Preston Pysh (00:05:54):
So for me, when you were describing that, it immediately sounded like a vector. Instead of the vector only pointing left or right, you’re able to kind of point it in many different directions in order to capture more data or more information inside of that, we’ll call it a thing right now. Is that an accurate description?

Andrew Fursman (00:06:13):
Yeah. For your readers who are hanging out or your listeners who are hanging out in front of a computer, there’s a thing called a Bloch Sphere. And if you google that, you’ll come up with this picture that’s basically exactly what you described. It’s like a vector that starts in the center of the sphere and reaches out and points to some place on the outside of it. And that’s the framework that people use to visualize what’s happening inside a quantum computer.

Andrew Fursman (00:06:38):
But I hesitate to go much further in this direction because in some sense I feel like a lot of the conversations around quantum computers really kind of dive into what is a quantum computer in the sense that it would be like if you asked me what a car was and I described that it was 1.2 tons of aluminum that was forged at this location and mined from this place. That’s all interesting and it’s all true. If you want to build a car, you need to know that.

Andrew Fursman (00:07:08):
But if you want to know why you should care about a car, it’s kind of the exact wrong place to be able to start, because in some sense, you should want to care about a car before you want to know how a car works. You need to be motivated to actually invest the time and understanding how to build one of these things by saying, “I have this problem. My horse is only capable of going a certain number of kilometers in a day. I wish there was a better way.” And that’s, I think, actually a much more interesting place to talk about, especially the overlap between my interests and your interests and how they sort of come together in this interesting world.

Andrew Fursman (00:07:44):
So in some ways, I think maybe talking about what quantum computers do is an interesting way in order to motivate people to care about how they work.

Preston Pysh (00:07:54):
Let’s pull on that thread a little bit. Let’s talk about the applications as they exist today for quantum computers, and then we’ll go a little bit deeper after that.

Andrew Fursman (00:08:03):
Well, the thing that’s really great is we don’t have to go deep into the applications for quantum computers because at present there aren’t many and the applications that do exist are really only capable of being run at a very small scale. So you could think of it. I like to use the analogy of if you had a calculator that only had one digit slot, you could appreciate the power of a calculator by using this calculator even though this calculator would probably not actually make you better at doing your taxes or more or less any useful thing that you can do.

Andrew Fursman (00:08:38):
And that’s where quantum computers really sort of sit today is, they’re at the stage where the quantum computers that you use presently, you can do the equivalent of two plus two. You can see it gets four. You can know that quantum computer is behaving correctly because it’s doing that. But if you try to do the equivalent of five plus five, you’ll get an error and you’ll realize this is sort of gone beyond the scope that we have right now.

Andrew Fursman (00:09:03):
Now, what we’re usually doing instead of one plus one, two plus two, we’re usually using these quantum computers and the quantum information that they contain to simulate the interaction of quantum information in the real world. What we’re mostly at 1QBit, the company that I’m working at right now, what we’re doing here is mostly thinking about how to build better quantum computers and how to help quantum computers be better at doing the things that we’re interested in using them for and building applications that really take advantage of these fledgling capabilities. And we’re almost entirely in terms of using these things for real applications, focused on the emergence of chemistry from physics.

Andrew Fursman (00:09:48):
And that’s extremely interesting because that’s really something where the way the universe actually works is that there are these physics properties that come together in nature and in reality, and based on how physics works, all of chemistry sort of emerges based on these calculations and the calculations that are vastly more complex than any traditional computer is likely to ever be capable of.

Andrew Fursman (00:10:16):
So what’s really exciting is if you’re interested in moving from a chemistry laboratory to a chemistry laboratory simulation in software, like I said at the very beginning, you need a computer which is good at things that classical computers are really bad at. And it happens that one of the strengths of quantum computers is manipulating quantum information and manipulating quantum information gives you a great insight into understanding how the world around us is constructed at that sort of chemical material level. So that’s one of the things that we’re really excited about.

Andrew Fursman (00:10:55):
In fact, if you listen to a lot of talks that are sort of primers on quantum computers, it’s often kind of noted anecdotally that the reason that people really think about quantum computers largely stems back to or credited to Richard Feynman, who was saying not we want to build a quantum computer initially, but we want to build a quantum simulator. And what he meant by that was not that we want to build a computer that can simulate the quantum world. He was saying we want to take quantum materials and build them into a simulating device.

Andrew Fursman (00:11:28):
So that’s really the original idea behind a quantum computer and the first applications of sort of what we call large scale fault-tolerant quantum computers are likely to also be in this realm. Now, more interesting probably to the conversation that I think we’re going to get into is the fact that after that idea was proposed but before the first quantum computer had been built, a gentleman named Peter Shor came up with this idea of using quantum computers to actually do something else that computers, traditional computers are not very good at, and that thing ends up being related to factoring which becomes related to encryption.

Andrew Fursman (00:12:10):
And that’s one of the other sort of well-known things that a quantum computer of a sufficient size is capable of doing and it’s certainly one of the reasons why people who are interested in cryptography and cryptocurrencies are close watchers of the progress of quantum computers, because as I think we’ll get into discussing, if you can do things that traditional computers are bad at, there is the possibility that what we considered hard tasks become easy, and that has ramifications for people who are interested in cryptography, privacy, and cryptocurrencies.

Preston Pysh (00:12:45):
So let’s go ahead and talk about that, because I think that’s where most of my audience has a genuine interest, is on the security side. So I’m not even talking Bitcoin, just talking encryption in general. And you think about e-commerce, you think about everything that’s going over the internet in a secure way. It’s doing it through encryption that is pretty much been standardized on in global finance. So when we start thinking about what those implications are and how disruptive quantum computing could be to that, we’re not just talking about Bitcoin. We’re talking about everything. We’re talking about Amazon. We’re talking about the big banks, literally everything. So talk about the implications of this. How far off from a timeline are we? And I know that’s insanely difficult to project, but just give us some of your thoughts on some of those ideas.

