Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #87: Elijah Meeks
173 Plays8 years ago
Welcome back to the PolicyViz Podcast. This week, I’m pleased to welcome Elijah Meeks to the show to talk about his work as a Senior Data Visualization Engineer at Netflix and the state of data visualization jobs. Elijah sparked a...
The post Episode #87: Elijah Meeks appeared first on PolicyViz.
Recommended
Transcript
Introduction to the Podcast
00:00:11
Speaker
Welcome back to the Policy Viz Podcast. I'm your host, John Schwabisch. And on this week's episode, I'm very pleased to have a senior data visualization engineer, Elijah Meeks from Netflix.
Role of a Data Visualization Engineer at Netflix
00:00:24
Speaker
Elijah, welcome to the show.
00:00:25
Speaker
Hi, John, thanks for having me. Great to talk to you. Of course, as most people, I'm sure who would listen to this show, know that you started a bit of a hubbub with the tweet a few weeks ago, followed up by a number of blog posts and some interviews on the Data Stories podcast and some other places. So I'm excited to dive in to talk about this date of the data visualization field.
00:00:46
Speaker
But I think before we do that, I'd like to give you a chance to talk a little bit about what you do at Netflix. I think for me, I'm just interested in what a data visualization engineer does at Netflix. And before I let you do that, let me just thank you for letting us download movies. That just gives me a lot of things to do on my commute to work. So thanks for that. But maybe you can start talking about a little bit what you do there at Netflix.
00:01:12
Speaker
Well, you're welcome, John. That is 0% of what a senior data visualization engineer does at Netflix is providing downloaded movies. I also don't provide the regular movies or any of that other stuff that you actually think of with Netflix. What I do at Netflix is I design and develop internal
Industry vs. Academic Roles in Data Visualization
00:01:31
Speaker
applications for
00:01:32
Speaker
analyzing the kind of data that you would expect Netflix to produce, which is data about how people become members of Netflix, how the Netflix application determines what shows and movies to put in front of you, how you are interacting with it, why you might leave, and those kind of things. And so analytics, the examination of analytics is a major
00:01:57
Speaker
aspect of the use of data visualization in industry. It's in contrast to what I did at Stanford which is becoming so long ago now that I feel kind of offering it up. At Stanford I did much more data visualization in service of research projects specifically in a field called the digital humanities which is the application of traditional quantitative techniques like GIS or network science or natural language processing directed toward
00:02:25
Speaker
humanity subjects like English, literature, philosophy, history, that kind of thing. And how many people do you have on the team at Netflix that does data visualization or the sort of work that you do? Well, there's different ways that organizations develop their data visualization talent. So some orgs like Uber have dedicated data visualization teams.
Internal Use of Data Visualizations at Netflix
00:02:51
Speaker
Netflix doesn't. We have data visualization engineers embedded in different teams that are directed toward different parts of what Netflix does. My team consists of myself and then Susie Liu, another data visualization engineer, and another couple of people who are more focused on more traditional JavaScript software development.
00:03:15
Speaker
And my role is developing, typically with D3, data visualization applications, building a bit of data visualization infrastructure, and then more general UI design and development for the applications that these data visualization products are embedded in.
00:03:34
Speaker
I see. So the visualizations themselves that you're creating, those are visualizations for other people in the Netflix organization to better understand how the algorithms are working or how other aspects of the business are working. Is that right? Absolutely. So we build products for internal clients and sometimes those clients are the scientists who are interested in designing better algorithms. Sometimes those clients are the non-member teams that try to increase membership at Netflix.
00:04:02
Speaker
And sometimes those clients are content or product teams that are more focused on developing new content or understanding the success of the content that we've either produced or licensed or the general health of the Netflix application itself. Right.
State of the Data Visualization Field
00:04:19
Speaker
And is there a separate data team that works on building applications to pull all the data from the streams and from users and gives those to you or is that part of the responsibility of you and your team as well?
00:04:32
Speaker
The larger tech companies in Silicon Valley, you're going to see massive resources dedicated to processing the data that comes through. I can't even remember the crazy amounts of data that are coming through anymore. We develop new technologies and we have lots of folks who are obsessed with the kinds of data stores and query engines and things that allow you to access this stuff in a really fast kind of way.
