Numlock Sunday: Philip Bump on How To Read This Chart
By Walt Hickey
Welcome to the Numlock Sunday edition.
This week, I spoke to Philip Bump of the Washington Post, who a few weeks ago launched his newsletter How To Read This Chart. I’m a longtime fan of Bump’s work at the Post; he’s a really compelling writer and an outstanding blogger and has been weaving data journalism throughout his work so organically for such a long time I wanted to have him on for a podcast version of the Sunday edition to talk about the state of the art and the newsletter.
We spoke about how charts took over the internet, how companies have managed to use increasing interest in data to suit their own ends in pitching mediocre polls, where the distrust of polling is coming from and the simple pleasures of weird charts.
This interview has been condensed and edited.
You have just started this newsletter called How To Read This Chart. It's probably, what, two months old at this point? It's already off to a really great start.
I appreciate it. That was the goal. Try and have it be compelling, and get people engaged. But yeah, I think about two months, which is sort of crazy how time flies.
In one of the early issues, you mentioned that you were looking for a combination of breeziness and topicality. Do you want to talk a little bit about what kind of motivated you to both start a newsletter in this space and also kind of what your angle is?
I do a lot of charts for my day job at the Washington Post. And one of the things that I have found is that often I will create a graph which to me seems pretty intuitive, but requires a little more explanation than I would've expected. That is absolutely a failing on my part, right? My job should be to make presentations of data that are simply intuitive enough that I don't need to have a lot of explication, but I also think that complicated charts can be visually interesting, and provide a lot of information once you dig into them a little bit.
What I wanted to do was create something, create a tool, in which I encouraged people to be more open to more complicated presentations of data. To offer up interesting, visually striking data visualizations that I could then walk through, and say, "Here's how this works. Here's why this is actually a smart way to present this, even if at first it may be somewhat intimidating." Over the long term, with the goal of having people just generally feel more comfortable with looking at data visualization, understanding how to pick out their own stories from it, and understanding how it can convey a lot of information in the way that words can't.
Yeah. I've enjoyed the approach and vibe of it a lot. I think that there are places on the internet that look at and evaluate and talk about charts a lot, and I think a lot of them will sometimes overengineer it. They'll be like, "Oh, look at this chart." And there are a lot of colors, there are shapes, there's a lot of stuff going on.
I think that what I've really enjoyed about yours is that you very much look at this as a process where there aren't really right answers. That there are just choices that are made. Do you want to kind of expand on how you view the process of creating a chart?
One of the things that I do too for my job is I do a lot of charts, right? So that means that I'm not sitting down and workshopping, A/B testing, different versions of charts. I'm like, okay, I'm doing a story on something that just broke. I need to make a chart. I did one that I included in the newsletter, which I thought was pretty good, which showed the evolution of votes for Supreme Court justices since the end of the Civil War, essentially. And so you start off, and I had different representations. I took all nine currently sitting seats and I sort of had little lines that wound their way through them with little nodes for where a new justice was added.
And so at the beginning, it's just, they're all by voice vote, and so that's just a little black dot. And then eventually you start to see actual votes for it, and first all the votes are overwhelming, like 90 to 10 or whatever. And then all of a sudden you start seeing them being pretty evenly divided. That was something that I didn't have a chance to workshop because I needed to get it up because this is when Breyer announced his retirement.
But I could then come back — and I didn't actually do this with this, I just sort of dropped it in a newsletter because I thought it was cool — but that's a good example of where you can say, okay, so how might you have done this in a way that provided more information? I got feedback from readers, like, this would be great if it's an interactive where you could mouse over it and see someone's name. That's absolutely true. I couldn't do that, in part because of the time crunch, but also in part because I couldn't then put it in the newsletter. There are all these boundaries and constraints. And one of the things I'm trying to do is get the voices of people who read the newsletter injected into the newsletters; I want people have feedback and thoughts to include that as well. Because part of the point is that I don't have all the answers.
Yeah, I like that a lot. I like it because you really are grounded in news. Your day-to-day job is all about kind of covering the news. That comes with different constraints than doing a long term; if you had spent a week making an interactive about that, it wouldn't have actually been as effective because sometimes just getting something out sooner is better.
No, exactly, you're right. I think too that one of the lessons there is that it doesn't have to be great, right? I mean you and I have experience in writing for institutions that have editors. But it is not the case that we simply write things and then let the editors clean it up. We try and write well the first time, right? This is a space where you don't have an editor. I don't have a graphics editor, right? Or when I do an interactive online, I don't have a data editor. Although I have sort of pitched that in the past as something that would be kind of cool. I mean, wouldn't it be cool if you had a lot of people who could do really basic code and then you could pass it over to a data editor, who could clean it up and make it, I mean, wouldn't that be cool? You could really expand the number of interactives you could do.
