Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #42: Ben Jones
140 Plays9 years ago
Welcome back to The PolicyViz Podcast. I’m happy to have Ben Jones on the show this week. Ben writes about data visualization on his popular blog Data Remixed and also helps run Tableau Public. Ben is a prolific writer, blogger, speaker,...
The post Episode #42: Ben Jones appeared first on PolicyViz.
Recommended
Transcript
Introducing Juicebox: Transforming Data into Stories
00:00:00
Speaker
This episode of the PolicyViz podcast is brought to you by Juice Analytics. Juice is the company behind Juicebox, a new kind of platform for visualizing data. Juicebox is a platform designed to deliver easy to read interactive data applications and dashboards. Juicebox turns your valuable analyses into a story for everyday decision makers. For more information on Juicebox or to schedule a demo, visit juiceanalytics.com.
Meet Ben Jones: From Background to Tableau
00:00:37
Speaker
Welcome back to the PolicyViz podcast. I'm your host, John Twabish. Thanks for tuning in. Very exciting guest today. I have Ben Jones from Tableau Software joining me. Ben, welcome to the show. Hey, thanks, John. Great to be here. How are you, friend? Doing well, thank you. Busy traveling all over the place, I see? Yeah, a little bit too much. Well, we all travel too much, but at least you're getting to go to some fun places.
00:01:01
Speaker
I was going to get some miles out of it that at some point, in theory, I should be able to use. To be able to catch it in, right? Yeah. Why don't we start by having you introduce yourself to listeners who may not be familiar with you or, of course, your great blog, Data Remixed.
00:01:16
Speaker
My name is Ben. I blog at DataRemix.com. That's something I've been doing for about five years now. That got me in a little trouble because I ended up getting hired by Tableau. Now I work there and help manage the product marketing team for Tableau Public, which is our free product that we put out there.
00:01:33
Speaker
for anyone who wants to tell a data story to the world. So that's a fun gig that lets me interact with people like the journalism school students here at University of Mizzou that I'll speak to in this next week. And so that's really kind of an honor to be able to talk to people that are telling the stories of our time using data and increasing numerical literacy really amongst the people around us. And so I also teach data visualization theory at University of Washington. It's my school class there.
00:02:03
Speaker
So that's me. Nice,
Tableau Public 9.3 Launch and Global Use
00:02:05
Speaker
nice. So let's talk a little bit about Tableau Public since it's a super active community. I actually have been attending the Tableau user groups here in DC the last few months. So what's new with Tableau Public? What do you guys have coming out? Yeah, good question. So we just launched Tableau Public 9.3. And for those of you who know Tableau, it's the same.
00:02:28
Speaker
Visualization software with some new features like the ability to union so when you have multiple data sets That all need to be combined together think of them stacking them vertically like maybe you got one data set for January one for February one for March and you really want to put them all together So there's a really fast and easy way to just you know to do that now using Tableau public and also we see such a global uptake of this product and
00:02:53
Speaker
We see people making visualizations and publishing them all over the world from Brazil or even places like Belarus, you name it, surprising places maybe you wouldn't expect. But we have a commitment to really try to improve things like mapping for different geographies. So we added zip codes for
00:03:10
Speaker
thirty seven european countries and uh... and japan as well and india districts and things like that so just really rounding out a lot of the mapping we updated the u.s. census data to twenty sixteen you know just in time for the presidential election so some new things going there on the feature side of things and then just uh...
