In 1947, André Malraux posed the idea of a musée imaginaire – an imaginary museum personalised to the individual – which brings together photographic representations of artworks from across the world and throughout history (Malraux 1967). This museum without walls has been seen as an antecedent of the ever-larger collections of digital artworks we can access via the internet, whether mediated by social media (Kelly 2010), virtual exhibitions (Catton / Smith 2021), or computer vision (Malevé 2021). As an increasing number of museums are making digital reproductions of their collections available online, the difficulties of sharing data and conflicting organisational policies have led to a plurality of online collections, many of which are stand-alone initiatives.
While the exceptional digital initiatives of the world’s ‘superstar’ museums are the subject to their own case studies and newspaper articles, the majority of collections share similar interfaces and user experiences, often relying on third-party software and content management systems (Gombault / Allal-Chérif 2021). It is these prevalent digital environments – the simple collection searches and tiled thumbnail images we use without conscious thought – that we have chosen to focus on (Drucker 2013). Mirroring Bishop’s analogue distant reading of modernist architecture, our paper aims to investigate “the mainstream, the mediocre, the déjà vu” – in short, the graphics user interfaces and everyday visual experiences that we have come to see as unremarkable (Drucker 2013; Bishop 2018).
To this end, computer vision is a useful tool for understanding the vast amount of visual data available, and it enables us to look beyond the largest organisations to the whole sector. This paper combines the use of computer vision on scraped digital collection screenshots with an analysis of the html code of over 50,000 European museum webpages, exploring their chosen hosting solutions, use of alt-text, other accessibility features, languages, and most prominent visual attributes. This methodology is then supplemented with in-depth case studies of websites that are representative of wider trends (Brey 2021; Schmidt et al. 2020). By combining ‘distant’ and ‘close’ viewing in this way, we aim to contribute to the ongoing debate surrounding the disjoint between the formalist and iconographic methods – prevalent in digital art history – and the critical theory more common to art history today (Moretti 2000; Näslund Dahlgren / Wasielewski 2021); Balbi / Calise 2023).
Aware of the homogenising tendency of digital representations of artworks, through analysing the digital environments and not the digitised artworks themselves, we have chosen to focus on born-digital artefacts (Manovich 2009). While our screenshots are a static representation of what is ultimately a dynamic experience, we have used computer vision of as just one tool amongst many, and it is used in conjunction with – not as a replacement for – visual and art historical analysis.
With third-party software solutions for digitised collections increasing in popularity, our paper asks how the prevalence of certain layouts and conventions might affect how we view digital images. What distinguishes the viewing experience of one online museum from another? And, what do these experiences assume of their audiences? Drawing from the discussions of the “white cube” galleries, we aim to situate our findings in a much broader art historical debate about the role of context on viewer’s perception (Filipovic 2005; O’Doherty 1986). Ultimately, we ask to what extent the dominant layouts and modes of interaction of online collections are changing the way we view digitised artworks.