Casey Miller is a data and graphics journalist at the Los Angeles Times. She enjoys turning data sets into easily understandable, unique, and engaging experiences for users.

Before the Los Angeles Times, Casey created data driven narratives at Mapbox and was an engineer at Vox Media in New York where she worked with various editorial teams to tell one-of-a-kind stories across a variety of platforms. Previous to Vox, she worked with teams at the Wall Street Journal, Los Angeles Times, and Atlanta Journal-Constitution. Casey graduated from UNC Chapel Hill in May 2014 where she earned a bachelor’s degree in Journalism and Mass Communication.

In this week’s episode of the podcast, we talk about Casey’s work at the LA Times, her background coming from Mapbox, and considerations when making data-driven maps.

The PolicyViz Podcast is now on YouTube! More past episodes being added soon. Subscribe to the channel to get more podcasts, tutorials, and more data visualization content!

Episode Notes

Casey on Twitter | Personal website

Live Wildfires Map

Los Angeles Times




Ben Welsh at the LA Times



Related Episodes

Episode #110: Ben Welsh

Support the Show

This show is completely listener-supported. There are no ads on the show notes page or in the audio. If you would like to financially support the show, please check out my Patreon page, where just for a few bucks a month, you can get a sneak peek at guests, grab stickers, or even a podcast mug. Your support helps me cover audio editing services, transcription services, and more. You can also support the show by sharing it with others and reviewing it on iTunes or your favorite podcast provider.


Welcome back to the PolicyViz podcast. I am your host, Jon Schwabish. I hope you are well, healthy and safe. I also hope you’ve been able to check out some of the content coming out on my website I’ve had a bunch of new blog posts go out, some new videos around creating visualizations in Excel. And I’ve also been hosting the daily series on the Clubhouse app All Charts Considered. If you’re not on the app, please send me a note, and we will get you invited, so you can join in some of these conversations.

On this week’s episode of the podcast, I’m very excited to welcome Casey Miller from the Los Angeles Times. Casey is a data journalist who automates and analyzes data around disasters, both natural and manmade, that threaten the daily lives of Californians. Now, in the past, Casey also helped create visual and data driven graphics at Mapbox and Vox Media. And I didn’t actually know about her background at Mapbox, so we talked about how that work has actually informed how she approached creating maps at the Los Angeles Times. We also talk about the graphics desk more generally at the LA Times, and we talked about how the graphics desk and the reporters at the LA Times work together to create different visualizations and their evolution to now using Datawrapper in the newsroom.

So I think you’re really going to enjoy this week’s episode of the show, I think you’re going to learn a lot about how they work at the LA Times, and I really hope that you’ll check out some of their work as someone who lives on the East Coast of the United States, I feel like the LA Times work doesn’t quite get there do. So I hope you will be able to check out their work, especially Casey’s work around these disasters, as mentioned, both natural and manmade, over in California. So I hope you’ll enjoy this week’s episode of the podcast, and here’s my discussion with Casey Miller.

Jon Schwabish: Hi, Casey, welcome to the show. Good to see you.

Casey Miller: Hey, Jon, good to see you too.

JS: I am very excited to have you on the show, always good to have folks from the LA Times, chatting with me, because I feel like, as an East Coast person, the LA Times doesn’t get quite the, I don’t know, doesn’t quite get the do that it deserves. So I’m excited to have you on the show and talk about the work that you’ve been doing over there. So do you want to talk about your background, and how you got into being a data journalist and the work you do at the Times?

