S4 E10: Rodolfo Novak (nvk) on the Coldcard Wallet
When Rodolfo Novak (nvk) designed the Coldcard Wallet, he considered his own security needs and expectations. The device, which has currently evolved to the third generation, makes no compromises in terms of privacy and security.
Yet in the quest for security and privacy, the Coldcard wallet turned out to be the least friendly hardware wallet on the market. In its current phase, only seasoned bitcoiners who run their full nodes and are comfortable with third party wallets like Electrum and Wasabi are prepared to unlock the Coldcard’s full potential.
In this interview, Novak gets to have the final word on some of the statements that have been made throughout the season. He addresses Slush’s criticism of physical security, Peter Todd’s comments on PSBT, Ledger’s categorization of security element chips, KeepKey’s focus on securing a trading platform, and Cobo’s focus on the needs of Chinese miners.
Also, Rodolfo Novak provides details on his work on CK Bunker and how it improves on the existing Coldcard security design.
Enjoyed this interview with Rodolfo Novak? Here’s how you can help the show grow:
- leave a comment;
- follow @BTCTKVR on Twitter;
- join the Telegram channel for discussions;
- donate $1 on Patreon;
- tip via Tippin;
5:25 – Why use a hardware wallet?
6:33 – A lot of people have lost their bitcoins in the early days
7:45 – Building your own Coldcard
9:08 – Why is the Coldcard special?
12:10 – What is great about the Trezor?
12:25 – The problem with Trezor
12:43 – Why is the Ledger interesting?
13:00 – What keeps the Ledger from being great?
13:08 – The pros and cons of the KeepKey
13:30 – What is nice about the Cobo?
14:45 – On Slush’s criticism of physical security
16:14 – Whe massive spectrum of secure elements
21:15 – Can airport security hack your hardware wallet?
24:05 – Ledger article “Not All Chips Are Born Equal” and the definition of security element chips
27:10 – Is going Bitcoin-only more secure and profitable?
30:13 – Why altcoiners don’t care much about security
32:17 – Different types of bitcoiners have different kinds of security expectations
34:45 – Can hardware wallet manufacturers keep track of the devices they sell to dox you or steal your coins?
37:16 – User interface wallets that potentially track you
39:30 – Is a KeepKey that you run via Electrum as secure as a Trezor?
40:10 – Is PSBT really safe?
45:03 – Can you check your balance on a Coldcard?
46:50 – Coldcard’s opsec
49:00 – Does Coinkite store any data about customers to link the serial number with the invoice?
50:28 – Coinkite’s merchant processor is proprietary
51:17 – Rolling the dice for more randomness during seed phrase generation
53:11 – What is CK Bunker?
55:05 – Shamir Backup on Coldcard?
56:12 – What is next for the Coldcard?
58:42 – Is it safe to never upgrade your Coldcard and keep it as cold storage?
1:00:56 – Secure elements designed for Bitcoin
1:01:30 – Is the Coldcard comparable with a dedicated general-purpose computer?
“LXMI is a European Cryptocurrency exchange whose name is inspired by Lakshmi, the Hindu Goddess of Wealth, Good Fortune and Prosperity. It’s one of the regulated and legal Cryptocurrency exchange.
On LXMI you can buy bitcoins with most fiat currencies and you can also do the trading for top Altcoins.
They follow the “Not your keys not your bitcoins” philosophy with their integrated non-custodial wallet which helps you manage your own private keys. So if you’re into trading, then you don’t have to worry about having your Crypto frozen by whatever political decisions, since you’re empowered to hold and move your coins around whenever you wish.
It’s great to have new players like LXMI that respect your financial sovereignty.
LXMI is launching in 2020 for more information please check out – www.LXMI.IO/
If you’re not trading, it’s recommended to move your coins to a hardware wallet or some other form of cold storage, and in this episode, you’re about to find why.
Please keep in mind that this is just an ad for a sponsor of this show. It’s not meant to serve as financial advice, and you’re responsible to do your own research before buying anything and act according to your own decisions. Embrace your financial sovereignty with agency and precaution.
“Phemex is a Bitcoin exchange with derivative trading options which focuses on speed, robustness, and maximum uptime. Built by former Morgan Stanley executives, it manages o bring simple and accessible Bitcoin trading. In 2020, Phemex will also add S&P 500 stocks, stock indexes, FOREX, commodities, and more.
Sign up today at phemex.com/bonus and receive a bonus of up to $72.
Please keep in mind that this is just an ad for a sponsor of this show. It’s not meant to serve as financial advice, and you’re responsible to do your own research before buying anything and act according to your own decisions. Embrace your financial sovereignty with agency and precaution.”
Vlad Costea (00:00:36):
This might be the last episode of Season 4, the 10th episode, which is about the Coldcard wallet. And my guest today is Rodolfo Novak who gets the privilege of having the final word on this. I’m not sure if I’m going to do any more episodes from this season, but if I do it will be small players. So of all the big ones, Coldcard gets to have the final word, which is quite an interesting position because there is a lot of reacting to do to what other people have said and I have sent some questions to Rodolfo to synchronize and make sure that everything is fine to discuss. And I’m really excited about this. So hello Rodolfo.
