Discover, September 30, 1992
Artur Ekert can feel secure. If his friends are within 200 yards of him, he can send messages to them in an unbreakable code. It’s not that his signals are immune to wiretapping; snoopers can listen in. It’s not that his code is too complex for anyone to crack; it’s actually fairly simple. Ekert’s secrets are guaranteed by the basic laws of energy and matter. They forbid anyone from cracking his code, ever. That includes the army, the CIA, and, assuming that laws of physics are really laws, even God.
Ekert, a physicist at Oxford University, is part of an international effort to create perfect cryptography. Instead of inventing new schemes for scrambling the alphabet or electric signals, these researchers use the weird behavior of subatomic particles to cloak information. The rules that these particles follow are the rules of quantum physics. In the quantum world, basic aspects of reality don’t exist in a well-defined way until someone observes them. Physicists such as Ekert exploit this fickle behavior: they send instructions for coding and decoding messages on a stream of particles. While the particles are in transit, the instructions don’t exist in any definite form. If someone taps into the stream and tries to “read” the particles, they don’t cooperate. Any symbol in the code can change to any other symbol, by pure chance. The wiretapper is out of luck, and the code remains secret.
***
Ten years ago the idea of practical quantum codes was just this side of insane, an amusing daydream for people with lots of tenure. Now prototype coding devices are sitting in labs, and several more are on the drawing board. Quantum cryptography may be on its way to reality.
Ekert’s coding scheme may be strange, but he hasn’t broken completely with traditional methods. Codes go back thousands of years, and they all share the same basic structure. There’s a message to be coded, known as the plaintext; there’s a way to encode it, called the key; and there’s the coded version of the message, known as the ciphertext. Sometimes the key is a device you can touch. In Sparta around 400 B.C., for instance, soldiers would wrap papyrus or leather in a tight spiral down a long wooden staff. They would then write a message straight down the length of the staff, rotating the staff after each line. Unwound, the message ATTACK AT DAWN might look like ACDTKATAWATN. To reconstruct the text, readers needed to wrap the strip around an identical staff.
You don’t have to be a physicist to see the weakness of this scheme. If the enemy got hold of a staff, they could intercept a message, read it, and send it on to the Spartans, who would be none the wiser. This is the basic weakness of any coding scheme: if the key exists anywhere as a real object, it is vulnerable to being captured or copied.
Keys don’t have to depend on physical props. Instead, they might consist of secret rules for shuffling the message. Julius Caesar used a simple rule: change A to D, B to E, C to F, and so on. At its heart, Caesar’s code is mathematical, just like modern codes. If you let numbers stand for letters, you can think of Caesar’s code as a scheme of addition. By adding a key to a plaintext, you get the ciphertext.
A message:
A T T A C K A T D A W N
changes to:
1 20 20 1 3 11 1 20 4 1 23 14
Add the key:
3 3 3 3 3 3 3 3 3 3 3 3
and you get:
4 23 23 4 6 14 4 23 7 4 26 17
which equals:
D W W D F N D W G D Z Q
Caesar’s code isn’t safe either, though. There’s no physical key to capture, but the rule itself can be deduced from the ciphertext. All a code breaker has to do is count how often each of the coded letters shows up, then compare the results with the statistical frequency of letters in the original language (in English the most common are, in order, E, T, O, A, N). The code breaker will quickly discover the pattern and figure out the key.
In 1918 an AT&T engineer named Gilbert Vernam discovered a coding scheme that won’t reveal any pattern. AT&T was using teletype machines that communicated, like primitive computers, with on-off signals, which can be represented by 1’s and 0’s. A letter would consist of five numbers. For a key, Vernam suggested, take a random string of 1’s and 0’s, a string as long as the message itself, and combine each bit of the message and the key according to the following rules: 1 + 1 = 0, 0 + 0 = 0, 1 + 0 = 1, and 0 + 1 = 1.
Plaintext: 1 1 1 1 1 0 0 0 0 0
Key: 0 1 1 0 1 0 1 0 1 1
Ciphertext: 1 0 0 1 0 0 1 0 1 1
(The numbers don’t have to be 1’s and 0’s, and the details of the rules for addition can be changed; what matters is the randomness of the key.)
Some years after Vernam introduced the concept of random keys, mathematicians were able to prove that no matter how long a ciphertext is, a code breaker can’t figure out what the key and the plaintext are, even though he may know the rules by which the ciphertext is determined. The randomness in the key wipes out the patterns in the plaintext.
