Digital objects and ecological politics

How are digital objects – such as hashtags and likes – involved in ecological politics? In the wake of the 2019 Amazon rainforest fires, journalists, media outlets, NGOs, and others commented on the role of hashtags, images, links, and other kinds of online devices in mobilising and shaping public concern. For example, Asad Rehman, director of anti-poverty charity War on Want, wrote in The Independent:

Such claims of virality were accorded prominence alongside announcements of the number of fires reported and the number of hectares burnt. For instance, the non-profit news source Common Dreams reported: ‘#PrayForAmazonia goes viral as Twitter users call attention to the “international emergency” of fires devastating Brazil’s rainforest’. They presented the hashtag as a device deployed by ‘social media users [attempting] to draw the world’s attention to the Amazon rainforest’ and to ‘[slam] the media for paying too little attention’.²

Others found digital objects were playing more troubling and unruly roles. Zoë Schlanger, environment reporter for Quartz, shared screenshots of search engine results for queries related to the fires, writing:

The media outlet Mother Jones published a piece titled ‘Stop sharing those viral photos of the Amazon burning’ with the byline ‘the Amazon is on fire, but the photos you’re seeing on social media don’t show it’ (see Figure 6.1).⁴ Other outlets explored the circulation of images which were considered misleading and provided tips on ‘how to spot inaccurate photos on social media’.⁵ Some questioned the effectiveness of celebrity-driven online crowdfunding initiatives, as a ‘noncommittal solution Americans can partake in from thousands of miles away’.⁶

Figure 6.1Tweet from Mother Jones on ‘fake’ viral photos. (Source: https://twitter.com/MotherJones/status/1166085507861876736. All rights reserved and permission to use the figure must be obtained from the copyright holder.)

Prompted by questions and concerns around the digital mediation and mediatisation of public engagement with the 2019 Amazon rainforest fires, this chapter considers the role of digital objects, platforms, and ‘methods of the medium’ in ecological politics.⁷ We draw on a series of empirical vignettes from a series of collective inquiries⁸ undertaken with the European Forest Institute, the Public Data Lab, DensityDesign Lab, and graduate students at King’s College London.⁹ Following research on the social lives of methods, we consider digital devices as ‘patterned arrangements’ which imply, perform, and attempt to (re-)produce (not always successfully) particular versions of the social.¹⁰ Such devices may be traced, scraped, and ‘repurposed’ for social and cultural research.¹¹ For example, from the perspective of a web page author, hyperlinks may be used to reference, acknowledge, or point to other pages, and be understood in relation to a broader background of textual referencing practices.¹² Hyperlinks can also be repurposed to disclose the material organisation of ‘issue networks’: groups of ‘heterogeneous entities’ such as ‘actors, documents, slogans, imagery’ configured around a ‘common problematic’.¹³

Repurposing digital devices in this spirit may allow for the examination of how things are organised online, not only considering digital objects (such as hashtags, links, or likes) as a sample or proxy whose value is derived from ‘standing in’ for broader societal relations, but also considering the role that such objects have in shaping social issues themselves.¹⁴ One may, for example, study how digital devices are involved in ‘configured, professional and publicized political culture’ when organisations actively participate in public displays of connection by linking to each other online.¹⁵ Digital devices may also serve to study the ‘formatting of issues’,¹⁶ and the ‘politics of association’,¹⁷ attending to how actors online form alliances and collectively advance their programmes.

Online data, devices, and methods may be repurposed to study not only ‘how well one is doing online’ (as often done by marketing professionals) but also who is dominant and who is marginalised, their concerns, positioning, and alignment,¹⁸ or even to study the displacements of politics and politics of displacements, such as when actors or institutions reframe issues to downplay their significance or to emphasise alternative priorities.¹⁹ Digital methods may be approached not only to study dominant voices but also to attend to marginal and excluded positions, with attention to the relations between, as Susan Leigh Star puts it, ‘lived experience, technologies … and silences’²⁰ in order to care for neglected, undervalued, and marginalised experiences.²¹