Andrew Fursman (00:13:37):
Maybe the first thing to say that almost sounds condescending, but I feel like if I just keep saying this to myself, then it really helps anchor me in understanding that we’re talking about quantum computers and not magic. So the refrain is quantum computers are not magic. And the reason that that’s really helpful is because there’s sort of all the crazy spookiness of the quantum world and in some ways the quantum world itself, when you learn some of the things that are just true as far as we can understand, quantum itself and it’s the ways that we describe it are very strange. And there’s a tendency to sort of just import the sort of mystical weirdness of the interpretations of the quantum world into quantum computing. And that’s a bit of a mistake.

Andrew Fursman (00:14:26):
It’s really better to think about it as a device that works based on the principles of the quantum world, but which is capable of doing some very specific things and is incapable of doing most things. So one thing that a quantum computer isn’t is a better computer. And I say that because computers are really good at almost everything and pretty bad at a fairly small number of types of things.

Andrew Fursman (00:14:56):
If you have a computer, you can calculate almost anything. And because of that, when people were initially trying to think about encryption and basically being able to communicate things securely, one of the first things that people had to really question is what are computers bad at? And essentially they took the approach of if we just make it so that decrypting our private information requires you to do something that computers are bad at, then in some sense it shouldn’t be that difficult to build a process where even if you have a computer, you’re incapable of reading my message because the message requires you to do something that computers are bad at.

Andrew Fursman (00:15:40):
Now, it’s even slightly more tricky than that because it not only needs to be something where let’s call it unzipping your message is something that computers are bad at. Also has to be something where actually locking these messages is something that your computers are good at. And so people I’m sure your listeners are really familiar with the idea of one way functions or sometimes referred to as trapdoor functions, but that’s kind of exactly what people did. They said if we can figure out what some of these different types of algorithms that are easy to do one direction and hard to do the other direction, then we can use that in order to make secrets. And as long as what’s hard for computers today is hard for computers in the future, then we should be able to rely on these as being generally good methods.

Andrew Fursman (00:16:26):
And that’s how most of what we think of is exchanging private information or keeping secrets kind of works. And it goes back to that first point that we were talking about, which is that it just happens that quantum computers are good at doing a small number of things that classical computers are pretty bad at. And one of those things is exactly related to that process of sort of unzipping that message without having the requisite knowledge that’s necessary to do it efficiently on a classical computer.

Andrew Fursman (00:17:00):
So that’s why sort of at a high level why cryptography and quantum computers are related, is because it’s not even just that it’s hypothesized that if you had a quantum computer, you might be able to do these things. In some sense the theory came ahead of the realization of the device itself, and it’s mathematically proven that if you have a device that looks like the kind of quantum computers that people want to build, then you’ll be capable of decrypting this information significantly better than could ever be possible with a classical device. And that’s sort of the interesting piece.

Preston Pysh (00:17:39):
So when I look at and I think this is important for the audience. You had mentioned the one way functions. I’m just going to explain this as simply as I can. And if you find that my definition is in error here, please correct me. But for folks that aren’t familiar with what we’re talking about, with Bitcoin, when you have a private key and you have a public address, think of it like this. If I told you my address to mail me a piece of mail, that’s obviously my public information. Anybody can know that. But to open the door to my house and to get in, there is a key that’s associated with that address. The person who has that private key can then open the house to come in.

Preston Pysh (00:18:14):
When you think about encryption and you think about how is that private key generated, it’s generated through a one way function. Think of it like if you provide some inputs into this function and it produces an output, it’s really easy to kind of generate that output. But to go the opposite direction, once you know what that output is and to figure out what the inputs were in order to generate it, it’s extremely, extremely difficult and the only way to figure it out is through a bunch of guessing. So that’s where the computer comes in.

Preston Pysh (00:18:44):
The computer is making millions upon trillions of guesses or whatever those numbers are in order to try to figure out what the original inputs were in order to generate that private key. So, Andrew, if I missed anything or you think that that’s a kind of a bad description, please chime in.

Andrew Fursman (00:19:00):
No, it’s great. That’s absolutely the definition of a one way function. And to make it even more real, you can think about it like this. One of the ways that we actually use this and specifically how sort of what’s at the core of the RSA method of privacy that’s commonly used is actually to say if you have a very large number that has only two factors, then being able to find those two factors is very hard. And yet being able to create a very large number that is the result of multiplying together two primes is extremely easy.

Andrew Fursman (00:19:39):
So think about it. If I give you a number. Let’s say we want to encrypt this conversation. Then my public key can be 8,633. What’s interesting is there’s only two factors that actually multiply together to make that number. I chose that number because of that fact. It’s to try and figure it out really just to say, “Okay, well, is it even? Is it divisible by two? No. Is it divisible by three? No. Is it divisible by four?” And we have to keep kind of counting that way until we actually find one number that divides into it. And of course, that gives you the other number.

Andrew Fursman (00:20:14):
So in this instance, even though it’s very easy to multiply 89 and 97, if I just give you 8,633 and ask you for that, it’ll take you a pretty long time to be able to give me 89 and 97, which are the only two numbers that are divisible or that that number is divisible by. So I think that sort of really helps to kind of get a sense of it. When you realize if I give you a pen and a piece of paper, it might not be trivial to multiply two numbers together, but you can do it so much faster than you can do the reverse, which is to break that number down into its components. And that feature, the fact that even for humans it’s easier to multiply it than to be able to find out what two prime numbers multiplied together to create this larger number, that’s something that is easy to experience, the one way-ness of that function. We built most of the internet on top of that.

Preston Pysh (00:21:10):
You know how it helped me really understanding it? So there’s websites that have various encryption functions on them, so like the SHA-256 encryption that Bitcoin uses. What I did is I went on there and there was just the text input. And you could literally enter anything, and that would produce the output of the hash function.

Preston Pysh (00:21:29):
So I just went in there and I played around with it. I typed the word the, and there’s the hash function. And so I wrote down what that hash function was. And then I changed it to something much longer, like a paragraph. And it’s still output a hash function that had the same number of string variables inside of it. Is that the correct terminology to use? But however many outputs there was, it was the same whether I put the word the in there or I put a whole paragraph or I put a whole page or an entire book, it’s still producing a hash of the same length. So when I went back and I put the word the back in, it produced the exact same output that I had used before.

