Is single-issue voting smart for elections

Lesson learned? Democratic resilience to digital electoral influence in the USA and Germany

Summary

The study examines the influence of systemic and situational factors on the different democratic resilience to external, digital electoral influence by Russia in the USA in 2016 and in Germany in 2017. The comparison shows that in particular the degree of polarization, the logic of the election campaign and the position of established media have explanatory power for the respective Russian meddling can be assumed. At the same time, the effects of individual factors on the technical level (e-voting) cannot be isolated, but only evaluated in relation to one another.

Abstract

The study examines the influence of systemic as well as situational factors on the different democratic resilience to external, digital election meddling by Russia in the US 2016 and in Germany 2017. The comparison shows that the degree of polarization, the logic of the election campaign and The role of established media in particular can be assumed to explain the respective Russian meddling. At the same time, the effects of individual factors on the technical level (e-voting) cannot be assessed in isolation but in relation to each other.

introduction

The Russian influence in the US election campaign in 2016 was perceived as an alarm signal by democratic states around the world. Four years later, in the run-up to the 2020 presidential elections, there are signs of similar occurrences (Wong 2019). The specific research interest of this study lies in the question of what influence certain features of the democratic systems had on the resilience to external, digitally occurring electoral influence during the 2016 presidential election in the USA and the 2017 federal election in Germany.

The case selection is justified by the similarities and differences between the two democracies: Here, institutional, but also more socio-political and informal system characteristics play a role. The USA and Germany, as liberal democracies, represent in principle attractive targets for external electoral influence by states such as Russia. Nevertheless, with regard to the dependent variable, resilience to external, digital electoral influence, they show in the study periodFootnote 1 shows a variance worth investigating. If this different resilience of two democracies is put to the test by the same actor, its effect, which is otherwise difficult to operationalize, can be made visible. In the context of this study, this should be done by means of a structured, focused comparison. Due to the small-N design, no claim to completeness can be made here: For the future, further analyzes will be required, which will illuminate the much larger spectrum of potential democratic varieties on the institutional and socio-political level in the sense of the study.

After a theoretical localization of the approach, the theory-based hypothesis formation takes place. The methodological approach is then specified. According to this scheme, the two countries and the democratic characteristics before and during the 2016 and 2017 elections are compared in relation to the reported Russian influencing measures.

The current state of research

The present work combines the sub-areas of International Relations (IB) and Comparative Political Science (VP). Even if, for example, the intervention of the USA in regional affairs of South American states during the Cold War (Cottam 1994) should not be concealed, the IB has so far examined more autocratic states that either “democracy prevention” (van Soest 2014) or “autocracy promotion” ( Burnell 2010; Bader et al. 2010; Tansey 2016). The common denominator of both concepts has so far been the regional focus, which is also reflected in the majority of the previous case studies, for example on Russia or China (Ambrosio 2007; Melnykovska et al. 2012; Bader 2015). Marianne Kneuer and Thomas Demmelhuber (2016, p. 778) conceptualize this export of autocratic varieties, which in some cases has already been viewed as a success, in the sense of regional "centers of gravity" with international appeal. According to Espen Rød and Nils Weidmann (2015), more and more autocracies such as China or Russia are flanking censorship and internet blockades with digital tools such as e-participation offers to reduce the “dictator’s dilemma” (Göbel 2013, p. 388). The regional focus of digital autocratic foreign policy is also increasingly evident in the control of social media discourses (Gunitsky 2015). An example of this is the extensive Russian involvement in the Ukraine conflict, according to observers a “cyber test battlefield” of the Kremlin (Cerulus 2019). In the case of the US elections in 2016, however, the envisaged goal was a geographically distant, liberal and consolidated democracy. Thus, the expression of the digital applies here democracy weakening / contestation to, which seems increasingly worthy of investigation.

You can use a Black KnightFootnote 2 (Tolstrup 2015) Disrupt elections even without a physical presence on site, undermine or even manipulate their integrity. This can go hand in hand with an ever decreasing use of resources, keyword asymmetrical conflict resolution. Another advantage is the extensive impunity in the digital space, which results from the problem of the attribution of a criminal offense (Knake 2010).

