WRR 4.4 1 AUGUST 2007
Of Algorithms, Google, and Snow Globes:
An Interview With
Computer Scientist David Dobkin,
Dean of Faculty at Princeton University
To understand the work of Professor David Dobkin, former Chair of the Computer Science Department, and Dean of Faculty at Princeton University, it helps to travel in the mind’s eye back to the 9th century CE, the golden age of Islamic art and scholarship, to a place called the House of Wisdom in a city called Baghdad. There, a mathematician and astrologer named Muhammad ibn Musa al-Khwarizm wrote the first book on the systematic solution of linear and quadratic equations. His book, Algebra, based on his own work and that of the Greek Scholar Diophantus would, ten centuries later, pave the way for the invention of the modern computer.
In 825 CE, al-Khwarizm wrote a treatise, On Calculation with Hindu Numerals. The treatise traveled to Europe, was translated into Latin, and al-Khwarizm’s name was translated to “Algoritmi.” Over time al-Khwarizm’s name took on a new meaning: algorithim or “calculation method.” Al-Khwarizm, intellectual, scholar, devout Muslim, and star-worshipper, most likely would not be surprised to see what scientists like Dobkin have done with his calculations because without algorithms this interview would not appear in the cyber pages of the Wild River Review.
Yet, for many of us who turn on our computers first thing each morning to e-mail friends and colleagues, read the newspaper, write our blogs, play the latest computer game, listen to music while we’re ordering books or cds or a car, the words computer science still conjure up the image of geeky men and women with thick glasses and pocket protectors. These mysterious human beings jealously guard their supply of necessary Bics, always ready to jot down the latest mathematical equation percolating in their brains. Or, we might imagine a team of grungy-chic, techno-whizzes with long hair, and a penchant for corporate perks such as lattes and M&Ms, working round the clock in the laboratories of Google and Apple.
Four years ago, Dobkin became Dean of Faculty at Princeton University, and his world opened up in new directions. A key player in the development of the sculpture garden, Quark Park, Dobkin recently talked about his life, work, and the importance of science in community.
WRR: On your website, you have a quote which expresses your philosophy. This quote appears in many languages except English. It roughly translates to say, “In the kingdom of the blind, the blind are king.” What do you mean by that?
That’s not quite right. The actual quote is: “In the kingdom of the blind, or in the land of the blind, the one eyed man is King.”
When I took my first administrative position as Chairman of the Computer Science Department I felt incredibly clueless. And I said, “Why did they put me in this job?” And then I looked around at my colleagues who were equally clueless, maybe even more so, and I realized that the quote was the best description of the situation I was living in.
WRR: And how did that turn out? Did you change your philosophy or does it still fit now that you’re Dean of the Faculty?
I think it’s a pretty accurate view of society. Unfortunately, on a national level we make mistakes and now we’ve put the blind person in charge.
WRR: Can you talk a little bit about your work as a computer scientist?
My specialty is three-dimensional modeling. The types of things I talk about are obviously much more basic than what’s going on in the forefront of the field, but one of the projects that I worked on recently was aimed at finding a three-dimensional version of Google.
Say you get on Google and you type in Quark Park. A tenth of a second later, Google comes back with roughly 15 million pages that reference Quark Park. Obviously, we’ve learned how to do that. It’s very easy to search using text string. What’s harder to answer is this question: What if I wanted to look in Google for a 3-D model of a chair?
Imagine that we’re fifty years forward, which is probably really ten years forward, and we’re teenagers, and we want to build our avatars (computer images of ourselves and our world) for a game we’re playing. And our avatar consists of building a model of a room so we need to find a chair somewhere. How do we tell Google that we want to look for a chair? We’ve already crawled the Web and have gotten somewhere between thirty and fifty thousand models of chairs.
How do we know that the image we’ve pulled up is a chair rather than an airplane? And when we find one chair, how do we find other chairs? We all know how to type, so typing in the word chair is easy. The problem on the front end is that constructing a picture of a three-dimensional rendering of a chair is not so easy. So, if we type in the word chair we would get pictures of chairs, some of them beach chairs, some of them desk chairs. And we get pictures of people: This is a picture of a chairman of the board (laughs).