There’s just one hitch, though. Say you and your cohort have agreed on a key and use it to send two messages. If an eavesdropper overhears both ciphertexts, he can use them to read both messages. All he has to do is add them together. The key gets canceled out in the process, and he’s left with the sum of the two messages. It’s then relatively easy to tease out both of them. A computer can systematically try different words at different places in one message and see what that would make the other message become. So the only way to maintain perfect secrecy is to use a new string of random numbers as a key for every message.
That might seem easy, but getting keys to the places they’re needed turns out to be a logistical nightmare. If a government intelligence agency, say, wants its agents to use Vernam’s system, it must either deliver long strings of numbers to the spies or else have agreed-upon secret rules for looking up numbers in a published source. During World War II a Soviet spy ring in Switzerland communicated with Moscow by taking numbers from a book of trade statistics. Each page was full of long lists of numbers—so many tons of iron imported, wheat exported, and so on. A spy would pick a page, row, and column at random to begin the key. To let the decoder back in Moscow know where in the book to look, the spy would encode within the message the numbers that would tell the receiver which page, row, and column to turn to.
Later, during the cold war, Soviet typists banged out reams of random numbers that were reduced to booklets the size of postage stamps. The lists contained some vestiges of order—the numbers alternated between the left and right sides of the typewriter, and since the typists were mostly right-handed, they tended to start with those numbers. Still, Soviets treasured these secret numbers so much that in 1956 they allowed their embassy in Canada to go up in flames rather than let firemen get a glimpse of the keys they kept in the building.
Vernam’s scheme is still the best one going: it codes messages on the hot line between Washington and Moscow. Both parties to a hot conversation use long, identical keys and discard numbers as each message uses them up. Such keys, of course, must be very well guarded—or if they’re taken from published sources, the agents who know the secret must be very loyal. And that’s the problem with Vernam’s system. In everyday life, it’s too easy to break security.
But just imagine how popular Vernam’s system would be if people could openly transmit their keys to each other, with complete confidence that no one could copy them. That, apparently, is a guarantee quantum physics can make.
The wonderful thing about quantum reality, from a code maker’s point of view, is that it just doesn’t make sense. Quantum particles don’t fit our ordinary commonsense picture of how things exist. We usually think of something that is as occupying one place at a time, moving in one direction at a time, and so on. But subatomic particles can be in several places and have several different amounts of energy at once. The different versions of a particle are said to be superimposed on each other, some strongly, some weakly. By applying equations that describe the particle’s behavior in a certain environment, physicists can say there’s a certain probability that they’ll catch it acting in a certain way.
Not until experimenters actually observe the particle does it “collapse” into a particular way of being. When observed, it decides which of many possible places it will occupy at that moment and in which direction it will be moving. But no one can measure every aspect of the particle at the same time. If you measure an electron’s position, for example, you forfeit your chance to know its velocity at that instant. Your measurement affects the particle, and the other information is lost. This cruel fact of nature, known as the uncertainty principle, shows up everywhere in the quantum world.
How can you harness uncertainty to code information? First, you have to pick some aspect of a particle’s behavior to use as your key. Take, for example, photons, or particles of light. Photons vibrate as they travel, which is the same thing as saying that the electromagnetic field of the space they move through changes. Some photons create a vertical electric force, pointing up, then down, then up again. Others create a horizontal force, moving from left to right. Still others create a force vibrating at some angle in between horizontal and vertical. When a stream of photons all vibrate in the same direction, they are called polarized light.
Say you decide to use polarization as a code. The different angles at which a photon might vibrate can represent different numbers: vertical polarization can represent 0, and horizontal can stand for 1. To use such a key, you have to be able to detect polarization. That’s easy. There are simple detectors in every drugstore: Polaroid filters that screen out sunlight.
Sunlight is unpolarized: its photons vibrate in every conceivable direction. Polaroid sunglasses let through light that’s polarized in one direction—usually vertically—and absorb the rest. If the photons that come through one lens of those sunglasses then go through another Polaroid lens oriented the same way, the light doesn’t change. But if you rotate the second lens, the light dims. When you turn the second lens 90 degrees from the first lens, no light gets through at all.
