S6 E9: Jonas Nick, Peter Todd & Alekos Filini on Bitcoin Development
Search this conversation
0:00Max HillebrandWe gotta hand it over to another rockstar panel with Vlad Costea hosting. Vlad, how are you my friend?
0:06Vlad CosteaHi there! I was about to say hello party people, but that's for the wrong podcast.
0:10Max HillebrandNo, actually, HCPP is a really, really big party, if you think about it. And we're having a blast here in meatspace, for sure. And it's awesome to get all of you from cyberspace involved as well. So Vlad, unfortunately, couldn't come. And neither could Jonas Nick, Blockstream cryptographer and wizard of Bitcoin, even though he doesn't have a long beard. That's a good to get this guy, actually. Jonas, how are you?
0:39Jonas NickI'm good. I'm very good. Good to be here. Took me a while to set this up technically because, as being a real cypherpunk, I don't do that very often. And my microphone broke, my driver didn't work. So I hope this will now work out for the next hour.
0:54Max HillebrandWell, we are up and running and I see you got your coffee too. Plus the whiteboard in the background, as any good hacker should.
1:05Jonas NickYeah, that's for my own safety. If I fail to understand concepts correctly, then this whole whiteboard might be able to help me.
1:15Max HillebrandYes, fantastic. So, Vlad, take it over.
1:21Vlad CosteaOh yeah, this is the Bitcoin Takeover, baby, so I take over. That's the expression. But do you hear me well? Because I didn't get to do a proper mic check and I'm not sure if the level is right or I'm louder than Jonas.
1:33Jonas NickYep, I hear you well.
1:35Vlad CosteaOkay. I usually do this testing before I get into conversations. This is Season 6, Episode 9 being recorded live with Jonas Nick. I wasn't sure if your name isn't Nick Jonas and I was about to crack a joke about the Jonas Brothers and where are Joe and what's the other one, Kevin or something.
2:00Jonas NickYeah, I'm sorry for that. I'm not really responsible for that name. But yeah, still sorry for that. And I don't know, it helps my anonymity set because if you Google my name, you usually find like this one guy from the Jonas Brothers instead of me, which of course has its advantages and also disadvantages.
2:24Vlad CosteaOh yeah, Max called us rock stars, but you have the name of a rock star and you work for a rock star company. I suppose we should talk about what Sergey said about Blockstream people in the previous panel. And he was kind of, you know, dismissive of the idea that people want to work for Blockstream and that you are wizards who seek for complexity for the sake of getting more funding. Or I'm not sure if that was the point. I wish he was still around to make it clear. But what is it like to work for Blockstream? Is it as obscure and occult as people claim it to be?
3:01Jonas NickI mean, Blockstream today is not that small of a company anymore. We're more than 60 people. I work mostly with the research team and I'm very happy to work with that team. Is it obscure and occult? I don't think so, because occult often means like doing things in secret and the Blockstream research team does very few things in secret. Most of the things we do is actually in public on GitHub, et cetera.
3:35But, I mean, one of our problems is that, like, everyone calls us wizards and this gives us a certain aura and this actually has the effect that people don't want to apply because they think you need to be a wizard to be at Blockstream. But I think the answer to that is everyone can be a wizard.
4:04Vlad CosteaThat's a nice way of putting it. And I suppose there's a lot of talent that doesn't necessarily want to work at Blockstream, but should work on Bitcoin projects and on the base layer and optimizations. And I happen to have taken a look at your resume, and I'm quite impressed. And I want to talk to you about this philosophy of making transactions smaller. So instead of making blocks big to fit more transactions that are likewise big, you're working on aggregation of signatures that will make transactions a lot smaller in these small blocks.
4:39And I hope I didn't say anything stupid with aggregation of signatures because I saw the look on your face, but I'm going to let you explain.
4:49Jonas NickYeah, so I think the general idea is that what we're doing right now in Bitcoin isn't very efficient. And there are ways to make it more efficient. And these ways aren't very contentious in the community. So we should just do them. And one of these things is the new encoding of signatures in BIP Schnorr and BIP Taproot, because right now signatures are more than 70 bytes and by having a new encoding scheme we can reduce that to 64 bytes and that's a very simple fix that has some effect. And we can do more with that.
5:32Like one of these things that we've been working on at Blockstream in particular is this idea of MuSig, where we are able to aggregate public keys into a single public key and this will be supported by BIP Schnorr as well and will be useful for multisignatures. And then the next step will be what you were talking about, this idea of signature aggregation, and for that you need a verification algorithm that's different to normal Schnorr signature verification. So this is not in BIP Schnorr, but something we'll certainly be working on in the future.
6:15Vlad CosteaThis is definitely fascinating and I'm happy that you mentioned Taproot and Schnorr because they have been discussed for quite some time. They have been in review. And I suppose the time is right to try some activation. I saw some Reddit posts by Greg Maxwell, who is explaining that this situation is already in a very advanced phase. And how do you regard Taproot right now? What is the current state of it?
6:44Jonas NickSo, the current state is that the cryptography parts have been recently merged. And now the question is of merging this pull request to Bitcoin Core that actually deals with this new output format. So right now the output format is SegWit version 1 and the new one with Taproot will be SegWit... or wrong. The current one is SegWit version 0 and the new one will be version 1. And the code to actually understand how to spend these outputs, that's currently just a pull request and that is being reviewed. And I haven't seen much active discussion there recently.
7:40And I have not reviewed this PR. I was mostly involved with the cryptographic parts. And I know that some people reviewed this PR very closely and have written tests for it and done even some, let's say, obscure but useful testing strategies like mutation testing on it. And so I'm relatively confident even without looking at the code that this might actually work. But I'm not so sure about the Bitcoin Core policies for merging these PRs. Maybe they're just waiting for some final ACKs on that PR from reputable people. But it's certainly a good time to talk about activation strategies.
8:39Max HillebrandIf you'd like to point out one specific question here, Jonas.
8:43Vlad CosteaBy the way, how's my connection coming through? Am I lagging?