Preston Pysh (00:22:03):
And that was kind of a light bulb moment for me to kind of really understand this one way function hash function that encryption techniques use so that it produces the same output based on the input that was provided in this public and private key kind of situation.

Andrew Fursman (00:22:19):
You kind of touched on two concepts which are really closely related, but it’s meaningful, and I think one of the answers to a question that will likely come up later where we can refer back to this moment, but encryption and hashing are very closely related, but slightly different.

Andrew Fursman (00:22:33):
And so probably for most of this conversation, we’re not actually going to be talking about hashing functions like SHA or MD5 people might be familiar with. Those are all like Merkle–Damgård construction based mathematical methods that are slightly different from encryption. And the reason is because exactly like you described, if you put in the, then you’re likely to get a fixed string of characters. And if you put just the one, like maybe a one character at the end of the, it completely changes what the hash that you receive at the end looks like.

Andrew Fursman (00:23:10):
So a small change makes it completely different. But what’s also interesting is because of the fact that it’s giving you a fixed length output, there is actually many different things that could be hashed into the same output. So even if you know exactly what the hash function is and had the ability to run it backwards and to be able to see the outputs, there’s many possible things that could have done that. So it’s actually not the case that even if you sort of know how to reverse that out, that you can be certain which of those inputs were used to be able to create it. Whereas with encryption and I think we’ll probably be talking more about RSA and elliptic curve cryptography and things like that, what’s interesting is that they are also based on sort of that complex mathematics and kind of a one way style function. But it’s literally a mapping where you can both put something into the encrypted format and then take it back to the unencrypted format, which is slightly different from hashing.

Andrew Fursman (00:24:14):
And so given that cryptocurrencies touch on both encryption and hashing, it’s important to sort of have that slight distinction because it becomes meaningful for especially quantum computers.

Preston Pysh (00:24:25):
Very fascinating point. I was not aware of that. Let’s talk about where some of this is today. The last I read on this big Google announcement, they had a 63 qubit quantum computer. On the surface of this, I’m looking at this as an outsider who really doesn’t understand any of this stuff in any type of depth. And I’m saying to myself, “This is getting crazy. Like they could probably do 100 or 500 or 1000 in the coming years.” In fact, I saw IBM is making an announcement that they’re trying to have a thousand qubit computer by 2023. So what is your opinions on that? How meaningful is that? Just in general, what are your thoughts?

Andrew Fursman (00:25:04):
My first thought is that when you said the G word there, my G powered home picked you up and started giving you answers about quantum computers. So I feel like Google really gunning for me from a couple of different angles here. Yeah, absolutely. The thing that’s really great about what we’re doing presently as a community of researchers thinking about quantum computers is that in addition to really thinking about the algorithm side, stuff like the SHA’s algorithm that we talked about before, or the modeling of the emergence of quantum information, there’s also great work that’s being done to actually build different types of these devices.

Andrew Fursman (00:25:45):
Some companies are interested in being able to take an approach that’s really kind of take the computing technologies that we currently have and make them quantum. And so you can think about Google’s approach of using superconducting materials to build computers. They’re actually not exactly the same, but they’re similar in terms of how you would build them to a traditional computer. There’s a lot of technology that’s come from building computers that were capable of reusing in that style of quantum device.

Andrew Fursman (00:26:16):
But there’s also some people who have taken a very different approach of saying start with something that quantum and make it into a computer. So people have been trying to build computers that work on photonics for a long time. But if you could build a computer that calculated utilizing photons, since photons are inherently quantum mechanical, you’ve kind of got a leg up that way. Or you might have heard of ion trap quantum computers. There’s a company called IonQ that’s making a lot of waves right now for the devices they’ve built. They’re actually about like the base that they have are atoms that are suspended in sort of a trap, is why it’s called an ion trap, and addressed with lasers. They work very different than how the computers that we’re familiar with operate. And again, that’s not surprising because quantum computers are just computers built a different way that have different strengths and weaknesses than the computer we traditionally use. So there’s a few different ways that you can build these devices.

Andrew Fursman (00:27:20):
At this moment, there’s this really exciting race that’s going on in order to figure out not just how many of these quantum bits you can produce, but also what ways of producing these bits look most scalable. Which ways are the most stable? Which ways are … You could think of as the tortoise and the hare sort of situation where someone can blast out the gates with something that’s very good in the short term, they get to one, two, three, four qubits before everybody else. But then it’s never capable of getting to the thousands of qubits that you might need.

Andrew Fursman (00:27:54):
I guess I try and give that introduction in order to frame the fact that my answer of I don’t know is based on the fact that there are so many variables that are happening right now where the first usable quantum computer might actually be based on techniques that we’re not aware of right now. Or it might be the fact that the thing that Google has just done to be able to show quantum supremacy, which is a term that was really coined in order to talk about the first time the quantum computer is able to outcompete a classical computer on the quantum computer’s home turf really, basically saying if we tilt everything in quantum computers favor, are we capable of giving a quantum computer any problem, whether it’s useful or not, that a classical computer can’t do better? Because, of course, if a quantum computer can’t even do useless things faster than a classical computer, then it’s going to be really hard to find useful things for it to do.

Andrew Fursman (00:28:54):
So this is sort of where we sit, is that now people have been able to show on a few different types of computing devices that they’re able to do fairly unuseful or let’s say not generally applicable functions that outcompete classical devices. So the idea that a quantum computer will never be capable of doing anything that can’t be done better, faster, cheaper on a classical device, that’s sort of out the window based on these latest results. But we’re still not in a part where we’ve actually built quantum devices that are capable of doing something useful better than we can currently do with classical devices.

Andrew Fursman (00:29:36):
And this is sort of a really interesting middle time where the success that we’ve had to date suggests that we should be able to get there, there being making quantum computers do useful things for us in society. But because we haven’t actually reached that point yet, knowing exactly what it looks like when we achieve that point is really something that’s up for grabs based on all the different approaches and all the groups that are working at it presently.

Andrew Fursman (00:30:04):
One thing that’s pretty clear, though, is that when we think about something like Shor’s algorithm, it’s not like we’re 50% of the way there or 80% of the way there. We’re sort of at like the 1%, 0.1%, 0.01% of the way there, where it’s kind of like going back to this idea of having a calculator with only one digit place. What are you capable of doing with that device? Understanding why calculators are valuable. What are you not capable of doing with that device? Your taxes.