CM: Sure, yeah. I’ll go back kind of far for this, but it happened when I was in college, as, I guess, for a lot of people, but I had attended a separate university my freshman year, and I transferred to University of North Carolina for my sophomore year; and when I got there, I thought that I really wanted to take a graphic design class, because it was kind of different than the things that I’ve been doing, and I wanted a change of pace, and it seemed interesting to me. And in order to take the graphic design class, I had to pre-declare as journalism major. It’s just where they were held in the school, that was the only kind of place to do it, and I was like, okay, this seems interesting enough, let’s try it out. And then from taking that class, I kind of found this data interactive kind of program where we learned a little bit of JavaScript, enough to make some basic interactive charts, and then we also had introduction to some back end programming, it was in Django at that time, I think it might be something else now. And so, I kind of fell in love with making projects that involve these tools, and geared towards storytelling, and I just really fell in love with it. And from then on, right out of school, I actually interned – what was then the data desk is now updated graphics desk at the LA Times, so come full circle there. And I worked at Vox for a couple of years doing a variety of things, including newsroom tools and also storytelling projects. And then I worked at Mapbox for a brief stint, got really well versed in those tools that I still use today, and I’ve been at the LA Times for a little over two years, mostly working on natural disasters kinds of related things. So I maintain our live wildfires map, and that incorporates a bunch of live fire feeds for hotspots and perimeters and things like that, as well as evacuation zones. And then I also kind of rehabbed our Quakebot, originally built by Ken Schwencke, and added some ShakeMaps to that, which was a fun add that I think everyone has enjoyed. So it’s kind of fun to see. We had an earthquake here earlier this week, I think, and it was kind of fun to look at the map after and see, yeah, all these people felt the shaking. So that’s always neat.

JS: So that’s interesting. So you interned at the LA Times. So you were living in LA, interning there, and then went to Vox and did these others things [inaudible 00:04:57] come back? Wow.

CM: Yeah. So I went to school in North Carolina, I was out in LA for the summer interning, then I was back on the East Coast, ping ponged a little bit between New York and DC, then made it back – then when I joined Mapbox, I was in the Bay, and then came back down here for LA Times. So I jumped around quite a bit.

JS: Yeah. So I asked you a while ago to do a video for the one chart at a time series on choropleth maps. I guess, we could talk about maps a little bit, but I don’t think I knew your background with Mapbox, and so I’m curious about taking the skills and the background of working at Mapbox, and now working in a newsroom where you’re creating maps for a general readership. So I don’t really know what my exact question is, I guess, it’s really, how do you think about blending those two, or taking the work that you did at Mapbox, and now focusing on that communication side of things?

CM: Yeah, well, it’s kind of funny, I actually joke sometimes that I use Mapbox more at the LA Times than I did when I worked at Mapbox. I joined a team there, I worked with – originally, it was a very small team. It was a narrative space team. It was going to be storytelling based. And I worked very closely with Lo Bénichou if you know them, and that was really cool, until I got dissolved. It became part of the marketing team, and then it was more so building demonstrations for people. I did get used to the tools, and it just like the outcome of what I was trying to do with them wasn’t quite aligned with what I wanted. It just wasn’t the type of storytelling that I really find engaging. It was more, let me prove how these tools are great, which enabled me, but now, to take those kinds of tools and skills that I learned and use it in my work, which has been really nice. I mean, it was, we, at the LA Times, when I joined, they had previously been using CARTO, which had been retired, and then had a smattering of other kind of maps and maps frameworks that folks had used for different projects, but there wasn’t a go-to solution. So print maps were done one way, they still are done a different way. But locator maps were one thing, it kind of depended on who was building the map, what they were comfortable with. But it was also challenging because we didn’t really have a single source, and we also basically didn’t pay for only, like a lot for any of these sources. We had the great ad hoc plans for the different things, and so, we kind of made a transition to Mapbox for our team on the editorial side, and on the business side, like [inaudible 00:07:36] the business side, but people who maintain the CMS at the paper, we use Mapbox in there now as well. So we have a much larger plan, and we know we’re all kind of centered on that and streamlined through there. So it was a no brainer for me, when I was building the fire map that it was going to be Mapbox. It’s a great tool for easily being able to implement scalability in a number of layers, and it’s going to work all the time, and a really good [inaudible 00:08:02].

JS: Right. So why don’t we talk about the wildfire map, for two reasons. So the one reason is anything you want to talk about with the technical parts of the map or just the process of building the map and clearly updating the map; and also sort of secondary question, which may be even more important, but maybe of little less interest to folks who really want to know how you do your day to day, which is, how do you think about a project like that, that is really meaningful, like, people can make life and death decisions based on what they see on that map. That’s very different than making a line chart of GDP growth or something like that. I mean, you’re making a map where people can really see their homes and how close a fire is to their neighborhood. So there’s really two questions there, and I’ll let you just tackle either one if you want.

