Keywords

1 Introduction

This paper focuses on global trolling, as it manifests itself on YouTube in the form of videos with overt satirical political content. While early research focused attention on deviant behaviors and malevolent trolling, more recently the focus shifted to satirical, ideological, collective, and political trolling from countries around the globe [e.g., 21]. Some scholars studied motivation behind trolling behavior [e.g., 1, 20], while others focused on the perceptions and reactions to trolling [e.g., 5]. Only some prior research on trolling can be generalized, because forms of trolling and types of perpetrators have diversified with time. One understudied aspect of trolling is its global reach.

There is research on specific cases of trolling from various countries around the globe, including USA [e.g., 6], UK [e.g., 23], China [21], Israel [20], Italy [2], New Zealand [15], and Russia [27]. Collectively, these studies raise a question about the extent and nature of trolling globally. Specifically, it is unclear to what extent motivations, perceptions, and reactions to trolling behavior differ from one country to another. One would expect both similarities and differences in global trolling, because research suggests that trolling behaviors differ from one socio-technical context to another; different communities and different platforms interact differently with online trolling. Furthermore, it is also unclear if global trolling can involve more than one country. It is possible, for example, that shared motivations or ideology can bridge over national, socio-cultural, and geographical boundaries on online platforms.

Studying global trolling seems to be timely and necessary because of the rise in media accounts of Chinese and Russian trolling, and the proliferation of the “state-sponsored trolling” phenomenon [e.g., 17]. Specifically, there is a need to address questions about global trolling, such as: What are the signs of and motivations for global trolling? What tactics and behaviors characterize global trolling? How do these resemble other trolling manifestations? How does culture impact global trolling events?

Further, research on collective intelligence processes mostly ignored the impact of trolling or national culture on these processes. Collective intelligence, as an umbrella term, refers to the knowledge that emerges from people collaborating, and it is discerned as shared intelligence. Levy [12, p. 13] defines it as “a form of universally distributed intelligence, constantly enhanced, coordinated in real time, and resulting in the effective mobilization of skills. …[t]he basis and goal of collective intelligence is mutual recognition and enrichment of individuals rather than the cult of fetishized or hypostatized communities.” He argued that technology enhanced the ability of collectives to pool together their knowledge through interactions. Collective intelligence fuels the most advanced inventions and is a powerful tool for innovation and growth. An applicable and known example of collective intelligence playing the role of a powerful tool is the telephone; it was ideas across generations that allowed for the evolution of the telephone. More recently, online platforms allowed for and capitalized on ideas such as wisdom of the crowds [e.g., 22, 24]. Scholars studied these processes and examined the outcomes of mass collaboration in knowledge creation, for example, by focusing on Wikipedia [e.g., 7, 25]. In the broadest of terms, research on collective intelligence needs to be done in order to seek understanding of human behavior in general, and in particular in trolling instances, or across cultures.

We designed a study that analyzes YouTube videos from sixty countries, as a case of global trolling, in order to address 2 research questions: (1) What are the features of this global trolling event? (2) What are the collective intelligence attributes of this global trolling event? This study is unique and timely in that it focuses on global and ideological trolling, analyzing internet videos as a medium of trolling, that together exhibit the outcome of collective intelligence process.

2 Background

There is no consensus on the definition of trolling or what even constitutes it; trolling behavior ranges in manifestations, meanings, contexts, and effects. Trolling is defined here as [5, p. 6]: “a repetitive, disruptive online deviant behavior by an individual toward other individuals or groups”. However, because we are focusing on the “America First” event, countries rather than individuals are at the center of our trolling attention. While early research focused attention on deviant behaviors and malevolent trolling, the focus has now shifted to satirical, ideological, collective, and political trolling from countries around the globe [2, 21, 27]. With regards to this paper, studies of collective, ideological and satirical trolling cases of American, Chinese and Russian trolls are most applicable [2, 3, 21, 27]. Trolling behaviors include provocativeness, intentionality, repetitiveness, pseudo-sincerity, and satire [19, 20]. Trolls can employ specific tactics to be provocative in specific situations [3], for example, through various outrage tactics such as lying, name-calling, insulting, or simply through the use of vulgar language [4].

