An Interpretive Framework for Contemporary Database Practice in the Arts

An Interpretive Framework for Contemporary Database Practice in the Arts

(Original, 2004), first presented at the College Art Association 94th annual conference, Boston MA.
Panel - From Database and Place to Bio-Tech and Bots: Relationality versus Autonomy in Media Art
Thursday, February 23, 12:30 PM, 2:00 PM
Hynes Convention Center, Third Level, Room 304
Chair: Marisa S. Olsen, University of California, Berkeley

Brett Stalbaum


There are two common notions regarding the nature (or ontology) of data and information that are important for us to think about when we are considering artistic practice with database. The first is the notion that information is disembodied from its subject, and the second is somewhat of a conflation of the terms "data" and "information". Political concern stemming from the first notion may be most responsible for stimulating "database art", but current art practice with database can be broadly divided into three generally recognizable, though not mutually exclusive modes of practice: database politics, data visualization (the latter related also to sonification, and haptics), and what I will term database formalism. The second notion represents more of a noise in our at-large cultural understanding regarding the meaning of the terms "data" and "information" that when clarified, may sharpen the critical focus on an aspect of data visualization practice. Honing these two notions will provide us with a critical basis for the interpretation contemporary database art practices, perhaps especially as they interact with emerging geospatial and location aware media practices. In this writing, interpretation is distinguished from definition and evaluation, as it is in the tradition of analytic aesthetics. I write from the perspective of a practicing artist; not a trained philosopher or art historian. Thus I demur, at least somewhat, on the issue of defining database practice (beyond the obvious), and I avoid any qualitative evaluation of the examples I give. I only hope to chart the terrain of a contemporary practice with which I am familiar, including the work of many colleagues and collaborators. I hope to form an interpretation of the approaches contemporary artists are taking to database that I hope will be useful in evaluating this territory.

Data Body and Data Politics

I will start by considering works that emphasize the contemporary consequences of disembodiment of data/information from its referent, regardless of whether we are speaking about the human body and its disembodied 'data body', or other material manifestations of reality and the data which refers to it. "Information" and "data", in this narrow context, are viewed as descriptions of the thing described, and are somewhat conflated terms. (See next section.) Christiane Paul patently describes the issues that seem to have been in play for artists surrounding the issue of disembodiment:

"In the digital age, the concept of 'disembodiment' does not only apply to our physical body but also to notions of the object and materiality in general. Information itself to a large extent seems to have lost its 'body', becoming an abstract 'quality' that can make a fluid transition between different states of materiality. While the ultimate 'substance' of information remains arguable, it is safe to say that data are not necessarily attached to a specific form of manifestation. Information and data sets are intrinsically virtual, that is, they exist as processes that are not necessarily visible or graspable, such as the transferal or transmission of data via networks."(174)

I will argue that the case is subtly yet importantly different, as this type of disembodiment is not actually a new phenomenon to the digital age. Information/data have always been disembodied, and in fact we do see that the interaction between the virtual and the real is more tightly bound today, and indeed is more materially generative (yet contra-abstract), than at anytime in history. Disembodiment is not the difference making difference that the digital age brings. In order to demonstrate this, I will take a double tact. First I will look into history for precedents of disembodied data and information, hoping to show that "disembodiment" is not a new issue just because we have entered a digital era. Then I will try to show that it is not the disembodiment of the referrer from the referent that creates the radical difference that the digital era has brought, but rather that it is the nature of distributed, high speed data processing that makes all the difference because it radically motorizes, automates and makes ubiquitous the potential for data and information to impinge on daily life. After presenting this idea, I will make reference to a few database artworks that I think map to the various assumptions outlined by Paul, which I think expresses an interpretive critical model in which artistic practice can be specified in terms of 'database politics'.