Preston Pysh (00:30:37):
So going back to your comment there, we’re at 0.01% of the way there when you’re talking about Shor’s formula and we’re talking about elliptical curve, being able to crack elliptical curves at that point. So if let’s just say IBM’s claim that they’re going to be at a thousand qubit here by 2023, if that would play out, let’s just say that that’s a valid assumption on their part, where would we be in that percentage if we’re 0.01% today, where would we be in that percentage if that was true in 2023?

Andrew Fursman (00:31:08):
Well, one of the things that’s really interesting to note, and again, I’m going to draw on traditional computers, classes of computers for this, is because of the fact that our individual bits are susceptible to some errors and computers are a little bit imperfect, when we talk about having a bit of information that’s stored in your computer, you actually store that bit a little bit redundantly. You can think about, many of your listeners will be familiar with that movie Minority Report, where there would be occasionally of the three people who are supposed to three, somebody is going to dissent. And this is why it’s good to have three people, because you can always check who’s the outlier.

Andrew Fursman (00:31:52):
But the less reliable any individual person is, the more people you need and the more redundancy you need in order to be statistically likely to be able to ferret out any errors. So with quantum computers, because they’re so unstable, it’s actually hypothesized that you could need on the order of a thousand or even more qubits to represent what we call one logical qubit, meaning you might need to take something like a thousand qubit IBM device and actually turn all of those qubits into a device that’s through something like the Minority Report style redundancy able to firmly hold the data of one qubit within it.

Andrew Fursman (00:32:39):
So that’s one of the first things, is that when we think about error correction or sometimes people talk about service code, essentially they’re all different ways of trying to make it so that you can believe that when your own computer gives you the value of a qubit, that it’s giving you the appropriate value that you want. That requires some redundancy. And that redundancy sort of instantly kneecaps the total amount of usable qubits that you have out the gate.

Andrew Fursman (00:33:08):
Then you have the next piece which is so let’s assume that a thousand qubit equals one qubit for sake of calculations. Then you might need a couple of thousand qubits or beyond in order to be able to do these types of calculations. We might need actually significantly more qubits than are currently available. And like I sort of alluded to, we might be at the point where the largest computers that we’re building today end up really becoming the foundation of one logical qubit for one of these large devices.

Andrew Fursman (00:33:42):
So if we need a thousand times more qubits than we might have in a few years, then you sort of have to be thinking about the growth of these things from both that error correction standpoint and the number of logical qubits that you need to go forward. And I should say some people even put the number as high as millions that you might need in order to do useful stuff. So we’re definitely, we’re not right around the corner from this. It’s not going to happen next week. But there is sort of a lively debate right now about in most technologies going from zero to crappy is kind of the hard part. And then going from crappy to good is something that happens quickly. And some people hypothesize that actually with quantum computers going from not existing to crappy might actually be the easy part and going from crappy to good to be really difficult.

Andrew Fursman (00:34:37):
A lot of that, again, depends on what type of quantum computer you’re going for, whether you’re looking at ion traps or photons or superconducting devices. And that’s why there’s no real easy answer to questions like how far in the future for such and such an application. It’s because as much as this doesn’t fit into a tweet, being able to say, “Well, depending on the overhead of error correction, depending on the total number of qubits from a logical qubit standpoint that you would need to run your device, and depending on what type of underlying quantum information you’re using, like a photon or an ion, the answer is very different. But all of the devices that we’ve built to date are really more like proof of concept for being able to show that we should be able to build an error corrected qubit.”

Preston Pysh (00:35:27):
So what you just said was so important to my own understanding of this, because in preparation for this discussion, I was doing research and I’m coming up with the questions like how many qubits is it going to take in order to crack elliptical curve cryptography? How many qubits is it going to take to crack 2048 bit RSA? And because that’s what SHA-256 uses.

Preston Pysh (00:35:48):
When I was finding the answers to these questions, I saw, oh, it’s 1,300 to 1,600 qubits to crack elliptical curve cryptology and it’s 4,100 qubits to crack the SHA-256. And I’m thinking to myself, “All right, so that number is not real high compared to these numbers that I’m reading that IBM and Google are saying they’re going to be creating here in the next three years.” And I’m thinking, “This doesn’t jive with the general narrative that I’ve heard about Bitcoin and just encryption in general and how it marries up with quantum computing.”

Preston Pysh (00:36:20):
But based on what you just said, now it’s crystal clear to me, because if IBM’s creating a thousand qubit computer here in the next three years and a thousand qubit computer might equal one reliable qubit when you’re comparing it to these numbers that I just stated, now all of a sudden things look like the 0.01% of the way there that you were describing earlier. It all makes sense to me.

Preston Pysh (00:36:44):
So when I’m looking at this and I’m looking at the problem that they’re obviously trying to solve, which is not an easy problem. And from my understanding, it all comes down to the ability in order to de-synthesize that information. So what are some of the ideas, and I know so much of this is theoretical, but how are they going to synthesize these 1000 qubit computers in order to get to this eventual 500 or 1000 qubits of processing speed?

Andrew Fursman (00:37:15):
Again, the nice thing is, although there’s so many unknowns, we can really take a lot of guidance from the traditional computers that were built way back in the day. I’m thinking about when computers were vacuum tubes, for example, we had the same sorts of problems where error correction was desirable and also where these machines were not capable. It wasn’t feasible to imagine building something, say, as powerful as an iPhone, but out of vacuum tubes, because it’s like the sorts of things where you hear like, “Oh, we just need to turn like all of the glass in the world into the vacuum tubes,” and then without enough vacuum tubes to do this stuff.

Andrew Fursman (00:37:55):
One of the things that’s pretty clear is that it’s unlikely that future quantum computers are going to be big versions of the small quantum computers that we have right now. And you also never know what you don’t know. So on the sort of pessimistic side, there are some people who point out the fact that there might be theoretical limits that we don’t understand that make it so it’s really impossible to ever build anything beyond kind of a proof of concept quantum computer.