So what we’ve done is to present you with the twenty things that are closest in our database to a chair. And then we ask you to grade the images: I like this. I don’t like this. I like that. I don’t like that. And then we have a way of comparing models, which we can do under the covers as it were - mathematically - to say if you like this, then the odds are you’ll like that, and that. We find if we come back and ask the user maybe one or two times, we’re able to provide a palette of good answers.
This gives us the ability to take the models that are found and let you, the user, slice pieces of them and attach them in other ways. There’s a software product you may be familiar with called Photoshop that Adobe makes, which people use for editing. So far, there’s not a three-dimensional version of that. This is what I’m interested in. I’ve worked on modeling faces with the idea of being able to morph back and forth. One of the things we can do with morphing is that my head could be turning. As it turns, it could be turning into your head.
We can also do timing so that halfway through I would still have my beard but I would have your hair and some of my facial features, but I would also have some of your facial features. For example, what we’ve done in the Google search function is create a picture of a high-backed chair, not unlike the one I’m sitting in. But, instead of legs the chair has runners from a sled. And one of the arms is a normal arm and the other arm is one that might be from a dining room chair, and some of the back is changed. But all the pieces fit together. It’s a bit disconcerting to look at because it looks like it must be organic. It must have been manufactured that way. But then it’s just a computer model.
WRR: The American public often perceives science as dry and boring and you’re talking about Google. You’ve also taught a class with sociologist Paul DiMaggio. It has a catchy title: “Sex, Money, and Rock and Roll: Information Technology and Society.” What did you hope to accomplish in this course?
I used to meet Paul many mornings when we took our sons to catch the bus to school. Over time, we started to talk and I noticed that some of his sociology interests overlapped with some of my computer science interests. So, after a while, we decided to gather some of our colleagues together into a lunch group we called GWA (Group Without Acronym). At GWA lunches, conversations would evolve with computer scientists talking about technical issues, and social scientists talking about policy issues and ways in which policies implement values, etc. This was during the Internet boom and we all enjoyed watching the technology grow as the web became a dominant force.
After a few years of lunches, Paul and I started thinking that we should broaden our audience, spinning the content of GWA into a freshman seminar, which we did in the fall of 2002. We were going to call the course Information Technology and Society or something like that, but I observed that many computer systems were written to support sex, drugs, and rock and roll. We decided that the three drivers of the Internet were the porn industry (sex), the financial industry (money), and music sharing (in 2002 Napster was big). So we proposed the title figuring that somewhere along the way, a bureaucrat would censor the name. The censoring never happened so here we are.
WRR: You were intimately involved in the conception and creation of “Quark Park.”
Yes. My son, who is now 12, was on a basketball team and they would practice one night a week. Invariably I would sit on the sidelines, and if another parent was sitting there we would get into conversation. One night Peter Soderman was that parent.
So we started talking and I misunderstood something Peter said. He was talking about the sculpture garden dedicated to writers called Writer’s Block. And I thought he was saying that Writer’s Block was very nice.
And so I said, “Writer’s Block wasn’t just nice. It was great.”
But what he’d really said was, “I did ‘Writer’s Block’.”
So we got to talking. And I said, “If you do it again, you should do scientists because it would be interesting and science doesn’t get that kind of play in public.”
He thought that was an interesting idea. And then I volunteered that I could find him scientists both through my job as dean and from my work as a scientist.
He introduced me to architect Kevin Wilkes, and occasionally the two of them would come and see me, and we would chat. What I didn’t realize was that, for the next two years, they were having conversations with lots of scientists and sculptors and artists. They’d come and give me a progress report, and they would give me a list of the people they were talking to. I never calculated how much time they took.
WRR: How do you see artistic interpretation of science as helping people understand and appreciate its complexity?
The projects at Quark Park were chosen to make science more accessible to people. If you read about science, you often get bogged down into details. Or, you think that the concept is so sophisticated that you’ll never understand it. Quark Park focused on issues that people can hope to understand.