It only requires a few examples from history to dispel the notion that disembodiment is a novelty specific to the digital era. Edwin Hutchins, in his study of how representations are propagated in systems of cultural computation, points out that the use of bearing logs in sea navigation dates back at least 4500 years, and that "Sumerian accountants developed similar layouts for recording agricultural transactions as early as 2650 B.C." (124) Cuneiform Tablets, a clay tablet inscribed with ideograms and numerals (multipliers), organized in the now familiar column and row format, formed the material basis for the disembodiment of material reality into a clay media for data storage of mundane business transactions. And certainly, the notation on a tablet of "18 unproductive trees" is no more the actual 18 unproductive trees than some contemporary individual's poor credit history (a common example of a 'data body') constitutes the breath of individual personhood. Yet, both such representations are similarly disembodied data representations utilized for economic control and management. In a loose sense cuneiform tablets were the first spread sheets, and one could go further to argue that the first written words and images instantiate a similar disembodiment of referent and referrer, not to mention the disembodiment inherent in language itself! This has been a constant issue in aesthetics from Plato (mimesis) through semiotics (sign as combination of signifier/signified), and in postmodern thought; perhaps most notoriously in the writings of Jean Baudrillard where the sign becomes ascendant and begins itself to relplace reality through precession.

Similarly, data has for a long time exhibited the quality of being fluidly transferable between forms of materiality in different representational media, and in fact transferal and transmission of data via pre-industrial 'networks' show that data transferal is in no way a novel phenomenon or a creation specifically of the digital age. Hutchins gives the chip log and the methods of using it as just one example of the propagation and transmission of representational states through various media. The chip log is device consisting of a reel, a rope line, and the "chip"; a piece of wood that would be thrown overboard to remain stationary in the water while knotted line was let out. The passage of time would be marked by crew members singing a hymn (maintaining the system's clock speed), and notations regarding the number of knots unrolled would be recorded in a log at a regular interval. The knots would measure the distance that the ship had traveled, from which the term "knots" as a measurement unit for maritime speed is derived. Importantly, Hutchins shows how the chip log was utilized to perform an analog to digital conversion:

"The log gave rise to a computational process that begins with analog-to-digital conversion, which is followed by digital computation, then either digital-to-analog conversion for interpretation or digital-to-analog conversion followed by analog computation." (103)

Through these conversions, the propagation of representations between various crew members aboard ship was enabled. Chip logs were utilized as dead reckoning instrumentation allowing the projection of the ship's future position on nautical charts; nautical charts which are themselves analog computers designed expressly for position-fixing calculations. Logs and analog-to-digital conversions allowed data to be transported, often in digital form, through a ship wide network of crew members utilizing different media to perform their tasks; for example from the memory of the log keeper into the log, then from the log to navigator who would project the future position of the ship onto a chart at some fixed interval, and then from the media of the chart to the mind of the captain who is responsible for the larger journey. The actual position of the ship in the material world was at stake.

Data and information have qualities of their own, as calculable symbolic representations capturing measurable aspects of material systems. Data and information are not only disembodied in some material form of representational abstraction from their subject (whether clay tablet or digital electric impulses), but can be recorded and transferred from one state to another, propagated from person-to-person in local, perhaps totally linguistic, networks of social computation, or from place-to-place via encoding into media mobilized by material transportation consisting of technology such as sailing ships, or more recently, undersea fiber optic cables. Importantly, this mobile property of data and information has been at play in human culture long before the digital era - perhaps as long as linguistic messages have been carried from place to place by foot and shared among different groups, and certainly since written (doubly coded) and numeric representations began to be transported. Additionally, the example of cuneiform as a particular clay media implementing informational disembodiment from the material world emerged well before the development of the algebraic analysis (as early as 1800 B.C.) and the discrete mathematics concepts (congealing nicely in the figure of George Boole in the 19th century), that would serve as the catalysts for the development of digital communications and computational technologies during the 20th century. The disembodiment of data and information from the real clearly predates the digital era.

Disembodiment does not mean that data and information, and the material reality they derive from, do not influence one another. In fact the case is rather the opposite, forming is the basis of the fundamentally materialist-formalist analysis I am trying to forge here. As I have indicated in the past:

"This position is supported by Paul Virilio’s theory of information as the third dimension of matter, (energy being the second), in that information and its effect on identity are not disembodied from the real, but rather become a integral part of the real world projecting directly into the body: a network of people hyperactivated by information machinery which has joined with the body no more or less conspicuously than the pacemaker or the telephone handset." (1998)