00:04:58
Speaker
And so most of that is handled by what are referred to here as data engineers. And eventually we'll step into write queries and process data as it comes into the application or maybe on our own for prototypes of certain data visualization products. But the productionalized version of these queries, the productionalized version of these sort of data pipelines will be written by people who know that area better than I do.
00:05:27
Speaker
Okay, very interesting. I mean, it's certainly interesting for me coming from sort of the other side of the country in a totally different sphere or different industry about how others are using data visualizations, which leads us to the next point of discussion, which is the thing that you've been talking about for the last few weeks, which is sort of the state of the data visualization field. Is there a data visualization field?
00:05:51
Speaker
Are people sharing skills and tools and products effectively and efficiently? So maybe to kick this off, you can try to maybe summarize the argument that you've made. And I think I may push back on a little bit, but I'll let you sort of kick things off and then we can fight from there.
00:06:10
Speaker
Sure, sure. It's been a month and a half now, I think. A long, long time ago, I tweeted something about how people leave data visualization and how it was a very sort of hamlet moment. There's something rotten in the state of data visualization. I thought that it would be acknowledged or found interesting by a couple of people and it was actually found interesting by
00:06:36
Speaker
quite a number of people in the field and that led to some vigorous debate and then my moving forward with a survey that I'd wanted to do for a while. And the results of that survey sort of backing up some of the ideas that I'd had.
00:06:50
Speaker
In creating that tweet, but also some of the pushback that folks had responded with, especially around like this idea of whether data visualization I think is a skill or is it a profession?
Complexity and Styles in Data Visualization
00:07:01
Speaker
I think everyone acknowledges that it's a skill. The question is, are there real data visualization positions?
00:07:08
Speaker
that are dedicated to data visualization, or is that a misapprehension of what data visualization is? Is data visualization just an important skill in use by other roles like UI engineers or data scientists?
00:07:22
Speaker
And so, you know, I mean, we got pushed back from prominent members of the community. I think that part of it was pushed back on this idea of there being a data visualization profession. So folks like Stephen Few said, no, there isn't a data visualization profession. It's a skill. And then part of it was pushed back on the presumed motivation for this kind of thing.
00:07:44
Speaker
Because I've also argued in other places about how we need to do more complex data visualization. We need to use more complex methods in practical data visualization problem spaces. So in industry you see quite a conservative bent and you have this
00:08:03
Speaker
desire on the part of your clients and industry to use methods that are good for numerically precise data visualization, so line charts and bar charts. I've argued in a number of different places that we should see more hierarchical databases, we should see more network databases because those kind of patterns you presume are relevant. I think I also saw pushback from folks who find it presumptuous, I think.
00:08:26
Speaker
that maybe I just want to do that because it's fun to do that kind of data visualization. And I'm not having fun making bar charts and line charts. And that really it shouldn't be about having fun and pushing the boundaries as far as data visualization goes. Really what it should be about is delivering insights and making sure that your visual display of information is as clear and accessible as possible.
00:08:49
Speaker
So there's obviously a lot to unpack there. And I think there's sort of two threads to this discussion. There's the discussion of the practice versus the career, as it were, or the position, as it were. And then there's the style of the visualizations or the types of visualizations that we're creating. I think when I first really got interested in this field, tower infographics were the big thing. And very quickly, those have sort of gone away. And then it was make everything interactive. And even now, we're starting to see people starting to pull away from some of the interactivity.
00:09:19
Speaker
But let's start with the first part because I think what struck me the most about the first couple of blog posts that you wrote was that, yeah, I didn't feel like there is a person who is the data visualizer. And again, I think what you've noted also is that we all have our own perspectives on the type of people that we work with and the people who we are. But I've always found that people who like or are good at creating visualizations, that's not all they do. They're doing research.
00:09:47
Speaker
they are doing the UI and the UX, they are doing the design as well. And that there isn't just a person who just does data visualization, also because the visualization part sort of sits on the bedrock of using data.
