Participants: Gabrielle Aguilar, Betsy Brossman, Illy Fajar, Paola Gardino, Laura Hoogenraad, Alice Novello, Stephen McNulty, Marjolijn van Raaij, Evi Roelfs, Ayoub Samadi, Nicoló Saracco, Esther Schoorel, Rachel Solis, Laura Temmerman, Sarah Vorndran, Olga Zvonareva.
Facilitated by Carlo De Gaetano, Sabine Niederer, Warren Pearce and Natalia Sánchez Querubín
“If we wish to avoid climate catastrophes, we must pursue a different future than the one we are on track for today. However, there is a problem: How do we move toward a future that we cannot imagine? Inspired by this problem, there is a unique task which accompanies fighting climate change: imagining what the world looks like in which we do succeed.”
Imaginaries is a term used to describe how people imagine their collective life. It includes sets of values, institutions, relationships, and objects that facilitate sense-making. The term is often employed together with words such as ‘social’ and ‘cultural’ (e.g. cultural imaginaries) and studied in the social sciences. In our project, we pair the idea of imaginaries with climate futures. Imagining what the world could look like — which is different from modeling the future — after climate change or in alternative settings to current conditions has been the work of science fiction writers, artists, speculative designers, environmental communicators, and also policy-makers. For example, Levy and Spicer (2013) study the role of competing imaginaries in shaping climate policy. These are ‘fossil fuels forever’, ‘climate apocalypse’, ‘techno-market’, and ‘sustainable lifestyles’. Some genres from the literature and arts have become recognizable such as afro-futurism and solar-punk. Each provides a version of the world that can be sensed and that may lead to sense-making, using Mathew Fuller and Eyal Weizman's (re-)definition of aesthetics.
In our research, we address platform imaginaries as how an issue is imagined by a platform. As a starting point, we take Noortje Marres’ (2017, p.132) idea of digital platforms as hybrid assemblages encompassing users, algorithms, cultures of use, and so on. We understand these platforms as agents in the creation and circulation of alternative climate imaginaries capable to bypass dominant climate imaginaries that are visible in the mainstream media.
To explore the question of climate futures imaginaries on social media — or rather of social media as a space where such futures are being staged and discussed - we propose a multi-platform research protocol. The platforms we focus on are Instagram, Facebook, Reddit, Tumblr, Pinterest, and Google Images. We approach them through the lens of three eco-fiction genres: afro-futurism, solar punk, and cottagecore.
What kinds of future imaginaries do Afrofuturism, cottagecore, solarpunk present online?
How are these imaginaries ‘put to work’ on the different platforms?
Taking a platform perspective, how do different platforms present climate imaginaries?
Using visual materials (top 200 images) for each climate imaginary genre and applying machine vision (Memespector - Google Vision API - features: labels and text) to distil different entities and settings in the images. Then close reading these image-label networks (using Gephi) and collectively exploring and annotating clusters in Miro. Using a combination of visual network analysis, close-reading, and word clouds (from posts captions and text extracted with machine vision from the images themselves) to analyze genres across platforms (e.g., solarpunk on Instagram, Pinterest, Reddit, Google Images, etc.). Additionally, we analyze platforms across genres (e.g., comparing Solarpunk, Cottagecore, and Afrofuturism on Instagram).
Instagram data was collected through Crowdtangle, querying Afrofuturism, Solarpunk, and Cottagecore (separately) with the custom date range: 1 January 2013 - 1 January 2022. The results were arranged by ‘total interactions’ using the filter available on the Crowdtangle dashboard view. Three separate .csv files (one for each query) were downloaded, setting a maximum threshold of 10k posts. The .csv files were uploaded to Google Spreadsheet.
Instagram images' direct URLs have the tendency to expire quickly. In order to retrieve the original images, a workaround was necessary:
Add an empty column after the column ‘Link’, add in the second row the string ‘media/?size=l’ and fill down.
Add an empty column after the previous one with the formula ‘=CONCATENATE’, and combine the link URL in the column ‘Link’ with the string ‘media/?size=l’
The new combined link redirects to the image file of each post, and can be used to preview the image directly in Google Spreadsheet (with the formula =IMAGE), or can be copy pasted in tools like TabSave to directly download images from a list of URLs.
