Benjamin Fry Interview · 2394 words posted 02/07/2003 05:03 PM

Benjamin Fry Benjamin Fry is a doctoral candidate at the MIT Media Laboratory. His research focuses on methods of visualizing large amounts of data from dynamic information sources. This work is currently directed towards “Genomic Cartography” which is a study into new methods to represent the data found in the human genome.p. Ben received an undergraduate degree from Carnegie Mellon University, where he studied Graphic Design and Computer Science. He was named in the “The I.D. Forty: Forty Designers Under 30” and his work has been exhibited at the Whitney Museum and the Museum of Modern Art in New York, as well as Ars Electronica in Linz, Austria. With Casey Reas of IVREA, he is currently developing Proce55ing, an environment for teachingcomputational design and sketching interactive media software.

since1968: Of computers, you have written “Ideas and emotions may be developed in this medium that are impossible to express in other existing media.” Can you elaborate?

Benjamin Fry: Heh… was that me? hm…

Maybe a bit more pragmatic and less pseudo-poetic… In my own work, the ideas and emotions have to do with enormous amounts of data. As I cannot draw such diagrams by hand, or process such data in my head, then I have to employ computation. In undergrad I studied graphic design and computer science, so I’m interested in trying to weave the two together as a way to visualize and understand more enormous amounts of data than can be dealt with using standard graphic design techniques… that is, I can’t sit down with Adobe Illustrator and layout a poster with all 30-40,000 genes in the human genome by hand… especially when the data keeps changing.

I think it’s very simple really, that the computer is simply another medium. So the above statement is just like saying “I can express things in painting that are impossible to express through photography.” The fact that the computer can be a tool, and perhaps a many-sided tool at that (employing all kinds of different software) often muddles things. There is probably a tendency to view the computer as a monolithic thing in and of itself, when really it’s this collection of parts (the set of software tools that you run on it, the types of things that you make with it).

since1968: Much of your work focuses on visually mapping the human genome. To a non-specialist, your work appears beautiful and rhythmic. What level of acceptance do your ideas have among geneticists?

Handheld Genome Browser Prototype Benjamin Fry: There are two tracks here. The first, more practical one is that I’m slowly making inroads with the geneticists and they seem to be very excited about the work. David Altshuler, who works at the Whitehead Institute here at MIT is a member of my dissertation committee, and I’ve been getting acquainted with their work so that I can have a more direct impact on what they’re doing.

This project was a result of my initial contact with them, and was about improving the representation of the concept being explained in a recently published paper.

More broadly, my research attempts to get involved in advanced information visualization work… that is, what’s next for info visualization. If I weren’t pushing that, then it wouldn’t be research and I shouldn’t be getting a PhD out of it. As it turns out, there is such a dearth of visualization tools for genetics that most of what they want is simply more basic representations of things to explain simpler concepts and that it need not be the most advanced visualization research going. The handheld genome browser, for instance, is much further out than say, something that’s just a better representation of the NCBI browser.

So in that case, I’m trying to find a good balance between more graphic design oriented projects (like what’s needed for their pragmatic day-to-day work) and more challenging ‘research’ level work that pushes the boundaries on what’s currently being done in info visualization. Here’s an example of a more graphic design type project.

That said, the second track is that many geneticists I have spoken to (at least claim to) enjoy the work simply for the aesthetic part that you mention, and the alternate perspective it represents, because it’s quite different from what they’re used to. At the meeting where I presented the haplotypes image above, I asked if they wanted to see some other work (a couple of the other ‘chromosome’ studies) and one person’s response was “Yes! We’re nothing if not curious!” I think that defines this group of geneticists, that they’re quite driven by their curiosity of things, and as a result are quite open to new ideas.

since1968: In the film Minority Report, Tom Cruise’s character uses a three-dimensional interface driven by hand motions to visualize and process data. It was one of the very few recent movies where I found the imagined computer technology both compelling and possible. Did the production designers incorporate your ideas?

Benjamin Fry: Nah, I can claim no ownership of that… John Underkoffler, a friend of mine who is a media lab graduate worked on the production design team (as science and technology advisor), and is responsible for how that interface works… He actually went through and made a concept movie and developed (after some research into other such systems like ASL) an entire hand-motion language system for use in the thing; it’s quite brilliant. He’s an amazingly smart and talented guy, and I think he was drawing on a lot of his experiences at the lab, where his research involved physical interfaces to computers, to develop the system.

Valence That said, he was responsible for including Valence in another scene that shows Tom Cruise returning to his apartment (Valence is running on his computer when he flips it on before watching movies), as it seemed to fit the movie conceptually.

since1968: Valence—that’s software you’ve written that allows you to visualize large sets of dynamic data. On your site, there’s an example of Valence processing the contents of “The Innocents Abroad,” by Mark Twain(see the QuickTime movie). The data appear organic; elegant but hard to quantify or verbalize. Do engineers resist data that won’t fit into a two-dimensional chart?

