S8 E11: Stepan Snigirev on Specter Wallet & DIY Hardware Wallet
Search this conversation
0:30Ad voiceoverThis episode is sponsored by Vaultoro. Keep on listening and you will find out more about how you can buy allocated gold when the Bitcoin bull run reaches its peak. This way, you don't have to deal with infinitely inflationary fiat or banks that freeze your account. Also, note that trading involves risks and the information presented is not financial advice. This episode is also sponsored by Wasabi Wallet. Go to wasabiwallet.io, download Wasabi for your OS and significantly boost your network level and transaction privacy.
1:39Vlad CosteaHello there and welcome to Season 8, Episode 11 of the Bitcoin Takeover podcast. This is a very special episode. I'm going to be talking to somebody who has spent many years in academia and has studied very interesting stuff with quantum physics and quantum computing, and has some lots of interesting ideas in regards to Bitcoin security. He's, I think, if I'm not mistaken, the lead developer of Specter Wallet, and is also one of the masterminds behind the Specter DIY project, which helps you build your own hardware wallet.
2:12His name is Stepan Snigirev, or Snigirev, depending on where you come from in the world. And he's a very brilliant man who doesn't get too much attention or credit, but among the developer circle, he's very much appreciated. So hello, Stepan.
2:28Stepan SnigirevHi, and thank you for having me. It's a pleasure to talk to you.
2:33Vlad CosteaLikewise. And I first want to ask you about your project with Specter. You told me while we were still doing the prep talk that it took you two years to write. And it's only at this point that you feel confident about releasing it out to the world and recommending it to people. How did you come up with the idea and how is it different from something like Electrum, that's very well known and has been around for so many years now?
2:58Stepan SnigirevSo if we talk about Specter Desktop, the software wallet, then it is, well, not two years old. It's more like one and a half maybe years old. So how we decided to write it: basically we already had Specter DIY, the hardware wallet that we were prototyping, and we wanted to have some kind of wallet support. And I looked at available options.
3:28And basically you could have something like Electrum and you could do something with Bitcoin Core, and that was the time when Andrew Chow actually started HWI, the Hardware Wallet Interface for all the hardware wallets to work with PSBT, and Bitcoin Core also was getting PSBT support. And Electrum back in the days was using their own transaction format that is not PSBT and it is not very well documented. So I was looking at it and thinking that, well, why don't we just take Bitcoin Core? And you can also run it as a full node, so you don't need to rely on any centralized services.
4:14And well, even though you could do the same with Electrum, it requires some extra steps. But what could be easier and simpler than just taking Bitcoin Core, like their reference implementation, and using that as a watch-only wallet? And it already had the support for descriptors, like the very initial support of descriptors. So we felt like we had everything that we need to build a nice UX around it. So the only problem of Bitcoin Core that we had back then and we still have now is that the user interface is not really friendly.
4:52So it is a great daemon, great implementation, very well tested, very well code reviewed. But as soon as you try to actually use it, like this Bitcoin QT, the interface, it kind of sucks.
5:04Especially if you try to do it with the hardware wallet, then it is even worse. So you can't really do it from the GUI. So you have to use command line here and there. So it is not very pleasant, let's say. So what we decided to do is the simplest thing ever. We take Bitcoin Core, we take HWI, the work that was already done, and we put them together, like glue them together, and put a nice web interface on top of it. So that's how Specter Desktop kind of appeared.
5:38And for us, it was very convenient because, yeah, you can have a browser-based interface that can display QR codes, scan QR codes. All the tech is known, so it's pretty nice and easy to develop. And then underneath you have functionality of Bitcoin Core that is like really well written and tested and reviewed. So we started that and then we started adding other hardware wallets because from the very beginning we were focusing on the multisig, just because, well, multisig security-wise is always better than single-sig, even though it is a little bit complicated. But I think we will get to it later.
6:20So we also saw that there are not many software wallets that properly support multisig. So even now, well, now there are more wallets that work with multisig, like Sparrow, for example, Gordian Wallet, or Electrum probably is the oldest one and the one that has the best multisig support. But back in the days, basically you had only options like Electrum, or nothing really else.
6:54Yeah, and somehow we quickly got support for all major hardware wallets, mostly because of HWI, because Andrew Chow did great work putting together this kind of translation tool between PSBT and custom transaction formats that are used by different hardware wallets. So from our side, again, it was very easy. You just take PSBT, you feed it into HWI, and you get all the communication with your hardware wallet. Yep, that's how it appeared. At first it was not very user-friendly in the sense that installation was difficult, so you had to do it through Python and pip.
7:44Then some time later we released the binaries, so you can just download it and run on your computer now. I think we added Windows support only after half a year or so, and it was really experimental. We didn't even test it, so we just tried and kind of put it together. And yeah, so you said that I'm a lead developer of Specter Desktop. It's not quite true anymore. So I started this project, but then like a year ago Ben Kaufman joined this project and started contributing, and now I think he is the main developer because, well, I do some pull requests here and there, and he is managing like everything, including the community, the bug fixes, issues, new functionality, everything.
8:36And I'm actually happy that I can focus on my side, on the hardware wallet side, and also help with Specter Desktop a bit. But it helps me a lot that it's not just me anymore, but there is a whole team working on that.
8:54Vlad CosteaSo I guess it's more correct to call you the creator as opposed to the lead developer.
9:00Stepan SnigirevOh yeah, yeah, something like that.
9:03Vlad CosteaAll right. So there seem to be a lot of wallets that work with PSBT nowadays. It's something that came to my mind while you were speaking. It seems to be very popular. Last year there were only two hardware wallets using it. I think the Coldcard was the first one, and then we had the Cobo Vault. And now we have at least three open source DIY projects that you can choose to use for yourself while making use of PSBT, which is interesting. And it makes me wonder, as I spoke with Peter Todd last year and he told me that there is nothing really special about PSBT, partially signed Bitcoin transactions, in terms of security.
9:45So is there something more to it? What is so special about PSBT?
9:49Stepan SnigirevWell, it's a BIP. And if it's a BIP, it is a standard. So we have these Bitcoin Improvement Proposals, right? And when something gets in there and gets adopted by the community, it becomes a standard, even though it is nothing special. So like Trezor is using protobuf, Ledger is using their own APDU commands. And internally you just need some kind of way to encode the transaction such that the other side can understand that, so like the hardware wallet. And the great thing about PSBT is not that it is a great standard or whatever, but that it was actually finally standardized and brought into the community and became adopted by many wallets. So it's really more about standardization, not about the thing that it is somehow better than any other implementation. And right now we have plenty of wallets that support it.
10:51So Wasabi, like on the software side, Wasabi I think started almost the first ones. Bitcoin Core, obviously. Electrum switched to PSBT. BlueWallet is using PSBT. Sparrow. What else? I think that most of the software wallets use PSBT now, because then in order to add support for all the hardware wallets, you just need to feed this PSBT into HWI, for example. And as far as I understand, many wallets do exactly that. Either they use directly HWI, or they use a similar implementation that converts from PSBT to like Trezor format or whatever. And if your hardware wallet supports natively PSBT, then it becomes easier. You don't need this additional translation layer. So Coldcards, Cobo, Specter, Bowser from Ben, SeedSigner, all these devices support PSBT.
11:54You can even use your old Android phone and install Electrum on that phone and make it air-gapped, and also Electrum as a signer uses PSBT. Yeah. So the main thing here is finally we have a standard that everyone can implement and then with minimal effort get integrated in the whole ecosystem.
12:17Vlad CosteaThat's interesting. And I like how you foreshadowed my next question, as I was about to ask you about the commercially available hardware wallets that you can build yourself. Because right now you can build yourself a Trezor, you can build from parts a BitBox02, and I think, and I'm not sure, if you can build a Coldcard. I think you can build it, but you cannot sell it. But why would users build a Specter DIY as opposed to one of these?
12:48Stepan SnigirevOh well, so the nice thing about do-it-yourself wallets is that, well, it's tricky. It's like a two-sided coin, right? So on one side you really need to know what you are doing, especially if you are writing your own hardware wallet like Ben or SeedSigner are doing. They just take the library and they write from scratch the whole thing. And then here it is a little bit tricky when you get to transaction verification, change address verification, secret storage and blah blah blah. But that's more on the developer side. For normal people, what is the trade-off?
13:29So if you are building a hardware wallet yourself, then you get off-the-shelf components, so you avoid supply chain attack. At least nobody knows that you're building a hardware wallet. Then you have to flash it yourself. This is the tricky part. So if your computer is compromised at the moment when you are uploading the firmware to the hardware wallet, then your hardware wallet can become malicious as well. So this is a really dangerous part. So if you're building a hardware wallet, keep in mind that actually it ends up to the security of your laptop at the point when you're uploading the firmware.
