Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #51: Kennedy Elliott
141 Plays8 years ago
I’m very excited to welcome Kennedy Elliott to this week’s episode of The PolicyViz Podcast. Kennedy is a member of the graphics department at the Washington Post and part of the team that recently won a Pulitzer Prize for their...
The post Episode #51: Kennedy Elliott appeared first on PolicyViz.
Recommended
Transcript
Introduction and Program Promotion
00:00:00
Speaker
This episode of the PolicyViz podcast is brought to you by the Maryland Institute College of Art. MICA's professional graduate program in information visualization trains designers and analysts to translate data into compelling visual narratives, benefit from the resources of a premier College of Art and Design while learning online.
00:00:20
Speaker
Earn your information visualization degree in just 15 months. Expert faculty includes Andy Kirk, John Schwabisch, Marissa Peacock, and Rob Rolleston. Learn more at mica.edu slash MPSInvis.
00:00:46
Speaker
Welcome back to the Policy Viz Podcast. I'm your host, John Schwabisch. I'm joined this week with Kennedy Elliott from the Washington Post, graphics editor galore, expertise. Hi, welcome to the show. Hi, thanks for having me. I'm excited to see you. Thanks, I'm excited to be here. At the new Washington Post building.
00:01:05
Speaker
I'm excited that you're here with me. I know, but we're after Washington Post building. It's quite lovely. Yeah, it's brand new. Brand new.
Pulitzer-Winning Project: Fatal Force
00:01:11
Speaker
So we have lots to talk about. Let's talk about the latest greatest. Well, maybe it's not the latest greatest project, but the latest greatest award, now called Fatal Force, the police investigations project you and many others worked on. Many others, yeah. But winner of a recent Pulitzer.
00:01:29
Speaker
Yes, it was very exciting. It's actually the third project I've worked on since I've been at The Post and has gotten a Pulitzer, which is incredible. But this one was definitely I had the most much more involvement in, which was really great. But there's a lot of people involved. I was just talking about this earlier today, but it's just a big collaborative project, obviously. And it took a long, long time to collect the data. And I honestly had no part in
00:01:55
Speaker
and collecting or calling police departments, or I can't take any credit for that, but I definitely helped with the development of the database, the front-end work and the front-facing. So for those who haven't, may not have seen it. Can you sort of describe the project itself and then talk about maybe a little bit of how it was pulled together? Because there's lots of different aspects to it. I would assume crossing lots of different departments of the paper.
00:02:17
Speaker
Yeah, well, it was mostly our national desk working on it. We looked into, astoundingly, in the U.S., we don't have any central tracking of when police shoot and kill civilians. And Wes Lowry, apparently, the end of 2014, I think, had an idea, or maybe the beginning of 2015 had the idea to track every single shooting that happened in America.
00:02:42
Speaker
And it just kind of took flight and we are fortunate to be at the Washington Post to have the resources to have researchers diligently track all this information. And a gentleman on my team who is very, very smart and great built a database for all the, an internal database for all their tracking, John Myskins.
00:03:03
Speaker
It was really helpful because we ended up calling police departments three and four and five times trying to get to the bottom of these details.
Role of Journalism in Data Collection and Impact
00:03:12
Speaker
You need to know how many times you've bugged. There's all kinds of metadata that is needed to even start tracking this data. We actually collected quite more than we ended up publishing just because
00:03:26
Speaker
We want to make sure we're representing everything correctly so the database itself is quite large yeah we track things like a person who died the race their age their gender this certain situational properties of the events like if they're running away or if they had a gun or fair you know what weapon they were armed with if any.
00:03:46
Speaker
All kinds of things like that and we build a big database and that database was great. We had a series of stories that ran throughout the year, obviously, but the database helped inform the stories moving forward, which is really powerful. I mean, it's, you know, hats off to our amazing researchers who are collecting and cleaning and making sure the state of was as accurate as possible talking to families.
00:04:07
Speaker
and police departments and all of that. But it's just really interesting as it grows. You're not even sure what kind of patterns are going to evolve through the year and that database helps inform that insight, which is cool. It's a living thing. Right. Still is. So how do researchers or reporters at the post use
00:04:27
Speaker
So is it within an internal tool or database that they can go and extract and then do stories off of that? Or do they talk to the researchers? Do they talk to you if they need to use it? What's that workflow? Now that it's sort of existing and living, how does it work from there?