Andrew Fursman (00:38:23):
But on the other hand, somebody could make a fault tolerant qubit tomorrow that’s inherently scalable. And then you’d be in a situation where you’re really much closer to these things. So hence the huge error bars on most of these conversations around know how close are we or how far away are we? It really depends on whether or not our unknown unknowns sort of fall in our favor as quantum computing enthusiasts or fall against us. And there’s really nothing to do to find the answer to that question except for either heads down do the work or probably pay attention and watch what sort of results are coming out of the leading institutions that are building these devices.

Preston Pysh (00:39:06):
So as an investor, I’m always trying to calculate what are my risks and I’m trying to think of that left and right limit. So what you just described, let’s just say on the one side of the spectrum we got, this isn’t even solvable from a physics standpoint. And it’s so far off that it’s just nothing to even be concerned with. On the left side of that risk curve, there could be a breakthrough in the coming year that could have a profound impact on just the speed that this could take place.

Preston Pysh (00:39:34):
Let’s play around with that narrative. Let’s say that that happened. How much time … Let’s say that that plays out. When you think about how the technology has progressed to date and the implementation of then doing this at scale to get something that could actually produce a thousand qubits of actual processing power, which might be a million qubit computer, once the breakthrough is achieved, what is that? Typically, a three year lag before they could actually implement and build this thing physically in the world that we would see it? Or is it a 10 year horizon?

Preston Pysh (00:40:10):
Because the reason that that’s such an important question for not only myself, but probably everybody who’s listening to this, is they’re thinking about the speed at which the adaption to that breakthrough would have to be built into. My understanding is that you’d have to then go to some type of multi-dimensional elliptical curve generation in order to be built into today’s private key generation, in order to protect against something like this. So what’s that timeline? I know this is another timeline question, but it’s so important to the investment thesis for people that are thinking about the risks.

Andrew Fursman (00:40:43):
Because of the fact that it’s not just about cryptocurrencies but cryptography just generally, it kind of is a question that’s analogous to, okay, we believe in climate change, it’s not going to do something catastrophic for us today, but you never know when it’s sort of reached some tipping point. So when is the right time to start taking action about something that you know is inevitable? You’re just not sure how far into the future it is. Probably the answer is that it makes sense to do something about it now.

Andrew Fursman (00:41:17):
Especially one of the things that’s really interesting to think about is this sounds more spooky than it actually is, but we’re kind of being hacked now by quantum computers of the future. And the way that that works is if you’re right now on, say, a Wi-Fi network or if you’re making a banking transaction and you’re doing it from a coffee shop or something, you’re sending information. It’s not like it’s in a tunnel that’s private only to you in some real sense. Everybody can see the information that you’re sending. It’s possible just to sort of flip open a laptop and see all the communication that’s going back and forth. We just can’t know what the contents of that communication is now, but we can store that, and in the future, we can use these devices in order to be able to decrypt it.

Andrew Fursman (00:42:08):
So one of the things that’s interesting is the question of what’s the shelf life of your secrecy. It’s especially important because my guess is that the NSA or some government organization might have secrets that they’re interested in keeping for a prolonged period of time. And whether quantum computers come out tomorrow or in five years or in 10 years that are capable of being cryptographically useful, those devices are going to be capable of doing something that you might not want if you’re somebody that’s keeping a secret.

Andrew Fursman (00:42:45):
Now, because of the way that the blockchain works or blockchains tend to work, there are some things where actually being able to know old secrets is not particularly damaging.

Andrew Fursman (00:42:58):
So it’s worth getting into what are the different ways that the blockchains rely on cryptography and which of those are specifically relevant to things that quantum computers of the future might do. And how much is that really a problem for people today versus not a problem at all? And what things are maybe not a problem yet, but we might want to be thinking about working on them because better do it now and sort of better safe than sorry? And we can get into all those topics.

Preston Pysh (00:43:30):
So when we talk about making things quantum proof, the thing that for me is just an intuitive response is double encryption, triple encryption, just using the same SHA-256. So if I take something, I encrypt it once. I take it again and I encrypt it again. Is that getting twice as hard, three times harder every time you do it, or is this some type of exponential curve when you’re double or quadruple or triple encrypting something?

Andrew Fursman (00:43:57):
A quick point on just the idea of course, we talked a little bit before about how hashing functions are a little bit different than encryption functions. So if we think more about say RSA, then one of the other things that’s really helpful is going back to kind of one of our first principles of quantum computers aren’t magic. So even if quantum computers are capable of doing something in some amount of time, it doesn’t mean that it’s capable of doing everything in no time.

Andrew Fursman (00:44:27):
So just like a regular computer might take some amount of time to be able to generate or answer a tough mathematical question, there’s still some amount of time. It might be eight minutes. It might be eight hours. But a quantum computer will still have a clock speed, sort of a pace at which they’re capable of doing their primitives. And they need to be able to do a bunch of those operations in a row in order to answer these sorts of questions.

Andrew Fursman (00:44:57):
Now, they might be trillions of times faster than classical computers, but that’s still not the same thing as instantaneous. So even if there’s something that a quantum computer is capable of doing relatively trivially with a particular key link, it’s likely the case that it would take longer to do the same thing for a longer key. So even though it maybe doesn’t scale the same way that we would expect and there’s a lot of questions about this in practice, like what is the actual device that’s working on this, what is the clock speed, what are the ways that these things can interact? But what’s important to know is that it’s still going to take some amount of time and there’s still some number of procedures that have to happen one after another in a quantum computer to be able to answer these problems.

Andrew Fursman (00:45:45):
And so, to your point, it might not be very sensible to, for example, a regular computer if you said like, “Okay, well, it’s easy for a regular computer to multiply something once, but what if I make it multiply it a million times?” The answer is that’s still really, really easy because computers are really, really good at multiplying.

Andrew Fursman (00:46:07):
And so in practice, it’s going to be interesting to see how much time did it take for, say, ions to use actual large scale quantum computer to be able to tackle, say, a particular key length of a whatever, a Diffie-Hellman style encryption method. It’s still likely going to be that there’s some scaling approach that says that as you make those things more and more difficult, that it still takes longer and longer for the quantum computer.