The significant difference making difference that does arrive with the digital era is the speed with which the relations between information technology and material systems are implemented: the move from the speed of hand inscribed clay tables, to ships, to trains, to telegraph, to the speed of light on fiber optic and radio networks. (This trajectory roughly paraphrases Virilio's analytic project.) The process has been a teleological one; the move from writing data on clay storage devices and the associated literacy to retrieve and utilize those notations in a local economy has progressed to 'writing' data in informatic media such CPU's, RAM, magnetic storage, optical and wireless networks, and of course this too assumes an associated literacy, in the contemporary case one required to utilize digital media in a global economy. As the transmission speed of the media becomes faster, the ability of data and information to impinge upon or embed itself in material systems expands. While clay-based inscription systems improved the management of a local orchards in Sumeria, information systems today, which wrap the Earth in fiber optic cable and paint it with electromagnetic carrier waves, facilitate the transmission of data and information around the world in milliseconds, allowing a global scope of impact for data and information. For example, as Geri Wittig points out regarding the relationship between geographic information systems and the Earth as a complex system:

"With the increasing use of GIS technologies in a wide variety of fields, including art, the data networks generated will disseminate into the expanding networks of information technology. I speculate these GIS generated data networks have the potential to act as bifurcations and coadaptive systems..." (2003)

This means that systems which operate, transport and calculate at the speed of light have greater power become co-operative in the distribution and creation of the real, causing the disembodiment of data itself to bifurcate into something more powerful and integrated with life on Earth due to the speed and intensity of data flows. This allows data and information to play a more immediate, acute, synchronized role in the daily life of persons, as well as non-human ecosystems and flows of materials. It is not disembodiment per se, but rather machinic catalysis of the relations between virtual and real that is the difference making difference in the digital era. Further it is the discrete properties of the digital that enable this speed, as well as enabling the exact quantification of information, ala Claude Shannon. It is the catalytic properties inherent in the material basis of digital technology that allows the analysis of the difference (that information is) to have a radical transformational impact on every aspect of culture, society, biota, climate, and to some degree, even geology. The disembodiment of information from its referent, which is an archaic and fundamentally ontological aspect of data and information, is now hyper-activated in real time at the speed of light. And indeed, it is the consequences of this speed which many artists working around the issues of 'database politics' have responded to.

A small but representative selection of artists who have notably responded to the sudden imposition of database as a mediator of power and social control include the Critical Art Ensemble, Natalie Jeremijenko, Graham Harwood, and Diane Ludin. The Critical Art Ensemble were perhaps the first artists to see the looming threat of database on matters of privacy and power, and to present issues relating to database theoretically in terms of an agent of social control. In their 1994 book The Electronic Disturbance, CAE states:

"As the electronic information-cores overflow with files of electronic people (those transformed into credit histories, consumer types, patterns and tendencies, etc.), electronic research, electronic money, and other forms of information power, the nomad is free to wander the electronic net, able to cross national boundaries with minimal resistance from national bureaucracies. The privileged realm of electronic space controls the physical logistics of manufacture, since the release of raw materials requires electronic consent and direction." (CAE, 1994)

While we do read here a direct reference to the concerns of disembodiment in terms of "electronic people", we also see a clear focus on new forms of pan-capitalist power and control over the economy through processes where "electronic space controls the physical logistics of manufacture." This inference on the part of CAE certainly maps to the notion of data and information as disembodied control systems of management, but disembodiment is placed in a context that makes the change less attributable to the original sin of disembodiment than it is to the speed and ease through which social power and control over the material world is deployed via contemporary, digital, highly distributed database systems. CAE's words may be the first shots fired in the art of database politics.