Satire and ideological trolling are of particular interest, given that they seem to be a ubiquitous part of online interactions [3]. Humor “is defined as an amusing social experience that “benignly” violates norms” [13, p. 3], and trolls are known to violate community norms malevolently [20] or use community norms satirically to promote their ideology [3]. While humor involves appropriate violations of communication norms, malevolent trolling involves aggressive and inappropriate violation of norms. What is appropriate is subjective and varies across cultures [13]. This context-dependent nature of humor becomes even more complex when considering also the sociotechnical context of this global trolling [18] in which the America First event took place. Unfortunately, there is no cross-cultural research on trolling [5] or satire trolling, and very little cross-cultural research on humor [13]. However, it was found that in collectivistic culture (China, Korea, and Thailand, for example) humor was used for group bonding and individuals used self-deprecating humor, while individuals in individualistic countries (such as Canada, Germany, and US) were more likely to use self-enhancing humor [13].

Research on collective intelligence attracted much scholarly attention. Nickerson and his colleagues focused, for example, on the effect of exposure to original ideas in crowdsourcing ideation [26], reuse for customization [11], and crowds designing an object collectively [16], applying theories of idea exposure and idea generation [11], knowledge and reuse for replication and innovation [16], and combinatorial conjecture of creativity and human conceptual combination [26]. They found that crowdsourced idea generation and the exposure to an original idea may affect participants’ motivation and their cognitive process. People are more likely to generate original ideas and filter out unoriginal ideas; however, it is often difficult for people to build on an original idea [11]. In another study they found that metamodels are more likely to be reused than models; designs that are generated from metamodels are less likely to be reused than other designs; metamodels will exhibit amplified reuse when created by members with higher levels of community experience; finally, metamodels are more likely than models to lead to designs similar to themselves, and therefore are less likely to lead to dissimilar designs [11]. In a third study, they found that a third-generation design of a chair was deemed more creative than the first, and features of the chair had been added, inherited and modified across the generations; essentially, crowd-based design processes are effective and highly encourage more creativity [16]. It is still unclear how people that participate in collaborative actions online can be productive or destructive, and which tactics are more constructive. It is also unclear how features of the collective outcome (be it a product, a design, an article, a video, or an answer) move through the generations of the combination of ideas process. Further, it is unclear what type of ideas should be attended to when it comes to generating good ideas by others. Above all, there is still much that needs to be addressed in this context, such as how creativity evolves across different cultures on a global-scale after being exposed to an initial and original source.

3 Method

In order to address the research questions, we choose a global trolling event that was reported in Vanity Fair on February 5, 2017 by Laura Bradley, who wrote an article titled “Europe is Trolling Trump”. What began as a European trolling phenomenon turned quickly into a global trolling events with dozens of countries involved from around the globe.

3.1 Data Collection

Data, in the form of brief videos, was publicly available online. Using a snowball method and following an initial sample from Bradley’s article (February 5, 2017), data was collected between February 6–February 24, 2017 on four separate dates. Only publicly available videos were captured and saved as files on shared folders for future analysis. We collected 100 videos and analyzed a total of 60 videos from various countries. Included in our sample are those videos that were published in February 2017 with the repeated theme, “America First [county name] Second”. Each video provides a parody version on Trump’s inauguration comment “America First” and then typically included humorous reasons why that country should be considered second.

Sixty out of these 100 videos were then uploaded into Nvivo 12, a software for qualitative data analysis. The 40 videos that were not included in our sample are those that represent regions, such as Europe or the Muslim World, and unrecognized countries and other entities, such as Mars, Westeros, Commander Geek, or Teen responses. All the videos are in English and they all begin with a variation of the statement “this is a message from the government of [country name]”. Typically, there was only one video per country, and in cases where there were 2 videos from a single country (India and Israel), we included only the first video that was published from this country. This resulted in 60 videos, each from a different country.