8:46Max HillebrandWorks very well. Okay, works very good. So, the question is, you know, as you said, the pull requests are very high quality, both in the code as well as in the review and all the testing and so on. But still, it seems that this is taking quite some time for further, you know, activation or merging. So you said that you had some quarrel with the Bitcoin Core process here. Where do you see some problems and how do you think that it can be improved?
9:18Jonas NickThe thing is that I'm not really actively working on Bitcoin Core, or I'm only working on a sub-project of Bitcoin Core, namely the cryptographic library, and I know how that works. So I don't have any problems with the Bitcoin Core PR policies and whatnot, I just don't know how they work. And the thing why it's taking so long is not due to some Bitcoin Core policies, it's just that this stuff just takes time and there are not that many people working on it and many people have other things to do as well.
9:58And like in the Bitcoin world, Taproot is very interesting. But it's not the most interesting happening all the time. So people work on other things too and there were some improvements to...
10:31Vlad CosteaMax? Jonas?
10:36Max HillebrandYes, hello, hello. There seems to have been some realignment of the Jitsi. I hope Jonas can drop in back in just a sec.
10:47Vlad CosteaOkay. I wasn't sure if it's my connection or something has happened on your side.
10:57Max HillebrandI think that it is on my side, but I will check some further things. So Jonas is back! Fantastic. Please continue.
11:08Jonas NickHello. When was I disconnected?
11:13Max HillebrandWell, Vlad, maybe you can pick up with the next question. You had something.
11:17Vlad CosteaYeah. So you're describing the inefficiencies with Bitcoin Core. And I think part of the answer that I got was that you don't work directly in Bitcoin Core or actively and you work on a different project. And after that, I think I kind of lost the plot. But I was about to ask you something which is very much relevant for this conference, because we are speaking as if we are at Breaking Bitcoin or something.
11:50But there are people in the audience listening right now, and maybe that they don't know much because this isn't a Bitcoin-only conference per se. It's more of a hackers congress. And there may be people who know nothing about all of these improvements that are to be made in Bitcoin. So how would you describe Taproot and Schnorr signatures? In very simple terms.
12:05Jonas NickOh, that's a good question and I'm the wrong person to ask that. But I can still try. So, as I said earlier, there's a lot of things in Bitcoin that we can still optimize and one of the things that we optimize with Taproot in particular is that, let's say you have a coin and there's a certain spending policy for that coin. And by spending policy, I mean something like Alice must provide a signature to spend that coin or Bob must provide a signature or Charlie and David must provide a signature.
12:54So I mean policies and sometimes this is also referred to as smart contracts. Now, one of the inefficiencies that we see is that even if, or let's say it like that, you can satisfy this policy without showing the policy. So you can say, "Here, I'm Alice, I provide a signature," and then I can spend the coin without showing the rest of this policy to the world.
13:28And this allows us to have smaller transactions, which are cheaper, we can put more of them in a block, but also makes Bitcoin more fungible, because then we don't have to show the whole policy to the world, only a small part of it. So this is the idea behind Taproot, and Taproot, I think, was first proposed beginning of 2019, I think, and this was like an additional optimization of the proposals that have existed before for doing this thing with only showing part of the policy, where you can do it in a really optimized way where you have one default spending path. And this one is very efficient. It's just a single public key and you can spend the coin by providing a signature for that public key.
14:34But inside that public key, a more complicated policy can be hidden. So if you need something like after time T, then a different set of signers can spend this coin, then you can have that hidden in a public key and nothing else. And this is like the main idea behind Taproot, that you can hide the policies inside a public key itself.
14:59Vlad CosteaYeah, so let me try to give an example if I understood well. Let's say that I'm going to have children soon. And I want to give my son 0.1 Bitcoin when he turns 18 or when he becomes literate enough to sign the transaction himself. And I put these two conditions in a multisig. And whichever one of them gets fulfilled is going to be written in the blockchain. But the final writing on the public blockchain is not going to contain all the conditions that I have set. So it's going to look just like every other transaction. Did I get that right?
15:40Jonas NickYes, until spending time. So you have one default path, and if it's spent with the default path, then it looks like a normal payment. But if you need to use another path, for example, something like after time T, after your son is 18, then you have to show that this path was actually committed to in that public key, and then you have to reveal it. But if there are more spending policies, you don't have to reveal them. You only have to reveal a single one. And yeah, that's one of the main things in Taproot.
16:15The other main thing is that in Taproot, we also switched to the Schnorr signature scheme instead of ECDSA, which means that we can actually, in this single public key, the default spending path, we can actually realize a multisig policy.
16:37And how that works is that if you have Alice, Bob and Charlie, you can say, okay, let's create a three-of-three multisig. Or perhaps a better example are Lightning channels, because that is like an immediate use case. So you have a Lightning channel between Alice and Bob, and a Lightning channel always needs to have some kind of multisig coin with Alice and Bob. And right now you have Alice and Bob's public key in the chain. And that shows the world that this is like a multisig setup, which is bad.
17:13And it's also more expensive because you have to actually write the public key, both public keys and both signatures in the chain. And with a switch to Schnorr signatures, you can simplify this, because what you can essentially do is just add up Alice and Bob's key.
17:30And then in order to sign it, both need to create a partial signature, and then they can add up this partial signature to get a final signature. And then they have this combined key, which looks like a normal key, a combined signature, which looks like a normal signature. They can spend the coin, and it looks like any other payment, but it requires cooperation between Alice and Bob, and no one can know about this from just looking at the blockchain. It's like a second big improvement with BIP Taproot.
18:03Vlad CosteaThat definitely sounds fascinating and I look forward to activating it on my node. Even if I don't use the features, I still think it's worth supporting for the sake of pushing Bitcoin development forward. And I was about to mention that a lot of people look into what else is happening in the space right now and they say, "Oh, Bitcoin has been the same for such a long time. There's nothing going on. There's no development. There are very few developers. Look at Ethereum, they have thousands of developers or whatever. They have DeFi," and look at this yield farming scam and whatever.
