Cherri Pancake | Anthropology & High Performance Computing

After a decade as an anthropologist studying the Maya peoples of Central America, Cherri Pancake shifted to CS and conducted pioneering work in usability engineering for high performance computing. Today she’s the VP of the ACM and Director of the Northwest Alliance for Computational Science & Engineering at Oregon State University. We talk about her journey from the social sciences to computer science, her experiences leading professional organizations, and how she thinks about her interdisciplinary work today and into the future.

Bio

Dr. Cherri M. Pancake is Director of the Northwest Alliance for Computational Research; she recently retired as Professor of Electrical Engineering & Computer Science and Intel Faculty Fellow at Oregon State University. She combines backgrounds in anthropology and computer engineering to address how complex software can best support the conceptual models and research strategies of practicing scientists and engineers.

Pancake was among the first worldwide to apply ethnographic techniques to identify software usability problems – an approach which is now mainstream – and she conducted much of the seminal work identifying how the needs of scientists differ from computer science and business communities. Over the past 25 years, she has served as PI or coPI on research grants totaling over $150 million from industry, not-for- profits, NSF, and Departments of Agriculture, Commerce, Defense, Education, Energy, and Interior. The methods she developed for applying user-centered design to improve user interfaces have been reflected in software products from Hewlett Packard, Convex, Intel, IBM, and Tektronix. As a leader of national standards groups, she developed procedures for consensus-driven design that expedited the adoption of community standards.

A Fellow of ACM and IEEE, Pancake currently serves as ACM’s Vice President and recently stepped down as Chair of SIGHPC, ACM’s Special Interest Group on High Performance Computing.

Transcript

Andrew Miller:             Thank you so much for talking with us today. The first question that I ask people on this podcast is what first got you interested in computer science or computing? I’m interested especially for you since your journey to becoming a professional computer scientist was quite protracted. Were there early signs? Were you always a computer person? Were you a design kind of person?

Cherri Pancake:            Well you’re asking in a very polite way because some people would say, “You just have trouble making up your mind.” It actually has been a very interesting journey. When I was in undergrad school, computers were still pretty new. I was at Cornell and they had some, but it was still definitely the mainframe days. My undergraduate degree was in environmental design. At that time, that was a new concept at universities. A new field of study. It was very flexible. Some of my peers focused more on urban planning. I got very interested by the anthropology side of it. The cultural anthropology in particular.

Cherri Pancake:            By the time I graduated I really had the equivalent of a minor in cultural anthropology. In fact, I was going to go to grad school in anthropology and I got married and things changed a little. Among those things, I ended up going to Peru with the Peace Corps. My husband and I did. Down there, I actually lived outside of a very small Indian community way up in the Andes, 20 hours from Lima for travel. I got very interested in applying some of the ethnological techniques that I had learned in school. That seemed to be a really good match. So much so that after I left Peru, we went on to Guatemala and my husband worked in forestry. I actually worked in anthropology. I became curator of the ethnographic museum in Guatemala that specializes in the highland Maya Indian culture. I was gone for almost 10 years.

Cherri Pancake:            Anthropology, in particular ethnology, was my great love. When we came back to the U.S. … My husband wanted to come back. I found that anthropology, to be frank, isn’t great for your resume in getting a job in many areas of the U.S. I was in the southeastern U.S. I started thinking about retraining in something else. I had gotten into computers a tiny bit while I was at Cornell, but later when I was in Guatemala, through the anthropological connections, I got involved in long-term study of communities. We had to do a lot of data collection and analysis using statistical packages. Not really coding, but still using the computers.

Cherri Pancake:            I got very interested in that and in the issues of representing indigenous languages when at the time there were only English character sets and how computers could be applied to anthropological tasks to make things work a little better. As I looked around to retrain, I decided I liked the systems side of engineering. Auburn University was opening up a new disciplinary area, computer engineering, and I was the first graduate student in computer engineering there. That’s how I got into it. I really have enjoyed that career also.