00:03:28
Speaker
As far as on the community side of things, it's just amazing how much people are just continuing to join this group of people trying to make sense of the data and talk about that and have this dialogue. We just are doing a new what we call iron vis contest, so that's going to happen. That's great to see a lot of creative people start to work with some of the same kinds of data sets and you get a real
00:03:51
Speaker
Appreciation for what can be done and what kind of talented people are out there. So it's a fun time of the year for us to see those submissions So those are some new things good good good So as with any sort of software product when you have a new release people get upset sometimes about different types of visualizations now one visualization type that you have written about and you and I have disagreed about our word
The Debate: Pros and Cons of Word Clouds
00:04:13
Speaker
clouds and
00:04:13
Speaker
So you have a great post from last October that talked about a whole bunch of different things, including how we should be more responsible when we critique, how there may not be sort of hard and fast rules when it comes to dataviz. But one thing that stood out in that post that you're being a little provocative, I think, was, hey, word clouds are not the end all, not the worst thing in the world. So I think we need to kick off the great word cloud debate of 2016.
00:04:39
Speaker
All right, so why don't you give me your rundown of your pros for word cloud.
00:04:44
Speaker
OK, so let me start by saying, I don't think word clouds are great all the time. I just think it's in everyone's best interests if we keep the solution set as large as possible. I also think there's a misconception that all data visualizations should be the most accurate possible choice at any given time. But what I've seen and what I've experienced is that a lot of times, a lot of precision isn't necessarily required.
00:05:11
Speaker
An example of that Don Norman gives in his book, The Design of Everyday Things is let's say you want to know what temperature it is because you're going to be going outside and you want to know if you should wear a sweater or not. So in that scenario, it doesn't matter if it's 52 or 55 degrees Fahrenheit or 55 and a half degrees Fahrenheit, right? Like in either case, no matter which of those three it is, you're going to be wearing a light sweater. So the task you want to get done is very important. And so is the audience. So let's say you want to show a visualization.
00:05:40
Speaker
What is the audience familiar with? What are they comfortable with? What do they know how and are trained to make sense of? I use a simple example of the 100 most common passwords. I used four different visualizations to try to show what they are. In my estimation, if I were just presenting this to a crowd of people and wanted them to get a general sense of what is in the set of 100 most common passwords, the word cloud was the best one because
00:06:07
Speaker
Yeah, it's hard to tell that password occurs 26% of the time more than 123456, for example, but it's the only visualization of the four I created that includes all of the passwords in the set of 100. The other ones just weren't complete in that regard.
00:06:26
Speaker
So questions like, does the audience have time to interact? What's the importance of the decision they're going to be making with regard to using this visualization? Those things make a big difference. You can come up with scenarios and cases where some of these visualization types work and they're on the list of
00:06:45
Speaker
you know, quote unquote bad visualization types. And I think we've been suffering because we've been afraid to use them since some people don't like them out there. So I just don't believe that, you know? That's my take on it.
00:06:59
Speaker
We, whoever we is, I don't know who we is, but we tend to take sort of a hard line sometimes when it comes to bubble clouds or what have you. I just think for me, the issue with the word cloud is that the sizing of the word, the
00:07:17
Speaker
character count of the word. For me it just distorts the view and I'd rather, I'm not ruling out something like a bubble plot you have in the post that I'll link to on the site. You have a bubble plot here, you have a tree map, you have a bar chart and the word cloud.
00:07:30
Speaker
You know, for me, I just still don't get like a great sense of the difference of the size, you know, the frequency when I look at that word cloud, which is I do get that from the bar chart. No, I will say, however, that, you know, there's a great quote from Amanda Cox from the time she says, you know, a world without bubbles or a world without circles is a world without joy. Right. And so there are these aesthetic things. There are these aspects of visualizations that, you know, can draw you in and keep you there. And the word cloud is certainly
00:07:56
Speaker
It's obviously a little different, a little more maybe aesthetic than a bar chart. I don't know. Yeah, so it's interesting you bring up the bar chart. So actually, of the four, yeah, I actually think it's the worst for the scenario I'm specifying again, which is
00:08:12
Speaker
I'm in a big presentation room, lots of people. I want them to understand that people use really stupid passwords, and here are the ones that they most commonly use. The bar chart only shows 16 of the 100, because you have to scroll down to see it. Critically, it doesn't include the word Batman, which I think is a big bummer. Probably use that. The other three all do. The other three all see it.