How such digital methods are put to work in the context of research and beyond the academic realm should be approached with both caution and care. What does it mean that researchers, public institutions, campaign groups, journalists, and others are using the web, social media platforms, and online devices not only for communication and engagement, but also for understanding and reporting on issues they work on? From the point of view of social research, digital methods have been considered ‘not our own’, bringing ‘alien’ assumptions into research.²² How does one untangle what repurposed digital materials bring to the study of social life, particularly in the context of fields which encourage an empirical sensibility towards how societal categories, groupings, and entities are produced and stabilised?²³ This is not just a case of asking whether digital methods can be enlisted to do the work of other methods (e.g. can tweets replace surveys?) or whether they may eclipse other research approaches (e.g. are application programming interfaces (APIs) displacing ethnographies?). One may also ask how online devices are involved in rearticulating relations between social life and social analysis and to what ends,²⁴ as well as how they might play a role in more inventive forms of social inquiry.²⁵

In the case of the #AmazonFires project, digital objects served to explore different perspectives on forest–society relations, as well as the role of online platforms and devices in organising these relations. Through a series of workshops and activities with the European Forest Institute (an international organisation concerned with forest issues), online materials were used not only to produce research findings but also to support collective inquiry around the actors, issues, and dynamics of online engagement with the 2019 Amazon forest fires.²⁶ This was undertaken together with journalists, civil society groups, policymakers, and others. In the sections below we examine how digital objects were involved in formatting, performing, and disclosing different kinds of ecological politics, before concluding with a look at how this may contribute to research on environmental events and empirical, conceptual, and theoretical engagements with digital ecologies.

What is happening? The algorithmic mediation of environmental events

We began our inquiry in media res, in the midst of significant international media coverage of the Amazon forest fires in August 2019. As a starting point we took hashtags which had been mainly associated with the event in predominantly English-language media coverage and used ‘hashtag snowballing’ – querying and gathering tweets using this initial set of hashtags to discover other ones associated with the fires. We also asked the European Forest Institute and their network of ‘issue experts’ for further suggestions. This gave us the following set of hashtags: #ActForTheAmazon, #AmazonFires, #AmazonRainforest, #PrayforAmazonia, #SaveTheAmazon, and #SOSAmazonia. We could also have taken a broader set of keywords, but we were particularly interested in what we could learn about the role that these prominent and widely used hashtags played in broader societal engagement. Indeed, such hashtags are considered to play a significant role in assembling and connecting different online posts into trending events.

Robin Wagner-Pacifici writes about how ‘events take shape’ through ‘concrete material and formal hosts’, such as ‘executive orders, letters, trials, handshakes, newspaper articles, photographs, and paintings’.²⁷ With this spirit, one may take hashtags as a kind of digital object involved in the gathering, making, and shaping of environmental events. Forest fires were implicated with the emergence of hashtagging as a social media practice. While there are longer textual and media histories of hashtags as what one might consider connective keywords,²⁸ the first widely recognised use of the hashtag on Twitter was said to be #sandiegofire, in relation to the 2007 forest fires in San Diego. This led to comments from Twitter’s founder that the platform ‘does well’ at ‘natural disasters’ as ‘massively shared experiences’.²⁹

Using the official Twitter API, we began collecting tweets containing any of the hashtags listed above associated with the 2019 Amazon forest fires using the Digital Methods Initiative Twitter Capture and Analysis Toolset, an open source project which aims to support ‘methodological diversity and epistemological plurality’ in working with social media data.³⁰ The collection starts with an initial ‘spike’ of tweets followed by a rapid tailing off over the coming days (Figure 6.2).

Figure 6.2Total number of posts per day associated with #ActForTheAmazon, #amazonfires, #AmazonRainforest, #PrayforAmazonia, #SaveTheAmazon, or #SOSAmazonia according to data collected from Twitter API using the DMI-TCAT tool. (All rights reserved and permission to use the figure must be obtained from the copyright holder.)

