Jock D. Mackinlay is the first Technical Fellow at Tableau Software. He believes that well-designed software can help a wide-range of individuals and organizations work effectively with data, which will improve the world. He is an expert in visual analytics and human-computer interaction who joined Tableau in 2004 after being on the PhD dissertation committee of Chris Stolte, one of the cofounders of Tableau. 

Mackinlay received his BA in Mathematics and Computer Science from UC Berkeley in 1975 and his PhD in Computer Science from Stanford University in 1986. His PhD dissertation described how to automatically generate graphical presentations of relational information including bar charts, scatter plots and node/link diagrams. Inspired by Jacques Bertin, he developed a composition algebra to generate a wide variety of graphical presentations and evaluation criteria to identify effective presentations. 

In 1986, Mackinlay joined the Xerox Palo Alto Research Center (PARC), where he collaborated with the User Interface Research Group to develop many novel applications of computer graphics for information access, many inspired by his composition algebra.  His key mentors were Stuart K. Card and George G. Robertson. They published three influential papers at the ACM CHI’91 conference on their prototype system called the Information Visualizer.  This prototype included the Cone Tree, an animated 3D node-link visualization of hierarchical information, and the Perspective Wall, a 3D focus+context technique that was effective for temporal analytics. In 1999, he co-wrote the book Readings in Information Visualization: Using Vision to Think with Stuart K. Card and Ben Shneiderman.

Toward the end of his time at PARC, Mackinlay was also a member of the PhD dissertation for Chris Stolte, who was working with Stanford Professor Pat Hanrahan on the visual analysis and exploration of large, complex databases.  Stolte and Hanrahan developed the VizQL specification language for data views, which extended Mackinlay’s composition algebra.  After graduation in 2003, Stolte. Hanrahan, and Christian Chabot co-founded Tableau Software to commercialize this research.

Episode Notes

Jock | Jock on Twitter
Tableau Customer Conference
Pat Hanrahan
Chris Stolte
Elissa Fink

Related Episodes

Episode #209: The Flerlage Twins
Episode #202: Lindsay Betzendahl
Episode #200: Do No Harm Guide
Episode #192: Eva Murray
Episode #188: Chantilly Jaggernauth
Episode #144: Neil Richards

Support the Show

This show is completely listener-supported. There are no ads on the show notes page or in the audio. If you would like to financially support the show, please check out my Patreon page, where just for a few bucks a month, you can get a sneak peek at guests, grab stickers, or even a podcast mug. Patrons also have the opportunity to ask questions to guests, so not only will you get a sneak peek at guests but also have the opportunity to submit your own questions. You can also send a one-time donation through PayPal. Your support helps me cover audio editing services, transcription services, and more. You can also support the show by sharing it with others and reviewing it on iTunes or your favorite podcast provider.


Welcome back to the PolicyViz podcast. I am your host, Jon Schwabish. And on this week’s episode of the show, I am excited to have Jock Mackinlay join me on the program. You probably know Jock from his work at Tableau Software. If you are a researcher in the data visualization field, you surely know Jock’s name from his long experience and his long body of work in the field. He has basically done everything there is to do when it comes to visualizing data. And so on today’s episode, we talk about his career. We talk about his work at Tableau. We talk about the various teams that he’s built and pulled together to help make that tool what it is today. So I hope you’re gonna enjoy this week’s episo de of the show. I’m sure you will. If you have comments or questions or suggestions for new guests, or other things that you’d like to hear about on the show, please do reach out. You can reach me on Twitter. You can reach me at the website. Or you can reach me at this YouTube channel, where you might be watching this video of my conversation with Jock. So on to this week’s episode of the PolicyViz podcast, here is my conversation with Jock Mackinlay.

Jon Schwabish: Hey, Jock, good afternoon. Well, good morning, your time. How are you? Good to see you again.

Jock Mackinlay: I’m fine. It’s good to see you as well. I hope that we can see each other in person sometime soon.

