< Transcript of PolicyViz episode #41 with Jon Schwabish interviewing Mike Pell >
Jon : Welcome back to the PolicyViz podcast. I’m your host, Jon Schwabish. I’m excited to welcome to the show this week Mike Pell from Microsoft. Mike is a Design Envisioneer, which is an awesome title at Microsoft in a group called The Garage. Mike, welcome to the show.
M. Pell : Hey, Jon. Thanks for having me on.
Jon : Great to have you on. I want to start by maybe having you introduce yourself and talk a little bit about your background. Then also maybe if can talk a little bit about what you do in The Garage at Microsoft, which in addition to being an awesome personal title, is an awesome name for a group to work in. It just sounds like you’re hanging around with friends building cool stuff. Let’s start there.
M. Pell : That’s exactly what we do, Jon, if you put it that way. The Garage is really a tiny team, but it represents a giant community around Microsoft worldwide, where we help employees and small teams really follow their passions, whether it has anything to do with their day job or not. We’re trying to help people get their best ideas out in front of other people who can help them develop the idea, and iterate. We collect data and help people move their ideas forward through garage.microsoft.com
We also do lots of internal hackathons. We’re helping to change the culture (of Microsoft), as much as we can. I’m sure you’ve been reading a lot about how Microsoft has changed in the last two years since Satya Nadella took over as CEO. I think The Garage has quite a bit to do with some of that cultural change.
Jon : You’ve been in this space of information visualization, computer graphics for a while. Can you talk quickly a little bit what your career trajectory has been?
M. Pell : Absolutely. It’s interesting to note that I started off as an artist. When I was going to school for art, I started programming almost by accident. At the time, I thought I could probably make more money doing this programming thing than with art. Luckily, I’ve always been able to do both the design side and the analytics and the coding parts.
I started off way back at the dawn of the Macintosh by founding one of the first Macintosh software companies called Beyond Inc., with Stuart Davison. We learned everything you could learn about running a software company – those were the good old days.
I’ve always been involved in things that were information-based and visual. Over the years, I’ve worked for companies like Adobe, lots of Silicon Valley start-ups, and now Microsoft for the last 15 years. All along the way, I’ve been very, very fortunate to be able to stay in the both design and coding types of jobs.
In The Garage, a lot of the things I do are to help teams assess what type of experience they’re trying to provide for people – whether that’s putting out a Maker recipe through open-source, or an Android or iOS app, or even creating a web service like the new one – Fetch. I don’t know if you saw Fetch; it’s very funny. You can feed it a picture of your dog and it’ll tell you what breed the dog is. Or if you put in a picture of a person, it’ll tell you what kind of dog the person is. That’s the kind of stuff that we do in The Garage, it’s very fun. So, you’re right, we’re having a blast.
Jon : Nice. You’ve been doing this a long time and you’ve seen the transition, as it were, from different types of interfaces. I know you’ve also been working on this idea of how to look at a visualization, or at some data, and get the bottom line, or understand the story right away. I think we were talking about earlier, what you like to call the moment of clarity.
This is a really interesting topic; I think we should spend some time talking about it. How do you define the moment of clarity, how does it differ across the different platforms and ways in which we visualize and collate data?
M. Pell : Specifically in visualization, there’s a point where we can put together any type of visualization on the spectrum, whether it’s something very factual or something that’s even data art, and identify the moment where the person who’s consuming that piece will either get it hopefully as the author intended or they’ll struggle with what was meant by it. In the time that you’re looking at a piece, whatever it may be, there’s a particular moment where most people will say they got it. It’s the a-ha moment.
As a Designer, or yourself of course, as somebody who does a lot of presentations and talks, you have to try your best to get your points to be very clear. Over my career, I’ve tried to practice what I call radical simplification, where I was getting rid of a lot extraneous things and focus on the clarity aspects.
The fact is people are very, very different. Every situation is different. What emotional state are you in when you’re looking at this. What’s your background? How deep have you been exposed to any of the information before?
The moment of clarity varies from person to person, situation to situation. I’ve always had this hypothesis that we could do a much better job of engineering the moment of clarity, as data scientists, Designers, coders, and people in this community. Whereas I think today we have a one-off approach. There’s a lot of room for us to improve.
Jon : Yeah, so how do you view the different ways in which we communicate data from your one-off graph, which may or may not be static, to a big interactive data tool? How do you get people to that moment of clarity when you have such different ways to communicate data and communicate information?