This spike coincides with media and user reports that the Amazon fires were ‘trending’. Twitter’s Trending Topics algorithm is said to prioritise novelty in the form of spikes or surges over overall volume of interest.³¹ The algorithmic mediation of environmental events can be read against a background of longer histories of ‘orderly expectations’ punctuated by improbable catastrophes in cultural representations of environments.³² For environmental phenomena to emerge as trending events on Twitter they must be exceptional, prompting a response which is an order of magnitude apart from what is usual.

Within our Amazon fires collection one can see Twitter users engaging with the platform’s trending algorithm in various ways, including: (1) as a publicity tactic (e.g. ‘let’s get this trending’); (2) as event, hook, or social fact (‘this is trending’); and (3) as failure, distraction, or displacement (‘as the fires burn, look at what is trending’). In the case of Amazon fires, Twitter users respond to, imagine, lament, anticipate, critique, contest, and attempt to intervene with the platform’s trending algorithm to obtain public attention – to register the fires as an internationally significant event. As various posts asked: Why are Spiderman, Trump, Sharknado, Miley Cyrus, or Jamie Oliver trending rather than the destruction of the world’s largest remaining rainforest? What does this say about Twitter? What does this say about us? What can we do about it? Tweet collections may be considered as authorised records of lively, interactive algorithmic cultures arising from an interplay between platform features and user practices,³³ rather than, say, a collection of opinions as one might expect from a survey. Algorithms take part in the making of environmental events not only through computational ordering, but also their reactive effects among users alive to the politics and consequences of this ordering³⁴ – including what constitutes a significant public event. The hashtags may be seen as part of an attempt to get the fires trending, as well as inviting particular forms of action: to #ActForTheAmazon, to #PrayforAmazonia, to #SaveTheAmazon.

What is engaging? Querying objects of engagement in environmental events

How might we characterise the 2019 Amazon fires according to this spike in Twitter activity? What kind of event is it? How are Twitter users engaging with it? Digital media scholars have suggested Twitter may be taken as a ‘storytelling machine’ to facilitate ‘remote event analysis’, for example by examining the most retweeted tweets in any given day to ‘tell the story of an event as it unfolds’.³⁵ This can be construed as a way of taking metrics and device data into account when making sense of online material, as indicated in the phrase ‘quanti-qualitative’.³⁶ Focusing on a ten-day period from 24 August to 2 September 2019 gave us a collection of 311,483 tweets. We looked at a selection of the top ten most retweeted tweets per day with the European Forest Institute together with Brazilian journalists and scientists and observed a wide variety of narratives, frames, and concerns, including, among others, Amazon fires as: distressing and moving event to pray for (signalled by the use of specific emojis ‘’); as call to action; as displacement of Indigenous communities; as celebrity cause; as space of misinformation; as cattle ranching and meat issue; as deforestation and investment issue; as foreign aid issue; as an environmental event receiving more attention than other comparable events; as scientific issue; and as political issue mobilising for or against Bolsonaro.

In the top ten most retweeted tweets per day, the Amazon rainforests are construed, variously, as ‘homes’ to people, plants, and animals; the ‘lungs of the earth’; Indigenous lands; part of ‘our planet’; agricultural sites; and Brazilian territory. Whereas encouragements to pray are most prominent among the most retweeted tweets of the first few days, over the following days, other kinds of calls to action become more visible as users were invited to sign petitions, to share messages, to elevate and give voice to affected communities, to go vegan, to debunk misleading content, to donate funds, to use the Ecosia browser extension to plant trees in Brazil, and to boycott and defund companies who are implicated in deforestation.

There are also voices and perspectives that remain marginal. For example, a graduate student group querying the Twitter data set for the names of all Indigenous communities and territories listed by the Brazilian ministry found that only two – Pataxó and Xingu – appeared in more than a handful of tweets. Among the most retweeted tweets, these two groups featured through the communications activities of climate groups outside of Brazil such as the Sunrise Movement and Fridays For Future Europe, showing how marginalised and local communities obtain visibility only through the voice of European and US-based organisations.

In addition to exploring posts according to ‘built-in’ platform metrics such as likes and retweet counts, the sharing of links may be taken as another way to characterise online activity associated with the fires. By unshortening,³⁷ cleaning, aggregating, counting, and reading the contents of the most shared links per day, one can observe a similar plurality of concerns, framings, entities, and invitations to action (Figure 6.3), as different kinds of links indicate different approaches to engaging with the fires.

