These are the three elements that are the foundation to exciting data visualization. Design isn’t just about polishing up data. Look no further than Apple, Inc., to see how foundational design can be in an organization.
On the other hand, data scientists must also align analysis with art. Artists ask the important questions business executives may not have considered. Plus, data artists excel at breaking down boundaries in creativity. No one understands this better than Jer Thorp, Co-Founder of the Office of Creative Research and Former Data Artist in Residence at The New York Times.
HH: What does it mean to be a data artist?
JR: There are two rough reasons why people call themselves a data artist. One of those reasons is people who are doing things kind of outside what we think of when we think of data science or when we think of statistics and maybe combining design to do data visualization and then there’s the second group of people who are working with the data as part of a larger art practice. Myself, I think I straddle both sides of that border. So, I run the office called the Office of Data Research where we do a lot of data focused R&D work for companies like Microsoft, Samsung and Intel. We do a lot of design work. We do a lot of data visualization tools to try to solve really weird and interesting data problems and then the other side of it is we also have an art practice where we build often physical, sculptures that can be built in physical spaces and museums and galleries. We just finished a long artist in residency at the Museum of Modern Art in Manhattan and we’re just starting a gigantic project to install a piece in the Boston Public Library. So, for us – and for me – the term data art is really wide-ranging.
Let’s talk about the “weird and interesting” part of data. How can the human side of data lead to innovation and effective change within an organization?
I use the term “humanizing data” a lot. In some ways it’s kind of a given. Data doesn’t exist without humans and if we think of data as the measurement of something, that act of measurement is by default a human measurement. We have machines that are doing the measurement as a proxy for us, but at the root of it, data is really a human thing. We’re producing it. I think where it really gets problematic is when we’re talking about data that is a measurement of humans. Even something as simple as location data or survey data that comes from customers or whatever the case it may be, there is something there that becomes ethically interesting and ethically complicated because we do need to consider the humans that are the systems from which that data is being generated. In a business sense, I think this is a challenge and an opportunity.
It’s a challenge because we can do things with this data very easily. There’s no permission form. I don’t even have to talk to these people. I can just use their data and away I go on my merry way. The problem with that is that if I cross a boundary that is either uncomfortable or negative in some way towards these people then I can breach their trust. That is a thing you definitely don’t want to be doing as a business because it can lead to a situation in which your consumer base loses trust in you. That’s why I also say data presents an opportunity because I think there are precious few companies right now who are seeing this as a chance to set themselves out from the crowd and say, ‘hey, unlike all these other organizations out there, we’re going to be fair with your data, we’re going to be transparent with your data and we’re going to use your data in a way that will make sure that you still trust us.’
I think we’ve been lucky in the last 85 years because we’ve been able to get away with a lot of things with consumer data without consumers being aware of it and now I think that’s changing. And, so these companies –hopefully there will be more and more who are putting the right foot forward and saying we want to be an ethical company – are going to have a tremendous advantage.
Research is the most important word. It really comes down to innovation at a true level. I fundamentally believe that you can’t have innovation without a certain amount of risk and without a certain amount of limitation. So, for lack of a better term, creative data exploration – or data art – provides an avenue to experiment and to try new things that otherwise wouldn’t be tried in the everyday course of data analysis or data science.
I believe there’s huge value in trying things. I have this phrase I use when I describe our work, which I call “question farming.” In a lot of cases what we’re doing is confirming the suspicions that we already had or trying to find answers. We hear a lot about using data to find answers. But, I think it’s just as important to use data to find questions. There are a lot of questions that we don’t even know how to ask yet. It’s those questions that lead to true innovation, when you get to say, “hey, we never thought about it this way, but what would happen if we try this new thing?” and it’s the nature of those questions that are not just available sitting in your bathtub. You need to try new things and you need to prototype and you need to take risks.
It’s not a particular surprise that if we look at the history of successful companies over the last 100 years, largely those are companies with strong research and development groups and I think that there are a lot of companies who we talk to who have trepidation about investing in R&D because there’s so much risk involved. But, you have to take a chance that it isn’t going to work. You have to be willing to understand it’s a risky investment.
Many analysts have to prove the quality and integrity of their data to their high level executives. But, they’re working in such a siloed structure that it’s kind of a struggle to present the data in a meaningful way that’s readable, but not based solely on the human factor.
What are your tips for bridging that gap and secure executive buy-in?