00:04:43
Speaker
Yeah, like I said, our internal database is a little bit more robust and we are tracking more information in 2016, which will be exciting as that progresses. There are quite a few people that have access to the internal database and obviously log data and make changes and parse things and etc.
00:05:03
Speaker
They're always welcome to use the front-facing database as well, which has the visualization component. That's improved a little bit this year. We have a new one for 2016. They can slice it anywhere they want and pursue any sort of investigation into the data that they want. They can also have the power to publish the data. Certain people will have the power to publish when new information comes in and they get it to a point where they're happy with.
00:05:28
Speaker
We limit that to a few people that are handling the data. Anyone in the newsroom can access the site at any time. Does it feel weird to you that
00:05:43
Speaker
The Washington Post is building a database that maybe the government should be collecting and putting out. Does that strike you as odd? Does it feel odd or do you feel like, well, it's just a gap that someone needs to fill? Yeah, that is exactly why journalism exists. I think that's exactly why we're here. It's really fun to be a part of something that
00:06:06
Speaker
And I hate to be cliche, but that affects change. The FBI has said that they're going to start tracking this data in 2017, which is incredible to be the reason why that sort of change is happening. Who knows if we'll be continuing in 2017 just for comparison purposes. But it's really incredible. There's a lot of journalism out there that affects that kind of change.
00:06:33
Speaker
You don't usually work on stuff like that, but it is really cool to do that. And I don't think it's awkward at all. I think that's why journalism exists and that's part of what we can do to help.
00:06:47
Speaker
So let's switch gears a little bit because we're talking about research here at The Post and you recently gave a talk at OpenVizConf on research in the data visualization field. Yeah.
Academic Research and Data Visualization
00:06:59
Speaker
So 39 researcher papers in 30 minutes. Yeah. A tour de force of the field. Yeah. So I don't want to ask you to lay down the rules here for listeners.
00:07:11
Speaker
How did it feel to sort of – I mean, did it feel like you were sort of taking yourself out of your comfort zone and diving into this academic literature? What was that experience like? Yeah, it was really uncomfortable because for me, I just – I mean, I kind of laid it out in the beginning there, but we were practitioners. I've suddenly kind of become aware of how much in the academic world exists outside of my
00:07:36
Speaker
You know, realm of knowledge and I'm so interested in it. I've been interested in it for years, but I don't really understand how expansive it is and how many wonderful things people are doing because I feel like our spheres don't really collide that often. Unfortunately, I just am really interested to see what other people have come up with that do completely different things that I do.
00:07:55
Speaker
But it's really uncomfortable reading papers and I don't have that kind of background. I think if my life had taken a dramatically different twist like really early on, I might be in that kind of environment. I've always been kind of interested in academia. So it is really awkward and I don't know the things that they know.
00:08:12
Speaker
And it's hard for me to evaluate their work, but I am interested in kind of reading how they fashioned their experiments and, um, the kinds of things that they tested for and the kinds of things that they're interested in, in pursuing in research and the kind of results that they came up with. So that was sort of my basic premise. Like I just wanted to relay what I read it rather than saying like, Oh, this says this, let's, I'm going to prescribe that rule. Right. Yeah. Yeah. Right.
00:08:38
Speaker
It was just sort of like an investigation into and like a lot of the studies were from so many years ago. I had a couple from like one from the two from the 20s and one from the 30s and 80s and 90s and you know, etc. And you just wonder how
00:08:55
Speaker
how visual literacy changes. Even today, academic papers can take so long to publish, you wonder what kinds of things people are now familiar with that just might not be the case when you started writing the paper. Yeah, absolutely. Did you have a favorite over the third line?
00:09:16
Speaker
I think there are 37. I think there are 37. But yeah, I had a couple of ones that I really liked for odd reasons or two that stood out to me. One of them was this one on line drawings and they're testing computer algorithms for, you know, if like the computer algorithm took to draw this like, you know, random object was as good as like an artistic like a human drawing. And that doesn't pertain to my job at all. I don't do any sketching. I like literally have never had a use case for this kind of thing ever.
00:09:45
Speaker
But for some reason, it was just delightful to me because of the way that they tested. I thought it was just really kind of novel. I started reading the paper. I was like, how the heck are they going to test this? And then they tested for screwdriver type of tools and weird kind of tool-y things that I'm not even that familiar with myself. But they put these little gauges on the drawing.
00:10:04
Speaker
And they got the mechanical Turk participants to rotate the gauge to kind of like match the angle of its dimension, like how they perceived where it laid on that object. I thought that was really funny. I don't know why I like that so much, but I did. And then there's another one by I think John Hare and some of his, you know, several other co-authors.
