Cultural analytics: a new way of analysing culture
Yesterday I bumped into presentation by Lev Manovich. After having seen our national soccer team win of Denmark (with some luck), I was in the same place where he was about to make his appearance. Ignorant as I am occasionally, I didn’t know at all what Manovich was doing. So some time before his presentation I asked some of the more academic folks at SETUP Utrecht what to expect. Since I couldn’t really form a mental image of a media theorist. I was told he wrote a classic book called The Language of New Media and he uses large datasets to analyse culture. Enough to get me interested. Here are my very non-academic observations (watch SETUP for a more academic article) of Manovic’s presentation. Which not only turned out to be a ‘very acceptable’ presentation but highly thought provocative as well.
Lev Manovich started his talk by stating the evident: we are using so much new media at this moment in history (e.g. 48 billion unique images on Facebook). So for Manovich us using new media is a huge and ever expanding jar in which culture is being created. So he has decided to analyse this amassing culture by the same means it is created — by computing. Thus cultural analytics was born. He continued by giving a brief overview of the evolution of his method of analysing. From reductive statistics — when statistics are applied on visual arts effectively you reduce the cultural artefact to data points which then can be plotted in a graph. To non reductive statistics to analyse visual arts. Here it got interesting.
Lev Manovich showed several of his on-going research in which he, assisted by his team, is analysing culture with non reductive statistics. He showed an image which contained a grid (left-right-top-bottom) of all pages of all issues ever published of Science Journal and Popular Science (See: Software Studies post with images). It almost directly it offered another dimension to these magazines. It became clear that Science journal was visually richer and became during the years less visually rich. Whereas Popular Science showed the opposite effect: from poor to richer illustrated. Everything is described on the lab.softwarestudies.com website.
Manovic, while showing his results, continuously made side-steps in which he explained the relative simplicity of how he came to these results. His tools of the trade are mostly open source and include programs such as ImageJ and Mondrian. He advocated everybody to start analysing yourself. Using something like a Macbook can get you decent results on oversee-able datasets. Although analysing may not proof to be easy to start with, something tells me this might become an important new instrument.
The cherry on the pie was a project in which 1 million Manga pages retrieved from the largest Manga-fan site were analysed. Based on the values of each individual manga page on brightness (X-axis) and entropy (Y-axis) they positioned each that page onto a graph. The results again gave a completely new dimension of understanding being neither subjective using words to describe Manga nor reductive like classical statistics. This image-set gives a nice overview of the results of the Manga analysis.
After his lecture there was the possibility to ask Manovich questions. But it might just be that most of the people present were still gasping for air to digest all of what Manovich had previously said. To me personally it only started to make sense long afterwards his presentation, while riding my bike home. The only thing I previously knew related to this topic was a Microsoft Live Labs-tool I came across some months ago called Pivot. I might give it another spin to start analysing culture myself.