CM: Yeah. So we, I mean, mostly myself, and Ben Welsh, and then a couple of other folks with the paper, talked a lot about the various data sources that we found for the different layers that we have on the map; and the different layers predominantly are hotspots, whether it be individual hotspots or aggregated hotspots, fire perimeters – we do have an air quality layer, which is kind of interesting to see. And then, we also have fire origin points, which is one of the most important layers; and then also the evacuation zones layer, which is the most difficult to maintain, but I think also the most important, and that was a newer add. I had started to add it at the end of 2019, but really it got a lot farther, more developed this past year. So I think part of it is challenging, it’s weeding out, and then also deciding which sources we wanted to use. So for hotspots, for example, there are five or six different sources, and the two that we picked, one is the fastest one, and one is the most human verified one, and it’s a combination of those things.

So we’re trying to [inaudible 00:09:58] data as quickly as we can, but also try and verify it as much as we can. So just for the background, there are a couple of different NASA satellites that gather hotspot information, and there is a NOAA feed that pulls both of those hotspot feeds from NASA, and then a person actually goes and looks at them, and verifies whether it actually seems like it’s somewhere that’s on fire, or if it’s going to be like, you know, that’s probably a group of solar panels or something. Because there are other things that can reflect and kind of look like it’s hot, but it’s not actually hot.

JS: Oh interesting, okay.

CM: Or not on fire anyway. And so, the other source that we have, the fast one, is a source that we have from Descartes Labs, and we work with them to kind of get a satellite source from them, which updates much more quickly, it’s a couple of times an hour versus twice a day. So I think kind of vetting the data is a huge part of it, was a huge part of it, I mean, still continues to be, I have to update [inaudible 00:10:58] I had to update one of my feeds last year, because they changed it. I would say honestly, an interesting key, but the least useful is the fire perimeters, just because they’re often outdated, and they’re not updated as frequently, showing you the extent of where the fires burned, which is an interesting thing to see, but it’s not really relevant in real time. Because the hotspots and evacuation zones are much more helpful when it comes to seeing what you need to know or what your family needs to know or whatnot, like in real time.

JS: Yeah, right.

CM: I will talk a little bit about the evacuation zones, it was a fun challenge, but also challenging. So unlike all of our other sources, there is not one main place to go for evacuation zones. They’re all put out by, well, that’s not one source anyway, but it’s much more disseminated. It’s usually the local fire department or the county office or the sheriff’s office sometimes, and you have to kind of do some strategic looking around. So sometimes InciWeb, which is a fire agency will have sometimes links, sometimes will write up where the evacuation zone is. Other times you have to Google for it and see if there are local county agency, has somebody tweeted about this link to the Facebook page for that county that has like a PNG of this map. And then you have to figure out how you’re going to extract it, and that’s sometimes easy, if they’re using an ArcGIS map and you can go in and you can find the layer, and you can go to the server, and you can query the server and pull down that GeoJSON which is ideal. Other times, it’s just a written two-sentence paragraph of started by the north side of this lake, and extends down to the old schoolhouse on the crossroads of Maine and 1st, and that would actually be an easier one. I’ve had some that are just really challenging to figure out, and it’s honestly my best guess. And I’m kind of err on the side of, well, I’d rather have it on there and it be a little bit off than not put on there at all.

JS: Right.

CM: And so, I’ve kind of worked to streamline that process a little bit, I wrote up some documentation for it the last summer, just because it was challenging also, in that, I was the one grabbing all these things, and I wanted to kind of help other folks learn how to do it as well, but we were in the throes of a huge fire season and me trying to kind of show folks in a way that made sense, but then also make sure all of these processes were updated, it was kind of too much to do at that point of time. So now we have some documentation, I think that this year other folks will be able to jump in and also find and add those zones too, probably with, you know, I’m sure we’ll have a Slack channel around it, and we’ll talk it through when there are questions, things like that. But we’re set up now in a better way to have folks contribute to it.