Andrew Miller:             Yeah. It’s interesting to think also just historically, computer science … Elements that we now think of as foundational were once interdisciplinary between other disciplines.

Cherri Pancake:            That’s right.

Andrew Miller:             Computer engineering, which we think of as a core component and a normal area of study, was new at that time.

Cherri Pancake:            Yes. More of what was being called computer science was definitely much more on the mathematics end at that time.

Andrew Miller:             Just jumping back to the anthropology thing, because this is fascinating and I think we’ll get more into it later as well, one of the things about anthropology is that you’re examining the artifacts that people produce and the roles that they play in people’s lives. It seems to me like there are interesting parallels to be drawn between that approach and looking at the roles of computing systems in people’s lives. Is that something that you were seeing at that time or is that something that came later?

Cherri Pancake:            Well it’s interesting you should ask because at the time, I was sort of hiding my anthropological background because I found in engineering schools … I was the first woman to be accepted for the graduate program and any engineering field at Auburn. I found that my colleagues all thought … They had all wanted to be engineers since they were like three years old.

Andrew Miller:             Right.

Cherri Pancake:            One of the things I learned in working with other cultures is that it’s important to make the people that you’re interacting with feel comfortable with you, even if you’re very different. For engineers, I focused on helping them get over the fact that I was the first woman first. That difference first. Then later on, when they would comment about how I had helped them see differently or that that was an interesting approach, then I would explain why to them.

Andrew Miller:             You’ve made this career switch. You’re in a computer engineering program and you’re starting to figure out, “Okay. What is going to be my topic of research? What methods am I going to use?” What was your thought process as you started to embark on independent research in this new area?

Cherri Pancake:            Well my own dissertation work was actually in embedded systems and software tools for embedded systems. It was my very first graduate student who got me interested in the problem of software tools for parallel computing. With my background, it was obvious to me, though not to her, that the first thing I would ask is, “What do we know about the people who will be using these tools and how they think about computing?” It turned out nobody knew. Most users of parallel computing and high performance computing have always been scientists or engineers. The ones who have had big problems that require lots of computing power.

Cherri Pancake:            At that time, there were only a few places that had high performance computers. Users went there for extended periods to do their work, but nobody really studied them or understood anything about how they approach their tasks. I decided to fill that gap. It turned out that I worked for … Gosh. 10 years, I think. I spent three months a year at the Cornell Theory Center, working with users, especially the experienced ones. That was when I truly got the marriage between computing and anthropology because I ended up using what really classical ethnographic techniques to watch how they work, to elicit how they were thinking about their problem and compare and contrast that with how you think about your problem versus how you coat software.

Cherri Pancake:            Guess what? No surprise now, but back then it was a huge surprise that these users thought about computers and programs very, very differently than the tool developers did. That turned out to be why they found software tools to be so totally unusable. I ended up working with many … Pretty much all of the companies that did major software tools at one time or another to help them understand the difference between what the tool developers were trying to support and what these users really needed.

Andrew Miller:             You found this way to put this anthropological expertise to use. How did you find a place to even publish this work?

Cherri Pancake:            I’ll say probably one of the things that in hindsight I regret most is that I didn’t spend much time on publications in those days. I was looking more about, “How do we take this knowledge and apply it to actually improve the software tools?” I was more on a crusade to convince tool developers that they needed to be providing better support for these users and to help service interpreters between the tool developer community and the user community. I got quite involved in a number of standards efforts that affected software tool standards and programming language standards for these large computers, but I didn’t do a lot of formal publications to be honest.

Cherri Pancake:            Sometimes people who aren’t themselves working in interdisciplinary areas feel it’s easy. It’s a cheat. I ran into a lot of people, some of them my own colleagues in the two computing departments I worked in, who thought, “If she really were a better computer scientist or computer engineer, she wouldn’t have to go out and work with these computational physicists and biologists.” Seriously. People had that kind of bias. I think to a certain extent, it still exists because people see … These aren’t the kinds of things we talk about in the conferences I go to. They’re not the kids of things that are written about in the journals I publish in. It must for some reason be second-tier.