Rodolfo Novak (nvk) (00:04:46):
Hey, thanks for having me.
Vlad Costea (00:04:49):
Yeah, I’m really happy that I got to convince all the big figures in the hardware wallet market to get on the show and not only mention what they promote about their products, but also what they think about the general state of the hardware wallet market, how they envision the future of private key security, and what they think about cold storage. My first question for you is pretty much the same one that I asked to everybody else who was on the show. Why would anyone use a hardware wallet as opposed to anything else like a brain wallet or paper wallet or some other DIY solutions?
Rodolfo Novak (nvk) (00:05:36):
For the average person, or even the advanced user, having a very simple electronic device that is capable of doing the Bitcoin operations you need and not being on a computer infested with viruses is a big deal. So even if you consider a Raspberry Pi, not connected to anything, with the right software in it—a hardware wallet—you’re already winning big-time. Your computer, especially nowadays, it’s just so complex, so full of stuff that it is impossible to really secure it unless you are a true expert. So having a device that’s made for it will remove a lot of the low hanging fruit regardless of the quality of that device.
Vlad Costea (00:06:33):
Yeah. Also a lot of people have lost their Bitcoins in the early days because they relied on their hard drives with the Core client or because they just didn’t think it would be valuable. So they didn’t care at the time. But I’m pretty sure—
Rodolfo Novak (nvk) (00:06:48):
Yeah, you’re getting for free all the tested and best practices of hardware wallets. We have all these competitors, all these vendors working on hardware wallets and everybody’s improving the whole process of doing Bitcoins safely. So when you get a hardware wallet, you’re inheriting all these best practices for free.
Vlad Costea (00:07:17):
Yeah. And there’s a lot of open source software. You can even choose which one you want to use. Slush even told me that there are over 50 clones of the Trezor on the Chinese market, which is interesting.
Rodolfo Novak (nvk) (00:07:30):
I wouldn’t use those. But yes, I take your point.
Vlad Costea (00:07:38):
It’s still interesting. And I remember writing an article for Bitcoin Magazine about how you published all the documentation to build your own Coldcard.
Rodolfo Novak (nvk) (00:07:46):
That’s right. Everything on Coldcard is available for you to buy off of any parts reseller. And the firmware is fully open source and the hardware schematics are also fully open source. So if you wanted to prove to yourself that what we say we do, we do, you could just rebuild it yourself and that should suffice your curiosity.
Vlad Costea (00:08:16):
And it’s not just about curiosity. It’s also about people in countries where they cannot order from Coinkite, which is based in Canada, maybe that they live in North Korea or something and they want to stay safe. And I don’t know how they can get the parts. Maybe they can build them themselves?
Rodolfo Novak (nvk) (00:08:36):
Exactly. Ultimately you give people a way of having the device even if you can’t sell it to them.
Vlad Costea (00:08:46):
Yeah. I think that’s the most useful part. There’s also of course the auditability of hardware and software because people can verify that you’re honest about your marketing and stuff like that. But speaking of this, why is the Coldcard special? Why do you think people should opt for the Coldcard as opposed to something else?
Rodolfo Novak (nvk) (00:09:11):
Coldcard was a product of our own needs. Neither of the main vendors were building something that would suffice our preferences and security requirements. So we’ve decided to make it ourselves. There was enough interest that we put it on the market. And people seem to like it, it’s a device that uses a secure element, but it’s open source. It has proper backup set ups, it works cold, it has a dice features, it has address explorer features. It’s got a ton of stuff that I’m not going to wait for vendors to come up with. So we’re just building the thing that we needed. I think that’s the best way to make a product.
Vlad Costea (00:10:08):
I agree. I remember when I first met you that you told me basically that you built the Coldcard for yourself and adapted it to the needs you thought that you might have according to your own mindset of security. And then it caught on as a commercial good that can be sold on the market. And I guess it’s successful. I see a lot of people who like it on Twitter.
Rodolfo Novak (nvk) (00:10:34):
Yeah. We sell a ton of them now. It’s a fairly well-priced device for the amount of features it gives you. The community knows that we keep on adding features and people request things that we often get it done for them. And other systems are also integrating it like Casa, [INAUDIBLE] and all that stuff. So it’s quite amazing it’s only been 2 years and a half now. I can’t remember when we launched it. There is a very robust ecosystem around it and I don’t see the competition still solving my needs, which is kind of interesting. We’re going to keep on going.
Vlad Costea (00:11:27):
Yeah. And before we move on and basically discuss everything about the hardware and the software of the Coldcard, let me ask you a question that has been around since Episode 1 and has become a tradition during the season to ask representatives of hardware wallet manufacturers to say something nice and something terrible or bad about their competition. And the four companies about which I’ll ask you are: Trezor, Ledger, KeepKey and Cobo. You can start with the Trezor and say something nice and something that you don’t like about it.