For each query. we downloaded the top (most interacted with) 200 images.
We downloaded Reddit data using Crowdtangle, querying Afrofuturism, Solarpunk, and Cottagecore (separately) with the custom date range: 15 November 2013 - 15 November 2021. The results were arranged by ‘total interactions’ and filtered by ‘’Photos’’, using the filter available on the Crowdtangle dashboard view. Three separate .csv files (one for each query) were downloaded, setting a maximum threshold of 10k posts.
From the .csv files we selected the top 200 (by interactions) Reddit posts. We added three columns after column Z, added “media/?size=l” to the end of the URLs in order to allow us to view the image alone in the Image Size column. We added “=CONCATENATE()” (only did that for one set of columns and then drag the small blue square down and it will automatically adjust) to the Concatenate column. We used the formula =IMAGE() to preview images directly in the CSV. We copied these new links and used the Google extension TabSave to download 200 images.
Data from Facebook had previously been collected by a team of researchers at the University of Amsterdam using the CrowdTangle data scraping tool. The three search terms were used to scrape data from publicly-accessible pages and groups, for the date range: 15-11-2012 to 15-11-2021, which generated a .csv file containing metadata from 10,000 posts, including the text content, their publication dates, numbers and types of reactions, numbers of comments, URL links to posts and usernames. Of these, the most liked 200 posts were subsetted for the purposes of this research.
For these 200 posts, direct image URLs were manually retrieved by copying them from the original posts into the .csv file. These images were then scraped using the Google Chrome extension TabSave. The resulting dataset contained 200 images. Posts containing multiple images were ignored and only the first image was collected, and where videos were included in the original posts, only screenshots of these videos appearing in the posts were collected.
Analysis was then conducted in two stages, as described above, utilizing “machine vision” and “human vision”.
We collected Tumblr data using 4CAT, by querying Afrofuturism, Solarpunk, and Cottagecore
with the timeframe: 15-11-2012 to 15-11-2021. The tool will scrape all the posts within the time range, up until (around) 100k posts. For the query Solarpunk we collected 9052 posts, for Afrofuturism we collected 7935 posts and for Cottagecore we collected 143445 posts. The post collection starts with the most recent post and goes back in time until 4CAT reaches the last published post or the scraping limit. In this sense, for the query Cottagecore the results were so many that we could only collect posts published since 2017.
The resulting .csv files were opened in Google Spreadsheet and arranged by number of notes, which is an indicator of all of the reblogs, likes, and replies that a post has received.
We selected the first 200 posts with more notes, and used the direct image URLs to download the images with TabSave. For posts with more than one image, we only downloaded the first.
The Tik Tok dataset was collected with the tiktok scraper. For each query, 200 most relevant videos with metadata were downloaded. We copied the image URLs in the column covers.origin to download the 200 cover images using TabSave. Data was downloaded on 12 Jan 22.
We collected image posts (referred to as “pins” within the platform) from Pinterest search engine results when querying ‘afrofuturism’, ‘solarpunk’ and ‘cottagecore’ as unambiguous terms (Rogers, 2019). We created a new Pinterest account to ensure that search engine results were not influenced by searches from searches on personal accounts.
Each term was used as a search query in the Pinterest search bar. To view more images on the computer screen, zoom out (by pressing Ctrl and - on Mac). Using the “DownThemAll” Chrome Extension, download the images from the results by selecting the “Media” tab, creating an associated “subfolder” and then selecting “download”.
To have a dataset of 200 images total, we downloaded roughly 300 “media” because DownThemAll also downloads advertisements and duplicates (as in various qualities of the same image). We opted to manually delete the lower-quality versions of the same files.
Nature is depicted as intertwined with architecture and infrastructure;
More focus on the things technologies enable than on what powers them (e.g., surprisingly little solar panels);
Minimal differences between platforms - though some focus more on the aesthetic, others on commentary;
Criticism of the aesthetic as being superficial: not enough focus on actual ideas on improving society.