18:43And how do you actually convince people that this is not... You know, I don't know which example I should give. Maybe Android development, where it's a good idea to have all sorts of developers working on your platform and do stuff independently. But with Bitcoin, you're supposed to be conservative and make sure that you don't break anything, and when you make a change, it should be very well written and it should not break consensus. And with decentralized systems in general, you should be very cautious. We have had bad experiences in 2010.
19:19And the worst one I think was in 2013 with that accidental hard fork that happened. And essentially, one of the goals of Bitcoin development is to allow Satoshi to return at any point and spend his coins, right? We don't want to move to another chain which makes it impossible for Satoshi to move his coins.
19:41Jonas NickYeah, I don't think that's... I'm not sure if that will ever work, because at some point it's possible that we have actual quantum computers. And then it will be possible for anyone with a quantum computer to spend Satoshi's coins.
19:57But in general, if people say that Bitcoin is somewhat slow in development, I mean, that's true, I think, and that's a good thing, but that only concerns the base layer. I think everything else moves very quickly, at least in my perspective. I have no idea about all the things that happen in the Lightning Network, for example, and all the things that happen in the hardware wallets, multi-signature space. I view Bitcoin and altcoins just as qualitatively different things. So one thing is censorship resistant and the other things are trying to do something else and perhaps don't try to be money, and perhaps they do interesting experimentation for sure, and there are better and worse altcoins, of course.
20:50So I think the answer to that would be just that these things are not really comparable.
20:59Vlad CosteaOkay, so let's get back to something that you actually do and concerns your everyday work. On your website, which is called nickler.ninja, you have a proposal to reduce the transaction sizes of Bitcoin with x-only pubkeys. How would you describe that? And why is it a good idea to reduce transactions and transaction sizes? And how can you do that without breaking consensus?
21:31Jonas NickSo the proposal is actually part of the BIP Schnorr proposal. So right now in Bitcoin, public keys... I can perhaps write it down, if you see that. So let me see. Does it work? Perhaps... 33 bytes. Yeah, I can see. Perhaps it's easier just to talk about it. So right now in Bitcoin, public keys are usually written to the blockchain in the compressed format, which means they have 33 bytes. And in an initial version of BIP Schnorr, we just kept the same public key format, 33 bytes. But we knew that you could actually remove the first byte, because the first byte is just a single bit saying on which side of the curve the corresponding point is.
22:28So this will get very deep, so let's perhaps not go there. So while working on BIP Schnorr we thought, well, I mean, we're thinking about this new proposal now, why don't we make it simpler and remove this first byte? So this is what we're doing now. It doesn't really add a lot of complexity. I mean, I think it just adds one additional check somewhere, and by that you save one byte. It's just something we did along with BIP Schnorr. And then the blog post also goes into detail of how to show that this is not less secure than having 33-byte public keys.
23:24Vlad CosteaOkay. Yes, so why do you think...
23:27Max HillebrandOh, there's Peter Todd. Oh my god. We just expanded the panel conversation. Peter Todd magically appeared. I wrote him like two days ago, inviting him to come on the TV at like 2 a.m. Prague time, which is this evening. He confirms everything, right? I aligned the schedule. All of a sudden, five minutes later, he says, "Max, fuck it. I bought a plane ticket. I will be there on Friday." And now he's sitting here. How are you supposed to organize something when you have crazy freedom-loving people like Peter?
23:58And Alekos as well. So I bring these two on specifically because I think they're, well, first of all, great Bitcoin contributors. And you know, specifically, Alekos is working on the Magical Bitcoin wallet, which is really cutting edge on many of the technical fronts. I consider it as one of the next generation wallets. And we spoke about Taproot, right? And how that type of new wallet might work with Taproot. So maybe you can speak a couple things about that specifically.
24:29Alekos FiliniYeah, I mean, I'm definitely not like a Taproot expert, so I don't know the crypto details. In general, for my project, which actually was called Magical Bitcoin Wallet until a couple of weeks ago, now it's been renamed to BDK because we merged with the BDK guys. So yeah, now it's...
24:47Max HillebrandWhat does BDK stand for?
24:48Alekos FiliniIt's Bitcoin Dev Kit. It was a project started by Steve Myers. Maybe you've heard about it. Yeah, it was originally called BDK and Magical. Now we've merged them together, but the name BDK remained. So yeah, speaking of Taproot, it's definitely something I'm looking at, because the goal of BDK is basically to make a library that can support potentially any kind of wallet, like generalized scripts, different spending policies and conditions. So Taproot would be definitely a good thing for that, especially in terms of privacy.
25:21So right now, if you wanna do like a very complex wallet, you're gonna have a very specific on-chain footprint, because you can see scripts that are very, very particular. So yeah, definitely looking into Taproot. I don't know how yet, so probably people smarter than me and other people, we're going to try to make it work somehow.
25:48Vlad CosteaOkay, so we have two Taproot experts. I suppose we should make them discuss and debate on something. Is there anything that you guys disagree on?
26:00Max HillebrandWell, Peter disagrees with most everything. So what are your disagreements with Taproot? Where do you have your quarrels with it?
26:07Peter ToddWell, I think the biggest quarrel is I haven't really read much about it. And, you know, I mean, I've been focusing on client-side consensus stuff for quite a while, so I haven't really paid too much attention to it.
26:19But, you know, even for client-side validated states, I think Taproot is still a major improvement, with, you know, multiple things like public key tweaks and so on. Yeah, you know, I think, like, reusing it as a, you know, as a module really, and then, you know, in the stuff I do, it's certainly going to happen. But, you know, it's one of those things where kind of let it take a while to gel and so on. And most stuff I've been doing has been written in Rust, so of course I'm waiting for a Rust implementation to happen. Yeah, you know, then to argue about consensus and so on, whether or not the two implementations match. But, you know, at some point I'll probably wind up using that. Isn't there already a Rust implementation out there? Does anyone know?
27:06Jonas NickThere's a Rust implementation of the crypto parts, and they're right now also including the new BIP Schnorr things. I think it's called rust-secp256k1. But it doesn't include consensus parts, and if it was, you shouldn't use it, of course.
27:28Max HillebrandYeah. Okay, Vlad.
