View transcript
- Hi everyone. Thank you for having me. I was very excited to be, as Jen knows, I brought my kids. We're in Boston and I brought my kids to the Exploratorium last summer and they were just like. So I got to be very cool because I was like do you know where I am going to speak? I'm going to give a talk at the Exploratorium. So this was a very very cool invitation. So I've been working with visualization for over a decade and one of the big passions for me in data visualization is the ability to take a lot of very complex information and data and somehow transform it into something that anyone can understand: kids, lay users. You don't need a PhD in statistics or computer science. You just look at stuff and stuff makes sense to you. And I think that's very powerful. But then when I was invited to talk about storytelling, one of the things I realized over the course of my career is that there are different kinds of stories you can tell with data visualization. And one of the insights for me lately has been about real time data and the kinds of stories we can tell with when you're dealing with real time data and how that might be different or not from regular visualization. So I'm gonna start with a small provocation. I'm gonna make a couple of very overreaching claims about data visualization and I'm gonna say that most visualizations show as the past. And the reason for that is that a lot of our data sets tend to be static, right. So they are talking about a world that was. And that's fine and that can be incredibly rich. However, for the first time I think for a while now, for a few years now, we have this ability to work with real time data and I think that changes the game a little bit. So one of the things is that real time data allows us to tell different kinds of stories. And I didn't quite realize that when I first started working with real time data. I was just like okay, we get to visualize this and it changes everyday. It's awesome. But then I started realizing how users take these visualizations and what they do with them. And so this is a little bit of what I want to talk about. I'm gonna talk about two projects. The first one I'm gonna talk about very briefly and then I'm gonna spend more time on the second one. They were very different. The first one is the digital attack map. This is something we did at Google with the Jigsaw people. Jigsaw is the more activist branch of Google and they really care about how can you use technology to address things like censorship worldwide? And so here they came to us and they're like you know, journalists have a big problem talking about things like DDoS attacks, denial of service attacks. These are like cyber crimes. These are things when it used to be just hackers, but now it's entire governments, will flood some server with so many requests for service that they basically bring down entire servers. Why is that a crime? Why does it matter? Why are governments involved in this? Imagine a situation where you have an election going on and you really need to get information out to people and if you have a way of flooding, of attacking servers, people can't get to information, right. So this becomes a way of weaponizing, of using technology for these not so great goals. So the media and journalists want to tell stories about this. The problem is this is something nobody sees. It doesn't have a shape. It doesn't have one villain. It changes all the time. It is so, it happens all the time, but it's kind of like you just go to a website and it says, oh, the website is down. How do you tell this story? So they came to us and they're like is there a way we can visualize what's happening? And this is when we created the digital attack map. So let me try to make this bigger. Okay, so digital attack map is just a way for you to start to see some of these patterns. And this is one day and you can see how different countries are attacking each other. There are different kinds of attack that are happening all the time. In fact, this data set is so incredibly huge that we can only show the top two percent of this attack. If we were to show more than this, you couldn't see the map at all. There are different ways in which you can visualize the same data. So is the source of the port? Is it the type of attack? There are different ways you can look at different days. Different days tend to look very very different. There are things like what you're seeing here in the U.S. where these are attacks that are domestic within a country and then obviously you have international attacks. Again, this is happening all the time and one of the things, so you can come to the map and you can see it in real time. How are we doing today? We're usually doing very very badly. And the thing that you want to do is we created galleries so that you can see different attacks on different parts of the world. And these galleries are really really important because as I said, this visualization is changing minute by minute, right. And so you want to be able, you want to design these visualizations so that you can capture moments in time. It's almost like you are doing a little bookmark, right, so that I can come back to it and point to it. And so, let me go back here. So we designed this thing. Oh, there's one more thing I wanted to show you because I think it's really important in terms of storytelling, which is at the bottom of the map. We have this thing that says, you know, embed the code. And this is specifically for journalists to be able to, or anyone else, but we designed with our audience was journalists, to be able to take this live map with a specific date, right, and be able to embed it in whatever story they were going to tell. They can obviously take screenshots too, but if they want a more interactive version of that, we were designing for this so that they could do that. And so what this does is that it gives journalists a very clear picture, or a somewhat clear picture, of what's happening. And it gives them the ability to story tell around this thing that, again, doesn't have really a shape, right, and it's changing all the time. And so journalists really started talking about this in terms of their censorship. There were wars that happened around the world and you can see the traces that these major events leave at a global scale. So this is one way that you can start to story tell with real time data. So a few special considerations that I think are kind of slightly different is when you're talking about real time visualizations, one of the things you have to think about is how does your visualization scale over time and what happens if the data changes? It could be that, so in the digital attack map when we designed the visualization, we had a certain volume of attacks. I said I could only show two percent of the data. Today that volume has grown a lot. So my visualization tends to be a little bit, even at the two percent, my visualization is very very overwhelmed, right. So as somehow who designs visualizations for real time and long usage, you need to be thinking about these things. How do you scale even your time, your time axis? How many days or months or years? How is that going, or decades? How