Addressing the Methodological Challenges that Cloaked Profiles Pose to Digital Observations

Abstract

Although digital ethnographic studies concerned with online misinformation have focused on analyzing the contents shared by “cloaked” profiles (concealed or fake identities), less attention has been given to the epistemological and ontological dilemmas that cloaked profiles pose to digital ethnography. This article deals with these issues by asking: how can digital ethnographers determine who and what we are observing? And how can we conduct online observations when confronted with cloaked profiles? Drawing on field research, this article argues that researchers would benefit from including more critical reflections on the presence of cloaked profiles and learning how to apply digital skills for how to unveil cloaked profiles. Such practices will challenge a commonly accepted ontology that online profiles represent human behavior and enhance researchers’ digital literacy and ability to recognize cloaked profiles. Finally, applying techniques to unveil cloaked profiles will arguably strengthen the hermeneutic process of knowledge production in digital observations.

Introduction

With the rise of internet technology, anonymous, manufactured, and nonhuman online user profiles have emerged. Accordingly, there has been a sharp increase in online sources such as pages and profiles used for the purpose of organized social media manipulation (Bradshaw and Howard 2019; Hao 2021). Moreover, the anonymity and disembodiment that internet technology offers allow people to carry out deviant behavior online—that either challenges commonly accepted social norms or regulations (Jewkes 2002) or contradicts authorities will or legislation through protest movements (Tufekci 2017). Furthermore, online sources with feigned or concealed identities can have aims of spreading hate speech and propaganda (Daniels 2009, 2013) and affect democratic political discussions by creating false impressions that a particular opinion has widespread public support (Bessi and Ferrara 2017; Zerback, Töpfl, and Knöpfle 2020). These facts have instigated epistemological questions on knowledge production in digital social research (Marres 2017). Digital ethnographic studies have pointed to epistemological challenges of how to conduct research when acknowledging the presence of cloaked online sources (Daniels 2009; Farkas, Schou, and Neumayer 2018; Kavanaugh and Maratea 2020).

Arguably, digital ethnographic methods have focused on directly observable misinformation and fake content shared by online profiles (Shapiro et al. 2021). Less attention has been given to the epistemological and ontological dilemmas that arise with the presence of cloaked profiles. The ontological dilemma concerns how we can know what and whom we are observing. The epistemological challenge concerns how to conduct research that makes claims of knowledge when acknowledging the presence of cloaked profiles. In what follows I draw on examples of my fieldwork where I used digital observations to conduct research on migrants without legal visa documents and their support network of European citizens. I use examples from my own study to illustrate the ontological and epistemological challenges that cloaked profiles pose to digital ethnography. First, I show that the preliminary research of my fieldwork uncovered a presence of cloaked profiles which created ontological dilemmas of how to know who and what we are observing. Thereafter, the focus of the article turns to the epistemological dilemmas: How can digital observations be conducted in a digital sphere when the observed entities might be cloaked? Because there is a scarcity of literature within digital ethnography that consider how to practically meet the challenges of cloaked profiles in a methodological design, this study contributes to the field by raising such discussions. This is done by assessing existing digital tools that can assist researchers in unveiling cloaked profiles. Thereafter I propose techniques that might assist researchers who seek to unveil cloaked profiles. Arguably, learning how to apply digital techniques for how to unveil cloaked profiles will enhance researchers’ digital literacy and ability to recognize cloaked profiles.

Marlow, Miller, and Roberts (2021), as well as reports from social networking sites (Hao 2021) have documented increases in cloaked online profiles over the past decade. Although reports from companies such as Meta are difficult to verify due to lack of transparency, the company reported that in 2021, approximately 5% of monthly active users were “false accounts” (Meta 2022). Both Twitter and Facebook have stated that they aim to reduce the number of fake profiles by using technology to uncover them and having other users report suspicious profiles. Nevertheless, this process has proved challenging. Even Twitter profiles marked as “verified,” and thereby promoted as authentic by Twitter, were later revealed to be fake (Porter 2021). Moreover, companies such as Meta aimed to prohibit “fake” profiles by introducing “real name policies,” meaning that those who wish to register a profile can only do so by using their “real” legislative name. Such policies have been critiqued for hindering anonymity, arguably a formative feature of digital culture (Marres 2017).