Summary of Key Findings
Escapism from expectations/constraints of modern, patriarchal society. Cottagecore is a female-dominated aesthetic, which shows how nature can be used as an escape from a patriarchal urban society.
Nature = female, nurturing, maternal
Urbanism (implicit) = patriarchal, industrial/capitalist
Advocates a simpler life without technology. Celebrate nature while we can.
Ironically, the nostalgic romanization of a simpler past without technology and the escape from an industrialized pace can often only be attained through technology, such as the user’s phone or computer-generated worlds, as well as through commercialized objects, such as frog merch.
While SOGI inclusivity stands at the forefront, people of color, disabled people, and unconventionally attractive body types come up short in this imaginary.
Findings and discussion per platform
Figure 1. All 200 image results for Facebook categorized by themes determined by Memespector.
Facebook images (Figure 1) were mostly fantasy photoshoot-style images, where individuals, primarily white females, posed in wilderness settings. This dataset also consisted of a large group of cottage-style houses, matching the typical cottagecore aesthetic. There were some meme-format images, though not as prominently as we saw on Instagram. Flowers and plants often appeared in this dataset, whether they were vines growing on a cottage, flowers weaved into a crown on an individual’s head, or plants decorating a home. ‘Floral designs’ appeared in 23.5% of results, 23.5% represented images with graphics, 60.5% portrayed ‘plaid’ in some way, and 30.5% of images showed ‘skin.’ There were some outliers, such as a celebrity section with images of Taylor Swift and Dakota Johnson, but overall on Facebook, we found images to be much more aligned with cottagecore themes than on Instagram.
Cottagecore on Facebook seems to respond to the climate emergency by advocating a return to nature: a celebration of the beauty of the natural world, while we still have it, living in a simpler technology-free way and drawing on the rural as an inspiration for how we live our lives. The data reflect this through their noteworthy thematic preoccupation with the natural world, depicted both directly and indirectly. In the former case, there is a wealth of images of rural landscapes and typical countryside dwellings (e.g. thatched cottages), seemingly spotlighting idealised, peaceful, and prototypical places to live, devoid of urbanity. In the latter, nature is also clearly drawn on indirectly as an inspiration for artefacts of modern life, as is evidenced by the numerous portraits of individuals interacting with nature - surrounded by, and often wearing clothing including, plants and animals (e.g. leaves and frogs). Indeed, where pictures of modern (information) technology are entirely absent from this dataset, nature is again drawn on as an inspiration for the design and use of everyday technology and interior design, with, for example, images of crockery in the shape of frogs and toadstools; while modern technology is used only for the dissemination of the cottagecore aesthetic - e.g. the creation of memes spotlighting a desirable aspect of this lifestyle, or software such as Minecraft being used to reproduce natural landscapes. Noteworthy too is the presence of images depicting affection between women against natural backdrops.
Cottagecore on Facebook, as on some other platforms, appears to advocate a rejection of the modern, urban lifestyle, as well as the rules and constraints of modern society, and a retreat to a (highly idealised), simpler lifestyle, such as that prior to the industrial revolution: free, physically, from the crowded spaces and suffocations of modern infrastructure and technology, and liberated social norms such as those governing gender and sexuality. This being said, it is noteworthy that racial diversity was almost entirely absent from the subset studied, with almost all images which included people only depicting young white, caucasian women, and without visible disabilities. Indeed, data from other platforms would appear to underline this as a known criticism of the cottagecore aesthetic as a whole (e.g., Kay, Boyce and Wood, 2021, p. 5).
Figure 2. All 200 image results for Instagram categorized by themes determined by Memespector.
On Instagram (Figure 2), we determined the key themes of our dataset to be grouped in the categories clothing, food, feminine energy, and most commonly, text-based images. In total, 50.5% of the images within this data set were text-based, or at least had some version of a caption within the image. This was the highest proportion of text-based images of all four platforms. We determined this to be likely due to Instagram’s higher than average maximum character limit for captions, which, as a result, allowed the hashtag #cottagecore to be used as a tool to earn more engagements on memes from various accounts, despite the image having nothing to do with the actual cottagecore aesthetic. Plants were the next highest keyword in the dataset and appeared in 42% of images.
