The NSA Probably Really, Really Wants a Quantum Computer

How long until the security agency gets one?

A forest of optical elements used in Delft University's quantum information experiment.  (Hanson lab@TUDelft)

This is huge: Physicists have figured out how to reliably transmit quantum information, a major first step toward the kind of quantum communications and quantum computing that would dramatically enhance our ability to secure networks, crack codes, and search databases.

These potential applications of quantum computing—and I'll get to what exactly that means in just a minute—help explain why government agencies ought to be interested in such technologies. And the latest news in this realm is significant. It comes from a team of nanoscience researchers at the Delft University of Technology in the Netherlands, where physicists successfully transmitted encoded quantum information from one computer chip to another, across a distance of about 10 feet.

This stuff is complex, but it gets easier from here: In classical computing, basic units of information are assigned values of either 1 or 0. But in quantum computing, the basic units of information, qubits, don't have set values. This distinction makes qubits more dynamic—they can do more stuff!—but it also makes qubits harder to transport reliably. And once scientists figure out how to do so, the future of quantum computing and quantum communications really comes into focus.

* * *

"In the whole field of quantum information, there are two different directions: One is quantum computing and the other is quantum communications," said physicist Steve Rolston, who is co-director of the Joint Quantum Institute at the University of Maryland. "If we start with quantum computing, this is the holy grail... The quantum computer is not just a faster regular computer. So far we've only identified a few problems that it's really good at: crunching numbers—which sounds like an arbitrary math problem, but that's how cryptography works—and searching databases."

In other words, a quantum computer is only exceptionally good at a few things—but those few things happen to have major implications for the future of intelligence, surveillance, and security. "That's why the NSA in particular is so interested in quantum computers and would like to have one, and probably would not tell anyone if they did," Rolston told me.

There's some debate over how advanced quantum computing already is. The Canadian startup D-Wave claims to have built a $15 million quantum computer, though many scientists have expressed skepticism about the technology. It still piqued the interest of Google, NASA and the Universities Space Research Association, which teamed up last year to install a D-Wave device in a NASA lab so that researchers can come by and test how it might advance machine learning. Earlier this year D-Wave's CEO was in Washington, D.C., shopping around his company's wares.

Quantum computing's code-cracking potential could undermine encryption protocols that now secure information across any number of networks—the kinds used by financial institutions, government agencies, etc. Quantum computing could also process enormous databases like no computer today.

But quantum communications represents "sort of the flip side," Rolston says. "Because of the laws of quantum mechanics, you can develop ways to communicate with people that are provably secure. It doesn't mean that people can't listen in, but it means you would be able to tell." Here's why: There's this rule in quantum mechanics called the No-Cloning Theorem. Rolston explained it to me like this: "Basically, what that says, is you cannot make an exact copy of an unknown quantum state."

So, think about this theorem in the context of copying—or recording—a tapped phone line. The way we transmit information now, you might bug a phone line you want to secretly record. The people using that phone line won't be able to tell you're listening in, and you end up with a recording of data was transmitted across that line.

But if that data is transmitted as quantum information, the No-Cloning Theorem won't allow a clean recording. It would come out garbled because of the ever-changing value systems assigned to qubits (compared with the reliable 1s and 0s of regular bits).

"As soon as you try to tap the line, you can't create the perfect copy," Rolston told me. "You can't make an exact recording without disturbing the information. If I make a measurement on any quantum system, I disturb it it in a detectable way. When we measure things, we mess things up. That's kind of the underlying concept."

This principle is what makes quantum information so thorny. To understand it or control it, you have to disrupt it—but disrupt it too much and quantum information will lose its mojo. You might start with a quantum computer, but if you poke around too much—in an effort to understand what it's doing, for example—it will effectively cease to be one. "It turns into a very expensive and useless classical device," Rolston said. "Like a regular computer, but not a very good one." And as scientists try to make bigger and bigger quantum systems, the likelihood of disruptions that sap the quantum magic is bigger, too.

Okay, it's not really magic. Rolston can explain: "Quantum mechanics describes the microscopic world of atoms and molecules—the stuff that we are made from. But because of interactions with the environment, the quantum nature quickly gets lost (very very rapidly for large systems)... and we use 'classical' physics to describe most everything. We could continue to use quantum mechanics, but it would be needlessly and impossibly complex, so we use classical physics concepts instead."

In other words, interfere too much with a quantum system you're trying to control, and we won't be able to use it as a quantum system anymore. "When you have large quantum systems, they end up acting classically," Rolston told me. "So the real challenge is building a big enough quantum system so it can do the calculations you want, but so it remains quantum. It's this kind of conflicting goal."

Quantum physicists have been obsessed with working toward this goal for 20 years, ever since MIT mathematician Peter Shor came up with a revolutionary quantum proof now called Shor's Algorithm in 1994. "Twenty years," Rolston says. "That just shows how hard it is."

And the difficulty academics have had wrangling quantum information suggests that the NSA probably hasn't mastered it yet either, he says.

"Would we know if they have? Maybe not," he said. "But given where technology is, the state of the art in laboratories around the world, I don't think it's possible that they're much far ahead of academic science. There are still a lot of problems we're working on."