Natalie Jeremijenko's and Graham Harwood's recent work with database share a consistent theme: an attempt to address the asymmetry of power between those who model and manipulate the world through data, (thus enjoying most of the rights to benefit from information garnered from that data), and those who are modeled and manipulated by data. A representative example of Jeremijenko's recent work is the Bit Antiterror Line project, which allows "every phone [home/cell/booth] to act as a networked microphone... For collecting live audio data on civil liberty infringements and other anti-terror events." The files are made available in a simple database of audio files on the bit antiterror line web site (Jeremijenko), one of which recounts the story of a stewardess who threatened a couple with arrest by armed Air Marshal if they continued to draw silly pictures and laugh at her. Harwood's 9 project is a website modeled around the simple square shaped layout of 9 media elements. It allows people to represent themselves, their neighborhoods, their identity, and their interests, via media elements arranged in this simple, easy to use layout strategy, including a notion of proximity and thus juxtaposition with neighboring 9's. The ease of use at the interface level belies a sophisticated custom database under the covers, coded by the artist. 9 encourages not only self representation, but the exploration of the self representations of others in a shared data commons creating connections between/within communities defined both geographically and informatically, while Jeremijenko's project creates a data commons as both an emergency antidote to, and cultural and social analysis of, the growing fascism apparent in the United States as the "War on Terrorism" progresses. As I write this (original draft, April 2004), CAE's Steve Kurtz is being investigated by a grand jury in Buffalo, NY, essentially for daring to make provocative art works with regulated biological materials. Although he (and CAE) have presented this work publicly in high profile art institutions for many years, his research and materials stored in his home became the subject of a wasteful and misguided anti-terror investigation after being noticed and reported by first-responders following the tragic death of Hope Kurtz from natural causes.

The prevalence of database in biotechnology research has led to many projects dealing with genomic data analysis or critique of the systems in which nature becomes private property. Diane Ludin's "i-BPE, i-Biology Patent Engine" takes on issues of intellectual property and ownership in the high-tech era by setting up a context where real United States patents on genes are themselves claimed as a kind of public property/context for remixing and play with the language of patents, resulting in a "aggressive take-over by i-BPE agents... i-BPE gene patents will return bio-rights to non-governmental, cultural agents for revision." (Ludin) In a presently unpublished manuscript, Ludin discusses, somewhat ironically, how speed has (with its own certain irony), allowed the disembodiment of data from its referent to return directly and literally to the site of our bodies, for which the only prior art is billions of years of evolution. "With the rise of ibiology the circuit between code and patent becomes part of the super speed ecology of Bio Capitalism. Ibiology establishes the next level of command and control culture where artificial selection becomes a post-human, globalizing, gene profit system." (Ludin) In Ludin's, and indeed all of the above examples, speed is the difference making difference that the art of database politics ultimately must address across a range of practice; regardless of whether the artist is using database as media to help along the emergence of shared understanding within a culturally mixed global culture, or responding defensively (with database) to the onslaught of database driven assaults on civil rights committed by corporatist or fascist governments.

Data Visualization, Beautiful Information and Sublime Data

A formal aspect of data and information that is often overlooked in culture at large is that the terms "data" and "information" have meanings that are quite different from one another. Although Dictionaries such as Webster's accurately define the terms; information as "an informing or being informed; esp., a telling or being told of something", and data as in "facts or figures to be processed; evidence, records, statistics, etc. from which conclusions can be inferred; information", (Webster's, italics mine), popular uses of the terms often overlap somewhat more than their dictionary definitions allow. Note that "information" is above embedded in the definition of data, across the semi-colon boundary behind which "conclusions can be inferred", but without a cadence or emphasis that would mark information's definitional difference with the same clarity as it is most commonly defined in computer science. Information as described above could easily be misread as synonymous with "facts or figures to be processed", even given position of the semi-colon. As I will discuss in the next paragraph, there is in fact an issue of transitory states. Nevertheless, information is most usefully defined as the conclusions or news of significant difference that is inferred from the logical processing of a collection of data. Data is defined essentially as being raw facts; whereas information is mined from processing those facts.

Of course, the situation it is not that simple. At any one time the same representations (I do not take "representation" to mean exclusively "visual"), might exist in different terminal states (as either data or information) on a larger conveyor belt of ubiquitous digital processing. A simple example: it is common for the output of one program (nominally "information") to be the input data for another, as in the unix command, ps -ef | grep brett, which pipes the somewhat lengthy output of the ps program (information about all processes) to the grep filter such that I might know only of my processes. In other words, information can become data to be filtered into more specific information. Another potential breakdown in the distinction occurs due to the graphical user interface, which does a better job of 'making invisible' the user's control data (another kind of input), for example in the form of pointing as interactive input (mouse clicks, mouse drags, etc.) These are definitely forms of control data input, but they are processed more invisibly than control commands given on a command line interface, because the visual half life of clicks and drags as pixel residue on the screen is not buffered as are commands that remain visible in the terminal shell (visible on screen) after being issued in a CLI. Nevertheless, ignoring interactive input and its own important implications, it is still true that data plays its most common social 'role' in the form of input to programs, and it is information that is derived from processing data as output; even if the information is later transitioned by being reprocessed as input back into some other program (potentially somewhere else in the world). The ontology of data and information as input and output is contextually mediated and transitory; existing alternatively between states of data and information. Yet data is still associated in an important way with input and information with output, even if the terms data and information are treated more loosely in culture at large, perhaps due to being seen adjacent to each other so often, a result of their status as quite inseparable siblings or perhaps a digital yin/yang.

