 2009; Reading, 2011). Thus, our first research question was as follows: can we, despite this alleged arbitrariness in ‘rendering social realities’, identify characteristic patterns with regard to search results obtained for different types of past events (RQ1)? Second, extant literature has intensively discussed the influence, and bias, of search algorithms (consider Pasquale, 2015; Rieder and Sire, 2014). In view of this, we asked ourselves: can we, in the Russian context and with specific attention to results obtained for past events, identify differences between Yandex and Google (RQ2)? Third, extant studies have frequently lamented that the embeddedness of search engines in social processes and institutions is relatively understudied (Graham et al., 2013; Hargittai, 2007). Against this backdrop, our third research question was as follows: to what degree, and in what ways, are search results obtained through queries for past events biased towards memory narratives supported by Russia’s elites (RQ3)?

Methods of data collection and analysis

In the process of grounded theory development, which we pursued in the 6 months between September 2016 and February 2017, we entered approximately 300 purposefully sampled research terms into the search bars of the two search engines and systematically analysed the nature and content of the first 10 websites found in the results. Our strategy for sampling search terms (for an overview of sampling strategies, consider Lindlof and Taylor, 2011: 112–116) broadly followed a combination of criterion sampling (we searched only for past events of national political relevance to Russia), maximum variation sampling (we searched for a wide variety of events from both immediate and distant pasts) and typical and atypical case sampling (once we had tentatively identified a type of event, we searched for similar and different events to confirm or challenge the propositions we had formulated). Our criterion sampling involved choosing at least one key event of Russia’s history from each century, starting from the Christianisation of Kievan Rus’ in 988, but focusing on the 20th century as the foundation of most contemporary states’ memory politics.

In the concurrent process of data collection and theory development, we went back and forth between reconsidering extant literature and analysing the empirical data collected. The process of grounded theory development was aided by two factors. First, both authors are Russian-speaking media experts, with one socialised and educated in Belarus and Russia and the other in Western Europe and the United States. The authors are thus deeply familiar with competing memory narratives both within Russia and in the West. Second, one author has a disciplinary background in memory studies and the other in communication.

One major threat to the reliability of our findings was search engine customisation. Both Yandex and Google personalise results according to the user’s age and gender, her geographical location, previous web and search histories and other factors (Van Dijck, 2007; Yandex, 2012). When collecting data for this study, we aimed to reduce the consequences of personalisation to a minimum. To do so, we deleted from our browsers all cookies and search histories. Since we conducted our analysis from outside Russia, we also switched off geolocation services on our computers to be able to use the Russian versions of the two search engines, google.ru and yandex.ru (both Google and Yandex automatically redirect users to the local version of the respective search engine, and using non-Russian versions of the search engines changes search results significantly). In the case of Yandex, switching off geolocation services was followed by the requirement to pick a Russian city from which the search would be conducted: we opted for Moscow.

To test the reliability of the search results obtained in this qualitative way, we decided to undertake a quantitative verification test and conducted trial queries for nine selected events, on both Yandex and Google. The queries were executed on a fixed day (9 June 2017), using the computer in Berlin on which we had conducted our research and five different computers located in Moscow. On our Berlin computer, we conducted 18 searches and thus obtained 180 links. Correspondingly, the five test individuals based in Moscow conducted 90 searches in all, obtaining 900 links. Prior to each search, all cookies and search histories were deleted from our Berlin computer, while Moscow-based testers conducted the searches with search histories and cookies on their computers remaining intact. By comparing the results gathered in this way, we aimed to check the degree to which the results we obtained on our Berlin computer, with geolocation services switched off, corresponded to those obtained on standard Moscow computers. We found that the average agreement of our results with the results obtained on Moscow-based computers was 89% for Google and 82% for Yandex. The same procedure was then carried out again, this time with personal histories and cookies deleted each time before a search was conducted on each of the six computers participating in the reliability test. Comparing the results obtained in this way allowed us to test the importance of personal histories and cookies for search results. Our findings showed that the agreement for Google increased to 92% and that for Yandex to 86%. This implies that while cookies and personal search histories do influence search results, their effect should not be overestimated. To summarise, we could not identify effects of personalisation great enough to threaten the validity of our propositions and typologies.