00:08:35
Speaker
Yeah, but let's say I wanted to put on a tablet and give to someone, and I wanted to show a timeline of how the usage of these passwords has changed over time. Well, maybe I might go for a bubble chart there because, hey, the bubble is a great way to filter, isn't it? If you're using a finger, that's a great landing pad, and then a bar that gets smaller and smaller and smaller that may be very hard to touch. So again, it depends on the scenario. And so I'm not a fan of saying, well, this is a bad chart type.
00:08:59
Speaker
Now, does it depend on how many letters are in the word and does that make it harder to tell if one is slightly more or less than another? Yeah, like if I look at it, I can't see if there's more Batman or Harley and they're very close in size. But again, maybe that doesn't matter. Maybe you're going to wear a sweater.
00:09:17
Speaker
whether it's 52 or 53. Maybe that precision isn't important for a specific scenario. It may be for others. You're in security, and you try to design some algorithm to optimize password protection or what have you. Yeah, the exact fraction of percent is important. But then I would argue, well, then make a table or have those values needed, even a bar chart. And when that amount of precision is required, it might not be good enough.
00:09:42
Speaker
Right. I mean, I think the other thing I'd say is that if I imagine this use case that you have where I'm giving a talk and I'm putting a word cloud up there, one of the things about putting a word cloud up there is that everybody does try to read all of the texts, right? And this is something that people have heard me rail against lots of times, which is putting too much text on the slides because people start to read it and all that sort of thing. Whereas if I created one of these other chart types and said, oh, I just want to highlight that X percent of people use Batman and Y percent use
00:10:11
Speaker
you know, Yankees, you know, then I would sort of focus on those pieces.
Choosing Visualization Types: Beyond Accuracy
00:10:16
Speaker
And maybe you could do that even with a word cloud, you know, sort of highlight those different words in different colors or something. I certainly agree that it's about your audience, it's about your use case. Yeah, I think it really does depend. I mean, at the end of the day, the concept that really I think is important to remember is something called a payoff function. So what we're essentially discussing is which visualization is better.
00:10:36
Speaker
of a set of visualizations. And so that's a question we're asking ourselves. Well, then, you know, if we were to approach that question from a rigorous point of view, such as they do in operations research, you'd have to try to develop some way of ranking them and some function. And so the question is, well, what factors are included in that payoff function?
00:10:55
Speaker
And I think anybody who's been working in the field of data visualization for any amount of time at all would say, well, the ability to comprehend the proportions accurately is a factor in the payoff function. I'd say it's the most important factor, right? Many, many times that's the most important factor. But I'm a big believer that it's not the only factor, right? Maybe like you said, maybe there's a humorous metaphor that really causes
00:11:16
Speaker
your audience to pay attention, like how much pie I ate versus how much pie I could eat, right? Showing that in a bar chart kind of misses the point, right? Frankly, showing a photograph of the pie is the way to show, and that happens to be a pie chart. The payoff function is something that I think includes a lot of different factors, and it's over simplifying the scenario to say, well, we're only going to go with the most accurate chart type.
00:11:41
Speaker
Now what you would end up with if you just
The Role of Interactivity in Data Visualization
00:11:43
Speaker
simply use that one factor in the payoff function is a really boring world where everybody has the same two or three kinds of charts for anything you ever do. And to your point, that kind of zaps the joy out of it. And I think sometimes there's such thing as good enough. Sometimes you get the accuracy that's good enough for the task.
00:11:59
Speaker
And why wouldn't you then pick a chart type that even maybe slightly sacrifices accuracy, but gives you a world of benefit in something like grabbing the audience's attention or leaving them with a metaphor that stays in their mind, right? Or makes them laugh. I mean, those are things that I think are important because we're talking about communicating to humans. We're not talking about communicating to robots.