The increasing digital dependency of liberal democracies also explains their enormous vulnerability on a social, economic and security level (Colaresi 2014, p. 237). Elections are attractive goals because they not only decide who will rule the country for the next few years, but also represent the core of the identity of a liberal democracy. However, democratic Estonia is also widely networked and therefore potentially easily vulnerable. It is not attacked because of its general power status or because of conventional resources, but - as in the case of the Russian Distributed Denial of Service-Attacks (DDoS attacks) on the country in 2007 - only if there is a specific reason for interference (Herzog 2011).

The USA and Germany, on the other hand, are characterized by great economic strength and their military presence in many crisis areas around the world (Silver 2019). In the case of the USA, this applies to its involvement in the Middle East as well as in Eastern Europe. Both regions are of central importance for Russia. On the foreign policy level, especially within the framework of the European Union, Germany can be seen as a possible opponent of Russia, not least because of the vehement enforcement of sanctions against the Kremlin as a result of the annexation of Crimea in 2014 To weaken elections. Despite the similar basic requirements mentioned, the USA and Germany showed different degrees of resilience during the study period. In addition to a possibly simply greater interest on the part of Russia in the US elections, this article assumes that the democratic differences to be examined have a decisive explanatory power.

The VP's research on democratic resilience online is still relatively young and has been advanced especially since the events surrounding the US elections in 2016. The work by Maria Hellman and Charlotte Wagnsson (2017) on European strategies to defend against Russian should be mentioned in particular Information was-Methods. Amy Pope (2018) also examined possible preventive measures for external electoral influence. The populism relevant in the context of the study was analyzed by Pippa Norris (2017) in the context of electoral integrity and assessed with regard to possible negative effects, but digital means were not included.

With the aspect of strict technical resilience in terms of electoral IT securityFootnote 3 studied Ben Buchanan and Michael Sulmeyer (2016). In contrast to this work, however, the present study also includes disinformation campaignsFootnote 4 and does not dwell exclusively on the purely technical level.

Building hypotheses

In the following, the independent variables are converted into hypotheses and derived primarily from formal system factors such as the party and electoral system.Footnote 5

Formal system factors

Polarization.

In presidential systems, the mostly smaller number of parties often makes governing difficult due to oppositional majorities. In addition, complex lines of conflict are usually not sufficiently covered as a result. Blockages can be the result. Amy Gutmann and Dennis Thompson speak of an "uncompromising mindset" (2010, p. 1125), which pushes political decisions to the ideological fringes, especially in two-party systems like the USA. The majority voting principle, which is mostly used in presidential systems, also contributes to the greater polarization hereFootnote 6 at: Whose The Winner Takes It All-Logic exacerbates tensions within the population and between the parties. The often lower polarization in parliamentary systems including proportional representation can, however, also have a negative effect on democracy due to a lack of programmatic differentiation in the form of disaffection with politics (Maier 2000, p. 15).

Polarization does not have to be anti-democratic per se, populist framed however, it undermines trust between the parties and the supporters (Mudde and Kaltwasser 2012). In particular, media providers with an often populist orientation that can only be accessed online promote this mutual mistrust in times of elections. However, there is a basic trust of the population in theirs long-established Media landscape, this can potentially offer protection against attempts to manipulate public opinion. Furthermore, the respective logic of the election campaign influences the polarization: election campaigns that are heavily focused on the personal interests of the candidates and less on content-related aspects offer stronger areas of attack for defamatory dirt campaigns in the sense of hacking (and doxing) or disinformation.

These theoretical considerations result in the following hypotheses, which are examined for their explanatory power in the course of the empirical analysis:

H1:

The greater the political polarization as a product of the institutional arrangement, the lower the resilience to external, digital electoral influence.

H1a:

The more populist this drifting apart of the respective poles, partly due to the candidate-based election campaign logic of a democracy, the more susceptible the democracy is to disinformation campaigns in social media or doxing.

H1b:

The more populist online media and the lower the reach of established media representatives, the more susceptible democracy is to disinformation campaigns in social media or doxing.

Contest of choice.

The high degree of political polarization can lead to a great deal of competition, especially in two-party systems in elections with the logic of the majority principle. Even in countries which, following proportional representation, have a multi-party system mainly as a result, this can be large. Nevertheless, this is usually moderated by the larger number of election winners (Scientific Services of the German Bundestag 2012, p. 7). The longer an election campaign is fought, the more sensible it is to exert external influence. If a candidate appears to be out of competition from the outset, an attempt could at most be made to divide the country more strongly through disinformation campaigns instead of manipulating the overall result too laboriously. From these assumptions the following hypothesis arises:

H2:

The longer and more competitive a democratic election campaign is, the more susceptible democracy is to influencing the election on a technical level (e-voting).