Benjamin Fry: I think there is going to be resistance to anything that’s not very understandable. I try to focus on things that are difficult or impossible enough to begin with, that it was too much data to show any sort of coherent visual representation of. That way, it’s less of an issue about “mine is better than what you were using” and more about “here’s a way to do what used to break in a 2D chart setting.”

For instance, 2D charts of the data in valence simply wouldn’t work. It’s a 200,000 word book, 15,000 words of which are unique. 7500 of those words are used more than once. So how would you make a chart of that many thousands of words and have it tell you anything useful? More important, how would you make the chart change over time? This is the sort of thing that breaks the chart model, so we need other representations and with them, different ways of thinking about data. in the case of Valence, I wanted a piece of software that responded to data “fed” to it, so that you could watch how the data was being digested over time, as defined by a set of rules (that’s the ‘organic information design’ part).

since1968: You’ve also used your ideas to map patterns of web site usage (e.g., anemone). It seems like one of the primary reasons to process data is to share it with other people and act on it. How do you extract data from an organic model in order to share it in a paper report?

Benjamin Fry: I use anemone to get a general qualitative picture of things, and use that to do more directed quantitative research into the specifics. So I start with getting a feel of what to be looking for by using anemone, then maybe using something like perl and Excel to crunch through numbers and get specifics.

Ideally, this process should be more integrated to truly be useful. So that’s currently a component of my PhD work… being able to move more seamlessly between the qualitative initial representation and the quantitative specifics.

since1968: Is there any chance you’ll release Valence in a format useful for managing personal data and ideas? Current software such as TheBrain isn’t especially compelling.

Benjamin Fry: I may. I have largely set Valence aside as I wanted to move on to other projects. Actually, that’s one thing that John Maeda forces us to do, otherwise it’s easy to stagnate and just beat one project into the ground. The ‘research’ part of it is done, and maybe more ‘product’ oriented things like what you mention will happen once I’m on my own.

You and Casey Reas have created Proce55ing, an open-source language based on Java that’s designed to allow designers and artists to explore multimedia. The results look exciting, but why not use Flash? Is the Proce55ing language any easier for a non-programmer to learn than ActionScript?

Benjamin Fry: This is a frequent question, and there are lots of answers. We need a good long explanation on the site that covers all the bits, but a rundown of the points:

Many more, but that’s probably more than enough… every environment has its own plusses, so I’d hate to have it become “proce55ing vs. Flash” because I think that’s silly. Nobody can ever “win” such a battle, unless it goes to Macromedia, with us being badly outgunned—multi-million $$ company versus two guys working in their spare time ;)

since1968: I recently spent several months in Southeast Asia, where most people access the Internet with handheld devices instead of desktops. Does the choice of device affect the complexity of the information you can share? What information tradeoffs went into the design of your handheld genome browser prototype?

Benjamin Fry: Ah, I think the limited space is a really nice thing, in fact. Such a tiny interface forces a rigid simplicity and minimalism. So for every decision, you have to be absolutely sure that it’s necessary to include that pixel or that shape on the screen. I think that makes for an interesting design problem.

For the handheld browser, I was using the NCBI browser and some others like the one at ensembl.org as my guide for what features I wanted available, but then used the notion of how Palm Pilot ui works (always on, simple clicks through pages of information) to make it as simple and clear as possible.

since1968: Most of my readers are small-shop developers struggling to decide how much they should worry about Rich Internet Applications or .NET Code-Behinds or simply hanging it up and moving to Tahiti. How do your ideas on data modeling affect their work?

Benjamin Fry: Could you elaborate on this one?

since1968: Sorry, lame question. What I really want to know isthis: most web developers use a simple graphing tool such as Crystal Reports, Java graphing with Swing, or ColdFusion’s built-in Flash charting to post relatively basic data sets to the web. To what extent is the type of data we visualize limited by the simplicity of the tools, and how soon should web developers begin looking for more robust tools to display more complex data?

Benjamin Fry: Ah, I see…

Part of the problem with the simplicity of the tools is that it sets the expectation for what types of things can be “seen.” My feeling is that as a result, the current expectation is set lower than it should be.

I’m most interested most in the type of things where the need has already been demonstrated, and that gobs of information are to be considered. Some of these applications are interesting (I find the genomics work fascinating) or perhaps more boring (credit card companies that want to people to see millions of transactions and understand patterns in them based on all sorts of criteria…) or even have questionable intentions or outcomes (John Poindexter’s ‘Total Information Awareness’ project).

As for when to look for the tools, it’s difficult to say. There are a small handful of companies who are making interesting tools/projects currently, but for myself I’d be more interested in having more people (designers and programmers) thinking about these issues and trying to build interesting things themselves. Maybe Proce55ing and things like it can start enabling that, or maybe it’ll take a few books or some people to set a few more good examples to set the standard higher.

since1968: Thank you for your time.

~~~

Read more about information design.

* * *