M. Pell : That’s an excellent question, and that’s exactly the question I asked myself. I’ve been collaborating over the last six months with a guy from the London Business School named Matteo Visentin. Matteo’s a visual designer, but he’s also a behavioral scientist. We’ve been exploring ways to measure and test how people react to the way that data is presented. I don’t want to try to make this into something that’s a template, because I don’t think you can frankly. But, I do think we can put together a set of principles or guidance that can help the authors, individuals and the teams put information together, or the pieces together, to help people get to the moment of clarity perhaps more quickly or efficiently than they have in the past.
There are so many different types of visualizations. There’s the very stock charting and things that you’d find inside of Tableau, or Power BI, or whatever. There’s also pieces that are much more thoughtful. You had Kim Rees on a couple of weeks ago. Certainly the pieces that Periscopic does are quite thoughtful and take a little bit to digest in some ways. Then there’s other pieces that are just purely meant as art. They’re based on real data but it might take some time to soak in, to understand.
For any of those things, the methods that we’ve discovered that you can use really vary from person to person, to the point where we feel like we have to offer multiple ways or multiple dimensions that you can consume this information in. That’s sort of been the key for me – there is no such things as one size fits all as far as moment of clarity goes.
We, as tool builders and platform builders and artists and Designers and data scientists, can do a much better job out of presenting information in ways that allow the person to go in at the altitude that they’re comfortable with. Whether it’s high level summary, low level detail, or how broadly or how narrow they choose to consume it.
Jon : Right. I think that last point is probably part of the key. In some cases, you are putting out a visualization that is for a specific audience. Think of an academic journal. You write your academic journal article. You know that the reader is sophisticated – well maybe a sophisticated reader who knows the field; they’re reading the article for that reason.
Then you have other pieces where you’re trying to hit multiple audiences, or a broader audience. How do you balance trying to meet the needs of all these different audiences with different interests and different skills and different sophistication? How do you get that moment of clarity for different kinds of users? You’ll probably just have the answer for this. (laughs)
M. Pell : Well, I kind of think I do. In the long run, the tools and platforms that we use will have the ability to allow you to construct pieces – I’ll just call them pieces of content, whether they’re charts of narratives, PSAs, whatever – that allow the reader, the consumer, to view them and experience them at different altitudes. As the author, you may pick one of those altitudes, let’s say, high level summary, to be the default. But there’s enough data to allow people to drill in. There’s enough cross references and linkages to allow them to explore the periphery and related topics.
I think that in the long run, the tools and platforms are really the things that will allow us to build what I call smart information. These pieces are very fluid and much more malleable in the way they can be consumed. In the short term, I think this comes down, unfortunately, to the authors.
We have to stop producing dead pixels, you know what I mean? We have to stop making things that can’t be explored and prodded and drilled into. I’m not saying you have to provide all of the capabilities of Tableau or something with each piece, but you need to have the ability to consume it in the way that’s right for you. That unfortunately in the short term is going to be on the shoulders of the authors of these things.
Jon : I’m curious, for a while there’s been interactive visualizations, but they’re really static visualizations that have some sort of interactivity just layered on top. It’s a chart with three lines on it where the visualization allows you click a point or a line and it highlights. I wonder whether you think that level of interaction actually helps people get this moment of clarity or if it’s just gratuitous, and going forward whether you think that type of interactivity is not going to be as common? Is it not going to be used as much because it doesn’t necessarily help people better understand, but just allows them really to just play with something?
M. Pell : Yeah, you’re absolutely right. I’m pretty against what I call putting different dimensions on rails. Isolating a few variables and letting people move some sliders around. That’s fine if it’s going to help you to get a little bit more of the detail in a piece. Really, it comes back to the design. This is all about people. It’s about what did the author or the team thinks the outcome of this is going to be. What were they trying to get across? Were they trying to be completely factual?
Let’s take, for example, how the New York Times is going to present an election result. They would like to be as factual as humanly possible in the presentation of that information, versus – well let’s take something like Periscopic’s gun piece that to me is more like a PSA. There’s a message there – intentional. They’re not hitting you over the head with it, but it’s pretty clear what their point of view was. As the authors, as the creators of information, we really need to focus on what we’re trying to get across.
I’m very fond of saying “just tell me”. When you’re designing a visualization, just tell me. Don’t make me sit there for minutes or hours trying to figure out what you’re trying to tell me because of the way that you’ve chosen to depict something. Oftentimes, we have been skewing towards beauty or visual interest rather than being clear.