Andrew Fursman (00:46:39):
But that might not be a very sensible way to be able to protect against quantum computers, because it might be the fact that quantum computers are so good at doing this compared to classical computers. But these encryption methods start to become kind of infeasible. Like it’s harder for a regular computer to do the encryption than it is for a quantum computer to do the decryption, in which case it still makes sense to try and find things that we talked earlier.

Andrew Fursman (00:47:07):
All encryption is, was recognizing that there are some things that computers can do in one direction much faster than they can do another direction. If there are things that are hard for a classical computer to do one direction and easy the other direction and quantum computers can’t do at all, for example, then that kind of a problem solved. And that’s what most people mean when they talk about post quantum encryption methods. They’re not talking about making the current encryption methods just that much harder. They’re actually talking about finding something where the crux of the new encryption method doesn’t align with the capabilities of proposed future quantum computer.

Preston Pysh (00:47:49):
Got it. And from what I’ve read and I didn’t read a whole bunch, but the little bit that I’ve read is the multidimensional elliptical curve generation is kind of what you just described in that it’s a different approach that would make it extremely difficult for quantum computing to crack. Is that true in the sense that in the way that I’m describing it there?

Andrew Fursman (00:48:11):
Yeah. In fact, the National Institute of Standards and Technology started, I think, with 69 submissions of potential new encryption methods for them to endorse. And I think it’s down to the third round and public review right now. It’s expected that within the next year or so, there will be the final release of these are the methods that we should all be moving toward in order to protect against these future problems. And actually, like you say, our current elliptic curve cryptography, which is a method that is favored now over RSA, it has a slightly different method where we talked about RSA is about multiplying prime numbers together is easy. But finding those prime numbers, again, it’s really hard.

Andrew Fursman (00:49:01):
Elliptic curve cryptography, which people might know about these, Diffie-Hellman is I think one of the ones that some people are familiar with. There actually are some, I want to say subtle changes, but they’re fairly complicated. And by that I just mean beyond my understanding. But they’ve got great names. I think the post-quantum supersingular isogeny Diffie-Hellman Protocol is something proposed by Microsoft in 2-16 and lots of work is being done within private industry. I went to school at the University of Waterloo and at the Institute for Quantum Computing in the Perimeter Institute. They’re doing lots of work on that method specifically.

Andrew Fursman (00:49:42):
But ultimately, I think when the initial standard for quantum resistant cryptography is released in 2022, they probably will have found a range of different methods, which, as far as we can see, are both the one way function still preserved for classical computers, and also that it doesn’t have this feature of what essentially comes down to the ability for something like Shor’s algorithm to be able to sort of unzip that information.

Andrew Fursman (00:50:13):
And that’s what we’re really looking for. And I suspect that that’s going to actually be particularly useful for some of these new methods. Although, I’m sorry, when I say these new methods, I mean new methods of encryption baked into cryptocurrencies.

Andrew Fursman (00:50:27):
But I think there’s also some clever stuff that’s being done. For example, I know that the P2PK and P2PKH addresses that were used really early on within Bitcoin, those are actually vulnerable to quantum attacks, although the P2PK addresses, I think, are just straight up vulnerable because they kind of expose the public key. And if you have the public key, that’s kind of like the equivalent of that really big number where you need to find the components that kind of go into making it up.

Andrew Fursman (00:51:04):
Once you have that exposed, then you can sort of attack it. I think that’s one of the real sort of interesting things, is have people made available the information that is itself susceptible to these quantum attacks, and in the case of Bitcoin, all of those original addresses are actually still are proposed to be vulnerable to near to term quantum computer.

Preston Pysh (00:51:28):
It’s interesting that you bring that up, because inside the Bitcoin community, there’s a best practice of not reusing the same public address, to use it one time and then to switch to another address. And this is really kind of what it’s getting at. Is there anything else that you would add on to that? I mean, I think you kind of already covered it, but is there anything else you would add to that?

Andrew Fursman (00:51:49):
Yeah, so that’s especially true for those P2PKH addresses where those addresses are vulnerable to quantum attacks once they’ve been used to spend Bitcoins, because up until the point when you spend from those addresses, the public keys aren’t made public. And so really, it’s only once the public key is available that you can tack it. So if you transfer your Bitcoins to a new P2PKH address, like you just described, then they shouldn’t be vulnerable to a quantum attack until you make that public key available.

Preston Pysh (00:52:30):
And just so people know, P2PKH, this is pays to public key hash is what that stands for versus just pays to public key without the H on the end with the hash. So just think of this as being run through another one way function is kind of probably the best way to think about that.

Andrew Fursman (00:52:48):
One of the things that’s interesting is I was just looking apparently fairly recently I saw a number of different consultancy reports on this, but it looks like somewhere between 20% and 25% of the Bitcoins currently in circulation are vulnerable to quantum attacks right now because they’re still sitting in these potentially vulnerable address styles as opposed to moving to the SegWit I think is the new method that sort of has already done a good job at making this a more challenging process.

Preston Pysh (00:53:21):
And that would be true if we had a quantum computer at 1,300 to 1,600 qubits because of the amount of processing power required to crack elliptical curve cryptography, is that correct?

Andrew Fursman (00:53:35):
Yeah-

Preston Pysh (00:53:35):
Or are you saying that that’s already in existence today?

Andrew Fursman (00:53:38):
It’s actually just more that there’s sort of the security by obscurity, which is the idea that if I don’t even know the large prime number, again, just using the RSA version of a one way function, but it’s more helpful than going into what an elliptic curve actually is. But if I don’t know your large prime number, then there’s nothing for me to find the factors of.

Andrew Fursman (00:54:03):
So that’s one of the ways of keeping these things secret. But, of course, people would probably prefer to say, even if you learn my public key, I still don’t want you to be able to figure out my private keys. And so because of this, I would like to move to another method. And those other methods can be the newer type addresses, which my understanding is that they’re already significantly better at dealing with this. And on top of that, I suspect that for the actual encryption, we’re going to find that some of these methods, especially the methods that are going to be endorsed by the National Institute of Standards and Technology, will just become integrated in.