JS: Yeah, that’s really interesting. So just to continue this thread as you look forward, and not in a good way, to the next fire season…

CM: Sure, yeah.

JS: Now you have this sort of infrastructure set up, and this is sort of a broader question, but how does the graphics team then interact with the journalist team, either for this project or for any project, just seeking a little bit more of information about how did the teams work together in the newsroom specifically?

CM: Yeah, I’ll talk a little bit about this project, and then I’ll kind of talk about it a little bit more broadly, too. So for the maps itself, unlike a lot of the things we do, it’s one of the few applications we maintain. So most of the graphics or graphically driven things that we do are standalone pieces, I mean, this is a standalone piece, but it’s continually updated. So for this one, we worked with a lot of reporters who cover fires and natural disasters to figure out what we want on the surface and how they wanted to use it. And also, there is a view of this map that is embeddable within stories in the CMS, so that’s something that gets used very frequently, especially for new fires that have evac zones and things like that. On the other side of things, there’s a mix between folks coming to us with ideas for stories that they kind of want to investigate, or are starting to report on and they really want to [inaudible 00:15:15] the beginning, and it’s not so much of a service desk kind of thing, it’s more of a we’re going to collaborate on this kind of idea.

And then there’s the other side of it too, which is we also pitch our own ideas, so there are stories that are homegrown on our desk, and folks do all the reporting and data reporting and all the graphics, and it’s, I don’t want to say isolated, but fully produced by our desk. Now, as far as traditional newsroom charts, basically, but things that bar charts, line charts, sometimes simple locator maps that are more of a request kind of thing, we are actually, you know, we’ve instituted my manager and the team that has instituted a service where we use Datawrapper to create these basic tools, we have a [inaudible 00:16:00] Datawrapper, we work with them very closely to get styles for our charts, both print and web, and we are starting to kind of create a production line that one of us [inaudible 00:16:12] every day. So it’s once every two weeks or so, you end up being like on call for helping reporters make these charts. It’s a lot less of a load than having someone come to you and ask, can you make me this custom chart. So I think we still want to help out, and it’s still in the purview of graphics, but we want to – we’re focused – I think, this allows us to focus more on the creative data storytelling that we really shine doing.

JS: When a reporter asks you, are they asking you to make the chart in Datawrapper, or they’re trained up in such a way where they’re asking you sort of more of a quality control check?

CM: Right now, it’s kind of a mix services, I think, ideally, we want to get to a place where all the reporters can make the charts and it’s more of a quality check type of thing, but it is a something new that we’ve instituted, there’s a mix right now. So there are some folks who have been making charts in Datawrapper for over a year, and there are some folks who have never made one and do [inaudible 00:17:08].

JS: Right. So does that mean then that the other tools that you use and know how to use, like Mapbox and JavaScript and D3 and these other tools, that mean those tools are used for the larger projects, the more in-depth projects – I don’t want to say in depth in that way. I don’t mean it that way, but like, yeah…

CM: Yeah. No, more technically, in-depth, yeah, I guess.

JS: Yeah.

CM: Yeah, I would say that’s right. I mean, I think we have – one of the unique things to think about this team is that it’s the data graphics team, which is the result of the formerly known data desk and the formerly known graphics team at the LA Times. And, as such, there’s a variety of people obviously who have different skill sets. So like, we have folks who are amazing at Illustrator and can do 3D graphics, and that is none of the skills that I have, and it’s really great to be able to work with folks and kind of pair on projects or smaller graphics or larger graphics to kind of figure out who has what skills to get the thing done most efficiently, or also learn new skills from.

JS: Right.

CM: I do use a lot of JavaScript, a lot of D3. We have a lot of folks who, I mean, myself included, but do data processing, and Python or R, it kind of varies. We do have like a [inaudible 00:18:30] system for our projects, so like static site generators that we use for most of the products we produce that or standalone, which is maintained by a few folks in the team. And there’s nothing wrong with using D3 for a smaller chart or something. It’s just we’ve tried to kind of get away from that because the overhead is so much bigger for spinning up a chart in D3 versus quickly going into Datawrapper or making a bar chart for someone.

