Mark P. Mills and Peter W. Huber
In transit across Manhattan on any given day are some 4 million letters, 3 million people, half a million motor vehicles and their contents, and half a million parcels—any of which may be carrying something lethal. Step by step, cities like New York must now learn to watch and track everything that moves. Airport screening is coming to much of the rest of civilian life; but it will have to be much smoother, faster, more accurate screening than airports have today, or life will just grind to a halt.
However much local governments may yearn for a one-stop federal solution to their new security problems, much of the initiative and innovation will almost certainly be left to them. And for obvious reasons, the challenge will be especially great in places like Manhattan, where the largest number of potential targets attracts the largest number of cars, packages, and people—every one of which could be a bearer of destruction.
In the post–September 11 world, we know we have to see the plastic explosives in the truck before they detonate, the anthrax before it’s dispersed, the sarin nerve gas before it gets into the air-conditioning duct—and not just see it but recognize it. Our imaging systems will have to distinguish between the scaffolding on a Wonder Bra and the wiring on a bomb. We have, of course, slow, conventional ways of doing all this in a forensic laboratory. But the challenge now is to do it in bulk and in real time. How do you even begin to do this?
Threat recognition is, fundamentally, an exercise in the analysis and transmission of information. Imaging systems and transponders will generate huge amounts of digital data; intelligent networks, pattern-recognition software, and remote databases will make sense of the data as it comes in, chiefly by comparing the image at hand against large databases of stored images. We can’t begin to screen the flow of people and goods properly without a massive deployment of digital technology to perform these functions for us.
The first step is to divide the civilian world into two, separating the trustworthy cooperators from the noncooperators, so we don’t have to search every car, package, and pocket. The airlines already make this separation. They already have detailed profiles of most travelers readily at hand—from credit cards, travel agents, and travel histories—and they effectively enlist us as deputy security agents when they ask us about the packing and custody of our bags after establishing our identity. The FAA uses such data, in a limited fashion, with the Computer Assisted Passenger Prescreening System that helps target passengers whose checked baggage should get a thorough examination. The same principles can work to screen vehicles, packages, letters, and most everything else that moves.
To understand how, consider the way quite similar tools are currently solving the problem of highway congestion. The tollbooth transponders now spreading rapidly across the United States—New York’s E-ZPass system is an example—track cooperative targets very effectively and save willing cooperators a lot of time, compared with the toll collectors or coin baskets of old. The relatively primitive friend-recognition network to which E-ZPass belongs now extends across all of New York and from New England down to the Mid-Atlantic states. Other networks are more sophisticated, like the new beltway around Toronto, a toll road without tollbooths: transponders, cameras, and license-plate numbers provide the billing information. Three years ago, Singapore implemented a much more comprehensive version of the same thing—a card-based toll system designed to curb the massive traffic jams that choke that tiny island-state. Electronic devices mounted on gantries above the road deduct the appropriate charge as the cars pass under them. Even systems as simple as these can be linked up to security networks, too, and can do much to enhance safety, because so much of security comes down to establishing identity and tracking patterns of conduct—just the sorts of things that the automatic toll collectors already do.
Quite similar technologies that address the congestion of people—and in the much more demanding context of maintaining security—are now up and running. The brand-new iris-scanning system at Schiphol Airport in Amsterdam is a dramatic example: you register, and then stare briefly into a lens instead of presenting your passport. The iris is the most personally distinct feature of the human body, even more distinct than a fingerprint. And yes, the toaster-size camera can see through your glasses and contact lenses. As with E-ZPass, nobody has to participate; you’re free to wait in line instead. An affront to civil liberties? Hardly. You have to volunteer—and then pay about $100—for the privilege of getting your eyeball into their database. The plan is to let you shop the airport stores that way, too, rather than having to swipe your credit card.