00:10:28
Speaker
And they're testing for another computer algorithm for assigning semantically resonant colors to different keywords. That was the very last one I presented. And I just thought it was kind of funny because their algorithm was pretty straightforward. It scraped Google images and I'm sure it had some sort of preference that it gave to certain colors in the Google images that I retrieved based on a keyword.
00:10:49
Speaker
I feel like that can open the door for some really cool things and it can also open the door for some really funny computer type errors. They eventually bet against their algorithm. They had a human person devise a palette that was also semantically resonant and they tested it against that one.
00:11:09
Speaker
And it's just kind of funny when you go through all that trouble to devise an algorithm and then you're like, yeah, but humans are still probably better. I don't know. It's just really cool. So now that you've read all these and probably have talked to a number of different academics, do you see a path by which academics and practitioners can
Challenges and Collaboration in Visualization
00:11:30
Speaker
or should, I mean, I think they should talk. I mean, I think we both agree that they should talk more, but do you see a path by which they can talk more?
00:11:37
Speaker
ways to inscribe incentives for one group or another to say, hey, we need this thing. You should be doing this. Or is it just sort of like this is the status quo, the academics are over there, the practitioners are over there? No way. I mean, I hate to just say that, but it's tough. I mean, I still feel largely ignorant of the needs and
00:11:57
Speaker
Limitations of the academic community and I'm sure they feel the same way about people like me And so I don't feel like I fully kind of grasp, you know, they're kind of culture environment And that's really fine. I just I'm not I don't feel like an expert at this point But I really I mean, I think conferences are a great way to kind of bridge that gap. I mean everyone is so
00:12:18
Speaker
I'm excited to share information, especially at places like OpenViz. I feel like everyone is just so happy to be in the same room together. It's such a great feeling. That really helps a lot. I don't honestly know what the best way to do it is. I think part of it is just me working in a bubble and I need to get out more when people publish things on blogs.
00:12:42
Speaker
Yeah, but clearly it's not just you, right? It's like everybody's in their own little bubble. Yeah, it's I mean, it's easy, you know, it's like working in a newsroom is hard. It's very driven by deadlines and stressful. And I'm sure academia has its own set of stresses and
00:12:57
Speaker
just easier to talk to your peers that understand your jargon and where you're coming from. But I guess I haven't really thought of what a good solution would be, but definitely just any opportunity for crossover. Yeah, any, any. Right. Good. So I want to switch gears one last time because we have, we've talked about interactivity in the past, so I want to talk about it again.
Interactivity and Storytelling in Data
00:13:18
Speaker
So what are your current sort of feelings on interactivity in data visualization?
00:13:24
Speaker
Is it always good? Are there limits? Is the field doing a good job with interactivity? Are we dropping the ball? When I say we, I don't really know what we mean, but okay. Is the field dropping the ball on the mobile side? Is the virtual reality, do you see virtual reality? I guess the one thing I've talked about with several previous guests is whether the gratuitous interactivity is worthwhile.
00:13:50
Speaker
You click on a bar and the data number pops up. Is that worthwhile? So do you have sort of thoughts on like where you end up putting your activity when you do your stories and where it's useful and where it's not?
00:14:01
Speaker
Yeah. Well, I mean, we definitely don't want to make the reader do additional work to gather information in a graphic. And I know that, you know, R.T. Say just, you know, had his came up with this great kind of mantra philosophy about not having interactivity unless it's absolutely necessary, which I think everyone would definitely agree with. And it's nice to have someone kind of establish that openly.
00:14:24
Speaker
I think in general, and this has been the case for the last several years, so this is nothing novel, but we're moving away from publishing these big databases where readers kind of interact with them and have to kind of extract what the most important things are to a place where we're kind of laying out what they should be getting specifically, just in maybe like little snapshots of the database.
00:14:48
Speaker
I think that when tools like D3 and how much JavaScript has really become an essential tool within the last few years in the newsroom, we get so excited that we have these new tools and we can do all these cool things and then we're like, okay, how do people really want to see this information?