Figure 6.3Top 40 most shared links per day, 24 August to 2 September 2019. (All rights reserved and permission to use the figure must be obtained from the copyright holder.)

As well as two peaks of sharing news media links on 24 and 27 August, one could observe petitions among the top shared URLs and social media sites (Twitter, YouTube, and Instagram), as well as NGOs who were active around the fires. The pages associated with these links proposed different framings of the forests and the fires. For example, the most shared link on the first day was an article from Argentinian media outlet Sitio Andino framing the forest as habitat and fires as an environmental disaster. The second was a 2006 article from El Pais framing the forest as livelihood and fires as a socio-economic crisis. The third was from NASA and showed heat maps and satellite imagery, portraying forests as planetary zones and the fires as planetary events. A Greenpeace petition which remained among the top shared URLs across the ten-day period contrasted ‘greedy’ and ‘land-grabbing’ agribusiness interests backed by Bolsonaro’s government on the one hand with Indigenous communities, biodiversity, and climate change on the other hand. By days nine and ten there was a rise of promotional content for Etsy merchandise, including bot-like regularly recurring posts.

The most highly engaged-with posts and URLs suggest that this collection is dominated by responses and reactions to the fires, as well as responses to these responses (e.g. accusations that other posts are misleading, reporting on geopolitical exchanges which frame the fires differently). If examining the most engaged-with content can be construed as a form of remote event analysis,³⁸ then perhaps the event being followed is not only fires spreading in the forests but also different responses to the fires spreading around the world: an algorithmically mediated, reactive archive of the cultural production of an international environmental event.

What is hashtagging? Hashtags as indicators of environmental issue composition

One may take the web and online platforms as a ‘collision space’ for different kinds of concerns around the Amazon fires.³⁹ The Twitter collection of the fires may be considered to exhibit a degree of ‘liveliness’ to the extent that quite different framings and variation in terms can be detected.⁴⁰ How might we unfold the archive to untangle and situate these different kinds of concerns? How might we look beyond what is most engaged with overall in the collection?⁴¹ Researchers in the field of science and technology studies (STS) have employed ‘co-word’ analysis as a ‘relational indicator’ in scientific text collections to explore connections among academic publications.⁴² Here, ‘co-hashtag’ analysis has been used as a way into the dynamics and composition of issues.⁴³

What is a hashtag? Hashtags form part of media interfaces and platform logics enabling interconnections between posts, users, and other digital objects. They may be used to indicate topics and events, emotions and actions, subjects and subtexts, aspirations, and asides. They may take on ‘different emergent qualities’ depending on how they are used.⁴⁴ We used hashtags appearing at least ten times in the collection as the basis for making a co-hashtag network: that is, a network which shows how hashtags co-occur together in posts (Figure 6.4).⁴⁵

Figure 6.4Co-hashtag network showing hashtags associated with posts containing prominent Amazon fires related hashtags on Twitter. (All rights reserved and permission to use the figure must be obtained from the copyright holder.)

We co-organised a series of collective interpretation moments with the European Forest Institute, including with Brazilian journalists, researchers, and policy experts. These led to the identification of various clusters of hashtags, providing an indication of the composition of the issue on Twitter (Table 6.1). The combination of different issue hashtags can be an indication of ‘issue hybridisation’⁴⁶ – for example, by connecting trending Amazon fires hashtags to #vegan hashtags. This may be taken as a deliberate tactic (‘our issues are related’, ‘hey vegans, look at what’s happening in the Amazon’) or as a way to situate the fires within a tangle of concerns that exceeds conventional or professional issue articulation (‘this is more than an environmental problem’, ‘what you eat has consequences’). The variety of concerns in the co-hashtag network may also indicate how these trending hashtags serve as ‘cross-cutting networking mechanisms’, producing ‘unpremeditated combinations across a variety of feeds and networks’.⁴⁷ Interpreting and annotating co-hashtag networks with our collaborators served as a way to explore issue composition as well as to situate and explore relations between different concerns.

