Working with Totalising Algorithms

Excited by the possible storytelling functions and forms that digital technology enables, I set out to foster meaningful encounters between author and audience in a digital narrative project titled We See Each Other. I had never considered the possibility of an invisible third party shaping these encounters, but they were there, ever-present and impossible to escape.

Pictured: We See Each Other home page

For many storytellers working online, large companies providing web services, like hosting or searching, play a constant mediating role which can shape stories in subtle but significant ways. This ‘mediation’ byweb services often requires users to participate in the reduction and over-simplification of people and concepts in order to easily present or index content. I refer to this as ‘totalising’, which can be challenging for practitioners concerned with fostering more democratic narrative experiences or challenging narrow and stereotypical representations of people, places or issues .

Many scholars (Morozov 2011; Jenkins, Ford and Green 2013; Noble 2018) highlight how structural inequalities and totalisation manifests on the web. For instance, Jenkins Ford and Green (2013) point out that meaningful participation online is ‘linked to educational and economic opportunities’ and Noble (2018) outlines the insidious ways in which search algorithms promote racist and misogynistic representations of people . But I want to provide a creative practitioner’s insight into how these power relations play out in digital narrative practice and examine some of the ways we can negotiate these issues.

The first encounter many creative practitioners will have with the underlying political structures of the web will happen early in the lifespan of a storytelling project. To create artwork online, creators must engage with some sort of intermediary service. In my case, during the construction of We See Each Other I set up the project domain and hosting through a leading web host provider. It seemed fairly innocuous at the time, but powerful intermediaries such as search engines and hosting providers shape how the Internet is used and who uses it. Morozov (2011, 209), for example, explains how people from countries such as ‘Belarus, Cuba, North Korea, Syria, Zimbabwe and certain areas of Sudan’ can face various unfair sanctions on the Internet simply because a large proportion of intermediary companies reside in the USA. For instance, the US government has a targeted policy to sanction particular former government officials and organisations in Zimbabwe (Morozov 2011, 209). This means American Internet companies should vet all their Zimbabwean customers, but because this ‘is so expensive and time consuming’ many companies end up banning all ‘Zimbabwean nationals’ (Morozov 2011, 209).

One company caught unfairly excluding a Zimbabwean organisation is BlueHost (Morozov 2011, 210): the very same hosting company I used to set up my digital narrative project. Because of the privilege I have of living and working in Australia I was able to use this competitively priced hosting service without interruption, but this may not be the case for all digital narrative practitioners.

The next encounter many creative practitioners may have with the underlying political structures of the web is when attempting to distribute their work. Making a work accessible to audiences means engaging with search engines, and in particular engaging with the most ubiquitous search engine, Google. Navigating this process can be difficult to negotiate for creative practitioners. Noble (2018, 100) explains, ‘what shows up on the first page of search is typically highly optimised advertising-related content, because Google is an advertising company and its clients are paying Google for placement on the first page either through direct engagement with Google’s AdWords program or through a grey market of search engine optimisation products that help sites secure a place on the first page of results’. This search engine optimisation (SEO) process can not only be expensive, but can also shape the work itself, as I discovered whilst constructing We See Each Other.

Currently, Google (2018) describes their algorithm as analysing ‘hundreds of different factors to try to surface the best information the web can offer, from the freshness of the content, to the number of times your search terms appear and whether the page has a good user experience.’ While Google’s algorithm uses hundreds of factors to determine page ranking, and it is not clear what the weight of each factor is, Google does highlight a few key factors; website speed, backlinks, keyword relevancy, submitting an xml sitemap, and editing page metadata. Backlinks and keyword relevancy are two SEO tasks I will explore in more depth because they are particularly hard to negotiate as a creative practitioner.

Google (2018) explains backlinks by stating that ‘if other prominent websites on the subject link to the page, that’s a good sign the information is high quality.’ This part of Google’s algorithm means the more that high ranking popular websites share links to creative practitioners’ work, the better the work will rank in Google, and the more traffic it will have. This can be problematic because it relies upon creative practitioners having connections to other influential and experienced  webmasters with high ranking websites, thus entrenching the same power structures present in the analogue world. Apart from the exciting storytelling possibilities, part of the reason I had (naively) turned to online platforms as a means for the creation and distribution of my work was because creators appeared to be able to make works that were directly accessible to huge audiences. But, the notion that online distribution is a democratic utopia obscures the wealthy companies shaping the Internet. In my case, because of the lack of connection myself and my co-creative team had to influential and wealthy domains, We See Each Other remains a low-ranking website resulting in a digital narrative that is less accessible to audiences.

