Picture, if you can, the following scene. It’s the year 2040. You wake up in the morning, and walk across your bedroom to your computer to check your email and some news websites. Your computer, your mail reader, and your web browser have some new bells and whistles, but all of them would be recognizable to a visitor from 2014: on casual inspection, not that much has changed. But one thing has changed: if, while browsing the web, you suddenly feel the urge to calculate the ground state energy of a complicated biomolecule, or to know the prime factors of a 5000-digit positive integer—and who among us don’t feel those urges, from time to time?—there are now online services that, for a fee, will use a quantum computer to give you the answer much faster than you could’ve obtained it classically. Scientists, you’re vaguely aware, are using the new quantum simulation capability to help them design drugs and high-efficiency solar cells, and to explore the properties of high-temperature superconductors. Does any of this affect your life? Sure, maybe it does—and if not, it might affect your children’s lives, or your grandchildren’s. At any rate, it’s certainly cool to know about.
Privacy and security are different as well in this brave new world. When you connect to a secure website—let’s say, to upload sensitive financial data—there’s still a padlock icon in your web browser; indeed, the user experience is pretty much the same as it was in 2014. But, you’ve heard, the previous mechanism that encrypted your data was broken by quantum computers, with their ability to factor large numbers. In place of the old mechanism is a suite of new ones: some of them classical, based on mathematical problems more complicated than factoring integers (like finding short vectors in lattices), and others that, ironically, use quantum mechanics itself to fix what quantum computers had broken. The technology of quantum key distribution, over half a century old, finally has a decent user base, although it only works over distances of a few hundred kilometers (quantum key distribution to and from satellites is expected to start soon). Truth be told, you never understood the old, broken security mechanisms, you don’t understand the new, quantum-proof mechanisms either, and in fact you continue to use your birthday as your email password, since you don’t really have secrets that anyone else would go to any lengths to steal. But you’re vaguely aware that things have changed in the world of encryption, and that the reason why they changed was quantum computing.
But wait, there’s more. Does your job require you to solve “NP-hard” combinatorial optimization problems—the infamous family exemplified by the Traveling Salesman Problem, of finding the shortest route that visits a collection of cities? For example, do you need to verify complicated microchips or pieces of code, or optimize a factory’s production schedule, or find hidden patterns in stock-market data?
Unfortunately, even though quantum computers are now available, it’s still disputed whether can they help you with any of this, and if so by how much. The debate centers around the so-called “quantum adiabatic algorithm,” which was first proposed by a team at MIT in 1999. Unlike the quantum algorithms for factoring large numbers and for simulating quantum physics, the adiabatic algorithm is heuristic—meaning that there’s no mathematical analysis guaranteeing it will run fast, and indeed there are cases where it certainly doesn’t. Often the best one can do is simply to try the adiabatic algorithm out, in head-to-head comparisons with the fastest available classical algorithms, and see what happens. So that’s exactly what people have done. In 2037, you read, a research team published a paper in Science finding a robust advantage for the quantum adiabatic algorithm over any extant classical algorithm, on real-world optimization problems involving tens of thousands of variables. Alas, in 2038, a different research team published a riposte in Nature (yes, Science and Nature are still around) finding that, for the particular problem benchmarks studied by the 2037 team, it’s possible to devise a classical algorithm that runs just as quickly as the quantum adiabatic algorithm. Then in 2039, the original team responded, proposing a new, carefully-crafted set of benchmarks on which the adiabatic algorithm outperforms the 2038 team’s algorithm. While the scientists duke it out, most people who need to solve optimization problems in real life simply continue to use classical methods, although a minority has switched to the quantum adiabatic algorithm.
Ironically, the biggest practical impact of the adiabatic algorithm may have been this: the effort to match its performance using classical computers has stimulated huge improvements even in classical optimization methods. More broadly, as they’ve done since the early 2000’s, the ideas of quantum computing have continued to generate dozens of “spin-off” insights for classical computation, as well as for ‘traditional’ areas of physics, like condensed matter and even quantum gravity. Although, not many people outside the research community know or care about these spin-offs.
Now, besides the applications to quantum simulation and cryptography, and the nebulous application to optimization and machine learning, the world of 2040 has one important additional application of quantum computers: one so astonishing that, back in 2014, no one could possibly have imagined it. Alas, because of that very fact, I can’t tell you what it is.
Reviewing the fruits of the quantum information revolution—a quantum leap in the simulation of quantum physics and chemistry; dramatic advances in both code-making and code-breaking (which, however, partly “cancelled each other out,” leaving us back where we were before); a small, debatable advance in solving NP-hard optimization problems; spin-offs for classical computer science and for physics; and of course, the application that no one expected—some commentators complain that quantum information turned out to be a massive disappointment. “We were promised the dawn of a new era!” they cry. “We were promised that quantum computers, once they finally arrived, would transform the lives of ordinary people just as the transistor, telephone, and internal combustion engine did for earlier generations. I mean, look at the popular magazine articles from a few decades ago. They said that quantum computers would harness the mind-boggling power of multiple universes, by trying every possible solution in parallel. They said the practical applications were limitless. They said that even if Moore’s Law was reaching an end, quantum computers would easily pick up the slack, giving us everything we would’ve had if Moore’s Law had continued unabated, and more. And now that billions of government dollars have been spent, and scalable quantum computers are at last a reality … now you tell us that these gizmos are mostly good for ‘quantum simulation,’ and a few other specialized tasks that excite you nerds, and might someday lead to important discoveries, but have little immediate bearing on the average person’s life? And some of you even knew this all along? Phooey! You quantum computing people should be hauled into court, for perpetrating a fraud on the public!”