Rodolfo Novak (nvk) (00:12:09):
Trezor was the first to identify the need for an open source hardware wallet and they have good cryptography chops. The problem that I have with it is that one it has not evolved to understand security needs and two it was completely unsafe physically. So it’s a tricky place to be at as a product. Then for Ledger, Ledger is interesting because it is fairly secure. The architecture is a known type of architecture in the chip and PIN industry. But it is closed source and I can’t make myself use a closed source device. KeepKey I believe is a Trezor clone. I’m not super familiar with the guts of it—it’s also closed source—but they do have a beautiful industrial design and I love the little cube or whatever you want to call that shape of the KeepKey. Then I think you mentioned Cobo. They seem to have a very nice industrial design. I’m not super familiar with their architecture, to be able to talk about it.
Vlad Costea (00:13:40):
Yeah. It’s actually interesting because I spoke with Lixin of Cobo and he had nice words to say about the Coldcard and also he was constructive in his approach and I liked that. He works for a company which develops products for the Chinese miners so that’s a niche in itself.
Rodolfo Novak (nvk) (00:14:19):
It’s a whole other universe. China is a whole other place for hardware devices and people in China don’t seem to be—at least from the impression I get and I’m told this from others—people are not as concerned about security there for some reason. So it’s interesting.
Vlad Costea (00:14:40):
Yeah. I think I ordered the questions just the right way because I did not anticipate your answers, but the way I sent them to you, the first one is about what Slush said and then what Ledger said and they relate to the criticism that you made about their devices. And I’ll go first with Slush because we did Episode 8 of the podcast and he said that in his opinion he has experimented with a secure element and has some prototypes of the Trezor with a secure element chip, but he came to the conclusion that he cannot make a design that is 100% auditable. And in his opinion, it’s more important to be transparent and chasing physical security in his opinion is kind of an illusion because you have to renew the set up all the time and stuff like that.
Rodolfo Novak (nvk) (00:15:36):
Yeah. Let me just get right in there. So that’s super misleading, because, one, Trezor is not 100% auditable because unless you have an electron microscope, you cannot tell what the die of the chip is doing. So you could have had the device swapped, you could have had a bunch of attacks that are advanced, sure, but not fully auditable. So that’s a very bogeyman kind of argument. I don’t think it’s very constructive to the industry because it creates just FUD. And then there is a massive spectrum of secure elements—many flavors. You can have the Ledger flavor of secure elements where it’s a smarter secure element and you run your code inside. That provides a lot of security to you because you have all your code, everything you do, inside the secure element.
Rodolfo Novak (nvk) (00:16:34):
Unfortunately those are closed source, so you lose the openness and auditability so that doesn’t work for me. Then you have this other branch of secure elements, they’re extremely dumb. They’re essentially fixed function secure elements. All the code is already pre-done in the die of the secure element. So you can’t really change it. You can only set settings on it. And that’s the chip we use. It’s super simple chip. And we only use it to hold the seed. And we don’t actually even let the chip see the seed. What’s interesting about that is we use an open MCU, just like Trezor does, to do all the Bitcoin operations with our open source code and do the encryption of the seed—also open source—and then put that encrypted seed inside the secure element.
Rodolfo Novak (nvk) (00:17:41):
So even if the secure element had a back door or a flaw, which is very unlikely, the secure element still doesn’t see the seed. So you’d still have to break the other MCU, the other open chip, to get the decryption key for that seed. So in our case, you’d have to break two chips to be able to see that seed. And then there’s the fact that the 508—we are using the 608 now—but the 508 already has public data sheets. The 608 requires NDA, but any security researcher that contacts [INAUDIBLE] can get that NDA signed and then do research on it. And also there’s plenty of labs out there that have already decapped the 608 and have full pictures of the die.
Rodolfo Novak (nvk) (00:18:38):
It’s such a simple chip that making the argument that 608 is a closed source bogeyman backdoor is insane. It is effectively as open as the MCU used on Trezor and Coldcard. I think it’s important that we focus on physical security because a hardware wallet should be able to offer you more security than a Raspberry Pi. Trezor has no physical security. I understand that’s their idea and they rely fully on the 25th word. I’m not sure that was part of their marketing back then, but it became so after it got owned. Even if it’s a marketing shift, it’s still a very good idea.
Rodolfo Novak (nvk) (00:19:41):
It’s still a very good idea to use that 25th word and not rely fully on their hardware. But to me, a hardware wallet needs to withstand a minimum amount of attack. So if you have your hardware wallet sitting on your drawer or something, you need to be able to survive basic attacking. You should be able to survive a maid swapping the device, that you can show the device was swapped so you don’t get a spearfished for your PIN. There is a slew of attacks that could happen that the secure element can help you out with. Even if you have the 25th word, you can still be a spearfished for your PIN. That won’t help there.
Rodolfo Novak (nvk) (00:20:27):
So I don’t think you could get away with making a hardware wallet without a secure element anymore. Is a hardware wallet that is as basic as a Trezor still better than a computer with viruses? For sure. You should still use a hardware wallet. But I think that the bar there is too low. I think we need to do better. Right now we’re in the third version of Coldcard and there’s still a lot of stuff I want to do. There are other chips I’m experimenting with, and you’re probably going to see another version of it in a year or two, because it never ends. This is a new industry. We need to make things better. There is no need to stop.