Since all of the light coming through the first lens is vertically polarized, you’d think that turning the second lens even 1 degree off vertical would shut all those photons out. But each photon is a quantum particle. Presented with a sunglass lens at a different angle, it has a certain probability of suddenly switching its polarization to match it. The further away from the vertical you turn the lens, the less likely the photon is to make the increasingly large jump to a new polarization. At 45 degrees, a photon has a 50 percent chance of getting through, which translates into the light you see becoming half as bright. If you turn the second lens 90 degrees, there’s no chance of any photon vibrating that way, and no light passes through.
Polarization detectors work in much the same way. If you know that only vertically and horizontally polarized photons are coming your way, you can use a detector to distinguish between the two types of photons with 100 percent confidence. But if a diagonally polarized photon comes along, it will confuse you. Let’s say you’re set up to let horizontals pass through. This diagonal photon is tilted 45 degrees from the detector, and so it has a 50 percent chance of flipping to a horizontal and setting off the device. But if it doesn’t go through, you have no way of knowing if it was diagonal or if it randomly chose to look vertical.
***
In the early 1970s physicist Stephen Wiesner, then at Columbia, theorized that polarized light could somehow be used to encode objects such as dollar bills to make them invulnerable to counterfeiting. But he was unsuccessful at getting a paper on the subject published, and so his idea languished until 1979, when a friend of his, physicist Charles Bennett of IBM’s research center in Yorktown Heights, New York, joined computer scientist Gilles Brassard at the University of Montreal to transform Wiesner’s idea into something workable: a random key created by polarized photons.
In their scenario, someone—Alice, as they like to call her—wants to send her friend Bob a message. First they have to agree on a key that’s as long as the message, so Alice fires dim light through filters and light-bending crystals to create a string of photon pulses. Each one is randomly polarized in one of four directions: horizontal, vertical, left-diagonal, or right-diagonal.
Alice sends:
/ \ \ | — \ | / — \
These pulses travel to Bob. Alice and Bob have agreed on two different ways to measure photons. They can set their detector to distinguish between horizontal and vertical photons—to represent this setup they use the symbol +. Or they can measure left- and right-diagonals—for this they use the symbol X. To preserve secrecy, Alice doesn’t tell Bob how to set his detector; he just sets it up randomly for each photon and records what he sees.
Alice sends:
/ \ \ | — \ | / — \
Bob uses:
+ X + + X + X X X +
Bob reads:
— \ | | / — \ / \ —
Bob and Alice then talk to each other. Alice can call Bob on the phone or they can have their computers communicate normally. In either case, they need take no security measures. An eavesdropper—call her Eve—can listen in to what they say. Bob tells Alice which settings he used, and she tells him which settings were wrong. Then they hang up. Alice hasn’t told Bob (or Eve) what the measurements actually were, so the key is still secret. Alice and Bob have agreed beforehand that both horizontal and left-diagonal will equal 1, while vertical and right-diagonal will equal 0. This number is their key, which they can now use to encode their message.
Alice sends:
/ \ \ | — \ | / — \
Bob uses:
+ X + + X + X X X +
Bob reads:
— \ | | / — \ / \ —
Bob keeps:
\ | /
Alice and Bob translate into:
1 0 0
If Eve tries to copy the key, she runs into insurmountable problems. As Alice’s light travels to Bob, Eve can’t split the pulses of light, because they’re so dim that Bob and Alice will notice the missing photons. Her only other option is to intercept the photons, read them, and send copies in their place. But quantum uncertainty forbids this. Since Alice doesn’t tell Bob how to read the photons, Eve has no better clue than Bob how to set her detector. She can guess, but if she gets a signal when using a + setting, she doesn’t know if it was actually a diagonal that slipped through. As a result, she doesn’t know what to send Bob. Half her guesses will be wrong.
Let’s say Eve detects a photon that looks vertical with her + setting, but it was actually a right-diagonal. Eve sends a vertical to Bob. Later, when Alice tells Bob which of his settings were correct, the setting he used to read Eve’s photon might be among them: Alice sent a diagonal, and Bob used X. But the photon Eve substituted was a vertical, so Bob had a 50 percent chance of reading it as a left-diagonal, although Alice sent a right-diagonal. Their two keys are likely to be different at that spot: Alice will have a 0, Bob will have a 1. Alice and Bob can compare a few numbers in their key, and if they find a lot of mistakes, they’ll know someone is bugging them. If the numbers agree, then they can throw them out and keep the remaining ones as their key.
Eve may try a different strategy: instead of reading the whole key, she’ll settle for some small part of it. The few discrepancies she introduces may slip by. After all, no detector is perfect. Alice and Bob may think her tapping is just a glitch in the communication line. “Perhaps she only gets ten bits, but those ten bits could make all the difference,” says Brassard.
