In their paper Artistic Data Visualization: Beyond Visual Analytics, Fernanda B. Viégas and Martin Wattenberg claim that artistic data visualizations “…must be based on actual data, rather than the metaphors or surface appearance of visualization.” What they seem to be saying (though I will admit I may be reading too much into this statement) is that metaphor cannot apply to any ‘artistic’ visualization directly derived from data. While the article is well written and researched, I will try to explain why I fundamentally disagree with this premise. [Note: In writing this it appears I misinterpreted their statement: see comments below.]
Metaphor simulates: it seeks to explain one subject through another. For example, metaphor is used widely in software, where it uses concepts from the physical realm to explain the virtual. It lies at the center of the use of the computer as a simulation engine, to borrow a concept from Lev Manovich. As Manovich has also said, an important innovation of computers is that they can transform any media into another. In The Anti-Sublime Ideal in Data Art, he critiques the arbitrary nature of most media transformations in the field of ‘artistic’ data visualization. Why use this particular form of visualization? What influenced its choice over another? The answer to both questions is intent, and I’m certain Viégas and Wattenberg would agree. In the relatively new world of data art, we (the viewers) expect to find rationale behind every decision, no matter how minute or unimportant it may seem to the artist. Intent is questioned every step of the way, and any breach in the logic determining the construction of a piece undermines its artistic integrity.
Metaphor drives intent, and vice versa. In data visualization, metaphor is used to translate raw data into visual form. The choice between a sequential representation or a network diagram is based on intent and reflects the interpretation of a data source. It makes a statement about the content, which, in the end, is always based on ideology. For example: “emotions are peaks and valleys” (as in Christian Nold’s Bio Mapping, shown above), or “equities are territories, and a market is an agglomeration of territories” (as in Wattenberg’s SmartMoney Map of the Market).
No visualization is a literal translation. Metaphor inserts itself where data is made to correspond with any form of representation. Since visualizations cannot be “objective” translations of the structure of a dataset, they must instead be metaphoric (or at least analogical). And as Denis Wood has explained, even scientific visualization such as satellite imagery is inherently subjective. Hence, any type of visualization has specific connotations, which may become metaphoric when seen in context of a specific data source. Metaphor in visualization works at the level of structure—it compares the composition of a dataset to a particular conceptual construct, and the choice of any visualization is always a matter of interpretation.
Data visualization necessarily speaks the language of the aggregate. Hence, art involving visualization must fundamentally make a statement contrasting the specific and the general. Artistic data visualization uses metaphor in a very particular way, namely to bridge the aggregate (the general) with the individual human experience (the specific), thereby making a statement on the human condition.
At first glance, Jason Salovon’s Every Playboy Centerfold claims to make an ‘objective’ visual comparison which reveals a certain ‘truth’. In the words of the artist, the piece “tracks, en masse, the evolution of this form of portraiture.” However, the claim he is really making (at least from my perspective as the viewer) is that our notion of beauty is based on uniformity. The visualization does not validate or ‘prove’ that idea. The only conclusions that can be drawn from it are indeed based on the aesthetic choices governing the centerfold spread—that, and the general shape of the human body. Yet, it seems clear that the artist is using a visualization—here, a multiplicity of specific objects—to make a statement on the human experience.
Gothamberg, a Turbulence commission including Martin Wattenberg among its various contributing artists, uses individual stories in the aggregate to metaphorically reference the life inside an apartment building. Its brilliance derives from the fact that the visualization is informed by the locations of the stories, which convey the shape of a building.
So Viégas and Wattenberg: “All these projects share one characteristic that distinguishes them from traditional visualization tools: Each embodies a forceful point of view. In a sense, the artworks derive their power from the fact that the artists are committing various sins of visual analytics.” And point of view, I would argue, is often communicated through metaphor.
Hi Christian,
I just read this blog entry. It’s a wonderful essay on the role of metaphor in data visualization. Ironically, I completely agree with what you’re saying! In the “Artistic Data Visualization” paper that you quote, Fernanda and I didn’t mean to rule out all metaphors–as you point out, they are essential to a lot of “core” artistic data visualizations. Instead (and this was confusing) we meant that using the surface appearance of visualization wasn’t enough to make something “artistic data visualization.”
We asked ourselves, what if someone took a visualization or diagram and erased all the labels? Or simply replaced them all? Should that fall under the category of data visualization? These aren’t hypothetical questions, since Simon Patterson (whose work we love) does exactly this in “J.P. 233 in C.S.O. Blue” and “Great Bear.” We decided that even though these works use the appearance or visual “metaphors” of visualization, they shouldn’t be categorized as displaying data.
Hope this makes sense… and thanks again for such an interesting article!
–Martin