Outcomes of searching the Internet for past events

Four types of websites: production of content and location of power struggles

In the process of data collection and grounded theory development, four broad types of websites emerged that differed markedly in terms of the type of content published, and with regard to how this content was produced: (1) pro-regime mass media websites, (2) encyclopaedic websites, (3) personal websites and (4) independent media and NGO websites.

Pro-regime mass media websites are curated by collectives of professional journalists that function as part of larger news organisations often also operating newspapers, radio stations or TV channels. In Russia, these news organisations are typically owned either by the state or by business elites with close ties to the country’s ruling elites (Toepfl, 2011). In the analysis, we considered as pro-regime mass media websites all those that at the time of research were reporting in exclusively positive terms on Vladimir Putin and his policies. Crucial for this study, websites of this type also tend to comply with the regime’s memory politics. Examples are 1tv.ru, the web platform of the state TV Channel One, and lenta.ru, part of the Kremlin-friendly oligarch Alexander Mamut’s media empire.

Encyclopaedic websites comprise sources that offer summaries of information on one or more subjects or areas of knowledge. Most big online encyclopaedias are created and edited by collectives of users, which implies that they have multiple editors with often competing views. However, this category also includes smaller reference and educational websites dedicated to one topic (such as a past event or a historical period), which are usually produced by a small team. Our findings show that encyclopaedic sources created collectively (for instance, Wikipedia) typically provide multifaceted accounts of the past, striving for ‘neutrality’ or ‘objectivity’, whereas reference sources produced by a small group of people tend to adopt either a pro-state or a critical perspective. The content of smaller encyclopaedic websites can thus either challenge official memory narratives or support them; in relation to bigger encyclopaedias, the power struggle is internal to the sources themselves, with so-called ‘edit wars’, that is, wars between editors who have irreconcilable views on particular topics, quite common (Dounaevsky, 2013). Encyclopaedias that featured prominently in our search results included the Russian version of Wikipedia and its replicas. Among smaller reference websites dedicated to one topic were anaga.ru, on military history and Russian armament, and stalin.narod.ru/stalin.htm, on the rule of Joseph Stalin – ‘the greatest genius of mankind’, as the Russian version of the website calls him. Both sources, the authors of which could not be identified, are in line with dominant historical narratives (even if at times their interpretations are more radical than those within the official memory politics).

Our third type is personal websites, including blogs. This type is usually centred around one authoritative personality, even when this personality is not the only author. Personal websites can be multifaceted, pro-state or critical of the regime. Some of the personal websites that appeared in our queries were critical of the regime or took a seemingly neutral attitude, which in the Russian context can mean some implicit criticism of the state in terms of the choice of subject covered: for instance, multiple deaths during Joseph Stalin’s funeral, absent from the official memory narrative, are discussed on Ilya Varlamov’s blog (varlamov.ru). Other websites are openly and aggressively pro-state, as is the blog rovego.livejournal.com, where the author discusses in one entry Donald Trump’s victory in the US presidential election and its benefits for Russia. The examples of such sources found in our search results thus show a whole range of possible political orientations and memory narratives.

Finally, the fourth category is independent media and NGO websites. What unites these, in the specific Russian context, is that they are produced by collectives of politically engaged individuals with a critical stance towards the authoritarian regime. Among the independent media websites featuring prominently in our search results was, for instance, the online news source Meduza. Independent NGO websites included those of the historical and civil rights society Memorial (memo.ru) and the human rights organisation Moscow Helsinki Group (mhg.ru). On websites of this fourth type, we came across memory narratives that resolutely challenge the regime’s memory politics.