3.2 Data Analysis

Based on content analysis we addressed the first research questions, while comparative case analysis addressed the other question. At the time we completed our data analysis many of the videos had been removed and were no longer available online. A coding scheme was developed from the data, using an iterative process of coding and discussion among the three coders. Each code was described and an example was provided to ensure coding reliability; codes with frequency of less than 10 instances were removed, as they were not significant enough for further analysis. Codes were grouped into four broad categories: trolling behaviors, trolling tactics, structural codes, and content. The unit of analysis for categories (trolling behavior, trolling tactics, and structural codes) was the video as a whole. For the content category, the unit of analysis for coding was 15-s intervals; coding involved assigning codes to each 15-s segment of the video. Two coders coded the data and intercoder reliability test was conducted on 10% of the videos by a third coder. Intercoder reliability was high at 91.4% with a Cohen Kappa of K = 0.829.

While we noticed in the content analysis some trends with the appearance of new codes over time and disappearance of others, we could not easily parse out trends over time, and a more nuanced analysis was required. We then wrote a brief case synopsis for each video in each stage; these were about half a page long each. Through continued comparative analysis between cases we identified themes that are common across the cases in a given stage, based on which we conducted a comparative case analysis between stages [14]. The themes that emerged include, for example, opening statements (e.g., “This is a message from the government of [insert country’s name].”) and closing requests (e.g., “We totally understand that it will be America first, but can we say [country name] second”). Tracing these themes through the five stages reveal a few that remained constant, such as the use of vulgar language, and others that varied from one stage to another, such as the nature of references to other countries.

The sixty videos ranged in length from 1:46 min (Russia) to 12:19 min (Germany), with an average of 4:07 min. There was a significant correlation between the length of video and country rank on Hofstede’s Individualism/Collectivism (r = −.66, p < .05) and length of video and country rank on Hofstede’s Indulgence dimension (r = −.29, p < .05) [10]. Collectivist cultures are considered to be less direct (more indirect) and less succinct (more elaborate) compared with individualistic cultures, which might explain why collectivistic countries had longer videos than individualistic countries [e.g., 8, 9]. Furthermore, longer videos were correlated with countries that suppress individual gratifications and regulate it through strict social norms. A little over half of the videos (52%) included an English narrator, most of the videos (84%) included subtitles in English and/or their local language, a closing request in the majority of videos (93%) to have their own country second (or even tenth in one instance), and half of them (52%) included an introduction that puts that video in the context of a local satirical TV show, in their own local language.

To address the second research question, we grouped the videos into 5 stages, based on assumed video publication date, to examine global collective intelligence change over time. We grouped countries into one stage solely based on our data collection dates. The first stage included the six videos that appeared in the Vanity Fair article on February 5, 2017 (stage 1); the second stage included the eleven additional videos that the Vanity Fair article added by February 6 (stage 2). Then we collected video through Google and YouTube searches on three occasions: eight videos on February 9 (stage 3), twenty videos on February 12 (stage 4), and fifteen additional videos on February 24 (stage 5).

Table 1 provides the names of the countries in each stage, along with average video length by stage. As we can see in Table 1, average video length varied between stages; yet there was no noticeable pattern of evolution, except that the range in the first stage showed the largest variety in length.

Table 1. Countries, regions, and length of videos by stage.
Full size table

4 Findings and Discussion

To gain a better understanding of the global trolling phenomenon, we present our findings and discuss them under two sections, each addressing one research question: (1) To what extent is the “America First” event a case of global trolling? (2) What are the collective intelligence attributes of this global trolling event across and within countries?

4.1 Trolling Trump as Global Trolling

To address the first research question, we identify the nature of this global trolling event by describing the trolling behaviors and tactics we found in our data and examining them in light of online trolling research.

