Big Data and the Art World: Mutual Art

moti-schniberg

Walid Raad peels back the onion of Artist’s Pension Trust and it’s parent Mutual Art, the operator of MutualArt.com, to reveal the big data operation being assembled. Artsy has received a lot of press for its art genome project but here Raad meets Moti Schniberg, the man behind Mutual Art and gets a lesson in how the firm is modeling the art market for private investors:

MutualArt.com is a website for anyone who loves the arts and wants to keep track of art matters in general. For example, if you like the work of Francis Bacon, you can sign into MutualArt.com and register this preference. You like Sophie Calle, Francis Alys, David Diao, or my work, you do the same, and the site’s algorithm is fantastic. It scans the net and delivers articles, essays, exhibitions, and catalogs that matter to you and only to you. Essentially, MutualArt.com is a database of its users’ preferences.

Then Moti tells me, “MutualArt has millions of users who have registered millions of preferences, and as such we are on our way to building the largest data set about the art world anywhere. Except that our data set remains a noisy picture, and in order to turn it into a clear picture, with reliable tendencies, clusters, and nodes, we hired Ronen Feldman.”

Of course they would hire Ronen Feldman. After all, Feldman is the computer scientist, the man renowned for having coined the term “text analytics” in 1995. And I later found out that Feldman is also a graduate of the most celebrated Israeli military intelligence unit: Talpiot. It’s the equivalent of taking Harvard and adding to it Berkeley, MIT, Caltech, Princeton, and Yale.

Anyway. Feldman’s job is to write algorithms for very large data sets. He works with what is called “Big Data.” Take any large dataset, filter it through a Feldman algorithm, and within seconds—as if by magic—all kinds of tendencies, clusters, and nodes begin to appear. These tendencies, clusters, and nodes help to formulate questions and answers. The questions and answers MutualArt was interested in are primarily about the art world. Lets take auctions for example: When is the best time to buy or sell a painting by Andy Warhol? Spring, summer, fall, or winter? If in spring, is it better to sell in March, April, or May? If in April, is it better to sell in the first half of the month or the second half? On a Monday or a Tuesday? At Sotheby’s in London or Christie’s in New York?

MutualArt has answered these questions, and the result is an instrument, a financial product that MutualArt will soon be selling to institutional investors.

Then Moti also tells me about other questions that MutualArt wants to answer, such as: How many articles written by what writers, in what art magazines, using what art language will affect the performance of an artwork coming up at the next Christie’s auction?

The last algorithm Moti tells me about, the one they are working on as Moti and I speak in 2009, is an algorithm about color.

That’s right, color. More specifically, color in postwar European art. What percentage of blue? What percentage of red? What percentage of yellow? What percentage of black must a Picasso painting from 1946 contain in order to increase in value by 37 percent over the next sixty months?

As he’s telling me this, I begin to see how the risk management techniques of Dan Galai are being combined with the text mining expertise of Ronen Feldman in order to put in place—not in five years, but today—complex, dynamic, and real-time prognostic models about anything having to do with the art world.