For instance, do male and female brains function differently? We all wonder how the brain works. Neuroscientist Tracey Shors took on the challenge. Her area of science focuses on how memories are acquired and stored in the brain. Obviously, the sculptor, Steve Weiss, who worked with her, couldn’t know all details of the science, but he created a representation of her work that could encourage people to understand basic concepts of her research.
So, I believe Quark Park crossed an important barrier. People weren’t sure if they were seeing art or science. Everybody believes that they like art and every body believes that they hate science. So if they’re not sure whether they’re looking at art or science, then this creates a better situation.
WRR: Can you talk a little bit about your project at Quark Park?
I worked with Kevin. We built the roof that covered the seating area and the stage. The science behind the design is the science of what is called “Freeform Surface Modeling.” Which probably doesn’t mean a whole lot to most people. But if you think of a geodesic dome, it’s essentially made out of what I think of mathematically as triangles all connected together, or a mesh of triangles that exist in three dimensions.
You could also think of building such a thing with sticks and balls where the sticks are attached to the balls and the balls provide the structure. Geodesic domes are very regular. When you do modeling and computer graphics, you don’t always want things to be regular. If I want a model of my face, for example, this is a fairly flat area but I’ve to get the peak of the nose and I have to get the details.
One of the ways to do that is to start with a geodesic dome and let the balls move, and by moving the balls you can add structure. By moving the balls, you change what is called the connectivity. So in a geodesic dome, you would typically have six sticks coming into every ball. But there’s no reason that number couldn’t be seven or eight or nine sticks. Technically, it means changing the underlying geometry or the changing the typology of the shape.
So we built something where our design initially started out flat, as a roof usually does for such a situation. And then we moved our equivalent of balls and sticks. In one area, the roof was very rippled, mountainous; and in another place it was flat. The covering created shadows, giving the feel of sun on a mountain range. This technique was used far beyond its original intention as a geodesic dome.
When you see computer-generated mountains in movies, they’re done according to the same ball and stick formula. The operational term that is often used is the notion of a ‘fractal.’ Designers often talk about fractal mountains. When you see a seacoast in an animated movie, it is often done by fractals. It’s a way to represent and control randomness.
WRR: What exactly is a fractal?
A fractal is a curve or surface of fractional dimension. We’re all comfortable with a line having one dimension. Or, the top of a table having two dimensions. But most of us don’t quite know what a 11/2 dimensional shape might look like, or 13/4, or 1.42357. Mathematicians have known about this for about two hundred years. It’s been popularized over the past forty or fifty years starting with Benoit Mandelbrot, and others who have taken the ideas forward. Basically, you take a line and make it so wiggly that it’s not clear if it’s a straight line or if it will become the surface of a tabletop. And there’s a mathematical way to measure the dimension of that line.
A very simple thing to do would be to start by pointing your finger to make a straight line. But, what if you bend your finger so now you have a line, a bump up, and a straight line. That’s still a line. But now, what if you could take each of the little segments of your finger and repeat the same thing so that you have a series of bumps and straight lines. If you do this over and over again, you suddenly get a jagged line.
For instance, if you look at a map, the coastline of England or Cape Cod is not a straight line. It goes into inlets and comes out. You can end up modeling things like that. The pattern that I just described comes out looking very much like a snowflake, an ever more refined snowflake. It has the wonderful property called “self-similarity,” which means if you take a magnifying glass and dig into the image, you’ll see that at any level of magnification the image is always the same. All you did was take a straight line, added a hill, and repeated it over and over again.
WRR: What about algorithms, which are always mentioned with computers?
An algorithm is a recipe for doing something. A computer program is an algorithm that tells the computer what specific steps to perform so that it can carry out a specified task, like creating a mountain ridge. My work there has been looking at ways (and preferably efficient ways) of doing these things. Sometimes, I actually implement my ideas by writing computer software. Often these are just pen and paper exercises to show that here is a way that one could complete a task without an actual implementation.
Many of the algorithms I’ve worked on have been geometric; some are complex and hard to describe. Most recently (before I became essentially a full time administrator) I was working on representations of three-dimensional geometric objects in a computer. The idea here is to have ways of presenting objects on a display that allow the user easy interaction and have a robust (mathematical) structure behind them so that the computer can facilitate this interaction.