JS: Yes. That would be lovely. That would be great. I’m excited to be able to chat with you. Thanks so much for coming on the show. So your spot in the data visualization world is very interesting, because you’ve had a long career of doing a lot of research. And then you basically helped build Tableau into what it is today. And so I thought we would talk about a number of different things. And I thought maybe we would talk real briefly about how you helped get Tableau started with helping Chris Stolte in his original dissertation. But then I want to focus a lot on the work that you’ve done at Tableau and where you see not just Tableau going, but the dataviz world going in the next, next few years. So maybe you just give folks like that quick, like that quick history. And then we can really dive into the work that you’ve been doing at Tableau for the last couple years. Last several years. Yeah.

JM: So I did my undergrad math, computer science at Berkeley, graduate work after being a programmer for a while at Stanford. And then I went off and started to be a research scientist at the Xerox Palo Alto Research Center in Berkeley. And then Pat Hanrahan, who was one of the founders of Tableau, is an expert in computer graphics. And I got to meet him when I was at PARC, and he introduced me to Chris Stolte. And Chris ended up having me be on his dissertation committee, because I was already doing work in Visual Analytics. And that was my start for Tableau. And it turns out, you know, Chris is very entrepreneurial. I ended up joining Tableau 2004. And then I, yeah, as you said, I had a really long career. I was employee six, Tableau, got to start both the design team and an industrial research team in 2011. And I’m now a technical fellow, which means I’m an individual contributor, but no longer a manager. But so I’ve had a very long arc at Tableau. It’s involving over 17 years.

JS: Yeah. So if like Chris and Pat are like the founders and the fathers of Tableau, you’re really like the grandfather of Tableau.

JM: So Chris literally used my PhD dissertation in his PhD dissertation. So there is some truth to that. I developed an algebra for automatic presentation. And Chris and Pat did a domain specific language called VizQL that also combined in the database part. That was sort of the golden spike that led to Tableau, like connecting Visual Analytics experiences to dig to actual data, particularly valuable data in databases led to the success of the company.

JS: So that’s the background. So 2011, you start the design team and the user research teams. So can you talk about, I don’t think we need to talk about the ins and outs of being a manager.

JM: No.

JS: But like, what were you looking for when you’re building those teams? How did you have those teams work together, work separately? I mean, you know, Tableau’s sort of goal is to help people get insights into their data. And there’s people facilitating that. So how, so maybe just talk about like, building those teams together, what you were looking for? And how that all shook out, I guess?

JM: Sure. So the core focus from Chris and Pat, originally, innovation at Stanford, was on to helping people answer questions with data, that that’s the understand part to help people see and understand data. And it is very cognitively challenging to do that. First of all, data has its own challenges. And then you’re trying to answer a question. And that is extremely, extremely challenging. So it made sense in 2011, as Tableau grew to hire designers, and user researchers that have an understanding of human cognition, and the first two people I hired one was Jeff Pettiross, designer, very good at this. And then John Kim as a user researcher, and he’s now a designer. He likes to design. You know, user researchers that does formal studies of people using software. Tableau has a great user research team. But yeah, it was trying to support that cognition, that extremely difficult effort to answer questions with data. Now, working with data is a really, really broad thing. And Tableau has grown up a lot. And so answering questions is only one part of what Tableau now supports. It now supports the entire, you know, enterprises doing all aspects of working with data. But that was the genesis of it. And so the start in 2011, in particular, the start was just to, you know, get professionals in to help us support people thinking, thinking with data.

JS: So can you describe maybe a couple of projects? I mean, I think, there’s a number of different things, right, that those teams are doing. There’s the internal part of helping build the tool into what it is, into what people see when they open Tableau and they make things. And then there’s also like a whole wing of folks that are doing research. And I’m sure there’s internal research, but a lot of what I’ve seen is like external stuff that’s coming up to move the field forward. So can you maybe talk about a couple of those projects on either design or the user research side? And how, I guess, you know, that’s a really broad question, but also like, how, what’s the balance, like in terms of like, we were improving the tool, and we’re helping the community sort of generally, the field moving?

