The Contingent and the Predictive – A Response to Lawrence Liang

Contingency describes a state in which change exceeds our ability to systematize or narrate it. When Liang calls for its foregrounding in the archive it seems important to add that our experience of contingency in the archive is altered by software in a number of ways: Firstly, our production of video is increasingly shadowed by the production of metadata.¹Increasingly brands like Arri and BlackMagic make metadata part of a video camera’s menu. Production workflows rely on these ‘sidecar files’ to order the massive amount of footage. Secondly, in the distribution of video, what appears contingent is increasingly a function of predictive analytic software (those that use previous user patterns to predict future ones). In the presence of these protocols it falls to the autonomous archive to provide a space for a counter aesthetic to emerge. Below I want to extend Liang’s discussion of the contingent by looking at how it plays out in two recent archival films. Out of this comparison the contemporary archive emerges as a space to re-imagine the relationship between the collective and the algorithmic, recovering the contingent from the predictive.

In archives like Pad.ma clips that would be parsed by software-curated playlists emerge as equals, reminding us that the web more generally is not an egalitarian space for all content.²Pad.ma is a growing autonomous archive of video coming primarily from India. The video archive thus offers a space in which software-as-participant-observer can be forced to operate in the real-time of cinema and thereby be made more visible. When a file is uploaded to Pad.ma its place in the ecosystem is markedly different than it would be on YouTube. Such archives provide a space outside of what Parisi and Goodman have recently called “mnemonic control” or a move towards a “pre-emptive” ordering of memory, resulting from the intertwining of big data and subjective experience of time.³Steve Goodman and Luciana Parisi, “Mnemonic Control”, in Patricia Ticineto Clough and Craig Wilse (eds.), Beyond Biopolitics: Essays on the Governance of Life and Death, Duke University Press, 2011.

Narrating Overproduction

Video recording is increasingly accompanied by the application of metadata ‘in camera’ both manually and through automation.⁴Chri Chen, Marrying Realtime Metadata with Live Events to Automate Production for Multi-Screen, Talk given at the 69th Annual NAB Broadcast Engineering Conference, April 15, 2015. This is a symptom of a larger shift to make our gestures database-ready. As cameras begin to capture these two levels of meaning, it follows that those in front of the camera will increasingly view their own performances as data to be cleaned for real-time indexing. If, following Foucault, the archive defined what was perceptible, then the in-camera application of metadata may begin to define what is performable. Dourish observers as much when he says:

“The materialities of database technologies shape the kinds of ‘databasing’ that can be done and imagined. If databases make the world, they do so not as ineffable abstractions but in material ways.”⁵Paul Dourish, “No SQL: The Shifting Materialities of Database Technology”, Computational Culture, November 9, 2014.

Couple this with Liang’s description of the “ecstatic overproduction facilitated by the digital turn in filmmaking” and the normative power of the archive begins to project itself forward in time. Overproduction tacitly invites increasing automation in indexing in which, “objects are classified according to pre-programmed definitions and specifications.”⁶Jordan Crandall, “An Actor of the Street: Events, Agencies and Gatherings”, SubStance, 40(3), 2011, (Issue 126), p. 1. It seems the autonomous archive no longer simply has to position itself vis a vis official narratives (e.g. national archives), but the un-official algorithmic narration of overproduction.

The Archivist’s Lung

To clarify how the archive might preserve the contingent, we might go back to the significance that Liang, via Steedman, prescribes “to breath”. In the two vignettes of archivists that Liang describes, dust and lungs figure prominently:

“From the dust that Michelet breathed to the DDT fumes that killed Roja Muthiah, we are confronted with the question of what it is that we consider of value, what we discard as debris, and the residue of value.”

He goes on to quote Michelet himself, who, speaking of forgotten documents in the French National Archives, claimed: “[A]s I breathed in their dust, I saw them rise up.”⁷Jules Michelet, “Preface de l’Histoire de France”, in Carolyn Kay Steedman, Dust: The Archive and Cultural History, Rutgers University Press, 2002. Breath, I would add, is crucially the only bodily function that operates under both the voluntary and involuntary nervous systems. As bots begin to open YouTube channels we might say that the archiving of video is subject to similar forces. When Liang asks:

“Does an archival instinct of the contemporary have the same connotation as the maintenance of an archive in the traditional sense?”

The answer would seem to depend largely on who is doing the breathing?

The archival impulse as exhibited by Michelet and Muthiah, and theorized more generally by Benjamin in the figure of the collector, is subject to a kind of embodied automatism. It was through sublimation and indeed compulsion, that these figures reassigned value to the residual. To embody and thereby resolve an impasse, Michelet and Muthiah sacrificed their lungs. Such critical gestures are tempered in digital collections: Each time we tag a face in a photo we get fractionally closer to a day when we won’t have to. What was seen as pathological behaviour in Muthiah (hoarding of objects) is the modus operandi of machine learning. We ingest/inhale consciously now, but should we lose consciousness, so to speak, there is increasingly a system in place that will keep the archive breathing. Benjamin wrote “To [the collector] falls the Sisyphean task of divesting things of their commodity character by taking possession of them.”⁸Walter Benjamin, The Arcades Project, Harvard University Press, 2002, p. 9. For better or worse, big data reframes the Sisyphean, making it profitable. Autonomous archives can speak to this process but not by following suit. When Pad.ma writes that, “The direction of archiving will be outward, not inward,”⁹See: 10_Theses_on_the_Archive [accessed August 10, 2015]. it points to the archive as a site of distribution that crucially does not rely on predictive analytics. And with the focus on unfinished files, the archive avoids binaries of raw footage and final cuts, inviting a more social processing of images.