27:36Vlad CosteaOh, so I'm the one who's supposed to save this awkward silence moment. Okay, so let's get back on track. I wanted to ask Jonas about his work on nix-bitcoin, because I think that's unique, and I'm not sure if he's the maintainer of that repository. But I don't know many people who use NixOS, so I think that's definitely special.
27:54Jonas NickYeah, so I'm the maintainer because I started it, and people verify my signature when they update nix-bitcoin. So the reason for doing this was, I think, as right now there are many Bitcoin node projects which try to bundle like Bitcoin, Lightning, CoinJoins and whatnot. And I think they all started with the same problem, that if you try to set this up yourself, your own node, then you want to do it in a way that's reproducible, and perhaps for others to use as well. And this is how nix-bitcoin got started. Yeah, not many people use Nix, that is true. But I like to joke that using NixOS is like the ultimate low time preference move, because it's just so much better than everything else.
28:54But it has a steep learning curve. That's like a downside if you really want to tinker with the code. What I looked for was what is the most systematic way to set up a Bitcoin node, and well, then this whole NixOS thing came up. So I'm really happy with it, and there are other brilliant contributors. This game has levels, because the cool thing about nix-bitcoin is you can treat, or with NixOS you can treat your infrastructure as code, but it's like real powerful functional code.
29:36So you have really powerful abstractions, and you can really abstract the things away pretty greatly. I think we have by far the best tests of any node project, and that's an interesting aspect, I think, because you can just run tests and then everything runs in a container and then executes some Python tests, and you can test multiple configurations and so on. So that's pretty cool. But of course...
30:06Peter ToddAll right. Yeah, for those of us who haven't looked at Nix much, what's your, like, 30-second elevator pitch for what it actually does that's different?
30:13Jonas NickSo Nix itself is a functional package manager that's aimed at being reproducible. And reproducible means something else than what we usually refer to as reproducible in the Bitcoin world. It doesn't mean binary compatible. It just means that it basically pins your dependencies by hash, and then, not the same bits, but at least it has the same dependencies. And so this is the main idea behind Nix. And you can use it on any Linux distribution, and on Mac as well, as an additional package manager to whatever you have, if you want to play around with it. And NixOS is an operating system built on top of that idea, which brings that to the next level, because then you just write a single declarative configuration.nix file. In the case of nix-bitcoin, you say: I want Bitcoin enabled, I want Lightning enabled, I want Tor enforced.
31:20You just set these options to true, then you run nixos-rebuild switch, that's the command, and then it completely builds your system according to your configuration.nix. So then that's really cool, because then you can have your whole system configuration in a Git repository, for example, to track it. And you can also roll it back easily, and you can deploy it from other machines easily. So, yeah.
31:52Vlad CosteaBut I think about this machine. How cross-compatible is it for hardware? Are there going to be some compatibility issues when you run it on different architectures of hardware?
32:06Jonas NickSo you can run it on ARM and you can run it on x86, and yeah, it works just like any Linux distribution. I haven't tried anything else yet.
32:17Vlad CosteaYou know, there are only two other people I know that use Nix, and they are Stick from Trezor and an Italian girl whose name is Daniela Brazzoni or something, and I think she works for Stick or something, I don't know. But they are the only ones, and right now I discovered that Jonas also uses it, which I think is interesting. I might try it. I'm a Mint Linux kind of guy, so...
32:41Jonas NickI mean, nix-bitcoin is more like, you don't really have to understand Nix to use it. You just have to change a configuration file, and that's mostly it. It's just if you want to install a node. And you can start with that. And if you want to go deeper into it, then you can make it your main distribution.
33:05Peter ToddBut is this though like Qubes, where you're separating things in virtual machines, or could you use Nix that way?
33:13Jonas NickSo I think so. What we're doing in nix-bitcoin right now, we use heavily containers, but not like Docker, just Linux namespaces and cgroups. And because you're essentially programming your system in a functional language, you can use abstractions and you can build your own Qubes, basically. Not with a graphical user interface and everything, but at least a... I think there's even a project which tries to rebuild some parts of Qubes with Nix, so that's not too difficult. And we do like a weaker thing. We don't use virtual machines, we use containers right now to really containerize every single system service that we have and put them in their own network namespace. Then we set up routing rules that only the services who are supposed to talk to each other can talk to each other, et cetera. So security-wise, So security-wise, I mean, it'd be as good as the Linux kernel, Linux kernel cgroups and stuff.
34:17Peter ToddYeah, yeah, yeah, it's so-so.
34:23Jonas NickBut at least it's reproducible. So you know if there's an issue with one of your dependencies, you at least know exactly what dependencies you have. And it's also like Qubes OS. Like in an imperative operating system, I would say it's also more or perhaps more likely to make mistakes because you misconfigure something or you configure something the first time, then you don't do it the second time, and etc. So, there are trade-offs, certainly.
34:56Max HillebrandSo, NixOS is an operating system, right?
35:00Jonas NickYes.
35:01Max HillebrandBut, in what language is it written?
35:06Jonas NickIt's an operating system built on Linux. It's just another Linux distribution. And some parts are written in Nix, some parts are written in C, I guess. Yeah, that's it.
35:22Max HillebrandAnd one more question for these dependencies. As you said, the dependencies are hashed and kind of ordered. How far down the supply chain basically does it go?
35:33Jonas NickSo this is an interesting question. It doesn't go as far as Guix, for example, G-U-I-X. And Guix is a bit of a similar project where you really try to get binary compatibility and to really reduce your trusted base or trusted computing base, which I think are interesting goals. But the Guix community is very small right now. And I think there's no reason why Nix couldn't do similar things as Guix and both projects could profit from each other. And so yeah, how far does this go? This goes very far. So you basically pin every dependency. And this recursively pins every other dependency, and so forth.
36:29Vlad CosteaYou seem content, Max. I'm happy that you got your answer.
36:34Max HillebrandYes, actually, no, that's a very good answer. So I'm happy.
36:39Vlad CosteaI'm actually excited that I get to have so many developers at the same time and I wanted to ask you guys, each one of you, so Jonas and Peter and Alekos, what is the Bitcoin project on which you're not working that gets you most excited and you see most potential in it?
