The Atlantic Monthly | February 2002

Notes & Dispatches

Washington Desk

Councils of War

Military spinoffs have transformed civilian life. The momentum right now may be running in the other direction

by James Fallows


There is no mistaking the excitement in Washington when world news originates here. Through the second Clinton Administration it was easy to think that a drive down Highway 101 in the San Francisco Bay area brought one closer to the real centers of power Oracle, Intel, Cisco than a drive along Pennsylvania Avenue from the White House to the Capitol. In Silicon Valley and Seattle the technology industry's leaders talked about the "withering away of the state," and in Washington the arrival of technology-driven prosperity was the central fact of political life.

However distant that seems now, the corrective reaction is, perhaps inevitably, going too far. I may be biased from having spent several years in Seattle and San Francisco before returning last summer to Washington. But now that the state is back, I am struck by the assumption here that if there is truly significant technology at the moment, it is the kind the military has used in Afghanistan. During the weeks when Taliban forces were collapsing, I did see three applications of technology with important economic, political, and even terrorism-related implications. Each was plain old civilian technology.

In one case the technology is e-mail, which has made possible the "open-source intelligence" movement. For decades diplomats and soldiers have bitterly joked that most important international secrets are likely to show up in the newspaper before they make their way through classified channels. Obviously, governments can still keep secrets. An illustration: three months after the terrorist attacks the Federal Aviation Administration was still enforcing strict "no-fly zones" ones forbidden to private noncommercial aircraft over three cities. Two were the terrorists' targets: Washington, the political capital, and New York, the financial capital. The third was ... Boston. Not San Francisco, capital of the technology industry; not Los Angeles, capital of America's image-making industry; not Chicago, capital of exposed skyscrapers. I asked Steven Brown, the FAA official in charge of airspace, why Boston? Because the planes that hit New York took off there? He said, essentially, If you knew what we know, you'd understand. What he actually said was "The vulnerabilities in Boston, those known to the public and others, are unique." Until we do know what he knows, there's no choice but to take it on faith. Maybe this is where Dick Cheney has been.

But the strictures secrecy requires can make it hard for armies or security units to get full, timely information in emergencies. One solution is to circulate non-secret information. In the mid-1980s the retired Air Force colonel  John Boyd  attracted adherents, especially in the Marine Corps, with his view that "fast feedback" loops were the key to military success. That is, the army that could observe and react to its opponents' movements the fastest would be the most likely to prevail. A young Marine captain named G. I. Wilson drew from Boyd's work the idea that the military should look for information as widely as possible. "It takes both unclassified open source resources and classified intelligence to win in today's information age," Wilson  wrote in the  Marine Corps Gazette  in 1995, with Major Frank Bunkers.

In practice "open-source resources" means what the best foreign correspondents and embassy political officers have always tried to keep abreast of, but on a bigger scale: reports in local papers, sudden changes in what's available in stores, snippets from the radio. Over the past decade Wilson and his colleagues have set up several electronic networks. The largest, called Access Intelligence (AI), connects hundreds of people in the defense, intelligence, law-enforcement, commercial, and academic worlds. It works like a normal list server or electronic mailing list: one person posts a message and everyone else receives it as e-mail moments later. The AI network often produces a hundred or more messages a day; recipients quickly scan the titles for subjects they are interested in. Although many AI members have security clearance, the material posted is strictly "open source" publicly available news reports or personal observations. That way the question of violating security rules won't come up.

AI has proved a valuable supplement to the slower, more controlled channels of official communication much as cell phones did for many civilians on September 11. For instance, Rick Forno, a computer expert who helps to operate the AI list, was in a building overlooking the Pentagon; he posted real-time reports about areas of damage and unfolding events before some of them appeared on CNN. Wilson, who is now a colonel based at Camp Pendleton, has relayed messages to ships' crews during (pre-Afghanistan) combat operations. "I can tell them what's being reported here, and they compare it to what they are seeing," he told me recently.

Open-source intelligence "frequently appears less valuable than classified information because it does not carry the classification mystique," Wilson wrote in 1995. "Because it appears less valuable, it is shared more freely and used more. The irony is by sharing it more the information's value and usefulness increases." Within the Pentagon, Wilson told me, reports that were posted on AI have been stamped with classified markings and used in briefings. An old trick of John Boyd's, Wilson said, was to get data into circulation by leaving it in "the head."

