Root Causes 57: Quantum Random Number Generation

Tim Callan

So we have done a series of, quite a few like seven maybe, podcasts on the topic of quantum computers and quantum resistance cryptography, which has certainly been a big topic on a lot of people’s minds and as I am out talking to people in the world and I know that you have shared with me the same happens with you, the topic of quantum random number generators, QRNGs keeps coming up over and over again and people naturally associate these two ideas in their minds. So, we thought it might be a good idea today to explain quantum random number generation, QRNGs, how they do or do not connect to post quantum cryptography and other matters.

Jason Soroko

Yeah. Sure, Tim. Absolutely. So as everybody knows, this is a PKI podcast and so random number generation is important because it is fundamental that entropy that it generates, that chaotic nature of random numbers is what makes for a good generation of a secret. In other words, if you are trying to do some good math, you know, I’m trying to very, very over-simplify this, but when you’ve got math that really depends on true random number generation then it’s very, very important to be able to do that as well as possible. So, I think in this podcast, Tim, there really are two questions here at least and the one that may seem obvious is the how. How do you do quantum random number generation? We’ll answer that. But it is also the why. We have forms of random number generation that we use classically but why do we need something else? Why do we need this quantum forum? And I think the why might be the more interesting question here.

Tim Callan

So, I think that’s great. And just for everybody’s background, listener’s background, we did do already a podcast on the topic of entropy and random numbers. It was our Episode 25. So, when you are done with this if you are interested in this topic go give that a listen because we get into some of the basics of random numbers, why they matter, how it works and what the classical kind of methods are. So, we’ve got a whole bunch of information on that. We are going to focus on the quantum problem today.

So first, let’s start with the how. So, a QRNG is different from an RNG in what way?

Jason Soroko

I think one of the most important things, Tim, is that it is essentially physically isolated. So not only is it truly random, right. Not only is it truly random, which is obviously a very important property - -

Tim Callan

That’s quantum physics, right? That’s part of what you get about it’s a wave and then it collapses into a particle, and you genuinely can’t predict where it’s going to be, so you get real randomness.

Jason Soroko

Right. So, in classical methods, we have seen poor implementations of pseudorandom numbers. There are all kinds of Google searches that you can do on that to find problems with randomness but there’s also good classical randomness typically done in hardware that uses various kinds of memes. And, Tim, you talked earlier about the league of entropy not that long ago where all sorts of forms of physical randomness that were able to be measured and generate entropy are used for cryptography. And these things are just great. But one of the things that they have in common that’s problematic is if you have an adversary that can also tap into that same stream, not only can they guess at randomness, right? In other words, (a) is there a problem with patterns in your randomness? Meaning it’s not random but there’s - -

Tim Callan

There’s some bias or there’s some other form of predictability, yeah.

Jason Soroko

And there’s also problems of too low of entropy. Meaning let’s say your randomness is generated upon things such as movement of a mouse or something like that.

Tim Callan

Right.

Jason Soroko

That in itself might not have enough entropy for the types of mathematics that you are doing. Then there’s also the problem of well, you know, if you’ve put a bunch of lava lamps on a wall and you’ve put a camera on it, you know, and I’m talking with the Cloudflare example here. Lavarand. The problem is the other people have access to that same stream of data.

Tim Callan

Right. Right.

Jason Soroko

So therefore, it’s an enclosed system.

Tim Callan

Yeah. University of Chili’s Seismic girl uses seismic activity in Chili and if you could figure out what size monitors they were tapping into and get the exact same feed you could get the same numbers. You could get the same inputs.

Jason Soroko

That’s right. That’s exactly right, Tim. So all of these things are great and in fact, they are used every day to good effect. The thing is I think that with the advent of a post-quantum cryptographic algorithms which require new forms of math to solve the similar problems that we were solving before just for different types of computers – we have a lot of podcasts on that so I won’t get into it. I think that there really is no direct connection between the usage of - -

Tim Callan

Right.

Jason Soroko

And people will probably shoot me for saying this, but I just have to say it. Quantum random number generation and post-quantum cryptographic algorithms are in themselves not directly connected. It’s not like one has to do with the other. You can use classical randomness for post-quantum algorithms, but I think the thinking is hey since we are gonna go this far with quantum we probably need for the most serious cryptographic use cases to push random number generation to the next level.

