Episode #182: Aaron Williams

Aaron Williams is a data journalist, scientist and visualization expert tackling inequity in data and design at scale. He’s currently a senior visualization engineer in Netflix’s Data Science and Engineering group and previously spent a decade as a data and graphics reporter—most recently at the Washington Post.

I invited Aaron on the show to talk about his fabulous dot density map and the code he published to an Observable notebook. But we ended up talking about so much more: his background, what it takes to use data responsibly and ethically, and his next career adventure. I hope you’ll enjoy the show!

Episode Notes

Aaron | Twitter | Website | Observable

Aaron Williams and Armand Emamdjomeh, America is more diverse than ever–but still segregated

Center for Investigative Reporting

Washington Post Police Shootings Database

San Francisco Chronicle

Los Angeles Times

Washington Post

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’m your host Jon Schwabish. Now, for the last few weeks on the show, we’ve focused on data and technology and the intersection of those two and how that intersection can punish and profile people of color and low income people and families. In some ways, that’s been a move away from the core content and the discussions I usually have on the show, but today we’re going to get back to some core data visualization work. I’m going to talk with Aaron Williams who was formerly a data journalist at the Washington Post, and that’s where he was when we spoke in late August. He is now at Netflix doing some great work there, I’m sure. And while at the post, Aaron produced some amazing visual journalism. In 2018, probably my favorite piece of his, Aaron and his colleague Armand wrote a story about segregation in America, and they created this series of dot density maps that I still think may be some of the best I’ve ever seen – just the combination of the text, the colors, even the background on the entire piece is really just amazing. So more recently, this was over the summer, Aaron published the code for that project to an Observable notebook so that others can use it and model their work after it. Now, we talk a bit about that of the Observable notebook and the code and some of that is a little over my head, but if you’re into JavaScript and you’re into Observable notebooks you’re going to love that part of the discussion. So we talk about Aaron’s work at the Post, we talk about what’s coming up for him at Netflix, and we also talk about conversations he’s had about challenges journalists of color face in the newsroom which is an interesting part of our discussion. So I’m sure you’re going to like this talk. I’m sure, if you’ve seen his work, you love it. So here’s my discussion with Aaron Williams.

Jon Schwabish: Hey Aaron, welcome to the show. How are you doing?

Aaron Williams: I’m doing well Jon, thanks for having me.

JS: Hanging in there in these, I think, strange times is an understatement.

AW: Yeah, no doubt.

JS: Well, thanks for coming on the show. I’m excited. You’ve got a number of great projects, and there’s a particular project that I want to dive into in some detail. You’ve got a new exciting opportunity coming up on the West Coast, and I want to talk about that. But before we get into all that, may be it’d be helpful for you to talk a little bit about yourself, your background and how you ended up at the Post on the graphics desk.

AW: Yeah. So again, I’m Aaron Williams. I’m currently a data reporter at the Washington Post. I’m originally from California. My background is in journalism. I have been working as a journalist for almost a decade now. Initially, when I started out in this work was very interested in focusing on sort of offers like a neighborhood reporter, neighborhood reporter in San Francisco. I remember, like, there’s one story in particular that I did where the neighborhood I lived in was just Excelsior neighborhood in San Francisco, the city had commissioned a report or a survey asking people in the neighborhood how safe they felt in their neighborhood, and so they had all these tables and graphics they used to not only do the survey but then kind of report what they found, and it was the first time I had kind of discovered the idea of data journalism, which was like using data that’s BEEN published by the city or by some kind of government, whether local, state or federal, and trying to find stories out of that versus how I had traditionally at the time been reporting which was like basically cold calling people and knocking on doors and people would be like, who the hell are you, like, which was just not a style I liked. So I was like oh if I could work with data from the city, not only does that, will I have more targeted questions to ask, that will hopefully mean I’m doing less of the kind of cold knocking on doors of strangers who are like, I don’t trust you. And I could actually like report on stories that weren’t just like floating there, happening in real time. It required kind of knowing the city’s publishing schedule around data, getting to know the public information officers and the data folks in the city halls of San Francisco. And so that’s what kind of set me on this journey, and that was like roughly 2009 I want to say, and so that was kind of my start. And then in the meantime while I was kind of working on my journalism career, I was living in San Francisco, it just seemed like if you weren’t coding, you weren’t cool. I don’t know if maybe there was like – I was just, this was at the height of Zynga, [inaudible 00:05:02] the behemoth tech company, just like Twitter had just moved to Selma, and so this was when all of that stuff was happening, and I remember thinking, oh I need to, like, I should learn some of the skills because, just in case, I want to change careers or, just in case, I want to try something new, I’ll do it. So I actually remember, I went to San Francisco State and I applied to take a minor in computer science, and my journalism advisor at the time was like, why would you do that, these are totally separate career choices, just like, why don’t you minor in history or philosophy or something. So I actually ended up minoring in philosophy which has some parallels to programming, but I didn’t take it upon myself to teach myself how to code. So during that time I was going to school and reporting, I was also on the side just like Codecademy and those kind of things didn’t really exists at the time, but I was – or they were really new, but I was just getting books off Amazon or out of the studio library, and I taught myself Ruby which was also the popular programming language at the time, and not so popular [inaudible 00:06:06].