14:12And in case of other hardware wallets, like normal hardware wallets like Trezor, Ledger and so on, you disclose your address, that you are buying the Bitcoin hardware wallet, so this means that you become a target for some kind of phishing attacks and things like that. Then you also trust that they, at their factory, are uploading correct firmware. So you're trusting that the employees of the company are trustworthy and then the whole firmware uploading process is fine. So it could happen at some point that some employee gets malicious and uploads malicious firmware to all the hardware wallets that he has access to.
14:56Well, obviously, these companies try very hard to avoid this kind of situations. But again, it's the question of your choice and your security model. Who do you trust more, the companies or your computer, that also could be potentially compromised? And this is also why we started by implementing multisig, because, well, you can mitigate many of these risks and many of these trade-offs if you combine multiple hardware wallets with different security models in your setup. So then you kind of exponentially increase the security and decrease the chances of the attack.
15:38Vlad CosteaRight, but my question was not necessarily about why users should build their own hardware wallet, but more like why should they build a Specter? Yeah, exactly. But if you're going to buy off-the-shelf parts from the internet, why build a Specter as opposed to a Trezor? What's the difference?
15:56Stepan SnigirevWell, I mean, to build your own Trezor, you actually need to go through the contacts with the manufacturer of PCBs, you need to solder things, you need to do plenty of extra steps. So you can't just buy a thing and flash the firmware there. So with Specter DIY and Bowser and SeedSigner, you actually get components that are off the shelf, like developer boards, and flash it there, and then there is nothing else really that is necessary. So in that sense, building your own Trezor might be a little bit problematic.
16:38You need to make more steps to get it assembled, so it is quite a challenge. So if we are talking about why they should pick Specter DIY against Bowser and SeedSigner, well, the question is more again about your security model. So just to cover a few features of all these wallets: so Bowser by Ben is based on the ESP32, the cheap, well, pretty cheap, I mean, inexpensive chip. I think it's like a few bucks. And the problem there is that...
17:22First, it is partially closed source. So, when you have this Bluetooth and Wi-Fi stack that is running on one core of the microcontroller, and then there is your code that is running on another one. So, you don't quite know what is running on the other part of the microcontroller. That is a bit scary. Then the other problem is that there are known vulnerabilities on this chip, how to actually attack it and extract the flash. And by the way, it's the same chip that is used in Jade, I think. Yeah, so that is also a little bit scary.
18:00And then another thing is that both these projects, SeedSigner and Bowser, they don't have a secure bootloader. So this means that anyone with a direct access, physical access to the device can read out all the information that is stored on those devices. And this is actually the main problem of the DIY projects, because this secure bootloader part takes a lot of work. So when we implemented that, it took us three, four months before we actually deployed this secure bootloader.
18:32So what it does, it locks the firmware on the microcontroller such that only correctly signed firmware by developers can be uploaded there, and it cannot be read out. So when you are connecting your hardware wallet to the computer, even though you shouldn't if it is air-gapped, right? But still, if you connect it, then you can read out everything from it if it doesn't have a secure bootloader. So that's probably the main difference. Otherwise, I think SeedSigner is only using amnesic mode. So basically, you need to enter your recovery phrase into the SeedSigner every time when you turn on the device.
19:13Bowser, I think, saves it on the chip. Specter DIY has multiple options, so you can run it either in amnesic mode or you can save it on the internal flash of the microcontroller, that is not quite secure. Or you can store it on the smart card, that is basically a secure element, secure chip. Depends on your particular configuration. Also, another difference is the price. So, Specter DIY is kind of expensive.
19:47So the discovery board that we are using is around 50 bucks. Then if you add a QR code scanner on top of that, that's another 40. Then if you add something like our extension board or some kind of card reader, then it will add another 20, I think. So in total it ends up in the order of 100, 150 or so. So it's not as affordable as, let's say, SeedSigner or Bowser wallet, because Bowser is like 20 bucks, SeedSigner is 20-40, I think, on that order of magnitude. So the question is also about the price.
20:30Vlad CosteaAll right, so what is the recommended stack that you must buy in order to build your own Specter DIY, and what is the price that is recommended for this?
20:42Stepan SnigirevSo in the GitHub repository we have a document called "shopping", I think, where we provide the links to where you can buy it. So the core part of the Specter DIY is the discovery board from STMicroelectronics. So it is a developer board that has a pretty nice screen, like four-inch screen, I think, and the microcontroller and everything that you need. So if you just buy this board, that is this first 50, then you already have a functional hardware wallet that can work either with USB connectivity or with SD cards, so air-gapped with SD card.
21:26Then if you add on top of that the QR code scanner, so we are using QR code scanner from Waveshare, so there is a particular model that we optimized for. It should work with other QR scanners as well, but would require some extra wiring. Then this is another 40, and then you can have the pretty convenient way to communicate with the host with QR codes. So you scan a QR code with this QR code scanner, then you sign on the hardware wallet and then you scan back the QR codes from the host.
22:01So similar to Cobo. Then another option is to add this smart card to it. So this is a bit tricky. We were trying to find a way to get it off the shelf. Unfortunately, we didn't find an easy way yet. So right now we developed an extension board for this discovery that puts in a single board both the QR code scanner, the battery and the smart card reader. And then you can use a smart card, you know, this plastic card similar to your government ID or banking card, to store your secret. Then you also cover this secure element area of hardware wallets.
22:45But right now the only way to get this extension board is from us. That's not great. But we are working on the support of normal smart card readers. So like what you can buy from Amazon or something. And then you could connect it to the discovery board and use that for secure storage as well. So right now, discovery board and QR code scanner, and then you are on the security level of something like Trezor, basically.
23:19Vlad CosteaYeah, that's interesting. So it's good to hear that it's relatively easy for you to acquire the parts that help you build your own. Maybe that it's not the most convenient to know that you have greater costs involved. So I suppose to some extent, it's more affordable to buy something basic like a Trezor, but at the same time, if you live under a regime which prevents you from buying Bitcoin products or you're not allowed to receive shipments from outside, you can say that you're building yourself a gaming console or something that you can also build with the general-purpose parts.
23:58And that's why I like the Bowser wallet, just because it has that extra plausible deniability. I think Ben is kind of geeky in this regard, because most people will not use Morse code to type in their PIN. But at the same time, I can see the value proposition because there are people who will not be able to build devices for themselves. They will not be able to buy them, so they will build them for themselves.
24:24Stepan SnigirevAnd I would say that another thing that we really try hard to maintain in Specter DIY is to give the user the choice of the security model. So as I said, you basically have many options how to store the keys. One is similar to SeedSigner, is amnesic mode. So whenever you turn off the device, it forgets the key. And then, well, no other wallets, like normal wallet, actually does it. So, right, Coldcard, Trezor, Ledger, they are all storing the keys somewhere, you know, like with different security models, but they are storing them.
25:02So you still can wipe the device every time, for example, if you are super paranoid. And I know a few people that do that, actually. But here it is like out-of-the-box functionality.
25:15Then you kind of need to either remember your recovery phrase or type it from your backup every time. So that's not very convenient. And then you also have other options for key storage. So you can store it in the microcontroller itself. This is like a Trezor security model. So also the chip that is not super secure, that can be hacked if you have direct access to it with some glitching or something. But yeah, you could do that. Another option that we have is you can encrypt the secret and put it into the SD card.
25:53And then this device will be able to load this secret from the SD card. But this makes the whole thing like kind of two-of-two, right? So you need both the device and the SD card that you put together and then you will get access to your secret. And yeah, the third one is storing it on the secure element, and then it comes more into the Coldcard security model or Cobo security model mode, where you actually have a secure chip that has some closed source parts.
26:28So these Java cards that we are using or these smart cards, they have an interesting, well, how to say, interesting architecture. So you can buy them off the shelf. So you just go to the online shop and order these cards, and they have secure chip in there, and you can actually write an application for the secure chip, and everything that we write for these cards is actually open source. So we have this Specter Java card repository where we put all the code, but it runs on top of proprietary closed source virtual machine that was developed by either Infineon or NXP or whatever card you use. But the thing here is, okay, we know that it is somehow secure, and we know that it is certified, and whatever it means, and we know that it is definitely better than any normal microcontroller that was built not for security but for performance.
27:30So we can use the security even though we don't know how exactly it is implemented and if there are bugs or not, but we can use it to protect our secrets better. And again, everything that we do on our side, we are in the same position as all our users, so we don't have access to the code, low-level code that is running on the secure elements. So, and the main thing here is, yeah, you need to decide yourself how exactly you want to store your secret.
28:06Vlad CosteaYeah, once again, it's funny that you foreshadowed my question, because I was about to ask you specifically about secure element chips. And on one hand, right now we have the Ledger approach, which means that you have one chip that's the most secure that you can get, but you cannot verify it. So you don't know what's going on in the background and Ledger doesn't know. That's the whole point. And on the other hand, you have off-the-shelf general-purpose security chips like the ATECC608 that you can find in projects like the Coldcard, BitBox02, the Passport.