Vlad Costea (00:21:12):
Yeah. Sometimes I think about this situation where you’ve passed through airport security checks and let’s say that you have your hardware wallet in the carry-on bag that you take with you into the airplane. And they make you take out all of the electronic devices and if you don’t, they will just see that there’s something there and ask you to take it out anyway. So they make you take out your laptop, your external battery, your phone and everything else. And I suppose that hardware wallets also have—either batteries or electromagnetic field—that gets emitted and will get discovered in the scan. So you’ll have to take it out. And I’m thinking sometimes can they have any sort of hacking devices that they use just like they have the ones for iPhones to unlock?
Rodolfo Novak (nvk) (00:22:08):
Yeah. In the case of Trezor, that’s 100% possible. For about $100, you can build a little box that you stick a Trezor in, and essentially you ROM dump it. So you take the whole memory of the main micro out and then you can try to break that later. But you can take that data out. That is not true for Coldcard and Ledger. We have very good defenses against that. Because we are using secure elements. You could still try to ROM dump. I don’t think it’s fully possible—nobody has provided a proof of concept on our open chip yet. It’s a similar family as the Trezor One, but it’s not the same. We also have more ECC defenses.
Rodolfo Novak (nvk) (00:23:03):
Anyways, the thing is that the seed’s still in the secure element, and you get a lot of those defenses as part of the package. Another thing too is if you’re crossing an airport, you probably want your 25th word, or not even take a hardware wallet with you, just take your seed or something. And then when you get to the other place you reconstitute your seed and then your wallet. But those are considerations. You have to walk through all these scenarios where you will have to surrender physical possession of the device either through absence, like the evil maid attack, or presence in a lawful situation where you’re not necessarily being beaten for your PIN.
Vlad Costea (00:24:00):
Yeah, that’s a fair point. But this leads me to a discussion about an article which was published by Ledger a couple of weeks ago, and it was titled, “Not All Chips Are Born Equal.” And in that article they distinguished between their security element chip, which is the ST33J2M0 . And they say it’s the true secure element and what they define the Coldcard chip as is a memory chip. What’s the difference?
Rodolfo Novak (nvk) (00:24:37):
You can get into a semantic argument about what is the definition of a secure element or not. So in their definition, a true secure element means that you’re running all the code inside of the secure element. So the 608 was our trick to get secure element security because the 608 has secure element security. So it has side channel attack defenses. It has a decapping defenses. It has a bunch of defenses. Are they not perfect, will somebody find a bug in them? That’s possible. But it’s the same for their chip too. If you have a secure element, it doesn’t mean that you’re infallible. It just means you have audited defenses against many types of attacks that are normally physical in nature.
Rodolfo Novak (nvk) (00:25:36):
So you have say, power differential analysis. So people can’t do a side channel read of you calculating things inside this chip. That is true for the to 608 that we use or for the Ledger chip. You have the [INAUDIBLE] cap defenses. It’s the same for us and for them. You have a tripping wires. You have a true random number generator, not just a random number generator. That means it’s an audited random number generator. And you use those to do some of the encryption and some of the communication between your secure element and other chips. So aside from the semantic discussion, the 608 that we use is quite robust. It’s a very good chip. So much so that we went from the 508, which was the previous version, to this new version that has a few more features. And we hope to use the newer version when it comes out that might have other defenses. At the end of the day it does provide security features and the manufacturer itself sells it as a secure chip. So aside from this semantic discussion, I think both of us have different approaches, but the idea is it’s still a secure element chip.
Vlad Costea (00:27:08):
Yeah. Also, the Coldcard was—I don’t want to say something mistaken—but I think you are the first hardware wallet manufacturer to release a Bitcoin-only product. I know that in the first generation you also supported Litecoin, but after this you dropped it and after you have promoted this Bitcoin-only approach, Trezor has released their Bitcoin-only firmware for the Model T and for the other one. And I think you have started this trend of having Bitcoin-only devices. And my two questions about this are, is this profitable and does this really help with security by reducing the attack surface?
Rodolfo Novak (nvk) (00:27:56):
It does. First, maintaining a massive codebase just means you have to do more testing and more bug fixing. So that means there’s more room for attacks. That just a reality of software. The more software you have, the more stuff you have to test. Again, it’s a little bit of marketing on the other devices because they’re still maintaining a massive codebase of shitcoin. The only difference is they’re making releases that don’t have the shitcoin, but that’s still a substantial amount of resources put in those or at least to review and test those. So just taking that stuff out doesn’t improve the security that much if you’re still putting all those resources on it.
Rodolfo Novak (nvk) (00:28:54):
In terms of profitability, we are a profitable player. We sell a lot of devices, but we are never going to be able to sell as many as, say, Ledger supporting all the shitcoins. So it’s an interesting sort of play. But we’re also not trying to be so consumer-friendly in a way, even though it is easy to use. We are focused on Bitcoin and actual hard security with open source. I don’t know where Trezor is in that spectrum. Because they do support all the shitcoins so they profit from that. But they also have to deal with all that stuff, so it’s not clear to me. But the reality is that Bitcoin is king and that’s where most of the HODLing is. So that’s what we’re going to focus on.
Vlad Costea (00:30:04):
Yeah. Also, I’m not sure if this question was part of the plan, but I have noticed that other shitcoiners, or altcoiners or call them whatever you want. They don’t care as much about security of their funds.