A good question for the impatient reader at this point would be "What does this have to do with contemporary database practice in art?" After all, there is no shortage of clarification regarding the distinction between "data" and "information" in engineering and the sciences. The answer is that the conflation of terms seems to pool especially commonly in the humanities discipline areas, such as art. To be fair, it is a common linguistic conflation in culture at large and this is indeed where artists operate, but I do think it merits our attention in any analysis of the works of artists who are working with database, and particularly for artists that are working specifically with data visualization, or the related disciplines of data sonification and data haptics (as in ambient computing).

Lev Manovich has made a very important observation about the aesthetic strategies of Data Visualization practice in an essay titled The Anti-Sublime Ideal in Data Art, (2002), in which he critiques contemporary data visualization practice in art as adhering to a pursuit of beauty in the transformation (or processing) of large datasets into the visual field: the "Anti-Sublime" aesthetic. Beauty is the pursuit of clarity, balance and transparent form, and data visualization is often pursued for the sake of understanding or making clear the behavior of data and the systems represented by data. Beauty in data visualization is opposed to the sublime: the condition under which the data overwhelms its viewer, and the viewer's senses are mobilized in a special kind cognition that allows them to carry on with the formation of an understanding that is, as it turns out, more likely to be satisfactory than a random guess. There are many names for this kind of cognition: intuition, anticipation, instinct, or a sixth sense. The sublime is of considerable interest to the artificial intelligence discipline in computer science. Human intelligence seems able to deal with the sublime condition and can continue to operate intelligently even when overwhelmed or subjected to context shifts, while discrete computational machines have not yet proven this ability. In a sense, the holy grail of artificial intelligence is to create machines that can behave with human-like intelligence when similarly thrown by excessive amounts of data under variable context.

Interestingly, the definitions of the terms "beauty" and "sublime" have also been culturally conflated, perhaps even more so, than the terms "information" and "data". Just as information and data are sometimes interchangeable terms in common usage, (often taken to mean information), the meanings of beauty and sublime are today similarly conflated, (often to mean beauty). The notion of beauty, revealing form and making cognizable, as the goal of data visualization art works dealing with large data sets is clearly described by Christiane Paul, writing of Benjamin Fry's 1999 work "Valence":

"The software visually represents individual pieces of information according to their interactions with each other. Valence can be used for visualizing almost anything, from the contents of a book to website traffic, or for comparing different data sources. The resulting visualization changes over time as it responds to new data. Instead of providing statistical information ... Valence provides a feel for general trends and anomalies in the data by presenting a qualitative slice of the information's structure. Valence functions as an aesthetic 'context provider', setting up relationships between data elements that might not be immediately obvious, and that exist beneath the surface of what we usually perceive." (177, 178)

I do not choose to wade into any aesthetic debate regarding the beautiful and the sublime in data visualization; I am sticking to my promise to hold fast to an interpretive framework in this writing. Lisa Jevbratt has written an essay titled The Prospect of the Sublime in Data Visualizations, responding in part to Manovich's use of the 1:1 project (1999, 2002) as an example of the anti-sublime aesthetic. (Jevbratt) For now, I merely want to point out that in terms of how we interpret the art practices engaged in data visualization, beauty as opposed to the sublime is the most critical contemporary interpretive framework in which such art may be evaluated aesthetically. The criterion for analysis shifts from the effectiveness of any particular visualization (and its ability to facilitate an understanding of the data through beauty), to the roll of the user or communities of users in interpreting a visualization via their own ontological thrownness, their own conceptual, computational or cultural methods for processing data, and their own ability to perceive when facing conditions of sublimity. At its extremes, the sublime analysis suggests that access to raw, unmediated data replace visualizations, and that communities should take democratic control of their own data interpretation in a way that best balances their exposure to quantities of data against their need to reduce it to useful information; all of which might only become practical if formal languages for processing data become standard educational assumptions for a baseline notion of what it means to be literate in post-industrial, high tech societies. Microsoft Excel(TM) can not save us. Artists might be able to play an important role in this regard: as guides in data exploration more so than as experts in data visualization.

