Looking to Give Executives
A Living Computer
By THOMAS PETZINGER JR.
CHRIS LANGTON is developing the
first "flight simulator" for CEOs, a piece of software so powerful it will
model turbulent markets, complex partnerships and other corporate vexations.
By adjusting the parameters, he says, chief executives will explore the
unseen hazards and opportunities ahead.
There's just one problem: Even the world's most powerful computer is
too small to run such a program. Dr. Langton's simulator can take flight
only by harnessing the processing power -- and possibly much of the information
-- in every computer inside the enterprise it's modeling.
What do you think of this week's installment of The Front Lines? Let
us know by joining in The
Front Lines Forum, a discussion area where readers post messages related
to this week's column.
Until recently, that seemed an impossible dream. But for the past four
months Dr. Langton has been secretly
test-driving a radical new Sun Microsystems product, which Sun publicly
unveiled last week. Dr. Langton, of
Santa Fe, N.M., thinks the product holds the solution to his challenge
-- though he acknowledges that the new technology harbors potential risks
as well.
Now 49 years old, Dr. Langton has
spent his entire career anticipating these developments. In the early 1970s,
as a programmer at Massachusetts General Hospital, he spent his free time
watching little squares on a computer display switch themselves on and
off according to the state of the squares around them, a game as old as
computing. As he tweaked the rules governing the blinking, stunning patterns
created themselves, some crawling across the lattice like tiny creatures.
In the next two decades, working at the University of Michigan and Los
Alamos National Lab, Dr. Langton repeatedly
found that a single computer program could model the dynamics of seemingly
unrelated systems, from ant colonies to economies. Complex populations,
it appeared, attained order through simple rules and local interactions.
Other researchers were reaching similar conclusions, giving birth to a
field called complexity science (whose ideas, as you've read here, are
influencing management).
DR. LANGTON moved full-time to the
citadel of complexity studies, the Sante Fe Institute, in 1994. His mission
was creating the mother of all modeling programs, called Swarm, to permit
biologists, economists and others to explore the deep laws of organization
uniting their disparate specialties.
As corporate people began dropping into the institute to bone up on
complexity, they saw Swarm as a tool for studying their operations, from
the movement of railroad cars to the structure of supply chains. In addition,
the Pentagon tossed money into the project.
But Dr. Langton walked away from
much of his funding in a dispute with the military over the breadth of
the project. And while Swarm eventually came to life, the cuts forced him
to abandon some of the features he had intended, including the ability
to run on many computers at once. By this past January, he had left the
institute.
These days there's no shortage of funding for computer scientists with
big ideas. He and some associates formed a company, called Swarm Corp.,
to develop new versions of the software for solving business problems.
A desktop PC could handle much of it -- opening the bottlenecks in transportation
systems, for example -- but others are so huge and complex they could easily
consume all the idle computing power in any organization.
Other problems, such as supply-chain management, might even require
computers from multiple companies to model one another's models simultaneously!
Enabling the new product to run on so many machines in parallel would be
a monumental task, if not an impossible one, for the tiny start-up.
Enter Jan Hauser, a principal software architect at Sun Microsystems.
A longtime student of the self-organizing properties of life, Mr. Hauser
met often with the complexity gurus of Santa Fe. Hearing of Swarm Corp.'s
efforts, he persuaded officials to bring Dr. Langton
into the inner sanctum in which Sun was developing a new product called
Jini (as in "genie").
JINI IS HARD to describe. Part language, part protocol, part
operating system, it's a computing "environment" intended to let any computing
device share programs and processing power with any other device. Whether
Jini will succeed remains to be seen; earlier efforts failed at Sun and
elsewhere. If the technology fulfills its potential, it could make sophisticated
computation as accessible as running water and machines" communication
as vibrant as a beehive's.
It's impossible to overhype such a world. Computers could be put in
charge of themselves, seeking their own connections in solving problems.
Networks could develop a life of their own, creating unimagined new structures
and processes (in the way that the Internet, through no central design,
creates new ways of doing business).
As a top Sun official told Wired magazine: "Our goal is to lose control
over the network, and make everyone else -- from Bell Labs to Redmond --
lose control, too."
But what of the risks? Any complex network, from a political system
to the electricity grid, harbors the potential to create not only order
but disaster-locking up, breaking down or spinning into chaos. Dr. Langton
says it's not premature to worry about confederations of machines outsmarting
their creators. "It's a little spooky," he says.
The shield may reside in the very tool he's now creating. The only way
to design against Jini networking catastrophes may be to run simulations
nearly as big as the networks themselves-simulations powered by Jini. "It
sounds so lifelike," says Dr. Langton,
"We're creating a system fully capable of self-reflection." |