Rodolfo Novak (nvk) (00:30:20):
No, because they’re going to dump it in a week. They’re holding on an exchange to dump it. Aside from some old altcoins—maybe there are some larger HODLers and stuff. Reality is, most of them are just gonna pump and dump it on the exchanges. So I don’t really see the true value in supporting them either.
Vlad Costea (00:30:42):
I know this is a Bitcoin-only podcast, but I see Ethereum people on Twitter and none of them seem to be concerned about verifying the amounts that they’re holding in their light wallets with a full node.
Rodolfo Novak (nvk) (00:30:53):
Vlad Costea (00:31:25):
Yeah. I don’t want to get into it either, but it’s just the general remark that Bitcoiners tend to be most paranoid about security and also invest in hardware wallets, steel plate storage methods and all sorts of creative ways of storing their private keys. Whereas on other coins they just don’t care. They might as well use a light wallet like Coinomi or something on their phone.
Rodolfo Novak (nvk) (00:31:51):
The way I like to look at it is Bitcoin is hard money so Bitcoiners want hard security. The rest of the stuff is just noise, so they’re going to just use noisy solutions.
Vlad Costea (00:32:06):
Yeah, that’s right. But we established that this is some sort of reaction episode to what others have said during the season. And in Episode 7, Peter Todd told me that he doesn’t use a hardware wallet even though he has played with a few of them. He owns a Coldcard, he owns a Trezor, but doesn’t use them and has some kind of virtual machine set up where he uses an operating system called Qubes and he doesn’t trust anyone else with providing any kind of devices. And I suppose that there are different types of Bitcoiners who have different types of expectations and expertise in regards to security. When you first start out and want to get your Bitcoins out of Coinbase, you’re going to have a different expectation from somebody who is basically a cypherpunk and can build stuff on his own. So what is the category to which Coldcard is trying to appeal?
Rodolfo Novak (nvk) (00:33:10):
So we have a lot of security researchers that actually use Coldcard as their wallet. We are the paranoid hardware wallet. We like real security, code, and everything. There’s always going to be extremely advanced users like Todd who know exactly where to drill on the laptop and burn it after so that nobody can retrieve their stuff after. But the reality is, most people don’t have the right knowledge. And even for the people who have the right knowledge, they have other things to do in their life. Aside from attack Bitcoin, which is Peter Todd’s full-time job. It’s unrealistic either in terms of knowledge or in terms of time for all categories of users to essentially not use a hardware wallet. I mean, I’m fairly knowledgeable. I could run my own wallet on a laptop and I chose not to. I chose to use to create Coldcard for that because a computer is just too full of holes. And I don’t think that that’s changing anytime soon, even using Qubes, which is awesome. I think it’s unrealistic.
Vlad Costea (00:34:44):
I remembered this point from—and you’re going to laugh when I mentioned the name—but I did an interview sometime in December, but Trace Mayer told me that he doesn’t trust any kind of hardware wallet manufacturer because they might keep databases of their customers and they can possibly associate the serial number with the amount of Bitcoins that you’re holding and possibly your identity.
Rodolfo Novak (nvk) (00:35:15):
So a few different things there. Let’s just address them. One, it’s true that hardware manufacturers could keep track of invoice versus device sent. We don’t, but it’s not impossible. Now what a lot of people do, especially large HODLers that I know of, they will either get a friend, a lawyer, or anonymously order the hardware wallets from us and sent to them. Essentially, even if we were nefarious, we can’t keep track of that. Now, two, the most important part here is we don’t know your key. So it’s impossible for us to derive your public key to then know your UTXOs. It’s literally impossible. And on top of that, we have this dice feature so that you don’t have to trust our seed generation. You can create your own seed with dice. So again, it makes it impossible for us to know your keys. So there is no way, even if we knew your serial number—which we don’t—for us to figure out your Bitcoins, it’s just not possible.
Vlad Costea (00:36:37):
Yeah. And this is not possible, mostly because you don’t run servers to which users connect. You expect them to use the Electrum or Wasabi or some other third party wallet as opposed to Trezor, which has their own setup. And I’m not a fan of their approach to use a browser extension. I don’t think that’s really—
Rodolfo Novak (nvk) (00:36:58):
It’s terrible security.
Vlad Costea (00:37:00):
And I think Ledger has a client that you install on your computer, but still you connect their full node and their server and they know at all times how many Bitcoins you own and stuff like that.
Rodolfo Novak (nvk) (00:37:14):
So yes this is a massive issue and that’s the reason why essentially we will never make a wallet that connects to our servers. So in that case, yes, it is possible for Trezor or Ledger to—I’m not saying they would do it—but it is possible for them to have a serial number attached to your invoice to then identify the serial number through their system, and monitor those wallets. Because they do have servers talking to their wallets. Now I believe Ledger is adding a full node to their Ledger application. I think that’s a huge improvement. And I hope Trezor does the same. I think one of the biggest issues with Bitcoin is UTXO privacy and if you get a hardware wallet and you go and you plug it in to the computer and into these server-based wallets that are a little bit easier to use that Trezor provides, you’re doxing yourself. I don’t believe they’re nefarious or anything like that, but you don’t know who’s listening. So I think it’s a terrible idea to dox your UTXOs to a service. Especially one that has your personal information.
