I had the pleasure of organizing the Ethics of Using Digital Media in Arts and Humanities Research Conference and to welcome delegates from various institutions at the University of Manchester on 22 February 2019. Below are the slides from my presentation, with captions. For more information about the conference, please visit https://digitalheritageresearch.wordpress.com/conference-2019/

This presentation stems from my doctoral research, in my project I explore the potential influence of social media audiences to museums’ brands. My case study is the Manchester Art Gallery and their takeover event that took place on January 2018.

For those of you who may not know what happened, let me give you a brief recap. The Gallery invited contemporary artist Sonia Boyce to create a new artwork and during the creative process, the artist and the Gallery staged a takeover event in collaboration with performative artist Lasana Shabazz and drag artists from Family Gorgeous.

The night was video recorded and the resulting artwork is called “Six Acts” – following the six performative acts of the night. One of these acts was the removal of Pre-Raphaelite painting Hylas and the Nymphs by J. W. Waterhouse.

The painting was replaced by a poster asking visitors to reconsider how the Gallery’s collection can be reinterpreted in our contemporary environment, particularly those works that represent women as a passive figure. It prompted visitors to leave their thoughts and the Gallery kindly provided post-its and the event’s hashtag for people to “get involved”

Online the Gallery posted about the event a handful of times, including this tweet that shows a video of the painting as it was taken to the Gallery’s store.

And for almost five days, few people went to Twitter to “get involved in the conversation”. Until almost suddenly, the conversation blew up on January 31st and February 1st. What happened?

Two articles were published on The Guardian, one titled “Gallery removes naked nymphs painting to ‘prompt conversation’” by Mark Brown and another titled “Why have mildly erotic nymphs been removed from a Manchester gallery? Is Picasso next?” by Jonathan Jones
The following days saw a torrent of comments, which were overwhelmingly negative and critical of the Gallery’s decision to remove Hylas and the Nymphs

My analysis included a thematic coding of the body of the tweets, which meant scouring through hundreds of posts. Unsurprisingly some of the responses were racist, homophobic, and very difficult to read (especially the ones that were made directly to individual staff members). So I made a decision to not illustrate this type of tweet in my thesis, even though for every other section about the themes of the conversation included an anonymised tweet as an example. And although I had some mixed reviews about this decision from my supervisory panel, I want to share with you why I did this.

My decision is mainly influenced by two ideologies: actor network theory (as described by sociologist Bruno Latour) and the ambivalent internet (as described by communications professors Whitney Phillips and Ryan Milner). These ideologies have influenced my approach to collecting and analysing social media data for my thesis, as well as presenting my results in and beyond the thesis.

I am adopting Actor Network Theory (ANT) in my project, as a theoretical and methodological framework. This theory (or ontology as its authors refer to it) recognises that everything in the social and natural worlds exists in constantly shifting networks of relationships. Some of the main tenets are those of generalised symmetry and of material-semiotic relations. This means that networks can be composed of human and non-human actors who have equal potential of agency within the network. From a methodological perspective, ANT asks the researcher to approach the network without prejudice and to let the actors define themselves and their relationships with one another.
In an actor network value or information (that is the basis of the relationships between actors) can be transmitted by intermediaries, or altered by mediators.

ANT affected the way I collected and analysed the tweets I collected for my research in two main ways. First by collecting tweets based on keywords that referenced the case study as an institution, thereby giving me the opportunity to study the event as it unfolded within the existing context of the Gallery’s digital space. And second, by analysing the data using metrics and themes provided by the technology and the users. By taking these considerations I followed a quantitative analysis based on the available Twitter metrics, and a qualitative analysis based on thematic coding using the users’ language as themes.

This framework means that in my writing process I should present the data as a reflection of the actors and their relationships with one another, which in my case, included a small portion of hate and bigotry.

I questioned whether reproducing the data in my thesis would have the potential to inflict harm (for example by using anonymised tweets to illustrate these instances of bigotry and hate)

This is where the Ambivalent Internet influenced my decision. In this text, Phillips and Milner describe our behaviours online as a form of folkloric expression – meaning, a form of expression that is continuously performed, contextually understood, and fundamentally hybrid. In this sense, folkloric expression can perpetuate bigotry and intolerance as seemingly innocuous everyday behaviours.

Furthermore, Phillips and Milner argue that such vernacular expressions (and their associated media like tweets or memes) are amplified thanks to the affordances of digital technologies. These affordances, as described by various new-media scholars, allow online participants to create, manipulate, re-appropriate, store and categorise mediated vernacular expressions. Moreover, these affordances allow participants to share their versions of vernacular expression freely, easily, and immediately.

