Ladies and Gentlemen,

Mikhail Simkin's article on modern art makes entertaining reading, but the 
idea that such a quiz could provide serious evidence to answer
the question "How good are modern artists?" with "The only difference between
masterpieces and fakes is the heavyweight name attached to them" reveals a 
fundamental misconception of what modern art is about. This is a good example
for how statistics can go wrong by ignoring the inherent ways of thinking 
of the subject matter to be analyzed. 

To begin with, here are some problems in the experimental design and its 
interpretation.

1. The quiz compares sometimes poor screen reproductions shrinked by a factor 
of several hundreds, I guess, from the canvas originals with fakes that have 
been created originally on the screen. So what's on offer here from the real
artists is not what the artists meant it to be, while Simkin's fakes have been
produced under the same conditions under which they are displayed.

2. Reasons for not recognizing "true art" could be different from what the 
author imagines. For example, how could a respondent, who doesn't know 
painting No. 4 but several other paintings of Malevich, be sure that the 
author didn't plagiarize Malevich's style skillfully to arrive at No. 4, even 
though it may as well be Malevich himself? Some respondents may have assumed 
(not too unreasonably) that Simkin did his best to lead them astray, and the 
best possibility to do that is to plagiarize well known artists while 
offering atypical originals. Therefore some artworks may have been nominated 
as fakes not despite but because they look "trueish" by some respondents.  

3. The choice of original artworks has certainly not been done randomly. 
Obviously the author has chosen artworks on purpose that are minimalist in 
the sense that they don't clearly reveal much technical skill (at least not on
the screen) and could intuitively be taken as produced by a child - or a 
statistician.

This leads to the deeper problems with the author's reasoning. The author has 
asked to identify the original works, but it is not the main purpose of good 
art to be identified easily, and by a majority of people, in such a quiz. Why
then should the quiz results tell us anything about the quality of the 
artworks? The appreciation of art is subjective, and a work that stimulates
a deep esthetic experience among 5% of the participants while it's quality 
remains unnoticed by all others, could legitimately be considered as having the
job of a masterpiece done properly under such difficult conditions. The data 
don't provide any evidence against the hypothesis that such a thing happened 
with the real masterpieces, so I don't see any relevant evidence against the 
artists here.

Furthermore, the fact that most of the masterpieces selected here are 
deliberately minimalist points to another insight missing in the article. The
discussion whether valuable art can be produced based on simple geometrical 
shapes or on a technical level that children could master is part of the 
background of modern art. At least some of the artists experimented very 
consciously with this idea, leading to some of the original artworks shown
in the quiz, and the fact that the best part of an uninformed audience at 
first sight has difficulties to distinguish their works from fakes would 
rather come as a confirmation to them. These artworks are not only what
Simkin showed; they played a significant role in a specific context which 
cannot be recognized in such a quiz.

In fact, these works paved the way for what Simkin did when producing his 
fakes. Having done what Malevich did is much more of an achievement than
producing something that looks like Malevich today. These artists have changed
our perception. Instead of ridiculing the artists, Mikhail Simkin should be 
grateful, because they made possible what he has done. His own "works" owe
quite a bit to Miro and Elsworth Kelly, and even though he doesn't consider
himself as a talented artist, I'm sure he had at least some fun producing 
them. The fact that some of them are quite nice is rather testimony for the 
success of the original artists than proof of their failure.  

To come to the conclusion, the idea that quality in art could be modelled by
an objective one-dimensional weight scale to be inferred from quiz results
alone and a test whether "true art" can be distinguished from fake on this 
scale is a totally inappropriate conception of a statistician who seems to
overestimate the power of his discipline in such circumstances hugely. (Note
that Simkin doesn't even attempt to define what "true art" is.) 

Actually, the "creative" use of the concept of "artistic grams" and 
weightlifting categories to explain away the fact that despite all flaws more 
than 60% of the answers were correct even after taking into account possible 
knowledge of some of masterpieces to some of the respondents lets me wonder 
how serious Mr Simkin himself takes his claims.

Best regards,
Christian Hennig
Department of Statistical Science (not art!)
UCL