Andrew Fursman (00:54:42):
It’s not that this isn’t a real issue, but the nice thing is because crypto currencies are not sort of created as one immutable form, instead, it’s sort of a process that a community of people agree on, the community can agree to change the way that these things work. And because, again, quantum computers aren’t magic, they’re not just about destroying cryptography. They’re about doing some things that classical computers do poorly well. So if we just change the hard thing to be outside of the realm of what quantum computers are good at, then that sort of solves the problem. And that’s probably the best practice, would be to both abandon those really old addresses that make public key available and then move to a method of encryption within Bitcoin that makes it so that even when quantum computers come in the future, even if they’re able to find your public key, that you’re still unable to be able to find the private keys that go with it.

Preston Pysh (00:55:37):
I’m kind of curious about the energy requirements on this. I know it’s kind of hard to have any idea what this might look like in 10 years from now as far as the technologies that kind of evolve and come out. But is this going to be something that’s very energy intensive, that only are at the Googles and the IBMs of the world, or is the idea that this is going to be not a huge energy demand?

Andrew Fursman (00:56:04):
Because of the fact that the first devices are likely to be pretty large, just physically speaking, they’re probably not going to be as efficient as you might imagine from sort of seeing one of these little quantum chips, because you might need a very large number of chips or ion traps or whatever else interacting.

Andrew Fursman (00:56:29):
But generally speaking, when you think about something like the energy requirements that we’ve all become familiar with from things like mining cryptocurrency, the idea of having millions of GP use or ASICs all sort of doing these hashing functions and taking up energy requirements in small countries, that’s not really the path that quantum computers are on. And in fact, I think the desirability is that there are some applications which might be amenable to both quantum devices and classical devices, but where quantum devices, one of the first things they might offer is significant decreases in the amount of energy required in order to get those answers.

Andrew Fursman (00:57:13):
And so you’ll see that a lot of people who are interested in quantum computers are actually interested in them to see whether very power hungry, large scale supercomputing tasks can be ported over to a quantum infrastructure, specifically because of the power save.

Preston Pysh (00:57:28):
Interesting. Okay, so for this question, I’m just kind of curious. Bitcoin, cryptocurrency stuff aside, what is something that in this space you’re most excited about or something that you’re looking at and just saying, “This is just going to be incredible to see it develop from here”?

Andrew Fursman (00:57:46):
Well, I kind of alluded to this at the very beginning, but I think that long before we’re using quantum computers to sort of rain on the parade of all people who love cryptography and crypto currencies, we’re going to be using them to do these really interesting practical things related to this emergence of chemistry from physics. And the reason that’s so interesting is because a lot of the ways that people have traditionally worked in material, I think sometimes referred to as material design, but which is traditionally actually been material discovery, is you kind of accidentally create Teflon and then you figure out what it’s for.

Andrew Fursman (00:58:26):
So there’s this kind of weird thing where I joke that sometimes drug discovery feels like go to the Amazon, lick a bunch of trees and report back which ones made you feel funny. That’s kind of the limitations that we have right now where there certainly are amazing companies and amazing researchers thinking about computer aided material design. But the reason that most of this work still happens in laboratories is because if we actually want to do experiments where we kind of ask the quantum world what the answer is to pouring this beaker into this bucket, easiest way that we can do that right now is just to literally do it.

Andrew Fursman (00:59:10):
It’s like a simulation by simulating it in the actual world. We think of the laboratory as being in some ways like a simulation environment. So that means that one researcher is capable of doing as many experiments as one researcher can do. One robot is capable of doing as many experiments as one robot can do. So you’ve probably seen those things where there’s an array of like 100 little pipettes that are sucking up liquid and then going and then pouring them into these little micro fluid storage areas. And these are essentially running 100 paralyzed experiments, which imagine the amazing benefit that you get if you’re a materials researcher and all of a sudden you can do 100 experiments when you used to be able to do one.

Andrew Fursman (00:59:56):
But we all know from using computers that computers aren’t about doing something where it’s like you try 100 things and you used to be able to try one thing. It’s more like you try 100 trillion things when you used to be able to do one thing.

Andrew Fursman (01:00:09):
So part of what I think quantum computers’ legacy, the early legacy of quantum devices are going to be is the ability to say you can combine all of the raw ingredients of reality into an infinity of forms. And we’ve explored such a small percentage of that. We kind of did all the easy ones, all the ones that kind of come about naturally. But there should be a near infinity of like exotic forms that we can produce.

Andrew Fursman (01:00:35):
And we found some of those by mistake. But now we should actually be able to say like I’m looking for a material has these properties, then simulate bajilions of materials and then be able to filter by saying which one of these is the stretchiest? Which one of these is the bounteous? You’re trying to build something like a car chassis. You would love something that’s stronger than steel but lighter than styrofoam. Does such a material exist? This is going to be a great way to be able to start doing that.

Andrew Fursman (01:01:07):
The way that I kind of hope that this rolls out is first you end up building these really small things like maybe catalysts, I usually say for the sake of being boring, that might help paint dry faster. So the most exciting thing about quantum computers is they’re going to help you watch paint dry, but for slightly less time. But that’s just the very beginning. Then it’s from building tiny catalysts that are very simple things that you toss in to make chemical reactions go faster or slower. Maybe you start being able to build new types of two dimensional materials, things that are kind of like sandwiches and stuff like graphite.

Andrew Fursman (01:01:42):
And eventually we hope that we can actually move into small molecules for drugs and drug design, and finally, really simulating down at a quantum mechanical level some of the biological systems to make it so that we can build better drugs and understand more how these things interact. I think all of that really gets well underway before we start being able to think about some of these more abstract uses of quantum computers, like the encryption stuff that we talked about, even though the encryption algorithm was written before many of the algorithms that are going to underpin the materials revolution.

Preston Pysh (01:02:20):
Fascinating stuff. What size of a computer are you talking? 1000 qubits for some of that basic just at a molecular level, like two molecules interacting? How many qubit computer are you looking at?