00:15:04
Speaker
How do you tell a story and all that? And so that's kind of a natural, kind of a natural course. SC tools being used like D3 a lot more for information retrieval. And so we, we tend to work really hard cleaning data and making sense of it in the background and doing like some just like
00:15:25
Speaker
throwing up crazy visualizations with the data and just trying to see what the data is showing. And then maybe for publication, we'll tone it down. Oh, I see. Okay. So it's kind of interesting. We put a lot of energy into just trying to test it, see how lasting it is, see where the fun spots are, what the interesting spots are. And then for publication, we'll have something much simpler and you would never ever know that we had some crazy
00:15:51
Speaker
Right, tone it all the way back down. Yeah, just so that we're relying on that wow factor or we're just trying to tell a really straightforward story that people are going to understand. So I definitely don't think interactivity is necessary. I think that a lot of these tools that can include interactivity like JavaScript and D3 and all of that, we use those mostly just to render shapes in the browser.
00:16:13
Speaker
That's a really powerful thing that D3 has given us and to do animation and that kind of thing. But we don't want to make the reader work too hard to get the information, obviously. So in a project like Fatal Force, so you're publishing this huge database.
00:16:27
Speaker
And so when you start working with the reporters, are you trying to guide them through how to like pick out the story so that it's not giving this just, here's a big database to the reader, but like, we're going to pull out some interesting stories. Are you helping other folks guiding them through the data and the tools that you're creating maybe internally and then whatever comes out later sort of stripped down?
00:16:47
Speaker
Yeah, sure. I mean, I think with Fatal Force, it worked a little bit in the other way around. I think people were really familiar with the data and they were helping me understand it more so that I could make recommendations on how to visualize it because
00:17:01
Speaker
I kind of came into that relatively late in terms of the group towards mid last year. So people had already been kind of hunkered down working on it for a while. But yes, if I'm kind of owning a data project from the beginning and I feel like I've been making reporting calls or I feel more familiar, like we have two computer assistant reporters on the team.
00:17:23
Speaker
Dana Ted, they're both really, really, really smart people. And they often are the ones that go really deep into the data. And they are the ones informing reporters like, you know, you can't parse this data this way, you know, it needs to be, you need to have, you know, different baseline or they just, they get really deep into it. And I definitely do that many times in several projects as well.
00:17:45
Speaker
because we're just trying to strive to tell an accurate story. Interesting. Let me ask you one last question then on storytelling. How important is it when you are working with the data, you're creating the visualizations, you're creating databases, how important is it to get the individual stories, to get the reader to connect with a person or a place or a thing, a noun? How important is that when you're thinking about a story and developing it?
00:18:12
Speaker
Very important. I think a lot of times that's why you'll see a lot of interactive saying like, you know, enter your zip code in and find this, you know, what matters to you. And we've gotten really good at writing robo techs customizable, but still scalable to people's locations or.
00:18:28
Speaker
demographics or whatever. So that's definitely, you know, when you feel like you have the opportunity to do that and do it well, and not just superficially, I think that really can make a difference. I, you know, I'm not sure if that's the, I'm not sure if that's the best way to do it, or if that's the only way, but it is definitely a case by case scenario, you can't do that with every single data set, obviously.
00:18:48
Speaker
But that's why journalism, some journalists are just amazing. They are able to connect in that way. Eli Sazlu is a series on food stamps and food hunger and poverty was so wonderful because he really told those stories from the heart. And so many other journalists are like that as well. I think that's timeless.
00:19:10
Speaker
It's interesting because you and journalists spend a lot of time trying to tell those stories. And you've done some work now sort of diving into what the academics do.
Cross-Benefits of Academia and Journalism
00:19:19
Speaker
And from my perspective, like researchers and academics who are doing data visualization could benefit really from talking to actual people and telling their stories. And so it's like you said, trying to get people to pop their bubbles a little bit and figure out all these different things. Interesting.
00:19:34
Speaker
Great. Well, thanks for coming on. This was fun. No, absolutely. It's super fun. And congrats on everything, right? National awards, big talks at conferences. Yeah. Appreciate it. Great. Thanks for coming on and thanks to everyone for listening.
Conclusion and Thanks
00:19:49
Speaker
Of course, if you have comments or suggestions, please let me know on the site or on Twitter or please review the show on your favorite podcast provider. So until next time, this has been the Policy Vis Podcast. Thanks for listening.
00:20:11
Speaker
This episode of the PolicyViz podcast is brought to you by the Maryland Institute College of Art. MICA's professional graduate program in information visualization trains designers and analysts to translate data into compelling visual narratives, benefit from the resources of a premier College of Art and Design while learning online.
00:20:31
Speaker
Earn your information visualization degree in just 15 months. Expert faculty includes Andy Kirk, John Schwabisch, Marissa Peacock, and Rob Rolleston. Learn more at mica.edu slash MPSInvis.