Similar technologies can separate innocuous packages from the suspect ones that need closer inspection. Most everything that’s shipped by UPS, FedEx, and even the Postal Service (if it’s dispatched through a postal meter rather than with ordinary stamps) is already bar-coded and repeatedly scanned while in transit. This is how the package companies maintain websites that can tell you exactly where your inbound package is as it moves through their system. But technology is now set to move far beyond the printed code and the optical scanner. Motorola’s BiStatix radio-frequency identification technology, for example, combines silicon with printed ink to embed smart electronic tags ubiquitously in packaging materials, so that packages automatically announce their whereabouts, without any human having to scan them. There are profound economic advantages in being able to “talk to” your packages the entire length of the supply chain—this is the heart of shrinking your warehouses and your inventory, and building a business around just-in-time deliveries.
The civil defense advantages are equally apparent: a package that begins in the hands of a trustworthy shipper like Amazon and is tracked carefully while in transit from New Castle, Delaware, to New York City can be admitted without more ado into a downtown office building and opened with confidence. The by-mail terrorist invariably prefers stamps, a handwritten address, and a late-night drop into an anonymous mailbox.
All of these tracking and screening systems that identify cooperating and friendly targets are prey to sabotage, of course. Smart cards can be stolen. With all these systems, you have to have intelligent networks and databases behind the wall to keep watch for wolves clothed as sheep. The machines have to look for patterns of activity, for daily habits—and for breaks in them—just as Citibank already does when it spots what looks like an unusual spending pattern on your credit card. Try to buy four cheap color TVs in a store in a high-crime neighborhood you never visit, and a computer somewhere deep back in the network is almost certain to initiate an immediate inquiry from your bank.
Today we use these systems, which are ubiquitous and affordable, chiefly for mundane economic screening, but they inevitably will be extended to track security as well. In response to the anthrax scare, public health officials are already taking steps to tap into the networks and databases that pharmacies use to track drug purchases and restock their shelves. When unusual numbers of people in a neighborhood begin feeling lousy, they’ll begin buying all sorts of conventional medicines; those purchases may signal a biological attack well before hospital labs can spot the pattern.
Thus, while talk of national ID cards inches forward in the federal government’s marble halls, identification systems are already evolving organically, and very much more rapidly, throughout the marketplace. Local governments should help set standards and facilitate collaboration with the private sector, so that building owners, garages, toll collectors, banks, package-delivery companies, and many others already in the track-and-screen business can effectively coordinate and integrate the identification technologies that they’re deploying anyway. Yes, to be sure, the ACLU will protest; but like it or not, this is the way things should go and inevitably will go, too.
Once you can identify the vast majority of people, cars, and packages that are innocuous and can be waved quickly through the gate, you can inspect more closely the few that may not be, and if necessary interdict them.
Dealing with the noncooperative targets is a five-step process. First, you project electromagnetic waves across a wide range of frequencies, from radar, millimeter-waves, and infrared heat through visible light, and X rays. There are other alternatives, too—magnetic pulses and acoustic waves, for example. Second, you carefully look for what gets through or bounces back. Third, you intensely analyze the same, crunching the numbers to turn the massive stream of return data into a coherent image. Fourth, you make sense of it, generally by another massive round of number crunching for pattern recognition, comparing the image at hand with huge databases of images stored previously. Fifth, if you don’t like what you see, then you kill it, disable it, or at the very least shunt it aside for closer and more leisurely scrutiny. If you see anthrax, say, kill it with a burst of gamma rays or an electromagnetic pulse intense enough to shatter DNA.
In its most familiar form, “sight” begins with sunlight and ends with eyeballs. But how do you “see” the molecules in plastic explosives, in nerve gas, in cocaine, or in anthrax DNA? A dog’s nose is pretty good at this form of “vision,” which is why places like airports are filled with canines these days. Chemical laboratories can work out the molecular composition of anything, given a big enough sample and enough time to run it through a lot of bulky hardware. But for mass screening, in real time, you project photon power—electromagnetic power—across a wide range of different frequencies, and if you do that cleverly enough, you can see most anything.