Andrew Fursman (01:02:33):
We are starting to do experiments right now where we have small non error corrected devices, so devices, kind of like you just described, that sort of in that from 20 to 100 qubit range. And part of what people are trying to understand is, are we capable of doing small enough calculations that if we run them a bunch of times, we can still kind of get the right answers because they’re sort of just quick and fast use of these devices.

Preston Pysh (01:03:03):
Like we ran it 100 times and 70% of the time this was the result, so that’s probably the actual result, is that what you’re saying?

Andrew Fursman (01:03:10):
Yeah, although it’s probably more like run it 1000 times and one time it gets you a really interesting result because that’s the sort of challenges that you’re working with. But a lot of what it really comes down to is that sort of this era of noisy, intermediate scale quantum devices where we sit right now. A lot of people think that we’re going to have to move into something where we have more like on the order of between 100 and 200 error corrected qubits. So that could be anywhere from best case scenario, you could find a very stable quantum bit that’s close to able to be its own lone qubit, or it could be 1000 or 10,000 times more than that, depending on what the overhead looks like.

Andrew Fursman (01:03:57):
But that kind of gives you a sense of if we’re thinking about 100 fault tolerant qubits or 200 fault tolerant qubits, that’s still pretty different than, say, the 10,000 might be needed in order to start to do interesting cryptography.

Preston Pysh (01:04:12):
Just so I’m understanding the math here, so 100 error proof qubit, is it about 100,000 qubit computer, is that right? Between 10,000 to 100,000?

Andrew Fursman (01:04:22):
Based on what we understand right now and the methods, some device has roughly those characteristics. That’s a pretty good sort of back of the napkin calculation.

Preston Pysh (01:04:32):
Okay, gotcha. I’ll tell you, Andrew, this has been super enlightening for me. This is fascinating stuff and I have a deep interest in the chemistry stuff, but I’m not so sure that my audience will. They’re definitely encryption people.

Preston Pysh (01:04:48):
This was excellent. This was awesome. And I just really appreciate you making time to come on the show. And for people who don’t know, I got introduced to you from Jeff Booth. I’m curious, how do you know Jeff?

Andrew Fursman (01:05:00):
Jeff’s a general man about town in Vancouver and probably Canada more broadly, so I don’t even know how I know Jeff. I just know that I know him.

Preston Pysh (01:05:10):
What a great guy, I’ll tell you. I’m a huge Jeff Booth fan. Andrew, thank you so much for making time. Real fast, give us a resource or something that you think would help out to somebody who’s curious about this field, maybe a book that does a great job, just kind of summarizing the high points and makes it accessible. And then also give us a hand off to where people can learn more about you.

Andrew Fursman (01:05:32):
Great. Well, first of all, I’d say no jokes, there’s a book called Quantum Computing for Babies. And what’s interesting about it is it absolutely does not assume that you know anything about quantum computers. It looks awesome on your coffee table. It shows that you’re a sophisticated individual who’s thinking about deep things, but it really does sort of start at kind of a fun place. And that’s, I think, a great thing just to dig into. If you want to have a book to get yourself going, that’s a fun thing to be able to pull up.

Andrew Fursman (01:06:05):
And maybe I would say the other thing is, you talked about your interest in chemistry. Maybe you can start imparting some of that to your audience because your audience should absolutely be observing the world of chemistry, because if chemistry is exploding because of quantum computing, then you know that it’s not long before encryption is sort of the next target out. I think of material science as being the canary in the coal mine for cryptography.

Preston Pysh (01:06:33):
I really like that point. That’s a great post or a milestone for people to kind of litmus test for them to pay attention to as this continues to progress, because like we said, that would be about let’s just call it 100,000 qubit computer that would be doing those calculations. And then pretty much next on deck is elliptical curve, at least the elliptical curve that’s being used for Bitcoin, which is the secp256k1 elliptical curve.

Preston Pysh (01:06:58):
If people want to learn more about you, tell them about your company or maybe a bio or something that they can read online.

Andrew Fursman (01:07:09):
Well, if anyone is interested in learning more about this, one of the things that I think is probably not to say that everything in quantum computing isn’t somewhat challenging, but one of the challenges that we’ve set for ourselves at 1Qbit is trying to make some nice infographics that we think does a great job of taking the best of what’s easy to communicate about quantum computers and making it as easy as possible, but no easier.

Andrew Fursman (01:07:32):
So if you go to 1qbit.com over the next little bit, we’re going to try and be taking people on a bit of a journey through these infographics that hopefully teach you what you need to know, that don’t take shortcuts, that are nice to say but inaccurate. And I feel like that’s really nice.

Andrew Fursman (01:07:49):
I also think that anyone who’s heard a little bit about quantum computers and kind of wants to check themselves, there is a researcher named Scott Aaronson who was involved with a web comic. And if you look at the talk, it’s basically a mom who comes and has a kind of conversation with her son saying, “Hey, I know you’re hearing a lot about quantum computers lately honey, and I think it’s time that we have the talk so I can help you understand much more about it.” If you’ve got five minutes and are interested in knowing what you think you know about quantum computers that might be wrong or some easy ways to start kind of digging into a little bit more meat, the talk by Scott Aaronson I think is great.

Preston Pysh (01:08:35):
Awesome. We’ll have that in the show notes. I looked up the book Quantum Computing for Babies, and I flipped through the digital pages there, and it looks hilarious and awesome all at the same time and …

Andrew Fursman (01:08:47):
Whurley and Chris will love that you’re promoting it.

Preston Pysh (01:08:51):
I think it’s hilarious and I love it. All right, Andrew, thanks so much for making time and coming on the show. This was really a pleasure.

Andrew Fursman (01:08:58):
Thanks so much for having me, Preston.

Preston Pysh (01:09:00):
Hey, so thanks for everybody listening to the show. If you enjoyed the conversation, be sure to subscribe to the show on whatever podcast app you’re using. We really appreciate that. And if you have time, leave us a review. Thanks for joining us this week. We’ll catch you next Wednesday.

Outro (01:09:14):
Thank you for listening to TIP. To access our show notes, courses, or forums, go to theinvestorspodcast.com. This show is for entertainment purposes only. Before making any decisions, consult a professional. The show is copyrighted by The Investor’s Podcast Network. Written permissions must be granted before syndication or before casting.