Figure 5. All 200 image results for Reddit categorized by themes determined by Memespector.
Our Reddit dataset (Figure 5) varied in topics, consisting mostly of the categories humans, nature, arts, fashion, and text-based images. Many of these categories overlapped with each other, for instance, human/fashion-related images showed humans posing in various cottagecore-themed outfits or costumes. Memespector determined that ‘skies’ appeared 16% of the time and the human body appeared 21% of the time. We only saw 16% of text-based images on Reddit, significantly less than we saw on Instagram.
Figure 4. All 200 image results for Google categorized by themes determined by Memespector.
Finally, Google Image Search results were categorized into clothing, humans, cottages/nature, interior design, homemade food, and text-based images (Figure 4). There were many naturescape images, images of cottages with vines growing up the walls, trees with fruits hanging, and housing interiors that incorporated elements from the natural world into the designs. These Google images very closely match what one may originally imagine when thinking of cottagecore, as most show a call-to-nature way of life, with very few outliers.
In total, of the 800 images evaluated across all platforms, 50.5% of them showed images of ‘plants,’ 24.8% showed ‘grass,’ 22.3% showed ‘tree,’ and 18% showed ‘sky,’ reinforcing nature as the key theme for cottagecore. While 24% of the images showed ‘font,’ meaning some kind of caption or text-based elements, usually in a meme format, we know that 52% of those were from Instagram alone. Other keywords and phrases that showed up in at least 13% of images were ‘happy,’ ‘building,’ ‘flower,’ ‘people in nature,’ ‘window,’ ‘art,’ ‘wood,’ ‘sleeve,’ ‘organism,’ ‘house,’ and ‘natural landscape.’
Coined by author Mark Derry in the novel Black to the Future (1994), "Afrofuturism," draws from Afrodiasporic experience and an Afrocentric Culture to re-imagine a future flush with art, science and technology through a Black lens. Through art, it radically reimagines past, present, and future.
The following three platforms were analyzed: Instagram, Pinterest, and Tumblr. Across these three platforms “Afrofuturism” places people at the forefront of the imaginary, depicting people in nature or people with technology. Afrofuturism draws from traditional African heritages, architectures, and fashion styles and fuses it with futuristic and imaginative technology. It does so in response to real past and present problems, including histories of colonialism and structural racism. Through art, it radically reimagines past, present, and future. The Afrofuturist imagination places black people, histories, and heritages at its center. Whereas in some examples histories of oppression are rewritten (think for example of ‘Black Panther’, an often-used reference within the genre) in others it seems to be erased altogether.
A major trope through which the link between people and technology appears in Afrofuturism is through the representation of the trope of the ‘cyborg’. Cyborg in the genre is imagined by merging common elements of cyborg imagery with pan-African or tribal elements.
On Tumblr, Comics are a prominent medium that allows for the use of imagination to shape this specific technoculture. It is also used as a community space. The genre of Afrofuturism on Tumblr is characterized by the representation of black women, who are often depicted as goddesses of cyborg-like people. However, the fact that events and open discussions are also present in the Tumblr top posts suggests that Afrofuturistic subjects are not individualistic figures, but they are uplifting their own communities.
The aesthetic of Afrofuturism on Tumblr imagines the person of the future as a black woman who transcends the barriers of gender and race and shifts current power relations. They take inspiration from great empires of the past, like Egypt, and incorporate traditional or tribal accessories and symbols to create the cyborg-person of the future.
On Instagram, digital art and Afro-anime are central techniques through which Afrofuturism is expressed in images. Digital art and Afro-anime are central techniques through which Afrofuturism is expressed in images on Instagram. Fashion is another important medium. Fashion is another central subgenre on this platform. Afrofuturist art on Instagram features the trope of the ‘cyborg’ and the Afroqueen and -princess. Women are also at the forefront of Afrofuturist depictions on Instagram.
On Pinterest, a wide variety of art forms are employed– ranging from graphic design to photocollage and there is an abundance of color compared to other eco-fictions genres. Afrofuturism on Pinterest heavily centres around envisioning people--all 200 images had a person(s) in it. It therefore places people at the forefront of nature and technology and is predominantly Afro-centric. The collection of images also prominently display women / female-presenting individuals. Technology draws from a lot of cyborg imagery but also highlights older/cultural technologies/artefacts. Art acts as an anchoring piece across all areas.