Data Visualization vs. Data Science Roles
00:10:00
Speaker
And so I worry about being able to create visualizations where I don't understand the data or not well versed in the data, even though people do that a lot. Right, absolutely. I mean, it's an interesting formulation, because it's like, if you think about it,
00:10:14
Speaker
You might then use the same kind of argument to say that there's no such thing as a data scientist, right? Because like a data scientist in a place like Netflix, while they'll query databases and such, they'll still rely on people who understand the fundamental summarization and processing of the data that they're working with better than them. There's this problem, I think, that we struggle with, which is trying to identify
00:10:39
Speaker
what's the necessary background knowledge versus what's the sort of specifics of what somebody's doing. So of course, I'm still going to write queries and deal with data and think about patterns in the data. But the majority of my time is spent thinking about ways to represent data that are going to enable the people using that data to make new decisions and to understand as much of the phenomenon as possible as they're examining.
00:11:09
Speaker
And I do think that fundamentally there's enough work in a role like that to justify the term data visualization engineer in the same way that I think there's enough work in processing data to justify the data engineer role. Even though a data engineer is gonna do some statistical modeling, they're gonna do some data visualization, a data scientist is gonna do some data processing, they're gonna do some data visualization. In the same vein,
00:11:37
Speaker
Yeah, I do UI work and I write queries and I even use statistical methods to try to aggregate and summarize the data. But I don't come up with research agendas. I don't sit here and say, I think this is what we should be doing with this AB text. I don't sit in STRAT meetings and try to figure out how to adjust the Netflix product to make it more valuable to users instead.
00:12:02
Speaker
Those people who do that come to me and say, well, we've been doing that. And these are the results. This is what the analytics on the website say. And this is how we're looking at it right now. What's a better way to look at this so we can take into account more factors? And that's where I add value to the process. And I think that that's a real role. And in fact, I think it's a super important role that I think that moving forward, I think that companies that don't acknowledge it and don't maximize it are going to not be as effective as other companies.
Visibility and Success in Data Visualization
00:12:31
Speaker
Right. Another thing that you mentioned in one of the blog posts and I think you found in the survey that you ran was, A, there's not a lot of data visualization people in leadership roles. And also this sort of split between people who are doing data visualization
00:12:47
Speaker
in organizations that maybe doesn't see the light of day. So the sort of stuff that you're doing behind the veil of Netflix and then people who are out and about to have a social media presence who are marketing either themselves or their organizations because that's part of the job and sort of, I don't know what words you're talking about, but maybe is there, are you thinking that there's an imbalance in those two parts of the field? Well, so I think that the lack of data visualization people in leadership, I think it's a real phenomenon.
00:13:17
Speaker
I think that it is based on the newness of modern data. I think it's just something that's going to change as things go forward. In the same way that when somebody becomes a manager at Netflix, they might have spent their time
00:13:34
Speaker
being a data engineer or a data scientist or in any of these other roles, UI engineer, they stop doing that. And it's a constant refrain across companies around here of somebody said, oh, I haven't written code in three months or six months or whatever. And they're so sad and they're not really. And so I'm not talking about, it's not like you continue to do that for the rest of your life. I'm talking about sort of preconceptions.
00:13:59
Speaker
and what you bring to the table when you're thinking about strategic development and these leadership questions. And so I think that's just a natural thing that's going to change that I just want to highlight and maybe hopefully spur leadership at organizations to try to prime that pump and develop those roles more quickly because it means that they've got a blind spot. As far as the other side of it, I think that, yes, right now I think our models for successful data visualization practitioners
00:14:28
Speaker
are too dominated by people who are in the public eye. And they're in the public eye either because it's their job like journalists or academics. An academic has to publish. The whole publish or perish thing means that you have to publicize your work. Or freelancers and consultants who need to publicize their work or they won't get new clients.
00:14:51
Speaker
And I don't think that it's so much that those roles are invalid in comparison to the kind of professional roles that I see in industry. It's that there's not 100% match between the skills that you develop in those kind of roles versus the kind of roles that you have professionally.
Biases in Data Visualization Tools
00:15:10
Speaker
And I think right now there's this idea that you have to be this sort of dynamic, charismatic,
00:15:16
Speaker
great communicator to succeed at data visualization when I'm sure that there are a lot of people who maybe don't like talking to people, don't like going to conferences, don't like tweeting, but are very good at developing visual rhetoric that I don't want excluded from the field. And so that's my biggest concern as far as modeling sort of successful narratives in data visualization.
