“Bikeshedding” is a term for the tendency to give minor issues disproportionate weight in the decision-making process. In a famous example from the naval historian C. Northcote Parkinson, a finance committee agrees to build an atomic reactor within a couple of minutes, but then deliberates for hours on what materials the clerical bike shed should be made of. Why does this happen? An atomic reactor is incomprehensibly complex, which means that most members of a committee have a difficult time wrapping their head around what it means. In contrast, everyone knows what a bicycle shed is and what it could be made of. Thus, everyone has an opinion about it.
For months, the little-used “Hairy/Furry/Shaggy/Fluffy” gene was The Art Genome Project’s bikeshed. Humorously, we often found ourselves debating its existence. We wondered how other controlled vocabularies delineate the furriness of artworks. We asked ourselves what the difference is between fur and hair or what to do in instances where you can’t tell if something is composed of faux fur, human hair, or real fur. Everyone on the team had some level of investment in it. We theorized that maybe people feel protective of hairy and furry things, i.e. their pets or other people. But then when we looked at the issue closer, we realized we were consistently preoccupied with whether this gene was supposed to capture content (what an artwork is depicting, e.g. dogs) or its physical qualities (e.g. furriness). When we stepped back and asked if every other appearance gene properly made this distinction, we realized that we had found our atomic reactor.
In reference to writing
, David Remnick, Editor-in-Chief of The New Yorker, has said “revision is all there is.”
One could say this is our approach to The Art Genome. We create but then we revise, revise, revise. We realized months ago that our appearance genes (which concern literally the appearance of artworks) needed a revision. In addition to confusing content and appearance, we found there were too many overlaps between genes. Also, some genes simply did not make sense to users. Additionally, some genes were heavily populated and others were applied to no works at all. So we went to work.
Our initial attempt at clarifying our genes looked pretty messy. Yet we kept at it. To bring some clarity to the picture, we decided to identify the genes we found most problematic (what we often refer to here as “worst offenders”): large, very general genes like Geometric, Emphasis on Line, and Thin Brushwork. We put them aside and evaluated whether they were even useful enough to keep. Secondly, we realized that to clearly see redundancies we needed to be more strict about what areas of appearance we were concerned with. As a result, we created the following divisions of appearance genes: Composition, Painting/Drawing Technique, Lines and Shapes, Overall Look, Concept, and Color Attributes.
We then spent a lot of time looking at artworks—analyzing shapes and lines; finding similarities and differences, in the attempt to provide conceptual clarity for this area of the genome and our users.
This is what our “map” of appearance genes looks like now.
Red indicates genes that we have decided to delete, green indicates new genes, and blue indicates genes that we are currently testing out.
We think these new categories of genes bring more rigor to Art.sy’s visual analysis, while leaving a lot of room to grow. Among other things, we’re excited that Emphasis on Line led us to a gene for Cloisonnism
; that a gene we had for Fluids led us to capture Fluidity
; and that we took the step of combining overlapping genes into individual genes, e.g. Sfumato and the “Hazy” part of Faded/Washed Out/Hazy have now become Haze
So what happened to poor “Hairy/Furry/Shaggy/Fluffy?”
Well, after a quick clean-up, we realized that it’s just not a very interesting way to describe works. We had two artworks on the site made using taxidermy: Kimberly Hart's Spectre 11
and Thomas Grünfeld's Misfit (Schwan / Esel / Nutruía)
and Spectre 11 also used feathers. And then there were a few images of furry things with hair-like fur
…Yes, we spent a lot of time talking, i.e. bikeshedding, about that dog. As for fluffy works… turns out there were none. This is all not to say that artists don’t make fluffy works, it’s just that a work’s fluffiness tends to be only one of many more important traits, and we cannot account for every single texture and phenomenon in the world. (Maybe we will have more of these works in the future and need to once again consider this gene.)
One final note: it’s important to ask why visual genes matter at all.
Appearance is important not only because it has an immediate impact on viewers, but also because certain physical properties—such as the mirrored quality of Michelangelo Pistoletto’s works
—might not translate in an artwork’s digital reproduction, even given a powerful zoom feature.
On a more theoretical level, concepts may inhere in appearances. Take this Max Bill work
, for example. Is the form of an infinity strip separable from the concept? Or, rather, is it separable from the concept most people have of the concept of infinity? What do we make of the fact that the fluid, undulating form is ossified in a material—granite—that is anything but? Or that the unbound vastness of infinity is tied up and anchored down by a heavy, immobile form? As Jerry Saltz has said, “Artists embed thought into materials.” In other words, the line between a purely formal quality (a visual trait you can see) and a concept is constantly being toyed with, exploited, and exploded by artists. For this reason, a few of our appearance genes actually became concept genes, such as Decayed/Aged/Deteriorated
A second, final note
: The title of this post is drawn from a now-famous 1928 essay
by Johannes Molzahn which asked Germans of the Weimar Republic to embrace the visual culture of the age. Despite its antiquatedness—it is, in fact, an almost utopian paen to mechanization—this essay shows that sometimes we need to be reminded of the most basic aspect of art appreciation: looking.