Still, the AI network doesn't get respect. "It's not popular with the intelligence community, because it doesn't cost anything," Wilson told me. (Forno and Bill Feinbloom, a former Green Beret, run it as volunteers, and it is free to all users.) "But you've got about three hundred people acting as individual sensors, from a whole variety of backgrounds. I may say something that seems commonsensical to a Marine, but someone who's a physicist will come back and say no, it can't have worked that way."

If the AI network is the application of e-mail to the military-intelligence business, a new company called Development Space represents the application of eBay to international aid. In the quarter century plus of the personal-computer age a few seminal applications have suddenly made computers necessities for new groups of people. The first was  VisiCalc,  the original spreadsheet program, whose introduction in 1979 gave small businesses a reason to own computers. The next was the coming of e-mail. And the most recent is  eBay , the online auction site. Whereas, for instance, offers a faster, more convenient version of a familiar shopping experience, eBay creates something that didn't exist before: a self-policing worldwide market matching buyers and sellers of even the most obscure goods. I am generally skeptical of "perfect markets" as laid out in economics textbooks, but an eBay auction for a used car, a signed baseball glove, or a new digital camera comes close. Those who want to sell have the largest audience of buyers; those who want to buy have the largest selection to choose from; and each party can judge whether to trust the other by means of a rating system based on past transactions (and a cautionary label on those with no record yet).

Dennis Whittle and Mari Kuraishi , two employees of the World Bank who had served around the globe, decided in 1998 to try to match resources and need just as directly in the public sector. Their first approach was bricks-and-mortar: a one-day Innovation Marketplace inside the atrium of the Bank's headquarters, in Washington. Normally proposals for Bank projects wend their way through a tedious multi-stage vetting process. On this one day anyone who worked for the Bank could set up a little booth, science-fair style, and make a pitch for a project; at the end of the day a jury would award grants to the best ones. More than a hundred teams made presentations, and eleven got awards, totaling $3 million.

Whittle and Kuraishi next persuaded the Bank to hold a two-day fair, with applications accepted from anyone who wanted to come and present an idea. More than 1,100 groups, from eighty countries, sent proposals. The heart of the program was letting people who knew firsthand about a local need or dream a well, a road, a small business explain what the money could do. A group of war widows in Bosnia, for example, offered a plan for a small, high-end knitting operation. The World Bank brought more than 300 finalists to Washington; and the forty-four winners got grants averaging just over $100,000 and totaling about $5 million. (The war widows won, and now they are prosperous, selling their output mainly to fashion houses in Europe and the United States.)

Electronic publicity explains the tenfold increase in applications. "Once this idea gets into e-mail circulation, it is amazing how fast it gets around the world," Whittle told me. "People who didn't have Internet access were contacted by those who did and encouraged to try. One Turkish guy was strutting around like a proud father at a Phi Beta Kappa ceremony five of the finalists had found out about the program from him."

Whittle and Kuraishi thought that if the concept worked despite the real-world impediments of getting applicants to one place at one time, it would work all the better if it were also implemented electronically. In 2000 they resigned from the Bank, and just as the Internet economy was beginning to falter, they created an online company,  Development Space , which began operation last month. Like eBay, it is meant to let the "market" in this case for development aid clear at minimum cost and with little or no bureaucratic interference. People who want money for vaccines, for an orphanage, for a small factory can prepare online descriptions of their projects, with help from advisers, if necessary, in drawing up business plans. Foundations and government aid agencies that intend to give money but also individuals who will give, say, $250 if they think it will help survey the projects and decide which to support. Various inspection and feedback systems will establish a track record, as on eBay, and follow up to see how the money was used.

A number of environmental foundations have approached Development Space to explore using this platform to find projects to support. If America's past wars are any guide, huge amounts of recovery assistance will soon be headed to Afghanistan, Pakistan, and who knows where else. This model could be a lower-cost, better-targeted way of getting it there.

Open-source intelligence and an eBay for foreign aid are extensions of the Internet's model of information flow. The third innovation comes from a company called  Athena Technologies , and it's an extension of the ongoing hardware revolution.

In 1992 a young South African named David Vos was preparing for his Ph.D. in aeronautical engineering at MIT. His dissertation project was to build a guidance system that would let a unicycle propel itself, with no rider. I have seen a videotape of his presentation. On an arctic day in Boston a shaggy, tired-looking graduate student in a ski jacket (Vos) hovers inches away from a unicycle, ready like a parent to reach out and support it. But the unicycle keeps itself erect and propels itself around a basketball court, responding to commands from something on top that looks like a cake box.

The mechanism inside the box was Vos's achievement: a system of inertial sensors and quick-response motors that could detect changes in the unicycle's balance eighteen times a second and issue the right corrective command. A tricycle is of course inherently stable, and a bicycle has a kind of stability when moving. But because a unicycle is always trying to fall over, most people cannot react quickly enough to control it, and no mechanical device had previously been able to.

