Saturday, December 22, 2007

Operation Everything (Operations Research)

Operation Everything

It stocks your grocery store, schedules your favorite team's games,
and helps plan your vacation. A primer on the most influential
academic discipline you've never heard of.

By Virginia Postrel
The Boston Globe, June 27, 2004

TO THE CONSTERNATION of his colleagues, Mark Eisner once told a
reporter that his discipline "is probably the most important field
nobody's ever heard of." Indeed, it's not one that's likely to come up
at dinner parties.

"I've been explaining for 40 years what operations research is," says
Eisner, who is associate director of the school of operations research
and industrial engineering at Cornell University. He defines O.R. as
"the effective use of scarce resources under dynamic and uncertain

That may sound arcane, but it's pretty much the problem of living --
and certainly the central problem of economic life. O.R. isn't
economics, however, though most economists have some O.R. training.
It's applied mathematics. Since its origins in World War II to its
recent resurgence fueled by the explosion in raw computing power, O.R.
has developed analytical models of the tradeoffs and uncertainties
involved in problems ranging from inventory management to police
deployment, from scheduling sports leagues to determining how many
people to call for jury duty.

Taking the kids to Disney World this summer? Operations research will
be your invisible companion, scheduling the crews and aircraft,
pricing the plane tickets and hotel rooms, even helping to design
capacities on the theme park rides. If you use Orbitz to book your
flights, an O.R. engine sifts among millions of options to find the
cheapest fares. If you get directions to the hotel from MapQuest,
another O.R. engine spits out the most direct route. If you ship
souvenirs home, O.R. tells UPS which truck to put the packages on,
exactly where on the truck the packages should go to make them fastest
to load and unload, and what route the driver should follow to make
his deliveries most efficiently.

At the park, O.R. can even let you skip the lines for the most popular
rides. For Epcot's new Mission Space ride, for instance, you can join
a "virtual queue" using the FastPass system introduced in 1999. A
computer issues a pass that tells you when to claim your spot at the
front of the line. But it doesn't just tell you to come back after an
arbitrary length of time, say, an hour and 15 minutes. Rather, to
calculate a return time for each guest in the face of constantly
shifting waiting times, the virtual queue's software takes into
account how many people are standing in the real line, how many are
already in the virtual queue, and how many of each group the park
wants to admit each time the ride opens up.

"That's the O.R. piece," says Irv Lustig, manager of technical
services for ILOG Direct, a software and O.R. consulting company
headquartered in Gentilly, France, and Mountain View, Calif. He
visited the parks in February to see how Disney uses O.R. and woo the
company as an ILOG client.

After decades in which the field's progress was mostly theoretical,
computers have finally gotten powerful enough to collect the data and
deliver the problem-solving solutions that O.R. has been promising
since the heady days of the New Frontier. Beginning in the 1980s, when
American Airlines demonstrated that airlines could save billions of
dollars using O.R. techniques to design their schedules, O.R. has
become an increasingly important, though largely invisible,
contributor to rising productivity.

"What we talked about when I was a young graduate student are still
the things that we talk about now, except then we could only talk
about them," says Jack Muckstadt, a Cornell professor who entered the
field in the early 1960s. "Now we can actually do them."

Indeed, when Irv Lustig got his doctorate in operations research from
Stanford in 1987, his thesis was controversial. Although it had the
obligatory theorems and proofs, it also included computational work
that some members of Stanford's O.R. department (now its department of
management science and engineering) thought too lowbrow.

By contrast, he says, today O.R. students who want to just do theory
have a hard time. "Everybody wants to know, `What does it mean on a
computer?"' says Lustig, whose work has included creating the National
Football League's schedule. "That's a big culture change."

. . .

O.R. started as a way of bringing scientific thinking to the complex
problems of warfare: How do you find enemy submarines? How many
bombers do you need to make sure a critical target is destroyed? When,
where, and with how many troops and what equipment should you make an
amphibious landing?

In World War II, scientists from a wide range of fields attacked
military problems with a potent combination of empiricism and
mathematical models. When airplanes came back riddled with holes from
enemy attacks, for instance, the intuitive response was to reinforce
the armor where the holes were. But, noted the scientists, those were
the planes that made it back. They didn't need more armor where they
were hit. The real challenge was to figure out the places that had
been hit in the planes that went down.

"It was a lively, informal, paradoxical exchange of ideas between
amateur and professional war makers and it produced some brilliant
successes," wrote James R. Newman in The World of Mathematics,
published in 1956, which cited O.R.'s role in simplifying supply
lines, providing a quantitative basis for weapons evaluation, and so on.

But O.R. didn't live up to its postwar hype, its implicit promise to
"solve everything." Militarily, it could attack certain tactical
problems but, as the Vietnam War illustrated, O.R. wasn't the right
tool for addressing strategic issues of where, or why, to fight. Even
for mundane business questions, like how to design sales routes or
what inventories to hold, O.R. specialists often lacked the data and
computing power to turn their models into practical results. By the
1970s, the Vietnam War had made O.R.'s military applications and
Pentagon funding suspect in universities, and businesses were
gradually disbanding their O.R. groups.

