It’s exciting, but not for the reasons we've been told about
Maybe, like us, you’ve heard about quantum computing and you think it sounds interesting but you’re still not sure how it works or what it means. Quantum-anything is hard right?
Turns out this isn’t true. We spent the last few weeks digging around in some of the more obscure corners of the internet, and discovered that the problem isn’t conceptual. The problem is that perhaps more any other subject in tech journalism and popular science, most of what’s written about quantum computing turns out to be pure, unadulterated bullshit.
Here’s the standard explanation:
In regular computing, we rely on lots of little switches called bits. A bit has a binary value, either 0 or 1, and when you combine lots of bits together you can store data and execute instructions. This is the stuff that makes your mobile phone go beep. In quantum computing, the switches are different. Instead of representing just a 1 or 0, as conventional processors do; they represent multiple values simultaneously. Here for example, is an explanation from the Prime Minister of Canada…
Very simply, normal computers work, either there’s power going through a wire or not — a one, or a zero. They’re binary systems. What quantum states allow for is much more complex information to be encoded into a single bit…a quantum state can be much more complex than that, because as we know, things can be both particle and wave at the same time.
This ambiguity, the ability of qubits to both ‘be’ and ‘not be’ is based on a quantum effect called superposition (sort of like the stuff you’ve heard about dead cats). There’s also a second quantum effect called entanglement, which allows you to link all the qubits together. When you put them all in state of superposition and entangle them, the spin of one affects the spin of all the others. Even if placed at opposite ends of the universe, a change in one entangled qubit would mean the others would dance in instantaneous perfect unison. Einstein called this “spooky action at a distance.”
When you arrange qubits this way, you can perform many calculations at the same time. A regular computer tries to solve a problem the same way you might try to escape a maze, trying every possible corridor, turning back at dead ends, until you eventually find the way out. But a quantum computer can try all the paths at once — in effect, finding the shortcut. The reason this is exciting is because the possibilities get really crazy, very quickly. If you linked a mere 1,000 qubits in a superposition that would allow you to represent every number from 1 to 10^300. The known universe has around 10^78 to 10^82 atoms in it.
At this point, our minds all start exploding.
It’s also where journalists start getting lazy. Cue breathless predictions about how quantum computers might give us whizzy new phones, improve weather forecasting, end traffic jams, allow us to simulate consciousness and solve world peace. Unfortunately 99% of this stuff is hype and it drives real world quantum computing engineers nuts.
The problem comes when you try to measure what’s going on with the qubits — when you open the box with the cat in it. Imagine a street hustler playing the cups and ball game, except instead of three cups, she has a 1000, and instead of a ball, each cup contains either a peanut or a cashew. When the cups are all face down there could be any combination inside them. They’re in a state of superposition. However, as soon as you lift one cup to reveal what’s underneath, it’s like banging your fist on the table; all the other cups fall over too. Sure, you’ve got a huge list of possible combinations but you can only access the list by making a measurement, which is a destructive event that produces just a single random outcome. In which case, you may as well use a normal computer or even just flip a coin.
This is the part you don’t hear about in the popular science articles. The secret is that quantum computing doesn’t actually rely on a qubit being both 1 and 0 at the same time. Instead it’s based on numbers called amplitudes which run on a spectrum, and can be positive, negative or even complex numbers. The goal is to choreograph things so that for each wrong answer, some of the paths leading there have positive amplitudes while others have negative amplitudes, cancelling each other out, while the paths leading to the right answer reinforce. Think of it like stirring up lots of patterns of waves, trying to work out the peaks and troughs, and channelling the flow toward the correct answer. And since we generally don’t know what the answer is in advance, it’s not just a simple matter of laying out a single channel but trying to stir up a set of waves in the open ocean that somehow all come together to make one big splash.
To make matters worse, it turns out that the qubit is the ultimate diva. It demands perfect isolation. The slightest vibration from a nearby atom can cause a qubit to throw a quantum tantrum, and lose its superposition. The difficulty is in maintaining delicate states of superposition and entanglement long enough to run a calculation — the so-called coherence time.