JM: Yeah. So Tableau, you know, with a dissertation at Stanford, had an academic tradition. So in 2011, we also started an industrial research team. And there’s a difference between user research and industrial research. Now I hired one person who did both really, really well. But industrial research is, you know, writing academic papers in academic conferences, also prototyping, doing prototyping work, which is hugely valuable. And literally, you can do user research studies on prototypes. So it made sense to start both of these teams at the same time. But the industrial research team is a small, you know, sort of forward facing thing that’s different, whereas the design and the user research teams are in the nitty gritty of building the actual software.

JS: Right. Right.

JM: They’re two different aspects. Which direction you want me to go now that I’ve made that distinction?

JS: So why don’t we start with the design piece?

JM: Yeah.

JS: And then we can go to the industrial research side.

JM: So there’s a relentless effort to first, you know, with that existing shelf experience, that is the core question answering part of Tableau, just making it easier. And I can’t completely remember 2011, but there’s two parts to it. There is sort of the data facing part, so connecting to data and whatnot. And then there’s the human, you know, augmentation of human cognition part of it, which has a visual part to it, but also the experience part. And so once the design professionals showed up, there was a relentless effort to sort of continue doing that. And later in Tableau’s history, of course, we added in additional products. And we’re continuing to do that for people with more skills. And also, right now, we’re also focusing on business professionals. And there’s people that certainly don’t have a lot of time to play with data, and may not have the data skills, data literacy to play with data. And so we’re pushing out that way.

JS: Right. Okay. So then on the research side, let me capture for you this way. When I worked at CBO years ago, you know, sometimes there would be requests from members of Congress to say, “Hey, we want to know the answer to this particular question.” And other times there was, when we didn’t have specific requests, when there are always specific requests, but some were shorter than others. There was what do we think, as an agency members of Congress would want to learn about? And what do they want to know about so they could do their jobs better? So I’m curious what that balance was like? You know, there are customer needs, and the research and the cognition research about how to solve those problems. And then there’s sort of what a lot of what I’ve seen is, well, maybe it’s just on the writing, but it’s like a little more general. It’s like, here’s a dataviz problem that we are trying to solve. So, I guess, my, if I was to, like, hone in on this question, it’s like, what does that day to day look like for those research teams?

JM: So the industrial research team is talented individuals who have expertise in various things, like statistics, computer graphics, etc., etc., along those lines. And the team that I built up, I had a passion about making sure it was very customer focused. So from a customer focus point of view, you start with problems that customers would have. And for Tableau, of course, it was very broad set of customers, lots of different industries and whatnot. Then the individual research scientists would have ideas about how to crack open a problem of one type or another. And there were two ways to do the research. One way was to actually build prototypes, because it is industrial research. And prototyping is a great way to do that. Sometimes prototypes can end up actually becoming part of the product offerings of the company. And then you can also write academic papers, which is, you know, getting academic papers through review committees is a way of honing the ideas, the understanding of the ideas, the methodologies about that. And so this industrial research group to this day is very, very good at both of those prototyping, and also, using the academic review process to hone ideas, but always with an eye towards ultimately building software that would help our customers.

JS: So you’re sort of at that point during that 20 years or so where you’re running, founding and running those teams. It sounds like you’re sort of looking forward, right? You’re sort of anticipating customer challenges. And I mean, I’m sure you’re answering certain challenges, but you’re sort of anticipating questions and challenges. And so I’m curious now that we’re in 2022, you have a slightly different role there. But now looking forward, what do you see? I mean, Tableau itself has gone over sort of at the aggregate level, tremendous changes, you know, recently being, you know, acquired by Salesforce and variety of other things. So I’m curious now, as you look forward, what do you see both in terms of the tool itself, in terms of the dataviz community, in terms of you know, the world, I mean, not the world, because we’ll never stop talking, but you know, yeah. So what, like, now looking forward, what do you see?