In the stories of Michelet and Muthiah, the body became a crucible for a basic incommensurability in society: the lack of popular historiography in the former, a dismissal of material culture in the later. By embodying an impasse between materiality and memory, they effected change. Thomas Csordas has commented,

“[T]he very possibility of reflexivity is grounded in embodiment. We can reflect on our experience because of the essential alterity that allows us to experience our own body as an object, an, ‘other’.”¹⁰Thomas J. Csordas, The Sacred Self: A Cultural Phenomenology of Charismatic Healing, University of California Press, 1997, p. 2.

In the non-predictive archive, the abstractions of metadata can begin to approach something like embodiment in the collective.¹¹For a discussion of “virtual embodiment” see Tom Beollstorff, “Virtuality: Placing the Virtual Body: Avatar, Chora, Cypherg”, in Frances E. Mascia-Lees (ed.), A Companion to the Anthropology of the Body and Embodiment, First Edition, Blackwell, 2011.

Archival Film as Symptom

For Liang, Los Angeles Plays Itself (2003) exemplifies a mode of archival filmmaking that plays off the inherent surplus in any image and, by extension, the contingent. Contrast this to The Clock (2010) by Christian Marclay and one begins to see the tenuous status of contingency in archival filmmaking. Los Angeles Plays Itself bears witness to a very human traversal of time. In it, LA’s history becomes a case study for the latent narratives that an archive might bear beyond the reach of metadata. The Clock in contrast has a kind of techno-determinism. It runs for 24 hours showing a succession of clips that more or less line up chronologically, a feat in and of itself, but very different in character: Marclay’s relationship to the archive seems mercenary compared to Andersen’s because he is concerned with the most banal kind of metadata: the timestamp. The surplus meaning of various clips is steamrolled by the premise of cinema-as-clock. Since no viewer can sit through the whole thing, what emerges is a mode of cinephilia with a death drive: The Clock is a film that could someday (soon) be made by software. While much has been made of the fact that the film was made by hand, it is certainly not outside the capabilities of video analytics to locate clock faces (analogue or digital) in video, thereby doing some, not all, of the material sourcing. It asks us to submit to a mode of viewing that is decidedly uncritical, estranging us from the source material as opposed to bringing us closer. Russell borrows Benjamin’s phrase “science out of magic”, to summarize the approach.¹²Catherine Russell, “Archival Cinephilia”, Framework: The Journal of Cinema and Media, 54(2), Fall 2013, pp. 248–9. The Clock captures a very contemporary ambivalence: we are entering an era in which it will be increasingly hard to discern if montage – sourced from database or archive – is human or machine-based. As CAMP has written,

“[Pad.ma], enquires about not only what is visible, but also about the backend in which machines or souls that propel or cast images and sounds in a particular way.”¹³See: http://studio.camp/event.php?id=218 [accessed 13 August, 2015]

The openness of such a space helps us better understand our evolving relationship to contingency and its others.

Recovering

While there has been widespread discussion of the fallibility of big data, the archive can perform a different mode of critique. Boellstorff has recently argued for breaking down the binary of raw and processed data by invoking Lévi-Strauss’s “culinary triangle” of raw-cooked-rotted.¹⁴Claude Lévi-Strauss, “The Culinary Triangle”, in Carole Counihan and Penny Van Esterik (eds.), Food and Culture: A reader, London, Routledge, 1997, pp. 28–35. The notion of “rotted data” avoids the chronology implied by the raw-cooked binary.¹⁵Tom Boellstorff, “Making Big Data”, in Theory, First Monday, 18(10–7) October 2013. Rotting, fermentation and distillation all exist between nature and culture and remind us of the materiality of their objects. Hence, if we apply Boellstorff’s thinking, the video archive is a space of ferment, accruing metadata through the social harnessing of automated processes, not vice versa. Watching how a video moves through an ecosystem like Pad.ma foregrounds the contingency of the prefix ‘meta’ itself. Metadata indeed, “valorizes some point of view and silences another.”¹⁶Geoffrey C. Bowker and Susan Leigh Star, Sorting Things Out: Classification and Its Consequences, Cambridge, MIT Press, 1999, p. 5. In big data analytics the focus on sample size and parallel processes often eclipse the situatedness of the inquiry, as though if the set is big enough, contingencies clean themselves. While this is how The Clock positions its viewer, we need to be able to imagine a different, interactive cinephilia in which we use sets to imagine, no just predict.

To this end, imagine an archive of the slow motion video shot in the year 2015. Slow motion is essential to cinephilia as it makes the contingent seem fated, the profane seem sacred. A decade ago the ability to shoot 240 frames a second was reserved for professionals with studio resources. Since September 2014, this capability has been conferred on anyone with an iPhone 6. As a result, the Internet is strewn with proud parents shooting baby’s first steps with technology that used to be reserved for Olympians. As the view counts reveals – slow motion has in the space of a year, been rendered banal. The archive of unwatched slow-motion asks us to reassess what for the first century of cinema was sublime.

Now imagine a fireworks display watched by 200 hundred iPhones recording every explosion in slow motion. On an isolated screen, life has never seemed more cinematic. However, compile every video from the 200 users and a dataset emerges that is purged of contingency: at 240/fps the ephemeral disappears. The resulting dataset could be used to create the exact same explosion (color, height, gunpowder) the following night. So long as this archive is accessed through atomized iCloud profiles it is socially mute, celebrating a cinema of one. It falls to the archive to recover contingency from such a set, lest a machine become the last cinephile.