In what follows, the commonly used terms such as fake, false, misinformation, and disinformation (Kalsnes 2018; Rogers 2013), are avoided. First, these terms suggest references to early studies of the online world which separated the virtual from the “real” physical world. This article follows the now common understanding of internet research as not separating between the “real” offline sphere and the online sphere (Rogers 2009). Second, terms such as “fake news” and “misinformation” are socially constructed and politically charged which may be used as discursive instruments (Farkas and Schou 2020). Third, even though profiles are created under a false identity, their shared content or claims may not necessarily be misleading. For instance, whistleblowers or protesters may see it as necessary to cloak their profile to share information. Hence, there is a difference between the verification of a statement that a profile makes and the verification of the profile itself. Although there are varying definitions of “social bots” and “bots” (Boshmaf et al. 2011; Yang et al. 2017), there are different forms of bots, some driven by algorithms and others by human interventions (Cresci 2020). Profiles and websites made by anonymous users or people who represent themselves as avatars, under fake names, instigate the same issues as automated bots: they are inherently about identity and the types of data social scientists claim to observe.

Within studies that have a human-centered ontology, meaning that researchers assume that the observed profiles are human (Marres 2017), it would make sense to refer to the observed profiles as informants or actors. However, this article seeks to challenge such an ontology by acknowledging that profiles may be more or less human-made and apply the terms “observed entities” and “cloaked profiles.” The term cloaked profiles draw upon Daniels’ (2009) definition of cloaked websites: “those published by individuals or groups that conceal authorship or feign legitimacy in order to deliberately disguise a hidden political agenda” (661). Although this definition suggests political motivation, cloaked profiles are in the following discussion defined as profiles published by individuals or groups that conceal authorship or feign legitimacy regardless of motivation.

This article begins by introducing the fieldwork that the discussion draws examples from, and the methodological approach of this study. Because the fieldwork encountered methodological challenges when recognizing cloaked profiles, the article moves on to discuss the ontological dilemmas that these profiles posed. Subsequently, the article explores the epistemological challenge of how to practically deal with the presence of cloaked profiles. I show that in my fieldwork, existing digital tools were found insufficient to address the challenge of unveiling cloaked profiles. Thus, alternative techniques that might assist researchers who seek to unveil cloaked profiles are presented. Last, the article addresses the challenges of applying such techniques. In conclusion, I argue that although there are additional layers to the discussions outlined in this article, the proposed approach for how to meet the challenges that cloaked profiles pose can strengthen the hermeneutic process of knowledge production in digital ethnographic studies. This is because gathering contextual knowledge of field sites, and how observed entities use digital techniques to represent themselves in different ways, will provide additional layers to our interpretations and thicker descriptions of who and what we observe.

Digital Networks of Migrants and Citizen Humanitarians

The discussion in this article draws upon methodological considerations made when researching migrants who reside in Europe without legal visa documents, and their support network of European citizens. The study set out to do digital ethnographic research within these networks made up of both migrants and “citizen humanitarians” who are European citizens who carry out support to migrants outside formal aid organizations (Pascucci and Jumbert 2021). The fieldwork for the study included observations of highly sensitive data because the act of residing and traveling in Europe without legal visa documents is unlawful. Additionally, the EU set out that states ought to criminalize those who facilitate such irregular migration (Carrera et al. 2019).

The study aimed to analyze the characteristics of support practices extended by citizen humanitarians to migrants. It was found that to observe the meaning-making and practices of the networks between citizen humanitarians and migrants, the digital sphere had to be included. This is because previous studies on citizen humanitarianism have uncovered how digital spheres are important for pro-migrant supporters to convey their views and they have largely focused on open Facebook groups (Morell 2019; Sætrang 2016). Moreover, with technology and the internet becoming increasingly accessible, studies have shown that the internet not only reshapes the migration process itself but affects other actors in the process such as support networks and those who aim to control migration (Sandberg et al., 2022). With this backdrop, the study aimed to research the networks between migrants and their helpers via Facebook groups and Facebook profiles.