Overall, we found that the most frequent codes are the main categories, with references to Trump (#198) and the use of Trump language (#193), as well as to trolling behaviors (#157); these appeared in all the videos and were coded more than once per video. Specific codes that appeared frequently include references to the culture of the (video) sponsor country (#190), and references to another country (#129), as well as the use of the hyperbole trolling tactic (#114), in which the video exaggerates one’s strengths or another’s weaknesses. Clearly the frequent references to Trump are unique to this case study and are expected as this is the subject of the videos. The frequent references to sponsor country and to other countries in all the videos is indicative of the global and international scope of this event. Finally, the common utilization of trolling behaviors, and specifically the hyperbole tactic in the videos, supports the argument that this event is an instance of (global) trolling.

Furthermore, typical trolling behaviors [19], such as repetitive, provocative, pseudo-sincere, and satirical trolling behaviors, characterize all sixty videos. Holistically, the videos exhibit a repetition of satirical provocation by mocking Trump’s inauguration speech and his “America First” campaign. The repetition occurs not only across videos, but also within a single video. The most extreme manifestation of repetition is in the China video that involved nothing but clips of Trump’s repeatedly saying “China” at various speeches and interviews. Similarly, a repetition of clips of Trump referring to “Denmark” appears as part of the Swedish video. This repetitive trolling behavior by all contributors continued for a three-week period, resembling the repetition pattern of other events, such as the case of Chinese collective trolling [21], or state-sponsored trolling activities [17]. The duration of the Chinese collective trolling repetitions was a few days, but state-sponsored trolling can last longer than a couple of weeks, with several peaks. The “America First” event continued for about three weeks. This might be because the production of video is significantly more time-consuming than simply posting a text message or image on Facebook, Twitter, or Weibo. Another possible explanation is that a global spread of an idea takes longer as it crosses geographical, national, language, and cultural boundaries.

As can be seen in Table 2, the “America First” global trolling event involved typical trolling tactics [3]. The most common tactics include the hyperbole arguments (23% of all tactics), exaggerating the weaknesses of either Trump, US, or other countries, or the strengths of their own country.

Table 2. Frequency and percent of trolling tactics.
Full size table

A typical example appears in the Sweden video, as the narrator, using Trump’s voice and intonations, argues: “Sweden is the best country of all of Europe. Better than the Netherlands, better than Switzerland, and especially better than Denmark.” This was followed by the derailment tactic (15%), which involved leading the conversation off track by latching onto an unimportant detail. The narrator started with an articulation of Sweden’s strengths, when mentioning Denmark, but switched into making insulting comments on the Danish people, saying that they are the “Mexicans of the Scandinavia”, and then included random clips of Trump saying “Denmark” at his various speeches, jumped into the “nuke Denmark” comment, mentioned IKEA, and finally followed with comments on the Trump organization, meatballs, furniture, and the wall bordering Mexico. Another common tactic was the politeness tactic (14%), which involved the use of “thank you,” “please,” or honorific mention, such as “Dear Mr. President”. The repetitive politeness mocks the president, but also addresses him with appropriate honorific; in this context, it adds a sarcastic tone in the opening of all the videos.

Other trolling tactics found in our analysis included insulting (9%), lying (7%), sarcasm (towards others (6%) or the US (6%)), misappropriation of jargon (5%), swearing (4%) and personal attacks (4%) (Table 1). These trolling tactics resemble those identified in individual satire trolls’ posts [3]. While each tactic on its own may not constitute trolling, the amalgamation of tactics and trolling behaviors, repeated over and over again, does.

Thus, we conclude that the “America First” case of global trolling resembles individual trolling behaviors and tactics, with global manifestation. Global trolling is less personal, but more ideological and collective.

