This is Data? Arguing with Data Baby

These IBM commercials are gorgeous, lavish examples of modern motion graphics from Motion Theory. Like some of the agency's earlier work, and a handful of other examples noted here, these ads show how code-literate design (could we call it the P factor?) is transforming this field. For all those reasons, I love this work; but it also really bothers me. I'll try to explain.

IBM - Data Baby from John Stanch on Vimeo.

The opening line of this voiceover says it all, really. This is data. Making that call - defining what data is - is a powerful cultural gesture right now, because as I've argued before data as an idea or a figure is both highly charged and strangely abstract. It makes a lot of sense for a corporation like IBM to stake a claim on data; this stuff is somehow both blessing and curse, precious and ubiquitous, immaterial and material. IBM promises here to help with the wrangling, but also, most powerfully, to show us what data is.

So, what is data here? In these commercials data is first and foremost material. It is a physical stuff. In Data Baby it wraps a little infant like some kind of luminescent placenta, drifting away into the air, thrown off in shimmering waves as the child breathes. In Data Energy it trails like a cloud behind a tram, and spins with the blades of a wind turbine. A lot of the (beautiful) animation work here has been devoted to simulating behaviour, making this colorful, abstract stuff seem to be tightly embedded in the world with us. What that means is both coupling it tightly to real objects, and supplying it with immanent dynamics - making it drift, disperse or twirl.

IBM Data Energy from Michael Suarez on Vimeo.

The second interesting property of data here - related to the first - is that it just exists. Look again at Data Baby, and note that there is no visible sign of this data being gathered (or rather, made). No oxygen saturation meter, no wires, no tubes, no electrodes. Not a transducer in sight. Not until the closing wide shot do we even see a computer. (This is fascinating in itself; IBM (or their ad agency) gets it that the computer is no longer the right image, or metaphor, for "information technology". Neither is the network; now it's immanent, abundant data.) In other words data here is not gathered, measured, stored or transmitted - or not that we can see. It just is, and it seems to be inherent in the objects it refers to; Data Baby is "generating" data as easily as breathing.

Completing this visual data-portrait are some other related themes: data is multiplicitous and plentiful, it's diverse (many colours and shapes) but ultimately harmonious and beautiful - in Data Transportation it looks like an urban-scale 3d Kandinsky painting.

Several things bother me about this portrayal. The first is the same is the reason I love it: it's powerfully, seductively beautiful, and this amplifies all my other reservations. The vision of data as material, in the world, is also incredibly seductive; my concern is that we get such pleasure from seeing these rich dynamics play out - that the motes wafting from Data Baby's skin seem so right - that we overlook the gaps in the narrative. This vision of material data is also frustrating because it has all the ingredients of a far more interesting idea: data is material, or at least it depends on material substrates, but the relationship between data and matter is just that, a relationship, not an identity. Data depends on stuff; always in it, and moving transmaterially through it, but it is precisely not stuff in itself.

You could say that I'm quibbling about metaphors here, and you'd be right, but metaphors are crucially important because they shape what we think data is, and what it does. Related to data as stuff is this second attribute; data that just is, in the same way that matter is neither created or destroyed, but just exists. This is crucially, maybe dangerously wrong. Data does not just happen; it is created in specific and deliberate ways. It is generated by sensors, not babies; and those sensors are designed to measure specific parameters for specific reasons, at certain rates, with certain resolutions. Or more correctly: it is gathered by people, for specific reasons, with a certain view of the world in mind, a certain concept of what the problem or the subject is. The people use the sensors, to gather the data, to measure a certain chosen aspect of the world.

If we come to accept that data just is, it's too easy to forget that it reflects a specific set of contexts, contingencies and choices, and that crucially, these could be (and maybe should be) different. Accepting data shaped by someone else's choices is a tacit acceptance of their view of the world, their notion of what is interesting or important or valid. Data is not inherent or intrinsic in anything: it is constructed, and if we are going to work intelligently with data we must remember that it can always be constructed some other way.

Collapsing the real, complex, human / social / technological processes around data into a cloud of wafting particles is a brilliant piece of visual rhetoric; it's a powerful and beautiful story, but it's full of holes. If IBM is right - and I think they probably are - about the dawning age of data everywhere, then we need more than a sort of corporate-sponsored data mythology. We need real, broad-based, practical and critical data skills and literacies, an understanding of how to make data and do things with it.

Read More...

Summary only...