Potential technical vulnerabilities.

The degree of digital voting practice is central here. To what extent is e-voting being used and which technologies are used for what? A nationwide uniform system could on the one hand be easier to attack or give perpetrators the opportunity to give up a hit to achieve the greatest possible effect. On the other hand, the different quality of the individual federal IT electoral systems can also make these more attractive as targets, since federal weaknesses could be exploited and at the same time targeted influence could be exerted (Buchanan and Sulmeyer 2016, p. 13). On the other hand, different modularities and technologies from different manufacturers also offer potential protection: although partial disruption of the choice is possible, manipulation of the overall result appears unlikely. The following hypotheses can be derived from this:

H3:

The more extensively a democracy uses e-voting, the more susceptible it is to external electoral influence on election day at the technical level.

H3a:

The more up-to-date the voting software, the more resilient it is with regard to exploiting security gaps.

H3b:

The more heterogeneous the electoral system and the electoral infrastructure, the more resistant the election is to any significant influence on the final result.

H3c:

The more heterogeneous the electoral system and the electoral infrastructure, the more susceptible the election becomes with regard to the disruption of federal subunits.

H3d:

The lower the nationwide security standards of the digital electoral infrastructure and the lower the federal control powers are, the more susceptible democracy is to partial attempts at interference.

Informal system factors

Creating a public awareness of the problem.

Democratic regimes can learn from negative examples from other democracies and thus develop greater problem awareness within the political leadership. This can also go hand in hand with IT preventive measures and legal regulations regarding social media platforms (SMP). Inter-democratic exchange weakens the efficiency of previously used autocratic instruments for the future, also on a technical level. If there has not yet been a corresponding reference event, at least it should to an influencing measure on the part of politics that has occurred who why and how the influence should be discussed in a timely manner and less the actual content in order to mitigate its potential impact. The last two hypotheses are thus:

H4:

The more active politicians and security officers publicly address them in the run-up to an election or after attempts to influence it, the more resistant this makes democracy to forms that rely on the element of surprise, such as B. Doxing.

H4a:

The more proactive a democracy takes preventive measures at the legal level, the more resistant it is to external influence on elections, especially to disinformation on SMPs.

methodology

As part of a structured-focused comparison, the hypotheses are tested for their explanatory power according to their operationalization. Many individual case studies lack the necessary theoretical focus in order to be relevant to the research area beyond the case. By means of a theoretically guided comparison, the hypotheses can be tested in a structured manner and thus made more connectable and fruitful for further research in the subject area. Focused The comparison is also because not all aspects of the respective case are dealt with, but only those that correspond to the theoretical interest of the work (George and Bennett 2005, pp. 67, 70). While polarization on the one hand, derived from institutional and thus more static factors, situational countermeasures in the run-up to the election are more in transition. Therefore sources of in front The actual period of study is used, since the respective system of government does not fundamentally change every year.

Formal system factors

Polarization.

Central indicators for this are the respective party and electoral system. However, since a two-party system does not necessarily lead to increased polarization, these must be viewed in the context of the respective institutional setting. In contrast to the concept of “issue voting” - or the often contrasted “party voting” (Highton 2010, p. 453) - H1a is less about the ultimate logic of the citizens' voting decision than about that of the election campaign . Their analysis is sought through a condensed representation of the respective culture of debate. If, for example, the opponent is attacked personally by the candidates particularly often, this speaks in favor of a person-based election campaign logic. In the case of H1b, any shift in the power structure between traditional print and often more populist online media is operationalized through opinion polls that have already been carried out in the respective countries. For the USA this would be shortened The New York Times across from Breitbart and for Germany the Süddeutsche Zeitung, DIE ZEIT or the Frankfurter Allgemeine Zeitung (FAZ) opposite Compact or similar providers of the populist spectrum.

Contest of choice.

To assess the length and strength of the competitive nature of an election campaign, national survey data at different points in time during the election campaign are used and compared with one another.

Technical vulnerabilities.

For this purpose, the e ‑ voting processes of the two countries are presented in a compressed form. This involves organizational (central vs. decentralized control, uniformity of procedures), but above all technical components (hardware and software) of the electoral process. In addition, the legal framework is considered, i.e. to what extent the technical elective components were subjected to regular checks and which standards were valid before the election.