Different labels are used to describe ethnographic approaches and techniques for researching digital spheres, field sites, and concepts. This article follows Hine (2015), who proposed ethnographic methods as ones for the internet and not of the internet. This entails a method for understanding the meaning-making constructed online, rather than one aiming to explore the internet itself and the goal is to capture a holistic account of the everyday meaning that the observed entities express (Hine 2015). These methods often provide detailed analyses of profiles and online subgroups’ shared content (Shapiro et al. 2021). The research applied the analytical concept of online networks to map the online environment within which these informants operate (Caliandro 2018). This type of research does not require direct interactions with informants. Instead, it involves engaging with the online field by mapping social formations, profile representations, and how informants construct meaning-making (Caliandro 2018).

The study set out to capture the observed entities’ everyday meaning-making, which relies on a process of making “thick descriptions” of the observed data (Geertz 1973). According to Geertz (1973), one can obtain thick descriptions by looking at the context and unveiling the layers that shape culture and social interactions. In practice, an ethnographer must scrutinize how the observed entities are able to express themselves and make meaning out of these observations. This process of meaning-making refers to the following two different layers of interpretation: (1) conveyed by the observed entities through their actions and (2) established by the researchers through their observations. This latter form is the focus of this article. According to Goffman (1959), individuals present themselves in certain ways through impression management, depending on the setting. In digital observations, profiles can be seen as acts upon the digital stage, where usernames, pictures, language, and different kinds of markers are used to manage an impression. Relevant to this study is how the digital sphere offers tools to construct an identity for those engaged in behavior that is not within the social or legal norms (Maratea and Kavanaugh 2012).

Arguably, internet technology has provided a safe space for connecting, as it allows for anonymous and disembodied social interactions (Durkin, Forsyth, and Quinn 2006; Jewkes 2002). As the available digital tools to construct profiles are continuously evolving along with technological advancements the possibilities for hiding identities can contribute to democratic practices, allowing activists and protesters to communicate and organize in digital spheres (Joyce 2010; McGarry et al. 2019; Tufekci 2017). Zeynep Tufekci (2017) has shown that many activists with cloaked profiles have had their profiles deactivated due to Facebook’s “real name policies.” Tufekci argues that cloaked profiles that draw less attention and are not clearly political, will not be targeted by such policies to the same extent. In fact, 20% of college students who participated in one of Tufekcis surveys, admitted that they use pseudonyms on Facebook (Tufekci 2017, 143). It should be noted that the discussion that follows in this article does not concern the informants’ motivations for concealing or constructing identity. Rather, it looks specifically at how researchers deal with the methodological challenges that cloaked profiles pose to digital ethnography.

Becoming Aware of Cloaked Profiles

The preliminary research phase in my study of citizen humanitarians and migrants in Europe uncovered the presence of cloaked profiles within the observed network. Not only could cloaked profiles be seen with the use of nonhuman avatars as profile pictures; additionally, the potential criminal aspects of activities and topics of discussion between the profiles could suggest that some of these observed entities would benefit from cloaking their identity. The first example from my study was when I suspected cloaked profiles in an observed conversation between two profiles. One was represented as a cartoon with a constructed name, the other profile represented a European citizen with a photo that depicted a person that matched the name. Although the profile with the cartoon profile picture appeared to be cloaked because it portrayed a fictitious character, the other profile which appeared to be a European citizen could equally be cloaked due to the various technological tools that exist for users to cloak an identity. Because the study set out to conduct unobtrusive observations, it was initially not an option to connect with these profiles to verify their identity. Hence, it became apparent that in this fieldwork, the presence of cloaked profiles ought to be addressed to some extent in the methodological design.

How Do We Know What or Whom We Are Observing?

The first methodological challenge that appeared when recognizing the presence of cloaked profiles was the ontological question: How do we know that what we are observing is what we assume it to be? Even in offline ethnographies, we might be fooled by those who hide their identity or tell lies, or we might misunderstand the social interactions and stories we are told. Researchers are not omniscient, making absolute knowledge about the research objects in an ethnographic study a utopian ideal. Qualitative interpretive approaches such as digital ethnography often assume a relativist ontology, meaning that researchers believe there are multiple truths of an observed phenomenon (Denzin and Lincoln 2018). Cloaked profiles in digital spheres are connected to concepts of identity, personhood, and representation and can be seen as a fundamental challenge for all interpretive qualitative social inquiries—both offline and online.

