‘The world is not only queerer than we suppose,” said JBS Haldane. “It is queerer than we can suppose.” Haldane was a biologist and something of a polymath (Peter Medawar, himself a Nobel laureate, described him as “the cleverest man I ever knew”), and whenever I read anything about quantum mechanics, it’s Haldane’s aphorism that comes to mind.
Quantum mechanics is the branch of physics that studies what goes on inside atoms. It is not for the faint-hearted, not least because it teaches you that everything you know about the physical, tactile world is wrong. “Our imagination is stretched to the utmost,” the great physicist Richard Feynman wrote, “not, as in fiction, to imagine things which are not really there, but just to comprehend those things which are there.” And at the quantum level, the things that apparently are there are seriously weird. For example: subatomic particles can be in two places at the same time – a phenomenon known as “superposition” – and any pair of them can be “entangled” in such a way that they can instantly coordinate their properties, no matter how great the physical distance between them. And the strangest thing of all is that since subatomic particles are the building blocks of matter, quantum physics is ultimately, the physics of everything.
So we live in a universe that virtually none of us will ever understand. Physicists, however, refuse to be daunted by this and have been casting about for ways of putting these quantum properties to practical use. In the process, their gaze alighted on that fundamental building block of digital technology, the humble binary digit (or “bit”) in which all digital information is encoded. In the Newtonian (ie non-quantum) world, a bit can take only one of two values – one or zero. But at the quantum level, superposition means that a quantum bit – a qubit – could have multiple values (one, zero and a superposition of one and zero) at the same time. Which means that a computer based on quantum principles would be much, much faster than a conventional, silicon-based one.
Various outfits have been trying to build one. The results are controversial but intriguing. On one test, for example, an allegedly quantum-based computer solved the travelling salesman problem (a well-known test problem in computation) in less than half a second while a conventional computer needed over 30 minutes to reach the same results. In other words, the quantum machine seemed to be 3,600 times faster than its conventional opponent.
Why is this significant?
Basically because processing power (which is a proxy for speed) really matters. Many real-world challenges (real-time language translation and breaking powerful cryptography, to name just two) currently remain unsolved simply because the necessary processing power is not yet available. And although silicon-based processor technology hasn’t yet run out of steam, we’re fast approaching the maximum density at which transistors can be etched on to a chip. So we’re going to need an alternative soon, and quantum computing is seen by some as the best bet. Which is why this week’s announcement by Google that a machine made by a Canadian company, D-Wave Systems, which is marketed as “the world’s first commercial quantum computer”, had shown spectacular speed gains over conventional computers. “For a specific, carefully crafted proof-of-concept problem,” Google’s Hartmut Neven reported, “we achieved a 100-million-fold speed-up.”
The company released a detailed technical paper with the riveting title “What is the computational value of finite range tunnelling?” on arxiv.org, and it will doubtless be pored over by hundreds of technical sceptics in the next few weeks. If speed increases on this scale are in fact achievable, then we are indeed on the threshold of a new era in computing. And, as usual, it’s not all good news. Up to now, for example, if you wanted to keep your information confidential, then the best thing to do was to encrypt it using really long (1024-bit) keys.
The margin of safety from doing that came from the fact that it would take conventional supercomputers thousands of years to crack the crypto. But if quantum computers can speed things up by a factor of 100m, then we will need to think again. Which is perhaps why when Brian LaMacchia, Microsoft’s director of security and cryptography, was asked recently what his wishes were for 2016 he replied: “quantum-resistant public-key algorithms”. D-Wave’s order book is doubtless commercially confidential, but I bet the NSA and GCHQ are on it ( via theguardian.com ).