28:41And I'm pretty sure that there are a few more. And it's a pretty popular chip. It's used only for one purpose. It can do a lot more, obviously. So it gets disabled for other uses. But at the same time, there are some trade-offs involved because it's still not 100% open source. It's just something that you can pick up and program yourself. And there's that project by Trezor that's called Tropic Square, and they're trying to build truly open source chips. And I hope that they succeed and they're going to turn the industry around and help us get a more transparent kind of security.
29:24And what is your take on this whole situation? When do you think that the trade-off is worth it? And when should you just accept that you will not have physical security and try to mitigate it with software solutions like a passphrase?
29:42Stepan SnigirevSo, yeah, secure elements are tricky, especially the whole security industry is a little bit too, like, security by obscurity is like the way how they approach all this stuff. So yeah, Ledger, as you said, is using the secure element, and Ledger actually knows and implements the code, and they signed the NDA. And from my experience with Ledger and what I know about their team, they're really, really good at hardware security. So the guys are actually from security industry, and they are also doing certifications and hacking and so on.
30:24So in that sense, I think that the fact that Ledger committed to their own implementation and they went through all this NDA and stuff is actually very good. Well, they will not try to open source anything, because I think that their security model is more like, okay, trust us that we know what we are doing. And I actually trust them that they know what they are doing in the sense of hardware security. So, and I would definitely recommend Ledger to people that are fine with the single-sig setups.
30:59But obviously, Bitcoin community as a whole prefers open source. And Trezor is like on the other, completely different side of the spectrum. So right now what they have is no hardware security whatsoever, so there are even known hacks how to extract the seed from a Trezor device. But the Tropic Square is really ambitious, but also really, really awesome product. So I talked to Pavol about their current state and like when they will be on the market and so on, and I will definitely integrate this chip into Specter DIY as soon as it is available.
31:40And from what I understand about the first step of their approach is that they basically want to build an open source version of this ECC, whatever the Coldcard is using, of this chip.
31:57So basically it will be just the key storage under the PIN code. And I think this is a very good first step. So, like, you don't overcomplicate the whole thing. So you start small. And then when you see the demand, you make more and more things. So I think that they have pretty good chances of succeeding and actually releasing the chip. As far as I understand, they already have prototypes on FPGAs. So now the question is to find the fab to manufacture this stuff. And yeah, hopefully in a year or so we will actually see them, like the first prototypes. That will be really, really awesome. And I think this is what Bitcoin as a community and like these companies that are working in Bitcoin can handle, because Bitcoin price is going crazy.
32:47So if you had like 100 bitcoins 10 years ago, you now have enough funds to actually manufacture the custom secure chip. So I think that maybe Trezor can handle that. So another interesting project to mention is Precursor by bunnie that was recently funded, and it is not quite a hardware wallet. It's more like a communication device that is also fully open source and FPGA-based, pretty expensive, for very paranoid people. But they are also trying to build an open source platform and secure, including some anti-tamper measures, and everything is also like flashed on the user side when you receive the device.
33:37So all these nice things of the open source. They have some problems with the components right now, but I think that they also have a very good chance of actually releasing the device, even though it will be a very niche device and pretty expensive, but still very interesting project to look at. Then regarding other available components, so this chip that is used by Coldcard, Passport and BitBox02. Well, one thing that worries me, it's built for things like printers and drills and other things.
34:20So like when you have a printer, for example, that has a cartridge, and you as a manufacturer of the printer wants to make sure that only genuine cartridges can be inserted in the printer, because this is how they make the money, right? And they put their chips like that, that's supposed to be like this key storage and authentication devices to make sure that only like valid cartridges are inserted in there. So it's kind of security chip, but it is not a highly security chip. So in that sense, from a hardware security point, I think that the chips that Ledger uses is more secure.
35:07And the chips that all other wallets use are kind of okay, but I think that Tropic Square definitely can take this market. There is another chip that we are looking at. It is an experiment by the company Maxim Integrated, the American company. So they released recently, like a year ago, a secure chip that is not certified but kind of secure. But the nice thing about that, it doesn't require NDA. So this means that it is the first ever chip that has some kind of hardware security, like anti-tamper and anti-glitching, all these kinds of security mechanisms that you can buy and program yourself, without signing any NDA, and you can open source the code as well.
35:54So these are like three interesting things that I would definitely look at: Maxim, Tropic Square and Precursor.
36:06Vlad CosteaYeah, you also mentioned bunnie, and I'm going to look into that after we finish the interview. I'm actually curious to see how that one works.
36:14Stepan SnigirevWell, that's not quite a hardware wallet, but yeah, it's really interesting approach for... well, they are trying to do this secure communication device. But I was seriously thinking about taking what they do, getting rid of all these communication chips, adding the camera and adding the Bitcoin crypto libraries in there, and then you can actually get a pretty, pretty nice hardware wallet.
36:46Ad voiceoverVaultoro, and that's V-A-U-L-T, like a gold vault, and O-R-O, oro, which is Spanish for gold, is a gold and Bitcoin exchange which offers instant swaps between hard money to over 31,000 customers from more than 95 countries. Vaultoro has offered Swiss privacy and security since 2015. Furthermore, the gold you purchase is your legal property, secured in your name, so even if something happens to Vaultoro, even liquidators could not touch your gold. If you want to become the custodian of your own gold bars, you can request to have them delivered to you, or simply trade them back to Bitcoin on the dip.
37:37Register for free in only 30 seconds and start trading only with hard money. Please keep in mind that all trading involves risks. This is not financial advice and you are responsible for your own decisions. When you are using Wasabi Wallet, your internet connection gets routed through the Tor network by default. This means that you get better privacy while using Bitcoin. Download it today at wasabiwallet.io.
38:02Vlad CosteaOkay, so you mentioned something interesting, that you would normally recommend to people using single-sig setups to use a Ledger. When should people use multisigs? Do you think that there is like a threshold for the amount of money that they hold, or a moment when they acquire enough expertise or understanding of how this works to start using multisigs?
38:35Stepan SnigirevYeah, you definitely need to understand how multisig works before you start using it. So my feeling is, for a normal person entering Bitcoin field, how you should handle that is probably you start with, as everyone, with storing money on the exchange, even though it is a terrible idea. We saw many exchanges shutting down and people losing the funds. So as soon as you at least understand how Bitcoin wallets work, you should definitely move to the software wallet. And then I think that adding hardware wallet to that should be as soon as possible. As soon as you have more than a few hundred euros, basically, you should definitely move it into a hardware wallet. And in that sense, I think Trezor and Ledger are fine. So I wouldn't recommend Coldcard for the single-sig, or like for not very experienced person in general, because it has a pretty bad UX. And in that sense, Trezor or Ledger Nano X are much better.
39:45And from the security perspective, like I'm always looking at hardware wallets from security, I think Ledger Nano X is currently the most secure hardware wallet in the market, because it is the only hardware wallet that has a secure element that controls the screen. The thing is that you need to verify what you are signing, and with all other hardware wallets, either they don't have a secure element altogether or they have a secure element only for key storage. But whenever you need to sign something or to display the transaction, this information goes through the normal microcontroller.
40:28And in that sense, Ledger Nano X, only X, S is different, has this functionality that the secure element is controlling the display. And this is really huge. And I don't know why they don't market it this way.
40:45They say Bluetooth security to go and blah blah blah, but instead you need to emphasize that, okay, this is actually when the secure element is controlling the display. This is like really huge. Yeah, so in that sense, that's the reason why I think Ledger Nano X is great. And then, as soon as you start using multisig, and you should start using multisig as soon as you have, well, basically significant amount of money on the storage. I would start using multisig with maybe 10k, but it's also different depending on the country, depending on how much you earn and so on.
41:24But I think that at some point, if you're really afraid to lose the funds, you should start using multisig. Because also, a few years ago, I did a research on the attacks on different hardware wallets, and what I found out is that over a year there were like more than 20 different attacks on different hardware wallets.
41:49And some of these attacks were even, I would call them, "fuck-ups". They were so terrible that the malicious laptop could steal your funds or lock your funds without actually direct access to the hardware. So, like, fully remote attack, and that is pretty bad. So if you think about that and understand that, okay, probably there are like dozens of attacks every year that can compromise your hardware wallet, how can I use them in a secure way then? So the thing is that the chance that there is an active attack, not fixed attack, on your hardware wallet, if it is one hardware wallet, at any time is pretty high.
42:39So probably there is an attack how you can hack it, either with direct access or remotely, but doesn't matter, there is an attack. But then if you combine multiple hardware wallets from multiple vendors, what is the chance that on two different hardware wallets there are two active attacks, like known to the malicious actor attacks, that can be exploited at the same time? It is like quadratically lower probability, right? So as soon as you add more and more implementations into your multisig setup, the chances that someone will be able to hack all your hardware wallets, all your signers at the same time, becomes negligible.