Being clear is hard. You know this. You have to present all the time. It’s very hard to be clear and concise in a short amount of time. Yet, you sort of have to isolate that. You need to figure out what is it that’s the most valuable thing to get across to people. How can you do that without ruining the aesthetic or the artistic integrity of a piece that you’re working on?
Jon : Where do you think that line exists between the clarity of presenting information and design? I’ll give you an example. I was re-reading a Medium post earlier today by Martin Wattenberg and Fernanda Viégas. I think it was early last year, critiquing the data visualization field. One of the examples they showed was a sort-of radial chart of things going on in the Middle East. You had ten or so countries going around a circle in a big radial chart. Alberto Cairo from Miami had redesigned it as more of a dot plot or a column chart. It got rid of the circles part because his argument was, well you don’t really —
M. Pell : Circles are evil. (laughs)
Jon : Circles are evil, right. You see these sorts of things all the time. Let’s get rid of the circles and let’s use the bar charts. Sometimes there’s an aesthetic beauty to it or something that’s attractive, but it’s not necessarily allowing you to dig in and see all the detail and all the minutia of the data. I’m curious, how do you think about the balance between the aesthetic that pulls the reader in versus maybe more something a lot clearer because it has bars instead of circles?
M. Pell : Well, quite often I talk about being beautifully clear. I completely agree with Alberto most of the time about trying to get back to the most clean and accurate representation of data as you can. But, you have to have some form of aesthetic or visual interest or people will just blow by it. Why would you even bother to stop to look at another line chart? Who cares? That said, there is a lot of room to introduce emphasis through call-outs, through the way it’s laid out, through helping people focus on the important parts of the chart.
I came across something on Flowing Data. Nathan Yau was talking about a “taxi ride volume during the Super Bowl” chart that had come out in The Upshot. Super interesting, it was a great example of helping people focus on an important piece of information. Right in the middle of this histogram-looking chart, it says, “Katy Perry half-time show.” This is back when the Seahawks and Patriots were playing in the Super Bowl last year. At half-time, you’d see there are not a lot of taxi rides going on because of the very exciting half-time show with Katy Perry that people were watching.
The thing that was very subtle in this was at the very end of the game, all Seattle Seahawk fans remember very horribly, there was an interception to end it. Then all of a sudden, there’s this giant spike in taxi rides after that because the fans were all like, “Oh, this game’s over with.” But, you know what? In that chart, I didn’t recognize the game ending moment because it all looked fairly normal. It was missing the most super-interesting thing to me as a Seahawks fan. There was no call-out for that. There was just a little thin gray line.
There was an opportunity there for the chart author to help me focus on the important parts of the data, without taking a point of view, without having an opinion about it.
I think we miss a lot of opportunities to point out the interesting parts of things. Maybe it’s not with circles, but the fact is, as Designers, as data scientists, as people trying to get this information across, we can’t be afraid to point what our insights tell us. A lot of times, people shy away from that. They’re trying to be too factual. They’re afraid to as some will call it embellishing. It’s not embellishing, it’s actually clarifying if you ask me.
Jon : Right, so you don’t let the data speak on its own; you help it out a little bit. You help the reader out a little bit.
M. Pell : Yes, because in this day and age, who has time to sit there and try to understand, try to dissect everything. It may not be your field of expertise – you may just be looking at something casually. Yet, the author would really like you to come away with something. Very often we miss that opportunity.
Jon : It’s interesting the way you talk about communicating data and information. You have this emphasis of talking about the user being a person, the reader as a person, the person who created it as a person. It calls back to an earlier episode of the podcast with Kim Rees and Mushon, where we were talking about empathy and data visualization. I wonder what you think about trying to get readers to really connect with the visualization and connect with the data and feel something.
I think there are different kinds of visualizations that get us to feel. Maybe in the taxi cab example we feel something because we all watch the Super Bowl, and a visualization on some microbiology thing that only four people in the world understand is a different type of thing.
What is the impact, or what is the effect of drawing out emotion and drawing out empathy? How important is that when people are creating visualizations and trying to tell stories with their data?
