For the last three years, I have been gathering art data to build the first database of all known works across the greatest artists of the twentieth century. Until the last few months, very few folks were aware of the Artnome project. Now, thanks to Oliver Roeder and FiveThirtyEight, thousands of people know about Artnome. In just a few short weeks, over 100 artists, art historians, data scientists, and economists have joined the Artnome community. Additionally, I have heard from professors, auction houses, collectors, and others looking to leverage the database or develop a partnership.
It's been great getting all this support and fascinating to see so much variety in the background and interests of the folks that show interest in the project. In conversations with my new friends, some of the same questions have been popping up, so I decided I would right a short FAQ answering the questions and taking the opportunity to again thank Ollie and the great folks at FiveThirtyEight.
Q: What is your end goal?
A: I started the project without a specific end goal. I was just very disappointed in how difficult it was to get information about the complete works of artists. Even as an art history student, we only looked at individual works or small clusters of works (Picasso's blue period, Monet's Haystacks, etc). I felt like I was only ever seeing the leaves and I was really curious to see the whole tree. We have extensive and easy-to-access online databases for Beanie Babies and Pez Dispensers, but no public database listing complete works across our most important artists. My hope was that by aggregating data for 40 artists, people would see that this is a solvable problem. However, a project like this can only scale when many people are involved. To facilitate this, I am building tools to make it easier for others to annotate the entries for each work. Which brings me to the next question.
Q: What value are you adding?
A: Aggregating data locked in a printed format and making it digital is in itself valuable. Once digital, we can perform all sorts of analysis that was previously impossible and discover new things about our most important art and artists. But digitization is just a starting point. My goal is to combine data across many sources to form a go-to digital record for the complete known works of every artist. The same way decades of boxscore data serves as the foundation for analysis of baseball, my hope is that the Artnome records will serve as a foundation for analysis of art and artists.
I generally divide the art data I am after into two groups: data that needs to come from an expert, and data that can be crowdsourced. Data, like dimensions, materials, and authenticity, need to come from authoritative sources: CR (catalogue raisonne), auction data, and museum records. I started with CRs, as they are key to attribution, establishing which works scholars agree are by an artist and should therefore be included in the database. However, for information on dimensions and materials, I prefer auction records and museum records, as they have prolonged direct contact with the work (not always the case for the authors of CRs). I think most people would be surprised how often CR data and auction data for the same work don't match on key areas like titles, materials, and dimensions.
Other areas, like research and provenance, are dynamic and lend themselves better to crowdsourcing. The provenance listed in a printed CR is out of date shortly after it is printed because artworks continue to change hands over time. The same is true for provenance taken from auction records, which also do not evolve as works change hands. This is where a Wikipedia-style community of folks crowdsourcing the continuous updating of records for artworks as they change hands is more effective than a single person tracking the provenance of all works in a printed book. Similarly, bibliographic information is dynamic and not static. Linking back to articles, books, and research that mention specific works is best crowdsourced by a community.
In summary, the value I hope to add is:
Aggregating the most reliable data across many different sources
Unifying the data into individual master records for each work
Digitizing the information so it is searchable and can be used for analysis
Creating a macro view of each artists' complete works
Facilitating the crowdsourcing of a continuously updated provenance and bibliography by building custom annotation software
Q: How are you going to make money off of this?
A: To me, worrying about monetization at this stage is a bit like worrying about what to name your band before you have learned to play an instrument. It is fun to fantasize about, but it is mostly just a distraction from the insane amount of work needed to make this project a success. That said, I have invested thousands of dollars and hours in Artnome and will need to think about ways to fund the project in order to scale it up. I believe many people would love to have a single database across many artists listing all known works. If I am right and can show a clear vision for building such a resource, and that I have the level of commitment needed to do so, I believe funding in some form or another should follow. I have already had several universities offer assistance and individual collectors offer access to CR and museum catalogues, which reduces the cost of sourcing this material myself.
Q: How did you get FiveThirtyEight to write about your project?
A: In 2007 I had another analytics project I started with my younger brother called Hoopism that focused on basketball data. Among other things, we created an interactive data visualization that showed the score for every dunk from every slam dunk contest, as as well as the corresponding video of the dunk. The tool went viral and we had a quarter million visitors within a month or two of starting the site. Youtube reached out to us to let us know they were big fans of our project and it actually registered as a nontrivial spike in traffic for them.
That experience gave us confidence that if you spend a lot of time and energy on a project you are genuinely passionate about, people will be interested in writing/reading about it. Around that same time we were doing a lot of networking with other sports analytics folks both online and at conferences like the MIT Sloan Sports Analytics Conference. The best thing to come of that was meeting Neil Paine, whom my brother and I have a competing man crush on. Neil is incredibly bright and genuine, and hanging with him at Sloan was always an annual highlight. At the time, Neil worked for the amazing site basketball-reference.com which is in no small part an inspiration for what Artnome could eventually develop into. He has since done a stint as a scout for the Atlanta Hawks and is now a Senior Writer at FiveThirtyEight.
After I wrote my first Artnome blogpost on winning 2nd place in the first fantasy art auction for cash, I joked with Neil and my brother Matt that I should offer my fantasy art services as an analyst to FiveThirtyEight. They both pointed out that Oliver Roeder had been doing brilliant writing on art and data and that my project aligned well with the FiveThirtyEight audience. Neil offered to submit it to the FiveThirtyEight Slack channel, but noticed it was already there. It turns out that my brother had already sent the link to several folks.
So that is a long way of saying that I did little to nothing to get the article; I just have really nice, supportive friends. From that point, it was really Ollie who championed the project, scheduled the interviews, and did all the heavy lifting of writing the article.
Q: How can I help?
A: I have been pleasantly surprised by all the folks offering to help in any way they can. It makes me that much more excited about the potential of the project. Currently, the best way to stay connected to the Artnome project is to join the community. I will be using the community to get feedback on article ideas and beta versions of tools, and eventually hope to leverage their help to expand the database. If you have something more specific in mind, I can always be reached at firstname.lastname@example.org.