Websites’ ranking by search algorithms

The four types of websites differed not only with regard to the memories they mediated but also with regard to how they were typically ranked by the two search algorithms under investigation. These differences were largely due to search engines’ basic mode of functioning. According to Granka (2010: 366–368), the three principal elements that make up search engines’ algorithms are ‘linguistic cues’ (for instance, how frequently a webpage contains the specific search term entered by the user), ‘popularity cues’ (identification of webpages with the highest number of inbound external links) and ‘user behaviour cues’ (identification of webpages on which users most frequently click). It was with regard to the latter two criteria that the four types of websites obtained in our results systematically differed: the number of inbound links (popularity cues) and their audience reach (user behaviour cues). According to these two evaluation criteria, encyclopaedic websites such as Wikipedia achieved the highest ranking (an exceptionally high number of inbound links and an exceptionally large audience). These were followed by pro-regime mass media (a high number of inbound links and a large audience), independent media and NGO websites (in the specific Russian context, these were less interlinked and less visited than pro-regime websites, see Toepfl, 2011) and personal websites (a small number of inbound links and a small audience).

Four types of memory events: representation in the dominant discourse

The proportion of results for each of the four website types varied markedly with regard to different past events. These events differed in terms of two characteristics: (1) their remoteness from the present moment and (2) their representation in the dominant (pro-regime) media discourse. Based on these distinctions, we identified four types of events: non-actualised memory events, actualised memory events, counterpublic memory events and emerging memory events.

Non-actualised memory events took place at a moment remote from the present and are absent from the dominant media discourse, which implies their relative unimportance to memory politics. Consequently, these events are rarely, if at all, conveyed by pro-regime mass media websites, personal websites or independent media and NGO websites. The querying individual’s view of these events will predominantly depend on how they are portrayed on encyclopaedic websites. It is important to emphasise that while a memory event cannot be non-actualised on an individual level (when the user types an event into a search bar, she ipso facto actualises it for herself), our categorisation refers to the respective event’s presence or absence in the dominant discourse and thus its importance to Russia’s memory politics, as reflected in search results. For instance, at the centre of Russia’s memory politics is the Soviet Union’s victory in the ‘Great Patriotic War’ (as World War Two is traditionally called in the Russian historical discourse), including Stalin’s role in achieving this victory. The Soviet leader’s crimes (e.g. political repression, forced relocations of whole ethnic groups) are not denied by the Kremlin, but are also not emphasised in the country’s memory politics (Gjerde, 2015). Accordingly, the fact that thousands of people were crushed to death in the crowds of mourners at Stalin’s funeral is absent from the dominant media discourse. Of our four website types, Stalin’s funeral (pokhorony Stalina) was almost exclusively presented at the time of data collection on encyclopaedic websites, and the very process of searching for this event on the Internet would actualise it for the querying individual primarily as it was represented on this type of websites. Significantly, in contrast to events of the 20th century, most events of earlier Russian history that are crucial for Russians’ identity (the Christianisation of Rus’ in 988, the reforms of Peter I in the first quarter of the 18th century, and numerous others) are largely absent from the dominant media discourse. This is partly explained by the fact that search engine algorithms privilege large online encyclopaedias and tend to crowd out other website types from search results (Granka, 2010: 366–368). However, this also allows us, with a certain conditionality, to consider these events as not central to the country’s memory politics.

Actualised memory events took place at a moment remote from the present, but are, in contrast to events of the first type, present in the pro-regime media discourse. Due to their presence in the dominant discourse and hence their importance to Russia’s memory politics, these events are usually conveyed by all four types of websites. However, memories of events of this type will primarily depend on how they are mediated by major pro-regime mass media websites and large encyclopaedias, with independent media and NGO websites and personal websites typically – due to their lower ranking by search engines (Granka, 2010: 366–368) – being crowded out from the first 10 results. An example of this type of event is the Soviet coup d’état attempt of 19–21 August 1991 (the August Putsch (avgustovskiy putch)), an attempt of members of the USSR government to displace Mikhail Gorbachev and take power in their own hands. While this event is not crucial for, it is undeniably important to the official memory politics, which is illustrated by the fact that at the time of data collection, Russia celebrated the 25th anniversary of the Putsch and both pro-regime and independent media websites dedicated stories to it. The emphases in these media, however, were different. While anti-Kremlin sources pointed to the failure of the Putsch as a victory for a new Russia wishing to free itself from the clutches of Communists (and hence, to put an end to the Soviet Union), pro-regime sources sought to avoid emphasising the demise of the USSR (‘the greatest geopolitical catastrophe’ of the 20th century, in Putin’s words), which came shortly after the Putsch, and accentuated instead the birth of the Russian Federation.