00:15:41
Speaker
Yeah, I wonder whether that's not unique to data visualization, right? So you've got Bill Nye the science guy in science, right, who's out there and everybody knows who he is. You've got Paul Krugman, you know, as an economist who's out there. There are lots of other economists who are also very smart, you know, also doing valuable work. And many of them are probably working for banks or other industries where their models may or might not be allowed out the door for the same sort of reasons that
00:16:09
Speaker
data visualizations aren't allowed out the door. So I think we're talking about really is evolution of the field as it were. So is there a difference between these other fields where the charismatic people who like to speak and are good speakers and good writers write books and those sorts of things and the data visualization field? I mean it seems to me like that's just you know part of what it is to you know sort of make your way to the top of a specialty.
00:16:32
Speaker
I think that is part of it for sure. I think that right now what we run into is that most of the people who are speaking for professional data visualization practitioners either are directly invested in certain solutions, and here I'm talking about folks who work for Tableau or work for Trifecta or any of these companies that I love, and I think these are smart people, but we have to acknowledge their biases.
00:16:55
Speaker
Or they're folks who used to work in this field and frankly haven't worked in this field in the last 10 years when things have changed quite dramatically with the tool sets and the expectations. And so here I think of someone like Stephen Few who provides extremely valuable work. And I own a couple of his books and I think he's right about a lot of the things he says.
00:17:16
Speaker
But Stephen responded to one of my articles, offended that I referred to him as a conservative view in data visualization. I think that's really odd because of course Stephen Pugh is a conservative view in data visualization. How could anyone think otherwise? He is focused on charts for numerically precise representation of data for busy executives, which is
00:17:38
Speaker
an important use case in data visualization. But it's like the important use case in data visualization from 1955. And there are other use cases now. There are other clients. There are other deep engagement over time with long-lived applications that are going to be extended. And that's primarily what I build at Netflix. I don't build figures for presentations to executives.
00:18:03
Speaker
And so a lot of these lessons that people are learning about professional data visualization that are extremely valuable only apply to a subset of the use cases of data visualization in industry. It's interesting that you mentioned not to pick on Tableau. I mean, we could pick Company X. It's interesting because last year when, just as the example, Tableau stock fell by some huge 50% amount or something like that last February,
00:18:31
Speaker
You know, I sort of tweeted out this graph of the stock price, which was just, you know, a remarkable sort of shift. And a lot of people were defending the company, which was fine. I didn't really have a take on the company. But it did spur this conversation about
00:18:46
Speaker
Are people in the data visualization field obligated to reveal or disclose when they hold stock in a particular company they have a financial interest in?
Constructive Criticism and Growth in the Field
00:18:56
Speaker
So we know when someone, you know, they work at Tableau or they work at Microsoft or they work at Netflix, but if they don't work at those places and they have a financial interest, is there an obligation then to reveal those affiliations?
00:19:10
Speaker
So I guess there's a lot going on in the field. I mean, it's funny that we've seen some ethics in data visualization and I haven't heard that offered up because I guess it's too controversial. I mean, it's odd. I think that I'm less concerned about that than I am about these sort of structural issues, which is that, first of all, I mean, I want to be very clear here because I don't have a financial interest in it and, you know, it's not going to affect my job. But I do want to be clear that I think Tableau does great work.
00:19:40
Speaker
I think that Excel does great work. I think Jorge's demonstration of that, the piece that he just published on Medium demonstrates it. It's not so much that I think that they're somehow holding the field back. I think the problem is when we forget that there are natural biases that are going to happen, people frame things within the tool that they are accustomed to.