Vlad Costea (00:38:49):
And if you go with the KeepKey—which is owned by Shapeshift—they make you KYC nowadays.
Rodolfo Novak (nvk) (00:38:56):
That’s a different model. It’s a different product. And I think that those are even free nowadays. So I don’t think they’re even playing in the security space anymore. It’s a whole different category. You’re just increasing your security exchange a little bit, but it’s not really playing to the actual security of your HODL hardware wallet anymore.
Vlad Costea (00:39:28):
Yeah. I like to argue sometimes that you can buy that KeepKey for $20 or whatever they sell it for these days and run it through Electrum and it’s still kind of a Trezor.
Rodolfo Novak (nvk) (00:39:40):
No, because it’s closed source. And it has all the security holes. I mean, I wouldn’t.
Vlad Costea (00:39:54):
Yeah, it makes sense though, because it hasn’t received all the updates that the Trezor has.
Rodolfo Novak (nvk) (00:39:58):
Yeah. It’s just too stale to be safe to use.
Vlad Costea (00:40:06):
Yeah. And I know that, for example, I spoke with @BTChip of Ledger and I think he designed the original Ledger and is responsible for the current designs too. And he said that he appreciates Coldcard because they brought PSBT to hardware wallets and he thinks that the future models of the Ledger are going to have PSBT. But on the other hand, Peter Todd told me that he doesn’t think that having the air gap with PSBT really makes much of a difference because you still connect the device to the computer with the SD card. So it’s indirect, but it’s still a connection there.
Rodolfo Novak (nvk) (00:40:50):
Yeah. So he’s very wrong about this one. So the USB stack is a massive clusterfuck. People who work with USB know this, it is just insanely big and messy. And it’s only getting worse. When you look at USBC, you just want to cry. And then you are susceptible—aside from the USB stack—you’re also susceptible to USB cable attacks like the iPhone stuff. You can have a nefarious cable there doing stuff. Let me go back to microSD first. So the microSD route—SneakerNet as it’s been known as for many years—it’s not perfect, but it’s a very tiny codebase. And it has very limited read and write capabilities and it’s not executing anything.
Rodolfo Novak (nvk) (00:42:02):
So it’s a much smaller attack surface, while the USB stuff is monumental. Another huge gain you get is retrieval capability. Assume that all devices are hackable. Eventually, somehow. If you are remote and the wallet is not connected to something, you cannot retrieve information you manage to attack. With the microSD, you’d have to physically go there and attach it. But say you found a flaw in a hardware wallet and somehow it can be exploited via the USB, you can remotely take that data out. That’s not possible with the microSD. There is a reason why so many secure facilities and important stuff is completely cold, like a nuclear power plant. Just think about the extent that they had to go with Stuxnet to be able to get inside a power plant, a nuclear enrichment plant, to do that attack.
Rodolfo Novak (nvk) (00:43:10):
And even then, it was not for retrieval, they were just trying to mess with the plant. So the increase in attack complexity and retrieval is very big when something’s not connected. And then you also gain the UX of better security hygiene. So you can have your Coldcard in a in a very secure location and you go there and you only plug it to power. There’s no laptops, there’s no electronics with you, you’re not being tracked or anything like that. And then you do your transaction and then you take the microSD out and you go back to the location where you’re going to broadcast it. That whole UX security hygiene is a huge gain as well. So it’s too simplistic to just think that microSDs can be attacked as well, therefore, USB and microSD equal the same. It’s not. There’s a lot of parts to this.
Vlad Costea (00:44:27):
That’s fair. And I didn’t know this about microSD versus USB security, but it makes a lot of sense because there is so much happening with the USB device and there is only so much you can do with SD cards.
Rodolfo Novak (nvk) (00:44:41):
Exactly. Again, especially when you’re doing the microSD card with embedded devices like a Coldcard it is a very simplistic approach. So there’s very few attack vectors there.
Vlad Costea (00:45:03):
Yeah. And let me address something that I did not like about the Coldcard. And I’m not sure if it can be fixed without a wallet of its own, but you cannot check your balance to see how many Bitcoins you have directly from the device. You have to connect to a third party wallet like Electrum.
Rodolfo Novak (nvk) (00:45:22):
Yeah, that’s true for any hardware wallet. Because you need UTXO information from the blockchain to give balance. So any hardware wallet that is not online—that’s not either a full node or an SPV client—cannot see balance. It’s not that the hardware wallets don’t want to show you a balance, it’s just that it’s impossible without connecting to something.
Vlad Costea (00:45:52):
Okay. Yeah, it makes a lot of sense. I didn’t think about it in these terms, but I was thinking of watch-only setups and stuff like that.
Rodolfo Novak (nvk) (00:46:01):
Oh, you can do that. You can do watch-only with a Coldcard and a Samourai or with a BlueWallet or Electrum or Core or really any wallet that supports watch-only you can export to the xPub and just watch it. Mind you that if you’re doing watch-only, you are doxxing your UTXOs to the node. If the node is yours, great. If the node is not yours, you’re still doxxing your UTXOs.