But that was kind of my start and to get into this whole world, and then you flash, after a brief stint at the LA Times, I ended up getting a job at Center for Investigative Reporting which is now called Reveal, and that’s where I really kind of started to build my skill set as a data journalist, that’s where I learned Python, that’s where I did my first real analysis and got really deep into D3 and data visualization and using all of those tools to do investigative reporting. And so I did that for a couple of years, I didn’t worked at the San Francisco Chronicle, and then about five years ago the Post reached out to me, being like, hey, we see what you’re doing over in the West Coast, would you to bring your skills to Washington DC. And at the time I was like, you know, I’m never leaving the West Coast, you’re going to have to drag me kicking and screaming out of here. But the Post had, you know, this was like maybe a year after the police shootings database had just been published by John [inaudible 00:07:06] and all those folks who worked in that project at the Post, and I was just enamored by the Post’s ability to take data visualization and deep investigative reporting to talk about something like really crucial society. And so I was like, I would be stupid not to try to go over there and try to do some of this work and work with those people. So I packed up all my stuff, moved 3000 miles across the country, and been at the Post since. So that’s kind of the long haul trajectory that got me here.

JS: There’s a lot of amazing parts of that story, but the one that strikes me is your journalism professor sort of said, why would you worry about coding, and that was only a decade ago, and how much things have changed, yeah, that’s amazing.

AW: I think about that all the time, because – and I’ve talked to that advisor since, like, we’re friendly – but I’ve told them, I am like, yeah, you were so off the mark, because it’s not just that learning how to code for journalism now seems like a no brainer, software just generally is a huge facet of all parts of society at this point. So you could learn how to code and work in business, you could be a journalist, you could be even a carpenter, you can, like, an architect, it doesn’t matter what skill set you use knowing a little bit of software engineering or coding will get you really far into all of those disciplines, and so, yeah, I think about this all the time now because I’m like, I’m glad I taught myself how to code because it’s like an alternate world where my [inaudible 00:08:39] I don’t know history, and then I just never went that route. My career, my life would be totally different because of that.

JS: Right. What’s also interesting about your origin story as it were, is that you came from a sort of true traditional journalism background whereas a lot of people who, and now are doing databases, they come from all sorts of different walks of life, they’re astronomers, they’re economists, they’re whatever. I’m curious when you are – one thing I’ve been talking about with people on the show and elsewhere is the difficulty that people who are – their background is more technical, so maybe they’re a social scientist or a mathematician or a statistician, and then talking to people is very foreign to them, it’s not something that they’re used to doing even though we all acknowledge that that’s useful to the skill of being able to talk to the people that we are studying or communicating with. And I just wonder do you find that you have this additional skill set where talking to the people that you are collecting data about or downloading from some other place is just more natural for you? And does that make your job both easier and the final product better because that part of your, you know, you’re able to just do that and have the experience doing that reporting?

AW: Yeah, I mean, I think that to some level my background as a “traditional journalist” certainly helps. But I think really the big skill that I think as a journalist that helps us do when applied to data journalism and data visualization that is really crucial is skepticism. And I say that because I think if you [inaudible 00:10:23] even this year, kind of at the kind of onset of the COVID-19 epidemic, you saw every armchair economist and data science student do a COVID data project, like, I’m not giving a number, March through June of this year, it just seemed everyone suddenly became an epidemiologist.

JS: Yeah.