Ten years later Athena, Vos's company, has produced a device that drives not unicycles or people, like the inventor Dean Kamen's highly publicized "IT" vehicle but airplanes, and with significant implications for defense. The device is known as  GuideStar , and it is about the size of a car radio. Packed with inertial sensors and logic circuits, it is capable of detecting and reacting to changes fifty times a second and of flying aircraft that are too tricky or unstable for human pilots to control. Vos made another video to underscore the point. In it an odd-looking airplane one big wing and no tail sits on a runway. Without a tail an aircraft would be even more unstable than a unicycle and, according to simulation models, would require such constant and immediate adjustments that even a skilled pilot would quickly lose control of it. But in the video this jet-powered tailless plane zooms off the runway and then circles several times before it lands, to the joyous whoops of Vos's team in the background.

GuideStar has civilian potential for instance, as part of the autopilot in small planes or airliners, permitting them to land in circumstances that overwhelm the pilot. Another device shown in Vos's videos suggests military and civilian uses alike. This is a vehicle, built by the Micro Craft company and guided by Vos's systems, that looks like a large smudgepot, with a cylindrical base and a vertical shaft, powered by a compact engine. It can take off straight up, maneuver itself around corners, travel at altitudes from treetop level to a few hundred feet, and land straight down. In the civilian world this could be a jazzy counterpart to Kamen's "IT" vehicle, delivering parcels rather than people. For the military it could also be a remote sensing device, far cheaper than current pilotless drone aircraft.

But what Athena has been touting since September 11 is that its GuideStar controls could be programmed to prevent any airplane from ever going someplace it should not. No airliner, we can assume, will ever be flown into a skyscraper again: the passengers will not let it happen. But in theory it could still happen with a FedEx or a UPS cargo plane. The coordinates of restricted areas and important buildings could be entered into the new guidance system, which could thwart a pilot's attempts to divert the plane. In principle the system could land the airplane at a military airfield if it sensed abnormal commands.

What do these innovations have in common, apart from reminding us of the fecundity of the high-tech world despite the Nasdaq's slide? They show two crucial traits of the civilian tech world in general, and these traits distinguish them from most military technology.

First, they are cheap. The open-source network is literally free to its users. Development Space plans to support the eBay model of foreign aid by taking a seven percent cut of all transactions, to pay for expert teams and authenticators much less than the overhead of most charities. The Athena controls are both cheaper and more powerful than current autopilots. "We come from the computer-industry mindset that the price has to keep going down," says Jeffrey Leonard, who is on Athena's board of directors.

It is easy to forget how important the race to cheapness was in creating the technology boom. Indeed, the Internet's main business problem is that users think content should be free. The contrast with military technology is sharp.

A B-52 bomber, for example, costs about $23,000 per flight hour just to operate; the B-2, which makes long treks to Afghanistan from its home base in Missouri, costs at least twice as much. During the Kosovo bombing campaign the United States reduced Serb defenses by firing HARM missiles, which lock onto the beam from a radar station and then destroy the station. The Serb army reportedly discovered that it could place microwave ovens in open fields and the HARMs would think the ovens were radar stations. Each oven cost less than $100; each missile it attracted cost $750,000. We pay any price for freedom, and the costs mount up. The idea of a race for cheapness has not spread from the civilian to the military world.

Second, these innovations don't try to replace what is best about human judgment and intelligence. The most popular breakthroughs in the commercial-tech market have let people do more of what they have always wanted to do: buy, sell, interact, explore. Open-source intelligence and Development Space follow this model as well. GuideStar does replace a human function, but a calculator-like one, at which machines should ultimately exceed human abilities as spreadsheets do, and language-translation programs do not. In principle the military would always prefer to use machines instead of men. Machines don't have grieving families; they don't need to be recruited and trained. Some of the most expensive boondoggles in military technology have involved attempts to mechanize the most sophisticated human abilities which include, surprisingly, the ability to detect patterns. Any human being can tell a camel from a car. Designing sensors that can reliably do so is very hard. That is why even in the phenomenal rout of the Taliban army, the bombing became effective only after special-operations troops, on foot and on horses, were there to identify the right targets.

This war began with a devastatingly brilliant bit of jujitsu, in which the very openness of our society and elegance of our technology were turned against us. The stages ahead will certainly call not for brute-force technical power alone but for a shrewd combination of human and technological abilities a lesson the military can take from the civilian world.