Andrew Miller:             Yeah. Speaking of professional societies and publication activities, you certainly have been doing publication and organizational service within the computing community since then. That’s the laugh of experience, I think. We talk a lot about the interdisciplinary stuff on this podcast. One of the things that the guests keep reminding me is that computing itself is growing and changing. Actually, there is a lot of opportunities for intra disciplinary conversations. You have been involved in various SIGs or special interest groups within the ACM and actually founded one or helped found the high performance computing one in 2011.

Andrew Miller:             I’m interested contrasting the early desire to just actually get the work to happen with I guess embedding it into the professional communities. What do you see as the connections there?

Cherri Pancake:            I early on when I was a grad student became a student member of ACM. Actually, I was a student member of Computer Society too. I was involved with them. With the ACM, there were more activities that we got engaged in more things at Auburn. One of the things I got involved with and later was involved with as a faculty member was the programming competition. What I saw was the way a professional society can help provide a cohesive effect among students who might not otherwise feel that comfortable approaching each other or becoming friends with each other, finding a common thing.

Cherri Pancake:            I thought the societies are great for doing that. Then later on, I got more active with different conferences and as you say, with publications in both of them. One of the gaps that I really noticed was in the area of high performance computing. A lot of the people that are in practice in high performance computing and in research are not actually computer scientists. They’re in occupational biology or computational physics or material science, forest ecology. All these different fields who have the need for high performance computing in order to develop the models and the analytics that they need. That means they’re quite isolated because often there are only one or two people doing their type of computational work in an institution.

Cherri Pancake:            How you find other people who share similar interests but most importantly, might be good resources when you run into problems or obstacles. That I think is a real challenge. It’s one I think that professional societies have a unique role to play. After talking about this concept to a lot of people from the HPC community, we already have a very strong conference that I had actually been general chair of, the SC Conference, it was obvious that other people felt that way too. I can’t tell you how many people told me, “Oh, we thought all along there should be a SIG HPC or something like that, but nobody ever started it.”

Cherri Pancake:            I decided, “Well why don’t I see if everybody really does believe it?” I got a couple of other people really respected in the community to work with me to form the nucleus. I can’t give too much credit to the staff at ACM because when I started asking them questions, I was so naïve about this whole process of forming a SIG and what would make it attractive or not attractive to ACM. I will say I also talked to IEEE and I talked to SCIAM because it wasn’t clear which of those three places are high computing performance group worked.

Andrew Miller:             Yeah.

Cherri Pancake:            ACM has the unique characteristic of being volunteer driven. It just looked like we were going to have a lot more freedom as a SIG than we would have as a technical committee under IEEE or as a think group under science. We decided to go with ACM. I think it’s been great, the extent to which the community has gotten behind it and the extent to which ACM has supported us as a new SIG.

Andrew Miller:             Yeah. It’s interesting to see the inner workings of these societies. I think a lot of people, and me included, are sort of in the dark about these mysterious organizations.

Cherri Pancake:            Yes. Until you have to do it, you stay in the dark.

Andrew Miller:             Yes. Maybe it’s better that way. I don’t know. One of the things that’s interesting about this high performance computing endeavor is precisely that even though there’s this cluster of people, this strong core of people who feel like, “This is one thing.” Even then, it still wasn’t clear exactly where it should live and where is its home.

Cherri Pancake:            Yes.

Andrew Miller:             One of the things that I’ve been thinking about when I think about these professional organizations are a good gathering place, but in some ways, they’re also a barrier. I’m in the ACM, but I’m not in the IEEE. That’s not a thing that I sense that you’ve really felt limited by.