Bennett and Brassard figured out how Alice and Bob could not only take away any shred of information Eve may cling to but also clean up the natural errors in their key as well—and do it all while Eve is listening in. The process is a little like playing 20 questions. Alice and Bob pick a random group of numbers in the key, say the second, fourth, seventh, and eleventh:
Alice: 0 1 0 0 1 1 1 0 1 0 0 1 0 1
Bob: 0 1 0 1 0 1 1 0 1 0 0 1 0 1
They tell each other whether there is an odd or even number of 1’s in their group. In this case, Alice has even, Bob has odd. That tells them there’s a discrepancy, and they can narrow down the group of digits until they find the flaw. Doing this randomly many times lets Alice and Bob have virtually complete confidence in their key. Yet they never announce what the digits really are.
“From this impure secret they can distill a pure one,” says Bennett. Alice and Bob get back on the public line and agree to look at random groups of numbers again (“Okay, Bob, now look at the fifth, twelfth, and twentieth…”). Once again they each see if the groups have odd or even numbers of 1’s. But this time they use this information to create an entirely new key. If there is an even number of 1’s in a group, they each secretly record a 0; if the number is odd, they write a 1. Bennett, Brassard, and Brassard’s former student Jean-Marc Robert have proved that this method will keep Eve from knowing the value of a single digit of the key, even if it is trillions of trillions of digits long. And once the random key is secured, so are the messages coded by that key, provided that Alice and Bob don’t use any part of the key more than once.
Bennett and Brassard’s innovations inspired Ekert to look into a different aspect of quantum physics. It’s a strange property of particles that Einstein first uncovered over 50 years ago. “Einstein was never very fond of quantum mechanics,” Ekert points out. As the great physicist looked for some flaw in the theory, he came up with a case in which he thought he had pushed the hazy game of chance to an absurd extreme—a case in which particles seemed to be telepathic.
Imagine that some physical process creates two particles that shoot off from each other in opposite directions. They can be electrons created during nuclear decay or photons created when an atom is heated. The laws of conservation say that their characteristics have to be equal and opposite to each other. If one electron spins clockwise, the other must spin counterclockwise. For photons, this means they have to be polarized at 90 degrees to each other. But as Einstein pointed out, quantum physics tells us that these photons don’t have any particular polarization if no one is measuring them. Theoretically, each photon in the pair can tootle along in this fuzzy state for light-years. If someone then uses a + polarization detector on one of them, he forces it to choose between being horizontal or vertical. If it registers as horizontal, then the other photon immediately becomes vertical. It’s as if the two photons communicate faster than the speed of light. This sounded absurd to Einstein, but the mysterious “communication” between such particles has since been proved by experiment to be real.
It occurred to Ekert that this paradox could create a perfectly secret key as effectively as Bennett and Brassard’s coding scheme. With some modifications added later by David Mermin of Cornell, his system works like this: Alice lives in New York and Bob in London, and an optical fiber links them. At the midway point between them, in the Atlantic Ocean, there’s a box containing a hot atom. It fires off two photons at a time, one of which travels through the fiber to Bob, the other to Alice. Say Alice reads her photon with a + detector and gets a vertical reading. Now Bob’s photon is horizontal. If he uses a + detector, too, he’ll get the right answer. But if he uses the X setting, the horizontal photon now has to choose which of the diagonals it will look like. Alice and Bob randomly set their detector on + or X each time a photon arrives. When they have enough readings to make a key, they check with each other to see which signals were a result of their using the same setting. They save only these good readings and use them to make a key. The situation is a little different from Bennett and Brassard’s scheme, since now Alice’s and Bob’s photons are supposed to differ by 90 degrees. In this case, they just agree beforehand which opposing orientations will stand for 1 and which for 0.
Eve is once again in trouble. She may swim down to the bottom of the Atlantic and read the photons traveling to Bob. But if she measures a photon with a + setting, she forces it to choose between being horizontal or vertical—and also forces Alice’s photon to pick the opposite direction. If Alice and Bob both use an X setting, each will be reading a photon that must randomly pick which of the two diagonals it will look like. There’s no guarantee now that Alice and Bob’s readings will be correlated. When they compare a few random parts of their key, they’ll discover that even when they used the same setting, they were sometimes recording different numbers—and just as in Bennett and Brassards scheme, they’ll know that Eve is back to her old habits.