In many cases, the ethical stakes are difficult to assess when digital media are involved. For example, stories may become convoluted when participants’ demographics are muddled and their motivations are not clear, as is the case with Twitter threads that lack the context to determine who is involved, who is being serious, and who is performing to draw out a reaction. From this perspective, it is difficult to discern the meaning of such observed content regardless of whether it was collected from its original creator or from a participant who repurposed it.

The technological affordances described earlier amplify content at an unprecedented rate and often to unsuspecting audiences. For example, liking, commenting, or retweeting contentious matter regardless of intent (such as to cause offense or as a joke) amplifies this content to the audiences of those who are enacting these behaviours.

Phillips and Milner argue that digital mediation further complicates amplification because it asks us to walk a fine line in deciding what to amplify and what to ignore in the service of our own critical analysis. So on one hand we may choose to engage with certain content to highlight cultural or social problems as a way to enact awareness. On the other hand, remaining silent may signal complicity with those same cultural or social problems.

To further illustrate this point, Phillips and Milner have recently published an article (2018) where they use an analogy of biomass pyramids to highlight the fact that as a society we tend to condemn ‘apex predators’ (tertiary consumers) when it comes to information disorder (meaning misinformation, disinformation, and malinformation), when in fact we should also be condemning and taking a stand to, or at the very least being aware of, those who are not at the top of the pyramid.
In other words, Philips and Milner (2018) ask that we need ethical reflection starting at the ‘primary producers’ level – in doing so we can stymie cumulative harm.

For example, they argue that this low-level or small harm can include posting snarky jokes about a developing story, retweeting misleading narratives ironically, making ambivalent inside jokes, @mentioning the butts of jokes, and jumping into conversations without knowing what the issues are. Individually these behaviours may not be worthy of consideration, especially when you have high level participants who are spreading information disorder (take for example some current politicians). Cumulatively, however, these harms provide the platform for these ‘apex predator’ to inflict harm at a higher level.

So I come back to this question – would reproducing instances of bigotry and hate in my thesis cause harm? The ethics of amplification help me navigate what I am seeing in my data and understand the specific social, cultural, and in some cases the political contexts of the users who are amplifying the narrative that followed the removal of Hylas and the Nymphs by the Manchester Art Gallery. I believe that the ethics of amplification can also help me navigate how I reproduce and report my findings, in particular those instances of abuse.

My concern in illustrating the abusive behaviour I observed in my data stems from my choice of using actor-network theory as my theoretical and methodological framework. That is, of using the data as it was collected, without modifications. I considered that such examples could cause potential harm to whoever read my thesis (who as observers would not have a choice of not reading the examples) and to the Gallery’s staff members (who were the victims of this abuse and could potentially relive this abuse upon reading the examples).
On a ‘meta’ level, I wanted to protect my thesis from being another mediator in a network that could perpetuate harmful behaviour. Instead I chose to be an intermediary.

In my decision to not illustrate harmful content I did two things: first I wrote in my thesis about my ethical stance and how it simultaneously shaped and is shaped by my research, and second, I reported abusive behaviours with Twitter ensuring that the user accounts who authored this content would be muted. In doing so, I protected myself from potential future harm from these accounts, without blocking them entirely so I don’t limit any potential future research.

My decision and my written acknowledgement on my thesis received mixed reviews from my supervisory panel. Although my main supervisor is very supportive of my decision, the rest of the panel had some practical concerns, such as questioning how much harmful content was collected (note it was less than 5% of the total dataset) and about making a political statement in my thesis (questioning whether this was appropriate given my theoretical framework does not involve ideologies of power).
I stand by my decision. I believe that if we agree that museums are not neutral, neither is – nor should be – our research.

To close the presentation I would like to recap what I learned from this exercise in ethics, as influenced by Actor-Network theory and by the Ambivalent Internet. Firstly, these decisions are highly contextual, starting with a research question that did not aim to study harmful online behaviour and followed by the consequences of a particular theoretical/ methodological framework, as well as the type of data chosen. Secondly, these decisions are part of an ethical reflection of how to present the results of this research and considering the potential impact of my actions. Finally, these decisions are highly reflective of my personal stance as a researcher and as a user, adding to the bias I conduct my research with and through which I employ digital technologies.

I leave you with this question one more time. If you use digital media in your research, what is the potential that it may cause harm through your research?

Thank you for your time! I hope this presentation was useful for you. Please reach out if you want to discuss further or if you have any questions.

This presentation was given at the Ethics of Using Digital Media in Arts and Humanities Research Conference on 22 February 2019, at the University of Manchester. For more information about the conference, please visit https://digitalheritageresearch.wordpress.com/conference-2019/