43:21The question is, it also increases the complexity of the backup and maintaining this nightmare. So I would say that if you have 1 million euros in Bitcoin, then you probably want to go to something like at least two of three.
43:40Vlad CosteaOkay, so what is your take on Shamir Secret Sharing? Because it's not supported by all hardware wallets. I think right now there are only two of them that do. But it's still interesting because it creates something like a multisig setup without being a multisig setup. And the benefit, I think, is that you don't pay higher on-chain fees. And until Taproot gets activated and you can use that, you also don't know which side of the transaction has spent or which parts of your backup were used for the spending.
44:14Stepan SnigirevSo I think that Shamir Secret Sharing is great for backing up the wallet. The problem with Shamir is currently, before Taproot and Schnorr are activated, you actually need to assemble all these shares on the same device. So if you are assuming that there is an attack on the particular device, particular manufacturer, then it is kind of a threat compared to multi-vendor multisig. Another problem: I recently looked into SLIP-39, the Shamir Secret Sharing standard by Trezor. And what I found that is really annoying is that current standard that Trezor and Cobo implement is not compatible with the BIP39, with the normal recovery phrase.
45:00So, I mean, my dream setup would be: I take my recovery phrase, my mnemonic, and I split it into Shamir shares. And then, like, if I want to combine it and put it on some kind of software wallet that supports BIP39, I can do this. And if I have access to device or software that supports Shamir Secret Sharing, I could do that as well. The problem is that you cannot do it right now. So either you use Shamir or you use BIP39. And this is the problem, because they are not compatible with each other. You need to make this stupid choice what to use: something that everyone supports or something that only two hardware wallets support.
45:42So this is the only thing that is really stopping me from implementing it into Specter DIY. So I'm still thinking how exactly to do that in a nice way, because generally Shamir Secret Sharing is a very good way to split your secrets and to secure your backups, but it's just the standard... I'm not very happy with the standard right now.
46:04Vlad CosteaYeah, plus you can do a Shamir and on top of it do a multisig.
46:10Stepan SnigirevWell, you can make multiple seeds and then you can split each of them, for example. Well, it becomes a nightmare in sense of storing these recovery phrases. I mean, you don't want to store recovery phrases or Shamir shares in the same place, so you need to distribute them somehow. So you need to get some safe places to put things. And I mean, I would never store anything like a Bitcoin key in my office, because many people have access there. At home, it is kind of okay, but it is just one place.
46:54And then deposit boxes, something like that. But then how many deposit boxes do I need? So Shamir on top of multisig, I think, is overkill and overcomplicating things. So I would say that either you do Shamir or you do multisig. Yeah.
47:13Vlad CosteaOkay. I've mentioned at the beginning of our interview that you have worked and have done academic research in the field of quantum computing. And I suppose that you're qualified to let us know what is the current state of development and how far we are from building machines that are able to decrypt SHA-256 and possibly break the cryptography of Bitcoin. Are we far away from that moment, or do you think it's ever possible?
47:43Stepan SnigirevEver possible? Yes. When is another question. So first, a small comment. SHA-256 is just a hashing algorithm that we use in different places, but it doesn't really add much security. So the main security instance is elliptic curve cryptography. So if we are talking about quantum computers breaking that, I see two possible approaches.
48:11Like, yeah, imagine I'm building a quantum computer. And yeah, I can easily imagine that because I did it. But anyways, so what I can try to do? So first, there are two algorithms that are really suited for breaking this kind of thing. So there is Grover's algorithm that breaks hashing functions and enables quantum mining, and then another one is Shor's algorithm that is breaking elliptic curve cryptography. So both are pretty difficult, and both require something like thousands of logical qubits and probably millions of operations on them, gates, quantum gates.
48:56So this is like the level that we need to achieve. So thousands of qubits, millions of operations. Current state-of-the-art quantum computers, for example by IBM and others, they operate with 5 to 10 qubits, and they're not very stable, I would say. So you can't do many operations on them, maybe on the order of 100. So we are at the level of 10 qubits and 100 gates. So there are like two or three orders of magnitude before we could tackle the problem of elliptic curves and hashing functions.
49:37So how fast it can develop? So basically what I observed in the industry is that when you build a new setup from scratch, it takes roughly like five years. So you start by researching what was done at the moment, you come up with ideas how to improve the whole setup, you start building it, because your old setup doesn't work anymore because, yeah, there are too many things to change. So you get a new setup, you start building it, you spend five years on building that.
50:08After that, you get something like order of magnitude improvement comparing to old five-year-old papers. So basically this means that every five years we can expect something like 10 times improvement on this number of qubits and so on. This means that basically we need at least two rounds, so like 10 years, in terms of qubits, and then we also need something like three, four rounds, so like 20 years, in terms of number of gates.
50:44And there will be problems on top, like all these physical qubits are not really great compared to logical qubits and blah blah blah. So I would say that at least 20 years we have before we can actually see something that could potentially break what we use today. So 20 years probably sounds a little bit scary, right? Not too much. But we will also see the development in quantum computing earlier, let's say quantum canaries or something like that. So, for example, there are many applications where we don't need these thousands of qubits and millions of gates, we actually need much less, let's say on the order of hundreds or so.
51:40And then we already have significant improvements in the lives. Neural networks will be trained much faster. We will get superconductivity stuff. We will start maybe doing quantum mining because for actually mining, you don't need to reverse the whole hashing function. You only need to find the solution that gets certain number of zeros in the beginning, and many other things. So you will see that quantum computers are coming, and the moment, what I know from my co-workers that work in the quantum field, they are really struggling just finding any practical application for the quantum computers, next-gen quantum computers that they're building right now. So I would say that we have around 20 years for sure. And then the question is how do we solve the problem that, okay, quantum computers are coming in 20 years. So what can we do about it?
52:46There is a whole huge research area, post-quantum cryptography, that could, well, bring us something reasonable. So right now many shitcoins are actually deploying these post-quantum cryptography algorithms, and the problem with them is that often these new algorithms that are supposed to be resistant against quantum computers, they are not resistant against classical computers. So like all these algorithms, many of these algorithms, very many, can be actually easily broken by a normal computer.
53:25And this is like why these particular algorithms shouldn't be rushed into Bitcoin in general. So altcoins, experimental, yeah, go for it, whatever. But for Bitcoin, we need something really reliable, and there are a few interesting algorithms that have nice anti-quantum features, let's say, so like post-quantum cryptographic algorithms that are also based on elliptic curves, and there are others that could be deployed. And we see that in principle nothing stops us from adding another signature scheme into Bitcoin. We are adding Schnorr signatures now, right? So, yeah, we already have experience with adding new signature schemes.
54:13So I don't see really a huge problem there. As soon as we have a very good and well-working post-quantum crypto that is suitable for Bitcoin, that can be developed even in 10 years, whatever, then we can start thinking about how to integrate it in Bitcoin. That's my take on it.
54:35Vlad CosteaOkay, and this might sound kind of like a stupid question, but right now we are in the process of transitioning from Elliptic Curve Digital Signature Algorithm to Schnorr. Do you think that this is going to help in any way with quantum resistance? Or is it just a useful exercise to integrate new signatures that in the future might become quantum resistant?
54:59Stepan SnigirevWell, I think that it has nothing to do with quantum resistance for sure. So, Schnorr signatures are very similar to ECDSA and they can't really protect from quantum computers, but it is a very good exercise on adding new signature schemes. So it's more like just testing how it could go on. But in general, I think that Schnorr signatures are really, really awesome. And I'm looking forward to them and I will rush into implementing Taproot and Schnorr as soon as it is there and we have kind of standards that we can implement.
55:40And it also helps a lot for hardware security, because some time ago I gave a talk somewhere about my fantasies about the perfect, well, not perfect, but next gen of the hardware wallet, and it actually requires Schnorr signatures. So what you can do, you can take a single piece of hardware and you can combine there different secure elements in different implementations. In one device you can combine multiple of them with different security models. For example, you can take one chip that is closed source, proprietary, running this smart card Java Card OS, for example, and use that.
56:20Plus, on top, you add something like a free and open source FPGA-based RISC-V chip that might be not so well tested, but also looks very promising, and use that for another key. And then you can have the third one that is something in between, like our Maxim chip or whatever, and it can have the third key.
56:50And then when you have these three secure chips in the same device, you split your secrets into three using, for example, Shamir Secret Sharing. And with Schnorr and Shamir Secret Sharing, you actually have more features. So if in normal Shamir Secret Sharing and normal ECDSA, you need to combine all the pieces together to get the key in one place, with Schnorr you can actually combine the signatures. So you don't need to combine the shares themselves. You can sign, kind of make shares of the signature on each of these devices, on each of these shares of the key, and then combine together the signature at the end.
57:33So this is like somehow similar to MuSig, but not quite. But this is a really, really interesting approach that I would be happy to see in the wild, where you can mitigate these trust issues to the particular secure element by adding more of different implementations.