Informal system factors

Creating a public awareness of the problem.

For this purpose, summarizing presentations of the rhetoric of relevant discourse carriers (politicians, actors from the circles of the security authorities) are created for both countries. In the case of Germany, this is intended to show the extent to which the events in the USA were repeatedly referred to in the run-up to their own elections and thus made it possible to sensitize the population. Due to the extensive use of fake news and disinformation on SMP in the US election campaign in 2016, a comparison of the regulatory Settings of the two countries shows that a democratic learning effect can also be assumed for Germany at this level.Footnote 7

Empirical analysis

According to the current state of knowledge, Russia used various measures to influence the US elections in 2016. These are discussed in more detail below.

Russian influence in the 2016 US presidential election

Hacking and Doxing.

In the so-called DNC hack (Democratic National Committee), the two hacker groups Cozy and Fancy Bear, who are assigned to Russian secret services, managed to break into the internal e-mail accounts of the US presidential candidate Hillary Clinton's chief of staff. Despite multiple warnings from the Federal Bureau of Investigation (FBI), the DNC reacted far too late to the risk of disclosing sensitive data to third parties to lose (Lipton et al. 2016). As a result, thousands of emails were published by Wikileaks and a website that was traced back to Russia. These emails contained e.g. For example, explosive material about Clinton's preference over Bernie Sanders, but also content that documented the party's anti-Catholic resentment. They also showed the extent of Clinton's financial proceeds from her previous Wall Street speeches (Lipton and Shane 2016), as confirmed by the then FBI Director James Comey in a Senate hearing in January 2017, the Republican National Committee also saw itself in Hacking attacks exposed in the run-up to the elections. In contrast to the DNC, however, no confidential information from the Republican Party and its presidential candidate and eventual election winner Donald Trump has been released to date. In addition, some of the Republican domains on older servers were no longer used at all (Greenberg 2017).

Hacking attacks and manipulation of IT by choice.

In the case of the Russian attempts to disrupt elective IT, the attribution situation is more controversial, but is now also to be seen as largely secure. Extensive access rights to the computers of individual employees of the companies responsible for the voting software were also targeted via phishing campaigns (including VR Systems) (Perlroth et al. 2017). In the run-up to the election, the hacking attacks were publicly discussed by the FBI, but the 21 states specifically affected were only informed of this one year later as a result of massive pressure (Collier 2017). Even if no decisive manipulation of the votes cast can be assumed, according to the Senate's investigation report from May 2018, manipulation or deletion of voter registration databases on election day partially disrupted the process. Even if this was limited to individual counties, reports of such incidents quickly spread beyond the borders of the states and subsequently undermined the integrity of the election (Perlroth et al. 2017).

Disinformation campaigns on SMP.

At an investigation hearing in the Senate at the end of October 2017, Twitter and Google announced that thousands of posts by Russian bots and so-called Trolls an estimated one-third of the total US population were viewed on their devices at the time of the election (Lee and Kent 2017). The reach of the 80,000 posts, according to Facebook, from approx. 120 fake Facebook pages that were traced back to Russia and addressed to approx. 29 million Americans, should be spread through sharing and Like have increased many times over. In addition to Facebook, the short message service Twitter was particularly affected: around 37,000 automated bots with connections to Russia were identified on the platform, and their tweets were displayed around 288 million times in the USA. There were also 2,752 accounts of the Troll army the Internet Research Agency (Lee and Kent 2017). Here too, however, their further amplification via botnets or retweets is missing.

In early October 2017, Facebook also admitted that politically motivated advertising worth more than US $ 100,000 was placed on the company's platform on behalf of actors allegedly from Russia (Isaac and Wakabayashi 2017). In terms of content, it covered a wide range of highly polarizing topics. According to Facebook, the majority supported the alt-right movement in particular, but also partially supported the alt-right movement Black Lives Matter-Campaign. Direct reference to the candidates for the US presidential election was mainly in the form of pro-Trump and contra-Clinton posts, but clearly outnumbered (Dawsey 2017). Unwitting agentsFootnote 8In the case of the USA, which also included journalists, was only made possible by the extreme reach of Russian content through its further dissemination and theming on SMP.

The US Democratic Resilience

In the following, the respective hypotheses for the USA are checked for plausibility with regard to their explanatory power for the expression of the dependent variable.

Polarization.