Additionally, the formal definitions of data and information imply another framework tightly coupled to the issues raised by the beautiful and the sublime. Data visualization practice is certainly bound to the transition of representations between states of being data and states of being information; and as Manovich points out, most contemporary artists working in data visualization are seemingly committed to visualization as information. This is essentially congruent with Paul's discussion of Fry's work Valence as well as her overall discussion of database practice; further implying that much data visualization practice in the arts today seemingly pursues beauty. Interpretively, we may extract from all of this that the pursuit of information is the pursuit of the beautiful and that the pursuit of data is the pursuit of the sublime. The former implies a struggle for understanding, the later an impulse for exploration, including the collection and generation of new data. How artists implement their forms of expression between information and data, and possibly in the transitory states between them, is an aesthetic issue that maps to the transitory states between the sublime and the beautiful. Speaking personally, this seems to be an unresolved area in data visualization as artistic practice, as well as in the related formal practice that I discuss in the next section.

Virtual and Materialist Data Formalism, Data Mining

In this section, my interpretive framework comes full circle back near the issue of disembodiment. In the first section of this essay, I believe that I was able to demonstrate that data and information have always been disembodied from their referent, and I did so by arguing from a materialist stance that views data as an important virtual reality that actually impinges on material reality. In a previous text titled Database Logic(s) and Landscape Art (original, 2002), I presented a more radical, though consciously very speculative and provisional view that data is embedded and operative within the actual through a process in which humans/data/Earth are inextricably implicated: humans mediate the landscape with the assistance of data about the landscape, and the data itself mediates that mediation, not necessarily intentionally, but in such a way that the actual material Earth now speaks through scientific data, thereby expressing a voice in conversation with human culture. In the same essay, I indicate how the term 'virtual' is also often misunderstood as referring to the imaginary interfacial illusions that computational systems can create, rather than (more appropriately) the abstract mathematics of reality (that can be modeled computationally, well beyond 3 dimensions), that in some sense produces the actual. In other words, the virtual is itself a real space of possible physical states for any system that crystallize into the actual, which is precisely what allows computational models of physical systems (such as engineering or atmospheric simulations) to have predictive power. I restated this case (it belongs to and is Manuel DeLanda's lifetime critical project) in order to suggest that artists should utilize the notion of the virtual for predictive or analytical practices that reveal knowledge about the world, or better, that emerge new behavior, exploration and experience. I think this holds for the humanities. I am in no way concerned if what is revealed functions as conceptual and performance art, and not as science.

There are many database art projects that demonstrate this analytical and productive practice which engage with data utilizing an ethos that maintains an interest in the embodiment (contra disembodiment) that is implied in the relationship between data and its material, actual, real world referents. Although I have avoided definition, I would argue that the preceding does constitute something close to a definition of database art in the bigger picture, the relationship to materialist embodiment being the key. In any case, it clearly fits into my interpretive framework for contemporary database practice as database formalism. These projects are interested in the actual materials that are modeled by data, and seek new, exploratory methods of interacting with the material world that reveal new knowledge about the materials, or our interactions with them, and that allow data to become a cooperative co-participant in the performance. For example, Lev Manovich's Soft Cinema (2001-) uses metadata to dynamically organize a Mondrian inspired screen layout for videos shot by the artist in his travels, in which "every clip is assigned 10 different parameters, which are both semantic and formal, so for example one is geographical location... how much motion there is in a clip, which is assigned a number... the contrast, the average brightness, the subject matter...", and so forth. (Manovich, 2003) The parameters are utilized by custom software to control the editing of the video clips and their organization in the layout, allowing data about the (video) data (the metadata) to manifest itself through being granted some level of decision making authority and authorship. Manovich's cinema edits itself; revealing itself in unexpected and often poetic ways that require one to apply a thrown and sublime mode of paradigmatic viewership to its interpretation.