Hence, it is important to question whether it is necessary to discuss cloaked profiles as a specific challenge for digital ethnography or whether they represent normative ontological issues found in all types of qualitative social inquiry. Arguably, the digital sphere is different, as it offers an opportunity for individuals to both present themselves and invent themselves using digital techniques (Jewkes 2002). As Markham (2005) highlighted, “the extent to which information and communication technology (ICT) can mediate one’s identity and social relations should call us to epistemological attention” (248). In offline observations, researchers use markers, such as body type, race, socioeconomic markers, or geographical location, to understand whom they are observing. Digital observations differ in terms of the researcher’s ability to interpret whom they are observing and the observed profile’s ability to be represented in multiple ways. Indeed, both online observed profiles and physically observed informants use markers to represent themselves in different ways. However, interpreting how digital profiles represent themselves is challenged by methodological obstacles in the digital sphere, associated with difficulties in connecting to the field, rapid technological advancements, and situating the context (Marres 2017; Hine 2015).

The increased presence and different ways to “cloak” profiles in digital spheres instigate questions of how researchers engaged in digital observations deal with these issues. Marres (2017) argues that in digital spheres, there exists a common human-centered ontology. This entails the assumption that we study authentic human interactions when we observe the relationships between digital profiles. Marres furthermore claim that even though we know that we are surrounded by cloaked accounts in digital spheres, “the idea that we interact as humans in digital settings is firmly upheld in public discourse surrounding social media and digital societies” (Marres, 2017, 73). In my fieldwork with citizen humanitarians and migrants, I started to challenge this human-centered ontology by reflecting on the fact that some observed profiles might be cloaked. When recognizing the presence of cloaked profiles, it also became apparent that a cloak might represent both non-human and human profiles.

Automated “socialbots” are cloaked profiles that mimic human behavior and aim to remain undetected. Socialbots have been shown to be unknowingly accepted in digital communities, particularly if users see that they have friends in common with these bots (Boshmaf et al. 2011; Freitas et al. 2015). However, unknowingly interacting with cloaked profiles is not restricted to automated nonhuman profiles. There are cases where authorities such as police and official security agencies have been setting up cloaked profiles, managed by officers to conduct investigations or surveillance. ¹ A Dutch migration scholar, Saskia Bonjour stated on Twitter that this revealed the importance of scrutinizing the profiles in one’s personal networks and looking for profiles with questionable identities. ²

Within migration studies, the challenge of cloaked profiles has been recognized by scholars who call for greater reflexivity around these issues while undertaking observations (Fischer and Jørgensen 2022). Studies conducted on online discourses on politicized subjects such as migration, nationalism, and racism have shown how these topics are prone to be infused by cloaked sources (Farkas et al. 2018; Daniels 2013). Nonetheless, when searching for examples of digital observational studies that concern topics of migration and anti-migration discourses, I found that few studies include reflections on the potential presence of cloaked profiles. For instance, in a qualitative nonparticipant observation study on anti-Islamic Facebook groups by Fangen (2021), she described what she observed as “authentic everyday conversations.” Fangens study includes excerpts from anonymous sources discussing highly politicized topics such as migration and Islam but does not discuss whether there are reasons to question the identity of the profile. On the contrary, the author pointed out that no contact was made between the researcher and the informants to ensure their anonymity. It should be noted that in many cases, such as this, it is important to ensure the anonymity of the observed entities, and it is not an option to actively seek to unveil the identity of the observed profile.

Nevertheless, conducting digital observations assuming that all profiles are representations of human subjects without considering the presence of cloaked profiles is an acceptance of a “human-centered ontology’ in online spheres. By not including critical reflection of potential cloaked profiles, one overlooks the potential nonhuman actors we might be observing. It furthermore fails to recognize the ways in which digital profiles use technological tools to represent themselves in different ways, which instigate questions of who and what we are observing. By not including critical reflections on the potential presence of cloaked profiles, we merely transfer the methodological assumptions from physical observations to the digital field—that we observe “authentic everyday conversations” (Fangen 2021) between human beings. We thereby fail to address the ontological challenges that cloaked profiles pose to digital observations. Although this article also recognizes that it is not always possible, necessary, or ethically correct to unveil profiles, researchers engaged in online observations should to a larger extent address the potential presence of cloaked profiles within datasets. Including critical reflections on these profiles in a methodology will contribute to a better understanding of what and whom we are observing in digital spheres.