57:57Vlad CosteaInteresting. And something else that I want to ask you about your academic background is: at which point did you decide that it's time to get into Bitcoin, and what was it that drew you into it? Because I suppose that it's easy to get sucked into all of this academic life and all of the hierarchies and all of that situation where you have to constantly do research and start taking guest lecturing positions and try to advance in your ranks and get acknowledgement from all the other peers. So what was it that made you give up on that and get into Bitcoin full time?
58:38Ad voiceoverHey, psst, hey. What's your plan for the next Bitcoin top? Unless you need the money to purchase something, you probably should not touch infinitely inflationary fiat. Check out Vaultoro and figure out to which extent hard money like gold and silver can help you preserve your purchasing power. You will be able to get back into Bitcoin as soon as the price hits a new bottom, and you will not be subjected to the arbitrary, inflation-driven volatility of fiat or fiat-backed coins. Obviously, this is not financial advice, and you should understand that all trading involves risks.
59:25Wasabi Wallet connects to your full Bitcoin node, and if you're not running one, it downloads block filters anonymously via Tor. In either case, you're getting excellent privacy. Download the software today at wasabiwallet.io.
59:53Stepan SnigirevWell, so there are two things. First, I was missing an immediate reward, let's say. So, basically, when you are doing some fundamental research, you don't even have a chance to see it applied in the wild. So, like, we are doing such abstract and, like, fundamental things that you have a chance that the work that is built on the work that is built on your work may become useful in the real world. But it is several years and steps ahead of what you are doing right now.
1:00:31So when you're building your new lab, doing the experiments, spending five years on all this stuff, then at the end you do something theoretically interesting, like proving something, whatever, and then at the outcome of this you get a paper. Okay, good if it is a paper in Nature or something. But it's just the paper. So it's not changing the world. It's okay, it's giving like more knowledge to humanity about how the world works and so on, but it's not applied anywhere.
1:01:05So that was something that I was really missing, and that was why I looked into Bitcoin. Well, I looked into Bitcoin also because I didn't trust the Russian ruble, and it was a saving, like the way to save my funds back then. But also I saw that all this cryptography and math and all these interesting, like, mathematical and cryptographic kind of tricks and tweaks, they are really awesome from an academic perspective, so you can do something interesting there. But you also see how it immediately applies to the real world, how it is changing the world. So Bitcoin changed the world. I don't think that there are any questions about that. And it is based on the research, on the cryptography, on the math and all this stuff. And yeah, it's really enjoyable to see that you can contribute and you can do something very interesting and it can be useful for people. So that was the first thing.
1:02:10And the second thing is, I didn't see any really bright future for myself in this field. So the thing is that I'm not particularly smart, especially when you go into really deep quantum physics stuff, and I was not super successful there. So there was no chance for me to get a top-tier lab, so like to lead the experiment that is like one of 10 top quantum labs, something like that. And if you don't get to this very small top of scientists, then you struggle with the funding, you struggle with the grants, with the equipment, with the place, with the labs, with the students, with all the resources. So basically, yeah, you need to be extraordinary to be successful in any academia work.
1:03:16Otherwise, you have to end up in the industry somehow. And yeah, Bitcoin felt just right for me. So yeah, a not very smart person can do interesting stuff. So I don't need to invent a new signature scheme, but I can do like hardware security and things. Yeah, now I feel like I kind of say that hardware security is easy, but sorry, I just wanted to say that, yeah, quantum physics is nice, but you need to be really, really top. And in the Bitcoin field, we have people like Adam Back, like Pieter Wuille, Andrew Poelstra, and other Blockstream guys that are really, really smart.
1:04:03And I don't speak the same language even, so they are on a completely different level. But still, on my level, I can do something useful. And this is really inspiring and really nice.
1:04:14Vlad CosteaYeah, I think I see you more like the Andrew Poelstra type of person. So you are a bit humble and you tend to lower yourself and your contributions, but you shouldn't. And it's also interesting that you presented academia as this place where it's all about competition and meritocracy, whereas in my experience, it was more about knowing which ass to kiss and finding the next professor who's about to retire, you know, going, taking part of his seminars and carrying his suitcase around and offering to do some grading on the papers that he receives from students. And when he retires, you're next in line.
1:04:54So, yeah. That was kind of my experience with the doctorate and those studies where I was...
1:05:00Stepan SnigirevSounds very different from my experience. So for me, it was always like, either you are the best and then you get all the perks, like all the funding and the labs and so on, or you are somewhere in the middle and then maybe you need to kiss some asses to get a position. But then you also don't have enough resources to do something like super nice.
1:05:23Vlad CosteaBut at the same time, there are two different layers to academia. There's teaching, and then there's research. And it's kind of sad when you get pressured to do both, because most of the times you compromise. And we haven't figured out a system to incentivize people to just focus on one side of it and just maybe allow somebody else to teach your research, or something like that.
1:05:47Stepan SnigirevI was also teaching when I was back in Moscow, and I actually find it pretty useful. No, yeah, it's lots of time that goes for the preparation and for teaching and so on. But the benefit is that you have a source of smart students. So when you are teaching the students, you can evaluate them, you can find smart ones, and then you can grab these smart guys into your lab. And this is like very valuable for the lab because all the labs are built on the shoulders of students, actually.
1:06:25So like postdocs and PIs and so on, they do mostly like freaking boring paperwork and ordering and all this stuff, and sometimes thinking about the direction of the experiment and kind of guiding the students what to do. But it is crucial that the students that build your lab are really nice and really good, and teaching actually helps a lot to have access to these smart students and to kind of grab them early before someone else grabs them.
1:06:54Vlad CosteaYeah, but then again, you did quantum physics and stuff like that, and I did philosophy. So I guess it's very, very different. So there isn't much research to be done. It's more like innovating in some regard. I think right now in philosophy the most popular ideas concern ethics and medical ethics and bioethics and stuff like that, and even animal rights and all of these ideas. I think these are the trendiest in this regard. And also social justice. There was a lot of it, and it was sickening to some point.
1:07:30Like, I got in academia in 2011, when I was a first year student, and I saw in about five to six years how we switched from "just speak your mind, and this is a place where we can debate anything" to making lists of stuff you cannot talk about.
1:07:53It was interesting. We're living interesting times. Anyway, let's get back to Bitcoin. So let's say that somebody buys today his first bitcoins, like a thousand dollars or something. What is going to be your recommendation for them to start securing their coins? Because obviously they should not be on exchanges. But what do you think should be the first step towards learning a healthy way of securing coins and possibly advancing towards something more advanced?
1:08:24Stepan SnigirevOkay, so here it depends on why you are buying Bitcoin. So if you are buying bitcoins just in the hope that you will get rich at some point, then probably the best way is not to store it yourself at all and find some kind of custodial or whatever, some fund that can store bitcoins for you, and you just have a piece of paper confirming that, okay, you bought 1,000 euros of Bitcoin. This is if you care only about the price. If you, like me, care about not trusting others to hold your funds, this is how I got into Bitcoin, actually.
1:09:03I was buying Bitcoin, and then soon after that, Russian ruble crashed by a factor of two, so I kind of saved all my money this way. And for me, it was crucial that, okay, I'm paranoid enough that I don't trust anyone else to store my funds.
1:09:24This is also why I started building a hardware wallet, because I don't trust anyone else, including like Trezor, Ledger and so on. So I decided, okay, I will better make something by myself. Yeah, so it's definitely not a recommended way for normal people. But let's say you're not very technical, but you don't want anyone else to be in control of your money. Then I think that even with 1,000 euros you probably want to look into some kind of cheap hardware wallets, maybe Trezor T or something, or Ledger Nano X, or even, well, Cobo... Wait, so how much do they cost actually? Around hundred, all of them, right?
1:10:09Vlad CosteaYeah, I think they're kind of the same price.
1:10:11Stepan SnigirevOkay, so 100 for 1,000 euros is probably too much. So what I was doing when I started, I actually had my old Android phone that, I was very lucky, had the broken wireless module. So I wasn't able to connect to the Wi-Fi with it. So I installed Electrum there, and I used Electrum as a signer, so as an air-gapped signer right away. Because basically what scares me the most is to store my keys, my Bitcoin keys and other keys, on the laptop where I browse the internet, kind of browsing internet and stuff like that. Because I know it will be compromised, it will be eventually compromised, because yeah, there are plenty of zero days, one days, vulnerabilities out there, and it's pretty easy to get something bad on your computer, let's say. So in that sense, I think either a cheap hardware wallet or even an air-gapped phone would work well.
1:11:20It probably is better even to use your phone instead of your laptop, because mobile operation systems have a better isolation between the apps. So if you are storing your bitcoins on your phone, either Android or iOS, it's probably safer comparing to your laptop, no matter what kind of laptop is it.