37:01Alekos FiliniWho's going to start? I don't know. I mean, one thing that kind of fascinates me is Lightning in general, like the whole ecosystem. I would like to learn more and play with it more, I would say. I don't know about potential because I mean, there are drawbacks, but it can definitely grow from where we are now. But yeah, I would definitely like to have more time to spend on Lightning stuff in general, because it's really, really cool and also moving pretty fast. So yeah, it's hard to keep up.
37:33Peter ToddYeah, I made a tactical mistake there, letting him answer first. Because, you know, I mean, I just used Lightning, what, three times today, to actually do useful things and it works. I mean...
37:46Max HillebrandMagic.
37:47Peter ToddI can't say that of much of the stuff I've done. OpenTimestamps works, but it doesn't have the same satisfaction as, you know, buying a beer. It's pretty magical how far it's kind of come. Maybe some interesting things to figure out more trade-offs about. How do you get payments to go through faster? How do you make them more reliable to go through? There's a lot of different space there. Of course, what we're doing is just adding trust to that. And those trusted solutions are working pretty well, marginally with trust.
38:26Max HillebrandYeah, that's always the thing, right? There's layers and then the trust trade-offs, obviously. And for me, Lightning is this natural answer. But to be more specific, actually, I will say Alekos's Magical Bitcoin wallet, or Bitcoin Dev Kit, because for me, it just packages all these new awesome things that we've built over the last couple years into a next-generation wallet. And that is so powerful, and I think it will really just realign what we actually think of as a wallet to quite a big extent. So miniscript, output script descriptors, and PSBT. That together is a big paradigm shift. So I'm super excited for that. Thanks. What about you, Jonas?
39:14Jonas NickCan I give two answers?
39:17Max HillebrandSure.
39:18Jonas NickSo the first one would just be a usable multisig. So I see there's a lot of progress there because I want to finally use multisig in a more user-friendly way and not just for geeks and without any danger of somehow not finding your coins anymore. So that's one thing I'm excited about. And the other thing I'm excited about is this specific concept of dual funding in Lightning. Where right now, if you open a channel with someone, you only use your own coins. And the idea of dual funding would be that both parties provide inputs to the funding transaction.
40:02And that way you have a CoinJoin, but even better, since, or if you use that channel and later close it, then the input and output amounts will be uncorrelated. So you can have an unequal amount CoinJoin, which right now, at least in my opinion, is quite a difficult thing to do correctly.
40:25Peter ToddYes, yes, I mean, and that's a good point too. How, like, even without any more scripting proof and so on, I mean, Lightning channels look like normal multisig when they go through the normal processes. And, you know, right there, you've done so much to confuse analysis and things like that. Just leverage that even further.
40:48Max HillebrandYeah, no. And as Jonas said, this is a small CoinJoin, so the common input ownership heuristic is broken. And specifically because we have uncertainty about which channel state is being published, the amounts are very difficult to correlate too. And now if you combine that with channel factories, that's, I think, going to be very, very interesting. Having actually more than two people providing these inputs, and, you know, we could potentially have hundreds of inputs in a channel factory opening. That's really going to confuse surveillance quite a bit.
41:21Peter ToddWell, you know, an interesting point with all this is I remember from last year I saw this point at a presentation how you can have a lot of variety in what exactly Lightning channels are between, you know, two entities. You know, one really seemingly crazy example being, well, you and I can operate a channel with a negative balance, as in I actually owe you money. But, you know, with a business relationship, that can make a lot of sense. It's just, you know, fancy accounting and it's not just the Bitcoin in the channel that's kind of securing it.
41:50But when you add that to, say, big entities creating these kinds of arrangements to make a bit more money on the routing fees and so on, well, what that'll look like for the rest of the network, it can just look like, well, we have two connections, but they're synthetic. You know, they don't actually match what's on the chain. And again, it's yet another way that you're losing, you know, you're losing data because you don't have insight into any of this.
42:22I mean, it's an interesting question. I don't know the answer myself, but it could be interesting to have capability in the Lightning Network to have, you know, data published about channel connections that doesn't actually mention anything about the pub keys or any of that or endpoints or so on, just to have that ability. Because, I mean, after all, in some cases, you know, maybe it's a bridge to Tether for all you know. You know, maybe Blockstream Liquid.
42:52Max HillebrandAnd I think that, that part plus that is good.
42:57Vlad Costeayeah, I was about to ask Jonas to provide the second answer to the question because he only gave us one.
43:04Jonas NickThere were two answers. One was like just multisig user experience and the other thing is dual-funded Lightning channels.
43:13Vlad CosteaOkay, I'm stupid. I admit it. That's why I don't work for Blockstream or something. But anyway, what is your guys' take on CoinJoins? Because I see a lot of debates in regards to their efficiency. And are they in their current state efficient, or should one use them to gain anonymity, or should we wait for more improvements to be made, as you said, with uneven amounts that get CoinJoined?
43:42Peter ToddWell, I mean, I'm the guy who actually came up with the name CoinJoin, so I'll be kind of the first to go say, you know, CoinJoin isn't that important compared to things like Lightning. But at the same time, I mean, all this debate about how inefficient they are and stuff, I think this is the point of how, you know, once you have good software and good kind of economics, good community, if you will, you can always use CoinJoin to make things more efficient, for the simple factor that, you know, if you and I need to go make a payment and we can arrange a CoinJoin, well, we can avoid the header on the transaction.
44:18You know, right there you save a couple bytes. Yeah, yeah, and that's with the Bitcoin protocol right now. So it will always be more efficient if you can make it convenient. Now, some of these CoinJoins might actually save you money, but there certainly are opportunities here to make it useful and it's just a matter of, well, how do you exploit them? I think where the inefficiency comes in is, well, do I want to do a CoinJoin, say, make my coins private? I'd probably be better off, you know, throwing them in a Lightning channel and routing through there instead.
44:50Or, you know, frankly, like throwing it on Zcash or Monero. That'll give you better guarantees for, you know, less money than a lot of the CoinJoin usage.
