Algorithmic worlds |
|
|||||||||
Search blog posts2009-08-31 Fractals in traditional art 2009-08-23 Gigapixel panorama 2009-08-21 Yellowstone's abstract art 2009-08-20 Pollock & fractals? A hoax. 2009-08-04 Kunstformen der Natur 2009-08-02 The Fibonacci fractal 2009-07-29 The art of Kris Kuksi 2009-07-22 Group exhibition 2009-06-20 Kusama's patterns II 2009-06-14 Kusama's patterns |
BlogA blog about algorithmic art and fractal aesthetic. Click here to subscribe to the RSS feed. August 20th 2009 Pollock & fractals? A hoax.A few years ago, Taylor, Micolich and Jonas claimed that paintings by Jackson Pollock display fractal structures. I remembered this after seeing a Pollock in at the National Gallery in Washington. Out of curiosity, I tried to find their paper. The first article they published was in Nature which is (maybe with Science) the most prestigious unspecialized scientific journal. First surprise, this is only a very short report, without any data, and without any reference to a more serious article. Then on the webpage of Richard Taylor, I found an article published in Scientific American, a usually good journal which publish vulgarization papers (instead of research papers). As usual in this type of bad science paper, the authors start rambling about relations between art and science, how Pollock deciphered the language of Nature, etc... Where it becomes funny (or sad, depending on the point of view), is that they claim that they can find a unique signature of Pollock in the fractal dimension, that they can authenticate paintings and even date them! Richard Taylor was asked to authenticate some paintings recently found by Alex Matter... Here is a part of the story. Fortunately some serious scientists, Katherine Jones-Smith and Harsh Mathur, investigated their claims and showed their fallacies. Here is an answer in Nature (you may not be able to see this paper if you do not have a subscription, unfortunately), and here is a more technical one, where they refute the "authentification" method of Taylor & al. In the first paper, Katherine Jones-Smith even displays one of her works:
which is an authentic Pollock according to Taylor's criterion. Interestingly, it shows that, while the box counting procedure allows to compute the fractal dimension of a fractal, it does NOT allow to prove that a pattern is fractal. I guess many of the "scientific" works finding fractal patterns everywhere should be revised. Taylor & al. replied to Jones-Smith and Mathur's article in Nature, displaying an interesting collection of unscientific arguments. They say: "Our use of the term 'fractal' is consistent with that by the research community. In dismissing Pollock's fractals, because of their limited magnification range, Jones-Smith and Mathur would also dismiss half the published investigations of physical fractals." First unscientific argument, the agrument from authority. Indeed I have no problem believing that half of the papers published on "physical fractals" (whatever it means) are crap. "Fractal description is physically reasonable because Pollock's technique involved a motion-dominated process at large length scales and a paint-dominated process at small scales, ... However, for Untitled 5 there is no physical reason to expect a transition..." The classical "My dreams are reality." argument: Pollock's paintings are fractal, and Untitled Nr.5 is not, because we expect them to be so, regardless of the numerical data. They end their reply with a cynical "We encourage further research.", whereas the raw data of their analysis of Pollock paintings does not appear anywhere. The saddest part of the story is in my opinion the fact that these people managed to publish a report in Nature without showing a single bit of concrete data. Still, the final lesson about the inefficiency of the box counting procedure to spot fractals is a very valuable one. comments powered by Disqus |
|||||||||
|