Science, Technology and Medicine in Society (STeMiS), Uncategorized

Platforming environmental imagery: the increasing role of algorithms in ordering knowledge.

Written by Warren Pearce

Top Google Images search results for [climate change] in six different countries. Source.

Environmental imagery online

How does one represent something that is both as ubiquitous and abstract as ‘the environment’? This question is becoming ever-more important and urgent, as human-caused environmental damage becomes increasingly serious and debates about political choices increasingly fraught. The ways in which these debates play out depend, at least in part, on the ways in which we think about the environment, which is itself a product of the information publics access on a daily basis. Recently, the importance of environmental imagery has been highlighted through the inappropriately jolly pictures used by the media in stories about extreme heatwaves. Such images play a crucial but underappreciated role in the representation of the environment, conflicting with or detracting from the story’s textual content. Yet there is more to environmental imagery than its use in the mainstream media. Despite the increasing importance of images in digital media platforms, most scholarship has remained focused on textual analysis, with the ways in which digital images appear and circulate remaining unclear. This digitisation of ecology and society means that social media platforms and search engines have become crucial sites for researching environmental communication.

Platformed digital imagery

In July, I led two research projects investigating the content and characteristics of online environmental images at the University of Amsterdam’s Digital Methods Initiative Summer School. These research projects moved beyond the mainstream media’s use of images to study how the different affordances and cultures of use found across digital platforms bring particular ways of seeing and showing the environment through our computers and mobile devices. The projects brought together international, interdisciplinary teams of researchers in week-long ‘data sprints’, recently defined by Janna Joceli Omena and colleagues as time-limited projects focused on social, cultural and media issues using data from the web and its technological environment. In this short post, I focus on just one of the projects: “According to Google Images”.

Homogenising the environment

In this project, the team focused on the Google Images search engine, where likely over 2 billion searches are conducted every day, making it arguably the world’s most important gatekeeper of everyday visual culture. We aimed to discover the difference that both environmental issue and location made to search results, comparing search results for climate change and biodiversity loss in six different countries. We found support for a previous research finding published by me and Carlo de Gaetano: Google Images is having a homogenising effect on the visual representation of both climate change and biodiversity loss. For climate change, generic representations of climate change dominate, with a small number of repeated stock images highly ranked across the different countries. Many of these images have a half-and-half “left/right” layout suggesting the current, often dystopian, situation on the left, with a more hopeful utopian future on the right (this implied temporality is sometimes reversed, with the more negative situation lying in the future). Examples of these images include earth in hand”, “landscape” and “tree”.

For biodiversity loss, there is also homogeneity, although with slightly greater diversity between countries. Stock imagery is less dominant, but half-and-half images remain important. Although biodiversity loss covers many different aspects of environmental degradation, the imagery focuses almost exclusively on deforestation. For both search terms, humans are notable by their absence in the images presented.

The algorithmic ordering of environmental knowledge

Overall, the search rankings seem to be more dependent on whether images closely ‘accord’ to Google’s vision of what climate change and biodiversity look like, rather than on the authority of the website hosting the image. This marks a significant, but rarely discussed change in the way that Google ranks search engine results, with ‘artificial intelligence’ computer vision algorithms playing an increasing role in filtering and ordering public access to knowledge on the web. This helps account for the homogenisation of imagery but does not account for how Google Images has ‘decided’ what climate change and biodiversity loss look like in the first place. While this is a subject in need of urgent further research, the images show the influence of historic scientific and political approaches that prioritise the global over the local, and the separation of nature from culture. We write more on this in our project report. You can also peruse the posters which you can explore the image rankings in greater depth, just like the attendees of the Summer School. More to come later in the year, as the team reconvene to write up the findings for a journal article.

Attendees of the DMI Summer School discuss the According to Google Images project posters (Photo: Warren Pearce)

Huge thanks to the data sprint team members: Maud Borie, Laura Bruschi, Ariel Chen, Daniele Dell’Orto, Matthew Hanchard, Elena Pilipets, Alessandro Quets, Zijing Xu

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s