The two-party system of the USA, which is linked to the principle of majority voting, promotes a strongly bipolar debate culture and thus an often confrontational starting point. On the other hand it does The Winner Takes It All- Principle of the right to vote a presidential election that is in principle highly competitive. Due to the constitutional order based on the separation of powers, coupled with numerous veto points, the often hardened front between Democrats and Republicans repeatedly leads to a political blockade (Kenworthy 2015, p. 16). In addition, the parties are increasingly bringing the permanent election mode into Congress. Especially if the presidential office and the majorities in Congress are not provided by one party (Helms 2017, p. 62), which was the case under former US President Barack Obama until the end of the 2016 election campaign.

From a statistical point of view, the degree of polarization in the USA has increased from approx. 40% under Eisenhower to approx. 70% under Obama.Footnote 9 This high degree of polarization between the parties and the supporters provided the ideal breeding ground for the Russian disinformation campaigns on SMP and in the interests of the DNC hack. A lower degree of polarization could have weakened their effects. However, the main reason for this was the populist, less topic-oriented character of the discourse: the Trump camp in particular did not shy away from personal defamation of the political opponent, which - driven by the alt-right movement - continued seamlessly in social networks (Hawley 2017).

Due to the apparently targeted exploitation of the US weaknesses identified by Russia, both H1 and H1a in the case of the USA can determine both the type and extent of the Leaks as well as the disinformation campaigns. H1b can also be largely confirmed: surveys by the Gallup Institute in September 2016 confirm the steady decline in trust of the US population in their traditional media landscape until the election: While in 2013 around 44% of those surveyed stated that they largely trust established journalists in the middle of the election campaign, it was only 32% (Swift 2016). At the beginning of the election campaign, the alternative, populist media channels were very active, especially on SMP. In the USA, the established media representatives enjoy many constitutionally guaranteed freedoms, but the increasing mass of onlineHobby journalists always under pressure. This, in connection with the losses resulting from lower advertising income due to the decline in sales figures (Meola 2016), had already led to the competition for the most clicks tightened. As a result, there was often little reflection on how leaks were dealt with, even by established media representatives. These were important success factors, especially for the dissemination of Russian content on Facebook and the content of the DNC leak. Thus, this aspect also weakened the US resilience to the respective Russian forms of influence.

Contest of choice.

In three of the last five US elections (including 2016), the difference in the popular vote was less than three percentage points, which reflects the generally high level of competition in the last elections (Tarrance 2017). In the case of both parties in 2016, at least the primaries were anything but one clear thing. However, when Clinton and Trump were finally established as the final opponents, it looked like a clear victory for Clinton for a long time: Only three of 28 polls listed in the period from the end of September to mid-October 2016 saw Trump ahead of Clinton (Villarreal 2016). Despite Clinton's apparent lead, the harshness with which both sides campaigned continued unabated until election day. Even if Clinton's apparently already certain victory could be rated as relevant for the absence of an attempted manipulation of the end result, the assumed nexus between the action of the independent and dependent variable in the sense of H2 appears less convincing than in the case of H1-H1b. Instead, potential technical vulnerabilities and the failure to create public awareness of the problem are blamed for this. The former are explained in more detail below.

Potential technical vulnerabilities.

In the run-up to the 2016 elections, the IT security infrastructure of the e-voting processes in the USA was still very susceptible to errors and hacking attacks, despite its long and controversial history. As a result of the events of 2016, the state of Virginia even switched completely to paper ballot papers again (Root et al. 2018). In 2016, with the Election Assistance Commission, the USA did not have a federal authority with sufficient powers to raise the patchwork landscape of different electoral systems and IT products used to a sufficiently high level of security. At the legal and regulatory level, this was complemented by specific failures by the Department of Homeland Security to inform the targeted states about the hacking attempts at an early stage. In addition, with Pennsylvania only one state is offered risk assessment before the 2016 elections. Although this is still not mandatory, it has been requested more and more frequently since the election of Trump (Starks 2017). Thus, this speaks for a considerable explanatory power of H3d for the partial interference attempts of Russia on the technical level.

