Tuesday, August 22, 2006

Data Art - Some Questions

I'm working up a paper on data aesthetics and creative practice, looking especially at visualisation (a kind of companion to "Hearing Pure Data," a paper written a while ago focusing on sonification / audification). At this stage all I have is a collection of questions and semi-formed hunches - so make of it what you will, etc.

  • Are we talking about data or information?Lev Manovich uses the term "info-aesthetics" and connects these practices to the notion of an "information society". What if we move back a step, and look at the relationship between data and information? Data is the raw material, the datum or measurements: information is the message or meaning constructed using those datum. Both terms get used (more or less interchangeably) around artworks doing visualisation, but I think we should maintain the distinction. Is this work concerned with rendering information - a known, formed message? On the surface at least it seems to be more interested in visual interfaces to data, downplaying or leaving open the interpretation of that data - its transformation into information.

  • As I argued in "Hearing Pure Data," presenting the data "in itself" is an impossible ideal; it is inevitably shaped, interpreted, formed, framed, etc., in any manifestation; in which case how does visual data art negotiate its own construction of information from the datasets it works with? Does it pass off its own interpretation and framing as "raw data"?

  • What about the constitution of the data itself? Data art seems to take a pragmatic and concrete approach - "the data is the data" - but any meaning constructed from that data must be inflected by the way the data itself was formed or gathered. This is stating the obvious to anyone working in the empirical sciences... how do data artists respond? In the wake of the AOL reSearch dataset affair, the issue of constructing information from data comes into sharp focus. It will be interesting to see how artists use this dataset (which as Marius Watz recently observed, they no doubt will). The ethics of data art?

  • Data art treats its datasets as generative resources: sources of rich structure, pattern and complexity. It seems that often the appreciation of these formal qualities of the datasets (or their visualisations) exists in tension with the content or referentiality of the data. There's a continuum: TheyRule leans towards referentiality and meaning; Ben Fry's Valence is more concerned with pattern (it's an exploration of a visualisation technique after all); The Dumpster sits somewhere in the middle.

  • Toxi blogged a while ago on the issue of access to quality datasets for creative visualisation. As the comments on his post show, this begs a kind of cart/horse question. Tom Carden writes: "once you've got the info vis bug, you feel like a guy with a big shiny hammer, but nobody will give you a nail." This brings us back to the same question: is this work about data as an indexical link to the world, or data as a generative device? Or both?

  • On a related point, there's a clear crossover between generative and data-driven art; the artists are often one and the same; the same tools are used. How can we think about the relationship between these practices? They seem to be complementary approaches to similar goals (visual and aesthetic complexity, the joy of the unexpected, etc): one builds a generative system from scratch, the other latches onto the most complex existing generative system (the world) and visualises that.

Responses to all this very welcome of course... stay tuned for more chunks of undersupported and undigested theorisation.


jpb said...

Excellent set of points and questions, really excellent.

My most immediate response is to this:

meaning constructed from that data must be inflected by the way the data itself was formed or gathered

If you're interested in this question, you might be interested in reading the writings of Donna Haraway (Simians, Cyborgs, and Women in particular) or Thomas Kuhn's Structure of Scientific Revolutions: both deal with the way that sexist or institutional bias shapes the way that data is gathered or interpreted.

Mitchell said...

Thanks! Of course this idea is Philosophy of Science 101, as well as basic experimental design (within the sciences). Data art often just seems to ignore it. It's like the digitisation of data helps conceal the fact that data is extracted from the world in specific & limiting ways. The popularity of social web datasets (flickr, del.icio.us etc) reinforces this sense... because the data is kind of immanent and already highly abstracted from the world.

infosthetics said...

as with most art works, aesthetic visualizaton exploits the characteristics of ambiguity and the presentation of multiple, parallel interpretations of meaning.

notably, this is exactly what traditional visualization attempts to avoid. therefore, one should not judge data art within the traditional framework.

Mitchell said...

I agree, I don't think data art should be critiqued as poor visualisation, & that ambiguity is linked to its focus on data (the undetermined) rather than information. But at some point artistic ambiguity can start to look like having-your-cake-and-eating-it... ie claiming an indexical relation to reality but ignoring the construction of that index.

toxi said...

Mitchell, thanks for the pointer to my post... It seems to me in art circles, "Data" is being mystified and re-interpreted as malleable medium just the same as "Code".

Since reading Merleau-Ponty (see the 1st quotation from the book) I'm more and more convinced data art (as you call it) has a generative / synthetic character, in terms of meaning, and is not necessarily a reflection of properties extracted from the real world.

This is of course fine, but unfortunately artists often suffer from a credibility & justification complex and hence tend to attach the label of "data visualization" to it. However, this can be a very dangerous thing to do since it then is sold (without further explanation) as "reflection" of true facts of the world. This in itself is then amplified by the publicity these projects gain in the blogosphere, with most blogs only providing linkage plus copy & paste excerpts of the artist's description instead of a critical examination of these pieces.

Mitchell said...

Thanks Toxi. I think you're right: data is the new code. Both ideas are part of the cultural imagination of new media technologies & practices. We (as a culture) apply ideas that help us grapple with the complex, abstract & malleable nature of these systems - and those ideas always have their own poetry, figuration, motivations, ancestries (cf "cyberspace").

Mitchell said...

Just for completeness, the paper that came out of this post (in part) is over here.