JM: So the first thing to say is, I’m actually super excited by the acquisition in 2019 by Salesforce, because Salesforce is very, you know, has a really long history, really successful company focusing on specific business professionals, sellers, marketers, you know, field service people, people like that. And so Tableau right now wants to grow to business professionals. And so this partnership with Salesforce is totally fabulous that way. So that’s the first part. The second part is, there are sort of two major vectors going forward. One of them was what Tableau was doing all the way back when Chris and Pat were at Stanford, which is the partnership between humans and computers. And the opportunity always and certainly today is that humans and computers have complementary but asymmetric skills with respect to data work. On the human side, I go either way, but I’m going to start with on the human side, we have a very rich understanding of the world. And in particular, individuals have a very rich understanding of their organizations. And comparatively, though, computers have a much more superficial understanding of the world, but they’re perfectly willing to dive into vast amounts of data 24/7 365. Humans need to sleep. Humans find data, you know, a struggle, boring on and on like that. So there’s this huge, huge opportunity around that partnership. And in particular, right now 2022, machine learning and artificial intelligence are slowly giving computers more and more capabilities to one type or another technically. And so that’s going to just enhance the partnership.

The other thing is collaboration between people. So the world is a really rich, complicated place. And so you might be know everything about your specific thing, your organization, but the world is really, really rich. So you do need to collaborate with other people. And because it’s still times a pandemic, I use the example of, you know, who knew that we were going to need to have an understanding of COVID data, you know, in 2019? Well, we didn’t know. But we’re now avidly reading experts on that kind of data, and, you know, taking the data and combining it with our organizational data, because it’s super important to the organizations. And so that’s a relentless trend. And so that is why collaboration is super important.

JS: Yeah. So on the machine learning and artificial intelligence, like for an everyday Tableau user, what, can you give me a sense of how they might expect machine either one of those sort of things to show up in their creation or their use of the tool?

JM: So this is a place where the partnership with Salesforce is really good. Salesforce did the negotiations to have Einstein be a brand element about all of this, but the basic idea here is, in your workflow, you could get recommendations from Einstein about what to do next.

JS: Right.

JM: Now, you know, humans need to bring their judgment to bear. They know the world about whether they’re actually going to follow the computer or not. But that’s the idea is that computer can provide recommendations of actions of literally not just decisions, but actions to do next, and that will speed people in their work. So the reality of it, so just to keep the drill down one level further, there are data professionals who are skilled enough to be able to do the machine learning and whatnot, but the delivery of the machine learning modules into workflow. And so you want to support the [inaudible 00:17:28], and also monitor those models to make sure that they’re still providing good recommendations to the individuals that are doing that. That is a current concern. That’s current work, research. I’m obviously not going to get into any sort of forward product announcements, but that is what’s going on.

JS: It’s interesting to hear you, hear the way you sort of frame that because what pops into my head is like, a show me tab, which we have now that’s like, for some select number of visualizations, but a show me tab that could be infinitely large, that’s not just on, give me examples of visualizations to create, but give me examples of other data analysis or other threads that I could sort of pull, you know, I’ve built this thing to help me reach this decision. With all these other pieces sort of circulating in and around Tableau, there are these other threads that I could pull, and so it’s maybe not necessarily just show me a graph type I could use, but show me a data analysis or a data field that I could go explore or a new data field that I could go explore.