In 2016, electronic voting machines were used in 43 countries that were more than ten years old and for which the corresponding software updates often no longer exist (Buchanan and Sulmeyer 2016, p. 14). However, in their analysis from 2016, Ben Buchanan and Michael Sulmeyer point out that the direct manipulation of a large number of votes would also have meant a considerably greater effort. The heterogeneity of the IT systems and products used can offer protection: IT expert Bruce Schneier argued against the thesis that at least a selective influence on voices in swing states could be sufficient due to a lack of technical hurdles: “If you look at the last few elections, 2000 was decided in Florida, 2004 in Ohio, […] so deciding exactly where to hack is really hard to know ”(quoted from Cole et al. 2017).

H3b and H3c are therefore assigned a plausible explanatory power for the disruptive attempts described: Due to the heterogeneity of the voting systems, the hackers did not focus on the supposedly hopeless attempt to manipulate the overall result. They also did not aim exclusively at swing states, but spread their commitment widely in order to increase the rate of infected servers and thus the extent of potential disruptions. With regard to the end result, the heterogeneous system thus acted as protection, but could not prevent disruptions precisely due to the human and technical weaknesses. Thus, H3a is also assigned a comparatively higher explanatory power than the already discussed H2 in this context.

H3 therefore offers the greatest explanatory power for the extent of attempts at influencing the election IT level: Although the numerous, often outdated e-voting components provided possibilities for disrupting them, the heterogeneous structure of the system also protected the election from a potential one Manipulation of the overall result. The last thing that will be for the US is the existence or lack of a public one Awareness discussed in relation to external electoral influence.

Creating a public awareness of the problem (awareness).

In the case of the USA, the the central failure of the government to show insufficient transparency in the public handling of the presumed Russian influence. Although the security authorities informed Obama months before the election of the suspected Russian influence in connection with the DNC, the administration shied away from public attribution. Insiders said Obama feared political interference if he were to be blamed on Russia. In addition, the government shied away from disclosing valuable sources of attribution and - in the event of a counterstrike - attack tools and thus possibly making them unusable in the future (Sanger 2018a). Within the scope of this study, the actual influence of the attribution of Russia, which was not made for a long time or was made late, can be examinedFootnote 10 as an external Black Knights cannot be assessed beyond doubt on the course of the election. Yet it reduced the resilience of American democracy precisely to the effects of the DNC hack. By not making Russian participation the subject of public debate for a long time, the government failed to mitigate its clout in good time. H4 is therefore given considerable explanatory power.Footnote 11,Footnote 12

The leading heads of the online services mainly affected responded to allegations of a lack of democratic due diligence for the most part with ignorance and defensive behavior. Therefore, only after Trump's election victory, due to increasing public pressure from corporations - and politicians, e. B. through the passing of the Honest Ads Act - initiated countermeasures such as the creation of a Facebook help center for disinformation (McNamee 2018). However, there was hardly any political regulation of the SMP to protect citizens' data before the election. This enabled external as well as national actors through the so-called microtargeting use the enormous amount of data for their own benefit and address precisely the desired target group with their messages (Papakryriakopoulos et al. 2017). Thus, in the case of the USA, H4a can also be certified as explanatory.

The German federal election 2017

Russian attempts to influence the German election campaign in 2017 were mainly found in relation to disinformation campaigns on SMP, which are described in more detail below.

Russian influence in the 2017 federal election campaign

A real-time analysis by the Securing Democracy Alliance showed that a large number of troll accounts on Twitter can be assigned to Russian influence operations before the election. They mobilized to vote for the Alternative for Germany (AfD), agitated against Islam or discredited Chancellor Angela Merkel, especially with regard to her refugee policy (Salvo 2017). In addition to posts that support AfD, Russian botnets are also aimed primarily at amplifying certain hashtags. The most striking example of this is the hashtag #MerkelMussWeg, which was increasingly used from August 2017 (Hegelich 2017). However, in this case too, pro-Russian and right-wing users continued to spread the news (Nimmo 2017).

In addition, Russian TV channels reported in a highly distorted manner about various incidents during the election campaign, such as protests at the US air force base in Rammstein, presumably with the aim of sowing distrust among NATO partners (Hegelich 2017).In this country, however, the channels are still assigned a rather limited range (Nimmo 2017). However, also actors of the so-called New rights acted as (un) witting agents like Kopp-Verlag, the magazine Compact or the AfD-affiliated weekly newspaper Young freedom (Kohrs 2016). Even if the Lisa case, which concerned a girl of Russian descent from Berlin who had allegedly been raped by migrants in 2015, was not in the actual period of the election campaign, it illustrates the logic of Russian disinformation policy, but above all underlines Russia's assumed interest in interference in principle also in German matters.