David Rokeby's Giver of Names (1990-) and George Legrady's Pockets Full of Memories (2001) both ask users to interact with real objects in the gallery space, which are scanned and input into a database system for further classification and comparison. While Rokeby's approach utilizes an AI computer vision technique and artificial language processing, and Legrady's uses a clustering algorithm designed to situate the personal objects offered up by the audience with their statistically nearest neighbors, both projects are literally concerned with the relation between real objects and how they are thus mediated (either by naming them or associating them with another) as they undergo analog-to-digital (material to reference) conversion, insertion into a database, and subsequent data analysis. Importantly, an emphasis on the materiality of the objects is maintained in the exhibition space. The materiality is directly experienced by the audiences who interact with Rokeby's collection of objects lying around the exhibition space that they may situate on a pedestal for scanning and interpretation by an artificial intelligence system. In Legrady's case, a personal object if offered up for analysis. Both systems connect rather literally with the real as an embodied space to be contextualized.

The near unification of referrer and referent is even more literal in recent C5 work, (a group of which I am a member), where geographic information system data (a digital 3D map of the landscape) is mined through the preprocessing of the primary data into a layer of metadata characterizing large areas of topography (currently the State of California), that can be searched via a relational database and related Java API. (The C5 Landscape Database API.) Mirroring the Input/Processing/Output pattern common in classic, non-interactive data processing, C5 takes input samples (collected with GPS), and processes them to identify the most similar landscapes to the original, but that exist somewhere else. As preparatory work for The Other Path (2004-) Geri Wittig set out on a month long trek along the Great Wall of China, starting in the northwest desert and following the Wall eastward to where it runs to the edge of the Yellow Sea. GPS data was collected from twelve separate trekked locations along the length of the Great Wall. Using pattern-matching search procedures developed at C5 (Amul Goswamy and Brett Stalbaum), the 12 most similar corresponding terrains in California were identified. After determining the blocks representing the most similar matching terrains in California, phase two of the Other Path search process identified discrete paths within those terrains expressing similar statistical characteristics, such as simple distance, cumulative distance, and elevation change. To do this, a swarm of virtual hikers, implemented as experimental features of the C5 Landscape Database API 2.0, were unleashed in the virtual California landscape to explore and generate tracklogs, which were then compared to Wittig's original "input" Great Wall of China tracklogs. The results of this search identified the most closely matching virtual tracklogs, which were then exported to tracklog files, uploaded to GPS devices, and physically realized by C5 in a performance of tertiary exploration (after the original, after database), of what is now known as The Great Wall of California. In this performance, walking works in the tradition of Richard Long, Hamish Fulton and perhaps even Dominique Mazeaud are reconceived as input, processed by via database applications that have been granted the ability to tell us where to go by outputting GPS coordinates that we are conceptually bound to follow with our feet. This generates alternative experience and exploration of the landscape at a time when everything (on the landform surface of the Earth) has already been explored and modeled. It emphasizes not the disembodiment of datascapes from their referents, but their intimate connection and productive capability.


I have outlined three modes of practice, database politics, data visualization, and database formalism (the latter contra disembodiment) in which contemporary database practice can be interpreted. The later formalist tendency, in which database is conceived as virtual context for implementing a data co-operative mediation of the world, perhaps most interestingly overlaps in the final analysis with database politics. Though largely apolitical at first glance, the formalist interpretative mode of database art practice is similar to that of database politics in that the goal of both is to realign the power of database to distribute the real, albeit for different reasons, as opposed to data visualization's dominant (but perhaps not universal) desire to better understand data. Though formalist practice may not self-consciously attempt to intercede in pan-capitalist distribution of power, data formalism and artistic data mining practices do conceive of agency returning back to the hands (or for C5 the feet) of the people who interact with such systems, although perhaps in a perverse way by tactically ceding a certain level of arbitrary control to the database applications themselves. But as long these are at least neutral with regards to power, ideally open source, and hopefully designed and performed by autonomous users of the systems in non-coercive ways, there are advantages to be found - perhaps even political ones.