00:12:22
Speaker
You know, the other thing I'd say just to shift the scenario a little bit, uh, instead of giving, uh, let's say instead of giving a presentation, I was writing a blog poster or whatever. I was posting the visualization online. We also sort of seem obsessed now with getting it all into one graph.
00:12:38
Speaker
And you have on your post, very simply, four graphs showing the same data. To me, it doesn't seem like we would have to always have one graph, that there are multiple graphs showing the same data, allowing people to interact in different ways. For example, I'm not a huge fan of state-level choropleth maps, because I don't think you get a... depending on the data and context and all that stuff.
00:13:01
Speaker
You don't always get to see the actual difference in the values, right? And Texas and California very big, right? But if you pair that same map with a bar chart, then you have sort of best of both worlds. People like maps. I get it. People like maps. They know where they live, and they get to focus in on that. But then you get to please the other group who just wants to know how big is this value of Virginia relative to California, and they get the bar chart.
00:13:27
Speaker
That's one of the things about a tool like Tableau and some of the other dashboard type tools out there that we have multiple visualizations showing up the same data allowing us to interact in different ways.
00:13:39
Speaker
Yeah, and that's where a lot of the times you could use one that doesn't have a degree of precision and include a table. Include a table in there or add a label. Those are ways to accomplish a few different tasks at the same time. I think this is going to help us get into our transition to our next topic because sometimes including a whole bunch of views together in the same screen can be a big challenge, obviously.
00:14:00
Speaker
I think you should include as many views as is helpful, but I've seen and I think we all have seen the tendency to maybe add more than is necessary just because it can be done. I'd say edit and remove until you get down to a list of visualizations in the view that I think are necessary and sufficient.
00:14:23
Speaker
So we were talking earlier before we came on about creating interactive visualizations for the sake of having interactivity. And I think a tool like Tableau that makes it so easy to create interactive visualizations and also the evolution of all sorts of tools of D3 and R and all these sorts of things.
00:14:40
Speaker
What is your take now on interactivity? Have we sort of gone too far and now everything's interactive because it's so easy to do and do we need to pull back? Or is it all good that we can have everything moving around and be able to click and hover all the time?
00:14:55
Speaker
I think interactivity is a tool we can use and I think it's common a lot of times. It's very helpful for example if you want to let someone drill down to the story in their own neck of the woods.
00:15:10
Speaker
Or if there's so much information in there that you really can't show it all at the same time, so it's helpful to give it to them in bits and pieces and then let them interact with the visualization in a way that lets them proceed through that story instead of overwhelming them all at the same time.
00:15:27
Speaker
So I think in general, interactivity is a very powerful thing to be able to include in a data visualization. I just don't think I agree with you. I don't think it's always necessary. And I think this transition to a world where more and more visualizations are being viewed on tiny phones has taught us that we can shed some of that. It's not always needed. And a lot of times you can get away with a static view or maybe a GIF or something else that's a little more hands off.
00:15:55
Speaker
I think that we're challenged right now because we don't really know how to do interactivity well on the phone. There's a lot of people innovating in this space and some great examples coming out, but I'm confident that interactivity in and of itself is a useful tool and I think we're going to continue to. I think we have an opportunity to bring those types of interactive elements to the phone in ways that
00:16:18
Speaker
makes sense to people in ways that people find a joy to interact with when they're sitting on the train commuting, right? As opposed to what it is right now is sort of can be a very confusing experience or it can be clunky and frankly it can not work, right? It can be a broken experience and so there's a lot of that right now out there.
00:16:37
Speaker
And I think it's because we've only seen the last couple of years here this huge shift of readership to mobile. It's a good thing. I think it's just challenging us. In some ways, it makes the field more exciting, right? Because we have a challenge now, which is to next generation coming up. That's what they use. That's what they're comfortable with. That's how they like to consume information. And so we got to do what we can to make sure that they can interact with data on their phones. And I think we're getting there. I just think we're not quite there yet.