4.2 Collective Intelligence in Global Trolling

We describe here the global trolling collective intelligence process based on our comparative case analysis, through five stages, addressing the second research question. Trolling Trump as a global trolling event evolved over a three-week period following Trump’s inauguration speech and involved videos from 60 countries. We identified five stages in which the videos were posted. Each stage featured patterns of references that evolved between one stage to another and an evolving format of the closing request. At first, videos followed a relatively simple format of “America First, ‘X Country’ Second,” as shown in the Netherlands videos, which was the first one posted. Then, in later stages, many renditions of this request appeared, including, for example: “America first, Netherlands Second, but can we say Lithuania third” (Lithuania, stage 2), “Unlike other countries we are not aspiring to be second, but 51st” (Czech Republic, stage 3), or even “Israel First, America Second” (Israel, stage 4). The same pattern was observed also on the eleven videos that emerged within regions in Germany. The Germany region videos, all in the 4th and 5th stages, displayed the same format of the opening and closing requests as other videos in these stages. Besides the evolution of the opening and closing request, three significant references were noticeable: (1) China; (2) the gay community; and (3) Muslims/Arabs. None of these references appeared in the first stage, but they start to appear in stage 2, for example, with reference to China in the Luxembourg video. An increasing number of videos began to incorporate these references, to the point that in the fifth stage, 10 of the 14 videos included at least one reference to China, the gay community, and Muslim/Arabs.

The opening statement and closing request are important components of this global trolling event. The closing request starts as a simple remark and becomes progressively more intricate across the stages. The videos in stage one simply stated “We totally understand that it will be America first, but can we say [country name] second”; only a couple of exceptions diverged from this format. One exception is Germany, which in addition to this statement introduced a joke that would be repeated numerous times later, suggesting “If you were wanting to push the red button on us, then this is where we are located on the map” and showing a map where they highlighted Italy instead. Another variation in closing request in stage one appears in the Belgium video, saying “We totally understand it is going to be America first, but can we say Belgium second, or tenth… we don’t care.” The rank order (and name) of countries included in the closing request attracted much attention in videos in later stages. Both these notions help show how collective intelligence is upheld within all of these videos across stages, perhaps because they build into the simple satirical statement a complex statement about international relations (which country is their enemy, which country has an important role in global affairs), and mock Trump’s impulsive behavior.

In stage two, most of the countries continue the pattern set up by the first group of videos, ending their videos either with the simple request, or with one of the two variations that were introduced in the first stage. For example, Luxembourg, Morocco, and Italy all included the bombing joke after their closing request. Again, a couple of videos here diverted in a way that affected videos in later stages. Specifically, Lithuania requested, “We totally understand if it is America first, Netherlands second, but can we say Lithuania third.” For the first time, the video directly referenced the Netherlands, the first country to post a trolling video, and by doing so, parodically gave them priority in global affairs.

In stage three, these specific variations on the closing request continued, while a few elaborated further. One video, from Iran, added to the intricacy of global affairs and referenced several countries that produced videos in prior stages, in their closing request, asking to be only ahead of their primary enemy, Iraq, stating, “We totally understand that it is going to be America first, and a lot of other countries already claimed to be second. We can’t compete with Netherlands, Switzerland, Portugal, Denmark, or other European countries, but how about Iran before Iraq”. The Czech Republic, in another video, took their closing request a step further by stating “not aspiring to be second, but 51st of the United States”. This exact joke was then repeated in later videos from Greece and Ireland, setting up a new trend in which various countries claim to became part of the United States, instead of competing for their rank globally.

In stages 4 and 5, the videos followed the general format of the closing request as in the previous stages but incorporated new ideas such as Israel (stage 4) claiming to take precedence over all the countries and to be ahead of the United States, showing the seal with “Israel first, America Second.” Egypt (stage 5) decided to outline the United States as a part of its bombing joke instead of a neighboring country. In the earlier stages the videos would conclude with the request “America First, X Country Second”, but in stages 4 and 5 it became a point to add additional comments after it. For instance, the United Kingdom (stage 4), after the closing request, added comments such as “Your greatest friend… seriously we love you… I need to tweet this…” and continued along those lines for 15 s after the request.