Plain old light will take you a long way in distinguishing shiny aluminum from dull wood. But if you push the frequency up quite a lot further, through visible light and on up into X rays, you can see broken bones; and if, in addition, you also analyze what bounces back, you can get quite a good profile of molecular composition, too. “Light” in millimeter wavelengths penetrates all sorts of clutter—like foliage, clothes, boxes, and Sheetrock—yet bounces off things we do want to see, like metals, plastic, and flesh, much like X rays, but at much lower energy levels that pose no threat at all to human health. (A company called Millivision has even developed a millimeter-wave “flashlight” that detects respiration at a distance—a concept originally pursued to locate wounded soldiers but equally useful in searching collapsed buildings.)
Use still more of the electromagnetic rainbow—infrared bands, ultraviolet, and so forth—and you see still more. Aim the “light” from multiple angles—as a medical CAT scan does, for example—and you can build complete three-dimensional images. From airport scanners to on-board radar for cars, these things aren’t matters of theory or laboratory speculation any more. A whole constellation of imaging products is now maturing into commercial viability. All it takes is enough different kinds of “light,” and electronic eyeballs to match—and powerful number-crunching microprocessors to make sense of the huge amounts of data generated from these multi-spectral, multi-dimensional scanning systems—and you can see through just about any truck, car, pocket, suitcase, or package you want, looking for just about anything you choose.
If the seeing and recognizing of threats isn’t highly automated, the prospect of gridlock will prevent screening from happening at all. So real-time networking will be crucial, as in the federal program that provides law-enforcement officers with fast, nationwide access to fingerprint databases, together with pattern-recognition software that fully automates the process of finding matches. Right now, real-time fingerprint recognition is moving out into the field as a wireless application, so that an officer who makes a stop can check prints in three minutes. All wireless, all fully automated. It’s headed for three seconds, and the faster we get there the better.
Now just do the same for irises. And the backscatter X-ray profile generated by Semtex. Or sarin gas. And dozens of other threat signatures.
The data volumes are staggering. The automated luggage scanner in the catacombs of the airport today, as it spins a highly advanced CAT scanner around your toothpaste and dirty linen, generates so much data that it would take about 4,000 dial-up modems to push it down regular phone lines. The task is so challenging that, so far, only a handful of U.S. airports do it.
The computing requirements are staggering, too. While it takes a fair amount of a highly paid physician’s time to interpret a hospital CAT scan, the FAA insists on a suitcase automatically scanned and analyzed by computer every six seconds. It will be the same for buildings, stadiums, bridges, and tunnels.
A decade ago, none of this would have been economically feasible. It is today. It will entail a lot of new investment; but the technology is there, or very close to there—real, commercial, functional. It’s going to get deployed, not only at airports but even more widely on private premises and, later, for municipal uses.
So what happens when you see and recognize something you don’t like? You kill it. Many new options are materializing here, too, more than a few of them employing technologies very similar to the sensor technologies.
Here’s just one example. Pump up millimeter-wave power high enough—at present, this takes something more like a television tube than a semiconductor chip, but arrays of chips will get there soon enough—pump up the power, and you can cook things, or people, or hostile microorganisms, at quite a distance. If you choose, you can make anyone within several hundred meters feel like his whole body is touching a very hot light bulb, encouraging him to run away, fast. The Pentagon calls this “active denial technology.” If you shift the frequency a bit and go for a longer pulse, you begin doing more lasting damage: “terminal denial,” the Pentagon might call it.
Such high-tech power-projecting gizmos may seem to have little relevance for maintaining security in the heart of New York—until you begin worrying about anthrax, say. Killing biologics is already a familiar objective in the biopharmaceutical industry. Blood extracts, for example, provide proteins that are widely used for health care (as in vaccines)—but possible viral contamination is a major concern. Technology companies have found ways to tune pulses of power to destroy any DNA that may find its way into such products, without destroying the proteins. Similarly, the food handling and processing industry projects carefully controlled pulses of power to kill biological causes of spoilage without damaging the rest of the food too much.