Murray McKeich - Generative Gothic

While I'm in Melbourne I've been trying to catch up with some of the artists who have made this town a new media hotspot over the past decade or so. I recently met up with Murray McKeich, an artist who came to prominence in the 90s here through the amazing (and sadly departed) Australian magazine 21C. He's refined a signature style, imaging urban detritus on a flatbed scanner, then grafting those elements together into surreal-gothic hybrids. In the hyped-up, Wired-style 90s, McKeich's images in 21C were startling; artefacts from a far more unsettling future.

In recent years, as I mentioned a while back on Generator.x, McKeich has discovered generative art. His approach, and the resulting work, are interesting in part because they are so different to a lot of what goes under that banner at the moment. At the core of his practice is a kind of heresy I find really appealing: McKeich doesn't code. He cheerfully admits to being "hopeless" at programming, and has no inclination to start. I had to fight the urge to talk him around, the way I do with students sometimes, leading them gently into the joys of Processing. But I had a feeling it would be futile, and besides, the processes McKeich has devised are coding, of a sort, and they are working beautifully.

Having accumulated a massive library of scanned-in source material, and discovering Photoshop's actions, McKeich began to experiment with automated processes; macros that would randomly pull source files into large multi-layer compositions. Hierarchies and groups of layers and sources provided a mix of control and randomness. He ran batch processes that would output many thousands of stills, then hand-picked the best to form very large image sets, with works like A Thousand Pictures of Footscray and DVN (detail above). The artists's motivation here, he insists, is pragmatic, not conceptual; he's interested in the specifics of an image, the moment of its impact, not in process for its own sake. For him generative techniques are essentially a matter of externalising aspects of his own process; computational studio assistants.

McKeich's next step was his discovery of AfterEffects, which he describes as "a superior imaging tool" to Photoshop - even for stills. With a more procedural approach, nested compositions, and powerful automation, AfterEffects remains his generative platform of choice. As a kind of bonus, it produces video. McKeich's procedural motion graphics use his signature palette of materials, but feel lighter, more ephemeral. In Maddern Square (above) we seem to skirt the edge of some dense conglomerate of street flotsam which is forever dissolving into itself.


Most recently McKeich has begun a new line of work - in one sense another brilliant heresy in the super-abstract context of generative art. He's been making faces, or rather, zombies. pzombie is from "philosophical zombie," a term for a hypothetical non-conscious human in a cognitive science thought-experiment. Like digital Golems, McKeich's pzombies (above - hi res) are cooked up from junk and grime, articulated by recursive coils of AfterEffects scripting. Smoke and mirrors, in a sense, but they have that visual impact McKeich is after; he shows them in large groups, which adds to the uncanny effect. The apparently infinite variety of these faces makes them both more intriguing and more unsettling than the usual science fiction clone-armies. While the artist might deny it, there's a conceptual hook here too; who are these portraits of, after all? Aren't these zombies the faces of those studio assistants, who work tirelessly through the night, those macros within macros. This time, they've been rendering themselves.

Read More...

Summary only...

Procedurally Hip: Generative Motion Graphics

The title sequence for the latest Bond film Casino Royale is one of its high points (which admittedly is not saying much). It's a stylish motion graphics confection: stencilled rotoscoping, shatters with simulated physics, prominent Saul Bass references. More interesting for me, some of it appeared to use classic generative techniques, shape grammars and branching recursion. Playing card iconography grows into curvy tendrils and later, Mandelbrot-like blobs. There's some general info online but not much on specific techniques - are there Inferno plugins for doing recursive geometry? Did the designers do some custom coding? The sequence was directed by Daniel Kleinman and designed by William Bartlett and Adam Parry at Framestore CFC.


Procedural aesthetics in motion graphics are nothing new - Saul Bass used (analog, optical) procedural techniques. But the Casino Royale titles coincide with a recent trend towards custom coded generative and procedural elements in commercial motion graphics. See for example the Nike 'One' ads by Motion Theory and more recently the Audi TT teaser from Universal Everything & Toxi (which now has a making of video). I'm certainly not the first to observe that code-based tools like Processing are gaining a foothold in genres dominated by proprietary tools like AfterEffects and Shake.

I've started talking about this trend with my students, as a way to argue for why they need to learn to code (and initially at least, learn Processing). It works as an argument because motion graphics as a form is, for many of my undergrads, the sexiest thing around. It's the high-speed flagship of commercial visual culture - where new styles emerge, proliferate, and are superseded within a few weeks. Which is why this generative trend (if that 's what it is) is all the more interesting: is motion graphics going to chew through generative techniques like they're last year's hottest AfterEffects filter? Or will tools like Processing actually change the practice in this field? Custom code holds the potential for more aesthetic diversity, but how will that fit with the trend-driven visual economy of commercial motion graphics? What about the open source ethos? Whatever, it's bound to look really cool. And that's the main thing, right?

Read More...

Summary only...