In examining this network, the prominent role of political personalities was observed, notably a series of heated exchanges between Bolsonaro and Macron and associated pro- and anti-hashtags (e.g. #vivabolsonaro, #MacronLiar). As well as ‘issue celebrities’⁴⁸ such as American actor #LeonardoDiCaprio announcing a donation of $5 million, one can also see more bottom-up fan culture interventions such as the ‘army’ of the Korean boy band BTS mobilising in support of Indigenous and conservation organisations (‘#ARMYHelpThePlanet’). Various hashtags indicate other ways in which the Amazon fires are connected to supply chains (#soybeans), consumption patterns (#govegan), trade agreements (#stopmercosur), and agricultural practices (#glyphosate). Hashtags may serve to invite, surface, and display relations between heterogeneous situations, issues, entities, and communities involved in the making of environmental events – including contestation around their meaning, significance, and stakes.

What is imaging? From visual misinformation to social lives of environmental images

If pedologists sampling soil in the Amazon forest support a ‘long chain of transformations’ enabling the circulation of references from field to report,⁴⁹ what forms of public knowledge and action are invited when online platforms, devices, and digital objects are involved in the making of environmental events? What role do images play?

Some of the most highly engaged with posts involve contested images and visual representations. For example, fact-checking account HoaxEye wrote ‘Not a single photo of #AmazonFires’, above images which are portrayed by another viral tweet as ‘sad images of Amazon fires’. The fact-checking account also provided source details for the inaccurate images (‘Spain, 2006’, ‘Argentina, 2018’, ‘Costa Rica, 2016’, ‘Mexico, 2019’), showing how images from other fires are misleadingly used to tweet about the Amazon fires. The same account highlighted Bolsonaro’s contention that Macron is sharing ‘fake photos’ (‘fotos falsas’) in order to ‘instumentalise an internal issue of Brazil’. In another widely shared post, US conservative commentator Dinesh D’Souza posted: ‘Many of the photos of #AmazonFires are bogus. They are taken from years ago – sometimes as far back as 2000. Another case of #FakeNews to advance the climate change hoax’.⁵⁰ US magazine Mother Jones posted: ‘The single most viral photo of the #AmazonFires is fake and you should really stop sharing it since it makes you look silly’, and the Times of India reported ‘These three most-viral images of #AmazonFires are “fake”’.

Our collaborators at the European Forest Institute were keen to know more about these claims about misleading images. When looking into the circulation of some of the most engaged with images in the Twitter collection (both within the collection and across the web and other platforms), we found that many of them were indeed from places and times other than the Amazon forests in 2019. One of the most engaged with posts contained a photo taken in 2016 showing a jaguar which had been rescued by the Brazilian army being held in the arms of a soldier wading through water. Many images of distressed, injured, and dead animals circulating in posts associated with the fires were from other events. Pictures of burning forests – including those shared by Macron, Leonardo DiCaprio, and the footballer Cristiano Ronaldo – were from other fires, as much as a decade and a half earlier. One of the most retweeted tweets contains an image of forest fires in Thailand. A highly shared video of an Indigenous Pataxó woman was from an arson attack outside the Amazon forest.

What is it that makes an image misleading? How might one distinguish between images being shared as direct representations of events and images which are doing other kinds of meaning work? When are images to be taken as evidence of an event, and when might they serve as ‘generic visuals’?⁵¹ Previous studies have explored the role of images in environmental and climate communications – from ‘issue animals’⁵² to ‘issue landscapes’.⁵³ Such images may serve to indicate types of animals, landscapes, and environmental events, for the purposes of issue work which often aims to enrol events as part of broader and ongoing issues (e.g. an oil spill as a form of pollution, as one of a series of spills as a result of corporate malfeasance). On the website of an NGO, an image of a forest fire may be to some degree substitutable. Similarly, images in news media coverage may often be selected from stock photography collections, with the image serving as content within a template or layout. Social media posts with images are said to have higher engagement rates than those without.

