[-empyre-] visualization as the new language of theory

Thomas LaMarre, Prof. thomas.lamarre at mcgill.ca
Thu Feb 4 06:50:40 EST 2010

One of the things that interested me about these models was the possibility to see the analytics as a kind of animation, or if you will, as a remediation of one medium by another.  This does present a challenge for how we think about empirical analysis.  Since the analytics stick close to the medium under analysis, the very status of the model seems to undergo a transformation.  As if this weren't a model but a conversion or a translation or remediation.

Yet I had some concerns about how we approach this empiricism, for I think, in a very positive view, that this is empiricism.  But it is hard to keep empiricism from turning into positivism without some questions about the process.

So, on the one hand, I truly appreciate these modes of visualization for allowing a kind of empirical analytics of movement and change.  They promise a great deal of sophistication and insight, especially for isolated films or collections of films. But on the other hand I am less certain about how this analytics operates at the level of decades and centuries.  For instance, if we are going to analyze a century of film in this manner, how do we decide what is a representative sample?  Given that so many films are not extant, will this affect the analytics?  With Japanese films, for instance, due to the firebombing of Tokyo, there aren't many prewar films today.  But what kind of analytics of a century of film could omit those films?  Also, would home movies be included, or is it only feature films?  But then, what of contexts where film wasn't primarily a matter of feature films?

I had similar doubts in the context of animation.  While I think it would be completely appropriate and useful to analyze 'slow' and 'fast' in this way, I wonder again how we would establish the parameters for the sample when doing comparisons or histories.  What are the material limits?

This does have implications for history.  I am totally in agreement that 'epochal thinking' - based on simple historical periodization - is a dead end.  But there are a lot of ways of thinking history that are not based on epochal thinking and periodization.  Foucault himself immediately retracted the term 'episteme' and gradually showed that there are different kinds of transformation underway at any one moment; he rejected that idea that disciplinary society follows sovereign society, etc.  Raymond Williams spoke of dominants, and Althusser of overdetermination to acknowledge the heterogenity of any historical formation.  These are very different ways of thinking history, none of them dependent on epochal thinking.

The idea of tracing gradual change historically I find appealing, but still I have to wonder about the boundaries or limits of the analytics when we're talking about decades or histories.  Is this a history of gradual change across the world?  Or is this another way of doing history of the West or of nations?

In Anti-Oedipus, Deleuze and Guattari present a model for thinking about continuity and process by way of three syntheses - connective, disjunctive, and conjunctive.  Maybe this is one way of taking into account the kinds of gradual continuous change that are now possible to measure with these visual analytics while acknowledging processes of disjunction, as well as another register of conjunction as well.

Just some thoughts.

On 03/02/10 12:01 PM, "Lev Manovich" <manovich.lev at gmail.com> wrote:

sorry - hit send accidentally before completing my post - here is the
right version:

There are many ways in which cultural analytics differ from what
people call "formalist" approach, but given the constraints of the
post size, let me just point out a couple here.

1. We are now able to visualize - and therefore better understand -
gradual changes over time at a number of scales - from a few minute
video of a gameplay to a century of film history. Google Earth allows
you to navigate space across scales - from the view of the Earth as a
whole to a Street View that puts you in a position of a car driver or
a passerby looking from a street level. In the same way, we should be
able to navigate through time, moving from the scale of a single
cultural artifact or its parts (such as a film shot) to the scale of
decades and centuries.

Visualization of gradual changes in visual and media culture over
longer historical periods is an idea that appears to us particularly
timely. Humanities disciplines, critics, museums, and other cultural
institutions usually present culture in terms of self-contained
cultural periods. Similarly, the most influential contemporary
theories of history by Kuhn ("scientific paradigms") and Foucault
("epistemes") also focus on stable periods rather than transitions
between them. In fact, relatively little intellectual energy has been
spent on thinking about how cultural change happens. Perhaps this was
appropriate given that, until recently, cultural changes of all kinds
were usually very slow. However, since the emergence of globalization
in, not only have these changes accelerated worldwide, but the
emphasis on continual change rather than on stability has became the
key to global business and institutional thinking expressed in the
popularity of terms such as "innovation" and "disruptive change." It
is time, therefore, for us to start treating "change" as a basic unit
of cultural analysis - rather than limiting ourselves to discrete
categories such as to "period," "school" and "work."

Here are couple of examples:


Thus, if we for instance take the hypothesis that in contemporary
anime characters move mush less than in the earlier works, not only we
can actually test this hypothesis to see if it true by quantifying and
measuring movement but we can also see how this parameter changed over
the years across dozens of works.

2. Another crucial advantage of using data analysis and visualization
is that now for the first time we can adequately describe many aspects
of art and media which previously we could only talk in a very vague
way. For instance, for a 100 years filmmakers, animators, critics and
theorists talked about movement in films and cartoons. But the natural
languages only give a few categories to describe movement - "slow,"
"fast" and a few others. In other words, natural language map
continuous qualities (such as movement) into few discrete categories.
Data analysis and visualization give a much better language for
describe such continuous qualities.

Here are two examples. The first visualizes movement patterns across
Vertov film annotated using linguistic categories:


The second visualization uses measurements of movement - and reveals
all kinds of amazing patterns in the film which were hidden when we
use natural language to annotate movement:

empyre forum
empyre at lists.cofa.unsw.edu.au

More information about the empyre mailing list