AW: And I found that infuriating for a lot of different reasons, one, because it just seemed there was this mad rush to be like, I have to make the definitive COVID dashboard. And a lot of it was I think in good faith, there wasn’t a lot of reporting being put out by the federal government, there’s a lot of uncertainty about what was happening. And so I think folks were trying, we’re actually eager to try to be like I want to do something about this. So I mean I think a lot of it was in good faith, but I think often what I saw people report, whether their denominator was really bad or they just, like, the data sources they relied on were super wonky, I just thought a lot of these things I saw kind of get popped up in those early days, as a journalist, the first question you’re asked is, is this data even legit, who produced it, how were they tabulating this, is it consistent over the timeframe that I’m trying to report or trying to visualize – these are questions that I think any economist, social scientist, anybody with any kind of data background was to ask, like, I think skepticism is the biggest trait a journalist has, and usually with skepticism that then means you have to talk to somebody who can either back your skepticism or say, oh actually, here’s something else you should add. So I mean I think that my background in terms of being able to talk to people certainly helps, but I really do think it’s the skepticism and the idea of not taking data on its face as legit, that’s the most key aspect of the role, I think.

JS: Do you think, I mean, aside from the COVID dashboards that were littering the internet a few months ago, do you think in general people are not as skeptical with their data as they should be?

AW: Oh absolutely, I mean, I think just generally, not just data journalists, but I think just when people think of data, there’s just this built in assumption that it’s without bias, like the data, whoever collected it was kind of done in this laboratory setting, and it’s machine run, and there’s no kind of built-in assumptions when, I mean, I think as you likely know, probably listeners of this podcast know, data is produced by people, so people’s assumptions get put into that data. The data you don’t collect is just as crucial as the data you do collect. And so I do think that because data science and data visualization, it’s kind of an interesting discipline right now because if you have both folks journalists and academics are deep in it, but then you also have artists and people who more want to kind of do more – not like they’re not backed in science, but I think there’s also kind of this artistic angle to it. And then you also have folks who are using data and visualization to just focus more analytics metrics and business building, and so all of these – and everyone kind of has built-in assumptions in all these different places, and things they’re willing to not break on, things they are willing to break on, and I think that all of that together just kind of creates an interesting conundrum of when should you be skeptical of the data. As a journalist, my assumption is always be skeptical. All data you collect has inherent problems, inherent bias, so it’s your job to figure out what those are and how much of that then impacts the thing you’re trying to visualize or the statement you’re trying to make. And just because of the bias there, it doesn’t mean that the data is bad, but you just need to know what that bias is before you move forward, otherwise you could end up publishing something that’s misleading. So yeah, but I think that there’s a total assumption that all data’s good. I think it’s changing, I think that with the rise of social media and kind of just we’ve seen some bad actors with the use of data, I think there is kind of now a more general consensus that not all data is good. But I still think that there is this underlying assumption that if someone hands you a dataset that is perfect, you should just work with it which I don’t think is true at all.

JS: Have you had an experience lately where you’ve been working with the dataset and you’ve been skeptical about it and you’ve gotten down some length of the project and said, I can’t use these data because of whatever reason and you’ve just abandoned the project?

AW: I mean, I think the closest example of that in recent memory would be the kind of early COVID data that was being reported by city, state, and county health departments. And it’s not that the data was misleading, it’s just that really early on, anybody who wanted to know, you want to know in my state how many COVID cases are there, how many tests have been done, just really kind of high level questions, and really early on, me and the folks working on these projects realized that data was just not there, or if it was there in some places, but not in other places. And sometimes they included how many total tests they gave and sometimes they didn’t or they did briefly and they removed the data, because it made them look bad. There was just all of this craziness happening with those early dashboard days of COVID that made it insanely difficult to actually say with some kind of authority or clarity what was happening. And so I think a lot of the early meetings I was in at the Post where we were talking about what was happening, like, we were collecting data around this, a lot of the talks were just me and Steven Rich who’s also data reporter at the Post, and others and [inaudible 00:16:09] who is also a data reporter, it was like the three of us basically like, we can’t use this, this is bad, because we couldn’t figure out how, like, we knew that, just looking through the data that it wasn’t collected in a consistent enough fashion for us to make a full-throated claim of what’s happening. Obviously, that changed. We then brought in more folks who helped us collect [inaudible 00:16:29] scrape and collect data from different health departments, and then that what now ends up powering the dashboard on the Washington Post. But those early, that March-April-May zone was just a lot of us trying to figure out, like, well, what can we say, because we just felt the data in some places is being reported, but it was only a handful of jurisdictions or they had only tested once or twice. So that was more just having to do with how the data was collected, less about the data just being outright bad or wrong.