Third, counterpublic memory events either took place at a moment remote from the present or are taking place about now. The name used for events of this type is a reference to Fraser’s (1990) theory of counterpublics. In contrast to the three other types, counterpublic memory events are in explicit conflict with the pro-regime media discourse, which implies that on pro-regime mass media websites events of this type may be either not mentioned at all or referred to using a fundamentally different term. At the same time, counterpublic memory events can be mediated by the other three types of websites: encyclopaedias, personal websites, and independent media and NGO websites. In some cases, and specifically with regard to transnational memory events, websites of foreign origin can also be found in the top 10 search results.

A first illustrative example here is the 1968 Red Square Demonstration (seven-people protest (protest semerykh)), a protest against the Soviet invasion of Czechoslovakia that took place on 25 August 1968. As is characteristic for counterpublic events, pro-regime mass media websites were absent from our search results. The query produced links dominated by encyclopaedic websites (Wikipedia and the educational portal Arzamas Academy) and independent media and NGO websites (such as the Memorial and the Moscow Helsinki Group). Another example is the search query ‘Annexation of Crimea’ (anneksiya Kryma), which produced results in sharp contrast to those for the query ‘Incorporation of Crimea’ (prisoyedineniye Kryma): the two queries relating to the same historical event actualised radically different memories. According to the official narrative of Russia’s elites, Crimea was not ‘annexed’ by the Russian Federation but ‘incorporated’ into it. Search results for the query ‘Incorporation of Crimea’ demonstrated a compatible number of links to pro-regime mass media websites and encyclopaedic websites, which made it an actualised memory event. Meanwhile, the term ‘Annexation of Crimea’ did not produce a single link to pro-regime mass media. Instead, our results were strongly dominated by independent media and NGO websites. In addition, what made this memory event distinct was its transnational character: the sources found in the results were predominantly of Ukrainian and Belarusian, rather than Russian, origin.

Finally, emerging memory events are current affairs that are highly present in the pro-regime media discourse and have the potential to become ‘proper’ memory events. They are covered by all types of websites except encyclopaedias, since encyclopaedic entries for these events often do not yet exist. Crucially, due to the fact that personal websites and independent media and NGO websites will typically be crowded out by search engine algorithms (Granka, 2010: 366–368), it is pro-regime mass media websites that will define the ways in which these events are framed and how they will be memorised (and, consequently, how they may be recollected later). The search process can be viewed in this case as a mnemonic practice directed at the future (memorisation), rather than at the past (recollection). Significantly, due to the fact that emerging memory events characterise recent occurrences that can have further developments, their representation in the search engines can develop as well. The way in which these events are encountered for the first time will, however, remain of importance. Since these events are conveyed primarily by pro-regime mass media websites, as found in the search results, the public’s first impression of them will be embedded in the dominant discourse, which will inevitably play a role in how these events are perceived in the future, irrespective of further developments. This effect is, in a way, similar to the idea of ‘flashbulb memory’ describing ‘the “photographic” quality of the recall’ (Hoskins and O’Loughlin, 2010: 105) of some events. At the same time, if the media attention wanes, some emerging memory events can also be forgotten.

