Ideas 2003: In Praise of Diversity: Jaron Lanier On Software Design

Walking out of his Manhattan apartment for his day job as lead scientist at the National Tele-Immersion Initiative, Jaron Lanier, the man who gave us the term virtual reality and was instrumental in developing the first virtual reality glove so that people could touch the digital universe, is only too happy to talk about what he thinks is coming in 2003.

“The thing I’m interested in now is a high-risk, speculative, fundamental new approach to computer science. I call it phenotropics,” says the 42-year-old Lanier. By pheno, he means the physical appearance of something, and by tropics, he means interaction. Lanier’s idea is to create a new way to tie two pieces of software together. He theorizes that two software objects should contact each other “like two objects in nature,” instead of through specific modules or predetermined points of contact. With this type of protocol, there would be no previous agreement or expectation regarding the character of the exchange. Each software object would be responsible for “learning” how to interact with the other.

Lanier believes that phenotropics would allow for the development of bigger systems, perhaps a million times larger than the ones we can build today, because there would be no need to build connections between them. With this approach, he hopes to get around what he sees as the “brittle” nature of contemporary software. “Software tends to break before it bends,” he says.

He has been experimenting with phenotropics for years and says he has recently made breakthroughs in understanding how feedback would work in phenotropic systems. “That’s the dreamy stuff I’m doing,” he says. “I can happily make the prediction that it will have no firm impact on next year.”

Less dreamy, however, is the philosophy behind phenotropics and its implicit criticism of the way we work now.

“Essentially, we’re in wartime, so I suspect that next year is not going to be about fun innovation, but rather about managing contraction, stiffer competition, lower budgets and issues related to security,” he says. But he believes that current development practices will not allow for effective security. “The way we build software is monolithic and antidiversity,” he explains as he strides through the streets of lower Manhattan. “If you have computers running identical software, that lack of diversity is a core [security] problem.”

Lanier looks to nature for examples of safety through diversity. “If we draw rough metaphors to natural viruses, the way they survive is through diversity,” he says. An antiviral agent that kills one virus will not kill another. By extension, an attack on one network would not be effective on another running a different species of software. But phenotropics would allow profoundly different systems to work together, thereby providing natural protection through diversity. “There is no security in monolithic networks,” says Lanier.

“Obviously, war has a big effect on everything,” he adds. “This could be an occasion to fix a lot of problems. It could be an occasion to have a political and social consensus on preparing the means for the next energy and transportation technologies but,” he concludes pessimistically, “that is not happening.”