Moreover, including critical reflections on cloaked profiles ought to consider how we use terms such as “fake,” “untrue,” or “misinformation.” Scholarly discussions that address the negative societal consequences of cloaked sources are closely associated with the term “post-truth.” Post-truth refers to situations where public opinion is shaped by emotions and individual beliefs rather than evidence-based facts (Ackland 2017; McIntyre 2018). However, such claims have instigated debates of what we mean by “facts” and “truth.” Farkas and Schou (2020) note that researchers should not take notions like “truth” and “fake news” for granted because “. . .these terms are deeply politically charged and constructed. They are not merely descriptions of the world, but discursive weapons used to intervene and shape social reality” (14). Thus, researchers ought to critically reflect on how they use terms such as “fake” and “inauthentic” when addressing cloaked profiles. These are important debates, particularly relevant to interpretive practices such as digital observations which assume that nothing is certain and that the researcher’s task is to interpret the observations through co-creation between the researchers own subjective interpretations and the observed (Denzin and Lincoln 2018).

How Can We Practically Meet the Challenges of Cloaked Profiles?

Although the ontological dilemmas that cloaked profiles pose to digital observations ought to make researchers include reflections on these issues in their studies—it may in some cases be insufficient. In my study of migrants and their support network, it became relevant to address the challenge of cloaked profiles in the data material in a more practical manner. First, my study included a relatively small number of entities, which made it feasible to seek to unveil cloaked profiles. Second, in order to understand what type of data the study could aim to uncover it was relevant to unveil whether observed entities were cloaked or not. Not only was I interested in whom I observed and their expressed meaning-making, but also what I observed as in the contextual layers and the entire network made up by observed profiles. Last, because some profiles clearly appeared to be cloaked, the study could not merely recognize some profiles as cloaked and assume that others were not. This would entail an assumption that all cloaked profiles are recognizable. This is arguably not the case as previous studies have shown that it is more likely that cloaked profiles appear “natural” to avoid being detected as cloaked (Boshmaf et al. 2011; Freitas et al. 2015; Tufekci, 2017). Hence, in my study, I found it pertinent to explore how to apply practical techniques for unveiling cloaked profiles.

As mentioned above, the practical ways to address the methodological challenges that cloaked profiles pose to digital ethnography have remained underexplored in the field. Thus, in my study, I first opted to explore available digital tools that could potentially help researchers who aim to unveil cloaked profiles in a digital observational study. The process is often recognized as “digital methods” and refer to the use of available “native” digital tools to conduct research. These tools use online and digital technologies to collect and analyze data and do not require going offline or transferring traditional research methods to an online field site (Rogers 2013). Caliandro (2018) has pointed out that such digital methods can provide valuable tools that complement digital ethnography.

The universities of Amsterdam and Indiana are examples of institutions that offer digital tools and methods that can aid researchers in unveiling cloaked profiles (Digital Methods Initiative 2021; Indiana University Observatory on Social Media 2021). The Observatory on Social Media at Indiana University has developed “Botometer,” “Hoaxy,” and “Botslayer” as useful tools to help discern with relative certainty if online profiles are performing automated actions, are part of a network, or if they are administered by a human (Varol et al. 2017). These tools are only applicable for investigating Twitter users, not allowing researchers who conduct observations on other types of platforms such as Facebook or Instagram to utilize them. That said, there are also tools developed to verify Facebook profiles (Albayati and Altamimi 2019; Gupta and Kaushal 2017). However, the complexity of applying these tools may lead researchers with limited knowledge of digital methods to refrain from using them.

All the tools mentioned above were found insufficient for unveiling cloaked profiles in my study of networks between citizen humanitarians and migrants. First, because they mainly focus on automated profiles, they may be more applicable to quantitative studies. Second, the digital tools mentioned here all use application programming interfaces (APIs), which are software intermediaries that allow researchers to extract data from one program to another. These tools are challenging to use because companies such as Meta have been shown to hinder researchers who use APIs to extract data from their online social platforms such as Instagram and Facebook (Perriam, Birkbak, and Freeman 2020). Hence, in my fieldwork, I opted to look for alternative ways to unveil profiles.