Vlad Costea (00:46:33):
Yeah. And it’s terrible for your privacy, which in itself is the preceding step for security.
Rodolfo Novak (nvk) (00:46:41):
Yes. If people don’t know how much you have it’s much harder to know if it’s worth attacking you.
Vlad Costea (00:46:49):
Yeah. And I want to mention that we spoke a little after I published that review in Bitcoin Magazine and some of the remarks that I made, you disagreed with them, and this is your chance to punch me back in the face for saying that the physical robustness of the Coldcard is not as good as the one on the Ledger, even though that was my perception when I was touching them, I didn’t try to smash them. But also before you answer this, I also want to mention that in terms of design and opsec, the Coldcard does look like a calculator until you take a closer look and you notice that the screen is too small and it doesn’t have any kind of plus, minus, multiply, divide buttons for operations. So do you think you could disguise it to make it look more like a calculator?
Rodolfo Novak (nvk) (00:47:45):
So I don’t believe in security theater. Reality is any advanced attacker or any attacker that’s coming after a hardware wallet, it doesn’t matter what I make a Coldcard look like, they will know it’s a Coldcard, because we’re going to have pictures of the product in a website. So you’re not really gaining anything. Nobody is stupid enough to believe that that device is not your hardware wallet. So it’s just not worth it. So might as well not try to pretend to be what it’s not. In terms of robustibility, Coldcard is made of PC—which is a polycarbonate—it is a fairly strong plastic, it’s actually stronger that some of the vendors, but these devices are not considered rugged. So any of these devices will yield under a hammer.
Vlad Costea (00:48:43):
Yeah. If you tried to smash them.
Rodolfo Novak (nvk) (00:48:45):
Exactly. And that’s a good thing because it’s a great way of destroying them if you have to destroy that device against security sanitation.
Vlad Costea (00:49:00):
So when I get on the Coinkite website and I order a Coldcard, are you going to store any data about me and the delivery address and stuff like that?
Rodolfo Novak (nvk) (00:49:11):
So we don’t have any link between devices like the serial number of the bag or the device itself—and the invoice. They are just random in a pile and they get shipped out. Now due to tax law in Canada, we have to keep invoices for years. There is no choice on that, but there is no requirement for you to put your real name on an invoice or send to your real house. So everybody can be called Satoshi Nakamoto in our invoices and that’s pretty fair. Essentially there’s really no link. We have to keeping invoice information. That’s true I think for probably most vendors of anything.
Vlad Costea (00:50:08):
I think the last reply that I got to this question was about Shopify, which is a third party shopping processor. And Cobo said that they don’t store the data themselves. And that goes to Shopify, but they still get invoice data just like you do.
Rodolfo Novak (nvk) (00:50:28):
We’ve built our shopping cart system many years ago before Segwit, before everything. And it was built with a lot of privacy in mind. And the information is kept by us. There is no third party involved unless you pay your credit card. If you pay with your credit card then you’re essentially doxxing yourself to Stripe. But there is no link between the serial number and the customer. If you want to pay with Bitcoin you put Satoshi Nakamoto and you send to your P.O. Box and it’s none of our business. We can’t really know who you are. There is no requirement for us to know who you are either.
Vlad Costea (00:51:12):
I remember when I first set up the Coldcard that I really liked that part where you can generate more randomness to your seed phrase by rolling a dice and inputting the result from one to six. And I think I rolled the dice like 100 times on that day. Just for fun to see if the device says this is enough. Just move on with your life to do something. Is there a recommended number of dice rolls or what do you recommend for security?
Rodolfo Novak (nvk) (00:51:46):
So again, into that sort of paranoid mode, which is where I think everybody should be in regards to their wealth. And especially regarding cryptography—we don’t trust the device itself. Never will for key generation. So you can let the Coldcard generate your seed. The code is open source. You can see, technically it is sound but why trust it? So what we do is we offer you the option of throwing dice—six face dice—to add your own entropy. You can do the full 256 bits of entropy with 99 rolls, or you can do 50 rolls, so that’s like half, or you can do just a few rolls and add some entropy at least to trust-minimize your seed generation. It really is to the level of your paranoia.
Vlad Costea (00:53:10):
I was just scrolling your Twitter feed because I remember you posted about something very innovative for which Francis Pouliot also praised you, but I don’t remember the name of it.
Rodolfo Novak (nvk) (00:53:22):
So it’s CK-BUNKER.
Vlad Costea (00:53:29):
And what is that about and how is it related to the Coldcard?
Rodolfo Novak (nvk) (00:53:33):
I always wanted a roll your own co-sign service. So essentially you’re like your own BitGo without doxxing your coins to a co-sign service. We essentially developed a open source little project that that does that. You can use a Coldcard as a hardware security module. Co-signing transactions from either another Coldcard or from other hardware wallets, anything that supports PSBT—partially signed Bitcoin transactions. And it’s a bigger project, so it supports a lot of modes and you can have different users and it has a web UI where it’s over Tor, but essentially you put it running on a machine with a Coldcard on it and then you get a web service essentially over Tor for you to do your co-signing. So you can tell it only co-sign transactions of one Bitcoin per week. And that’s all it’s going to co-sign. So now you can, you can have a little bit more security with more parties involved, and that party is essentially you, so you’re not losing privacy.