Cherri Pancake:            No, but it actually touches on what I think is one of the real issues that’s going to be very important going forward. That is what does it really mean to be a member of a professional society? Are you a member of just one? Are you always a member of multiple ones? I don’t want to get too much on to my soapbox about this, but you know, computing isn’t a fraternity of geeks anymore. For the last 20 years or so, the number of people working in computing-related work has exploded and many of them … Maybe I’d say most of them have little or no formal computer science background. A heck of a lot of these people who are actively working with the new … Especially the newer technologies, like cyber currencies, autonomous vehicles, big data, IOT, you name your favorite ones. They were trained in other disciplines and they don’t think of themselves as computer scientists.

Cherri Pancake:            I think that’s the next great opportunity for ACM is how should we get these people more involved? Maybe membership is not the answer. Maybe it needs expanding our notions about conferences and workshops and other kinds of interactions rather than focusing on membership.

Andrew Miller:             That sounds to me like someone with an anthropological perspective.

Cherri Pancake:            Gee, I wonder why.

Andrew Miller:             Right. This idea of these membership affiliations and categories and how we move between them is really fascinating. I teach a research design course here at my university and one of the most complicated and fraught discussions is when we get to ethnography, ethnomethodology and anthropology in general, it just seems to flummox people. It’s a totally different way of thinking and knowing. I imagine that you’re someone who’s had a lot of experiences like this. How do you reconcile for people? How do you help people from a computing perspective translate and see the value and understand how you think?

Cherri Pancake:            That’s a great question. I guess I’m more used to explaining it to people who have some notion of the scientific method. I am going to approach it from that standpoint. If you know the scientific method, the whole point is to form a hypothesis. Then you test it. It might be right or it might be wrong, but that’s how you learn is by forming a notion and then testing it. Ethnology in particular, when you do field work, that side of anthropology as opposed to physical anthropology that studies bone characteristics or some of the others. When you’re dealing with the cultural stuff, the whole key when I went into a community and what I was trained to do is you’re supposed to purposely go in with no hypothesis in mind.

Cherri Pancake:            It’s almost the antithesis of the starting point of the scientific method. Instead, what you do is you go in and you just observe. The key difficulty, the big challenge there is how do you control bias? We all have it. Lots of biases, we don’t even recognize. A lot of what the training involves is how do you recognize your own biases and how do you get around them so that you know you’re actually observing what’s happening in front of you or what people are telling you without an unnecessarily subjective interpretation?

Cherri Pancake:            Then after you collect a lot of data, then you start looking for the patterns in those data. In many ways, it’s similar to data mining if you will that you don’t start out telling them to look for this. You say, “Look for the data. You tell me what you’re going to find.” That’s how you identify what’s really going on in another culture because it doesn’t make sense to develop a hypothesis ahead of time based on your culture when you’re going into such totally different worlds. I think it’s an exciting way to think about things in that it actually just like knowledge of the scientific method can help somebody in business or English literature or comparative religion understand a different way of looking at the world.

Cherri Pancake:            I think this ethnographic approach can also help you look at the world in a different way. If for example, the first time you get together with the community, rather than going in with, “I want to find people who will work with me in X,” you go in with, “I want to see what everybody else is doing and see who inspires me to think in a different way from the way I’ve been thinking.” To me, that’s the true value of interdisciplinary work is that because we’re all trained to think about problems in different ways, working intensely and really hearing from people from another discipline expands my ability to approach problems in new and creative ways.

Andrew Miller:             Well thank you. We should figure out a way to package you up and have you as a module. It gets at the other aspect of this is what’s fascinating you today? What things are you learning about today that you’re putting into your own research?

Cherri Pancake:            In a way, I think I’m one of the luckiest people in the world. I am director of a research center, NACSE, the Northwest Alliance for Computational Science and Engineering, which I founded 20 years ago in order to be a place for doing interdisciplinary research. Really interdisciplinary research that a normal center focuses on starts out with a topic that we’re going to do. We get all the people, whether they’re from one discipline or many, who are interested in that topic. This one has focused on where new and advanced technologies might be able to be applied to help scientists and engineers do their work more effectively.