For decades, the academic discipline retreated to theory. Scholars
built their reputations on mathematical proofs, largely abandoning
empiricism or real-world problem solving. Some O.R. veterans blame the
pure-math imperialism common to many theory-based fields for this retreat.

"Some time in the 1970s or 1980s, O.R. was in a sense hijacked by
mathematicians who insisted on imposing their view of rigorous
mathematics onto the field. This placed much less emphasis on modeling
and empirical work," says Richard C. Larson, a professor of civil and
environmental engineering and engineering systems at MIT and for 15
years the codirector of the institute's Operations Research Center,
which recently celebrated its 50th anniversary. "In some OR journals
today, the only empirical data are, `Date of submission' and `date of

Other O.R. scholars argue that theory was the only way to advance the
field in a world of scarce data. "A paper would start, `Here is an
interesting problem. If I had all these data, this is what I could
have done.' So the problem was challenging, but the focus was on
theory, because the data to support it did not exist," says David
Simchi-Levi, a professor of engineering systems at MIT.

But in the 1990s, the data became available. Now corporate information
technology systems collect unprecedented amounts of data -- on costs,
sales, and inventories, in itemized detail and real time. Wal-Mart and
Procter & Gamble, for instance, know exactly how many 200-ounce
bottles of liquid Tide Free have sold in which stores today. That
information in turn determines how many new bottles are shipped from
which warehouse tomorrow.

. . .

Simchi-Levi exemplifies the new generation of O.R.
scholar-practitioners. He entered the discipline as a theoretical
mathematician "focusing on algorithms and the theory behind different
logistics problems," but was drawn into applications in 1992 when the
New York City school district called, looking for help with its bus

Intrigued by the enormous potential of applying O.R. techniques to
logistics problems, in 1995 he and his wife Edith, a software
developer, started a Chicago-based company called Logic Tools to apply
O.R. techniques to supply-chain problems. Tweaking such mundane but
strategically critical decisions as where to site plants, when to
restock, and so on, can provide enormous productivity boosts.

In his work, says Simchi-Levi, mathematical theory and business
applications complement each other. "When I go to a company, or when
we develop a new product," he says, "I am familiar with the state of
the art in terms of engines and algorithms. For instance, the
inventory positioning technology is very, very recent, even in academia."

As ubiquitous as it is invisible, O.R. is a crucial ingredient in the
productivity surge often credited to information technology. "The real
driver of the productivity resurgence that we've had since 1995 has
been the way the technology has allowed changes in business processes
and the reorganization of work," says Erik Brynjolfsson, an MIT
economist. "For every dollar of [information technology] there are 9
to 10 dollars of organizational change, human capital, and other

O.R. is only part of that story, since companies often have to make
major organizational changes to reap its benefits. But without O.R.
problem-solving, many management innovations couldn't take place.

"Having data doesn't give you productivity. Having better decisions
gives you productivity. So if O.R. is all about the science of making
better decisions, then this is clearly an area in which we'd like to
claim preeminence," says Michael Trick, a professor at the Tepper
School of Business at Carnegie-Mellon and the former president of the
professional society INFORMS, the Institute for Operations Research
and the Management Sciences.

Trick's consulting projects include designing each year's Atlantic
Coast Conference men's and women's basketball schedules. Arranging 16
games among the nine men's teams may sound easy, but it requires
systematically sorting through hundreds of millions of possible
combinations looking for the best way to satisfy dozens of conflicting

"You don't want to play too many consecutive home games. You don't
want to play too many consecutive away games. You have to make sure
that every team has the same number of weekend home games," explains
Trick. "There are various games that have to be played -- Duke-North
Carolina is always played on the same day. And then the TV networks,
who are paying for all this, have strong views on how they would like
games to line up, so they can create a successful TV schedule. You
don't want to have all the good games on the same weekend. You want to
spread them out, so that every weekend there's a hot ACC game. All
those things go into play."

Thomas L. Magnanti, the dean of engineering at MIT and previously the
codirector of the institute's Operations Research Center, is
optimistic about the field's future. Until recently, most O.R.
scholars worked either in business schools, where the field is usually
called management science, or in departments of O.R. or industrial
engineering. Now, he says, departments like mechanical engineering and
electrical engineering are hiring O.R. specialists.

Magnanti calls O.R. "a liberal education in a technological world."
Just as a classical education once prepared students for a wide range
of endeavors, from theology and science to diplomacy and warfare, he
argues, so the habits and tools of O.R. are widely applicable to
contemporary problems.

"You can do finance today, manufacturing tomorrow, telecommunications
the day after. You can move from field to field and make contributions
that have impact on all those fields," says Magnanti. "We do health
care. We do criminal justice. You name it, we do it."

Virginia Postrel ( is author of The Substance of
Style: How the Rise of Aesthetic Value Is Remaking Commerce, Culture,
and Consciousness (HarperCollins).

No comments: