Jan Bot: a Surrationalist Historiographer
Reflections from a digital scholar on a filmmaking bot
Since I first encountered Jan Bot in May 2017 in the context of a presentation at Eye Filmmuseum’s Collection Center I have been fascinated by the videos it generates. I especially felt that they resonate strongly with ideas I am developing about the relation between artistic and scholarly approaches to digitised film collections.
Jan Bot’s videos consist of clips from Eye Filmmuseum’s Bits & Pieces collection and of intertitles based on text snippets from contemporary news items. Bits & Pieces was established by the Filmmuseum in the late 1980s as an initiative to preserve film fragments that the curators could not identify but nevertheless deemed worthy of preservation because of their striking aesthetic qualities.¹ The content for Jan Bot’s intertitles is sourced from news items retrieved using Google Trends. Both the film snippets and the trending news are analyzed automatically using automatic video content analysis and keyword extraction, respectively. The results are subsequently used to generate new works that seek to match film and text semantically.²
Borrowing a term from philosopher of science Gaston Bachelard, I qualify Jan Bot’s exploration of scientific approaches to automated content analysis as “surrationalist”.
Currently, Jan Bot creates approximately twenty videos each day. The resulting videos — typically 20–30 seconds in length — juxtapose the footage and text material through a frenetic montage and, in their elucidation of one another, make for unexpected associations, while also frequently bordering on the nonsensical.
By creating these videos, Jan Bot invites us to reflect on the meanings assigned to archival films in digital archives, social media environments, and in relation to online news consumption through the lens of generative, experimental filmmaking. Beyond this, the videos’ poetic approach to data-driven content analysis (of archival film and online news) also offer an urgent artistic counterpoint to current deployments of digital methods in archives and, by the same token, digital scholarship in film historical research. Borrowing a term from philosopher of science Gaston Bachelard, I qualify Jan Bot’s exploration of scientific approaches to automated content analysis as “surrationalist”.
I will focus on two aspects of Jan Bot’s work, namely machine learning as applied to archival film for metadata creation, and analysis of film circulation on social media platforms. With regard to the first, I argue that Jan Bot challenges the evidentiary, scientist inclinations of current video analysis software and turns it in a cinephile direction that harkens back to classic avant-garde filmmaking. As for the latter, I suggest that Jan Bot’s experimental reason has devised a compelling method for elucidating archival film’s circulation and reception on social media, potentially to the great benefit of scholars who research historical and contemporary film distribution and reception.
In order to explain my fascination with Jan Bot’s videos in relation to current digital scholarship and my reasons for wanting to make these points, let me first tell a bit more about my scholarly background and the research I was doing at the time I first encountered the videos.
Deformative Criticism in Digital Scholarship and the case for Surrationalist Film Historiography
At the point in time I saw Jan Bot’s videos for the first time, I had just defended my dissertation at the University of Amsterdam — Film history in the making. Film historiography, digitised archives and digital research dispositifs (2017). In the context of my research, I had produced a critical overview of emerging digital techniques and methods for film historiography, while discussing their epistemological underpinnings.
My reason for undertaking this work was that I had observed how film and media historians were in recent years increasingly beginning to make use of computational, quantitative techniques to analyze and visualize patterns in digitized archival films. Using digital methods to produce data and graphical representations — such as diagrams, graphs and interactive maps — scholars were slowly beginning to produce evidence in their research in new ways, for instance for the purpose of style analysis of historical film data, or socio-historical analysis of film distribution data. In doing so, scholars began adopting tools, representational practices and forms of reasoning from the sciences.³
Scholars should begin to push data-driven research in new directions and pursue an experimental, reflexive approach to film history that engages poetic forms of moving image appropriation.
Considering this development from the perspectives of philosophy of science and film historiography, my research’s aim was to elucidate how knowledge production in specific research traditions — stylistic history, film philology and socio-economic cinema history — had changed as a consequence of digitization, while also indicating future directions and methods I thought were currently missing. In the dissertation’s final chapter, I argued that scholars should begin to push data-driven research in new directions and pursue an experimental, reflexive approach to film history that engages poetic forms of moving image appropriation informed by cinephile and surrealist theory.
Using the playful terminology of scientist and philosopher Gaston Bachelard I suggested that there is a need for a “surrationalist” (surrationalisme) approach to film historiography that concurrently aims to critique and advance contemporary modes of knowledge production in film archival research.⁴ Influenced by psychoanalysis and surrealism in the 1930s, Bachelard coined this term to articulate a dialectic approach that embraced logical and formal deduction’s rationalities and scientific methods, while constantly questioning them to open new paths for scientific discovery. In Bachelard’s words, this implied to ”take these forms — after all purified and economically arranged very well by logicians — to fill them up psychologically and put them back into motion and life again”.⁵
Digital scholarship should enable and encourage the languages, technniques and cultures of data, statistics and algorithms to intermingle with poetic gestures of moving image appropriation and cinephile appraisal, as a way to develop experimental forms of reason.
In doing so, Bachelard argued that surrealist poetics could play a crucial role in creating an ”experimental reason” as a scientific pendant to the surrealists’ “experimental dream”, with which to question science’s formalized methods, assumptions and rigour and develop imaginative alternatives. Inspired by this proposition, I argued that digital scholarship should enable and encourage the languages, techniques and cultures of data, statistics and algorithms to intermingle with poetic gestures of moving image appropriation and cinephile appraisal, as a way to develop experimental forms of reason.
In making this suggestion I was, in addition to Bachelard’s concept, also inspired by recent work of film scholars Katherine Groo, Kevin Ferguson, André Habib and Catherine Grant, among others. Drawing on early surrealist, cinephile film theory and verfremdung strategies in their use of digital methods for video analysis, these scholars have in recent years underlined the great value of defamiliarizing films as objects of study through poetic gestures and idiosyncratic readings (or through “videosyncratic” experiences to use Marijke de Valck and Malte Hagener’s great term) to expose, reflect on and challenge current technical conditions for historical interpretation.⁶
As Grant has put it, poetic appropriations that make otherwise familiar objects look a bit strange, for instance by using video editing to juxtapose or associate film clips in unexpected ways, can “unsettle a ‘professional cosiness’ of traditional historicism” and yield new critical perspectives on film analysis.⁷ While the scholars mentioned and cited here do not all explicitly use this term, this trend in film studies may be seen as calling for a deformative criticism, that deploys digital tools to produce readings of films against the grain of traditional, historical scholarship.
Instead of seeking to fit archival films into classic notions of stylistic schools or linear accounts of film history’s development — which cinephiles traditionally did when cinephilia emerged in the 1920s to legitimize film as an art form with its own history — such work instead reclaims a cinephile tradition to highlight idiosyncratic and anecdotal experiences of film viewing that do not fit into such accounts.
In this regard, artists play an important role. In developing such an approach, scholars look beyond film studies to a long tradition of experimental filmmaking — from Joseph Cornell’s Rose Hobart (1936) to Gustav Deutsch’s Film Ist series (1998-) that through artistic interventions create — as Habib has formulated it — a:
…movement back and forth between historical understanding of the past and (…) reactivation in artistic works [which] is essential for illuminating our knowledge and enriching our experience of film history and, in a more general way, our apprehension of what is in the film archives.⁸
Taking the cue from the notions and concerns of the scholarly and artistic work I have sketched here, I was convinced that in making this suggestion, I could contribute to opening new avenues for a fruitful dialogue between digital film historiography and artistic practices. However, when making this suggestion in 2017 I also observed that while there do exist several inspiring examples of visualizing data relating to classic films in media art, very few artists seemed to seek out archival film material to appropriate them artistically with algorithmic means.⁹ For these reasons I was enthusiastic to encounter Jan Bot’s work because I found it to fulfil the ambition of challenging digital film historiography’s modes of knowledge production while offering imaginative, cinephile alternatives for experimental reasoning.
While there do exist several inspiring examples of visualizing data relating to classic films in media art, very few artists seemed to seek out archival film material to appropriate them artistically with algorithmic means.
Having since had the time to see Jan Bot develop I share a couple of observations on two aspects in which I think Jan Bot achieves a kind of deformative criticism particularly well, while creating a new experimental reason. First, I discuss in the following sections how Jan Bot’s artistic strategies challenge notions of film as evidence in contemporary metadata creation in moving image archives that rely on automated content analysis. Subsequently, I argue that the resulting videos may be considered as more than primarily artistic productions and may also be seen as highly topical for contemporary research on film distribution and reception.
The evidentiary visual regime of automated content analysis in audiovisual archives
Jan Bot’s work is particularly interesting in how it challenges the notions of evidence at work in automatic tools for metadata creation in moving image archives. In order to understand why, it is helpful to first consider a few core aims of film cataloguing and current developments in automatic indexing. Written catalogue descriptions contained in archival film’s metadata remain at the core of film heritage institutions’ catalogues and offer a basis for scholarly research and reuse of footage in new productions. They are made, in most cases, by archivists following rigorous guidelines in order to achieve the best possible accessibility and retrievability of moving images and related materials.