JS: Right. This is interesting because it’s actually a good segue to the project I want to spend a little time talking about, which is your segregation and diversity maps that you published at the Post a couple of years ago – these dot density maps that are just, I just think, spectacular. And then a few weeks ago you posted the code to an Observable notebook, and so I want to give you a little bit of space to just talk about the Observable notebook and how folks can use it. But maybe before we get into that, you can talk a little bit about the project itself, and then maybe we can also talk about the missing data because the project has, as I recall, six different race and ethnic categories; and of course, there’s a lot of diversity across and within those racial groups that are not captured by our standard federal datasets. How did you think about having a category for Black people and a category for Asian and Pacific Islander people? How did you think about these different groups that are, you know, they are missing pieces, they are missing people who report their race in maybe different ways and capturing and not capturing the diversity of the country?

AW: Yeah absolutely.

JS: Yeah, so that was a long question, and I sort of like…

AW: No, no…

JS: It’s like, here you go, just talk about it.

AW: Yeah, no, no, it’s just great. I got what you were kind of setting out, I think I got this.

JS: All right.

AW: So yeah, the project, the story’s called America is more diverse than ever – but still segregated. It published May 2018. It was me as well as my colleague, Armand Emamdjomeh, who’s currently an assignment editor for the graphics desk at the Post, we’ve been longtime friends and colleagues for years now. And so this project was kind of born out of my desire to bring some data, some clarity to the question of what’s happening in the country. I live in Washington DC, before that I lived in Oakland, California; and in both instances, these are cities that have seen rapid change both in the demographics of the cities, the racial demographics, as well as the kind of class demographics of those cities. And having moved from the Bay Area which was – it is still the most expensive place on the planet, but it was nearly the most expensive place on the planet, they have rapid gentrification. So then moving to DC where kind of the same thing was happening, I had all these kind of conversations with friends of mine who were just kind of like, oh yeah, I remember when this neighborhood used to look like this but now it looks that; and wow, the Black population of this part of the city is getting pushed out, and now this is the demographic group that’s taking that over. So my question was like, okay, well, could we actually use data to show that. So that was kind of the genesis of the project. I didn’t want it to also measure the idea of segregation which when you look at these dot density maps are really great, but we’re just, we’re showing where people in theory live, we’re not actually measuring segregation. Segregation is the actual separation of people along racial lines. And so what I wanted to also figure out is that could we not only show where different demographics have been and are over time but can we then add a score. This is like, hey, this block of the city is actually like, if you look at the distribution of people in it is fairly isolated by one racial group, and because of that, that is in some way showing segregation in a way that’s just showing where Black and White and Latinx and Asian folks live. That’s part of it, but we’re not actually scoring the blocks that those people live in. So that was the second aspect of the project.

So yeah, and so, I use census data from the 1990 census, 2000 census in 2010, so the decennial census years as well as the latest American Community Survey data which at the time was the 2012, the 2016 five-year release. And my colleagues Dan Keating and Ted Melnick helped me out a lot because since the geography changes over time, so they helped me standardize to the 2010 census geography, so that way we could have an apples to apples comparison of the same blocks. Yeah, so what we did there was like, look at a couple of different places. I wanted to kind of understand the history of segregation in America, and I spoke with a researcher at American University, his name is Michael Bader, who gave me a helpful way to think about it, which was kind of like, you kind of have three ideas of segregation, you kind of have the legacy of historical segregation that was set up right after the Civil War during the Reconstruction Era. You then have – so you have white flight that happens in the 50s, 60s, 70s. And then you kind of have the new, since the 70s and beyond, you now also have kind of these rapidly diversifying suburbs. So a great example of that is Northern Virginia and kind of how communities like Annandale and Fairfax and [inaudible 00:22:15] and those communities have rapidly become more Latinx and Asian-American over the last 20-30 years. And then you have places like Houston, Texas that, to some level [inaudible 00:22:26] level of segregations because of just the fact that the city was kind of developed later compared to Chicago or New York or DC, because of that Houston kind of has also its own unique mix of racial integration and segregation.

So I wanted to [inaudible 00:22:42] so I tackled Chicago, DC, Houston as three examples of how segregation and racial integration works in the US. And so, I did dot density maps around those three cities, and then finally we just published all the data for the entire US in a matte box vector map that allows you to type anywhere in the US, see where you grew up, see how either integrated your neighborhood is or how segregated your neighborhood is. And hopefully, that then led you to either have conversations with your local politicians or your family or spark something in how you felt. So yeah, but that was the genesis of the project, out of all the work I’ve done as a journalist I think I’m most proud of, it’s the one that got me to this podcast, it’s still one that folks often ask me about, and so I’m really proud of it in it. I think it really shows the power of what you can do with data and visualization to tackle something as insidious but crucial as racial segregation.