Cherri Pancake:            Where most centers start around a theme, this one started around more of a direction of how to apply technology to the real means of practicing scientists and engineers. Originally, I thought it was only going to be software tools. In fact, that was where our focus was probably for the first seven or eight years. To do that, we would work with all different communities of scientists and engineers. For example, just at Oregon State, we’ve had collaborators from seven different colleges. That’s every college, including Veterinary Medicine. Everybody except pharmacy.

Cherri Pancake:            Along the way, it turned out that more and more, they were being challenged by the difficulties of managing big data that was very heterogeneous. Collected by people in other disciplines, so they didn’t necessarily know some of the rationale behind why certain things were collected, done in a million formats. Different kinds of systems and how you actually can combine and make sense of that kind of data. We moved into that for about eight years. Then it turned out another need was emerging and that was how you collaborate, especially in an interdisciplinary context, when people are in very different geographical locations and intermittent over time.

Cherri Pancake:            The common theme to everything we do ends up being is there a real need and a large enough community that has this need that we can have a built-in market for whatever comes out of the work? Also enough people who will actually work with us because we do everything through iterative and user-centered techniques. That means we have to have users willing to put in the effort with us. The way that’s turned out in the last three or four years is we do decision support systems, right? At this time. Ask me again three years from now and it may be very different. It’s endlessly fascinating. I am always working with new communities and being able to learn, like I said before, different approaches to problems.

Andrew Miller:             Yeah. When you’re talking to people from these partner communities, have you learned things about how to talk about your approach or how you convey your findings that could apply to people communicating across disciplinary boundaries more generally?

Cherri Pancake:            Well I would say the first thing is to do what you have been doing during this interview, and that is focus more on asking questions than on telling people things. I heard a quote once about, “I never seem to learn as much when I’m talking as when I’m listening.” That’s really true. A lot of times, it’s tempting if you develop a tool … I certainly have been tempted to do this. You see another community where you think it might be useful. It’s very tempting to go there and say, “I’ve got this tool that I think will solve your needs.” That’s like inventing the hammer before the nail. Now I have to look for a nail to hit. I think this kind of communication where you start with, “What do you need?” Rather than what I can do for you is actually … has been very useful for me, at least.

Andrew Miller:             I’m interested whether you feel like you yourself have changed as you’ve shifted across these different disciplinary boundaries. Do you feel like an anthropologist at heart in an engineering world? Do you feel like you are an interdisciplinary person now?

Cherri Pancake:            First of all, the first part of that. Have I changed? Absolutely. I think change is really healthy in that if we’re not changing, we’re falling behind. I find that to be something extremely positive. As I mentioned before, every new group I work with, the tsunami researchers actually may have expanded my worldview more than any other group that I’ve worked with because they had to deal with things that happened very quickly that were massive catastrophes. They’d have to gear up to form ad hoc teams that would go into the areas very quickly after an event. There was no organizational structure with the staff member who did this for them.

Cherri Pancake:            I learned so much from them about how thinking of each other as resources and not thinking of each other as fixed teams, but totally malleable made me approach things very differently. For example, in terms of my work with volunteer organizations as well as my work with other disciplinary groups. I think that interdisciplinary researchers have a harder time in some way because as I mentioned, sometimes by your own colleagues, you might be a little undervalued, but in the end, I think we win a lot more than we possibly lose from that because we had the chance really every day to be expanding our repertoire of how to be creative and how to think about problems. Ultimately, I think whether we’re practitioners or researchers, how effectively we can solve problems is what will determine our success.

Andrew Miller:             Well I can’t think of a better way to finish than with that inspiring message. Thank you so much for joining us today.

Cherri Pancake:            Well I was delighted. You made me think, too. There you go.