Germany and its democratic resilience: Who is afraid of russian bear

Do the hypotheses put forward also explain the different degrees of Russian influence in the case of Germany?

Polarization.

Overall, the political polarization in Germany is classified as much lower than it is in the USA (Stelzenmüller, in Rokahr 2017). With the electoral system of personalized proportional representation, the political system differs considerably from the majority electoral system in the United States. Here, a greater representativeness of the parliamentary balance of power and the promotion of smaller parties through the 5% hurdle, but at the same time protection against Weimar conditions the goal (Schmidt 2016, pp. 45–47). For elections this means, on the one hand, more winners and losers than in the USA, and on the other hand, due to the need for subsequent coalition negotiations, power relationships are often more unclear.

In addition, the spectrum of parties in Germany - despite decreasing demarcation features of the Social Democratic Party of Germany (SPD) and the Christian Democratic Union of Germany (CDU) (keyword political apathy) - much more diverse than in the USA, which means that a broader spectrum of topics can potentially be covered (Woyke 2003, pp. 480–481). Also in response to this increasing convergence of the Popular parties In 2013, the AfD gained considerable influence and a large following, especially in the east of the country until the start of the election campaign (Heimbach 2016). Their radicalized tone was primarily evident in social networks, their most important communication medium. That even before the election topics like hate speech moved into the focus of German politics, is to a not inconsiderable part of the populist form of discourse among AfD supporters directed against refugees, migrants and Chancellor Merkel (Rosa, in Sapper and Kaspar 2017).

For Germany, H1a can at least be justified in relation to the identified disinformation campaigns: Due to the polarization of the online discourse, which had already increased before the election, the corresponding lines of conflict were already established for Russian fake news during the election campaign period - e.g. B. Right to asylum against xenophobia. However, that it does not become a species Merkel leak came, is mainly due to the discussed, fundamentally lower polarization of the German system. Not only did the institutional characteristics of the populist voices moderate more strongly than in the USA, but also the still larger ones IssueBasedness of the German election campaign. Despite the shift in discourse caused by the AfD, even personal defamations of the candidates, for example Merkel's in connection with the refugee crisis, were essentially based mainly on the topics that dominated the election campaign (Kortes 2019). The separation of roles between politicians as private persons and officials appears to be more pronounced in Germany than in the USA.

With regard to H1b, there was also the fact that the traditional media landscape in Germany still enjoys a far greater degree of trust than before in the USA (Forbrig 2017), despite efforts to the contrary by AfD and PEGIDA supportersFootnote 13 (Keyword lying press). The DNC hack was able to achieve its nationwide impact primarily because it was also picked up by established media in the analogue sphere. Due to the limited reach of populist media in Germany, something comparable would have been more difficult here.

Contest of choice.

While in March 2017, after an initial hype about SPD chancellor candidate Martin Schulz, both candidates received similarly high approval ratings according to the ARD Sunday question, Merkel continued to increase her lead from April 2017 (largest difference: end of July 2017). Already in the late spring of the election year, hardly any observer doubted her re-election, and there was no longer any mood of change. Despite historical ties between the SPD and Russia, Martin Schulz and the SPD appeared to be no alternative to Merkel first had realistic chances of victory and Secondly due to its programmatic proximity to the Kremlin, should be hoisted into the office at all costs (Taylor 2017). The most exciting and, in the sense of the contested nature, the most polarizing question remained the actual performance of the AfD until the end. The topics they propagated, such as the refugee crisis, largely shaped the culture of debate in the print media and TV (debates). The in the sense of the made a decisive contribution to this success as agenda setter "Negative Campaigning" The party's election campaign was deliberately polarized (Schneider 2017).

H2 could thus explain why Russian intervention at the technical level on election day would not have been very promising: a manipulation of the result in favor of Schulz would probably have caused too much attention based on the previous polls. Here, too, technical aspects appear to be of greater relevance to explain resilience before technical influence. In the case of Germany, the lower level of competition in the election could explain Russia's generally lower or, at some levels, non-existent striving for influence.

Potential technical vulnerabilities.

While in the USA E and I votingFootnote 14 are now approved in many federal states, there are no comparable options in Germany. After the partial use of voting computers during the 2005 federal elections led to election review complaints before the Federal Constitutional Court (BVG), the Federal Constitutional Court (BVG) ruled against the use of e-voting in its so-called voting computer ruling. Reaffirmed by a commission of inquiry in 2010, this judgment was still valid nationwide, also with regard to the 2017 election (Scientific Services of the German Bundestag 2015, pp. 10–12).

