89 The Art Genome Project
By David Eicke, December 2014.
What is being organized? Artsy.net carries the ambitious mission of making “all the world’s art” accessible to anyone with an Internet connection. This is not only challenging purely from a scale perspective, with the number of artworks in the world daunting even if it were not being incremented constantly, but it is also challenging in that “art” is a nebulous term. Creators of music and literature often refer to themselves and each other as “artists.” The same goes for dancers and other performers. Will their works be included? The current collection seems to be mostly visual art, with some architecture and design objects included.
Artsy’s mission is to be carried out by their Art Genome Project, which is the organizational engine that powers their search and interactions. The name was inspired by Pandora’s project, as was their term for their organizing process: “genoming.” Genoming is not yet automated and still costly, so Artsy selects the art that is to be “genomed” carefully. Their first priority is the works featured in galleries with whom Artsy has contracts. Galleries pay to have their work organized and searchable on the site. Those works, then, must be genomed quickly in order to keep the company running. Artsy’s engine also takes in works from museums and other institutions who do not have contracts with them, but many of those institutions have image-rights concerns, and not all their artworks can be published. In other cases, the images of the works are simply too low-quality to be displayed.
Why is it being organized? Why organize art? The simplest answer is to educate. That said, art has been being organized into movements and -isms for a very long time. The Getty Foundation even created an authoritative art vocabulary called the Categories for the Description of Works of Art a few decades ago. At first glance, Artsy seems to be reinventing the wheel. However, the organizing system Artsy uses is unique in that it facilitates a special kind of interaction with its body of published works.
The way resources are organized on Artsy is a cross between a hierarchical structure and a graph structure. They have over 1,000 characteristics (which they call “genes”) to describe their resources. These characteristics can have to do with art movements, formal qualities, techniques, subject, etc. The emphasis here, however, is on relationships between works of art. For example, one of the genes Artsy uses is “eye-contact,” and if you have a photo taken last month where the subject is looking directly into the camera and an oil painting from hundreds of years ago where the subject’s eyes are looking at the painter, those two can be one click away from each other. No other organizing system could facilitate that sort of easy link between two such disparate works.
This free-flowing linkage between works enables the “berry-picking” model of knowledge seeking, where a user searching for something doesn’t necessarily have to know what he or she is searching for. A user could begin her exploration with only a vague notion that she enjoys this long-legged rhinoceros sculpture by Salvador Dali. She may not know what she likes about it, but she will see his other work there. Maybe she finds a painting she likes in the “other works by Dali” section, and she clicks on it. Then the characteristics of this painting are listed in the interface, and she is free to click on any one of them. She might click on “Surrealism” and find more works from that movement. She may click on “waterscapes” and find other oceanic imagery. She is free to explore and discover art in a self-directed way and free to discover what she likes and why she likes it. The director of Artsy’s Art Genome Project says the system was intended to parallel a professor who is adept at “riffing” on things.
How much is it being organized? As mentioned above, Artsy currently uses over 1,000 characteristics (“genes”) to describe its resources. These characteristics can describe anything from the art’s form to the art’s subject to the technique used to create the art. Experts assign these genes to the artworks and then assign those genes a weight from 0 to 100, depending on the salience of the characteristic within the work. Aside from the genes, the art is described in terms of physical dimensions (how much space it takes up), whether it has been sold or not, its gallery, its price (if for sale), its creation date, and, of course, who created it. Having such a rich set of descriptions has allowed Artsy to create a public API for developers to use all of this information as they see fit.
When is it being organized? Description of Artsy’s resources is an ongoing process. Their ingested collection of art is much larger than their published collection. Most of the artworks are waiting to be genomed, with some of them waiting for permissions or image-rights paperwork to process. Another factor in determining when something is organized is the signing of new contracts with galleries. Works from galleries with contracts have first priority, and Artsy experts genome those works as they come in.
While these experts are assigning genes on a rolling basis, they are also drawing upon hundreds of years of art history scholarship when assigning them. For example, the Arsty experts did not come up with Dadaism as an organizational concept. So, in a way, some of these works were organized long ago.
How or by whom is it being organized? Artsy has a team of art historians and experts working to describe the resources that Artsy has ingested (and those that it will ingest). They have done some experiments with image-recognition software, but its descriptions are simply not rich enough to facilitate the sorts of interactions the organization is trying to facilitate. The strategy of employing experts has its obvious downsides, however. It does not scale well, and it is reminiscent of Yahoo’s early strategy of employing librarians to describe web content. There will also be inevitable biases in human resource description.
Other considerations. With such a grand ambition, one thing that may stand in Artsy’s way of becoming an authoritative organizing system in the art space is that they are for-profit. Even if they are able to avoid too much bias in the interest of revenue generation, the perception remains that they are less interested in classifying art for educational purposes and more interested in making money.