JS: Well, I’ll just say this, so I saw you talk about this project at OpenVis Conf I think in Paris a few years ago, and I remember asking, you know, you had chosen this sort of dark gray background for the whole thing, and I was like, why this dark gray background, and I still remember your response was just because it looks whack, it just looks good. But there is a design element to this piece that I think is really striking – do you want to talk maybe a little bit about that – and also maybe folks should definitely look at this project, maybe before they – maybe they pause here and take a look at the project if they haven’t seen it. But just also the choices of color that you use throughout I think just are striking in the way they pop off the page.

AW: Absolutely. All right. So yeah, the color choices I picked for the maps were, we deliberately chose a dark background, and the reason for that was just the fact that we use what a lot of social scientists use when they use census data to break up race into six categories, so we chose Black, White, Hispanic, Asian, Pacific Islander, Native American, and then kind of everybody else, and so everybody else is includes anyone of two or more races and we chose a race that did not fall into any of those categories and things like that. And so we had to pick six distinct colors, and so Armand Emamdjomeh, we were playing with several different color palettes until we decided on these six multi hue colors that we felt were a good representation of the data that weren’t obviously racist in their color choices, and that were to some level of colorblind safe as well. It was like a no easy task. It was insanely difficult to choose six colors that met that threshold.

And so when we initially did the data or visualized the data, we had it on a white background and it looks fine, it looks very cool. And if you go to the Observable notebook, you’ll see it’s also on a white background just because that’s how Observable is designed. But we did, we were just messing around one day, and we decided to place the data on kind of a soft matte black background, if you will, and the colors just really just popped right just because of the sheer contrast between light and dark. And I think that we just decided that’s kind of the aesthetic we should go with for this project, because, you know, and also you look at a lot of Washington Post projects and just a lot of news site projects in general, it’s just a lot of serif type on black type on white backgrounds. Most news websites to some level fit the same kind of genre palette of what you see on a news article website. So by going with white type on black, we just felt it kind of bucked the trend of what you would normally see on a news website. And to your point, when you asked me the question at OpenVis, why did I choose it, I just thought it looked really dope. At the end of the day, I’m a hip hop kid at heart, and so I was like, hey this looks really cool, I would put this in my studio or in my apartment or whatever. So yeah, but color is a huge part of this project I think too, and even when we created the diversity layer to actually measure the level of segregation in every census block in the data, we chose kind of a very traditional green to purple diverging color scheme, and that was much easier to do because we’re kind of only grading on a zero to one diverging scale which allowed us to kind of keep the colors fairly simple. But we really tried to create palettes that were really striking, that kind of popped off this black background, and that would hopefully keep you scrolling and clicking on the project.

JS: No I agree, I mean, I think the pop is what does it for me, right? When I look at this map of the entire United States, it’s the red really pops off, the yellow pop off, and also the areas where there are not a lot of people living, like the Rocky Mountains, there’s just this sort of emptiness in the country which I think also, like, the lack of data, as you sort of talked about, alluded to earlier, the lack of data is part of this story.

AW: Absolutely. Yeah, it was actually really funny, I remember one of the comments after we published the story. It’s like someone who was like, you really want me to believe that all those dark empty spots, like, people don’t live there. And I was like, yeah, those are mountain ranges, I mean, there might be some people, and they would [inaudible 00:28:25] but not to the point where he’d zoom out to the entire American continent that you’re going to see, your North American continent, that you’re going to see a dot pop up. It was just so funny because, yeah, because I then could also, if you look at this map, the “black belt” through the south really comes through really strikingly. So the kind of huge Hispanic population that populates the American Southwest, and obviously California, it really emerges. And then another kind of feature of this project I really enjoy is that if you look around the Phoenix, Arizona area and kind of going north into the Dakotas, you can actually see where the large Native American populations are. So you see the Navajo Nation’s really big concentration right there in Phoenix and then kind of going up into the Dakotas, the assorted Native American First Nation groups that are there. So I mean, I think again it really kind of shows the power of this data, I think if you were to ask me [inaudible 00:29:26] in US kind of which racial groups populate which parts of the country, we kind of know this to some level because of just being American and kind of knowing kind of the setup of the nation. But to see it actually put on a map where you can actually identify, oh I’ve been there, oh this matches my experience; or maybe, like, hey, I never actually thought about the concentration of this people group in that area, what’s the history behind that. I think that that’s where this project I think really kind of elevates the discourse around where people live in America.

JS: Right. Can you talk a little bit about the Observable notebook and, I guess, why you decided to put the code up in Observable and how people can use it for their own dot density maps?