In most ways Ekert’s setup provides the same level of security as Bennett and Brassard’s, but in one respect it’s better. If Alice and Bob use Bennett-Brassard cryptography to get their key, they each have to store the key as regular, non-quantum signals in a computer or in writing until they send their message. In that interval Eve could break into Alice’s computer or her safe and copy the key.
In Ekert’s system, Alice and Bob receive “entangled” photons. As long as they don’t try to observe polarization, they can store these photons without disentangling them. There the photons stay, in their quantum indecision, until the moment Alice and Bob need them. Eve would spoil both sets by trying to observe one of them.
As Brassard points out, “Theoretically this is very nice, but from a technical point of view it is outrageous.” There is no way to store photons intact—the best mirrored trap would lose them in a fraction of a second. Ekert cheerfully admits that Brassard is right. These researchers don’t let today’s practicalities slow them down. They’re confident that technology will someday catch up with their ideas. And when it does, they’re confident the schemes will work. In fact, for a long time Bennett and Brassard were so sure they could actually build a cryptography machine that they didn’t feel the need to make the effort. “If you have an experiment that you’re sure is going to work, there’s no point in building it,” says Bennett. “If you know that the North Pole is there, what’s the point of actually going to it?”
Still, says Brassard, “we kept meeting with mounting skepticism, so we decided just as a daring move to build it.” Bennett and his graduate students handled the hardware, and Brassard’s team figured out the software. The prototype sits next to Bennett in his small office at IBM, looking a bit like an aluminum coffin. (“One of the jobs I suggested to my grad students was to climb inside the box to make sure that it didn’t have any light leaks,” says Bennett.) Under the lid is an unassuming array of crystals, detectors, and a green light-emitting diode, all of which are controlled by Bennett’s PC. The coffin prepares pulses of light at one side, and photons travel across the table to detectors on the other. “Alice, Bob, and Eve live in the computer,” Bennett says, patting the hard drive, “and they’re honor-bound to look only at their own data.” As Bennett and Brassard already knew, the system provides Alice and Bob with a perfect cover. “There was never any doubt that it would work,” says Brassard, “only that our fingers might be too clumsy to build it.”
Bennett and Brassard have issued a challenge to any skeptic to come to their lab and stick any eavesdropping equipment whatever into the light beam. No one has taken it up. And yet they also admit how absurd their device really is. They need several minutes to transmit a few hundred digits a grand total of 12 inches. “It would be silly to use this,” says Brassard. Some of the problems are minor: being theoreticians and not engineers, they used off-the-shelf equipment that isn’t the most efficient. But the real problem is with light itself. It might seem obvious to use optical fibers to carry the signal, but the signal tends to get lost because the bends and twists in a fiber often spoil the polarization of light.
Researchers have been pondering ways that Bennett and Brassard’s machine could be employed as is. As it turns out, it could be used when two parties want to negotiate something without giving away any information about themselves in the process. Governments and businesses, for instance, often want to come to joint decisions without giving away their secrets. Negotiating arms treaties and forming partnerships are just a couple of examples. With the Bennett-Brassard box to keep the secrecy, the two negotiating parties would merely have to agree on the question to be asked. Then, in response, each side’s data would be put into a quantum code and sent back and forth in small parts, with each bit of information subjected to a number of manipulations. By a process of elimination that’s far more involved than the method Alice and Bob use to distill a cryptographic key, they would ultimately discover the answer to their mutual question without anything else left over in their computer that could help them discover anything more. The short distance wouldn’t matter, since two people could go into a room, plug their laptops into the machine, and start making deals.
The hope for long-distance transmission isn’t dead, though. This year Ekert used quantum cryptography over a distance of 200 yards. To get it to work on a larger scale, he had to give up using polarization, although the basic idea is unchanged. Instead, he joined with two researchers, John Rarity and Paul Tapster of the Royal Signals and Radar Establishment, who are experts on how photons can interfere with themselves.
In a new scheme, a device known as an interferometer replaces the polarization detectors. Photons entering the interferometer strike a beam splitter, which gives them a choice of taking two paths: there’s a 50 percent chance that a photon will go either way. A short path leads to one detector, and a longer path to another detector. There’s one catch: mirrors deflect photons taking the longer path so that the two paths cross before they reach the detectors. There, at the intersection, another beam splitter gives photons a second choice: they can either keep going in the same direction or switch over to the other path.