45:02Max HillebrandYeah, I think all very good points. And it's a bit tricky, right? Because, as you said, making a payment in CoinJoin is usually where we get a lot of efficiency, or where we can potentially get a lot of efficiency gain. But so far, the only CoinJoin implementation that does this is JoinMarket, right? And even for here, only one user out of the CoinJoin can do the payment, only the taker, not the makers.
45:28Peter ToddI'm pretty sure with JoinMarket there is actually a mode where you can say, "I need to do this payment," and just do it eventually. But I can't imagine how many people actually use it. But say if I wanted to put some of my income off to an exchange, that would be a perfect use case because I know rent's going to be paid in a week and I can wait. Not a real common thing.
45:50Max HillebrandYeah, exactly. And so, for example, in Wasabi specifically, one of the issues is that it's not a payment CoinJoin, right? Because it relies on ZeroLink. It relies on and it must have equal amounts on the cryptographic coordination layer. Otherwise, there is no privacy. And this is a big downside because now you cannot make a payment. You have to CoinJoin to yourself and then make a payment. So instead of just having one payment transaction, you have two, right? So this is inherently more inefficient. And then the question is, how do you structure this type of CoinJoin rather efficiently?
46:24And so what Wasabi consciously chose here was a high anonymity set, so a large base of users, so that the transaction ID overhead becomes, like, we use that a lot, as well as batching different CoinJoin pools. So in Wasabi, there's a 0.1 pool, a 0.2, 0.4, 0.8, 1.6, 3.2, exponential. And all of these could be independent pools, you know, each creating their own transaction where there's only the 0.2 or only the 0.8 or what. But in Wasabi, this is batched. So under one transaction, all of these pools are batched, and then we get the efficiency of transaction batching.
47:04So these are two of the things that we considered in the early Wasabi design. But it's far from efficient and cheap. It will probably still cost you a lot for, first, on-chain fees, but then you also have to consider coordination fees, which specifically in JoinMarket can be actually quite expensive.
47:21Alekos FiliniIt's probably even worse for people that have different wallets, that they have to make one first transaction to send to Wasabi, then one transaction to JoinMarket, then another transaction to get everything out.
47:30Max HillebrandYes, yes, yes.
47:33Peter ToddYeah, yeah. And I think that might be a little underappreciated by many people how, you know, and you see this in Lightning wallets, like, you really want your wallets to be able to support many different things, purely to avoid having to do many transactions, to get things in and out.
47:49Max HillebrandWhich is also why output script descriptors are so important, because now you can kind of import just one private key specifically from that wallet into another and then sign.
48:00Alekos FiliniYeah, probably the problem there is also the different scripts. Basically what I was saying before, Taproot here would do that, because if I make a CoinJoin, I'm the only one with a, I don't know, three of five multisig, and then I send something to another three of five multisig, then it's probably easier to link people. So Taproot here really helps, because every input, even if it's a kind of, I don't know, strange script with timelocks or even just a simple multisig, they will all look the same. Yeah.
48:26Max HillebrandYeah. Interested in Jonas's answer, actually.
48:31Jonas NickSo, I must admit I'm not that knowledgeable about CoinJoins, because I know that's today a huge field, but I've never been really convinced that it works so great, I must admit. I've talked earlier about saving one byte, so I don't really like making more transactions than you have to for getting your anonymity set. I hope in the future, but this is the far future, that Bitcoin can advance to a state where CoinJoins are really, really usable. I mean, one fix is confidential transactions, but that's very hard to get into Bitcoin itself.
49:17I want to shill nix-bitcoin again because we recently added JoinMarket. So I'm going to play with CoinJoins at least soon, because I wanted to have DeFi on nix-bitcoin and get my yield with JoinMarket by providing liquidity. And I think one thing that people who are interested in this should look at is this WabiSabi idea by Wasabi.
49:47I'm a really big fan of that because that allows you to split your inputs and outputs you provide to the mix in a way where the coordinator doesn't learn about it. And more importantly, I think, also in the context of HCPP, because I know that a lot of people are talking about that, is with the same scheme, WabiSabi, you can build really efficient anonymous Bitcoin banks. So that's, I think, a cool side benefit of this WabiSabi idea.
50:20Max HillebrandYes, yes, and I got to give Jonas credit here because we were researching WabiSabi for a year and a half or so. And we were at the point where we found out what we want to have, so that the user can register multiple inputs without revealing them on the communication layer to the coordinator, can register multiple outputs without linking each of them to the same user as well, and the coordinator must not link inputs to outputs, right? And so we had that figured out, that this was what we want, and we didn't know the crypto for it. So we got Jonas on one of the research calls and he was like, "Yes, Signal paper, keyed-verification anonymous credentials, that works perfectly," and it was a solved case. And since then, we've kind of refined it and implemented it. So thanks, Jonas, for actually coming up with that.
51:08Jonas NickYeah, I just did the easy parts. I mean, the hard parts are actually putting the parts together such that they work. And I know, I think you people are still implementing it, right?
51:17Max HillebrandYes, the crypto implementation, at least the proof of concept in C#, is pretty much done. It's going to be slowly now upgraded with more tests to production-ready code. And we're currently doing a lot of research on transaction structure and then graph analysis, which turns out to be much, much more difficult actually, because there's much more design space, right? Like, how are you going to structure one single transaction and then an entire graph of transactions in a way that is efficient and brings decent privacy? It's actually very difficult to come down to an optimal solution. So it's a lot of work still, but it's worth it.
52:05Vlad CosteaCan I take a dive into shitcoinery and ask you guys what you think about something, which might be a bad idea, but I still want to know why?
52:13Max HillebrandGo for it.
52:14Vlad CosteaSo yesterday Litecoin activated extension blocks with Mimblewimble. And we know what that means if it works on Litecoin. They have been working with a Grin developer and they want to have a mainnet launch in a couple of months. So why are extension blocks a good or a bad idea? That's the idea that concerns me the most.