Yeah, FiveThirtyEight was a shop like that. That was a really good time.
There you go. But one of the things is trying to encourage people to just sort of jump into it, right? You don't have to worry about it being perfect. In the same way that I've been doing this long enough now that I can do something fairly quickly that is effective and isn't necessarily buggy. But the way you get there is by just sort of jumping into it and doing as much as you can, and having fun with it and making it. And here's the other thing that I'll say, I will put out graphics that I think are cool that I recognize are going to be complicated to people. You know why? Because I do this whole day, and sometimes I want to just have fun with it. And I think that's okay too in a lot of circumstances.
Yeah. And that reminds me, a few weeks ago you had a particularly neat visualization, I pulled it up, it's a keogram, right? There was a guy who put a camera on his roof. Do you want to talk about that one?
It was really cool. A guy I follow on Twitter retweeted it, it's this astronomer who lives in the Netherlands. And literally exactly what you said. He put a camera on the roof of an outbuilding in his house, and just pointed straight up with a fish eye lens and took a picture every 15 seconds for the entire length of 2021. And what results from that is you get this really cool pattern, this sort of hourglass pattern of when night falls — obviously less night over the course of the summer, more night when it gets to be winter. You can see how the sky is different colored at different times of the day. So twilight and dawn are much more blue because the sun isn't reflecting off the clouds as much. You can see when the moon is moving through. The phases of the moon are depicted in it.
There's all this information. It's just this little rectangle with this little black hourglass sitting on a blue field. But when you look at it, you can pick out so much stuff. And then he, of course, as an astronomer, he finds a lot of stuff that you don't necessarily see. He goes through it and he picks out the movement of the constellations through the skies. He has this giant version of it you can get as well. And just, it's fascinating. It's just such a good use of visualizing data. It's not visualizing data intuitively. It's just a picture of the sky. But all you have to do is orient it by time of day and by time of year, and then all of a sudden you have an infographic that really tells you a lot about literally how the world works.
I really loved it. Again, even as just like an art piece, it's really cool.
I feel like there's been more conversation about chart crimes than it has in the past. You've been in the game a while now, and have definitely kind of seen good visualization, bad visualization, malicious visualization at times. That also seems to be a theme at some points in your newsletter. How has the internet's interactions with charts kind of evolved over the past couple years?
Good question. I mean obviously it is something that's been powered by the connective capabilities of social media to a large extent. I think also there is a sort of nerdiness that emerges on social media more than in other places, just by virtue of the fact that it's grounded in technology, and age groupings and so on. And of course, you and I probably operate in nerdier circles than most people. Like my sister who spends a lot of time on Instagram, I'm not sure she sees a lot of chart crime tweets, right? Which isn't meant as a disparagement to her, she's just got different, more normal interests than you or I do.
Part of it is not only the fact that you have that connectivity, which obviously we're all familiar with by now with the internet, but you also have a lot more people who are trying to present information. One of the things that's been fascinating over the course of the past decade or so is the ways in which corporations have really latched onto the idea of using data presentations to sell stuff. And so they'll do like these shitty polls, of like, "We polled 14 Airbnb members, and seven of them found that Airbnb is awesome." And they'll pitch those to Forbes contributors who go on to money laundering crimes.
Alleged money laundering crimes...
I didn't name names! But there's this pattern wherein corporations have figured out that people are compelled by data visualizations, and by sort of weird esoteric data points. And so they gin these up, but what that also means is that there's this big influx of visualizations and attention being paid to these things. A lot of media outlets who are trying to move from being crusty old newspapers to doing more interesting things online, there is this new attention being paid to how you present data for a lot of different reasons. I think that provides a lot more people who are skilled in this to some extent, and a lot more people who are paying attention to it, which I think is probably part of it.
I love this point and I actually want to spend more time here just because it's rare that I get to be like, "Hey, you also see this, right?"
It is wild just at how regularly I will see press releases, I will see corporate blogs, I will see all this kind of stuff that will either be hinged on a poll that they ran because they would like to get into the press, and they realize that's a decent way to launder a talking point. It's almost interesting kind of watching some of the techniques of data journalism kind of get co-opted a little bit by some of these corporate actors.
You cover politics, and I'm sure that you see this time and time again in that space in particular too. You pointed out a LinkedIn poll a few weeks ago, that was just like, why are we talking about this? And this made it into what?