Looking beyond a focus on the representational (in-)fidelities or faithfulness of images, we traced their circulation and social lives within and beyond our Twitter collection in order to obtain a richer picture of ‘instances of image usage’⁵⁴ and ‘sites of audiencing’.⁵⁵ As Hito Steyerl comments, digital images are not just about ‘the real thing’ but also their ‘own real conditions of existence’ including their transformation, recontextualisation, appropriation, exploitation, and dispersion.⁵⁶ When accounting for the technicity of ‘networked images’,⁵⁷ one may also learn about the features and vernaculars of online spaces and platforms in which they circulate.⁵⁸ This shifted the focus of our collaborative interpretation activities with the European Forest Institute from ‘misleading images’ to ‘media recycling’ and tracing and telling stories about the social lives of images without necessarily assuming they are used in a way which is misleading.⁵⁹ We explored various approaches for analysing and redisplaying social media images in terms of their content, ordering, circulation, and audiencing.⁶⁰ For example, taking a forest fire image that the Agence France-Presse (AFP) characterised as misleading as a starting point, we looked at variations, visually similar images, and images which contained this image throughout the collection (Figure 6.5), as well as tracing the texts and contexts in which these images were shared.

Figure 6.5Tracing variations and different visual articulations of pre-2003 burning forest image shared by Macron and others in association with 2019 Amazon fires. (All rights reserved and permission to use the figure must be obtained from the copyright holder.)

In tracing these variations and image variants one can find different visual formats, indicating some of the ways in which they are put to work in different situations. There are several near copies with minor modifications such as cropping or stretching (group A in Figure 6.5). There are also screenshots of the image posted indicating the context of the online space in which it was posted such as Instagram posts and tweets (group B in Figure 6.5), often including additional textual elements. These posts include practices of ‘screenshot debunking’, where images of news articles and tweets flagging the image as potentially misleading have been used together with hashtags like #FakeNews and #misinformation. These kinds of images are also overlaid with fact checking or ‘credibility labels’ on platforms where they are shared. In looking into these labelling practices, we found that not all versions of an image are treated equally: one widely circulating version of the image on Facebook has a warning from fact-checking organisations, whereas the same image posted by Macron contains no such warning. One can also see memefied versions, with other visual elements overlaid or juxtaposed, repurposing the original image for satirical takes on the event (group C in Figure 6.5), often with an affected dimension, which may be read in a similar vein to ‘reaction gifs’ (e.g. animated images used to express emotions in online conversations).

Digital objects, methods, ecologies

What can digital objects tell us about ecological politics? The empirical vignettes in the sections above examine how digital objects may serve as a way to explore: (1) the algorithmic meditation of environment events and how the 2019 Amazon fires were organised, staged, and narrated as trending events; (2) which kinds of concerns, framings, entities, and invitations to action are ‘doing well’ online by following and querying objects of engagement such as posts and links; (3) how relations between different concerns and groups are invited and displayed by means of the interpretation of co-hashtag networks; and (4) the roles that images played in online activity around the fires, looking beyond misleading photographs to the social lives of networked images and what they disclose.

What kinds of understandings of environmental politics can digital methods enable? Another recent study suggests that following prominent hashtags shows how the 2019 Amazon fires were construed as a planetary problem – for example, with framing such as ‘lungs of the earth’ – framing Bolsonaro as responsible and animals and Indigenous Peoples as victims.⁶¹ Delving further into the Twitter archive provides further insights into how digital objects were involved in the making, contestation, and negotiation of meanings, representations, and relations around the fires. As well as surfacing different kinds of actors, concerns, and invited action (witnessing, boycotting, donating, debunking, amplifying, sharing, signing), the Twitter archive was used to elicit different conceptions of what the Amazon fires were as an environmental event.⁶² Through collaborative interpretation exercises with our civil society and journalist collaborators, digital objects in the archive served as a prompt to rethink the making of environmental events online and what it means to talk about ecological governance, stakeholders, science communication, and misinformation in this context. Amidst the many lively responses and counter-responses in the archive, the absence of live reporting involving affected communities, human and not, present at the scene of the fires is even more conspicuous.