An event that can be used as an example of emerging memory events is the detention and subsequent arrest in November 2016 of Alexey Ulyukayev (arest ulyukaeva), Russia’s then minister of economic development who was accused of soliciting a bribe from Igor Sechin, the executive chairman of the state oil-company Rosneft. Ulyukayev’s arrest was top news at the time of data collection, which resulted in pro-regime mass media websites absolutely dominating the search results. The story has developed significantly since the arrest, but the initial impression (framed by pro-regime media) is imprinted in the public’s memory and cannot but influence the perception of the whole case. However, only time will show whether Ulyukayev’s arrest will become an actualised (or any other type of) memory event.

Differences between Google and Yandex

A comparison of the results between Yandex and Google did not point to strikingly different patterns with regard to any of our categories. Contrary to our initial expectations, our typologies, and the propositions formulated on the relationship between type of memory event and results obtained, were broadly valid for both search engines. However, the patterns we identified in search results may change significantly as a consequence of the Russian government’s regulatory measures against Yandex that appeared to be on the horizon at the time of research. Increasing government pressure on Yandex (Daucé, 2017) may result in independent media organisations further losing their influence. This would significantly change the consequences of Internet searching as a mnemonic practice with regard to the three types of events where independent news organisations play a somewhat or highly important role: actualised, counterpublic and emerging memory events.

Reproduction and reinforcement of dominant narratives

When embarking on this study, we expected Google and Yandex to actualise opposing or at least very different memories of past events. We based our expectations on the study by Jiang (2014), who compared search results for two search engines operating in China, Baidu and Google, and found an overlap of only 6.8% in search results. Our assumption turned out to be wrong (RQ2). Why do our results contrast with those of Jiang (2014)? This may be partly due to the fact that our study investigated different search engines in a different authoritarian context. However, our project’s research design, distinct from Jiang’s, may also have contributed to the diverging results. While Jiang (2014) compared the extent to which two search engines included the same websites in results, our study compared the extent to which they featured the same, relatively abstract, types of websites. We adopted this approach because we found that within Russia’s political context, each of the four website types would mediate specific types of memory narratives. For instance, while the first type (pro-regime mass media) would typically disseminate memory narratives supported by the regime, the fourth type (independent NGO and news websites) could also be expected to feature narratives that challenged the regime’s take on the past. Compared to Jiang’s (2014) study, the analytical framework developed in this article thus served particularly well to shed light on the degree to which search engines as technological artefacts articulate the power struggle over competing memory narratives between ruling elites and oppositional groups in Russia (RQ3).

While we found next to no indication that the algorithm of Yandex, compared with that of Google, privileged regime-loyal types of websites over critical ones (RQ2), critical sources challenging the regime’s memory narratives were not highly present in the results. Particularly with regard to the types of events that constituted key elements of the official memory politics (actualised and emerging memory events), both Google and Yandex accorded extensive presence to pro-regime mass media. In contrast, personal websites were virtually absent from the search results for all four event types. Independent media and NGO websites, as the only explicitly critical sources, were typically accorded only low or medium visibility, with the exception of counterpublic memory events. This leads us to conclude that within Russia’s context and at the time of research, the search process, as a mnemonic practice, tended to reproduce the dominant memory discourse (RQ3), irrespective of whether the algorithm of Yandex or that of Google was deployed (RQ2).

The bias of both search engines towards dominant memory narratives appears to be rooted in the technology’s very ‘nature’, that is, in the ways in which algorithms typically establish the relevance of websites (Beer, 2017; Rieder and Sire, 2014). If two of the principal elements that make up search engine algorithms are the number of inbound links and the amount of incoming traffic (Granka, 2010: 366–368), the high degree of interconnectedness and extensive reach of a small number of websites that are closely tied to, and financially supported by, a country’s most powerful institutions and elites implies the existence of power inequalities. Importantly, search engine algorithms will not only reproduce these power inequalities (by privileging websites that are already popular) but also increase them. Through an act of searching for a past event and clicking on a website, the user will not only be exposed primarily to dominant memories but also contribute to the website’s popularity and thus influence future search results. This will, in all likelihood, reinforce the dominant narrative, in line with Van Dijck’s (2007) argument that memories are not simply mediated by technologies of memory: they shape and are shaped by them (p. 21).