It's in New York Times for God's sake, right.
Is that a local newspaper? Where is that based?
I live outside of New York City, and yeah, a lot of my neighbors get it. Yeah, I hope you've noticed that I go out of my way to insult The New York Times whenever I can in the newsletter. Not out of any ill will, but just because it's funny.
There's one coming up in the one that's dropping this Saturday. Anyway! No, you're right, and part of that, I mean, bear in mind, you are familiar with political polls as well, but for years it's been a problem that you'll have political journalists who'll pick up internal campaign polls and treat them as serious, when those are the same thing, they're just marketing pitches using this front of data. You have that exact same factor which comes into play in doing politics reporting.
You have a lot of journalists who simply don't know how to spot bullshit, right? They just don't know how to be like, "I'm going to treat this with some caution." Instead they're getting pitched on, "Hey man, we have this new insight into what's going on." It's the exact same thing, it's just corporations have figured out how to do this.
And it costs very, very little money for them to do it. If they find a reporter who isn't particularly savvy in assessing the validity of data collection mechanisms, then they get a bite pretty easily. "Hey, I'm going to the exclusive. LinkedIn did this poll. And we found that X." And someone's like, "Whoa, that's cool. Thanks for the exclusive. Send me the data." And then you and I get it, and we're like LOL. What the hell is this?
Again, I'm not trying to judge. In the same way that if someone asked me to be an opera critic for a day, I would totally embarrass myself! I just don't know it. I don't know that world. But there's a reason why, if someone's putting on a crappy opera, they might want to have me come and do the review, right? Because they know that I'm not going to understand what's going on. And so I think we see a lot of that with corporate pitching in particular.
It's great that you point out again, this isn't to knock the folks who do it because maybe they're credulous of a poll, but they're also like they take no bullshit elsewhere. It's just kind of almost an expertise thing where knowing what's a rough poll, knowing what is a specious source of collecting responses, knowing what is, oh, the sponsor of this is perhaps a little sus. That is a skill.
I think that your beat at the Post has been really good at kind of calling some of that out in the past.
Yeah. I mean, I think it's fun, right? I mean it's fun to have to catch Tucker Carlson, who I do not assign any good faith effort on his part. It's fun to say, "Hey, actually this graph that you used is nonsense," and to call that out. I mean obviously his nightly audience is slightly larger than my newsletter audience.
A little bit.
You don't know. You don't know. My newsletter audience could be two million. I will say this, it's definitely a percentage of two million!
Rounded to the nearest two million, sure!
That's right. We're even. I think there's value in it, but also in recognizing how, through intentionality or accidentally, the people can offer you information that you should be skeptical of.
I felt like at the beginning of my career, maybe 10 years ago or so, the challenge was convincing people that there was value in writing about polls, that there was value in writing about this kind of data, that there was value in making all these kind of viz and putting this investment into it.
Now I feel like, battle's won, congratulations, high five, victory lap, champagne.
But now the challenge is like, oh, we've got to be smart about how we actually end up using it, because those tools have also been — whether it's through a profit pursuit, or whether it's through political pursuit — we won one battle and now, oh wait, there's actually a whole lot of cleanup that we’ve got to keep doing.
No, you're right. I think that there are a lot of ways in which data presentation right now is sort of a risky endeavor. From the visualization standpoint, it is obviously the case that we get a lot of dubious, questionable presentations that make their ways in front of us. And I think we're calling it out.
I think polling itself is something that we could spend a whole half hour talking about this separately. But I think polling itself is something, that the skepticism of polling has been weaponized in the same way that say skepticism of a coronavirus vaccine has been weaponized, right?
That you can pick out particular ways in which you can cast out in order to try and undercut the whole thing. It is much more common that we see that with polling, in part because you're always going to find a poll that gives you information you don't want. One of the things I thought has been fascinating in the course of the past several months has been this spate of new polls showing that President Biden's approval is quite low, and just the number of people, including the president, who are saying, "Oh, you can't put any stock in that."
And it's just like, come on, man. How are you — and I made this point in the newsletter — you are asking us to say, here is the best information we have on the vaccine. It's imperfect. There have been mistakes that have been made. But we understand generally what things look like and we need to have confidence in that.
And then on the same side, you're saying I ignore these polls because they're bad for me. It's bullshit. You can't do that. That's not how it works.