00:17:01
Speaker
Yeah, I think I like the way you talk about the joy of interactivity. I think there is a joy, like a really good interactive visualization. We could obviously name tons of them. That where you do, you get joy out of interacting with it. I think where we tend to fall down is we have some sort of, I don't know, a line chart. And you're allowed to click and hover. And it shows you the data. It shows you the data point, right? But that doesn't really help you better understand the story. It doesn't help.
00:17:29
Speaker
There are use cases, I think, where if you and I are sitting down at a table and we're exploring some data set, that may be useful. I want to know what the values of this point. But if I'm writing a newspaper article or I'm writing a blog post, I just want to show you that this trend has gone up over time. And the mobile thing is a whole other issue. So I'm curious, as Tableau Public, have you guys started to try to figure out what are people doing with mobile when it comes to Tableau?
Optimizing Tableau Visualizations for Mobile
00:17:56
Speaker
How are they trying to tackle those challenges? Yeah. So we've seen this coming. This has been something we've noticed here for going back a few years. And so we've seen some people, actually some authors, it's kind of a nice thing. A lot of times they can be the ones to be creative and innovate. But we've seen some authors, for example, figure out how to use custom CSS styling to swap between a different version of the dashboard for the different form factors that their reader might be.
00:18:21
Speaker
approaching the visualization with. And that doesn't work for everyone because of their various CMSs and so forth, but for a lot of people that does work. And we've been taking it upon ourselves really to look into this exact thing, mobile being so important for the people that we see day in and day out using Tableau Public. And so we're definitely looking at a lot of things here and have been working on a feature that we're pretty excited about hopefully releasing sometime later on this year that really lets
00:18:48
Speaker
People do just that, figure out what the experience is going to be like for these different devices and customize a single workbook so that they can deliver that experience. Even if it means taking some things out or making it look, maybe you want to show the big map on desktop. But then for a phone, maybe really the consumption needs to be in a list or a bar chart. And so you can customize that. That's the kinds of things we're looking at adding to the product here. Just so that out of the box, you can
00:19:16
Speaker
You know, using click and drag can create that exact kind of experience where a reader can get something that just works for them no matter what device they're using. And that would allow you to make those selections within a single Tableau workbook, or you'd have to have multiple workbooks.
00:19:31
Speaker
Yeah, so exactly. Instead of having a bunch of different files that you're using, CSS to swap, which actually some news organizations like, for example, the Sydney Morning Herald down there in Sydney, Australia, they do a great job at the CMS level of changing what content they're putting.
00:19:46
Speaker
But this would allow you to do that at a workbook level so that in some way it simplifies that workflow. So you just publish that one workbook and then you're able to take the single embed code and then the server is doing the smarts of figuring out what to serve up to the reader when they get there.
00:20:03
Speaker
That's interesting. Well, that'll be big, so we'll certainly look forward to
Tapestry Conference: Data Storytelling Across Fields
00:20:08
Speaker
that. Before I let you go, I want to ask you about the Tapestry conference that took place a couple months ago in Colorado. You and Robert Casara and Ellie Fields at Tableau organized. This is the fourth year? Yeah, that's right. That was the fourth annual conference, right. You want to tell folks about it? I'm not sure everybody knows about Tapestry. It's sort of a close-knit, small attendee list.
00:20:28
Speaker
Yeah, but it's a it's a great show. So yeah, we joke around and call it more like a retreat than a conference So we're really fortunate to have a great relationship with the folks at IRE and NYCAR In fact, I'll be hopefully dropping in on the IRE office here while I'm in Columbia, but
00:20:44
Speaker
We create this conference that gives people the ability to have a very small boutique experience. It's in a cool location. We park it right in front of Nykar so that journalists don't have to pay a lot extra to get there. They're already flying to the location and whatnot. So it's usually just one extra hotel night for them.