Pictured: ‘We See Each Other’ search dominated by viral video from Real Housewives series

The second potentially challenging SEO task for creative practitioners worth discussing is identifying and using keywords. Identifying keywords that are most relevant to the content of the project is important to ensure that the website ranks highly when people search these terms. However, using these keywords is a highly rigid task. It involves using the keyword phrase in the title of the work. The title of the work must employ a header html font tag to be recognised by the algorithm, and the keyword phrase must appear frequently enough in the body text of the page or post (but not too many times as Google may penalise the site’s ranking for keyword spam) (Patel 2018). For artists trying to dismantle dominantrepresentations which define people, places and issues in totalising and stereotypical ways, moving away from definitive phrases is key. Therefore, researching popular keywords and optimising the content to match these phrases can undo some of the work towards less stereotypical and narrow representations. I have faced this struggle with keywords in my own work. The overall aim of We See Each Other was to move away from limiting and totalising representations of people from refugee backgrounds and towards more ethical encounters between people from refugee backgrounds and audience members. I asked myself what keywords might apply to the stories on the website. ‘Australian stories’, ‘refugee stories’, ‘stories about family’, ‘stories about war’ and ‘stories about choices’ all seemed somewhat relevant. But no matter how many phrases I came up with, none of them truly seemed to capture the project or the stories. Privileging a few words seemed absurd given the diversity of the stories and the storytellers, and using popular keywords  would have forced me to resort to privileging words and labels I was actively attempting to resist.

Even for creative practitioners whose work was not created with the specific goal of dismantling dominant or stereotypical representations, their engagement with SEO contributes to the way stories and people are found and framed. Noble (2018, 13) points out that:

the near-ubiquitous use of search engines in the U.S. and perhaps worldwide, demands a closer inspection of what values are assigned to race and gender in classification and web indexing systems, and warrants exploration into the source of these kinds of representations and how they came to be so fundamental to the classification of human beings.

In this statement, Noble highlights that classifying and labelling human beings and creative works, is a fundamental part of the way the web operates presently. Noble (2018, 1) contends that this is harmful because it can ‘reinforce oppressive social relationships’. Noble (2018, 14) goes on to describe her experience of searching the key terms ‘black girls’ on Google and discovering that ‘hotblackpussy.com’ was the first hit. Noble is clear that Google has a role and responsibility in this algorithmic oppression, but content creators like me looking to distribute through Google also have a role to play in constructing the way humans are classified and the ways content is framed. As creative practitioners we must examine the ways in which we choose to classify our authors, characters and subjects in order to rank highly in Google. It not only frames the way audiences interpret our creative work, it also has an effect on the way certain groups of people are represented and found in search engines.

Not all SEO tasks are so restrictive though. For example, editing the metadata or descriptions displayed on Google, Facebook and Twitter can provide opportunities to extend the way audience members might experience their encounters with the authors or narrative and make the experience more meaningful. Without editing the metadata, words from the titles and body text of a URL are selected by a bot to create a short description of the page. These descriptions appear on Google search lists and underneath Facebook or Twitter hyperlinks. Taking control over this element of SEO allows creative practitioners to rewrite and reshape how their work is framed. In my case, writing a few sentences of description allowed the authors and I to frame the work in a way that challenged narrow representations of people from refugee backgrounds.

Pictured: The metadata for We See Each Other as it appears on Google

Editing metadata is also an opportunity for creative practitioners to extend the interactive functions of their work beyond the domain of the site. During the construction of We See Each Other, manipulating the metadata allowed me to frame the encounter between author and audience from the moment someone found the site via social media or search engine. It also allowed me to construct further meaning  and metaphor for the audience to interpret when they shared the work over social media. So, when sharing a particular link to We See Each Other via Twitter or Facebook, the user’s friends or followers will see a link which reads, ‘I chose to see the authors of We See Each Other.’ The description underneath this link is then targeted toward their friends or followers, asking them directly if they will choose to see the authors too.

Pictured: ‘Screenshot of backend of seeeachother.com whilst editing metadata’

Editing metadata is not only an opportunity to continue to shape the creative work, it is an opportunity to shape how people (authors and characters) are represented and classified, rather than ceding agency to bots created by a powerful company, like Google.

Through my own experience, I have observed some of the challenges creative practitioners face when working in digital spaces to foster more democratic narrative experiences or challenge dominant stereotypical representations in their work.This is by no means a comprehensive list of these challenges, but it is my hope that by outlining these challenges, this part of the digital narrative construction process will become more visible. Other creative practitioners will have much to contribute to this conversation, particularly as the challenges we face will grow as algorithms and digital tools continue to evolve at a rapid pace. By observing and critiquing the political context of the digital tools and services we use, together we can make more informed choices about how we engage with them and gain more control over the effect they have on both our work and society more broadly. In particular, I believe that negotiating our engagement with these tools and services more critically can assist us to foster more democratic narrative experiences and challenge dominant stereotypical representations more comprehensively.

References:

Google. 2018. ‘How search algorithms work.’ Accessed February 1, 2018. https://www.google.com/search/howsearchworks/algorithms/.

Jenkins, Henry, Sam Ford and Joshua Green. 2013. Spreadable media: Creating value and meaning in a networked culture. New York: New York University Press.

Morozov, Evgeny. 2011. The net delusion: The dark side of Internet freedom. 1st ed. New York: PublicAffairs.

Noble, Safiya Umoja. 2018. Algorithms of oppression: How search engines reinforce racism. New York: New York University Press.

Patel, Neil. 2018. ‘SEO copywriting: How to write content for people and optimize for google.’ Accessed February 1, 2018. https://neilpatel.com/blog/seo-copywriting-how-to-write-content-for-people-and-optimize-for-google-2/.