Querying the Internet as a socially framed mnemonic practice

As our findings illustrate, search results are largely predicated upon how users phrase their queries. This is most obvious in the outcomes of the queries ‘Annexation of Crimea’ and ‘Incorporation of Crimea’, which – despite referring to the same historical event – produced fundamentally different results that tended to transfer recollecting individuals to webpages that actualised opposing memories of the event. Against this backdrop, it can be argued that within the Russian context, the political outcome of querying as a mnemonic practice is much less affected by the power of distinct algorithms (Yandex vs. Google) than by specific search terms that individuals enter to recall past events. The crucial question in this case is which of the two search terms do Russians predominantly use when seeking to recall what happened with Crimea in 2014? Yandex provides a tool (Yandex Keyword Statistics) that helps answer this question by comparing the popularity of queries across geographical spaces. Specifically, it offers provisional forecasts for how many ‘impressions a month’ advertisers can expect to obtain when using a specific term as one of the keywords for their ad. These figures indicate that within Russia and as of May 2017, individuals sought to remember the ‘Incorporation of Crimea’ approximately 12 times more frequently than the ‘Annexation of Crimea’ (21,566 vs. 1717 expected impressions). In contrast, in Ukraine users tended to remember the ‘Incorporation of Crimea’ 1.4 times less frequently than its ‘Annexation’ (720 vs. 520 expected impressions). These figures illustrate how even the seemingly individualistic practice of querying the Internet is highly affected by the ‘social frameworks’ (Halbwachs, 1992 [1925]: 35) within which it occurs. Consequently, as Olick (2008) argued, within any society

powerful institutions […] clearly support some histories more than others, provide narrative patterns and exemplars of how individuals can and should remember, and stimulate public memory in ways and for reasons that have little to do with the individual or aggregate neurological records. (p. 156)

Search engine companies, and the algorithms they operate, thus form only one of many powerful institutions that constitute the social framework within which querying the Internet as a mnemonic practice is pursued in a society. The outcome of this practice is also affected by the memory work of multiple other powerful institutions, including pro-regime and independent mass media, NGOs and online encyclopaedias.

Future research

This exploratory study has several limitations. First, it is grounded in a single revelatory case study, basing its findings exclusively on empirical observation from Russia. Future research is required that will substantiate the conclusions and categorisations presented in this article by replicating a similar research design across other contexts. Second, this study has adopted a qualitative approach to analysing the results obtained through search queries. In contrast, future research could develop and deploy quantitative designs building on the typologies of websites and memory events developed in this study. Quantitative analyses of the degree to which different types of websites dominate search results could be deployed to further interrogate the bias of search engines, and to trace how they represent different events over time. The latter type of analysis could explore the conditions under which specific memory events transform from one type to another, for example, from a non-actualised to an actualised memory event. Third, this study has addressed only past ‘events’, while personas, issues and processes of the past can be equally important to a government’s memory politics and a social group’s collective memory. Future research could investigate these types of engagement with the past using a search engine. Finally, this study has been able to dedicate only a brief section to theoretical reflections on the consequences of Internet searching as a mnemonic practice: further theorising on this type of engagement with the past is required. The political relevance of these lines of research is underlined by the fact that at the time of writing, Russia’s ruling elites directed at Yandex highly effective legal measures (Daucé, 2017) that cannot but influence citizens’ engagement with the past through the mnemonic practice of searching the Internet.

Footnotes

Acknowledgements

This work has greatly benefited from the comments by two anonymous reviewers, as well as the participants of the Mnemonics 2017 gathering, especially Aleida Assmann and Ann Rigney. The authors also owe deep thanks to their student assistant Daria Kravets and the five Moscow-based participants of the reliability test.

Funding

This research was supported by an Emmy Noether grant sponsored by the German Research Foundation DFG.

