The National Exhibition of Fine Arts (CNEFA) is the most prestigious government backed art competition in China. Designer Xiang Fan and computer scientist Shunshan Zhu believe they have come up with a data-driven formula to help artists hack their way into winning the contest: “Paint a young woman sitting in front of a window with a cow, enjoying a patch of sunshine.”

The formula may sound ridiculous, but the research was a very serious look at a contest run by a government with a history of keeping a tight handle on media, including artists and artwork. Of course, any art contest will be driven by the biases of the judges, but without being able to examine all of the winning work as a whole, those biases typically remain hidden. In this case, the biases reflect not just a single contest and a handful of judges, but the biases of a government that has dictated the artistic production of 20% of the world's population for the last 70 years. Many have speculated that the tight control of the Chinese government may be leading to increasingly homogeneous and formulaic submissions to the contest. The impact of this goes well beyond the contest, as the winners of the contest are instantly "canonized" and typically given high profile academic positions where they further influence the production of art in China. As Fan described it:

When I was a child in the 1980s, my Dad hung out with some artist friends who had won this contest. I know he admires them a lot and is proud of their accomplishment. He would take me to the exhibitions with him, though he never had the intention to push me into art. Having grown up attending the exhibition with my father, I saw how prestigious the National Fine Art Exhibition is. Thirty-five years later, I visited this exhibition in 2014, and surprisingly I found that some of awarded works looked the same as the paintings I saw as a child. It was hard for me to comprehend how little things had changed, as I knew this award could mean a lot in tenure-tracking in education and pricing in the art market.

Rather than just speculate or take sides, Fan and Sam (Shunshan) took a scientific approach and let the data speak for itself.

Fan and Sam looked at the 2,276 winning paintings from the competitions ranging from 1984 to 2014. The first contest was held in 1949, the year that communist China was founded. The 2nd through 5th contests were held in 1955, 1962, 1964, and 1980, respectively. Starting from the 6th contest, which was held in 1984, the contest is held every five years. Sam explained:

We focused on the awarded works from 1984 (6th) to 2014 (12th). The reason is that from the 1st to the 5th, either there is lack of information for the paintings, or we could only find catalogues in black and white, or the awarded works could not even be called art.

There was no list online of winning submissions, so the duo resorted to the arduous task of locating printed material in university libraries to track down the winning entries.

Once the source material was gathered, they manually tagged all the works to create a database that includes the artist, the name of the work, the dimensions, location created, color, and subject matter. Subject matter tags included things like “woman,” “soldier,” “landscape,” “still life,” etc.

Though many suspected that the winning artworks in the contest were becoming increasingly homogeneous, the scale of the contest made it difficult for anyone to know for sure. For example, there were over 500 winning artworks in the 2014 contest alone. In order to really find any visual trends, Fan and Sam needed to develop a custom tool that would let them see all the works at once and filter those works using different criteria. Inspired by media theorist Lev Manovich and his project imageplot, the team set about building their own custom tool for analysis.

Sam, a computer scientist, developed the interface they later named "AwardPuzzle" using Processing 2.2.1. "We used its Java mode to build the standalone versions for Windows and MAC boxes, and used the JavaScript mode to build the web versions for browsers running on desktops or mobile devices. Basically, the programming language was Java, JavaScript, and HTML."

Sam explained that before they started the project, they were thinking about questions like: "Will artworks with brighter colors be more likely to win?"
"Do larger works have a better shot at winning?" "Does certain subject matter improve odds of winning?" While the tool helped them explore those questions, it also unlocked many new questions.

"We found new questions like 'How can some artists win multiple times for artwork that looks very similar to their previous submissions?'" Fan, a highly talented information designer, was not surprised by the discovery of new questions through visualization and quoted designer Ben Schneiderman, saying “visualizations give you answers to questions you didn't know you had.”

Below are some of the artworks Fan and Sam discovered that won multiple prizes (in some cases in the same year) despite being remarkably similar.

The two nearly identical paintings above were both painted by the artist Zhanfeng Wei and both won a prize in the same year of the contest, 2014. All Artists are of course known for repeating themes in their work, but getting two awards in the same year for nearly identical subject matter seems a little strange.

Artist Yue Liao won for the painting Noon in 1999 and then again for a nearly identical painting titled Newborn Calf, submitted just five years later. I suppose if Damien Hirst can have over 1,000 spot paintings and Richard Prince has his Nurse series, why shouldn't Yue Liao be allowed a Calf series? In all seriousness, this feels more like an oversight by the judges to me than the artistic exploration of a theme; in this case, a cow. If it were the only oversight, it might be comical, but many pairs like this pop up throughout the contest results.

Looking across the winning entries through Fan and Sam's interface also makes it apparent that the subject matter across all the entries rarely departs from realist paintings of farming, military, and domestic scenes. These themes played heavily in the Socialist Realism period of Chinese art, when artists were told what they could and could not paint. Apparently the themes have stuck. Fan points out the irony of the farming theme given China's rapid urbanization starting in 1985.

Yaoyi Shen Zunyi Conference, 9th, 1999 — **Yaoyi Shen**
*Zunyi Conference,* 9th, 1999

Yaoyi Shen Zunyi Conference, 11th, 2009 — **Yaoyi Shen**
*Zunyi Conference*, 11th, 2009

With artist Yaoyi Shen, we see two paintings that share the same theme, some of the same figures in the same poses, and in this case even the same title, Zunyi Conference. These two winning entries are particularly bold as they are two of the largest paintings to win the contest. One would assume that someone could recall the 1999 version ten years later in 2009 when the second painting won. But then again, perhaps due to the large number of similar paintings submitted each year, they all start to blend together. It would appear that Fan's casual observation that not much has changed in 30 years of the contest is supported by the data and information made available by the team's clever interface.

Their are a great many more insights that can be found via Fan and Sam's AwardPuzzle interface, from geographic trends to a propensity to lighter colors and use of reds and yellows.

Winning Entries seperated by brightness and hue

Winning entries sorted by geography of submission

Sorting artists by number of times they won the contest

Fan and Sam ultimately decided the best way to share their insights was to make the tool publicly available so that others could make these connections on their own and see for themselves what drives the winning entries. You can also watch a demo video they created to explain the interface below. If you are able to identify other trends using their tool, let us know, we'd love to hear them.

Blog