00:21:04
Speaker
And that's because we want to bring together people, not just journalists. We're looking at academics. We're inviting people and practitioners like myself or tool builders, you know, people like we had some folks from a variety of different tools that came this past year. But the idea is to say, hey, can we get everybody together, you know, or a group of people together in one room and have a discussion of talking about data stories on the Web and what that looks like and where we are right now. And so we had some great presentations this past year by folks like Enrico Bertini, you know, fellow
00:21:34
Speaker
Data podcaster and he showed off a tool that he was creating that really allows you to do interesting Visualization searches of review data and whatnot and then we get Jessica Holman up there from the University of Washington who's researching Really fascinating ways to make comparisons like if you need to go from here, you know to the 7-eleven down the street like what's that like in
00:21:57
Speaker
in terms which you can we've, you know, number of football fields or something like that kind of a thing, right? Making those comparisons. So that was another example of a presentation, you know, we pick a left field speaker, so someone who challenges us to think differently. So this year, we had Nick Susanis from the world of comics, because you know, in a lot of ways, that's kind of the same thing as a story that's being told in different views.
00:22:18
Speaker
Right. Helps us and makes us think about different ways. So we had a bunch of great presentations this year. They're all up on the website tapestryconference.com and we're our YouTube page, as you can see them. And we stream live too. So for the folks that can't either make it out to the location we are at or want to view it as it's happening, you can just plug in and watch the live stream. So it's kind of become a little, yeah, there's a lot of repeat attenders, you know, people, offenders, tendering.
00:22:42
Speaker
Well, we're looking forward to having you come back after a great presentation you did a couple years back out there in Annapolis. It's kind of turned into, feels like more like a class reunion at this point. It's a neat thing. It's very rare that you go to an event now where everybody's in the same room tweeting about the same thing.
00:22:58
Speaker
You're always picking a track or trying to hustle between this conference. We take it all out and it's just let's sit down altogether and get to know each other. There's 120 people-ish, and so you get to know all the faces by the time you're done. That's what's neat about it. I attended the first two, and Scott McCloud, who gave the closing keynote at the first one in Nashville, I think for me was probably the best presentation I've seen live. It was just a great talk.
00:23:26
Speaker
And you're exactly right. The people are sort of doing different things forces us to sort of think about how we use data and how we use visualizations to, you know, I hesitate to use the word story, but I'll use the word story here. You know, that's a whole, we'll have a whole other show you and I can talk about stories. Well, we, that's how we, we always, we love also to the person that's going to shake it up a little bit. So last year we had Kim Ruiz come and talk to us about, you know, how she hates the term data storytelling. We loved it. You know, it kind of,
00:23:54
Speaker
We get a little high on data and storytelling and it all becomes all great but sometimes you need to be reminded about what it's not good for or we had a great talk by Eva Galena Rosenbaum this year about what if you don't have data or if the data is so sparse that you really can't say much. How do you communicate that well? We try to include things that rattle our cage a little bit too. It's not just a bunch of congratulating ourselves about being great data people.
00:24:20
Speaker
Wait a minute, what are the counterpoints here and what are the things we need to remember so we cool our jets a little bit and get back to realizing that there are some limitations to this form of communication?
00:24:32
Speaker
Great. Well, really good, Ben. Good luck with your travels. Thank you. I'm excited to see what happens with Tableau Public. The social or the mobile platforms will be really interesting to see. So we'll be checking that out. So thanks for coming on the show. This has been really interesting. I appreciate it, John. Thank you so much for having me. Thanks, everyone, for tuning in. This has been fun. Please let me know if you have any comments or questions. And of course, rate the show on iTunes or your favorite podcast provider. So until next time, this has been the policy of this podcast.
00:25:13
Speaker
Again, thanks to our sponsors, Juice Analytics. For 10 years, they've been helping clients like Aetna, the Virginia Chamber of Commerce, Notre Dame University, and U.S. News & World Report create beautiful, easy to understand visualizations. Be sure to learn more about Juicebox, a new kind of platform for visualizing data at juiceanalytics.com. Also, check out their book, Data Fluency, found on Amazon.