Background
The bet that contributed to top-level science
Leiden statistician Richard Gill aided an important Delft quantum experiment. “For fun, more or less.”
Bart Braun
Wednesday 9 September 2015
Artist’s impression of quantum entanglement. A Dutch experiment appears to have delivered the definitive proof this entanglement exists.

The academic paper isn’t finished yet and the measurement data hasn’t been published yet either. Nevertheless, the news was too big to stay put. Scientists from Delft, Barcelona and Oxford made international headlines because they have done a Loophole Free Bell Test. The words “Nobel” and “Prize” have already been heard.

What kind of a test, did you say?

OK, let’s take step back. Physicists use quantum mechanics to describe the behaviour of minute particles, a system of mathematical comparisons that works extremely well: predictions have been correct to way beyond the decimal point and it has led to all sorts of neat applications, such as computers and lasers. The disadvantage is that the conclusions often feel counterintuitive to normal people.

For example?

Quantum particles can have what’s known as a superposition: they can be in two different states – or they are in two different places at the same time – as long as you’re not looking. Because if you take a measurement, the superposition suddenly collapses and the particle will end up in one of the states.

Quantum particles can also be entangled: in that case, they belong to each other and so do their properties. If you measure one particle, the other particle loses its superposition and the properties it has in that event are also fixed. Electrons have a property called spin. If you measure the up-spin of one electron, the spin of an electron entangled with it will be a down-spin.

And why is the scientific world getting all excited about this?

It doesn’t matter how far apart the two particles are.

So?

How does the other electron know it should be on the “down-spin” right at that moment?

Are they somehow communicating with each other?

The problem is that if such communication exists, it must have been faster than light in the Delft experiment – and Einstein says that’s not allowed. Mind you, Einstein was not a fan of quantum entanglement; he called it “spooky action at a distance.”

I’m on holiday, I’m going through my suitcase and I see my left glove. I immediately know that the glove at home is my right glove, even though I’m miles from home. And Einstein thought that was spooky?

No, that was precisely one of his arguments against quantum mechanics. Isn’t it more likely that there is a hidden variable – the glove factory, in this example – that would explain the shared measurement results? That brings us to Mr John Bell, of the eponymous Bell Test, who, back in the sixties, calculated that there was a maximum quantity of correlations between two particles that you can explain with that kind of hidden variable. That limit was probably exceeded in the Delft experiment.

“Probably”?

Let’s assume, for now, that the test set-up itself was correct – that’s something that peer review will go through with a fine comb. Depending on how you calculate it, the chances that the Delft researchers are wrong are either 2.5% or 3.9%. Most scientists feel that anything below five percent is acceptable, but physicists tend to have much higher standards. ‘‘It’s borderline”, says Leiden statistician Professor Richard Gill, “but they have done it very carefully and even announced in advance which statistical method they would be using. They haven’t cheated with the statistics.”

Why is Mare interviewing Gill rather than the authors?

The researchers aren’t talking to the press: major academic journals don’t allow it until the paper is on their pages. However, the acknowledgement and the reference list mention Gill: the paper uses statistical methods he introduced.

Gill adds: “About fifteen years ago, I made a bet with someone who claimed that his computer program could imitate a Bell test. I said it couldn’t and, as I wanted to win, I studied the underlying statistics, the martingale theory.

The condition I set at the time for his experiment– and the one they used in Delft too – was randomising: you toss a coin to decide which measurements you do and which ones you don’t do, to eliminate the effects of memory and time. I published that theory in some rather obscure places but in recent years experimental groups have picked it up and improved it – which it needed. I just wanted to win a bet while they are hoping to make physics history.”

Next week, Gill will be visiting the Delft lab. Wouldn’t he rather have been co-author? “I’m just happy I had something they could use. It was something I did for fun, more or less, and now it’s being incorporated into top-level science. So I’m not a co-author, but I am part of the larger scientific community that has accomplished this. And who knows what’s still to come? I think that, with smarter statistics, you could get clearer results from the same measurements.”

But if you’re looking for higher significance, shouldn’t you just keep measuring?

That’s like asking an Olympic runner to get his socks on. “They work with highly sensitive equipment”, Gill explains. “The measurements can be disrupted if a lorry drives past or if someone trips over a cable in the lab – anything. And the longer they measure, the more noise those disruptions will cause. Delft has 245 measurements now: if it were easy to do more, believe me, they would have done it.”

In other words: this work is difficult. And that’s also the reason why it’s taken decades to do this experiment. Loads of Bell tests have already been done, but there’s always been a way out, a loophole, for anyone who didn’t really want to accept quantum mechanics.

Researchers would measure light particles, but not all the light particle that were released, so they couldn’t know whether the particles they were measuring were representative for all light particles. Or the two particles they wanted to measure were so close together they could, in theory, communicate. Or -

-But no such loopholes in Delft?

In two Delft labs, with a distance of 1.3 kilometres between them, the scientists shot lasers at electrons in a diamond. Those electrons each emitted a light particle and those two light particles came together at a third location, where they became entangled – and consequently the electrons that emitted them did the same. Then the scientists performed measurements on them so quickly that the communication between the diamonds was impossible.

So, quantum mechanics are correct. But hadn’t that already been proved by all the predictions that were right, and all by those applications of quantum theory?

This can mainly be regarded as physics’ finishing touch, but physicists love precision. Besides, there’s a great application for it: quantum encryption. You can use those entangled electrons to generate a series of numbers to scramble and unscramble a text. The sender and receiver both have the same series of ones and noughts, because the electrons were entangled, but no one else could possibly know that number. If a malicious hacker tries prematurely to get a glimpse of the key, the electrons will lose their superposition and you’ll know someone is up to no good. A quantum encryption code is, by definition, secure. Even if you received it from the malicious hacker himself.

You can already buy this kind of quantum encryption: they can be produced in different ways. However, there’s a theoretical catch to it: that hacker could sell you a device that says, in huge letters: “WITH SUPER-REAL QUANTUMS!” but could give all his customers the same code of noughts and ones that he has already fed into it, so he can read everyone’s secrets.

How can you prove that the device really does what the manufacturer says it will? Well, with a loophole-free Bell test.

In 2011, the academic journal Science published an article about the race between various research groups to execute the test. “Suddenly, this test, which was originally a philosophical experiment, has a practical use with commercial rewards”, a Spanish researcher said at the time. Now we’ll have to wait and see whether those rewards will go to Delft.