1:11:40Yeah, and as soon as your amount of funds reaches something like maybe 10k or something, I would say that, okay, if you are holding N amount of money, then you probably want to dedicate something like 1% to the security of this money. So if you hold 1,000 euros, then basically you don't have many choices. Doesn't really make sense to use a hardware wallet or order a hardware wallet, because you will spend 10% of your money on that. In that sense, in that case, a normal app would be nice. So if you are not paying with Bitcoin, well, what mobile wallets I can recommend that are non-custodial?
1:12:32I personally use Phoenix. So Phoenix from ACINQ, this Lightning-only wallet that is pretty interesting and supports both payment to Bitcoin and Lightning addresses at the same time. And they can merge this functionality really nicely. So that is a pretty good option, I would say. Electrum, I don't quite like the user interface there, on mobile especially. Then what else? BlueWallet. BlueWallet, I had some problems with that on the phone that is really, really cheap phone. So like the user interactions in general was a little bit laggy, so I don't know.
1:13:23Yeah, for 1,000 euros, I think Phoenix is fine, Green Wallet is fine, something like that. I didn't try many software wallets. Then when you go to something like 10k or so, I think that it's definitely worth getting something like Ledger Nano X or something like that. Trezor T is also fine. And I think they... Well, the thing is that the hardware wallets... Ledger has really great hardware. If you are worried that someone will break into your house and steal your hardware wallet and try to hack into it, then Ledger is your choice.
1:14:05If you don't care about that so much, but you care more about Bitcoin logic and proper verification of the transactions and in the future integration into multisig, then Trezor T is much better. So, like, there are two different companies with completely different kind of pros and cons. One is really hardware security focused, and another one is really like making Bitcoin standards, verification, open source and everything like that. So I don't know what to choose. I mean, it would be perfect if they could work on something together, but it will probably never happen because they don't quite like each other.
1:14:43Yeah, so hardware wallets should come after you reach a few thousand k. Then when to go to multisig is probably when you go to 100k or so, and then you definitely want multi-vendor multisig. And on the way you should definitely research and try to understand how all this stuff works, because you need to know what is your recovery phrase, what exactly it is. Like maybe not on the how you derive keys from it, but at least understand that, okay, there is this words, they are a human-readable representation of your entropy, you can add this password, what exactly this password does.
1:15:29So like, why this password was introduced, how it affects your security, all these kinds of things. So when you're getting more and more comfortable about that, and when you start understanding, okay, why do I want to go to multisig? Because I don't trust a single vendor, for example, or because I want to do multisig with my friends, such that we together control the funds and one can spend the funds of our common wealth and so on, or common fund or something. Yeah, so every step should come with a reasoning why you are doing that.
1:16:15Vlad CosteaYeah, that's very similar to what Peter Todd told me last year. And actually, he was a fan of getting used iPhones and wiping them, because he said that the security on them is very good. So it's just a good idea to get a wiped iPhone. Actually, get a used one, wipe it yourself and update it with the latest operating system and run only a mobile wallet on it. He said that in terms of plausible deniability and privacy, it's kind of the best approach, because almost anyone, I mean, nobody would be surprised if you own an iPhone in your house.
1:16:54Stepan SnigirevI would be very scared of using anything that was used before. So, I mean, you don't know who did what to your phone. And for complex systems like phones and laptops, there are so many ways how you can hide malware there that I would probably try to avoid using used phones.
1:17:19Vlad CosteaYeah, but when you wipe it, unless there is some kind of hardware modification to the device...
1:17:25Stepan SnigirevYeah, yeah, hardware modification is pretty, pretty easy.
1:17:30Vlad CosteaYeah, but when you connect to Apple services, I think it checks the hardware and it tells you when something's not right.
1:17:37Stepan SnigirevYeah, it could be. I mean, I don't quite like the Apple products, even though I'm on MacBook right now, but how they lock everything. So I don't know. I personally would go with Android phone, but if he thinks that iPhone works better, then it's his choice. Yeah, so you need to make your own choice.
1:18:10Vlad CosteaYeah, it's kind of funny because this discussion about Android versus Apple is kind of the same as Trezor versus Ledger.
1:18:20Stepan SnigirevWell, I think that Ledger are not trying to lock everyone. I mean, you know that you can actually write an app for Ledger, right? So if you are not happy with the functionality of their Bitcoin wallet, you can write your own Bitcoin wallet for Ledger. And then you have all the benefits of this hardware security and NDA framework that they developed. And then on top, you put your own logic.
1:18:56Vlad CosteaWell, I think you can also use the Ledger with Wasabi, they have added an integration, and I'm pretty sure that Electrum finds it. But I don't think you can set it up. I think Ledger has made it so that you must connect to their service to be allowed to do the initial setup. And there is no, not that I'm aware of, there is no other way to install the firmware updates other than connecting to their servers.
1:19:22Stepan SnigirevOh, that might be true. Oh, yeah. Then in some sense, Ledger is locking everyone to their servers. But they also have, yeah, so in that sense, I understand. So both Apple and Ledger have a security model that, okay, guys, we know better, so you better connect to our servers and we do all the verification for you. So the approach, users are stupid and we are smarter. And in many cases, it is true, but it's also a little bit frustrating and annoying. And Trezor and Android is another approach, where do whatever you want with your device, it's your responsibility.
1:20:02Vlad CosteaYeah. There are people out there who advocate, like, most of the time on Twitter, and tell people that they should not use hardware wallets, because if they buy them, they can be subjected to supply chain attacks. And if they build them themselves, it's most likely that they have no idea what they're doing and there can be a lot of malevolent code which gets injected.
1:20:26And they're gonna be like, "Why don't you just set up a multisig with nothing but open source software like Bitcoin Core and Electrum, and you print the stuff, you print your QR codes with the private keys with your printer offline, or you also keep your backups on CDs and stuff like that." I think JW Weatherman has a system that's called Yeti Cold, and he promotes it. But at the same time, he scares away newbies from buying hardware wallets and tells them to not use them at all. Why do you think that using a hardware wallet is a good idea and most users should do it?
1:21:07Stepan SnigirevSo, supply chain attack, I see the point there. So, yeah, obviously when you receive the package from Ledger, Trezor or Coldcard or whatever, you disclose to everyone that you're working with Bitcoins. In that sense, DIY hardware wallets are better, because this is common kind of off-the-shelf components, developer boards, so you don't disclose that you are using it for Bitcoin. And my personal take is that DIY hardware wallets are better than laptops because the attack surface is smaller. Because when you look at your laptop, there are plenty of components there.
1:21:50And many of these components have even internal memory, can be tampered with, can be replaced, swapped, and so on. And also the whole operation system. I mean, it's written, some parts of this system is written without security in mind. So if you are writing the hardware wallet, you always think about what the attacker will do. And in case of Linux system especially, imagine you're writing a video player or some other thing that has nothing to do with security, but you can introduce a vulnerability that will enable the attacker to get into your computer, especially when you're thinking about the device drivers.
1:22:33So you can easily exploit the device driver. I think there was a talk at Black Hat some time ago, a couple years ago, that you can hack into the Linux system by abusing the driver of the smart card reader, because the Linux system is expecting that your smart card that you normally use for authentication shouldn't be malicious. And they wrote it in a way that basically there was a way to hack into your system using this kind of device. Also, like USB in general, like you are plugging something into your computer and it can tell the computer what is it.
1:23:13So it can tell that it is a keyboard or it is a mouse or whatever else. So there are plenty of ways how I can compromise your computer. And in that sense, if you are using the hardware wallet, then the code is really minimal. So just the binary of Ubuntu installation is 1.something gigabytes. The firmware of any hardware wallet is around 1 megabyte. So, and obviously it has some relations to the code as well. So basically this means that on your laptop, you have 1,000 times more code running, or not maybe running, but it is there that can be exploited.
1:23:57So it's like 1000 times more attack surface. And so this is, I think, the main problem. Also regarding the JW Weatherman and Yeti Cold, I don't think that the guy understands the standards. So he was trying to use Bitcoin Core with hardened derivation and he was struggling why we cannot support it in Specter Desktop. But damn, it is hardened derivation. I mean, if you know what is hardened derivation, then you know why we cannot freaking watch the funds. So in that sense, I wouldn't trust this particular guy because I don't think that he knows what he's doing.
1:24:36There are other standards, for example, the Glacier Protocol, I think, where you are supposed to use multiple laptops, also air-gapped, and also wiping them or using Tails, that is amnesic operating system that forgets everything after you turn it off. It is pretty difficult to kind of set up, to operate and so on. But I am sure that there are people that use that because this is how they, well, this is their security model. And I think that for normal people going with laptops and multisigs is not very convenient.
1:25:22And I think that there are plenty of ways how you can screw up your setup if you are using something complicated. So, as far as I know, there is more Bitcoins lost because of the user errors compared to the hacks. So, you should not only think about how you can be attacked, but you should also think how you can screw up yourself, how you can shoot yourself in the foot. Because this actually has more chances compared to someone getting into your hardware wallet or into your setup. So yeah, you also need to think about: do you feel comfortable about your setup, and can you easily reproduce it and recover from any kind of failure, like earthquake, burning the house or whatever else. So yeah, that's my take.