52:44Peter ToddI mean, the way I would go with that is essentially that's just a way of doing a protocol upgrade with an increased block size. I think whether or not that's good or bad depends on, well, ultimately, part of it is, well, what's increasing block size, and can the system likely handle that? I mean, if I owned Litecoin, one of the things I'd be thinking about is, well, is this crypto really going to work? Because at some point, I mean, assuming this is a soft fork, and I have to look at the details there, but if it's a soft fork, the amount of coins that go into the Mimblewimble part will eventually get to the point where it's economically relevant.
53:24Which also means a bug in that is economically relevant. And like it or not, this is a whole bunch of much more complex crypto than Litecoin normally is. I mean, a sort of standard I like to talk about is, well, could a drunk fine arts graduate go and figure out how to go audit this? Because I'm a fine arts graduate.
53:46Max HillebrandAnd usually drunk.
53:49Peter ToddAnd, you know, I could not do that for Mimblewimble. And I mean, frankly, no one really can. All they can really do is just audit the math and say, "Oh, yeah, that looks like it works." Whereas it's so easy to go audit the amounts. So this is really a fundamental change in Litecoin's economics. It'll take a while to happen, but it's a big deal. And I couldn't say Bitcoin should do this because it's unclear how well that math works.
54:17Max HillebrandDoes anyone know more details on this extension block specifically? So you move your Litecoin in this case from regular scripts into Mimblewimble scripts or in this Mimblewimble extension block. Can you ever spend this Mimblewimble coin with a regular Litecoin script in the output? So can you move out of the extension block back into the base currency?
54:42Alekos FiliniI mean, if it's a soft fork, you should be able to do that somehow. But I don't really see it.
54:47Peter ToddWell, so the thing is, like with a soft fork in this, the miners would have full insights into the extension block and what's going on and so on. And they would essentially just say, "Well, you know, there's this pool of money that looks like an anyone-can-spend, but if it follows the rules of the Mimblewimble block, well, then let it go somewhere else." Which is not unlike, say, Blockstream's Liquid thing. It's just replace, you know, miner consensus with a bunch of guys who are pretending to be banks, but we're not going to use that term. And, you know, again, that's the key thing with Mimblewimble. The money locked up in that thing, because it by default is an anyone-can-spend, if the math of the Mimblewimble implementation breaks, it doesn't actually mean the math breaks.
55:40It could also mean there's just an exploit. I mean, just like any of these other things, you can steal money en masse and bad things can happen. And you want to make sure you actually get this right. Or, you know, you can see the price of Litecoin crash.
55:57Jonas NickI think it's pretty cool. I wasn't aware of this. I mean, of course, this wouldn't be an option for Bitcoin in the midterm, perhaps in the very long term, because of course on the base layer, we do get efficiency improvements and the hardware gets better and the internet connection gets better. So perhaps at some point we can soft fork in bigger blocks, but that won't be in the next few years, I suppose. So I'm happy that there are projects that are experimenting with this.
56:36Vlad CosteaYeah, I think it's definitely interesting that for once they are doing something different from Bitcoin, as opposed to just being the testnet or copying code that gets developed by Core, right? But I suppose it's relevant in the context of a hackers conference, as opposed to Breaking Bitcoin or something else that would be hardcore Bitcoin. And I think this is close to the end.
57:10Peter ToddI mean, I will go say, you started this off saying shitcoin, and what Litecoin is doing doesn't quite clearly fit the description.
57:21Vlad CosteaOh yeah, I think it's definitely interesting. But a lot of people have very bad ideas in regards to what Litecoin is and what it does. They consider it basically some sort of inflation for Bitcoin's supply because it's not much different. It has some of the same properties. It doesn't have the same security.
57:43Peter ToddI mean, I think like Adam Back in particular, especially earlier on, he's really pushed the idea of, well, these different coins effectively inflate the supply of Bitcoin. I don't know, I'm not that convinced by that argument. But that doesn't mean I necessarily think an infinite supply of coins is a good thing.
58:10Vlad CosteaOkay, so I see we're getting towards the end. I want to ask Max if someone else is supposed to come in and take over.
58:16Alekos FiliniYeah, it doesn't help. He was asking if somebody else is supposed to come.
58:21Max HillebrandOh, someone is supposed to come. Yes, Pavla Holcova or something. Czech names. She's coming soon, in two weeks she will be down here. Let's see, but I actually think maybe already. I'm not sure. She will come, for sure.
58:44Vlad CosteaI mean, I could ask a couple more questions, but I'm trying to be fair and not take up too much time.
58:55Max HillebrandOh, no, no, that's all right. I think we can make a smooth handover now over the next couple minutes. So, yes, Vlad, any final rapid-fire rounds to finish this off, maybe?
59:09Vlad CosteaYeah, so what does it take to become a Bitcoin Core developer, and why should one pursue this goal?
59:20Alekos FiliniI'm just going to start by saying that I'm technically not a Bitcoin Core developer because I don't have any commit in Core, or actually in any other major Bitcoin project. I think maybe in c-lightning. But yeah, I don't know. I just think if you like Bitcoin, if you have passion for the topic, just start a small project and then try to contribute to open source stuff, get people to know you. I mean, this is basically what I've done. I guess it would work for others too.
59:54Peter ToddI contributed quite a few years ago and then haven't for a while. But based on what I've seen, I'd actually go say that's kind of the right approach. Like, go find something within the entire ecosystem that makes sense for you. Don't just focus on, like, well, should I contribute to Bitcoin Core? Frankly, I mean, Bitcoin Core, I don't really want to say stagnated because it's not really true, but the really key things that could be going on, I think it's a smaller share of the entire ecosystem than it used to be. You know, it used to be like...
1:00:27Max HillebrandThat's probably good. I mean, it's specialization of that.
1:00:29Peter ToddYeah. Oh, it's like...
1:00:31Max HillebrandLike Bitcoin Core is trying to get node and other stuff.
1:00:33Peter ToddYeah. And, you know, five years ago, there's just more stuff to be done in Bitcoin Core relative to the whole ecosystem. Whereas now, you know, it's just one part of many. I mean, particularly, of course, Lightning. You know, Lightning's... there's more code complexity and stuff in Lightning than all of Bitcoin Core. So, you know, find the thing you're interested in and don't feel like there's more value in one thing than another. You know, figure out what excites you and what you're good at.