In what follows I use examples from my fieldwork to show that similar studies that use digital observations might benefit from including alternative digital techniques to unveil how profiles represent themselves. In other words: studies that might benefit from including practices outlined below are ethnographic digital observations of relatively small networks on Facebook that not only analyze who the observed entities are but also aim to include interpretations of what we observe online. It should be noted that the relevance of the practices discussed below will depend on the type of data and analysis in each specific research project.

An Ethnographic Approach to Unveiling Profiles

Including practices from other disciplines is common in ethnographic methods, especially when there are specific research problems that lack methodological guidelines (Ploder and Hamann 2021). In this article, a handbook for journalists concerning digital source criticism (Grut 2021) served as the inspiration for unveiling cloaked profiles. Journalists often employ certain practical steps and useful digital tools to verify the content they use as sources. The techniques described in the abovementioned handbook are associated with “open-source intelligence” (OSINT) and seek to verify online content, which are overt and publicly available (Glassman and Kang 2012). Because companies such as Meta limit researchers access to their own data, these techniques are useful as they allow researchers to unveil cloaked profiles through publicly available data.

The techniques are presented in five steps. The first entail critical reflection and a thorough scrutinization of the observed entities’ representation. Some studies may find this step sufficient. The second step involves mapping the digital traces of an online profile through multiple sites/different platforms. In my fieldwork where there was a relatively small number of observed entities, it was possible to type in usernames found on Facebook to other platforms such as Instagram or Twitter. By doing this, we may be able to recognize whether the account displays a coherent identity across platforms and sites over time. The third step consists of using search engines and simply typing in the names and user handles of accounts or websites that one wants to unveil. Applying the moniker which the profile is represented by in a search engine such as Google can quickly provide fruitful information to indicate whether this is in fact a human being or not. The fourth step involves using the source’s published content, such as profile pictures, and conducting a reversed image search in search engines, such as Tineye, Bing, Yandex, or Google. This allows researchers to upload pictures and search for the websites where the images initially were posted, if any (Grut 2021). In my study, this step was more relevant to profiles that had profiles pictures that appeared to be natural and not to profiles, which used pictures of cartoons or other avatars. Finally, as official bodies have started registering cloaked profiles, it is possible to look at such databases and run them up against collected data to search for cloaked profiles. One example is Facebook’s and Twitter’s provision of a list of clocked profiles to the United States government, which they made public after the investigation of potential Russian interference in the country’s political and electoral processes (U.S. House of Representatives 2021).

These steps for unveiling representation are not exhaustive and do not offer safe passage to unveiling all relevant information in the collected data. They merely illustrate practical tools available for ethnographers. Moreover, due to the unstable infrastructure of online tools and techniques, which constantly change and develop, it is impossible to provide a final and complete set of methodological practices that ethnographers can apply. Nevertheless, practicing how to use digital techniques, such as the ones outlined above, will enable researchers to conduct online credibility assessments, a critical area for enhancing one’s digital literacy. Such digital literacy refers to a person’s ability to identify, evaluate, create, and use information or other types of content in digital spheres (Cordell 2013; Walton 2016). Enhancing one’s digital literacy are specifically important for researchers concerned with digital observations because research has shown that those who are the least able to recognize cloaked sources or misinformation are also not aware of their own limitations. This entails that people with low digital literacy are more prone to believe and spread online misinformation further (Lyons et al. 2021). Additionally, applying the above-outlined techniques in a digital observation will reveal subsequent layers of representation. Such contextual data may assist the researcher in determining what type of data the observation is capturing, and it adds additional knowledge on how to interpret the meaning-making expressed by the observed entities.

Challenges of Gathering Contextual Data

Learning techniques for how to unveil cloaked profiles may be difficult, due to the fluid and rapid changes in web infrastructures, the lack of access to data from social networking sites, and ethical concerns related to privacy and consent. As social networking sites have restricted access to their platforms (Perriam et al. 2020; Shapiro et al. 2021), acquiring information about their users is shaped and conditioned by the site’s technological structure. For instance, Farkas et al. (2018) illustrated how they were unable to unveil the originators behind Facebook pages, due to Facebook’s technical settings that prohibit users from seeing the identities of the page owners. As contextual data are pertinent for unveiling cloaked profiles, not being able to gather this data may hinder the unveiling process. Scholars have stressed the importance for researchers to access data from commercial social networking sites and that the commercial actors themselves ought to facilitate researchers access to their data to a larger extent (Shapiro et al. 2021).