Since film heritage institutions began to systematically build collections and develop catalogues in the 1930s, they have traditionally catered to two types of film historiographies: a history of film as an art form, and of film as source material for social and cultural history. Initially, archiving initiatives emerged primarily from cinephile film clubs and societies but have, in more recent decades, become increasingly professionalized and scientifically codified. Especially since the 1970s, metadata creation has increasingly followed well-defined taxonomies to produce accurate descriptions of the content, physical characteristics and life of archival objects to secure their retrievability. Typically, such taxonomies adhere either to international standards defined within networks of film heritage institutions — in particular those of the International Federation of Film Archives — or specific institutional guidelines.
While metadata are mostly created by humans, they are in recent years increasingly generated with various software for automatic feature extraction and, subsequently, metadata creation. For instance, by using machine learning to automatically detect objects, locations and concepts in digitized film material. An important reason for taking such an approach is to make the workload lighter for archivists as metadata creation is extremely labour-intensive. Furthermore, it makes it possible to analyze historical patterns in archival material as an evidentiary basis for researching events, individuals and social movements.
The development of such tools has a decades-long history in broadcast archives, that have been experimenting with artificial intelligence for metadata creation much longer than film heritage institutions.¹⁰ One explanation as to why broadcast archives have a longer history in this regard is that in addition to wanting to lighten the task of metadata creation, such institutions in many cases cater directly to television production or, in some cases, need to generate profits from footage sales as a substantial part of their business model. For this reason, they need as much metadata as possible on their archival material’s content to optimize access and sales.
Having lagged behind these developments, film heritage institutions are slowly catching up and developing similar methods for film collections. On a European level, the current project I-Media-Cities sees a coordinated effort among nine film heritage institutions to develop software for recognizing objects, places, buildings and persons in order to make archival film accessible for socio-economic historical research. Likewise, the Danish company Vintage Cloud is currently developing film scanners that can produce such metadata in real time, presented by the company as Smart Indexing.
While the algorithms of such efforts are still not very robust and applied primarily on an experimental basis, they hold great potential for film heritage institutions and, when they work well, their affordances in terms of improving access are obvious. Yet, beyond practical considerations of lightening the tasks of metadata creation and improving access to film archives for historians, one may also question their limitations.
In particular, such methods impose a specific regime of vision that favours clear indexicality in the quest to recognize objects, places and people. While this may help semantic content retrieval and produce a great variety of useful entry points, it is less productive in drawing attention to image details or gestures which in unexpected or accidental ways may appear beautiful or intriguing. In other words, this regime of vision can be said to neglect an idiosyncratic form of appreciation that highlight surprising details and moments, something which a more classic cinephile contemplation of films has traditionally valued and would arguably be better in discerning.¹¹
Jan Bot’s Surrationalist 8 o’clock news: the Bits & Pieces as absurd ‘fait divers’
Jan Bot’s approach to feature extraction and metadata creation as a basis for classifying and appropriating videos operates according to a logic that challenges the applications I have discussed above and makes an interesting intervention in this regard. This approach strikes me as a form of deformative criticism that harkens back to classic avant-garde filmmaking and, in doing so, opens an avenue for a cinephile, idiosyncratic mode of appreciation.
Instead of adhering to scientifically codified archival standards and taxonomies and pursuing the development of robust algorithms for feature extraction, Jan Bot taps into the shared consciousness of social media and online news consumption to create metadata as a starting point for playing with the evidentiary status of archival film, using the Bits & Pieces material. Making use of Google Trends, Jan Bot extracts words from current news items, produces tags and associates words with the contents of the Bits & Pieces. Each source is indicated in the data associated with each video. Subsequently, this data forms a basis for automatically generating video appropriation works consisting of video material and text snippets from news items. Finally, the videos and data are gathered in the JAN BOT_ CATA.LOG.
Instead of adhering to scientifically codified archival standards and taxonomies and pursuing the development of robust algorithms for feature extraction, Jan Bot taps into the shared consciousness of social media and online news consumption to create metadata as a starting point for playing with the evidentiary status of archival film.
The tags associated with the video material can be meaningful in rather literal ways, and in some cases suggest strong indexical relations between images and text. For instance, news about the British Royal family are often associated with clips of high society, like in the video called “2018–11–14.005-queen_elizabeth.mp4”.