AW: Absolutely. So for the listeners of this podcast who are not familiar, Observable is an online web based notebook environment that uses kind of its own – it uses JavaScript, but it’s kind of its own flavor of JavaScript. If you know JavaScript you can mostly work in it, but it kind of has its own kind of set of keywords and syntax it also provides, but it’s a really great way to explore JavaScript code without setting up a webserver and kind of your own coding environment. You just all you have to do is go to observablehq.com and start coding, and it works entirely in your browser which is really nice. And so, again, when I published this project two years ago, I think probably the first question I got was how the hell did you do this. And I think a lot of people thought I used traditional techniques. So a lot of folks who have got built dot density maps in the past have used tools QGIS or used R or Python which have really mature statistical and GIS tools for doing this exact kind of work. But a lot of folks are surprised that I did it entirely in JavaScript, and so which I think really kind of, for me, was also I wanted to, in some way, you know, I use all those languages, I use Python, I use R, I use QGIS, I love all those tools. But at the time I was just really deep into JavaScript, and I wanted to see if there was a way to do kind of the level of sophistication in terms of GIS analysis that you would do with Python and R but in a language like JavaScript that also, because it’s in the browser, offers a lot of kind of advantages that Python and R deny. And so I told people I did it in node JS and JavaScript and they were like, well, can you write a blog on how to do this. But as you probably recall, not only had I just probably published a project, I was getting ready to talk about it at OpenVis Conf.

So I was not really in a position to also do a code deep dive. Also if for anyone who’s ever published a behemoth of either an analysis or a project, you’re usually not eager to go right back into that code, you kind of want to never think about it again [inaudible 00:32:33] all this time thinking about this one thing. So I was also kind of eager to not look at the code anymore. Also I wrote this code for a while when this project started, I was kind of doing it in some ways, it’s a passion project. So I was kind of doing it in my off time between stories at the Post, and it wasn’t until February of 2018 that my editor at the time, Katie Hank, kind of gave me the full – she was like, I’m taking everything off your plate, you can just go deep on this. So that’s when I brought Armand in, he kind of had to read my tea leaves to figure out what I was actually trying to do with the code, and then together we finished the project. But that code was not written for public consumption at all. It was written [inaudible 00:33:15] in kind of my – and I think I’m a pretty competent software developer, but I was not writing this code to be used by anybody else but myself.

So there’s just all these things that were going to have to happen in order for me to even explain this code. Also I have tried, I have tried and failed at least, I want to say, 15 times, to start a blog about my work. And I have just come to the conclusion that I don’t think I want to blog. I think I [inaudible 00:33:45] I don’t want to maintain a blog, I don’t want to promote it on Hacker News or on Twitter. I just, I can’t, I don’t want to do it, like, me working at [inaudible 00:33:57]. So I was like, okay, and so if you go to my website, like personal website now, it’s just like [inaudible 00:34:04] a very brief bio and a link to my Twitter, my GitHub, and I think my LinkedIn, I think that’s it. So I was like, oh man, [inaudible 00:34:12] turns into a blog, that’s more coding, I don’t want to code anymore, [inaudible 00:34:15] I was so tired of wanting to code.

So basically I kept telling people I’ll do it, but then I didn’t do it. But basically, once Observable came out, there now came this huge opportunity, because I wrote the code in JavaScript to actually begin reporting it over to a web environment. So Observable, I mean, it got announced a while back by Mike Bostock. He’s also the creator of D3. But I think it kind of, like, the beta came out maybe just over a year ago or a year and a half ago, something like that. But I remember, I started using Observable for just even my own production work, and it kind of has, and it still is in a lot of ways, the main way I kind of explore data using JavaScript. I just find it incredibly comfortable to use, and I love that you can use kind of the full ecosystem of NPM which is node package manager, it allows you to use all these tools that normally you need to run a server and have JavaScript setup on your computer which requires you setting up a development environment, working in terminal. With Observable, I can just literally open up a tab in Firefox or Chrome and just start doing it. So the upfront cost of setup is so much lower now. And so, I was already using Observables for some of my own work. And then finally, it kind of came to a point where I was like, you know what, Observable had matured a bit. I had since moved from the graphics team to the investigative desk at the Washington Post which kind of meant I wasn’t doing the kind of day to day reporting that I had to do, that I was doing as a graphics reporter. I kind of had a little more time on my hands as I was kind of investigating and doing more – and then just also that job required less kind of front end development. I was doing a lot more analysis and stats in Python and R, so I wasn’t writing front end code as much. And so I was like, you know what, I’m going to take some time to actually break down exactly how I did this. And it took me a while to do it, because again, I hate blogging. So in Observable, I was effectively creating a blog.