JS: Yes. In fact, the computer can easily monitor the data and notice things that to draw human attention to, which then might lead to questions that a person wants to answer, you know, super relevant questions to their organization. And, you know, like, obviously, a lot of business professionals don’t have the time to actually answer those questions, but they will delegate the answers, you know, to the analysts in their organizations or whatnot, or, which is why, by the way, I think that Salesforce acquiring Slack is super exciting for Tableau, because it’s a natural place to do that kind of delegation or that kind of collaborative work with people. It’s already established user experience for that. But yes, for me, personally, I find that the deeper questions to be the most interesting ones to try and support. That was how I ended up on Chris’s dissertation committee. It’s cognitively challenging to answer those questions. And by the way, you know, the 17 years I’ve been at Tableau, those deeper questions are hugely valuable to our customers from getting to good, using data to getting to good answer on deep questions brings huge customer value. And so it’s super valuable that way, in addition to all the routine data work.

JS: Right. From the perspective of the tool, is there a, well, how, I know there is, but like, how does the tool help lead people to those deeper questions? Right? So it’s really easy for me to pop in, if I think about like HR data, right? Just because, you know, the Salesforce link is kind of obvious one, you know, I could pop in my HR data. I can do some basic tabs. But how does the tool then encourage me or, or lead me to ask these deeper questions?

JM: The questions occur all the time to people. So it’s not so much the computer leading to deeper questions. It’s the computer helping a person quickly decide whether, you know, first of all, getting answers to simple fact questions really quickly, and then deciding whether some question is actually worth, is worth deeper exploration or not. And because it’s the human that has the knowledge of the world that that has to be brought to bear, rather than the computer, but the user experience needs to be designed so that you can quickly, essentially do triage on the question, and then spend your time focusing on the ones that have depth to them and might have high payoff to them.

JS: Right. Right.

JM: And I never said the word triage before now, but it is exactly, it is exactly right. It’s like the software needs to be set up for that to support that kind of triage, not just for the individual, but also for the collaborative efforts going on that.

JS: Right, right.

JM: So that’s my perspective. But the computer can certainly, absolutely help with that. Because, you know, the marshaling up the data, that COVID example is a good example. It often, like the deeper questions often involve bringing together data, starting with some data, and then bringing in more data in completely unanticipated ways such as, you know, the data curators didn’t, you know, they go after the routine data that’s going to help the organization. But deeper questions often lead to this kind of on the fly modeling that data modeling that has to happen to get to the answer.

JS: So there’s oftentimes I see in the data science, dataviz community, this sort of argument more or less about whether there should be like a person should be able to do everything, right, sort of what I call the unicorn kind of person, or like, is it about teams? And I get the sense from you that it’s really all about teams. And so can you talk a little bit about like, yeah, from your perspective, like, what’s a way, I’ll make a little more concrete, actually. So someone’s listening to this podcast, and they’re the one person at their organization that’s using Tableau. And they love Tableau, and they’re doing analysis, and they’re trying to get more buy in at the organization, maybe for the tool itself, but maybe just more generally for like better dataviz, better data communication. So for that person, they’re trying to build these teams like, how do you see that person? Like, what’s their path forward to build these teams and build this, you know, this sort of network of people who have this shared interest to ultimately ask these deeper questions?

JM: Yeah. So, you know, what popped into my head was the Tableau conference, and oh, boy, do I wish we could have in person Tableau conferences, again? Because the brilliance out of Lisa thinks brilliance about starting the Tableau conference was that the person that you just described would show up at the, you know, like I was there at the very first Tableau conference, you know, 160 guests plus the Tableau people. But they were already sharing their knowledge and their best practices about how to get their organizations to use data more effectively. It’s a very rich set of things. Tableau’s specific responsibility is to build the software. And the good news is, of course, the visual part of Tableau means that you can make the data visible to people. You can tell really good stories about it. But there’s a lot of rich, it’s like, we don’t have time in this podcast to go into all of the different ways. But what I’m fundamentally saying is, if you’re one of those people, you know, when Tableau conference starts to be in person again, come because you will find your fellow people in different industries, but the ideas are fungible. The techniques are fungible. And the thing is, is that companies are all at many different places in the sort of the sophistication with respect that they do that data. But you will find people that are given your organizations where it is right now. And literally some of the Tableau documents like, we have a thing called Tableau blueprint, which is to help organizations do this sort of thing. And part of the, there is a self assessment of where you are, where your organization is right now, and what proven next steps are for moving forward.