WRR: You’ve segued from the roof you built at Quark Park to explaining the science behind the roof. There are twelve other scientists at Quark Park, each with their own specialty. One of the things you mentioned earlier was about getting science out to the community. What would you like the public to learn from this?
I felt a mix of emotions at Quark Park partly because of the beautiful art that interpreted the science. By its very existence, Quark Park and the exhibits within presented enough information for people to question just how and why science affects their daily lives. And the hope is that some people who visited got curious enough to Google a scientist or their work, or perhaps they made a visit to their local library.
WRR: How do you feel about Quark Park being a temporary exhibition?
I feel mixed. On one hand I think it’s a wonderful idea, first writers, then scientists; who knows maybe philosophers down the line. I don’t know what comes next or where you go. On the other hand it’s a lot of work. Although I know that some of the work has gone on to find a permanent home.
WRR: Earlier, we talked to Freeman Dyson and he said mathematics is a language of nature. Would you agree or disagree with that statement?
I don’t know if its mathematics or physics but I think there’s a place where mathematics and physics merge. But I guess I would actually disagree in the sense that I think of physics or biology as being the language of nature, and mathematics being something that was created by humans to explain that concepts that become too hard to explain using natural things.
WRR: You are now Dean of Faculty. Has it taken you away from your science?
I’m just starting my fourth year as Dean of Faculty so I’m over the culture shock of starting what really is a completely new life. It has, in a way, completely taken me away from my science. I spend my time in Nassau Hall wearing a tie and jacket worrying about the university rather than spending my time as a computer scientist, which usually means sitting in my pajamas in front of a terminal.
On the other hand, part of what I am trained to do as a scientist is to solve problems, and that’s what I’m doing now. The difference is that the problems I solve as a scientist usually involve moving symbols on a piece of paper or inside a computer, and the problems I solve now involve people.
Yet, the job itself is fascinating because I can watch a whole city operate and understand how the parts fit together. Again, this is not unlike being a computer scientist and looking at a large piece of software with its inter-working parts and wondering how those parts fit together.
WRR: In addition to all your accomplishments and awards, you are also known in certain circles for your world-class collection of snow globes, which you call paperweights. Is there a scientific reason for the name change?
WRR: How did you start collecting paperweights?
In the fall of 1975, I was on my way to a conference and I stopped in an airport gift shop. I saw a paperweight from the World Trade Center and decided to buy it. On that trip, I went through a few airports and bought a paperweight in each, and then I was hooked.
WRR: What do you like so much about them?
I like the motion of the snow; I like that they are inexpensive so people feel fine getting one for me. And paperweights are the quintessential form of kitsch.
WRR: What are some of your most prized paperweights?
I have a paperweight of Jesus walking on water from a Bible Book Store in Nashville, a few high-end musical paperweights (including one from Graceland that plays Love Me Tender), salt and peppershaker paperweights, a paperweight that was made by a friend for my wedding using a candy jar, a bride and groom from a garden shop and filler from a bean bag chair.
WRR: You have a room in your house devoted to your collection. What does your family think of this?
We’ve learned over the years to live with each other’s idiosyncrasies.
WRR: The Martha Stewart Living show took an interest in your collection. Has there been any follow up?
I haven’t heard back from them. I think you need to start a grass roots movement to get me on her show, perhaps a petition drive.
To help you out, we’re including a link so that readers can mail their endorsement of your collection to Martha... http://www.marthastewart.com/page.jhtml?type=learn-cat&id=cat20269
Readers can also see your demo reel online.
WRR: But let’s return to Quark Park. In the interest of science, can you tell us what a quark is?
No. It’s an elementary particle. (Laughs) I think I’d be hard pressed to go any further. But you’ve talked to Freeman Dyson, he can tell you.
Celebrating the mysteries of science and art, Quark Park is a collaboration of Princeton-area visionaries, scientists, artists, and architects including Templeton-prize winner, Freeman Dyson. Over the coming months, Wild River Review will be running a series of interviews with many of the players in this one-of-a-kind sculpture garden...