First of all, data and integrity go together hand-in-hand. With the data that we work with, even though it skews to a more creative axis, data integrity is fundamental to our work. We are always very careful to make sure that the data that we’re using is sound; and that we’re representing it in the right way, that we’re aware of its biases, that we’re aware of its errors, that we’re aware of its missing data and so on, and so on. Part of the answer to the question is to be honest about those types of things. There’s no such thing as perfect data. One of the things that I always found that instills a little bit of trust in data visualization or a data presentation is if there’s some honesty about those types of issues.
I have a principle I sit on when I’m doing data visualization, which I call the “Ooo-Ahh” principle, which means that a good data visualization should do two things at the same time: The first thing it should do is capture people’s attention. That’s like the “Ooo” moment. The second thing it should do is teach somebody something, which is the “Ahh” moment. When you’re presenting data to the CEO or whomever your stakeholder is, there’s a balance that has to be achieved because you want to invest enough in the “Ooo” that they’re not just going to skip over the figure or the chart that is the “Ahh” moment. You want to show them something that’s going to be engaging. But, you can’t do that in sacrifice of “Ahh.”
The reason why there tends to be conflict between data and aesthetic is that the mistake is to sacrifice clarity for aesthetic. That doesn’t have to happen. You can have your cake and eat it, too. So, we can have a data visualization that carries all the information that we want, but it adds some visual flavor and design treatment, which makes it so that it’s more memorable, it’s more attractive, it’s more readable.
In previous talks, you talk about the lack of dialogue between three elements of data; Science, Art and Design. Can you tell us a little bit about how organizations can address this issue?
Multidisciplinary data is something I truly believe in. Envision those three circles in the Venn diagram – the more overlap there are the more productive the result is going to be. Starting with the boundary between data science and design: I was having a conversation with a large organization the other day who was talking about how great it was that they just finished a project, it was all working, and then they brought it to their design team and they were amazed by what the design team was able to do in a couple of hours with this thing to make it better. I turned around to them and said, “If you think that’s good, imagine if you had decided to work with the design team from Day One.” What if this project had design as one of its elements? You probably would have ended up with something of inordinate magnitude than you actually did. I think one of the misconceptions that people from the data world have about design is that design is just about making things look good. But, design is a lot more than that. Design is a way of thinking and by bringing designers into the process early is that the results will get better and better and better. We just have to look at Apple to understand how foundational design can be in an organization. That’s one of the reasons why they’re so successful is that design is baked into the organization from the ground up.
On the other end of the Venn diagram is the art. I’ve been a huge advocate of recommending companies do artist residencies, which I think are such an incredible opportunity for everybody involved. Bring an artist in for six months, set them up at a desk and they will come in and work with your data and your employees and make something incredible. One of the things that the artists are really good at is asking questions that you may not have thought of and they’re all so good at breaking down boundaries in creativity.
This is a thing that has a deep history and it has a deep history that really works and it’s not just something to do for fun. I’ve been spending a lot of time over the past year with this woman named Lillian Schwartz. She was the artist in residence at Bell Labs for 35 years between 1960 and the early 1990s and a lot of the transformative things that came out of Bell Labs at the time were little pieces that Lillian had assisted in, starting with answering questions and also doing some real work on these projects as well.
Jer, it’s very clear that you love what you do, or at least you very much enjoy it. But, do you ever struggle or suffer from data fatigue? What kind of tips would you provide analysts who are struggling with this and maybe aren’t able to discover new questions or answers to the data that they’re processing and reading?
There are a couple of answers to this question. For individuals who are working with data and for executives who are working with teams of people working with data is to find ways to continually make the data fun. What it might be about is that every second Friday of the month there’s a Hack Day, in which you bring in a totally new data set that maybe isn’t related to a current project. It could be one of your teammate’s location data. It could be something pulled from the Internet or your email history from the last two years. The idea is to get into data to exercise their muscles that have been lying dormant for a little bit.
At the center, we’re always trying to find these small tasks and sets of data to use as small experiments to get our minds off the “real job.” That has been really effective for us. It’s important to remember that data analysis isn’t all about sitting in front of your computer. Any data set has a lot of real grounding in the real world. There are resources you can read and things that you can watch to be more informed of where the data came from and that will allow you to do a better job of visualizing it or analyzing it or whatever it is you might be doing.
I should take a moment to shout out to one of my best friends who runs DataKind, which tries to take really great scientists and pair them with non-governmental organizations that have data problems. DataKind will plan a Hack Day where they’ll bring in data – maybe it’s from a cancer organization or a company that’s doing irrigation work in Africa and is need of data scientists to read the data for them, but they don’t have the money to pay them. So, it’s a nice way to tackle a really hard problem, promote team building and also do some good.