00:20:04
Speaker
to be aware that I use D3 all the time. And so therefore, there's a whole grammar graphics lineage that I struggle against all the time when it comes to building data visualization. And not that there's anything wrong with grammar graphics, of course, when it's enabled a huge amount of processual database, but it's still a frame that if I'm not aware of,
00:20:22
Speaker
is biasing the kind of work that I do. That said, I think that we're remarkably unself-reflective in data visualization for being such a young field. We're remarkably naive about these things. I think it's for positive reasons. I think what we've been trying to do with the field is be positive and community-oriented and not be as critical as we could be because we want to lift people up
00:20:50
Speaker
But I think as a result, we haven't gotten very good at
00:20:54
Speaker
being critical. You painted an interesting picture in my head of whether there's a curve of the evolution of a field where, again, you just look back the last 10 or 15 years, we had the tower infographics sort of came and went. We have the birth of D3 and the interest in Tableau, and then of all the D3 sort of drop and drag tools like, I don't know, high charts and click and quadrogram and all those.
00:21:20
Speaker
and how the criticism works both within the field and then from outside the field and I wonder whether there's sort of this evolution of the field where in the next few years maybe we seem to be coming back from interactive to static and whether the critique of the field will evolve in a similar kind of way and that we'll see now we're sort of everybody's being nice to each other but maybe that's about to change or maybe it will change?
Is Data Visualization a Profession or Skill?
00:21:49
Speaker
Yeah I think that you know
00:21:50
Speaker
whenever you see movements where there's a lot of novelty, it's natural for criticism to recede into the background because people don't know how to criticize. How do you criticize things that have only been around for a year or two? How does anybody tell me that I use the wrong particles in my particle Sankey diagram when no one's ever done that before? I think that we're seeing a maturity of, like you were saying, this procedurally generated data visualization that's enabled interactive and animated data visualization.
00:22:20
Speaker
That makes it less flashy and so naturally people feel like oh well I don't have to do animated interactive stuff I can do static stuff because it's not so important the flashiness isn't rewarded because people are accustomed now to animated and interactive stuff and maybe that's why we're seeing more of a focus on impact and insights and all of the I words are coming back into data visualization so yeah, I think it's natural I think that there's nothing very controversial about
00:22:49
Speaker
that and yeah and I think that a part of it is this question of professionalization and I think that the question professionalization frankly it's still an open one. I think that there's a profession here but it could be that there isn't it could be that
00:23:02
Speaker
that five years from now there are no data visualization engineers, that it really is a skill that UI engineers have or data scientists have or people who are doing more communication style pieces have. Do you think the innovation in the field, at least the outward innovation, so again, the two of us at least have different roles or different places in our organizations, but do you think
00:23:26
Speaker
um outwardly at least publicly that the innovation and data visualization has changed dramatically like i look back at the year when periscopics gun piece came out and pitch came out with their drone piece and there were a few others that year that seemed
00:23:41
Speaker
in some ways like a high point of innovation of totally new things. And I wonder from your perspective whether we've continued to be as or more innovative or we've sort of slid back into, well, let's just go and make more conservative, more Steven few. I mean, I would agree with you. He is clearly on the conservative side of the field, but are we worried too much or more so about bar charts and line charts and not pushing the envelope as much as we were maybe three or four years ago?
00:24:11
Speaker
I think so. I mean, it's funny. It seems that way. I don't have any empirical evidence of it. It feels like we're not radically developing new methods of visual expression of data, like we were two and three years ago. But you know, it's hard to it's and I mean, I think that that's why everybody's so taken by
Interest Shift from Visuals to Insights
00:24:30
Speaker
naughty Bremer. I think she's a real throwback with all of her filters and
00:24:34
Speaker
text on curves and things like that. I think that she's very sort of mythical in that way, right? It seemed like there were people like that showing up every six months. And I can't think of anybody who's really sort of taken the scene like she has and just said, hey, I had these ideas about stuff and I'm going to push boundaries. And I've heard some people say that's because all the good talent got hired up and locked away. And now we're making bar charts in the Netflix minds. And I think that while that might be the case to a certain degree, I think that it's
00:25:04
Speaker
It's one of these art imitating life or life imitating art kind of question. I know that I've grown less excited about pushing the boundaries of visual expression of data and patterns. And I've grown more interested in how to measure impact and how to structurally speak to concepts of insight and information and things like that.
00:25:29
Speaker
But again, this is the problem that we run into when we talk about this field is because we haven't developed structural language around the field, because we haven't developed a professional dialogue, then it's just this constant chorus of
Lack of Professional Societies in Data Visualization
00:25:41
Speaker
discussions. My feelings on it, your feelings on it, Lynn's feelings on it, everybody's got feelings. I mean, we don't have any sort of consensus and we don't have any sort of mechanism. We haven't developed any mechanisms to drive consensus.