Among the more striking matches that, to a great extent, appear accurate in a way that traditional feature extraction strives to achieve, one may consider December 23’s video “2018–12–23.002-tsunami.mp4” that links snippets of news items about the Indonesian tsunami to stormy sea images.
For the most part, however, the semantic word-image relations of these matches and the indexical relations they suggest are in a traditional sense broken if not nonsensical. Take for instance the video below about the micro-blogging platform tumblr and its decision to ban adult content:
Broken news
Yet, at its most nonsensical, Jan Bot’s flickering and flashing intertitles still succeed in feeling like a compelling, imperative call to watch and to try to make meaning of the film material. This is achieved by giving cues about recent news events that may resonate in our short-term memory and is arguably also a result of the videos’ insisting aesthetic devices. For instance, the intertitles’ consistent use of all caps — often considered the internet’s equivalent to shouting — may convey the feeling of a newspaper boy shouting out the latest news sensations, of online breaking news banners, or of the sometimes absurdly cryptic clickbait headlines of the tabloid press, designed to create curiosity or astonishment. Thus, the videos’ intertitles attract viewers by offering details and teasers about events that they may recognize, while never offering explanations or entirely accurate illustrations. In other words: Jan Bot’s news are not breaking, they are broken.
I believe said features of Jan Bot’s videos can be traced back to classic avant-garde filmmaking’s surrealist and dadaist plays with meanings of words and objects and that doing so may illustrate how they ultimately achieve to reinstate a cinephile, avant-garde attitude in automated content analysis. In particular, watching Jan Bot’s associations between film material and tags, bears an uncanny resemblance to the avant-garde classic Ballet Mécanique (France, 1924), created by Fernand Léger assisted by Dudley Murphy. Centered around the enigmatic news item “Pearl necklace worth 5 million stolen” presented in different variations, the film offers a frenetically paced and radically repetitive visual exploration of shots of industry, objects, working people, Chaplin and Kiki de Montparnasse’s grimaces, ultimately turning the news item’s hint of a narrative frame into an absurd faits divers.¹²
In several aspects, Jan Bot’s videos recall Ballet’s repetitive montage and elliptical evocation of a news item and drama that is never resolved. In a way, the videos feel as if Leger had created silent film newsreels with algorithmic means. For an illustrative example of this resemblance, consider for instance the clip below from Ballet (from approximately 00:09:15 to 00:12:00):
As in Ballet, the intertitles of Jan Bot’s videos announce factual events, but rather than offering viewers enough details to make sense of them, they take them as a departure point for mystifying current news items ad absurdum while inviting viewers to contemplate the images and fill in the gaps. In this respect, Jan Bot’s dissemination strategy of its own works is particularly eloquent in underlining this strategy: every evening one of its videos is published in a synchronous Twitter, Facebook and Instagram post at 8pm (CET) and thus offers an alternative to the local 8 o’clock news. In this way, the videos point back to the users that create and influence news agendas through the trending topics, using the Bits & Pieces material as a prism for defamiliarizing the newsfeed’s content and current approaches to feature extraction. This invites users to critically reflect on the logics and desires of their news consumption while questioning what evidentiary role archival footage may play in relation to current events.
To conclude, one may say that considered in isolation, Jan Bot’s word-image juxtapositions at first glance appear as strictly nonsensical, stochastic dada poetics that are in themselves highly enjoyable for their unpredictable juxtapositions. Yet, regardless of whether the material is illustrative of the events the algorithms associate them with, the videos also invite critical reflection on automated content analysis and contemporary news production and consumption.¹³
By the same token, the videos produce a surrationalist, deformative alternative to current feature extraction and video content analysis in archives and digital film historical research, by playfully challenging their scientific emphasis on indexical accuracy through generative, poetic strategies. Thus, Jan Bot’s reflexive, experimental reason points us to the impossibility of exhausting archival films semantically and neatly fitting them into contemporary news narratives. Instead, it highlights the underlying contingencies of archival film’s evidentiary status and the distance in time between the material’s past expressions and present interpretations of it created with software.