But the one thing that Observable allowed which is something I wanted to show people with this project was like, it’s one thing for me to just copy and paste code to a code block on a blog. And it’d be like, hey, if you can read this code, this is how it works. But by using Observable, you can actually run the code in the browser which I think kind of totally shifted the power of actually writing about it, because now not only am I explaining how I did it, you could literally in real time see the code running, you can go in there and change the code. If you want to change the colors I use, if you want to change the scale of the map, that’s all available to you directly in the browser. You can manipulate the code in real time, and that didn’t exist in Project Publish. So I think the power of moving it to Observable is that it allowed me to not only explain the code but it allowed for anyone who’s interested in it to literally go and poke around the code and do their own, like, manipulate it, so they can hopefully get a better understanding of how it works. So yeah, I mean, I think without Observable, you likely would not still have seen this code because it would have required me to set up a blog which again, as I’ve said, I decided I am really bad at doing it.

JS: Do you have a good name in your head for a blog, but you just didn’t get that far?

AW: [inaudible 00:37:33] I mean, my webs [inaudible 00:37:36] my website is acwx.net. So I guess, it’d be slash blog and I don’t know. I could probably think of some like really like, I don’t know, nerdy [inaudible 00:37:50] on. Yeah, I don’t know man, this is…

JS: Only certain people would understand, would get the – there’s an inside joke in the name of the blog,

AW: Yeah, exactly. Yeah. And I don’t know, man, I mean, it was funny as my Twitter handle about Aaron that I use basically everywhere, I actually came up with that because in Firefox if you type about colon config, you can adjust the browser, the browser settings under the [inaudible 00:38:18] in Firefox. So that’s where my username actually comes from. It’s like about colon Aaron, like as if you’re getting under the hood of my [inaudible 00:38:28]. But I came up with that handle when I was 19-20 or something. And dear listeners, I am not 19 or 20 anymore. That was a long time ago. So I’m kind of done with the overwrought, like, look at me, I understand that clearly you know how to code, I will need to hammer that over in the name of my blog or my handle. I just haven’t changed my handle because it now just kind of stays. I’m just like, I’m not [inaudible 00:38:54]

JS: Yeah, that’s right. It becomes a stay that you can’t get rid of, right?

AW: Yeah, basically, what I’m trying to explain to everyone listening is I’m just incredibly lazy, I don’t want to do any more work I absolutely have to do. The only reason why I was able to publish this code for your consumption is because a bunch of other very smart people built an entire notebook environment for me to do it, that’s like, [inaudible 00:39:17] they made it so much easier to do it.

JS: That is great. That’s great. Okay, so we’ve learned a few things about you today. So before we wrap up, I want to just talk for a minute about your leaving the Post and, at least in the short term, virtually heading back to the West Coast. Do you want to talk about the new gig you have coming up?

AW: Yeah, absolutely. So I’ve been at the Post for about five years, and I decided I wanted to try to explore how to use data in just a different environment, and also think about a lot of my work has focused around racial equity and data equity. And so I wanted to see what it’s to tackle these problems in just an entirely different domain. And so, I recently accepted a job at Netflix where I’ll be working on the content science team as a senior visualization engineer. I can’t really go into too many details about what the job entails because I have not started the job yet. And hopefully, when this podcast comes out, I am still employed with that job. I have no reason to think otherwise, but 2020 is crazy like that. But [inaudible 00:40:21] I think it’s just a huge opportunity. I think when you work in the data science and data journalism space, it sometimes felt like to me that a lot of us who, particularly as data journalists, we were kind of all playing a really high level version of musical chairs where we were all just getting each other’s jobs, where like, one person might work at the Post, and then they might work at New York Times, and then they might go to a Chicago Tribune or an LA Times, and you might go back to the Washington Post, back to New York Times, you might go to a nonprofit, ProPublica or the Marshall Project. And I have nothing but respect and admiration for all of my peers at those organizations and journalism comes out of them, but I think as I was thinking about my career, I wanted to try to just see, well, what else is out there, what else could I do. And for me getting a similar style job at another news entity wasn’t really what was calling me, I wanted to try to see, well, could I bring journalistic thinking and thinking around equity to a space that maybe is inherently journalistic, or maybe inherently, like, it isn’t like a news business by design. And so Netflix, it’s this behemoth of a company that is having an insane year because of COVID, and they have a lot of data. And so when they first reached out to me, I actually had no idea that they were even doing DataViz work. I knew that Elijah Meeks used to work there. He’s a very well-known D3 and DataViz person, and I knew a couple other people Susie Lu who’s also a really fantastic database developer. But I really didn’t really know that like they were looking for people who had kind of my skill set to apply to some of their work around content specifically. And so, yeah, so we talked about it, and it was kind of an ongoing process for a while, but they finally came back to me saying we’re really eager to explore the ideas you’re in. I felt like it was just a really great time for me to just try my hand at something different. So yeah, so it’s still very new. I’m still finishing up my 10-year at the Post. And then I’ll be starting there in a few weeks. But yeah, it’s going to be interesting to see what it’s like to apply these kind of skills in a different domain.

