 Today I'm going to talk about some data visualization, which was like my first love in terms of things that I shared on the internet. When I was a PhD student I started a blog called ifweassume.com. One of the things that drew me into science as a career was this love of data visualization, artistic appreciation for what numbers and science and data looked like. One of my favorite visualizations came out of a data set that the Tate Modern, an art museum in London, published, which was 65,000 pieces of art. It was the metadata for all the art in their collection, so who the author was, when it was acquired, when it was created, what it was made out of, and I got really fascinated by the dimensions of the art itself. So I made this graph, which I guess is a little complicated. It shows a density map. On the x-axis we have the width of the piece, and on the y-axis we have the ratio of the height to the width. Really tall pieces, compared to how wide they are, are up top, and really squat pieces are on the bottom. So the color of the pixel represents how many points, how many pieces of art fall within this space. This horizontal line in the middle, those are square pieces. And I loved this figure because there was so much structure and detail. I thought this was a fascinating graph, and to me it told an interesting story about the sizes of art and the choices that people are making. And when I published this it went nowhere. It didn't get any traction, nobody seemed to like it. So I made a new visualization, which was much more artistic. Instead of showing the actual data themselves in some sort of really quantitative, scientific presentation, I just made this wire diagram a bunch of boxes that represent the width and the height. This is all the pieces up to, I think, about three by three meters. My favorite part is the box that's clearly portraits, that's clearly paintings and things that are a meter by two meter or something like that portrait. I would argue this is a terrible graph. This might be beautiful data in its representation, but it's a terrible graph. You can't measure anything from this. You can't make any quantitative reasoning based on this presentation. Instead it's, this is just art. And what I love about this is this is art, which has become data, which has then gone back to being art. This is what I love about the data is beautiful subreddit. Subreddit is a complex place, but data is beautiful has consistently been one of the best places, I think, on the internet for appreciating quantitative reasoning and science and numbers and has helped push graphs and charts and maps into the common lexicon. My feeling is that that subreddit has had an outsize impact on our cultural appreciation for data. And this experiment with the Tate Modern Gallery's metadata made an interesting connection that data and art can be somewhat indistinguishable. So a game that I like to play in some of my talks that we'll play here today is art or data. Can you tell the difference? So the way this works is I'm going to show you two images and you just decide A or B, which one is data, and which one is art. All right. First we have these two sort of wire frame looking things. They're both really messy jumbles of wires, black lines on white background. What do you think? Okay. The answer here is B. B is a trace of somebody's mouse motions when they're using a coding editor. A is an art installation, which is supposed to be a giant representation of a spider web. All right. Let's move on to the second one. Here we go. Here's two very sort of abstract bar chart looking things. What do you think? A or B? Which one is the data? So the answer is in this case A. A is a data. This is actually a visualization I found on data is beautiful, which took all the references to color in Alice in Wonderland and scaled these bars to all the different colors. And B is a painting by the American painter Ellsworth Kelly. All right. A couple more. A or B? Which one is data? I think this one's really hard. In this case, A is data. It's actually a representation of a graph that I made. This tile of colors represents the position of bestseller books over time. Maybe I'll do a video on this visualization soon. This was a really cool project. B is actually again another painting by Ellsworth Kelly. He's got a series of these. They're really fascinating looking. And the last one maybe is a little easier, but I really like it. B is the data here. This is actually a plot from my friend Ethan Krusey's PhD thesis. And A, again, actually is Ellsworth Kelly. Now this might seem like a big love fest for Ellsworth Kelly. He died a couple years ago. I think the very sort of quantitative feel of his work is really compelling. But I think the point here is that sometimes data looks like art. And that's part of our appreciation for it. That's part of what I think has caused data is beautiful to have millions of subscribers and for graphs and maps to become really part of our conversation about what's going on in politics and the world. A lot of scientists push back on this notion that they have to worry about their graphs being beautiful, that the presentation is the most important thing and that the science and the numbers themselves aren't the most important result. And I sympathize with that. We want the science and the results to speak for themselves, but they don't speak for themselves. We speak for the results. If our writing is unclear or if our graphs are confusing, people aren't taking the message away. As a scientist, I have to make graphs and charts that help make measurements that help convey specific results and meaning. But often the presentation is in a conference talk or a video or maybe on a poster, but we're probably not trying to convey a specific quantitative result. And instead really we're trying to tell a story. I would argue that the data visualization we do as scientists is not dissimilar from the data journalism that's going on or the data appreciation that happens in the data is beautiful subreddit. The reason to make a beautiful graph is to tell your story. And whether you like it or not, the public and your colleagues as a scientist are going to resonate towards figures that are easy to understand and easy to look at. So there you go, art or data, maybe it's both.