Vlad Costea (00:55:00):
Also, does the Coldcard have stuff like Shamir backup and stuff that Trezor uses?
Rodolfo Novak (nvk) (00:55:08):
Yeah, so we have, since the launch, a secure backup method. So you can essentially encrypt your private key and other settings of the Coldcard into a microSD card. And then you get the 12 words to decrypt that micro SD card. That’s very nice for you to migrate between your Coldcard to another or for you to just create a safe backup that you can put somewhere else. We don’t have Shamir yet. We plan on doing Shamir shares at some point for backup as well. It’s just that there’s a lot of other features that are more pressing that people want. We actually had a release today of another firmware with more interesting features. We’re trying to sort of do the roadmap based on need as opposed to want.
Vlad Costea (00:56:12):
That’s a fair approach. But what is next for the Coldcard? How do you envision the next iteration of the wallet or some other firmware update that adds cool stuff to the existing models?
Rodolfo Novak (nvk) (00:56:26):
So just today we launched a scramble PIN key pads, so you can scramble it on the screen. You can now show a QR code when it’s over USB. Oh, this is a big one for me. When you create an Electrum or Bitcoin Core file from Coldcard, now you can choose the account number. So you can choose different derivation paths essentially. That’s nice because if you’re migrating from a Ledger or Trezor that had different accounts on their software it’s very easy enough for you to import that seed and create new Electrum files with those different account numbers. So just a click. There’s bug fixes to do. There is a lot more Multisig support. There is more integrations with other services and products.
Rodolfo Novak (nvk) (00:57:28):
There’s the Shamir stuff we want to do. The list is long and prosperous. It just never ends. I hope to one day do Lightning, and maybe some of the coin mixing from the device. We want to see how people use the HSM bunker feature and improve there as well. And then there is hardware itself. We were doing more experimenting with different secure elements. There is very interesting stuff that are not public yet, but coming out of different providers, the different chip vendors that they’re are giving us to try. We want to improve the UX on Coldcard. Definitely look into different industrial designs for it. This is what we want to do with our lives all day. It’s sort of like a never-ending pursuit here.
Vlad Costea (00:58:41):
I agree. And for every new security feature there are going to be many more, not necessarily bugs, but ways the exploit. So you have to fix all the time and keep up with the hackers, some of them ethical, some of them stealing your Bitcoins. But I wanted to ask you something. Let’s say that I buy the Coldcard today and I hold on to it for 5 years, and maybe use it as a cold storage way. You think it’s secure enough to not update the firmware and just leave it like that?
Rodolfo Novak (nvk) (00:59:17):
It’s impossible for me to know what the bugs or we will be. So it’s not possible to give a future-proof answer to that. If your physical storage is fairly safe and you are using your 25th word, long term, it’s pretty reasonable. It’s likely more reasonable than a computer. Is likely more reasonable than some of the other vendors. Coldcard Mk1, even though it has some exploits is still an order or two of magnitude more secure than a Trezor. So we just never know what kind of attacks will come. Again, this is a very young industry. We haven’t had secure elements made for Bitcoins. We’re still using generic stuff. So it seems important to keep pushing until this industry as a whole matures a bit. Maybe in 20-30 years, it will be more obvious and more stable. Like where is your footing in terms of security. But I don’t think that that’s necessarily a choice or an option now. I think now the best thing to do is to keep pushing and sort of don’t expect things to just remain safe or stable forever.
Vlad Costea (01:00:57):
I think you gave us a very interesting hint there regarding secure elements specifically designed and dedicated for Bitcoin.
Rodolfo Novak (nvk) (01:01:06):
I think it will eventually happen. There is some conversations being had with different vendors, but the scale is not there yet for this kind of R&D. So as Bitcoin scales we have a lot more economies of scale and we can make more interesting things that are only allowed at that scale.
Vlad Costea (01:01:33):
So my last question to you is, to which extent is owning and using a Coldcard comparable to having a dedicated general purpose computer?
Rodolfo Novak (nvk) (01:01:46):
It’s a whole different level. Unless you are a, well, even if you are a very advanced user, it’s still in my opinion, much safer to use a Coldcard than a computer. There’s just no comparison there.
Vlad Costea (01:02:10):
Okay. I think these are all of the questions that I had for you.
Rodolfo Novak (nvk) (01:02:15):
Thanks for the opportunity to talk about this stuff and clear out some of the conversation around hardware wallet security, secure elements and all that stuff.
Speaker 2 (01:02:33):
Yeah. And thank you for joining. Is there anything else that you’d like to add or promote regarding your work with Coldcard?
Rodolfo Novak (nvk) (01:02:40):
No. Just follow Coldcard wallet on Twitter and join the Coldcard telegram group so you can stay up to date with this stuff. Submit bugs if you find them help us improve the documentation. Keep on buying the product. We keep on making it. It’s as simple as that.
Vlad Costea (01:03:03):
Okay. Thank you very much.
Rodolfo Novak (nvk) (01:03:07):
Thanks for having me.