1:26:19Vlad CosteaYeah, I've actually had a funny experience with the Coldcard because I booted it up a year later and I realized that I have no idea how it's used. So I needed to watch a YouTube video to figure out something just because I forgot.
1:26:35Stepan SnigirevYes, with the Coldcard the user experience is not very nice. I mean, I still respect them a lot because they were the first movers in the air-gapped field. So, like, Coldcard, I think, is the first who introduced the air gap and I think it is a really nice approach. But user experience could be better. So the hardware is also, as far as I know, breaking sometimes accidentally and stop working. So buy two Coldcards if you want to go with the Coldcard. Another thing that worries me a lot, but it doesn't only apply to Coldcard, but also to our project, is when you have only few contributors. So, Coldcard's firmware is mostly written by Peter Gray, the CTO of the company. NVK is more like a spokesperson or something. And also in transparency, in the sense of their GitHub repository is closed for issues, I think. So if you try to open an issue there, you can't really.
1:27:51So this worries me a lot. And in that sense, also the fact that Passport took their code and started working on top of that also worries me a bit because it's basically based on the work of a single person. I mean, in our case, it's also pretty similar. So, like, the whole logic of Specter DIY is written by me. So I'm the main contributor there, like 90% of the logic is written by me. And this is also the security problem. And yeah, another underlying layer and the bootloader and stuff is written by another guy in our company who is actually a hardware security expert. And so like one guy writes one part and I review that and then I write another part and nobody really revisits it.
1:28:50This is very scary. So in that sense, Trezor and Ledger have much better reputation and history. And the size of the company also matters, like 15 engineers is better than two, I suppose. Yeah, but I think that I lost the track of what we were discussing.
1:29:12Vlad CosteaOh yeah, no worries. But I think that this situation can be mitigated if you get the funds to pay for security audits and people to look at your code.
1:29:23Stepan SnigirevYeah, either that or if you actually build something that people start building on top. So like Trezor, for example, let's take Trezor. The cryptography library, elliptic curve crypto library, is used in majority of hardware wallets. So Coldcards, Passports, KeepKey is just a fork of Trezor. Then there are plenty of hobby projects as well that build on top of that.
1:29:52So they really developed something that is used by the community and that is also therefore reviewed by the community and also attacked a lot. So I think that Trezor has the largest number of disclosed attacks and mitigations on them. So this is really interesting and nice. And I also know that they have a really big and nice team working on that, but also many eyes looking at that. So my hope is that our project will also evolve into something that people can build upon. So you can take the library and build on top of that and then as soon as more projects become more and more popular we will get bug reports, contributors and things like that.
1:30:43And yeah, so Bowser is using the Arduino library that we developed. SeedSigner is using our Python, MicroPython library that we developed. So we are getting there and I hope that it will continue, and this is kind of our way to grow the community and become more secure.
1:31:04Vlad CosteaThis is really counterintuitive because in every other industry, if somebody starts building on your grounds and on your research, it's kind of like, you know, competition. I didn't want to call it competition, but it is competition. But in open source, the fact that somebody forks your code and uses it for a different purpose and gives it another life and establishes a different user base only means that your code is getting more reviewed and getting more robust and tested for different use cases and gives you greater confidence that you're doing a great work, or else helps you improve on something that you developed and has issues that you did not figure out when you first wrote it.
1:31:50So I think that from this point of view, open source is just incredible. And I very much appreciate the people who engage in this type of development. But I'm not even sure how most of you make money because most of the times the donation system doesn't work. And most people, when they can get something for free, they get it and they don't think about the people who wrote the code and spent maybe a few years of their lives doing research and doing practical code writing. And I don't think there are many sponsors in this field yet, because even the core developers sometimes struggle to find sponsorships. I think the only way that you can make money is through consultancy, but I'll let you develop more on this idea.
1:32:37Stepan SnigirevYes, so when you are doing open source, it's hard to monetize that. So what we decided to do, as I said, is more like to build a platform and to use it mostly for, well, to get money from the enterprise, actually. So our focus is, okay, we have Specter Desktop, that is a great software Bitcoin wallet focused on multisig and hardware wallets.
1:33:07We have Specter DIY, that is our hardware wallet platform that works well with all this air gap and all these interesting entropy features and stuff. And then we build on top of that something that companies need. So, for example, if you are running some fund or even the custodian or any kind of exchange or whatever, so you are running a company and you are holding Bitcoins. Hardware wallets don't work for you anymore because hardware wallets are built, like normal single-user hardware wallets are built around the assumption that, okay, you have the user, the user provides you the PIN code and then he can do whatever he wants with the funds, and it doesn't work in the corporate setting. So what we do, we take our open source, MIT free platform and we build on top of that some custom functionality. For example, the additional authentication layer, that hardware wallet will not sign the transaction unless you provide certain authorizations from multiple people from multiple departments.
1:34:19And only if it verified that, okay, the security officer signed and then the operator signed, then it will sign the transaction. So it's more going into this fancy HSM-style mode and customizing it for more complex scenarios that are not really relevant for normal hodlers and Bitcoiners, but is very appreciated by the companies. And this is our kind of monetization strategy right now. Okay, first we build nice tools that work for the community and for Bitcoin. And then we add some bells and whistles for enterprises that can pay the bills for us, and then we can continue improving our open source free software and hardware, and the guys also get all the benefits of this interesting functionality.
1:35:18Vlad CosteaYeah. It's definitely something groundbreaking from this point of view. And I look forward to seeing how far this can get because to some extent, it feels experimental, but at the same time, you look at how Microsoft was making fun of Linux like 15 years ago, and you had Steve Ballmer calling it like a cancer or something. And now they are actively funding Linux projects because they figured out that they can use these open source operating systems for testing of features, and they can also integrate some of the functionalities.
1:35:55So it's really incredible to see that this has gotten to this point where it's inevitable and it's part of how software gets built. And you also have very competent, what should I call them, alternatives to software which costs a lot of money, like Office platforms, like Photoshop, video editing, everything that you can think about. There is an open source alternative which might not have the same nice user interface yet, but it has kind of the same functionalities.
1:36:33Stepan SnigirevAnd I think the nice thing is also that these open source tools, when you start using them and, okay, for example, let's take Blender, the 3D software, right? I start using it. I see that there is some feature missing and then I can make a pull request and add this feature. And if it is nicely done, it will be merged. And I can't imagine making a pull request to Photoshop or to Maya or whatever else, software that I pay like a few thousand euros per month or however much it costs. So it's also, free open source software motivates people to contribute back however they can. So either this is donation, this is not very reliable, but a different story is with pull requests and developers. So you can get your product better by giving it for free, and this is insane.
1:37:29Vlad CosteaOkay, so let's get back to the topic of Specter. And I'm going to ask you what is next for the Specter software wallet? And what is next for the DIY hardware project?
1:37:42Stepan SnigirevSo, recently we added Miniscript support to the hardware wallet. Yeah, this is like a side effect of this enterprise story. So now in principle we already support any kind of scripts, including timelocks and Lightning Network and whatever else, but our software wallet doesn't support it yet. So we need to do that, so this is the first. I think you should be able to do interesting Miniscript, like Bitcoin scripts, custom scripts, with the software and the hardware. That is what I'm really looking forward to.
1:38:20Then what else? Another thing that I'm looking forward to: so for the hardware, right now, if you want to use the secure element, the smart card, the only way is to get this extension board that we have.
1:38:46So I'm looking forward to a completely DIY setup where you can just plug this normal smart card reader to the device. Then our DIY option will be fully functional, like everything that you can do with the extension board, you can do with that as well. What else? So on both hardware and software sides, we are looking at Liquid support. That's something interesting, like the second layer. So Lightning is not on the roadmap yet, because we focus mostly on kind of hodling and cold storage and air gap, and it is extremely hard with Lightning obviously, basically impossible.
1:39:33Yeah, then another interesting thing is, I think it was two years ago when we were hanging out with the Trezor team and we were writing this CoinJoin standard for hardware wallets, how to verify that all the inputs and outputs, which inputs and outputs belong to you and which not.
1:39:56So I think that having something like automatic CoinJoin support in the hardware would be really nice. It could play nicely with something like either Wasabi or JoinMarket, whatever else is out there. The question is right now who will be the first CoinJoin server that will implement the standard. It is already in the SLIPs. So I think it's SatoshiLabs' improvement proposals, SLIP-19 and 20. That would be very cool. What else? Taproot is coming, right? And with the Taproot, there are plenty of things to work on both on software and on the hardware.
1:40:41For example, MuSig support is not even close to be standardized. And we really need to work hard in that direction, so that the standard is actually reasonable. And it will actually require collaboration between multiple hardware wallet vendors. So we need to stop fighting with each other, but actually collaborate because it is often interactive. And this means that we need to find a common way to pass data over to other hardware wallets to do this MuSig stuff. Yeah, what else? I think that's already enough.
