Why Google’s Quantum Victory Is a Huge Deal—and a Letdown

They finally did it. After years—no, decades—of declaring their hopes and dreams with hardly any practical results, researchers in the quantum computing community have delivered on a promise. Or have they?

Last week, news leaked that researchers at Google and other institutions had solved a problem on a quantum computer 1 billion times faster than a classical computer. Google did not respond to a request for comment, but according to a draft manuscript describing the experiment, they have realized “quantum supremacy,” an achievement that “heralds the advent of a much-anticipated computing paradigm.”

The reactions from the rest of the quantum community, however, have been downright contradictory. In an email, physicist John Preskill of Caltech calls the work a “truly impressive achievement in experimental physics.” Cautioning that the leaked manuscript is just a draft, mathematician Ashley Montanaro of the University of Bristol says this is a “genuinely exciting moment,” and also that the work is “not of practical relevance.” Dario Gil, the director of research at IBM, disputes the notion of quantum supremacy itself, calling the term “misleading” in a statement to WIRED.

So which one is true? Did the Google team just clear a steep technological threshold, or has it performed a largely useless experiment dressed up in clever branding? The truth lies somewhere in between, and captures the competing tensions at play in the quantum computing world.

On one hand, these researchers managed to execute an extremely complex experiment built on painstaking mathematical proofs and years of hardware development—an undeniable achievement. Yet the experiment brings them no closer to the money-making applications that the quantum community has promised, where the computer’s unique number-crunching capabilities will reveal new molecules for better batteries, drugs, and more, at speeds that would put normal computers to shame. It’s both a major victory and a bit of a letdown.

LEARN MORE

The WIRED Guide to Quantum Computing

To understand this, let’s unpack what the Google paper actually describes. The researchers ran the experiment on a quantum computing chip named Sycamore, which contains 54 tiny objects called qubits that they can program to represent the digit 0, 1, or a weighted combination of both, called a superposition. The researchers apply voltage pulses and microwaves in various sequences to the qubits, changing their values according to rules set by quantum mechanics. String several such pulse sequences together, and you’ve written an algorithm for a quantum computer.

In the quantum supremacy experiment, Google’s scientists designed some pulse sequences that essentially turn their computer into a random number generator. They then made the quantum computer spit out millions of numbers. Though the numbers look random, they are meant to still fall into a pattern prescribed by Google’s algorithm. So they next checked to see if the numbers obeyed that distribution. They did.

On its own, this task is basically an excuse for a quantum computer and a supercomputer to race each other. Its practical implications are minimal. But this is the first race a quantum computer seems to have won. The supercomputer could not verify that the numbers obeyed the distribution in time.

The scientific effort behind this experiment began more than a decade ago, with researchers outside Google. In 2004, Barbara Terhal and David DiVincenzo, then at IBM, developed the initial mathematical evidence that qubits might compute this specific task faster than classical bits. It wasn’t clear anyone would ever try to perform the task; capable quantum computers didn’t even exist yet.

“I thought it was a neat idea, but a largely theoretical idea,” says Terhal, now working at Delft University of Technology in the Netherlands. “We submitted a first draft to a computer science conference, and it was rejected. It wasn’t like ‘Oh, this is a good idea!’ Not at all.”

University researchers in the US and Canada developed the first prototypes of the qubits Google used, built of tiny superconducting circuits, as early as 2007. Other researchers, including John Martinis, who now leads Google’s research team, improved the design and figured out how to connect them and make them compute. At long last this collective effort has produced the analog of a symphony in which the musicians had to write all the music, invent all the instruments, and then sit down to play.

The paper represents a majestic scientific production. But at this point, with massive corporations and startups alike developing these elusive machines, the focus has pivoted away from scientific achievement toward economics. For several years, researchers have hinted at the nebulous goal of quantum supremacy as a harbinger of when the financial spigots will fly open. First coined by Preskill in 2011, the term was meant to describe an era in which the devices became faster than classical computers in general. So did Google achieve quantum supremacy?

“I don’t know, who gets to decide what ‘supreme’ is?” asks Terhal.

She thinks no single experiment can demonstrate quantum supremacy; it will need to come from a series of successes. The Google experiment is more of “a stepping stone,” she says.

Some researchers prefer not to focus on supremacy at all. IBM, for example, describes its computer’s performance using more grounded, perhaps even old-school metrics, based in part on accuracy tests. This gave them “a road map to show improvement in systems over time,” says Jay Gambetta, a quantum computing researcher at IBM.

Another problem with supremacy is that classical algorithms can sometimes catch up, says physicist Robin Blume-Kohout of Sandia National Laboratory. Google may have bested supercomputers for now, but eventually someone could find a clever way for a traditional computer to win a rematch.

Some researchers just dislike the term itself. “There’s been a fair amount of discussion in the community whether we should be using ‘supremacy’ because of its nasty connotations,” says Blume-Kohout. One alternative is “quantum advantage.”

There’s also the possibility that Google’s claim won’t hold up. The scientists wrote in their paper that the quantum computer could complete the task in 200 seconds, versus 10,000 years for a supercomputer. Clearly, they did not run a supercomputer for 10,000 years to verify their claim. Instead, the researchers ran simplified versions of the task on both the quantum computer and the supercomputer and compared the results. When the quantum computer succeeded at the simpler tasks, they trusted that the device also executed the more complicated task correctly. While Terhal personally finds the method convincing, she says some experts consider it “a little slippery.” To convince the skeptics, she thinks Google should offer independent researchers access to their device to to run the experiment themselves.

What’s clear now is that quantum supremacy is no longer a useful term. Researchers need to develop and agree on more descriptive ways to talk about their machines, says Blume-Kohout. “Quantum computers are complicated,” he adds. “You can’t boil them down to just one number.”

The true meaning of Google’s experiment is something simpler. They made hardware that works well enough to execute, arguably, the most complicated task to date. It gave the money-losing field a legitimacy boost. “If you’re a skeptic, this makes it hard for you to maintain that position,” says Montanaro. But in the larger scheme to build a profitable quantum computer, it remains unclear whether they’ve gotten past square one.