are you going to scale that, right? Galleries, as I pointed to the gallery, I think galleries are really important. One of the things I really like when I go to museums is the you are here map, right. Helps me place myself. To me, those galleries are the you are here in time. So it's kind of like this is today. This is what I see today. Is today a typical day? I don't know. I have to look at other days to decide how typical or atypical my day is, right. So the gallery gives you very quick access to sort of historical data. Your portability needs may be different. So like I said, because we were trying to empower journalists, we were trying to make it very easy for people to take portions of this visualization and put it somewhere else. And then the other thing that's really sad, but you have to think about, is what happens when the data is not available? What happens if things go down? What happens to your visualization? Because remember, when you have a nice static data set, chances are it's almost always going to be there. But when you have a live stream, you're probably pinging some other website or service, right. So design for failure. How does you visualization fail gracefully, right? Okay, now let's switch gears because I have very little time. I want to talk about wind. This is a project that Martin my, thank you, Martin my colleague and I did off hours. And we were interested in what wind looks like. Wind has been visualized for hundreds of years. More recently whenever you talk about scientific visualization of wind, it's been like vector fields, which are these fields of arrows, right, that try to average out both the speed of wind, but also the direction of the wind. And they work somewhat well, but I look at this, my visual gestalt doesn't do it. I'm like okay where is most of the wind? What's going on? And we wanted to visualize wind in a more dynamic way. So we wanted to put this on the web. We wanted anyone to be able to look at the wind. And so this is the wind map that we created. And this is real time. So this is the wind all over the U.S. And one of the things we here is that we didn't aggregate the data and average the data nearly as much as what you do on a vector field. So we let these, we let your eyes resolve the picture. We gave you more detail and more texture and more messiness, but unintuitively, I think your eyes do a better, my eyes do a better job of resolving that picture than the vector field, right. I have, we hit so many dead ends here. You have no idea. I have a whole other talk about what it took to get here. But anyway, once you have a map and you can play with it, right, oh I always do Boston. I shouldn't do Boston today. I should go closer here. Okay, so it lets you see overall broad patterns, but it also lets you see details. Now remember another thing I said was really important is the gallery, right. So we also have a gallery of wind. And part of the reason we did this, galleries, because we ourselves had never seen images like these before. We didn't know what the shape of the wind was like and this was the reason we did this project is because we wanted to see the shape. In fact, we were so enamored with the shape of the wind that we didn't put any color. We could've. We let go of one of our really precious dimensions in this visualization, which is color because we were so enamored again by the shapes that were coming up. And it changes a lot. Like on one day, a day like this, like Canada is stealing all of our air, right, which is saying a lot because the mountains usually prevent that from happening. And then you have a day like this. Hurricane Isaac making landfall. And this was the first time after we created the wind map that it was live. It was happening during a hurricane making landfall. And immediately we started getting e-mails from people. So someone sent us this e-mail saying I'm here in New Orleans, I'm looking at your visualization, I'm praying that this thing passes and that everybody's okay. And it was a very powerful moment. And to me, it speaks again to the power of real time data. We had never had this experience before where we're all looking at this thing together and we are able to actually interact with the people in the path of this really dangerous situation. It wasn't until we had our own like, this is Sandy, where we were in the path of the Hurricane and being directly affected by it that we also realized how powerful it was. Now think about this. This data is the same as this data. It's exactly the same. And so as I was saying, there were a lot of design decisions here that I think help bring this data to life. But again, back to the real time nature of this data. We started getting these very different kinds of e-mails sent to us. So like farmers look at the wind map before they decide how to spray their crops. Scientists look at the wind map, so both for bird migrations for butterfly migrations. Teachers look at the wind map and when they are talking about forecasts with kids. Pilots look at the wind map. We started getting e-mails from both commercial pilots and military pilots. And we're like whoa, no. No, do not look at the wind map because we make it very clear. The wind map visualizes surface wind, which is zero to 12 meters high. You do not want to look at that wind when you are flying an airplane. And pilots know that, but we were like do we need to even be even more clear about this? So we actually put a disclaimer on the wind map page saying please not use the map or its data to fly a plane, sail a boat, fight wildfires because we kept getting these e-mails right. To which we got one of the following responses. Yeah, yeah, I see your disclaimer and all, but please respect the power of this visualization in promoting the prevention of wildfires, okay. So to me, this is one of the things where I was saying in the beginning, I think real time data changes game a little bit, right. It changes the game because it's not only describing a world that was, it's describing your world today. And so it allows you to take that and to do things with it. We put the wind map up as a art piece. That's what it was in our minds. And then people took it, took ownership of it in many many different ways, which was really interesting to see. The wind map is also used as a memento. So people ask us for prints of the wind map for the day their babies were born or the days they got married or whatever other really important date, right, because again it's kind of unique, right, that power of the day, that moment. The technique has been taken and used all over the world. So before the wind map, no forecast T.V. stations would show the wind the way we do and today they do, right. We didn't obfuscate. We had the ability if we wanted to to obfuscate the code. We didn't. We're like we're just gonna leave it here and whoever wants to use it, use it. And so people did, which we were obviously very happy to see. It also ended up at art exhibits all over the world, right, which I think again speaks about this power of bringing these sensors and these data sets and creating things that allow you to relate to these data sets in a very easy, useful, relatable way. And that's it. Thank you.