Finally, Afrofuturism on Tumblr, Pinterest and Instagram does not directly refer or draw on discourses of climate change. Unlike in the other eco-fiction genres explored in this project, the link between nature and technology is hence also not a central theme of Afrofuturism. Rather, this genre concerns itself with a future defined by people, culture and art. Furthermore, the past and present are generally considered within a negative light. This is in contrast to Cottage Core, which takes a nostalgic approach to the past.
a) staging a photoshoot (e.g. cottage picnic) or sharing an Afro-futuristic artwork;
b) reframing existing scenes & objects via photo + tagging (e.g., cool building that looks solarpunk);
c) sharing architectural renderings and animal crossing scenes.
These imaginaries are put to work by:
Offering a way into alternative realities. Images cannot tell the whole story. Rather, they offer windows into other realities and appeal to the affective. One can feel how it would be to be there and play with an embodiment. Also, they inspire DIY engagement with these imaginaries: through synthetic renderings, art, design, fashion, craft, memes, etc.
These visuals circulate in ambivalent spaces, where mockery, irony, and critique occur. Critique comes forth often via 'climate memes'. When studying climate futures imaginaries online, one needs to take into account these dimensions of ambivalence.
Pearce, W., and C. De Gaetano (2021) Google Images, Climate Change, and the Disappearance of Humans. Diseña (19). Doi: https://doi.org/10.7764/disena.19.Article.3
Pearce, W, S. Niederer, S.M. Özkula, and N. Sánchez Querubín (2019) The social media life of climate change: Platforms, publics, and future imaginaries. WIREs Clim Change (10)e569. Doi: https://doi.org/10.1002/wcc.569
Castells, M. (2004). The rise of the network society. Blackwell Publishing.
de Zeeuw, D., & Tuters, M. (2020). Teh internet is serious business. Cultural Politics,
16(2), 214–232. https://doi.org/10.1215/17432197-8233406
Ghosh, A. (2016). The Great Derangement: Climate Change and the Unthinkable. Chicago, IL: University of Chicago Press.
Guattari, F. (2014). The Three Ecologies. Bloomsbury Academic.
Iossifidis, M. J. M. & Garforth, L. (2021). Reimagining climate futures: Reading Annihilation, Geoforum, Elsevier. https://doi.org/10.1016/j.geoforum.2021.12.001
Kennedy, T. M. (2014). Sustaining white homonormativity: the kids are all right as public
pedagogy. Journal of Lesbian Studies, 18(2), 118–132.
Mies, M., & Shiva, V. (2014). Ecofeminism (2nd ed.). Zed Books.
Mohanty, C. T. (1984). Under Western eyes: Feminist scholarship and colonial discourses. Boundary 2, 12(3), 333–358. https://doi.org/10.2307/302821
Parkins, W. (2004). Out of Time: Fast Subjects and Slow Living. Time & Society, 13(2–3), 363–382. https://doi.org/10.1177/0961463X04045662.
Reggev, K. (2020). What Exactly Is Cottagecore and How Did It Get So Popular?, Clever. https://www.architecturaldigest.com/story/what-exactly-is-cottagecore.
Tsing, A. L. (2015). The Mushroom at the End of the World: On the possibility of life in capitalist ruins. Princeton University Press.
For the DMI project that started this inquiry, see: http://climaps.eu/#!/map/mapping-cli-fi-scenarios-book-covers-with-landscapes-issues-and-personal-narratives
For an introduction to the Climate Futures project by the Visual Methodologies Collective, in which cli-fi novels were used to train AI, there is this podcast interview with Sabine Niederer, Carlo De Gaetano and Andy Dockett, by Morgane Billuart on Spotify and Soundcloud.
Climate fiction genres glossary: https://grist.org/fix/afrofuturism-to-ecotopia-climate-fiction-glossary/
Data sets (links)-- SabineNiedererDMI - 22 Feb 2022