youtube

spotify

apple_podcasts

[00:00:10] Demetrick Ferguson: Hey everybody, welcome to another amazing interview. Another amazing hashing it out interview real hashing it out interview. Today we are joined with the team from Zama Kazimiye. And we're mostly going to be talking about fully homomorphic encryption, but we're going to go into those things. Today we're joined with Jason Delahaye and Mr. Morton.

[00:00:32] Jason Delabays: Hello, everyone.

[00:00:33] Demetrick Ferguson: Welcome.

[00:00:36] Demetrick Ferguson: Tell us. Tell us who you are. Give us a little bit of background. What? What? You know, what were you doing prior to crypto? When did the crypto freight train destroy your life? Like all of us. And you picked up the pieces now and you're working with Zama. So give us a little bit of your your guys background.

[00:00:55] Jason Delabays: So I'm Jason. I I've been working in a web development agency as a product manager for, like five years. And then in 2020, 19, 2020, during the DeFi summer, I got hooked into crypto first crypto Twitter. Then I read everything I could, I read Balaji, all the all the guys writing about crypto and definitely got hooked. And so I quit my job, and I started a project on a darknet, a layer two with a friend. We we tried different different stuff on the darknet, and but it didn't work for me as a founder. But I, I met Rand, the CEO of Zama, as an advisor of the company I was trying to launch. And he talked about his pH. And the relationship between pH and the blockchain and the product he wanted to build. And so I went to work for for Zama. And I've been there for one year now, working in the same team as a Morton here in the blockchain team.

[00:01:58] Demetrick Ferguson: Very nice. Very nice. Gordon, how about. How about you?

[00:02:02] Dr. Morten Dahl: So I've been in secure computation privacy for some years now, was actually working with Rand before, and then I went to do my own thing for a few years and then came back to join the blockchain team and tried to bring cryptography to the what we're building here.

[00:02:17] Demetrick Ferguson: Cool. So background and cryptography. And now you're kind of applying that here at Zama.

[00:02:23] Dr. Corey Petty: I love. I love that like we've gotten to the point of, passed the threshold of legitimacy where a good portion of people in strong academic backgrounds are able to understand how they can contribute and excited to join and to do so in order to kind of maybe set the stage on what Dharma is and how fully homomorphic encryption fits into it. Can we talk about a little bit like on the idea of what Zarma is and then what offerings it has, and then it's particularly kind of the blockchain offering that you all are focusing on and how fully homomorphic encryption fits into that.

[00:03:01] Jason Delabays: Yes, I'll take this one. So

[00:03:03] Jason Delabays: There is a company founded four years ago doing a research, in fact, fully homomorphic encryption and open source. We are an open source company. Everything we do is is a fully open source. You can find all we do on our GitHub and website. And so we are bringing a new cryptographic primitive for to to, to production historically this so what is F8 maybe first. So pH is very simple. It's the ability to do computation over encrypted data. So you take some data, you encrypt them so you cannot do anything. And usually what you can do is store it and nothing else. You can send the data, someone decrypt it and you can just store it, but you cannot work on the encrypted data. Fag is this magic thing where you can still do the computation while the the data are being encrypted. And so it's the often called the holy grail of cryptography because you don't need to trust anybody. You send your private data to a server. The server does this computation without reading your data, sends it back to you, and only you with your key can decrypt it. So this is a basic idea. And historically this has been very slow on computation heavy. And so there is a company dedicated to make it easily available for developers and improve the computation time and the scalability of the tech. That's the whole idea.

[00:04:35] Jessie Broke: I'm going to ask the question as an electrical engineer, what is the hardware infrastructure look like to support that sort of use case and also scaling that out?

[00:04:43] Jason Delabays: So right now. So our core product is a rust library called TFS. And right now it's optimized to run on CPU because it's the most accessible hardware. But we are working to make it available on GPU. It should come this year and then on FPGA. And we are also working with hardware accelerators to have Asics because of course Asics will get us a new order of magnitude in terms of scalability. So it works everywhere. But of course, it's slower on the, on CPU.

[00:05:18] Dr. Corey Petty: When you envision this on a blockchain, right? Like if you think about a blockchain network that has some level of fully homomorphic encryption as a part of its core protocol, meaning that like the computation that the blockchain is doing or the execution that's happening on that blockchain is done on encrypted data. What hardware resources are you expecting that individual nodes to have? Are they going to be home like, you know, gaming enthusiast hardware, good GPU, good graphics card, whatever. Is it going to be? Raspberry Pi? Is it going to be something that's closer to something that's run in a data center?

[00:05:57] Jason Delabays: Oh, no. It will be mostly hardware around in data center. So you won't be able to run an Fahy validator on your on your laptop, at least in the beginning. So this year, if you want to launch. So for blockchain we call this product an EVM. So it's an Ethereum virtual machine with this new module doing Fahy. This can only run on the very big CPU in data center. So next year on GPU on the right after probably Asics. So it's not a technology that will be like on everyone's computer, we guess for for the validator part.

