Joshua Davis outside his Long Island studio.
In the pale-skinned and over-caffeinated world of Internet designers, Joshua Davis is a rock star—the Tommy Lee of digerati. A self-taught Flash master, he rose to prominence in the ’90s with his work for the now-defunct Web design firm Kioken, as well as his personal sites: dreamless.org, a once hugely popular message board for the design community, and the open source Flash laboratory/showcase praystation.com. Nowadays, the heavily tattooed 34-year-old appears frequently on stage in a tank top, lecturing to packed audiences. But at his modest, three-story Colonial Long Island home/studio, which he shares with his wife and 2-year-old daughter, Davis might feel more comfortable hanging with Marcel Duchamp than J.Lo (though she happens to be a client).
Chaos is currently his work’s dominant theme. He not only encourages randomness, but also manages to control it. He’s been dabbling in artistic mess, on and off, since the mid-’90s. While studying illustration and communication design at Pratt Institute, he liked to freeze or bake paintings to see how the stress affected pigment. One of his favorite inspirations of late is Jackson Pollock. "I’m not as crazy about his work visually as I am conceptually," Davis says, referring to Pollock’s mastery of accident.
Images created as backgrounds for Nokia’s N-Gage phone represent the evolution of an artwork "trained" by Davis’s Genetic Aesthetic software. Between images 01 and 06, the software learned which arrangements are most visually pleasing, in response to the designer’s rating system. "It figured out black-and-white balances—05, for instance, cuts the space wonderfully, whereas 01 doesn’t seem to be anchored anywhere," Davis says.
John Maeda, a multimedia guru not known for extravagant compliments, praises Davis as "one of those people who don’t just talk, they make stuff. Davis, Ze Frank, Yugo Nakamura—they contribute to the collective knowledge of things that are made digitally." For its March 2005 issue, fashion glossy Zink magazine put Davis’s riot of floral allusions and bubbles on its first illustrated cover plus three spreads. "He is truly an amazing artist," says Denise Dandeneau, assistant to the publisher at Zink. "He is not only a genius with Flash, but is a great guy and very personable."
Davis’s primary tools are generative composition machines—applications and systems, written in Python and Flash, that automate his workflow. Color Stripper, the first of his machines or algorithms (as he alternately identifies them), outputs 64 colors based on transitions between three user-defined base colors. His Random Controlled Color application employs hues chosen by Color Stripper to fill blanks in scanned, hand-drawn line art. With either application, a click or touch of a button enables Davis to edit out colors singly or generate new iterations. The third machine in his arsenal, Composition Generator, outputs finished works populated by items from an asset library—a folder housing a collection of images Davis draws by hand for each project.
"There are a lot of people doing computer-based generative design, using the program to draw the art," he says. "My spin on that takes artwork that I’ve already predetermined. I define its assets, what it can use, how many objects I would like it to paint."
He bounds each machine with rules for using the asset library. "I have a program that generates forests," Davis explains. "I’ve defined seven possible trunks, six possible leaves, and fifteen possible things that can live in those leaves. All I do is say, ‘Here’s the canvas, give me twenty trunks.’ And the program places those trunks within the prescribed field. Then the trunks take over and say, ‘I need to populate myself with leaves.’ Then the leaves wake up and say, ‘Okay, I have to do this.’ It’s like a domino effect in which you’re defining a pattern of systems where one thing, set into motion, can trigger an entire series of events."
The key drawback, Davis admits, is noise—the composition machines have a very high failure rate. For "Accident Happy," an exhibition that ran from March 2 to April 11 at Alfred University in upstate New York, Davis produced prints with assets by artists Derrick Hodgson (cuddly yet sinister animal forms, eerie floating mouth-breathers), James Paterson (scribbled organic skulls, tentacles), and Davis himself (jellyfish). Outputting the prints was comparatively simple, but Davis still had to sift through hundreds of haystacks for a few golden needles.
To combat compositional noise, Davis and his studio team—partner Branden Hall and four staff designers (average age: 28)—have set out to design software they call Genetic Aesthetic. Now undergoing beta testing, it’s a neural network that rates a machine’s compositions on a scale of 1 to 8. "As I feed the interface my pattern structures, it generates a form and asks me to rate how well it did," Davis says. "Over time, the network becomes imprinted with what you think is an aesthetically beautiful arrangement. The rating system evaluates a composition based on various inputs such as scale, rotation, placement, depth, and the particular assets used. It will look at where things were composed and understand how to make favorable decisions in the future, and still deliver deviations."
Genetic Aesthetic’s other major element will be a data file that interprets its aesthetic success. Davis refers to this as the software’s genetic component; the aesthetic imprint can be extracted and disseminated. "Once we get to the point where we can distribute this software, and people can imprint it with their own aesthetic preferences, we can extract that DNA imprint," he says. "Then we can mate two or more of these imprints to create an entirely new aesthetic; additionally, we can train the resulting genetic string."
As Davis found out, even a well-trained composition generator can wreak havoc. For a show in April of graphic designers’ wallpaper at the Barcelona gallery Maxalot, he submitted a machine-generated image from a series called Once-Upon-A-Forest. The 8×19-foot image file elicited a harried phone call from the printer. The program generated 50,000 vectors, subsequently covered by a large opaque object. "He thought I was either playing a practical joke or was torturing him," Davis laughs.
Despite the seemingly unstructured nature of his projects, "We get a lot of corporate clients—Nokia, Sony—that want to use this stuff," Davis says. "It’s good for them, because we write applications that are constantly morphing and sort of have a life of their own, so content always appears to be fresh." For Sony Vaio (available in Japan), for example, Davis’s studio devised a freestanding desktop application with calendar, time, and MP3-playing functions. Its random abstract animation changes color with the seasons.
The composition generators can output to a wide variety of media, so Davis can quickly create multiple options for print, broadcast, or the Web. For the 2004 Red Hot Chili Peppers tour, Davis delivered 16 versions of background animations for a single commissioned video project. "They thought I was crazy," he says of the band. And he’s now talking to a small, independent fashion label called Tekstil, figuring out whether the generator can pump out textile patterns.
Davis knows his work may rankle designers using dyed-in-the-wool methods of trial and error. He’s ready with counterarguments: "I’ve had some people say that I’m no longer the creator of the artwork. But I created the algorithm, and the artwork assets the algorithm uses, so it couldn’t be any more me. People take issue with the fact that my role at one stage is that of critic. But I’m still the same artist I’ve always been; it’s just that my weapons have changed. If beauty is in the eye of the beholder, well, we’ve programmed the beholder."
Rich Hoxsey is a freelance writer, artist, and cook living in Brooklyn.