In other words, quantum computing is a fundamentally new and different way of doing computing, and a really difficult engineering challenge. This means that a quantum computer is not going to make your smartphone run faster or lead to the next generation of voice assistants. It’s not going to be better at playing Go, scheduling airline flights or proving theorems. In most cases, doing these things with quantum computing is more hassle than it’s worth.
So why do it then? Why, despite the difficulties, has the race to build the first practical quantum computer become one of the grand scientific challenges of our time, involving thousands of physicists and engineers at dozens of research institutes scattered around the globe?
The reason is that there are two specific tasks a quantum computer would be amazing at. The first is cryptography, the big money prize in the quantum computing business. Most modern day cryptographic systems rely on the extreme difficulty of factoring a product of two large prime numbers. The special nature of a quantum computer means that it should be able to factor these products, making current encryption systems obsolete. This would be a Very Big Deal and it’s why spies, bankers and tech companies are all throwing so much cash at it. The second thing quantum computers would be good for is simulating other quantum systems. That would revolutionise our ability to run simulations of biological systems, understand superconducting materials and figure out quantum chemistry, not to mention quantum physics itself.
As of early May 2017, researchers have only been able to build fully programmable 6 qubit computers and more fragile 10 to 20 qubit test systems. Industry insiders believe that somewhere between 30 and 100 qubits is where quantum computers start to have commercial value, and that these systems are likely to be for sale within two to five years. The most promising projects are at Microsoft, IBM, the University of Oxford and possibly some labs in China. The frontrunner though, is a team at Google working under a guy named John Martinis. Their latest chip has six qubits arranged in a 2x3 configuration, proving that the technology works when the qubits are side by side as opposed to a straight line.
And then there’s D-Wave, the company that keeps on making headlines. Earlier this year, there were a bunch of press releases saying they’d built a quantum computer with 2000 qubits. However while their chips rely on quantum effects, they’re not true quantum computers. Instead they rely on a process called annealing, which involves a whole lot of superconducting loops whose magnetic fields interact with each other. To obtain a solution you start with lots of energy so the magnets can flip back and forth easily. As you slowly cool, the flipping magnets settle as the overall field reaches lower and lower energetic states, until you freeze the magnets into the lowest energy state. After that, you read the orientation of each magnet and that’s the solution to the problem.
Technically, while these do count as qubits, they’re not properly entangled, their quantum states are more fragile, and their manipulation is less precise. So while D-Wave do use quantum phenomena in their devices, it’s doubtful whether they can ever be used to solve real-world problems exponentially faster than classical computers. Essentially, D-Wave has decided to go for as many qubits as possible as quickly as possible, without worrying about their lifetime, coherence or error correction. Think of them as taking the dirty approach, where most others groups are trying to take the clean approach. It might work out — as dirty approaches often have in the past — but they’ve taken a massive risk and as yet there is no strong evidence to indicate the D-Wave systems are little more than an unconventional computer architecture not dissimilar from other fast systems.
So the next time you see an article saying that some bank has invested in quantum computing, or that some research lab in China has achieved quantum supremacy, dig a little deeper. Hopefully we’ve given you enough to be a bullshit detector. If you’re ever stuck, check out Scott Aaronson, whose work we have plagiarised liberally in this post. He runs a blog calledShtetl-Optimised and it’s the last word on all things quantum computing.
Perhaps the most important lesson to take from all of this is that quantum computing is less of a technology thing, like a microchip, or a virtual reality headset, and more of a science thing, like the Large Hadron Collider or mapping the genome. If we get it right, we’ll be able to break most of the world’s cryptography, simulate the universe at the atomic scale, and gain crucial new insights about quantum gravity. Those are some pretty amazing applications.
But it’s also something fundamentally new, a problem worth solving in and of itself, for science. In the same way we’ve always built incredible machines to confirm our theories, it seems worthwhile to build a scalable quantum computer, and thereby prove that our universe has this immense computational power beneath its surface.
That seems like a pretty decent goal to us.