[00:06:40] Dr. Corey Petty: Do you see? Is

[00:06:41] Dr. Corey Petty: That just a product of the applicability of the mathematics today? And there's a potential road based on some level of innovation that gets you to commodity hardware at some point. Or is it something that's like, kind of fundamentally constrained to large hardware?

[00:06:57] Jason Delabays: So our view on this is, you know, there are several different techniques to do to compute unencrypted data. You have MPC, multi-party computation zero knowledge proof and a SSC, and the F8 MPC is bound by latency, so you cannot increase the speed of light at some point. So this is like physically bound. In FHA we are bound by computation. And as you know, with the Moore's Law, even if it's not increasing as much as in the past, computation is still improving. So FHA is fundamentally a bet and the improvement of computation over the long term, over the long term. So I cannot tell what's what will happen in ten years, but we are sure that it will only go up in terms of scalability. So we hope I hope that one day it will be enough for for FHA to run on your smartphone. But we don't exactly know when. Maybe. Morton, if you are more insight on this probably answer. No.

[00:08:00] Dr. Morten Dahl: Oh, sorry. Yeah, my my audio broke off for a second. Yeah, I think so. As someone, we're also playing on different levels here. So one is the hardware. We are interested in Asics, as Jason is saying, and how that can improve performance GPUs as well. But we also have a research team that's looking at improving the fundamentals of the of the homomorphic encryption. So coming up with new algorithms, finding new ways of representing the data. Some of this the slowdowns is because there's a some blow up in the size of the data when you when you do the encryption. So in a typical encryption, the ciphertext is roughly the same size as the as the plaintext. But for moving encryption there is a small blow up that makes the ciphertext larger. So coming up with new algorithms for processing the data, representing it in smaller ways and so on. There are some of the research effort that we're also doing that that will bring inherent or kind of get around that, that doesn't that doesn't we're not inherently bound by, by the compute limitations of this. There is a research to be done at the algorithmic level.

[00:09:03] Dr. Corey Petty: Yeah, I tend to think about this with the experience I've seen so far within SSC or Zero Knowledge. And that is like early in the blockchain world. The academic representation of zero knowledge was was has been around for a long period of time. It was just wasn't very applicable. And I think a lot of the research within Zcash made it very made a strong incentive to then make it applicable once a few innovations hit and it became into this kind of threshold of of applicability with limited use, of course, they were able to do something which had a use case for people, which then spawned spurred a tremendous amount of research and subsequent innovation into an area of zero knowledge, applicable zero knowledge. And now we have this kind of massive ecosystem of research and applied engineering of zero knowledge within the blockchain ecosystem and broader for that matter, because it's a lot, a lot of people to look into. How can we use this? How can we grow it, how can we make it more efficient, etc.? Do you see this, this road into fully homomorphic encryption as a parallel to that, that storyline and what we're kind of at the I'd say beginnings of applicability with respect to fully homomorphic encryption.

[00:10:22] Jason Delabays: Yes, definitely. So, you know, our name is Zaman. I so first in 2020 was the company was founded. It was. We thought that I would be the first market for this with machine learning, etc.. But then we realized that there was only one industry in the world that desperately needs confidentiality and was already used to slow

[00:10:45] Jason Delabays: Stuff. And so that's why today blockchain is our the perfect go to market for fhea as a technology. And in four years we have already seen every every two years we do a ten x improvement in the in the computation time. Plus we abstract away all the complexity. And so it's very easy now for developers to, to use fat as a tool because they just need to be the normal developer. They don't need to be cryptographer in addition to being developer in order to manipulate these new concepts and cryptographic primitives. And we announced the EVM last year in ETH in Paris. And since then we are seeing like there are three companies today who have raised money using our tech and who launch a product and network for real in production. So the first one is called Phoenix. Phoenix. Another one is Inco Inco network. And a third, a third one is called the privacy. It's a, it's a little bit different distributed computation. But we are just seeing the first three startups building on top of our tech. And a lot more are coming this year. And so some of them will succeed a lot. Maybe not all. But we are sure that we are beginning of something. And also institutions are looking at this tech because with FHA, they can, for the first time use all the benefits of a public decentralized blockchain, plus having some confidential information. And, as you know, like in normal business the normal business transaction, you need to hide lots of data because that's the normal way to do stuff. And so with FHA,

[00:12:28] Jason Delabays: You can do it very easily.

[00:12:30] Jessie Broke: What's the biggest,

[00:12:32] Jessie Broke: Encrypted data use case for what you guys are trying to do? So, like, top of my mind. Like, I would immediately want to use it with healthcare information. But I guess what are some specific, I guess, sectors that you guys are seeing a lot of interest spark up from?

