Randomness rules the very fabric of reality. So it only makes sense that scientists have figured out how to use nature’s randomness as a tool in our mundane world.
Random numbers go hand-in-hand with encryption and online security. But when is something truly random? There are computer algorithms that strive to generate randomness, but these can often be cracked or reproduced, or require a non-random input. Then there are various random things that happen in the real world that can be relied upon, like the balls used in a lottery drawing. But often times if you really understood the laws of physics, perhaps you’d be able to predict these ball’s behavior. Scientists at the National Institute for Standards and Technology (NIST) are planning on using the weirdness of quantum mechanics to create something truly random.
“This experiment is random so long as you don’t believe in [faster-than-light] travel,” Peter Bierhorst from NIST told Gizmodo.
In order to create something random, it must be truly unpredictable. That means that even if some enemy understands everything about how the random number generator works, they still can’t predict the outcome. That’s quantum mechanics in a nutshell—even physicists would call it unpredictable.
The most basic assumption of quantum physics is that the tiniest particles have innate properties, but you cannot reliably know the value of those properties between creating them and measuring them. If something can be produced in one of two possible innate states, you can’t accurately know what the state will be before you measure it.
But there’s always the possibility that your device that produces or measures the particles can introduce some level of predictability. So, this team took another idea from quantum mechanics, called entanglement. If the two light particles are prepared together at the same time and interact with one another, then they become entangled, and quantum mechanics won’t allow you to understand one without immediately understanding the other. There will always be a correlation between the two particles.
That might make this whole thing sound not very random. But actually, it confirms that you’re measuring true quantum randomness. So long as your light particles are demonstrating statistically that they’re entangled, then you can use their outcome as a quantum random number generator.
The experiment creates this setup with an L of fiber optic cables, each side around 130 meters long. At the intersection of the L, a pair of light particles are prepared in an entangled state by a laser shining through a special crystal, giving them one of two polarizations. Each is sent to a detector on either end. You just have to assume that nothing travels faster than the speed of light. The physicists must ensure that the random particles and the correlation between the pairs of random results come from the measurement, and not from someone sending information faster than light speed between them. Since nothing travels faster than the speed of light, that shouldn’t be a problem.
They published the paper today in Nature.
There’s a lot of post-processing that goes into this, Stefano Pironio from the Université libre de Bruxelles in Belgium writes in a Nature commentary. True, he says, the generator is “the most meticulous and secure method for producing randomness that has ever been demonstrated.” But it’s also huge and takes a lot of work just for a few random computer bits.
Generating the randomness requires actually detecting the entangled photons with the experiment. It took around 55 million photons and 10 minutes in order to generate 1,024 random binary digits. Computers store their data in binary code, rather than in base 10, so 1,024 random binary digits could be any one of 21024 numbers (its’s a kilobit).
Bierhorst told Gizmodo that perhaps there’s a future in which something relying on this method could be scaled down. On the shorter scale, perhaps they can integrate it into the random number beacon that NIST offers. It’s more of an idea that can inspire other random number generators.
I never told you the experiment would be easy, practical, or cheap. I just said it would incredibly random.
Correction: The piece previously called 21024 numbers a kilobyte. It’s a kilobit (technically a kibibit). Sorry about that.