Documenting Film Heritage’s Social Media Dissemination in the JAN BOT_ CATA.LOG
In addition to offering alternative entry points to the Bits & Pieces material and producing a critical perspective on metadata creation with machine learning, Jan Bot’s productions also carve out a space for reflection on online dissemination of archival films. Through defamiliarization, it elucidates how the viewing contexts of social media constitute a changing web of semantic relations, while producing a significant trace of these contexts. In doing so, Jan Bot has imagined an interesting way of documenting the dissemination of digitized film heritage online and leaving traces of its programming context that may offer interesting perspectives for scholars researching film distribution and reception on social media.
For film scholars, understanding how social media environments attribute meaning to the films that circulate in them is important in order to understand the transitional nature and changing specificities of archival film and its reception today. In recent years, film scholars have increasingly studied films’ past and present circulations across physical and virtual sites, to yield better understandings of how films acquire meaning through intertextual or intermedial relations to other films or media outlets, especially in the scholarly HoMER network. This research strand’s focus has emerged and become gradually stronger since the mid-1970s and is prominent especially in the fields of early cinema studies, new cinema history and media archaeology.
By studying the exhibition contexts of films and attending to how relations to other media are established and negotiated, such research develops insights into film’s configurations and specificities as a medium in phases of technological transition. While film and media scholars are beginning to develop extensive statistical research projects on recent film production and circulation — see for instance the impressive research in Australian media scholar Deb Verhoeven’s Kinomatics projects — there have been few coordinated efforts for understanding the intermedial relations between archival film and new media outlets that develop as a consequence of film’s circulation on social media.
Meanwhile, film heritage institutions increasingly strengthen their social media presence as a way to nurture and improve the circulation of their digitized collections and engage with broader user groups. Archival films now circulate in Facebook and Twitter newsfeeds where they are related to current events and news items through curated posts as part of social media strategies. Typically, film museums post digitized archival clips online as a way to offer historical perspectives on current news, engage with communities, or simply to show a beautiful new digitisation. For instance, the picture below shows an example of a travelogue from Sumatra, posted by Eye Filmmuseum on its Facebook page in November 2018 for its beautiful imagery of bridges and rivers — currently counting 729 views.
Such forms of access are increasingly embraced by film heritage institutions that consider and cite view and share numbers as metrics that reflect the success of their digitization and online dissemination projects. It is great for access purposes that archival films now circulate more widely and can be seen in situations that would have been unimaginable just a few years ago. Yet, beyond giving an indication of the popularity levels of archival films online, likes and shares say little about the kind of viewing contexts and receptions that social media offers for watching films.
Jan Bot’s videos offer an interesting meta-perspective on this development. By working with trending topics and the Bits & Pieces as source material, it creates — as discussed in the previous section — videos that conflate archival films and the newsfeed into one singular expression. In doing so, it gives an impression of the news items the footage may have been associated with on a given day.
Moreover, the same footage appears in multiple videos, and is assigned different meanings in each. In this regard it is again relevant to consider Jan Bot’s own social media strategy. The video posted every day at 8pm (CET) is accompanied by the following standard message — beginning with the computer programming crash course idiom ”Hello world” — in which the subject changes from day to day (in this case the video from November 14, 2018 concerning former US Deputy National Security Advisor Mira Ricardel):
🤖 Hello world. Today I generated this video about Mira Ricardel. I hope you like it! Read more about me and my algorithms on jan.bot/20181114.
Through this generative approach, Jan Bot makes changing relations between films and news outlets explicit through continuous re-appropriation while leaving a trace of the news stories they may have been experienced in relation to in its database. When following the updates closely one begins to discern how the same footage is related to widely diverging news items and becomes aware of their shifting meanings. Among numerous examples, take for instance the videos “2018–11–14.007-fallout_76_review.mp4” and “2018–11–14.008-when_is_black_friday_2018.mp4” that in large part rely on the same footage of an aircraft to illustrate news about, respectively, the computer game Fallout 76 and Black Friday.
Becoming aware of such connections between videos, one realizes how Jan Bot makes visible changing patterns and desires of news consumption and information streams and the circulation and changing meanings of archival film material in relation to them. In this respect, Jan Bot has — through its experimental reason — found a way of elucidating and inscribing ephemeral moments of contemporary dissemination and viewing of archival films online which, I believe, scholars would not have imagined.
Because the videos excel in making these relations visible, they may be thought of as traces that scholars can study to nurture novel insights into the dynamics of contemporary film heritage dissemination. Thus, the JAN BOT_ CATA.LOG can simultaneously be considered an interesting collection of avant-gardistic appropriation works while it may also be approached as a highly valuable resource of inscriptions of algorithmic film programming on social media.