[00:12:47] Jason Delabays: The first one is finance. So finance is

[00:12:49] Jason Delabays: Very large. But the most obvious one is is finance and finance and tokenization to be more specific, because right now there is lots of assets that cannot be tokenized because they require at least a part of the information to be private and encrypted. And in addition to that, they need to do the computation on encrypted data. And so tokenization of of some assets in private equity, for example health care is another use case, but it's more web two for the web two world. And

[00:13:22] Jason Delabays: We will probably do it once we have the hardware accelerators. So, you know, probably one 1 to 2 years. But right now, in the short term, it's mostly finance.

[00:13:34] Demetrick Ferguson: So I like that you guys, you were in I before it was cool to pivot to I. You can you can keep that. Put that in your hip pocket. All right. So I kind of want to, like, unpack the how a little bit, if we could. Right. And try to find analogies to to bridge to the audience's, like, how powerful is it to compute on encrypted data, right. Like, I don't think I'm still trying to wrap my head around is like, wait, I'm receiving a bunch of encrypted stuff and I can compute with it. To me, I'm really trying to make that end meet because I'm like, If I'm receiving scramble, how do I how do I even know how they scrambled it to compute anything at all? Right. So I kind of need the layers peel back just a little bit if we could like put me in the one on one of Fahy, because to me it's not making sense in my brain. How are we doing this?

[00:14:32] Dr. Morten Dahl: I can take a stab with that. So the particular encryption scheme produces ciphertext where you can you can maybe see it as a, as translating the plaintext into a different

[00:14:47] Dr. Morten Dahl: Representation where it's hard to read, but you can still do the computation. I don't know if a good

[00:14:54] Dr. Morten Dahl: Analogy could be we have to be a bit careful here, but but because the analogy can be too simple, but imagine turning a number into a Roman numerals. That makes it hard to read, but you can potentially, potentially still do computations. Well, is there the difficulty or the difference here is that for morphic encryption, there is a private key that allows you to go backwards. So basically it allows you to go from this difficult representation that can still do computation using algorithms, but it's hard to make sense of. And back to a representation that actually makes sense.

[00:15:30] Demetrick Ferguson: Okay. That's understandable. I can do math with Roman numerals.

[00:15:36] Dr. Morten Dahl: I mean, maybe that's simplifying it too much, but there is a lot of research around how. So we're basing this stuff on well understood primitives. So hard problems in mathematics. And then on top

[00:15:49] Demetrick Ferguson: Well understood. Let's put that in quotations. This is

[00:15:53] Dr. Morten Dahl: Well, from the from

Episode hosts - Demetrick Fergurson, Corey Petty, Jessie Santiago

Dee, Corey, and Jessie host an insightful interview with Jason Delabay and Dr. Morten Dahl from Zama, focusing on fully homomorphic encryption.Delabay, with a background in web development and crypto, delves into his journey into cryptography after meeting Zama's CEO, exploring the potential of FHE in blockchain technology. Dr. Dahl, with expertise in secure computation, discusses the application of his cryptography background at Zama. The team elaborates on Zama's mission to make FHE easily accessible and scalable for developers, highlighting the difference between FHE and other cryptographic techniques like zero-knowledge proofs. They also discuss the challenges and future prospects of integrating FHE into blockchain, aiming to enhance privacy without compromising on trust and security.

<p>Hashing It Out is a highly-technical bi-weekly show that brings you the pioneering developers driving the Cypherpunk movement. In each show we take a deep dive through privacy, sovereignty, and network state technology with some of web3's most influential innovators.</p><section class="footnotes-container">&nbsp;</section>

Hashing It Out is a highly-technical bi-weekly show that brings you the pioneering developers driving the Cypherpunk movement. In each show we take a deep dive through privacy, sovereignty, and network state technology with some of web3's most influential innovators.

Dr. Corey Petty

Jessie Santiago

Demetrick Ferguson

Hashing It Out

<p>Dee, Corey, and Jessie host an insightful interview with Jason Delabay and Dr. Morten Dahl from Zama, focusing on fully homomorphic encryption.</p><p>Delabay, with a background in web development and crypto, delves into his journey into cryptography after meeting Zama's CEO, exploring the potential of FHE in blockchain technology. Dr. Dahl, with expertise in secure computation, discusses the application of his cryptography background at Zama.&nbsp;</p><p>The team elaborates on Zama's mission to make FHE easily accessible and scalable for developers, highlighting the difference between FHE and other cryptographic techniques like zero-knowledge proofs. They also discuss the challenges and future prospects of integrating FHE into blockchain, aiming to enhance privacy without compromising on trust and security.</p><section class="footnotes-container">&nbsp;</section>