ORCID iD

Andrei Zavadski

References

Amoore

Piotukh

(eds) (2016) Algorithmic Life: Calculative Devices in the Age of Big Data. London: Routledge.

Beer

(2017) The social power of algorithms. Information, Communication & Society 20(1): 1–13.

Bidder

(2012) Russia battles online foes and freedoms. Spiegel Online, 10 August. Available at: http://www.spiegel.de/international/spiegel/growing-russian-internet-power-both-a-boon-and-worry-to-kremlin-a-849125.html (accessed 2 August 2017).

Blacker

Etkind

Fedor

(eds) (2013) Memory and Theory in Eastern Europe. Basingstoke: Palgrave Macmillan.

Boletskaya

(2016) Google mozhet nazvat’ sebya liderom runeta [Google can call itself the leader of Runet]. Vedomosti, 30 May. Available at: http://www.vedomosti.ru/technology/articles/2016/05/31/643012-google (accessed 2 August 2017).

Bourdon

Kligler-Vilenchik

(2011) Together, nevertheless? Television memories in mainstream Jewish Israel. European Journal of Communication 26(1): 33–47.

Daucé

(2017) Politicheskiye konflikty vokrug interneta v Rossii: keis instrumenta ‘Yandeks.Novosti’ [Political conflicts around the internet in Russia: the case of Yandex.Novosti]. Laboratorium 9(2): 112–132.

Dounaevsky

(2013) Building Wiki-history: between consensus and edit warring. In: Rutten

Fedor

Zvereva

(eds) Memory, Conflict and New Media: Web Wars in Post-Socialist States. London: Routledge, pp. 130–142.

Etkind

Uffelmann

(2014) Editorial: digital mnemonics. Digital Icons: Studies in Russian, Eurasian and Central European New Media 12: i–v.

10.

Fraser

(1990) Rethinking the public sphere: a contribution to the critique of actually existing democracy. Social Text 25–26: 56–80.

11.

Fredheim

(2013) Detecting memory events. East European Memory Studies 14: 9–14.

12.

Garde-Hansen

(2011) Media and Memory. Edinburgh: Edinburgh University Press.

13.

Garde-Hansen

Hoskins

Reading

(eds) (2009) Save As … Digital Memories. Basingstoke: Palgrave Macmillan.

14.

Gjerde

(2015) The use of history in Russia 2000–2011: the Kremlin and the search for consensus. East European Politics 31(2): 149–169.

15.

Graham

Schroeder

Taylor

(2013) Re: search. New Media & Society 15(8): 1366–1373.

16.

Granka

(2010) The politics of search. Information Society 26: 364–374.

17.

Halbwachs

(1992 [1925]) On Collective Memory. Chicago, IL: The University of Chicago Press.

18.

Hargittai

(2007) The social, political, economic, and cultural dimensions of search engines: an introduction. Journal of Computer-Mediated Communication 12: 769–777.

19.

Hellsten

Leydesdorff

Wouters

(2006) Multiple presents: how search engines rewrite the past. New Media & Society 8(6): 901–924.

20.

Horsti

(2017) Communicative memory of irregular migration: the re-circulation of news images on YouTube. Memory Studies 10(2): 112–129.

21.

Hoskins

O’Loughlin

(2010) War and Media: The Emergence of Diffused War. Cambridge: Polity Press.

22.

Introna

Nissenbaum

(2000) Shaping the web: why the politics of search engines matters. The Information Society 16(3): 169–185.

23.

Jakubowicz

(2009) Remembering and recovering Shanghai: seven Jewish families [re]-connect in cyberspace. In: Garde-Hansen

Hoskins

Reading

(eds) Save As … Digital Memories. Basingstoke: Palgrave Macmillan, pp. 96–114.

24.

Jiang

(2014) The business and politics of search engines: a comparative study of Baidu and Google’s search results of Internet events in China. New Media & Society 16(2): 212–233.