Conclusion
Jan Bot’s videos are great fun to watch on their own, and they also offer a welcome — critical — counterpoint to current software-driven knowledge production in film archives. This approach may be qualified as surrationalist insofar as Jan Bot makes an interesting double movement by simultaneously engaging in surrealist play with the scientist ambitions of machine learning and traditional evidentiary notions of film as historical source material, while producing new types of historical traces with a different evidentiary function. Stripping archival footage of meanings that have hitherto been assigned to them with automatic feature extraction, foregrounds the footage’s potential and contingent meanings and creates a repository of works which may be considered as inscriptions of contemporary film viewing on social media.
Ultimately, by achieving this, Jan Bot ties in with scholarly calls for a reflexive, deformative criticism and its ambition to interrogate and question the socio-technical foundations of digital film-historical research through poetic gestures. Far from the usual fare of current feature extraction and film data visualization, it makes things strange and in doing so — to recap Habib’s words — “enrich[es] our experience of film history and, in a more general way, our apprehension of what is in the film archives”.
In this sense, I feel Jan Bot’s experimental reason has truly managed to open new artistic avenues for exploring film archives in a way which speaks very directly to contemporary concerns of film scholars and digital film historiography — probably much more directly than Jan Bot is (and will ever become) aware.
Footnotes
(1) For an interview with Bits & Pieces current curators on the initiative’s history and selection criteria see Christian Olesen, “Found Footage Photogénie: An Interview with Elif Rongen-Kaynakçi and Mark-Paul Meyer”, NECSUS — European Journal of Media Studies, 4 (Autumn 2013)
(2) For a discussion of Jan Bot’s algorithms see Pablo Núñez Palma’s Medium post November 28, 2018, ”Jan Bot’s step by step. The filmmaking algorithm explained”.
(3) Most emblematic in this regard is film historian Yuri Tsivian’s pioneering database and stylometric research project Cinemetrics, see: cinemetrics.lv.
(4) Gaston Bachelard, L’engagement rationaliste (Paris: Les Presses universitaires de France, 1972[1936]) 12.
(5) Gaston Bachelard, op.cit., 12. Original quote: ”C’est de reprendre ces formes, tout de même bien épurées et économiquement agencées par les logiciens, et de les remplir psychologiquement, de les remettre en mouvement et en vie”.
(6) Marijke de Valck and Malte Hagener, “Down with Cinephilia? Long Live Cinephilia? And Other Videosyncratic Pleasures” in Marijke de Valck and Malte Hagener (eds.), Cinephilia. Movies, Love and Memory (Amsterdam: Amsterdam University Press, 2005) 11–24.
(7) Catherine Grant, ”How Long Is a Piece of String? On the Practice, Scope and Value of Videographic Film Studies”, (Paper presented at the Audiovisual Essay Conference, Frankfurt Filmmuseum/Goethe University, November 23–24, 2013)
(8) André Habib, ”Le cinéma de réemploi considéré comme une ‘archive’. L’exemple de A Trip Down Market Street (1906) et Eureka(1974)”, in André Habib and Michel Marie eds., L’avenir de la mémoire. Patrimoine, restauration, réemploi cinématographiques. (Villeneuve d’Ascq: Presses Universitaires du Septentrion, 2013) 151 (emphasis in original). Original quote: ”mouvement d’aller-retour entre l’intelligence historienne du passé et sa réactivation dans les oeuvres artistiques, est essentiel pour éclairer notre connaissance et enrichir notre expérience de l’histoire du cinéma et, de façon plus générale, notre appréhension de ce qui se trouve dans les archives du cinéma…”.
(9) See for instance film scholar Kevin L. Ferguson’s very exciting work on film data visualization inspired by contemporary media art practices: http://www.digitalhumanities.org/dhq/vol/11/1/000276/000276.html
(10) For a succinct historical overview of this history, see Brecht Declerq’s Medium post November 28, 2018, ”Jan Bot or a new step in the demystifaction of feature extraction technology”.
(11) For a discussion of historical and contemporary cinephile viewing habits see in particular Christian Keathley’s excellent book Cinephilia and History, or the Wind in the Trees (Indiana University Press, 2005).
(12) François Albera, L’Avant-garde au cinéma (Paris: Armand Colin, 2005) 78.
(13) In this respect, as both Brecht Declerq and Elif -Rongen Kaynakçi point out, Jan Bot may actually be considered a rather logical extension of the Bits & Pieces initiative’s underlying assumption in that it does not seek to identify the fragments.