00:25:54
Speaker
I wonder why that hasn't happened. I was asking Enrico and Moritz about this. Why don't we have a professional publication? Why don't we have a professional society? Why is it that
00:26:06
Speaker
All of these conferences are so very different in the demographics of who's attending and how they sort of laud successful practitioners. So let me give you an example that I think might, well, I don't know if, maybe it'll shed some light on this, but part of my job is to help people with their visualizations, but also with their presentation.
00:26:28
Speaker
techniques in their PowerPoints. And what I have found is that there's been a demonstrated value to having better graphs, better visualizations, different visualizations. But I don't think the same has been shown of four presentations, four PowerPoints that there is a demonstrated, I mean, I would argue there is, but I don't think people in general have recognized, at least the people that I tend to work with and consult with, recognize the importance of changing how they present.
00:26:55
Speaker
So that being said, I don't do as much sort of hands-on PowerPoint design that as probably I would like because I don't think it's been demonstrated as well.
00:27:04
Speaker
same time, about six months ago, a few friends of mine launched the presentation guild, which is exactly the sort of membership certification, society sort of thing that you were just mentioning for data visualization. And I think even though these two fields are sort of moving at different paces, it is interesting to me that it's taken this long to get a group up and running that is actually trying to do exactly what you're talking about, but for the presentation field. So
00:27:33
Speaker
Perhaps it just takes time. Perhaps it takes just a core number of people to actually push this over the edge and say, yeah, we need to have this thing. I do think so. I don't think that there's anything radical about the state of the field. I think it really is just a natural evolution of something that was expensive and uncommon before.
Evolution and Resistance in the Field
00:27:56
Speaker
and has progressed to the point where now you have a lot of folks who have moderate level of talent. And that's reflected in that data visualization survey. As skewed and as skewed as that survey is, I'm sure it is. But you see this real youth bulge in that survey of people who have been doing data visualization for three or four years. And it's really interesting because there's also this huge bulge in the people who've been doing data visualization for 10 plus years. And I think that that to me
00:28:25
Speaker
speaks to the entire state of the field where you have a lot of folks who have been doing this for a long time and naturally have. Like I was saying earlier about biases you know biases that that developed because that was how things worked when they were learning how to do these things and I'm not some kind of.
00:28:49
Speaker
I don't believe that we throw out the older generation and we have nothing to learn from the older generation. I think Tufti is still required reading. I think that Bertan is still required reading. I think Stephen Few is required reading, but I also acknowledge that they were developing concepts
00:29:07
Speaker
at a time when the tools and applications were very different than they are now. And on top of that, the number of practitioners was very different. And just when you have an increase in the number of practitioners, that doesn't just mean that there's more. It means that you do have different issues at play. And I haven't seen anything that has really dealt with that. And so yeah, I think it's natural that the field is just at this inflection point, and it just needs to understand itself. But at the same time,
00:29:36
Speaker
I mean, I'm not all kumbaya. I think that there's a lot of regressive elements in data visualization right now who don't have any value for this position that there is something up, that there is a tension.
Standards and Aesthetics in Data Visualization
00:29:54
Speaker
And, um, and I just don't agree with them. I fundamentally don't agree. I think that, that things are changing. The field is growing and changing. And I agree with you. I think it's really interesting, this idea that we're equivocating now on the value of interactive and animated graphics. And I think that's great. I think it's really exciting. And I hope that my personal interest is in trying to revisit the value of, um, iconography.
00:30:20
Speaker
sort of traditional infographics and all of that stuff that doesn't fit within the grammar of graphics paradigm, all of this very sort of hand-generated auto-neurath kind of stuff. So let me close with this. Let me put you in the position of the president of the Society of Data Visualization Engineers and ask you, what would you like to see, you know, the top two or three things you'd like to see over the next, let's say, three or four years? That is interesting.