Tim Callan

Right. So, this is a good point. Let me just say it again to be really clear about it. The two technologies are not in any way interdependent. They are both examples of where quantum computers can do something better than classical computers. In the one case, we think that a quantum computer can be a better random number generator. In the other case, we think that a quantum computer can break open cryptography in a way that a classical computer cannot. They are both instances of where quantum computers are or will be superior and they are both things that are involved in the world of cryptography, but you don’t need one for the other or vice versa. However, advances in quantum computing make both technology types viable and will bring them both into the mainstream and ultimately, they are both going to be part of the cryptographic landscape that we are all working within.

Jason Soroko

Yeah, and for those of you who are in this business who are tying directly QRNG to your post-quantum algorithms, we’d love to hear from you. But, for now, on this podcast the assumption really is they are two separate things, but they, definitely can work hand in hand together. I’ve seen it to the point, Tim, it’s quite interesting where some of these – and this is how we are gonna get into the how part of it. I think we’ve been previously talking about the why, but in the how, it’s quite interesting to see those new patents and new innovations and really, really good engineering has made it to the point where you could have a hardware-based quantum random number generator that is small enough to put into an IoT device.

Tim Callan

OK.

Jason Soroko

And that makes possibilities quite interesting.

Tim Callan

Yeah. Absolutely. So, I mean what is the right leveled technical explanation of the how on this? Like how do they do this?

Jason Soroko

There have been a few different forms of random number generation over the years. This is not an entirely brand spanking new topic. I think what’s new is exactly what I just said, which is the miniaturization of one particular form of it. So, for those of you who are maybe physics, you know the history of physics with the arguments between Neils Borh and Einstein about the uncertain. That’s quite interesting. If I just bypass the physics course for a moment and get a really, basic use case of how this is typically done now, especially in the miniaturization of it, what has changed in terms of engineering if not physics to really get to the point where we are. The experiment that could be done is this. You take something that could emit a single photon, which is quite difficult to do, you know, typically from a laser or something like that. The idea behind it is if you pass it through a substrate that is at a 45-degree angle more or less and if you actually have that to the point where you are able to polarize that substrate and have that photon pass through it there is, there is with a whole bunch of other very, very important controls within the physics experiment that I won’t get into right now, the interesting principle is that there is a pretty much - - or to be more accurate an absolutely guaranteed 50/50 chance that the photon will actually pass directly through it or actually reflect - -

Tim Callan

Bounce off it.

Jason Soroko

Exactly right. And so because it’s a 50/50 chance, it really has no bias and so therefore if you can pass enough of these photons through you have a really, really good set of random 1’s and 0’s essentially. If you have one of the detectors as representing 1 and the other representing 0, you essentially have a series of 1s and 0s that are what physics would tell us now is truly, truly random. So that’s an interesting - - it’s a very, very simple experiment and it’s very, very - - it’s something that because of just the laws of just the crazy, crazy randomness of quantum which, you know, we have to have other people on to truly explain what the heck is going on within the chaos of that mechanism but the beauty and the simplicity of that is such that engineering has now gotten to the point where, you know, single photon emitters, detectors that are very, very accurate and the miniaturization of these types of electronics has now gotten to the point where these things are now viable solutions. They don’t just have to sit in a university laboratory to prove a point. They can now be used to generate entropy for real-world cryptography.

Tim Callan

Yeah, and they don’t need to be super cooled to 10-degrees kelvin, right? You can have them out in the pragmatic world doing their work in your device that’s just in whatever environment it happens to be in.

Jason Soroko

That’s exactly it, Tim. So, you know, I think that the thing to keep in mind here is people will say, well big deal. We already have hardware-based randomness. And you know what? I think that’s the point here, Tim. I think that’s the point is I don’t think that classical hardware-based random number generators are going anywhere.

Tim Callan

Sure.

Jason Soroko

I think that QRNG, quantum random number generators, as I just described are really going to be for the hardcore, I absolutely need to know that this can’t be fudged with random number generator.

Tim Callan

Yeah.