The use of software is only permitted when the votes from municipalities and states are transmitted to the Federal Returning Office. According to statements by the regional returning officers, this is mainly the PC-Wahl software from the manufacturer Vote IT. However, they also stated that they often do not know exactly what tools the municipalities are using to transmit votes. According to the Chaos Computer Club in September 2017, the mostly very outdated software was characterized by a large number of potential points of attack for manipulating the at least provisional election results (Gruber and Horchert 2017). The Federal Returning Officer stated, however, that the result, which was quickly made public on the evening of the election, was based on these processes, but not the official final result. The paper votes are still decisive for this.

Thus, due to the outdated software, H3a speaks for a lower resilience in Germany to technical electoral influence or disruption. However, the fact that there was no (attempted) influence or disruption at regional or federal level - at least according to the state of public knowledge - is primarily due to the far fewer technical possibilities in terms of H3 than in the USA (H3c). The predominantly analog voting process also moderated the greater homogeneity of the software used compared to the USA (H3a). In the case of Germany, there was a certain heterogeneity only in terms of the transmission of votes. Due to the effect of H2, Merkel's predicted large lead over Schulz, any manipulation of the final result that this potentially enabled would have caused too much attention and, as a result of studies, would have lost the intended effect (H3b). The apparently weak transparency about which communal unit uses which aids to transmit votes is also moderated, especially through the effect of H3 (H3d). Thus, in the case of Germany, H3a-H3d are more mutually dependent in interaction with H2 and their mode of action cannot be assessed in isolation from one another. However, H3 can also be checked for plausibility for Germany as an upper thesis.

Creating a public awareness of the problem.

In contrast to Obama's public attribution of Russia to Russia very late, German actors discussed the possibility of this months before the election.Footnote 15 As early as November 2016, the then President of the Federal Office for the Protection of the Constitution, Hans-Georg Maaßen, stated: “Last year we saw that the Russian side was influencing public opinion formation in Germany. This could also take place next year. And we are alarmed ”(quoted in n. N-tv 2016a). In the period that followed, politicians from various parties repeatedly made such statements: “We are experiencing disinformation campaigns that can also be preceded by attacks on the IT of the government, parliament or media companies” (de Maizière, quoted in n-tv 2016b). In addition, cyber attacks on public institutions, mainly from Russia and China, increased, according to CDU Interior Minister Thomas de Maizière at the beginning of December 2016 (n-tv 2016b).

In the situation report on IT security 2016 prepared jointly by the Federal Ministry of the Interior and the Federal Office for Information Security, reference was made to the cyberspace threat identified primarily from Russia and Asia and the increased alertness of the authorities in the run-up to the election was highlighted (BSI 2016) . Not only the DNC hack served as an empirical reference event, but also the Bundestag hack from 2015, for which Fancy Bear was also held responsible (Beuth et al. 2017).

In the sense of H4, one can speak of a proactive attitude on the part of the relevant actors to raise public awareness of external electoral influence.Footnote 16 It is assumed that this situational preventive measure contributed to preventing measures such as doxing in Germany due to the lack of a surprise effect.

With regard to Germany's resilience to disinformation campaigns, the significantly higher level of data protection than in the USA should be mentioned, which is, for example, political microtargeting difficult (Papakryriakopoulos et al. 2017), but not impossible, especially for third parties. The political parties in Germany agreed in the run-up to the election to forego this means, only the AfD reserved the option (Brien 2016). As in the USA, freedom of expression is strongly protected, but only within the framework of existing laws: Hate speech was thus already addressed much more strongly by politicians during the election campaign than in the USA. The legislative initiative for the Network Enforcement Act, which was passed on June 30, 2017, illustrates the efforts of the Federal Government to also prevent developments such as in the USA at the legal level (German Bundestag 2017).

Another countermeasure was the offer by the BSI to the ten leading parties, so-called in the election campaign Penetration tests on their online systems, which was also used by everyone (Schwirtz 2017). Thus, H4a can also be assigned a relevant explanatory power, in the sense of democratic resilience to measures such as doxing, but also to an even more pronounced disinformation policy of Russian actors on SMP, which existed but was far lower than in the USA.