Photons, of course, pull a nonsensical trick inside the interferometer: they take all possible paths at the same time. Nothing is definite until the photon reaches a detector, at which point it collapses, or recombines, on one path and strikes one of the two detectors at random. Before then, researchers can only assign each path a probability. But they can manipulate those odds, just as they can manipulate, with polarization equipment, the probability that a photon will vibrate at different angles. In this case, researchers adjust the length of the longer path in the interferometer by minuscule amounts, so that when the photon splits and goes two ways at once, the “part” of the photon going the long way arrives at the intersection out of sync with the part that went the short way. In the jargon of physicists, the photon’s paths are phase delayed—“phase” being how far it’s gone through its cycle of vibration.
Using phase delays, researchers can make sure that a photon can’t choose to hit one of the two detectors. If the phase delay is exactly half a vibration, the photon’s “parts” can’t recombine at one of the detectors: the “part” of the photon that took the long path, say, will be vibrating up while the part that took the short path will be vibrating down. The probability of vibration cancels—which means there’s zero chance of the photon’s going down that path and hitting that detector. The photon can’t choose that option any more than a vertically polarized photon can pass through a filter turned in the horizontal direction.
Ekert, Rarity, and Tapster gave Alice and Bob identical interferometers. Each interferometer had two detectors, labeled “1” and “0,” and each interferometer also had a phase-delay switch that Alice and Bob could throw on or off. The researchers then replaced the hot atom that sent out photons at the bottom of the Atlantic with a laser. All the photons emitted by a laser have identical frequencies, but in this case a photon hits a crystal that sends out two new photons in its place. These child photons travel through the optical fiber to Alice and Bob.
Unlike their parent, the child photons don’t have a fixed frequency. Like the polarized photons Ekert used in his first setup, the children’s frequencies are entangled in a quantum fuzz. All you can say is that their frequencies add up to the frequency of the parent photon.
When each pair of photons reach Alice and Bob, they can throw the phase-delay switch on or off. The switch changes the length of one path in the interferometer. Since this device is making a measurement, it forces the photon to select one frequency. And as in Einstein’s original paradox, forcing one photon out of its quantum limbo instantly forces the other to take on a correlated frequency. Ekert, Rarity, and Tapster have arranged the interferometers so that if there’s no difference in the phase setting Alice and Bob are using (both have their switches on or both have them off), they will both register a 1 or a 0, as long as they are both reading correlated photons. If the settings of their phase switches don’t match, they can get disagreeing numbers. As in the earlier schemes, all Alice and Bob have to do is let the laser feed them photons, switch their phase delay randomly on and off, and then confer to throw out the readings that were the product of different phase-delay settings.
Eve is in the usual bind. Sitting at the bottom of the ocean with her interferometer and trying to intercept Alice’s photons, she has no idea whether to turn her phase-delay switch on or off. Say she turns it on and gets a 1, then sends a substitute photon to Alice. By making her measurement, Eve has forced the photon traveling toward Bob to register as a 1, as long as he sets his phase switch to the “on” position, the same way Eve did. But say Alice and Bob both turn their switches off. When they confer, they’ll know they used the same setting, but Alice may get a 1 while Bob gets a 0. Comparing numbers will tell them that Eve is still trying to make their life difficult.
***
Ekert, Rarity, and Tapster ran their phase-delay system successfully for the first time in March. For simplicity’s sake, they kept the whole affair inside their lab, with the 200 yards of optical fiber wrapped up in rolls, and with Alice, Bob, and Eve existing only as computer software. The reason their device works over a longer distance than Bennett and Brassard’s is that the effects of interference survive over much greater distances than those of polarization. But they didn’t try to use the code beyond 200 yards because the technology doesn’t exist yet to make it possible. The farther a photon travels down a fiber, the more it spreads out and becomes harder to detect. The optical fibers used in phone lines have amplifiers to boost signals as they go along. But boosting a signal that’s supposed to be part of a secret key would have the same effect as Eve trying to measure it—the code would be spoiled.
Ekert wasn’t discouraged before, and he won’t get discouraged now. With some improvements, his device may work over a distance of 50 miles, and if optics keep improving it might be possible to get up to hundreds of miles. “One should never underestimate technology,” says Ekert. “When Bennett and Brassard started thinking about this, it was science fiction, but here we are. People are now coming from industry. British Telecom is looking at it, and we have to evaluate whether in, say, the next five years it would be feasible to have quantum cryptography. Whether the cost involved will be worth it is another question.”
Copyright 1992 Discover Magazine. Reprinted with permission.