I think that we're seeing, and we saw Trump obviously spend a lot of time trying to downplay and denigrate polls that showed him doing poorly, and I think that's damaging. I do. Because polls are, for all the complaints about going to a diner and talking to voters, and yada, yada, yada, polls are talking to 1,000 people at one time, asking them more narrowly tailored questions, and not having a lot of space for follow up, obviously, but that's what polls do. I think for people that reject them out of hand, I think just moves us more into this space of uncertainty where all sorts of bad things can happen.
There are always good polls, there are always bad polls. I will just say, in your newsletter, How To Read This Chart, one pervasive idea that I like is that it almost comes off similar to like Wirecutter where it's like you're evaluating the different tools and trying to figure out what the best way to talk about it is, and how to use it and all that kind of stuff.
It's not like saying that the tools are invalid, or the tools don't work, or the tools always work, or this is a bad tool. It's acknowledging we're all coming from the position that this is a cool tool to have and a cool tool to use. Here are the best ways to use it.
And also every time you make a bar chart, I get a commission, echoing the Wirecutter model.
I mean, the goal is to some extent — and part of this is informed by Edward Tufte, who we are invariably going to get to as part of this conversation. He does these data visualization things and is sort of like the godfather of all this stuff. When I worked at Adobe as a designer back 20 years ago, I went to one of his seminars. And the thing that always struck me about that is that you're sitting through the seminar, and it's like the morning— Have you been to one of his seminars?
I've not, no.
So the morning you go through it — at least in Silicon Valley 20 years ago — but you sit through this really interesting assessment of how data visualization works. The O-ring story from the Space Shuttle Challenger, and all these various ways in which presenting data is important. And then you have a break, and then you come back in the afternoon, and it's all about how to make cool charts and PowerPoint. And at the time, I skipped it. I was like, I don't give a shit about making cool charts for PowerPoint.
But I realized it's brilliant from a business model because that's how he gets all these companies to pay to send people to his seminar, right? They go to seminar, and they say, "Hey, this is going to help me make better PowerPoint charts." And they're like, "Okay, fine. We'll pay the $1,500," or whatever. It's genius. But it's also value, that's the value he's added to people.
One of the things I want do with the newsletter as well is give people, to some extent, some familiarity with the tool sets that are out there so they can broaden their approach a little bit. I will also at times slide in little recommendations. There was this spiral chart that was in the Times that got a bad rap, and I found a link to how to recreate that thing using R, which, no one that reads the newsletter is using R to make data visualizations. But A, it's there in case they do and want to, but B, it also makes them feel like, okay, I'm part of this kind of cool insider club that knows how to make visualizations in a pretty advanced way now. Which then gives people the confidence to do so, to explore it, ideally, hopefully. I don't know how many more charts I've brought into the world through my newsletter. Hopefully one or two!
That's so smart. Because again, you never want to be in a course where you know everything. You never want to be in a course where it's completely out of your bound. You want to be in a course where you get a little bit of exposure to things that are a little bit beyond your grasp at the point and things that you have well in hand. I like the confidence that you put in your audience.
It's just really good stuff, man. Is there any particular thing that you've had a chance to write about that you feel proudest of in the past couple weeks?
I mean, the keogram one was good, just because it was not specifically news-centered. I don't know, just one of the things I try to do with the newsletter, which hopefully I succeed at to some extent, is to make it actually fun to read. It's super easy to write about a chart in a way that no one is interested in it. I'm not going to say it's hard to do it in a way that's a little more compelling, but I try and make it like I want it to be something where someone who is not really familiar with charts picks it up and at least finds it funny, even if they don't really care about charts or even if they don't really necessarily get the charts at the end of the day. I feel like I've done a pretty decent job with that. I think that that's the thing which I am probably happiest about is that people who I know aren't nerds on charts will come back and be like, "Hey, at least it's fun to read," which I think is a step in the right direction anyway.
Where can folks find it? Where can folks get a chance to subscribe? Where can folks find you?
Yeah. I'm on Twitter @pbump. The chart, I mean, I think if you search How To Read This Chart, I would hope it comes up pretty high in the results by now. If you go to wapo.st/readthischart, that'll actually take you directly there and you can sign up. Obviously it's free. I'd be very happy if you subscribe to the Washington Post, of course. But it's a free newsletter, it teaches a little bit about charts. And if you stick around after the break, we'll teach you how to make the excellent PowerPoint presentations, so you can send me $1,500 for this seminar too. It would be great.
Sign up for Numlock News and wake up to the best stories in data every day.
If you have anything you’d like to see in this Sunday special, shoot me an email. Comment below! Thanks for reading, and thanks so much for supporting Numlock.