Keeping computers in check

Efficient? Yes. But computers that aid in urban planning and other large-scale systems may miss the little details, researcher warns. Images: A look at Scalable City

Daniel Terdiman Former Senior Writer / News
Daniel Terdiman is a senior writer at CNET News covering Twitter, Net culture, and everything in between.
Daniel Terdiman
4 min read
SAN DIEGO--Computers can be mighty tools in creating video games or simulations of urban environments, but humans must maintain control over the machines to keep the results from running amok, warns techno-philosopher Sheldon Brown.

As head of the University of California at San Diego's Center for Research in Computing and the Arts, the academic researcher and industrial consultant is responsible for developing and overseeing projects that demonstrate the power of algorithmically created content. Just as important to him, however, is stressing that computers might miss the little details that matter--forgoing real-world livability in an algorithmic attempt to maximize the number of buildings in neighborhoods, for example.

Scalable City

"Just because we can do something, doesn't mean we should," says Brown, whose center at UCSD aims to create new art forms from the development of digital technology. "When we build tools, we have to think deeper and broader about what that tool is doing."

Nowhere is that tension more clear than in Brown's "Scalable City," an early version of a software application that explores ways in which researchers in fields like video games and urban planning can take advantage of computational power--and in the process automate work that previously might have required dozens or hundreds of animators and artists.

On its surface, Scalable City shows how computers can utilize algorithms, or automatic programmed procedures, to design layouts for new urban or suburban environments in limited spaces. Relying on so-called "L-curves," which Brown said are similar to fractals, the multimedia system can rapidly take digital representations of barren landscapes and fill them in with gracefully curved roads and neighborhoods full of new houses.

"Algorithms do things well that they're set up to do, but we were around for centuries without computational processes."
--Sheldon Brown, UCSD Center for Research in Computing and the Arts

From a macro perspective represented by an aerial view, such designs present what appears to be an efficient maximization of available space, Brown said.

But Brown cautioned that at a micro, or surface, level, computer systems can't know, practically, what environments work for people. Thus, Scalable City demonstrates how an algorithm designed to maximize the development of a previously barren landscape could well result in individual streets so jammed with buildings that few could actually be habitable.

It's tempting to focus on what things look like from the big-picture level at the expense of investigating the micro, he acknowledged. But humans enamored with the power of computers would be wise to closely examine what the implementation of that power means and keep control of it through vigilance, he said.

While in many ways a metaphor for responsible management of powerful computing resources, Scalable City will also have practical applications when it's completed in a year or two, Brown said. Among those uses are helping to create new game design methodologies, including the development of new tools and systems that will facilitate high-level player control and engagement.

The basic Scalable City technology could, Brown said, be an attractive and powerful application for the implementation of large-scale video games. That's because the program is based on the idea that a computer will someday be able to efficiently develop all kinds of assets--game characters, structures, environments and the like--in real time and extrapolated from simple data sets. The end result: Game publishers may very well look increasingly to computers to automate the design and implementation of many of the small but vital details of such games.

'Seductive entrapment'
In some ways, Brown is pointing out that humans are already playing the role of the data set.

"Our lives in general are now really a part of algorithmic processes," Brown said. The "major part of our lives are for producing algorithmic data and then interacting with it. We've become I/O, input/output."

But today, he argued, people are increasingly relying on algorithms to do the work we used to do ourselves, and while that can be good, it's not necessarily a positive development.

"Cities are part of our computational processes--(things like) water systems, and roads and information about us," he said. "Algorithms do things well that they're set up to do, but we were around for centuries without computational processes."

Thus, Scalable City is in some ways a metaphor for showing the power of these algorithms and showing what can happen when computers start to take over the decision-making processes.

"I'm trying to exaggerate the artifacts of algorithmic processes," Brown said, "and show that this power can be seductive entrapment."

For example, video game companies like Electronic Arts or Activision are looking for ways to create new titles without having to spend tens of millions of dollars on artistic teams that can create material suitable for new high-definition displays.

"As an artist, I need to be working with (new) video game technologies and taking advantage of them," said Brown, who consults with leading game publishers such as EA. "So I think about them, but I only have a tiny bit of funding. I don't have $100 million for this. So I have to come up with these methods that are more efficient and clever and which help us figure out how to squeeze more out of these systems."