Jason Soroko

And keep in mind, Tim - - and you know what? I’m gonna make one more point here about this which is we were talking about some of the weaknesses of hey if you can tap into my randomness stream then maybe you can guess at what number I chose, etc. etc. Keep in mind that even QRNG has its issues that must be solved with engineering. In other words, hey, since it’s a laser, since it’s a photon, could I interfere potentially with the emitter? Could I interfere with the detector? Could I somehow cause an interference pattern that causes a bias in the movement of the photon? All those things absolutely are true. I think if you were to do a search right now on the actual physical and commercial implementations of this technology that’s where a lot of the thinking has gone into. Which is quite interesting.

Tim Callan

Right. Right. Sure. Because it is light after all. It is affected by electromagnetic fields and yeah, you could imagine problems either intentional or accidental interfering with the randomness of these photons for sure.

So I feel like you touched on a valuable point that’s worth hitting, which is that just as with quantum computing we are not predicting anytime in the near future that these quantum surrogates are going to eliminate your classical binary computing versions, right? Just like we don’t think that we are all gonna stop using 10 gated chip computers, along the same lines we don’t think that your old-fashioned RNG card is going to go away. What will happen is there will be specific circumstances, specific use cases, specific applications where the quantum device is either much better and it really matters or where your need, in the case of the randomness, your need for randomness is so extreme that you are going to go the extra mile. But for the most part, that’s not gonna be the case, right? If I’m generating a cert that I’m gonna put on my relatively low-value employee device on their laptop and if the value of what’s gonna be stolen on that laptop is reasonably limited them, I’m probably not going all the way to that extreme. My off-the-shelf RNG is probably just fine. And at least for several years, that’s how we expect the market to play out. You agree?

Jason Soroko

Yeah. I see that that is what this is, Tim. And that’s why we are podcasting about this now because it is a topic. People are asking us about it.

Tim Callan

Yeah.

Jason Soroko

I think most people think that there is a direct connection with I need to have a QRNG when I’m doing post-quantum cryptographic algorithms. It’s not the case, but on the other hand, it’s a very interesting subject. I think it’s the new standard in real, real randomness. Randomness to the point where we can trust it, which is not a world we’ve moved in in a very long time. But it is absolutely for the most paranoid use cases probably now and probably quite a bit into the future.

Tim Callan

Yeah. It’s just a fun subject. I don’t know why. Something about randomness and entropy I just find fundamentally fascinating. So I always love to explore what is going on in that part of the world just because it’s just always really interesting.

Jason Soroko

Crazy thing though is Tim, with all this quantum stuff so many of the properties and, you know, this could be a podcast that we may or may not want to do but the whole - why does this work the way it does? Why is it 50/50? You know? What causes quantum entanglement and what the heck even is that anyway? I mean back in the early part of the 20^th century when Neils Bohr and Einstein were having their arguments, goodness gracious, that was disturbing to them, and I think it is still disturbing to us. The world is a crazy strange place and I think I’ll leave us with this, Tim. I think the world we live in, you know, with TED Talks and all these other things that make you feel like everything has just been sorted out and guys like Elon Musk have enough money to just buy the solutions to everything, we just don’t live in that world. And I think as human beings we just still must have a lot of humility for the laws of the universe that we - - I honestly think we are just still starting to begin to understand it.

Tim Callan

Oh yeah. I completely concur at that last point. You know, as we look back upon how people thought about physics 100 years ago, we realize how primitive it was and 100 years from now people are going to say the same thing about us. As they are using their super-spiffy quantum computers to solve problems that we can’t even conceive of solving today. So yeah. Thank you, Jay, for a fun talk about a fun topic and, you know, QRNG, as we said at the top of the show, it just comes up a lot and I find myself explaining it a lot. So that’s why I think it’s good that we get this out there so people can listen and really get how these technologies connect to each other.

Jason Soroko

Yeah. And for those of you obviously who are really interested in this subject, there is a ton of material out there. Unfortunately, it’s all aimed at physics students, so it’s not really brought to the layman’s terms but that’s what we are trying to do here so for those of you who are physics students who are laughing at our over-simplified examples, please have mercy on us and try to understand we are trying to reach a different audience.

Tim Callan

Right. Yes. And maybe have mercy on me because I am just trying to understand some of these things. So, with that, probably a good place to leave it. Jay, thank you very much.

Jason Soroko

Thank you, Tim.

Tim Callan

And this has been Root Causes.