In response, the quantum computing researchers protest that, from the very beginning, they’d tried to explain the real scope and limits of quantum information technologies to anyone who would listen, but their voices were drowned out by a drumbeat of hype. As the quantum computing researchers see it, they were maltreated twice: first when many journalists, investors, and funding agencies ignored their sober scientific assessments in favor of snake-oil promises; then a second time when, in a cruel (if predictable) twist, they were the ones who got blamed when the sober assessments turned out to be correct.
For all that, the researchers insist that the construction of general-purpose quantum computers really was an inspirational triumph, one of the greatest triumphs ever of the human intellect. For one thing, they say, the new ability to simulate Nature at the atomic scale, efficiently and programmably, is nothing to sneeze at—indeed, this ability could lead in a generation or two to a revolution in medicines and materials. “But,” implore the researchers, “just for a moment, try to set aside the applications entirely. Instead, marvel that quantum computing turned out to be possible at all! Marvel that we now know how to harness the true computational capacity of the physical universe. Marvel that the exponential vastness of Hilbert space is no longer just a creature of the physics textbooks; it’s been given agency, brought to earth like the Promethean fire. Marvel that all the confident predictions that quantum computers were a fantasy, impossible even in theory—that deep down, the universe is ‘really classical,’ or has ‘correlated noise’ that diabolically conspires to kill quantum computation—marvel that all those ideas are now strewn like corpses across the intellectual landscape. I mean, have you read the defenses by the erstwhile quantum computing skeptics: their lame protestations of how, when they’d said ‘impossible,’ they didn’t really mean ‘impossible’? Wasn’t the satisfaction of vindicating quantum mechanics itself worth the entire cost of building a quantum computer—nay, ten times the cost?”
As you sit in your bedroom checking the news, you reflect that the point might go even further than these researchers say. In 2040, so many of the problems we faced in 2014 have gotten worse, with droughts, floods, and hurricanes causing misery all over the world. And outside of pure mathematics (which is doing fine), believers in the so-called “End of Science” can find plenty to crow about: there’s still no experimental evidence for supersymmetry, no Moon colonies, no evidence for life beyond Earth, no cure for cancer (or even for obesity), no fusion power, no radical life extension, no human-level AI, no jetpacks. And yet, you remind yourself, at least the world now has scalable quantum computers. That’s something—a little ray of hope in the darkness. And who knows? Maybe those quantum computers will even be used to design a new generation of solar cells, and thereby help create a sustainable future, if humanity ever gets its act together.
As magical as it all sounds, this is the wondrous science-fiction future that my sixteen years of research in quantum computing and information lead me to believe is possible. Assuming, of course, that we actually do build scalable quantum computers.
- Back around 1950, even the people working on the first electronic computers couldn’t imagine most of the applications of computers that fill our lives today. Do you think that the author is suffering from a similar myopia and shortsightedness with regard to quantum computing? (Or is the situation different today, precisely because we now have such powerful and pervasive electronic computers, and a quantum computer will only be useful for those tasks where it can outperform them?)
- Do you think there’s something special about the subject of quantum computing that makes it hard for journalists to get right? Or is it just the usual difficulties of reporting on any new technology?
- What have you read or heard about the possibilities of quantum computing?
My essay, “How Might Quantum Information Transform Our Future?” attracted some insightful comments, both on the BQO site and on my own blog (and this summary will deal with both of them). I’m not sure that there was any single theme, but here were some highlights:
Commenter hljohnston complained that my article put too much blame on the media, funders, and investors for “snake-oil promises” about quantum computing, and not enough on the physicists themselves. In response, I accepted the criticism, and speculated on the possibly-unfair reasons why I might give my colleagues the benefit of the doubt—e.g., because I know that they actually understand the issues and were just trying to simplify things for a popular audience, or because I recognize that a *tiny bit* of optimistic self-delusion helps them to do productive science.
In answer to my question about why quantum computing is so hard for journalists to get right, commenter Joel David Klassen wrote, extremely perceptively I think, that “[r]eporting and popularization relies on narrative, and it is very difficult to have a narrative without local realism … Essentially, quantum computing has inherited the PR baggage that quantum mechanics has had since its inception, the complete departure from classicality.”
Gil Kalai, a colleague and longtime commenter on my blog, wrote, “[t]his is a lovely article and I find it entertaining and enjoyable in spite of me being one of the ‘skeptics’ who think that superior computation through quantum computers is not possible.” Gil’s position has long been that “diabolical, correlated noise” will conspire to kill quantum computation. In response, I’ve always said that I hope Gil and like-minded skeptics are right, since scientifically, that would be far more surprising and interesting than a “mere” success at building scalable quantum computers. But I don’t expect them to be right.
Commenters Jay and Vlad both took me to task for understating the practical importance of quantum simulation, while Juan Miguel complained that I didn’t mention emerging technologies at the border between quantum information and “conventional” physics, like quantum metrology. Commenter Sid K argued that, considering the history of classical algorithms, we’ll probably see many more advances in quantum algorithms, once we get actual quantum computers to test them out on. In the opposite direction, commenter Hanno thought I was overoptimistic about the practical potential of quantum key distribution, arguing that conventional public-key cryptosystems like NTRU will likely outperform them in practice.
Rahul, a longtime commenter on my blog, wrote “I read the article twice & am totally lost in the depths of its sarcasm,” a comment that I relished greatly.
On the other hand, commenter Denver Bob wrote, “I’ve begun listening to your predictions about quantum computing and assuming they are anticorrelated with what will actually happen — this strategy seems to be working.” However, when challenged to name a prediction of mine about quantum computing that turned out to be wrong, Denver Bob was silent.
New Big Questions:
- When a new technology gets overhyped in the press, so that readers come away without understanding its limitations, how much of the blame usually belongs to journalists, and how much to the technology developers themselves?
- How much, really, would the ability to efficiently simulate quantum systems transform our future?