00:30:50
Speaker
Put my money where my mouth is. Well, first off, you've already been elected president of a mythical society I just created. So congratulations. You all have wonderful taste. What do we need to see? So in that case, let me let me create some committees. Okay. All right. So that'll that'll be how I declare it. So now that we've got this together, one of these committees has to be focused on developing mechanisms for evaluating impact within
00:31:19
Speaker
an organizational setting. I think that's a key problem that I experience all the time with data visualization. Our mechanisms for evaluating data visualization either are overly academic, they rely on sort of these cognitive, cog sci kind of stuff, gestalt kind of stuff.
00:31:36
Speaker
or they rely on experimental practice that you can't do inside an organization, or they rely on too large an audience. Internal applications don't have audiences of thousands. They have audiences of half a dozen to maybe a couple of dozen. I'd like to see us develop some standards for how to evaluate impact of data visualization, specifically within an organizational setting.
00:32:04
Speaker
I don't know how to make it a really sort of. A committee. Simple, slogan-y kind of way to say it. But what I want to see is I want to see an integration of aesthetic value.
00:32:14
Speaker
back into organizational development of data visualization. So right now, I think people are way too comfortable saying, oh, well, that's aesthetic. And so I don't deal with that. And so as a result, what you end up with is you end up with every data visualization you see inside an organization using the same horrible 10 color scheme when there's only three categorical variables that they're representing. You see people wondering why they made some kind of dashboard and nobody's using it.
Need for Better Training and Creativity
00:32:39
Speaker
and therefore tacitly acknowledging that they work within an attention economy inside their organization but they're not taking that into account when they're building the applications. And just more generally accepting that when we say something is attractive or has a sophisticated look and feel or has a complex look and feel that sometimes those are impactful purely for that reason within an organizational setting.
00:33:05
Speaker
And I think that that is specifically at odds with this sort of the theme of data visualization in industry in the last 10 years. I would extend some of that. I mean I don't think it's within an industry or within an organization. I mean sometimes the visualizations that are nonstandard are not as familiar to people.
00:33:28
Speaker
engage readers and users in a way that the traditional graphs don't. The one that comes to mind is the measles heat maps from the Wall Street Journal where you could have easily presented these patterns in measles infections by state over time as a line chart and you would have seen 51
00:33:48
Speaker
squirrely lines, but as a heat map, it engages you and you see it in a different way. And maybe you're more inclined to look at it, because it's just different. That's right. Yeah, I think so. And I think that we really do need to deal with that. And I think that finally, and it's a point I've touched on a couple of other places, let's develop some fundamental training for how to effectively create and also for audiences to effectively use some broader categories of
00:34:18
Speaker
data visualization, and here I'm specifically referring to flow diagrams, hierarchical diagrams, and network diagrams. Complex data visualization is anything you want it to be. I know that. But let's focus on a few of these fundamentals and
00:34:37
Speaker
You know what I would love? I would love for Steven Few to write a book about how to make and read Sankey diagrams, how to make and read force-directed network diagrams, and how to make and read dendrograms, circle packs, and tree maps. That's what we need. We need somebody to tell us here are the, because all of those things that I've just described have been around for a long time are fundamentally straightforward representations of certain types of data and
00:35:05
Speaker
I still cannot point to a sort of best of class or a bulleted list of the important things to keep in mind when building them or reading them. Interesting.
00:35:16
Speaker
Well, you just gave someone a PhD dissertation, so hopefully someone will pick that up. No, it has to be shorter than that. Nobody's going to read anybody. It has to be a really aggressive pamphlet. What we have to do is we have to do for sophisticated, complex data visualization what Lin-Manuel Miranda did for Hamilton.
Podcast Conclusion
00:35:35
Speaker
We have to rehabilitate it, except that Hamilton didn't deserve it.
00:35:43
Speaker
But we need a musical version for data visualizations. We need data visualization the musical.
00:35:49
Speaker
Well, I look forward to it. I'll buy tickets first. Elijah, thanks so much for coming on the show. This has been really fun and I look forward to this conversation continuing. Thanks for inviting me, John. Me too. Thanks everyone for tuning into this week's episode. I hope you enjoyed it. I hope you'll share your thoughts and your comments. You can find both Elijah and I on Twitter and various social media channels. This has been the Policy Vis podcast. Until next time, thanks so much for listening.