Perhaps the (few) naysayers have been wrong: Google recently announced test results for a D-Wave “quantum computer” that far surpass the capabilities of traditional computers. So, why the scare quotes on quantum computer? And why not celebrate this apparent revolutionary device, which garnered headlines such as “NASA, Google unveil a quantum computing leap” and “Google’s quantum computer can blitz a normal PC”? Here are a few reasons to reserve judgment for now.
D-Wave: No Stranger to Controversy
D-Wave has claimed for years that it manufactures quantum computers. In August of this year, for instance, the company issued a press release titled “Announcing the D-Wave 2X Quantum Computer.” Granted, however, the company doesn’t claim that it has achieved a “generic” quantum computer—that is, what most people think of when they read stories about this potential technology. D-Wave claims that its products implement so-called quantum annealing. I’m not going to get into the technical details, as they really aren’t relevant.
What is relevant, however, is that testing of D-Wave products has been a back-and-forth battle, with claims of superior performance relative to classical computers later giving way to claims of no real advantage for the “quantum” approach (I leave it to the reader as to whether the scare quotes are warranted). For instance, Wired reported in 2014, “New research suggests the commercial quantum computer sold by Canadian company D-Wave Systems isn’t faster than the PC on your desk.”
The problem for the layman (i.e., pretty much anyone who isn’t well versed in both quantum mechanics and computer/algorithm theory) is that interpreting the technobabble—particularly when it’s (justifiably) simplified by journalists—is nearly impossible. Readers are thus tossed back and forth by the competing claims with nary a sound footing on which to make their own judgments. In the world of classical computing, by contrast, the controversies are much less fundamental. Sure, there might be some debate about whether x86 server chips are better than ARM server chips, but the data presented in a typical pro or con argument is fairly familiar. Even despite nuances like performance versus power consumption, an average user can typically make some decent evaluation of the competing claims. It’s the “fuzziness” (in more ways than one) of quantum-computing claims that make this topic so difficult to navigate—and so prone to wild predictions of a coming quantum revolution.
Getting a Few Things Straight
A supersonic fighter jet is faster than an economy car. So what, right? An economy car sells for somewhere in the neighborhood of $10,000. A fighter jet costs a lot more—on the order of $100 million, if you consider the case of the boondoggle F-35. That’s a 10,000x price premium for the speed, to say nothing of the cost of fuel, maintenance, storage and so on. But what’s this got to do with a quantum computer?
Well, consider that according to ZDNet, “Google says its D-Wave 2X quantum computer is more than 100 million times faster than a conventional PC.” Assume it’s true. What’s the price premium on a D-Wave 2X versus your run-of-the-mill desktop (say, an HP Pavilion)? Although pricing isn’t readily available, one has to figure it’s at least a 1,000x difference, if not more. In addition, an HP Pavilion operates just fine at a balmy 80°F. The D-Wave 2X? Try “an operating temperature below 15 millikelvin, near absolute zero and 180 times colder than interstellar space,” according to D-Wave. The operating costs for the supporting infrastructure alone in this case would blow out of the water the power costs for running a desktop PC at full throttle.
Data center operators, for one, already recognize the important distinction between absolute performance and performance per watt per dollar. This difference is driving the ARM-versus-x86 battle for the server, with ARM attempting to sneak in on the basis not of higher absolute performance, but more-efficient performance. Whether it achieves that task is irrelevant; the point is that energy and economic efficiency are the critical metrics. So, even if a D-Wave quantum computer wipes out a desktop PC on some computing problem, it’s dead in the commercial water if its costs and complexity are so high that the same amount of money can buy and run a nearly equivalent (or better) classical computing system.
Moreover, the ZDNet article drops a number of qualifiers about the Google results. For instance, according to Google’s Quantum AI Lab head Hartmut Neven, “While these results are intriguing and very encouraging, there is more work ahead to turn quantum-enhanced optimization into a practical technology. The design of next generation annealers must facilitate the embedding of problems of practical relevance.” So, in English, a quantum computer that can beat a PC at calculating how much fun is in a barrel of monkeys may be cute, but its value is lacking. Neven added, “We are optimistic that the significant runtime gains we have found will carry over to commercially relevant problems as they occur in tasks relevant to machine intelligence.” No doubt, some researchers continue to be optimistic about the commercial relevance of flying cars, cold fusion and virtual reality, too.
Wait for the Contradicting Research
Scott Aaronson, an MIT professor and theoretical computer scientist, notes that the paper citing these seemingly astounding results for the D-Wave 2X include caveats that make the claims all but inconsequential for skeptics. For instance, the paper says in its summary section, “More work is needed to turn quantum enhanced optimization into a practical technology.” The average Joe might well ask, “So what good is this contraption?” Some good may come out of it, but interesting abstract properties without any real application are insufficient. The plight of graphene is a similar example in the materials domain.
Aaronson added that “I still have no idea when and if we’ll have a practical, universal, fault-tolerant [quantum computer], capable of factoring 10,000-digit numbers and so on.” He doesn’t deny the value of the D-Wave 2X as far as it goes, nor does he claim that quantum computing is impossible or will never be practical. He does, however, try to tone down the hype that surrounds these results. Future testing may bear out the Google results, contradict them or simply show that they are too narrow to be of value beyond a particular domain. For D-Wave, however, the goal may have been accomplished: a big headline that draws attention to its products, regardless of whether they’re “true” quantum computers.
Revolutions tend to be predictable in hindsight, but those that are advertised well in advance are prone to fizzling. Quantum computing has been on the technology prognostication list for years, and D-Wave has tried to steal the show. The test results for a D-Wave 2X device run by Google seems to indicate that quantum computing is more than just hype, but the caveats are too big to ignore. Moreover, the ultimate question is whether a quantum computer—whether from D-Wave or a company that has yet to emerge—can beat classical computers in a practical sense. That is, does the quantum solution to a problem cost more than the equivalent classical solution? Is a computer that’s lightning fast yet extremely difficult to program worth the hassle compared with a slower yet cheaper alternative that anyone can use? As with the case of Moore’s Law, the situation involves more than just what’s possible—it’s more about what’s economical.