1:41:20Vlad CosteaThat's a lot. Yeah, it is a lot. And I think your... I don't think that heading is a good verb, but you are coordinating quite an interesting project, and you have started quite an interesting movement because you found some parts that were missing in Bitcoin Core and in Electrum, and you built them under the kind of user interface that even people who are new into this space can understand and can use. I've seen a lot of good reviews of the Specter wallet. And I've even seen some enthusiasts who are building their own hardware wallets with the DIY specifications.
1:42:03I think it's really good. I'm proud of the fact that we get more open source in this space and a lot more interesting projects. And I don't know, I hope that in the coming year, we're gonna have maybe another interview and talk about some other integrations that you have made and other plans that you have, because this space evolves so fast. And last year, I've recorded an entire season about hardware wallets, and at the time, I was not aware that there is the Specter project.
1:42:37But I did interviews with Trezor, with Ledger, with KeepKey, with BitBox02, Cobo, Coldcard, also Peter Todd and Lazy Ninja, because Peter Todd is kind of a good person for security. He understands how it works and he can give some advice. And Lazy Ninja is known for hacking some hardware wallets. So he has found vulnerabilities in the BitBox02, Coldcard, and what else? I think he disclosed them all and he has earned some nice bounties, I hope, at least. What I wanted to say with this is that maybe the questions are the same, but the answers have changed. And this space evolves very, very fast.
1:43:19And there's a lot happening. And there are lots of maybe changes that get brought by user demand and the greater demand for security. And I think that the standard has been raised for the amount of security that users get in free wallets, which I think is incredible. And I'm grateful for your work. And I feel ashamed right now because I'm not sure if I have any more questions for you, except maybe how can people follow you? And how can they contribute to Specter or some other projects that you work on?
1:43:56Stepan SnigirevWell, I'm not sure that people will be really interested in following me because I don't tweet for several months. I mostly tweet only about the releases of Specter DIY and retweeting about the releases of Specter Desktop. But anyways, on Twitter, I am on Twitter, Stepan Snigirev is my handle. Then if you're interested more into like contributing, then we have a GitHub organization called Crypto Advance. This is the name of our company before rebranding, let's say. And it has the repositories for Specter Desktop, Specter DIY. And I think that right now, if you just type Specter Desktop into Google, you actually get into the GitHub repository.
1:44:49That is nice. And yeah, there are plenty of like issues, some documentations, and also we have a group on Telegram that is also written in the repository. And also people can just directly message me on Telegram.
1:45:11Vlad CosteaYou said something that made me chuckle a bit because you said that your name is Crypto until you're rebranding. Like, doesn't crypto stand for cryptography?
1:45:22Stepan SnigirevWell, yes. The name of the company is Crypto Advance. And after Specter, both Desktop and DIY, became popular, we actually registered the website specter.solutions. And so this is our main website where we put all the marketing stuff and also link to the downloads and things like that. So we are seriously thinking about changing the name, but it's pretty difficult from organizational perspective. And I mean, personally, I don't care how our company is named. I mean, what matters is only what we kind of publish and write and what we stand for. And in that sense, it doesn't matter if it's Crypto Advance, Specter Solutions or whatever else.
1:46:19Yeah, in that sense, the GitHub organization is called Crypto Advance. The website is specter.solutions. And yeah, on Twitter, Specter Wallet is also another handle that is managed by Moritz, our co-founder. Yeah.
1:46:41Vlad CosteaOkay, so thank you very much for taking your time, Stepan, and I look forward to next releases and possibly doing another interview in the future.
1:46:50Stepan SnigirevSounds good. Thank you for inviting and for asking all these interesting questions. So it is almost two hours already and it was a lot of fun. Thank you very much.
1:47:16Vlad CosteaYeah, likewise. Bye.
1:47:17Stepan SnigirevBye-bye.
1:47:18Ad voiceoverVaultoro has a 100% track record with fully audited and insured gold bullion that are secured in a top-tier tax-free Swiss vaulting facility. It also features a generous affiliate program, OTC trading, physical delivery and pickup, or trade back to Bitcoin in seconds. Register for free and check out the ways in which you can trade hard money and preserve your wealth.
1:47:46And if you want to help this show, sign up using the vaultoro.com/bitcointakeover link that you can also find in the description. Keep in mind that this is not financial advice. All trading involves risks and you are responsible for your own decisions. Wasabi Wallet's innovative CoinJoins will make your bitcoins more fungible. So if you accumulate more than 0.1 BTC, you can mix it with other users to remove all traces about their whereabouts. So it's like putting multiple fingerprints on your dollar bills and it becomes impossible to determine the last few owners of the money. Download Wasabi Wallet today and start coin joining!
1:47:46================================================================================ Production notes: Remote one-on-one interview, both speaking English (guest is a non-native speaker), with a short music intro and pre-recorded sponsor ad segments for Vaultoro and Wasabi Wallet at the start, mid-roll and end. Correction: anthropic/claude-opus-5.5 via NanoGPT in 7 windows with a whole-episode speaker/glossary plan, the show notes, the Qwen3-ASR full pass (142 chunks) and 66 Scribe v3 re-checks of doubtful spans; [unclear] resolve pass: 2 of 2 flagged paragraphs re-checked. Automatic checks: word count vs raw, timestamps, seam duplication. Fixes: Removed music/BF-WATCH TV and stray "Thank you"/"All right" hallucinations; labelled opening ads as Ad voiceover; Guest A → Stepan Snigirev; restored dropped text (Bitcoin QT sentence, Gordian Wallet multisig support, Cobo/Coldcard list, factory trust, "trust more" phrase, transaction verification) from second ASR/re-check; fixed Specter, Vaultoro, Andrew Chow, PSBT, protobuf, APDU, Coldcard, Cobo Vault, Trezor, SeedSigner, watch-only, BIP, Gordian; split long multisig paragraph at 00:06:20; Guest A → Stepan Snigirev; restored dropped speech from second ASR (Jade vulnerability, secure bootloader, wallet-to-computer, Waveshare, price, smart card, "But the thing here is", "better. And again...", Bitcoin community open source, "also doing certifications"); fixed Specter DIY, Bowser, air-gapped, Cobo, Coldcard, Trezor, Ledger, ATECC608, BitBox02, FPGA, bunnie, anti-tamper, Pavol, 4G→40; removed "Thank you." hallucinations and "unmanageable"; added split timestamps 00:27:30, 00:31:40; Restored dropped speech from Qwen/Scribe (Coldcard/Passport/BitBox02 chip, "printer, because this is how they make the money", Maxim "doesn't require NDA", "money on the storage", "Trezor and Cobo implement", "I could do that as well", "safe places", "I did it", Bitcoin wallets sentence); Vaultoro ad relabelled Ad voiceover, ad copy restored, Vlad's question split out; fixed Trezor, Ledger Nano X, Coldcard, Specter DIY, SLIP-39, Shamir Secret Sharing, SHA-256, Shor's/Grover's, IBM, gates, anti-tamper, malicious actor, multisig, bunnie; removed stray "you"; Guest A → Stepan Snigirev; Restored dropped text from second ASR/re-check (post-quantum/altcoins, "closed source, proprietary", Maxim chip, Vlad's full question, ad copy, academia passages); fixed Schnorr, ECDSA, MuSig, RISC-V, Vaultoro, Adam Back, Pieter Wuille, Andrew Poelstra, Russian ruble, "app/web"→lab, "kiss some asses"; split Vaultoro/Wasabi ads as Ad voiceover; removed "Thank you. Thank you. Bye." hallucination; relabeled Guest A as Stepan Snigirev; Speaker labels fixed (Guest A → Stepan Snigirev; 01:07:53 Vlad); fixed turn boundaries at 01:16:52/01:17:19/01:18:56/01:19:22; names/jargon: Trezor, Ledger (Nano X), Cobo, ACINQ, BlueWallet, air-gapped, NDA, ruble, JW Weatherman, multi-vendor multisig, "lock"; restored dropped words from second ASR/Scribe (1,000 euros price sentence, hardware wallet/"don't have many choices", "installed Electrum", "enable the attacker", "shouldn't be malicious", "Trezor and Android"); removed hallucinated "Thank you." and "Bye."; split long monologues; Guest A → Stepan Snigirev; fixed JW Weatherman/Yeti Cold, Specter, Coldcard, NVK, Trezor, Tails, Miniscript, MuSig, SLIP-19/20, SatoshiLabs, multisigs, funds; restored dropped words from second ASR (Passport, SeedSigner, reputation, MuSig support, etc.); removed "Thank you."; reassigned turn boundaries at 01:29:12, 01:36:33 and final "That's a lot."; Specter/Coldcard/Cobo/Moritz/Stepan Snigirev fixed; restored dropped Twitter-ha