But if we deploy all this screening and tracking technology right here at home, the ACLU will doubtless argue, aren’t we stepping into a horribly Orwellian future, a future utterly devoid of any privacy? In fact, we are not, and here’s why.
Most of the screening of the future will be entirely by machine, and the machines can be set up to respect a whole lot of privacy. As our pattern recognition gets better and better, more and more can be waved through the checkpoints without any human involvement at all. Most information need never be pulled out of the digital loop for human scrutiny. When you get good enough at detecting threats, then you invade privacy only when you should, and at no other time.
And even then, the surrender of privacy can remain, by and large, an opt-in choice. Nobody is forced to step through an airport metal detector, or send luggage through the CAT scanner in the basement. It’s a choice that goes with stepping into an aluminum tube called an aircraft—and the same choice will probably have to come with stepping into a cement tube under the Hudson River or onto a steel-and-cement edifice that spans San Francisco Bay. Civil libertarians won’t like the argument at all, but bridge and tunnel tolls already have the effect of excluding those who can’t or won’t pay, and government buildings already make screening a condition of entry, as do a growing number of public schools and museums. The wave-through screening systems, intended to speed passage through an otherwise resistant—properly resistant, lawfully resistant—gateway will survive the inevitable constitutional challenges.
The equally inevitable health and safety objections won’t stop the new screening technologies, either. Some are indeed dangerous—airport luggage screeners have to be carefully shielded, because X rays are up in the ionizing frequencies that can tear electrons off atoms and rip molecules apart—most notably our own DNA. But despite frequent theorizing to the contrary, the lower-frequency beams don’t ionize; they pack just enough punch to penetrate and reflect but not to separate electrons from atoms entirely. And happily, the new focus on altogether real threats appears to have made the public a good bit more skeptical of junk-science health scares.
Anyway, we really do want an Orwellian future—not in Manhattan, but in Kabul.
We all look for a quick end to al-Qaida, and we hope anthrax is over. And we certainly hope that a sufficiently quick, unambiguous, and violent victory in Afghanistan will impel other governments that might wish us ill to rein in their homegrown terrorists themselves and so avoid the fate the Taliban met in backing al-Qaida. Perhaps some of this will happen, but ultimately wars like this one can’t really be won, in the conventional sense of the word. Terrorist “wars” will continue, in one form or another, for as long as we live.
Many pundits insist that technology won’t win such wars; only ground troops will. They have it exactly backward. It is only our technology that will let us survive the enduring war against terrorism.
We are destined to fight a never-ending succession of micro-scale battles, which will require us to spread military resources across vast expanses of empty land and penetrate deep into the shadows of lives lived at the margins of human existence. Their conscripts dwell in those expanses and shadows. Our soldiers don’t, and can’t for any extended period of time. What we have instead is micro-scale technology that is both smarter and more expendable than their fanatics, that is more easily concealed and more mobile, that requires no food and sleep, and that can endure even harsher conditions.
Technology in hand today fundamentally changes the calculus of war, especially wars that in any other era would have become wars of human attrition, wars that favor the side most willing to dispatch young men into the line of fire.
The next-generation sensing technologies are far smaller, cheaper, and more powerful than today’s spy satellites or SR-71 Blackbird spy plane. The old ones were fine for spying on the military-industrial complexes of nation-states. But now we have sensing technologies that bring to the battlefield abroad, and to the vast arena of civilian defense here at home, the same wizardry that transformed the mainframe computer into the Palm Pilot, the television tower into the cell phone.
Equipped with such sensors, the Predator “remote-piloted vehicle” (RPV), about the size of a pterodactyl, can see through fog, foliage, and snow, can see objects buried underground, can see and track bullets back to their source, and can shoot Hellfire missiles. The Prowler, roughly half the size, and the high-altitude (65,000 ft.) Altus boast similar capabilities. Military RPV development programs are now focused on fully functional bat-size and even butterfly-size RPVs, which have already been built. AeroVironment’s electric-powered Black Widow typifies a new family of tiny fliers, with two-mile range and live color video downlink. The company is now developing a wing-flapping, dragonfly-like Microbat that weighs half an ounce, including camera and telecom downlink.