JS: Right, absolutely.

AW: [inaudible 00:42:46]

JS: Yeah, it sounds great, and I trust you’ll get a free Netflix subscription too.

AW: I hope so. I mean, yeah, I’ve been watching a lot of Netflix, as is everyone right now, so definitely, that’s another thing I’ll take on. But yeah, and I think, I alluded to this earlier in the podcast, but data is everywhere and working with data is everywhere. And I think in the same way that my advisor was like, why would you be a data scientist or why would you learn how to code if you want to be a journalist, which now seems crazy to say. The way I’m kind of thinking about this is that it’s a lot of people are now even asking me, why would you as a journalist go work at an entertainment company [inaudible 00:43:31] I feel like my entire career has always been about taking skills that don’t seem like they match, and then finding a way to make them match. I think certainly what you, as I also talked about [inaudible 00:43:45] bringing skepticism, trying to bring justice and equity to work, I mean, I think being able to apply that in a space like Netflix, in a place that reaches millions of people and a company that is deciding the kinds of content that we watch in our homes, there’s a lot of power in that. And so for me I think it’s kind of interesting, I think on paper it might seem like I’m doing an entirely different job. But I think in some ways to me, it feels like actually this is like a natural conclusion of the work I do. It’s not actually a totally different role. In many ways, it’s thinking the kind of same concepts and ideas I’ve played with over my career and just kind of taking it to a different direction. But it’s not different in that it’s unrelated. It’s just not in the same industry. But the work is very much Congress. So I think that that’s the way I’ve been thinking about it.

JS: Interesting. Well, it sounds like a great gig. And hopefully, you’ll be able to share some of the work that that you do and comes out of it, or at least share the improvements in Netflix that come out, and we’ll have a chat again in a little while [inaudible 00:44:46].

AW: Yeah, absolutely, yeah, totally. Yeah, hopefully, like I said in my little Twitter threads, I don’t think I’m ever going to stop being a journalist. So it remains to be seen what my workload is going to be like, but I do hope I can get a little bit of freelance work here and there and publishing stuff, because I don’t think, as you can imagine, just because I stopped working full time in media doesn’t mean I will stop thinking about stories [inaudible 00:45:13] thankfully, there’s stuff Observable that allows me to kind of play with code and publish stuff pretty easily. So certainly I think I’m going to try to get more stuff out on that in Observable specifically. And then hopefully, [inaudible 00:45:28] places. So yeah, so we’ll see, but yeah, I’m excited, it’ll be a nice change of pace. So hopefully, when I come back on the podcast, I’ll have more to talk to you about.

JS: That’s great. Well, I’m looking forward to it. I’ll book you right now. We’ll do it again.

AW: Yeah, man.

JS: Thanks so much man. This was great. It was great to chat with you.

AW: Yeah absolutely. Thanks for having me. I can’t wait to hear the chat live.

JS: All right. Thanks man.

AW: Thanks man.

Thanks everyone for turning into this week’s episode of the show. I hope you learned a lot. I hope you will check out some of Aaron’s work from the Washington Post. And if you are able and interested, you go over to his Observable notebook and take use of the fact that he’s opened up this code for you to use, it’s an amazing resource. If you would like to support the podcast, please share it with friends and family, use your social networks to tell folks about it, write reviews of it on your favorite podcast provider. The show is on all major podcast providers including iTunes, Spotify, Stitcher and Google Play. If you’d like to support the show financially, head over to my Patreon page where just for a few bucks a month, you can get a mug, you can get key insights and new stuff before anyone else finds out about it. I email patrons every month to tell them what’s coming up on the show. But in any case, make sure you tune in next time for the podcast, more great episodes coming your way. 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. The PolicyViz website is hosted by WP Engine and is published on WordPress. If you would to help support the podcast, please visit our Patreon.