25.

Keightley

Pickering

(2014) Technologies of memory: practices of remembering in analogue and digital photography. New Media & Society 16(4): 576–593.

26.

Keightley

Schlesinger

(2014) Digital media – social memory: remembering in digitally networked times. Media, Culture & Society 36(6): 745–747.

27.

Lindlof

Taylor

(2011) Qualitative Communication Research Methods. Los Angeles, CA: Sage.

28.

Miller

(2012) Istoricheskaya politika v Vostochnoy Evropie nachala XXI v [Historical politics in Eastern Europe in the early 21st century]. In: Miller

Lipman

(eds) Istoricheskaya Politika V XXI Viekie [Historical Politics in the 21st Century]. Moscow, Russia: Novoye Literaturnoye Obozreniye, pp. 7–32.

29.

Olick

(2008) From collective memory to the sociology of mnemonic practices and products. In: Erll

Nünning

(eds) Cultural Memory Studies: An International and Interdisciplinary Handbook. Berlin: Walter de Gruyter, pp. 151–162.

30.

Olick

Robbins

(1998) Social memory studies: from ‘collective memory’ to the historical sociology of mnemonic practices. Annual Review of Sociology 24: 105–140.

31.

Parfentyeva

Papandina

Nosonov

(2016) ‘Yandeks’ perestanet pokazyvat’ novosti nezaregistrirovannykh SMI [Yandex to stop showing news from unregistered mass media]. RBC, 20 October. Available at: http://www.rbc.ru/technology_and_media/20/10/2016/5808d3d99a7947d8a61cf78c (accessed 2 August 2017).

32.

Pasquale

(2015) The Black Box Society: The Secret Algorithms That Control Money and Information. Cambridge, MA: Harvard University Press.

33.

Reading

(2011) Memory and digital media: six dynamics of the globital memory field. In: Neiger

Meyers

Zandberg

(eds) On Media Memory: Collective Memory in a New Media Age. London: Palgrave Macmillan, pp. 241–252.

34.

Rieder

Sire

(2014) Conflicts of interest and incentives to bias: a microeconomic critique of Google’s tangled position on the web. New Media & Society 16(2): 195–211.

35.

Rutten

Fedor

Zvereva

(eds) (2013) Memory, Conflict and New Media: Web Wars in Post-Socialist States. London: Routledge.

36.

Ruvinskiy

(2017) Plokhoy i khoroshiy ‘Yandeks’ [The good and the bad Yandex]. Vedomosti, 22 September. Available at: https://www.vedomosti.ru/opinion/articles/2017/09/22/734888-plohoi-horoshii-yandeks (accessed 30 November 2017).

37.

Spink

Zimmer

(2008) Web Search: Multidisciplinary Perspectives. Berlin: Springer.

38.

Toepfl

(2011) Managing public outrage: power, scandal, and new media in contemporary Russia. New Media & Society 13(8): 1301–1319.

39.

Toepfl

(2016) Innovating consultative authoritarianism: Internet votes as a novel digital tool to stabilize non-democratic rule in Russia. New Media & Society. Epub ahead of print 22 November 2017. DOI: 10.1177/1461444816675444.

40.

Van Dijck

(2007) Mediated Memories in the Digital Age. Palo Alto, CA: Stanford University Press.

41.

Yandex (2012) Novaya platforma poiska Yandeksa s personal’nymi rezul’tatami: ‘Kaliningrad’ [‘Kalinigrad’: Yandex’ new search platform with personalised results]. Habrahabr.ru, 12 December. Available at: https://habrahabr.ru/company/yandex/blog/162245/ (accessed 2 August 2017).

42.

Yin

(2014) Case Study Research: Design and Methods. Los Angeles, CA: Sage.

43.

Zimmer

(2010) Web search studies: multidisciplinary perspectives on web search engines. In: Hunsinger

Allen

Klastrup

(eds) International Handbook of Internet Research. Dordrecht: Springer, pp. 507–522.