JS: Right. So, I want to just get back to, just want to get to one more question before we wrap up, which we could talk about this one I think for a while, too. So you’ve been in academia. You’ve been in the private sector, and you’ve built all these different teams with all these different skill sets, and sort of built a culture of collaboration. And so I’m curious for younger folks who are interested in the data visualization field, and I was about to say computer science graduate students, but I think that’s not like the right framework, because it’s really, lots of different people come to dataviz from lots of different places. So when folks are thinking about their careers, what is your thought about academia versus the private sector versus the nonprofit sector? And like, I mean, so that’s a pretty like general question. But I think if people are, you know, thinking about where do they go once their education is done, what are your thoughts on, you know, what people should be thinking about? What are the questions they should be asking themselves as they start to look for careers in data, data viz?

JM: Yeah. The space is really, really big. And you’re absolutely correct. It’s not just computer science. People, like, for example, my son, oh, it’s his birthday today. Happy birthday, Gavin. And of course, now, I can’t figure out what his age is. But he works currently at Amazon. And he started out with a degree in economics. Well, so, for example, which, of course, obviously very data oriented, but he did need to learn some computer stuff. And he also, of course, partners with lots of other people. Frankly, you can proceed forward, if you have that sort of a data oriented passions, you can proceed forward with academic profit or nonprofit companies. The data is affecting everywhere. And everyone understands that it’s really important. And then the challenge is, are you more data oriented, then you can learn, you can learn the data, the parts of it, or the, you know, the query parts of databases or whatnot? Or are you more interested in a specific domain? In which case, then that will drive you forward out of college because, you know, one of the challenges of people who are the unicorn data scientists, in other words, a person who’s a unicorn and knows how to do, you know, the statistics, the machine learning, the queries on the database, you know, the programming and all those sorts of things, unfortunately, that unicorn doesn’t actually know about their organization. And they have to, you know, they often have to team up with or spend a long time learning about a particular organization. So that’s on the data end of the spectrum. And then part of the reason why the Tableau community is so rich and vibrant, is because people never thought that they would really be able to deal with data but really cared about their particular organization discovered that well, if the software was easy enough to use, they could actually do stuff. And that’s the passion on the other side of it. So, yeah.

JS: Yeah, yeah. That’s great. I can see your passion for and I can see the passion really for the collaboration, which I think is just so important in the way that we work and communicate data. Yeah.

JM: Yeah.

JS: Jock, thanks so much for coming on the show. It was a treat chatting with you, and hopefully be able to chat in person soon.

JM: Absolutely. Looking forward to seeing you in person soon.

JS: Thanks.

JM: Bye.

JS: And thanks everyone for tuning into this week’s episode of the show. I hope you enjoyed that. I hope you learned a little bit about the history of data visualization, the history of Tableau, where the company is now, and where it’s heading in the future. If you would like to support the show, please consider sharing it with your friends, your family, your networks. If you’d like to financially support the show, even just for like cost of a cup of coffee, you can go over to my PayPal page, or you can go over to my Patreon page, get some goodies over there. Your support helps me afford the transcription costs, the sound editing cost, the video editing cost, the web hosting, all the thing that’s necessary to bring the show to you each and every other week. So until next time, this has been the PolicyViz podcast. Thanks so much for listening.

A number of people help bring you the PolicyViz podcast. Music is provided by the NRIs. Audio editing is provided by Ken Skaggs. Design and promotion is created with assistance from Sharon Stotsky Ramirez. And each episode is transcribed by Jenny Transcription Services. If you’d like to help support the podcast, please share and review it on iTunes, Stitcher, Spotify, YouTube, or wherever you get your podcasts. The PolicyViz podcast is ad-free and supported by listeners. If you’d like to help support the show financially, please visit our PayPal page or our Patreon page at