A related issue is how the lack of access to data from commercial social networking sites has created a need for researchers to establish cloaked profiles themselves to conduct hidden observations. This has instigated debates on whether some research activities violate the terms and conditions of online platforms (AoIR 2019). In a lawsuit in the United States, Sandvig versus Barr, the researchers were charged, because they had established several cloaked profiles to analyze discriminatory behavior online. Notably, the court found the research to be legal and dismissed the case (United States District Court for the District of Columbia 2020).

Markham (2005) has noted that conducting unobtrusive online observations can involve severe ethical pitfalls, particularly concerning consent, confidentiality, transferring data, and data storage. Such considerations were particularly relevant when observing migrants who experience highly precarious and vulnerable life situations and observations of potentially criminal acts committed by European citizens who help these migrants. To avoid issues of data storage of sensitive information only field notes were taken. As for informed consent, this is a much-debated and highly challenging task for researchers utilizing online observations. In my study I considered reaching out and connect virtually to obtain consent. That said, consent is an ongoing process and not always possible or ethically advisable (EASA 2018).

When collecting online data there are also concerns of privacy regulations. Privacy regulations, such as the General Data Protection Regulation (GDPR) set out by the EU in 2018, reflect an ambitious and broad approach for collecting digital data. The GDPR is particularly relevant because social networking sites have adopted these regulations as their global baseline for privacy (Shapiro et al. 2021). Within the GDPR, processing personal data is allowed if it is in the public interest and academic expression, as well as ethnographic research have been recognized as such (EASA 2018). However, one main challenge of the GDPR is that much of the regulation can be open to interpretation, making it rather ambiguous and difficult to navigate. Those concerned with these issues can refer to accessible data in government frameworks for ethnographic practices and the related useful guidelines (Dilger, Pels, and Sleeboom-Faulkner 2019; Jiménez 2018).

Additionally, when researchers venture into the quest of unveiling profiles, it changes the role of the researcher which call for ethical considerations. Ethical considerations are the base of methodological designs and when we use techniques to capture contextual data, ethics ought to be considered. This was found particularly challenging in the fieldwork, due to the precarious situation of the migrants and the fact that facilitating unauthorized migration can be prosecuted in the EU. Exposing how the migrants and the citizen humanitarians used cloaking techniques to connect and act digitally might have unintended negative consequences. Hence, ethical considerations were made of how the study could potentially expose the digital techniques used by the observed migrants and the citizen humanitarians. This implied that in my final research I would not describe uncovered cloaking techniques.

The above outlined techniques for unveiling profiles are inspired by “Digital source criticism” (Grut 2021)—initially aimed at assisting journalists. Including such techniques to a digital observational study ought to consider how the ethical considerations that journalists need to make are different from research ethics. Journalists do not need to obtain ethical approval prior to conducting investigations—which researchers in most cases are required to. By obtaining ethical approval, a researcher has demonstrated that they have adhered to the ethical standards within that research community. A journalist on the other hand may set out to collect data straight away. Gans (2018) argue there are similarities between the ethical practices of journalists and qualitative researchers such as sociologists. Both strive to be accurate, avoid plagiarism and violations of agreed-upon research procedures (Gans 2018, 47). According to Gans journalists have increasingly become concerned with verification of online content—compared to sociologists. Moreover, he suggests that the two disciplines could learn from each other. Following these arguments, it is possible for researchers to take inspiration from journalistic techniques—and simultaneously uphold ethical standards and methodological principles in their respective research discipline.

That said, journalists deal with the concept of what is the “truth” and what is “fake news” differently than researchers. The journalistic quest to deal with “fake news” has been seen as a response to restore societal trust in journalism and that misinformation is a threat to the journalistic practice. Researchers, on the other hand, are encouraged to problematize what they mean when referring to “fake news” and discussing “true” information (Farkas and Schou 2020). Unlike journalists, who can seek to unveil the “real” compared to “fake” news, a researcher concerned with ethnographic practices make considerations of what is “real” and how to make claims of evidence. In qualitative inquiry, such as digital observations, relativist ontologies tell us that there are multiple truths (Denzin and Lincoln, 2018).