JS: Yeah, sure. In your experience, since you’ve been doing this a while, and I was literally, before we started talking, I was on another call about interactive graphics, and since you’ve moved from, I guess, sort of, like the general graph from D3, from a more [inaudible 00:19:13] custom graph to Datawrapper, what is your take on interactivity when it comes to creating graphs, I mean, when D3 sort of first became really popular, everything was interactive, and I feel like we’ve pulled away from that a little bit, and now with tools like Flourish and Datawrapper, we’re sort of getting a little bit back to that, but only because it’s a little bit easier now to build those. So what’s your take on the balance between interactive graphics versus static graphics?

CM: Yeah, I think, honestly, for most graphics, I think they don’t need to be interactive to a large degree. I do think that if you really need to be able to surface granular information, having – there’s a difference between having tooltips for something and like animating bars, like, I think it’s really a scale of how much do you need to show and I think that having the ability to drill down and show more information is definitely useful. But I don’t know, if I think that I will use the animating bars again, but the scaling out the bars, unless you’re trying to show change is useful. So I think a lot of times you just need that bar chart or that line chart, if you’re really trying to use the same chart to kind of show differences over time or show how the difference between these two things are, that’s when animation really comes into play. And I think that for us, and I think that you see that in a lot of the more custom graphics that we do, but we don’t need it for a lot of these simpler, just as good, but simpler charts that we do, it’s just the animation wouldn’t add anything to it, and often it would even make it a little bit obscured in what we’re trying to show.

JS: Right. And then how do you think about that interaction of the interactivity or the animation with what I presume is a large share of your readers who are going through your content on mobile device?

CM: Yeah. It’s kind of funny, because I’ve always thought about mobile first, and I don’t say that as like, I’m better or anything, it’s just when I graduated from school, that was the thing. Mobile first was the thing that was drilled into use, and I use my phone all the time, and I know that I’m one of those people who will more likely see the graphic from any news site or what have you on my phone but on the computer unless I’m seeing it while I’m working. And I think that I put, we all put a lot of time into it, I prefer to start there and kind of think about the interactions, like, it’s definitely challenging with some technologies, like, I was working on a project with some folks recently, we were trying to, instead of having an auto playing video, we were trying to allow you to physically scroll through the video, so you would scroll down the page and they would advance the frames for the video. And it worked fine on desktop, it did not work well on mobile. The scroll itself, scrolling through the video worked okay, but we are trying to have like these interstitial points where cards would come up and you’d see more detailed information, the interactivity you’re speaking of, and the way that mobile registers the scroll is totally different than the way the desktop registers the scroll. And you don’t get every frame you get when you started, and like when you ended, and you just skip the spans of time, that’s how we were keying these things to these locations, and it just totally flopped. And we were on deadline for this thing, and so we kind of retooled how we wanted it to work and tabled that, like, okay, we learned a lot from that, we’ll come back to it, I think it’s a cool idea, not for this project.

JS: Oh, interesting. So in that case, it really was, and primarily because of the, sounds like because of the deadline, it’s like we’re not going to put this out, because it’s not going to work on mobile, yeah, that’s interesting.

CM: Yeah, we switched to, it’s the project we did recently on Pico Boulevard. So we featured a number of different places, businesses, so restaurants, some more traditional businesses along Pico Boulevard in Los Angeles, and showing how they’ve changed over the year of the pandemic, and some of the reporters wrote little anecdotes to go with that, and pulled quotes from the business owners, and we had one of the videographers actually drive down all the Pico, so like, from the coast down to downtown, and we did use that video, but we ended up cutting pieces of it to show where the locations were, and instead of scrolling through it, we had kind of like, it was the slide, like, scroll up a slide thing, like, the videos would auto-play and the information was kind of being [inaudible 00:23:52] the next video will auto-play rather than you scrolling through the whole video yourself. But it was like really cool to explore, and I think that having the time to kind of look into that and realize that there is something there, even if we couldn’t turn it around for this project, I think we’ll definitely want to use it for something in the future.