For one, formalist database practice is in alignment conceptually with the ubiquity of database in our culture, perhaps encouraging individuals to develop related expertise for apolitical ends (recreation, hobbies) that produce ecologies of knowledge that become useful when political conditions become too onerous for the majority of people. Formalist practice could be aware that discovering the possibilities and building novel alternatives (especially when done so by communities instead of for them), might be just as effective as directly resisting the distributed, nomadic power of systems of mass subjugation. Also, database formalism allows aesthetic analysis to move toward and explore truly interesting, purely formal issues of database itself as a medium. For example, the relational database model trades maximum processing efficiency for the ability to maintain ad hoc queries, which may be consequential in terms of how the material world is ultimately mediated in particular instances. All three of these conceptual modes of artistic practice with database are important of course, and they certainly overlap in practice. None is mutually exclusive.

Interpretively, there is perhaps a fourth mode of practice that it may be argued that I have ignored. The only other mode of database practice that is perhaps not necessarily some derivation founded in database politics, data visualization, or a database formalist practice is seemingly a multimedia practice that assembles and processes a 'database' of multimedia materials, mixing or remixing them into some other media forms such as web video, animation, real time video processing, music, etc. The multimedia assumption insists that the core of digital media art practice is manifest as pixels on a screen, or some other output such as speakers, or as interaction at an interface that produces some kind of visceral or otherwise magically mediated experience. The mediation is viewed as ultimately flowing from the identity of "the artist" of course, who is assumed to produce some kind of political awareness or aesthetic/cultural experience in the minds of the audience. Often, this kind of very traditional orientation toward art practice does not consider the elements in the database as being datum with their own ontology, and suppresses data's identity into being mere media elements or samples to be processed, remixed, and assembled by the artist in an expressive configuration of individual artistic style and message. Media tools such as digital video editing and multimedia authoring platforms are commonly employed, and often these are used pretty much the way that their designers (large corporations) intended them to be used. There is no reason to think that software can not be used by artists in other ways (in fact, there are many delightful examples on, but in practice such conceptual repurposings of code are all too rare. When such repurposings do happen, they transcend multimedia and map to conceptual art practices (often termed "software art"), and I suspect that my categorical distinctions regarding database practices would support these. But I am veering dangerously toward making an evaluation of multimedia practices here. That is not my goal, so this is a good place to conclude.


  1. Critical Art Ensemble, The Electronic Disturbance, Autonomedia, New York.
  2. Jeremijenko, Natalie, Homepage for the bit antiterror line project, accessed April 25th, 2K4.
  3. Jevbratt, Lisa, The Prospect of the Sublime in Data Visualizations, YLEM Journal, Artists using Science and Technology, Volume 24, Number 8, August 2K4.
  4. Ludin, Diane, i-BPE project website, accessed June 6th, 2K4.
  5. Ludin, Diane, Deep Harmonization i-BPE, unpublished manuscript, 2K4.
  6. Manovich, Lev The Anti-Sublime Ideal in Data Art, (2002)
  7. Manovich, Lev, Lev Manovich / Interview at DEAF 2003, quoted from a video
  8. interview, selection transcribed by myself.
  9. Paul, Christiane, Digital Art, (c) 2003, Thames and Hudson Ltd, London, ISBN 0-500-20376-9
  10. Stalbaum, Brett, Aesthetic Conditions in Art on the Network: beyond representation to the relative speeds of hypertextual and conceptual implementations, Switch, the new media journal of the CADRE digital laboratory, 1998,
  11. Stalbaum, Brett, Database Logic(s) and Landscape Art, Noemalab - t ecnologie & societa, 2002-3,
  12. Webster's New World Dictionary and Thesuarus, Accent Software International, Macmillian Publishers, Version 2.0 - 1998, Build #25

This essay is dedicated to the memory of Eric Gray, who is responsible more than any other for helping me establish my interest in computing as a young person. In 1981, Eric showed me a war dialer he had written in BASIC on a TRS-80 computer, along with custom hardware enabling his tape drive remote control output to perform pulse dialing on the plain old phone network, which he was using (while his parents were away, of course) to war dial for local modem connections to hack into. I was hooked. And the hours of playing "Adventure" did not hurt either. On behalf of your family and friends, we love and miss you Eric.

Also, thanks to Warren Sack. I wrote this after presenting and hanging out with him in Karlsrue in January 2004, talking about these kinds of things, and it is really very cool that we both ended up presenting on Marisa's panel together. Tad and Helen too:-)


site | contact