Conclusion

This article discussed the ontological and epistemological dilemmas that cloaked profiles pose for digital ethnographic methods by drawing upon examples from my own fieldwork. First, I discussed the ontological dilemma of knowing what and whom we observe in digital observations and whether these challenges could be considered normative issues found in all social inquiries. Arguably, the digital field sites and the presence of cloaked profiles pose new methodological problems. The discussion showed that digital observational studies that do not address the challenge of cloaked profiles fail to recognize how digital profiles use technological tools to represent themselves in different ways. Such considerations should make researchers include critical reflections of who and what they observe in digital spheres. However, critical reflections can be insufficient for digital observations where there is a clear presence of cloaked profiles and where the presence of cloaked profiles will be of relevance to making claims of who and what we observe. Hence, the article discussed the epistemological challenge of practically conducting digital observations when acknowledging cloaked profiles. Depending on the type of method and analysis in each specific research project, researchers will, in some cases, benefit from practically engaging with how to unveil cloaked profiles. Thus, I presented digital techniques for how to unveil cloaked profiles and argued that researchers will enhance their digital literacy by practicing such techniques. Such techniques entail following digital traces of observed profiles, examining associated metadata, and cross-checking profile names with databases of revealed cloaked profiles. Finally, the article highlighted the hindrances to applying such tools, as ethically ambiguous and, at times, not even necessary.

Unveiling cloaked profiles is a complicated task that requires digital literacy and critical thinking skills, as well as the ability to situate the observed profile within a social and political context. Hence, what this article suggests is that by developing skills and actively engaging with how to unveil profiles, this will in turn enhance researchers’ ability to recognize cloaked profiles. Introducing critical reflection on the presence of cloaked profiles is a good start; however, it will not enhance researchers’ digital competence in unveiling profiles. This can only be done by actively familiarizing oneself with relevant sources, where to look, and what to look for. Practicing these skills will sharpen the researchers’ eyes and add to the methodological toolbox for how conduct digital observations.

Despite the strengths of this article, the discussion is not void of limitations. The ontological and epistemological discussions here merely reflect parts of what is embedded within the ontology and epistemology of cloaked profiles. As such, the article points to, but does not engage in, discussions of how cloaked profiles change our ontological and epistemological understanding of the ethnographic field. Arguably, what has been discussed here evokes considerations for what we understand as digital ethnographic methods and what we may claim to know. Hopefully, this article has triggered interest and may encourage future debates on these issues within the ethnographic field.

Moreover, this article has argued that ethnographers should observe and seek to unveil layers of representation that construct single observed entities. Critics may argue that this indicates a sort of methodological individualism, in which the researcher’s goal is to look at single actors’ intentions and motivations, reducing the observations to individual entities and findings. However, this is not the discussion’s intention. Rather, a critical reflection on the presence of cloaked profiles and learning digital techniques will arguably provide researchers with additional context and engagement with the field. In turn, this will strengthen the hermeneutic process of data interpretation, for which the researcher aims to establish an understanding of the “whole” by looking at the parts (Elster, Føllesdal, and Walløe 1983). In this discussion, the “whole” is the entire observational study, and the parts represent the contextual information on how profiles are represented, which we can gather using digital techniques. Gathering contextual data that unveil layers of representation will provide the researcher with thicker descriptions of how observed profiles use digital techniques to present themselves. It will also provide the researcher with supplementary interpretations of the field site, which will enable thicker descriptions of the observed social phenomena. In the fieldwork with citizen humanitarians and migrants, understanding how, and to what extent, observed profiles used digital techniques to cloak identities allowed for thicker descriptions of the networks. Furthermore, it allowed the analysis to include contextual information on how each observed entity constructed their expressed meaning-making. Hence, using digital techniques to unveil representation will allow the researcher to understand the different parts of the observed phenomena and adjust the overall interpretation accordingly.

Discussing methodological dilemmas that cloaked profiles pose to digital observations will add nuance to the picture of what type of data we can capture and the ontological assumptions of whom we are observing and add value to digital ethnographic data interpretations. As there is a scarcity of scholarly literature that addresses these matters, this article has sought to create a debate and propose ways for researchers to potentially address these complex and multifaceted dilemmas. Accordingly, the discussion only tackled a few central dilemmas, and the complex nature of these issues implies that further scholarly discussions are the most fruitful way to address them. Therefore, more research that discusses epistemological challenges instigated by the presence of cloaked profiles is highly encouraged.