Referring to another country and to other videos was not limited to the opening and closing statements. One common reference was to China, typically in an insulting tone. There were none in stage 1, but by stages 2 and 3, there were brief references to China by Luxembourg and India, and in stage five, 7 of the 12 videos incorporated a type of insulting or degrading trolling comment towards China. Mocking both Trump’s and the other counties’ fixation with China is the video from China in the fourth stage, which only included a repetition of one word “China”. Another common reference to other countries in the videos involved direct reference to early videos and their respective countries, by either insulting the country or citing their video. In stage one, all the videos insulted the Netherlands, as did many of the later stage videos as well. Insulting the Netherlands developed into a relevant feature across all stages of the videos. In stage five, Armenia directly quoted Turkey’s video from stage four with the comment of “the country not the bird” and also directly quoted Bulgaria’s video (stage three) of them “wanting to be tenth.” There were also many instances where countries from later stages would insult countries from previous stages in their videos such as Iran (stage three) insulting India (stage two). Iran’s closing request is also another example of this; they stated “We can’t compete with Netherlands, Switzerland, Portugal, Denmark, or other European countries…” which refers to all the videos in the first stage. With this occurring, ideas can be directly traced from source to source.

Unlike the closing request, which became more complex with each stage, until it reached stagnation by stage five, the general format of the opening request did not change much until the fifth stage. The format of the opening statement is as follows: “This is a message from the government of [insert country’s name].” In stage five a variation was introduced by the Hong Kong video, saying instead “This is a message on behalf of Her Majesty’s request because the government of Hong Kong is too busy right now.” This introductory statement criticizes their own country, and as such it became a way to excuse late adopters, who joined the trolling game later. Syria’s video followed that same presentation by stating, “This is a message on behalf of all people from Syria because the government is too busy.” At this stage, Taiwan had also changed their opening request to say “This is a message from the government of China, but you can just call us Taiwan,” turning its opening request into a satirical comment on a controversial international affair.

Collective intelligence in global trolling resembles and enhances prior research in other settings, media, or platforms, and other creative processes and outcomes. As others have shown, creativity in mass crowd production is a result of combination of ideas [16], as well as the influence of the original idea [26]. In this global trolling event, the evolution of videos over time picked ideas from the first videos, early videos, and from later videos, combining them into new creations of videos. The impact of the original idea, in the form of the video from the Netherlands, continued to influence the later videos in content, style, and ideas that were incorporated into all the videos. The impact of this first video was evident in subsequent videos not only through the replication of ideas, such as opening and closing requests, but also in making direct reference to the video. Furthermore, in global trolling we identified similar processes of replication, innovation, and customization as others have discussed in the context of Thingiverse 3D models, for example [11]. The use of Trump voiceover in the videos illustrates replication that was adopted by video creators across the globe throughout the various stages of the global trolling event. At the same time, the closing and opening remarks were customized for each country and each region with specific manifestations, as well as stage-related variations. Innovation occurred over time, not only through country-specific customizations, but also in the introduction and through the evolution of major references to gay people, China, and Muslims.

This global trolling event, using YouTube videos, while resembling collective intelligence processes, results in a different outcome – trolling on a global scale while using videos. The repetitive nature of ideas in this case is the backbone of trolling. Sharing ideas in videos, across the YouTube platforms, triggered both shared creativity and pooled skills of video creators, from 60 countries across the globe, enabling them to join the event, by posting their unique creations on YouTube. Together they troll and their trolling videos creation are more sophisticated over time. Cultural and geographical differences became instrumental in the construction and evolution of trolling globally; instead of barriers to knowledge sharing these differences became the building blocks of collective intelligence.

5 Conclusions

This study examined a global trolling event, “America First,” with the intention to identify whether global trolling exists, and if so, what trolling behaviors and tactics characterize global trolling, and what are the specific cultural manifestations of global trolling. We found that this is indeed a case of global trolling, exhibiting repetitive, provocative, pseudo-sincere, and satirical trolling behaviors, across all videos, regardless of sponsored countries. The study also examined collective intelligence processes; we found that similar to prior research, processes of innovation, replication, and customization were evident, as well as clear impact of the original idea, from the first Netherland’s video, and throughout the five stages. We found support for these processes in video creation, while prior research focused on text or designs; across geographical, political, and national boundaries, while prior research largely disregarded these boundaries; and in online settings with naturally evolving creative processes that spans three weeks.