M. Pell : Hugely important, absolutely hugely important. As a Designer, I’m all about people and emotion and understanding what people are thinking and feeling, and trying to use it. As Kim mentioned a few weeks ago, even as a data scientist, she’s not opposed to trying to draw on emotion to get people engaged. This is hugely important. I am also a big fan of persuasion. A lot of people would say “there’s no room for persuasion in data visualization”. I would say that’s nonsense. People even like to call it data storytelling sometimes.
The fact is we should employ all of those methods and tools and all of those techniques that the great storytellers and playwrights and people who make movies have used for so many decades, to try to get whatever that is that’s important. As an author, I’m for whatever helps get my point across, so let’s use everything at our disposal.
As Designers, we’re not above lying, cheating and stealing to get our point across. I think the same holds true in any type of information communication, information visualization, or data visualization for that matter. We need to draw on all the talent and all techniques we have to get our points across. I don’t think you’re going to bias somebody too much, unless you actually try to do so. If you try to be completely persuasive and you’re trying to really overtly twist the data to support your point of view, that’s another thing.
If that’s what you want them to do then great, go do it. If you’re just trying to present information clearly, I do think there is a big need to draw on the people part of it, on the emotional part.
Jon : You touched on a couple of things there. Persuasion has this negative connotation to it, but visualizations are almost always trying to persuade people that the thing that they’re showing is true. The unemployment rate is going down. This thing is going up. There’s all levels of persuasion. I guess the difference is whether you’re being honest with the data; trying to persuade someone that your perspective is correct, given that you are using the data honestly. Then there’s persuasion where you’re distorting the data, either visually or in the way you’ve created the final product with the data itself and you’ve distorted it in different ways.
I’m also curious, when it comes to persuasion, when it comes to getting people to connect emotionally to visualization, how important it is to connect at the individual level. You’ve mentioned the Periscopic guns piece. One thing about the guns visualization is that it has a line for each individual in their data set. How important is it for a creator to convey those individual stories, and maybe not always aggregate into six bars? Show more of the individual observations, the individual stories.
M. Pell : Again, it’s always context-dependent. Whenever possible, I would always tend to try to connect with the individual person, whether it’s in the data or consumption. If I’m talking about a very big concept or very difficult world issues and I can’t make a personal connection with what you’re telling me, then chances are I’m going to move on to something else. Super-important for you, the author, to try to get me interested as an individual in what you]re talking about so I can dig in and get all the benefits of what you’re trying to do.
In some cases, I don’t think it’s possible to involve an individual person as well as the Periscopic piece did. There’s a certain point where you have to abstract things. So, why not use other elements. Not every data visualization has to be on a plain white background. There’s a lot of things that we can do from the design aspect – from sound and motion, to the way that it’s staged. All the cinematic elements come into play.
Even going back to that Periscopic piece as an example, the very beginning is very cinematic – dark background, things start happening very slowly, and then the data starts to draw itself. It’s a story unfolding. All of a sudden, I get the moment of clarity because at about the sixth person as it starts to accelerate it hits me. Wow, they’re talking about a ton of people dying and so much tragedy in loss of years of life. It gets you in a cinematic way.
Perhaps even the most mundane, straight up, data-pure type of representation could actually be presented to help make an individual connection with me the consumer. For example, as we were talking about before the show, things like SandDance use every single pixel on the screen to represent a unique piece of data, which may be a person.
Jon : For those that don’t know, SandDance is the new visualization project from Microsoft Research and The Garage. Do you want to talk about it real quickly before we wrap up for people who haven’t seen it or heard of it yet?
M. Pell : Sure. SandDance is a really great piece of visualization work done by Steven Drucker and Roland Fernandez at Microsoft Research that you’ll be able to play around with for free. They have been working on this for years to help people more easily explore data. To me, it’s a great example of a new genre of tool at our disposal called a data explorer. It’s super fluid in order to make it easy to move through data sets. You can look at your own data or play around with data sets that are already built in.
SandDance allows you to look at things very quickly, move on, explore, play what-if, and try different things out superfast in a way that we’ve never been able to do before. They’ve constructed this tool in a way that can be extended and are going to continue to iterate on it and experiment. It’s really about helping the visualization community try out new ideas, and hopefully tell better stories with data when they’re using that particular approach.
Jon : Great. Well, on that note of new tools to explore and trying to maybe be a little bit more aesthetic and beautiful in our visualizations, even with the boring stuff, Mike, I want to thank you for coming on the show. This has been really interesting.
M. Pell : Thank you, Jon.
[ Editor’s note: light editing of the transcription was required to make this more readable ]