These devices and many others like them are already well past lab-bench theory; they are close to the point where they can be churned out at low cost and in large quantities, like artillery shells. Video cameras are down to lipstick-size; million-pixel deerfly-size cameras are under development. Until quite recently, the best infrared detectors were the size of a coffee can, and they strained to detect an engine’s hot exhaust, but today’s thimble-size units can now detect the microscopic heat difference between a red and a white stripe on an American flag from four miles out. Sensors to pick up light or X rays or millimeter waves—the same sensors that can “see” threats in airports or at the gateways to bridges and tunnels—are now built from the same materials as microprocessors, in the same factories, using the same core tools. Advanced radar will soon be far cheaper and more ubiquitous than binoculars and telescopes once were. It will be built from arrays of single-chip integrated transmitter-receivers, which defense contractors can incorporate by the hundreds or thousands into different platforms, weapons systems, and munitions.
Manufacturers are now etching sensors alongside microprocessor, memory, and transmitter on a single semiconductor chip, and before long they’ll be able to build by the bucketload complete sensor modules—with built-in laser, memory, and CPU—that are no larger than a grain of sand. Dispersed along roadsides, hills, and trails, they will report just about anything that may interest us—the passage of vehicles, the odor of explosives, the conversations of pedestrians, the look, sound, weight, temperature, even the smell, of almost anything.
The cost of such chips is plummeting, as semiconductor costs generally do—down tenfold in the last decade, with another threefold drop projected in the next few years. Whereas yesterday’s military technologies always grew more expensive, today’s get progressively cheaper, even as their performance doubles and redoubles every few years.
A decade ago, most of this would still have sounded far-fetched. But today we have in place a trillion-dollar infrastructure of semiconductor and software industries, with deep roots as defense contractors. Two years ago, the Department of Defense established the Sensor Information Technology Program to develop the software that will manage distributed networks of communicating micro-sensors. Does anyone now doubt American code writers’ ability to write such software? Or their motivation to do so?
Can the other side turn these same technologies against us? No chance. Building and using them requires a digital infrastructure and a digital mindset. The very notion of someone waging a “digital jihad” is oxymoronic. Even if our enemies steal these devices, they can’t get them running, or keep them running, without developing a tech-savvy population. And in any event, it’s a pretty straightforward matter to lock up digital technologies from the inside. You just build in a secure, Y2K-like bug, with whatever terminal date you like, and right on schedule the weapons will turn into plowshares, or sand itself.
However advanced our power to destroy, we obviously can’t bring down the other side’s skyscrapers—they don’t have any. This kind of enemy never will. So our longer-term objective must be to infiltrate their homelands electronically, to the point where we can listen to and track anything that moves. We can then project destructive power precisely, judiciously, and from a safe distance—week after week, year after year, for as long as may be necessary.
Properly deployed at home, as they can be, these technologies of freedom will guarantee the physical security on which all our civil liberties ultimately depend. Properly deployed abroad, they will destroy privacy everywhere we need to destroy it.
It may seem anomalous to point to micro-scale technology as the answer to terrorists who brought down New York’s tallest skyscrapers. But this is the technology that perfectly matches the enemy’s character and strength. It can be replicated at very little cost; it is cheap and expendable. Small and highly mobile, it can be scattered far and wide—across Manhattan, the richest place on earth, and also across the Hindu Kush, the poorest. It can infiltrate, image, track, and ultimately destroy at the peasant-soldier’s scale of things. It can win a war of attrition.
It is a horrible vision. It gives us no joy to articulate it. But at home and abroad, it will end up as their sons against our silicon. Our silicon will win.
Read article at CityJournal.com
© 2002 City Journal