1:01:03Max HillebrandYeah, and you know, one important thing is that being a general free software contributor, specifically Bitcoin Core, does not mean that you have to be a coder. Funnily, even officially looking at the Git log, I'm a Bitcoin Core contributor because I fixed a typo in the Taproot pull request to libsecp. So, you know, that counts. But I agree with you. I mean, Bitcoin Core is a super important project. Super important. But the question is, if you contribute to this specific project, is that the most important thing that you can do with your time? And in many, many cases, probably the answer is no. Just because the Bitcoin Core process will probably need very specialized knowledge to make a meaningful impact. I'm sure maybe some UI work or some, you know, test cases and everything.
1:01:52Peter ToddYeah, my own CheckLockTimeVerify patch to Bitcoin Core, which was an upgrade to the consensus, that was probably like five lines of code, maybe 50 lines of tests. And what was it, two years or something of like politics? Like, the skill set does not look like a normal software project. You know, frankly, most of what I contributed to Bitcoin Core was not writing code. It was peer review and analysis and so on. You know, there's much more interesting stuff on my blog than anything I actually directly wrote in Bitcoin Core. Yeah.
1:02:28Alekos FiliniI think too many people see making commits on Core as some kind of achievement, like a validation, you need to do that. But I don't know, I mean, for me personally, I don't think this is the case. Maybe I will in the future do something on Core, maybe I won't. I don't think this is a measurement of how much of a Bitcoin developer I am. What do you think, Jonas?
1:02:49Jonas NickYeah, I think most important, as you've already said, is passion. And the second thing is luck, because you need to find someone who funds you for the things you're passionate about. And there are quite a few people passionate about Bitcoin, and there's only so much funding. So if you don't have that luck immediately, then when you're really passionate about it, then you also need patience, because then you need to constantly reapply, re-ask, try to get a grant, and then hopefully at one point you'll be able to work on what you want and also get paid. And if you don't get paid and you do it in your free time, that's great too. There are many people who do that. But don't freak out about it and burn out. That will be my suggestion.
1:03:40Vlad CosteaOkay, so I see that the next people, specifically Max, is around and is about to take over. So thank you very much for joining this conversation. I'm very happy that I got to meet you all because I saw you a year ago in Berlin. And coincidentally, I'm wearing the T-shirt that I received during the Lightning Conference. I saw all of you at the Lightning Conference and it's ironic that there's a global pandemic which prevents us from meeting in person right now, but I'm glad that we get to do this.
1:04:13So thank you very much for participating. Thank you, Jonas. Thank you, Peter and Alekos and Max. And I'll see you around. I think I'm going to host tomorrow morning once again. And there's going to be John Carvalho. Hello.
1:04:13================================================================================ Production notes: Live-streamed panel at the Hackers Congress Paralelni Polis (HCPP) 2020 in Prague, with Vlad Costea and Jonas Nick joining remotely via Jitsi and Max Hillebrand, Peter Todd and Alekos Filini on stage in person, conducted in English by non-native speakers, no music intro and no sponsor reads. Correction: anthropic/claude-opus-5.5 via NanoGPT in 4 windows with a whole-episode speaker/glossary plan, the show notes, the Qwen3-ASR full pass (85 chunks) and 50 Scribe v3 re-checks of doubtful spans; [unclear] resolve pass: 5 of 5 flagged paragraphs re-checked. Automatic checks: word count vs raw, timestamps, seam duplication. Fixes: Speaker repairs: opening is Max, Vlad's reply split; Jonas/Max/Vlad boundaries fixed (00:00:54, 00:01:33, 00:08:39-00:08:46, 00:11:08). Removed hallucinations ("The show is brought to you...", "Thank you for watching/Please subscribe", stray "Thank you", "This is a podcast about Bitcoin", "This is a very important topic", duplicated phrases). Restored dropped speech from second ASR/re-check (MuSig, "in public on GitHub", Breaking Bitcoin/Bitcoin-only conference, "committed to in that public key"). Fixed Vlad Costea, HCPP, Blockstream, BIP Schnorr/Taproot, MuSig, ACKs, blocks, Jitsi, 2019, meatspace; Removed hallucinations ("Join the conversation", Bitcoin.com production line, "Thanks for watching", "Thank you for watching... Bye" x2, stray "Thank you"); restored dropped speech (Schnorr key addition, Alekos/Max BDK exchange, Peter's Rust/consensus remarks, Jonas Nix reproducibility, Vlad hardware question); fixed BIP Schnorr/Taproot, rust-secp256k1, nix-bitcoin, nixos-rebuild switch, Qubes, Bitcoin Dev Kit, Magical, CoinJoins, altcoins, "Okay, Vlad"; speakers reassigned (Guest A=Jonas, B=Max, C=Peter, D=Alekos; Peter's Nix question, Vlad's hardware question, removed spurious "most popular platform" line); Speaker repairs (Guest A=Jonas, B=Max, C=Peter, D=Alekos; Vlad/Max boundaries at 00:36:29-00:36:39; 00:37:33 Peter; 00:47:49 Max); restored dropped speech from second ASR/Scribe (Vlad's question, Jonas's first answer, Peter's OpenTimestamps line, negative balance channel, Tether, Wasabi pool batching, Taproot multisig point); removed hallucinations "www.bit.ly/BitcoinDealers", "Episode: The world is a world of people."; fixed Guix, Qubes OS, nix-bitcoin, CoinJoin, ZeroLink, JoinMarket, WabiSabi, PSBT, Bitcoin Dev Kit, Blockstream, Zcash/Monero, "as Jonas said", heuristic, solved case, keyed-verification anonymous credentials; Mapped speakers (Max, Jonas, Peter, Alekos); fixed Mimblewimble, Litecoin, Liquid, Adam Back, libsecp, Taproot, CheckLockTimeVerify, c-lightning, commits; restored dropped speech from second ASR (Jonas tail, Max questions, Peter "I mean, I will go say"); removed "two weeks" and bit.ly hallucination; split turns at speaker changes.