JS: Right. And so, now you’ve got this experience, and presumably, it’s not just you’re part of the graphics desk, but it’s also the journalists, it’s also the, I assume, there’s – I don’t know anything about this, but I assume there’s kind of like a back end person or people who have more experience of how to piece all this together for the next time.

CM: Yeah, kind of, so yeah, we worked with a number of reporters on this story, and we are on the business desk, and then it was a group of three of us from the data and graphics team, but I will say kind of like how [inaudible 00:24:47] earlier with us having different skill sets, there are a couple of us who also do back end work on this team. So it would probably be either one of the two of us who was kind of doing that work, will probably be on the next story where we might use this kind of tool, but we also – and I think another different view, if we weren’t working on it, we would probably talk to a partner with the person who was working on it for a little bit, just to share the knowledge that we’ve gained.

JS: Right. So one more question on this balance of these different platforms, because you and I were on a chat on the Clubhouse app a couple of days ago, and the topic of print came up, and so, now you’ve got mobile, you’ve got desktop, and you’ve got print, and a decade ago or so, my understanding was that newsrooms were doing sort of three separate things for each of those different platforms. So there’s still a big print circulation of the LA Times, let’s just take the story that you just did, so how do you think about the balance between that versus the desktop mobile experience versus the print experience?

CM: Yeah, that’s a great question. So I think, like many other newsrooms, the mobile and desktop experiences, as far as technology is used, very similar. It may be scaled differently or things like that, but for print, it can be different. So, like I mentioned, we have Datawrapper for a lot of the print charts we do, and that works for maps as well, for simple maps. So locator maps or things we’re trying to show, points on a map, it’s more challenging with things like choropleth or any kind of overlays you have. So if it’s a simple map, we can often do it, Datawrapper can export for the web and for print, and it’s a different style, but it works for both. And for more complex things, it is a different project. So I worked on a couple of weeks ago, I worked with one of our reporters, Doug Smith, on pulling the new tsunami evacuation zones for the LA County area. And so, this data was updated for the first time in 10 years, and they based it on a different large event than the previous ones [inaudible 00:27:04] it used to be based on once in 500 years event. It’s now based on a one in 1000 year event, so it’s a bit bigger, and the evacuation zones are a little bit larger. And I made this web version using Mapbox, and now we need to do a print version, and you cannot export print from Mapbox, it doesn’t work that way.

JS: Right.

CM: So I will work with one of the folks who is a little bit more well versed in – I can get around in Illustrator, but I am by no means an expert, so somebody who’s a little bit better at Illustrator than I am, who has more experience working with print to do these kinds of things, and so we’ll kind of collaborate on that, which is, I would say this is not super common, a lot of the graphics we do for the paper, we can do with Datawrapper, but every once in a while, there is something that needs to little more custom, that is a whole different process.

JS: Interesting. Wow. Okay. Well, I think you’ve got your work cut out for you, and it sounds great. I mean, the wildfire map, although it’s not something I really have to worry about here in Virginia, I do find it really interesting, fascinating project. So thanks for working on that, I’ll just put it that way. And it sounds like there’s so much data getting pulled together, that is really incredible. So Casey, thanks so much for coming on the show. It’s been great chatting with you and learning about the work over there at the Times.

CM: Yeah, thank you so much for having me. I had a great time speaking.

Thanks so much for tuning into this week’s episode of the show. I hope you enjoyed that. I hope you learned a little bit about how the LA Times graphics desk works, and I hope you will check out Casey’s work and the rest of the team over at the LA Times. So until next time, this has been the PolicyViz podcast. Thanks so much for listening.

A number of people help bring you the PolicyViz podcast. Music is provided by the NRIs, audio editing is provided by Ken Skaggs and each episode is transcribed by Jenny Transcription Services. If you’d to help support the podcast, please share it and review it on iTunes, Stitcher, Spotify or wherever you get your podcasts. The PolicyViz podcast is ad free and supported by listeners. If you’d like to help support the show financially, please visit our Patreon page at