Mark “Murch” Erhardt and Mike Schmidt are joined by Lloyd Fournier to discuss Newsletter #213.
The Bitcoin Optech Podcast and transcription content is licensed Creative Commons CC BY-SA 2.0
Releases and release candidates
Notable code and documentation changes
Mike Schmidt: Welcome everybody, Bitcoin Optech Newsletter #213 Recap. We’re going to go through the news and some of the PRs, and we have a special guest today as well. I think it makes sense to do some quick introductions. I’m Mike, I’m a contributor at Optech and also Executive Director at Brink. Murch?
Mark Erhardt: Hi, I’m Murch, I work at Chaincode Labs. I contribute to a bunch of different educational initiatives and sometimes also manage to do some code in Bitcoin Core.
Mike Schmidt: Lloyd, do you want to introduce yourself for the folks?
Lloyd Fournier: Sure, yeah. I’m Lloyd, I work on Bitcoin research. I’ve been doing that since around 2019. I went to my first Scaling Bitcoin conference and was inspired by the other attendees and became a fully-fledged bitcoiner, and now all I do is think about Bitcoin and work on Bitcoin. I have my fingers in many different honeypots. One of them is the Discreet Log Contract (DLC) effort, and especially on the research side of DLCs. I’m really excited in practice for the technology to be able to make bets on things, do trading on things without third-party custodians at least. Yeah, so I’ve been really interested in that. I’ve been contributing to the research side of things and we just published a paper on it that I’m really excited to talk about.
Using Bitcoin-compatible BLS signatures for DLCs
Mike Schmidt: Excellent. Well, yeah, thank you for taking the time to join us and discuss that paper and your post to the mailing list. I think we have a diverse set of technical experience on the call, is my suspicion, and so I think maybe if you could explain DLCs generally and maybe also get into existing DLC technology, and then we can kind of get into why what you’re proposing here has pros and cons versus what is largely being done today.
Lloyd Fournier: Yeah, sure. So, yes, DLCs. Well, you can make all kinds of funny conditions with Bitcoin Script, well, perhaps not all kinds, perhaps there’s some really funny kinds that you can do on Ethereum that you can’t do on Bitcoin. But most useful kinds of conditions you could want to do, conditional transfers based on revealing secrets, transfers based on time and transfers based on having a certain threshold of signatures, you can all do that in Bitcoin, but then there’s this question of how you do conditional payments based on real-world events, right, because that’s the holy grail of conditional payments, is to basically map your coins conditionally moving to someone else based on a real-world event. There is of course no way for the Bitcoin blockchain to inspect the real world, so you have to somehow cryptographically map the real world into Bitcoin.
So, a proposal was made by Tadge Dryja, an ingenious proposal, called Discreet Log Contracts, was the name of the paper, and it was a very simple idea, which was totally out of character for crypto-based ideas at the time, or the other ideas that were rather complicated to do this. And the idea is basically, yeah, two parties put coins into an output on the Bitcoin blockchain and create some transactions that will be enabled should an oracle attest to that outcome or the state of the world being in a certain way. So, the oracles, the users would look at what the – some kind of, let’s say, an oracle announces an event, an oracle would announce that it will attest to an event in the future. When that announcement is made, the parties would be able to construct a set of transactions each not yet valid, but each would become valid should the oracle then attest to a certain outcome, one of the finite outcomes specified in the oracle’s announcement. And that is at a very high level how DLCs worked.
At a slightly lower level, the way I like to think about it or talk about it is that we put coins into an output and we verifiably encrypt signatures under some key, and each key corresponds to a certain outcome. And so, when/if that outcome is attested to, we are able to decrypt that signature and then put that transaction on the blockchain. The verifiable encryption we use is called an adaptor signature, which allows you to create a signature that is encrypted, encumbered by requiring some decryption key, and those decryption keys are the things that the oracle reveals, in this case when some outcome occurs. So that skips over some things, but that does it for how things started and how things are today. And I’m just going to check in, because my internet is bad, that everyone can still hear me and that I’m not just talking to myself.
Mark Erhardt: Yes, we hear you.
Lloyd Fournier: Perfect.
Mark Erhardt: So maybe one comment.
Lloyd Fournier: Go ahead.
Mark Erhardt: The interesting thing here is of course that the oracle attesting to an event makes this a public announcement, and the parties using that attestation do not have to tell the oracle that they’re going to use it, but can totally depend on it without the oracle even being aware of what it is being used for. I think that’s pretty cool.
Lloyd Fournier: That is absolutely correct and that is a crucial point that I missed. I sort of alluded to it when I talked about how complicated other schemes were for doing this on Ethereum and in other places, and even in pre-DLC Bitcoin proposals, the oracle would always be involved in the contract or the oracle is supremely involved, in that it takes data from the actual world and actually puts it onto the blockchain, in some cases in Ethereum, like there will be actual numbers and prices stored on the actual chain in perpetuity in some of those ideas.
So, this is much more elegant, and this data never actually gets on the chain, it is simply used; the announcement data, the actual real-world events are mapped to cryptographic objects, and those cryptographic objects are used to enable or disable signatures. And all this can be done by the two participants in the contract and do not require the oracle to be aware of the contract. And indeed, if things go well, the settlement and the creation of the settlement and settlement of this contract can look like normal payments, especially in a taproot world.
So yeah, we can now perhaps move on to what our most recent research is, is that we managed to make practical, I would say this is the major outcome, although I think there are some really interesting things inside this, but the major outcome for DLCs is that we can have the oracle attest using Boneh-Lynn-Shacham (BLS) signatures, which are signatures on a different curve, different properties to the elliptic curve that Bitcoin uses for its cryptography. And the interesting thing about BLS signatures is that they are deterministic, and deterministic perhaps is not the right word here, there is a single valid signature for every single message under a certain key; whereas with schnorr signatures or ECDSA signatures, there are many valid signatures for a particular message.
So, if I have one message like, “Attack at dawn”, and I want to make a signature of it with a schnorr signature, I can make that signature and people can verify it, but there will be many valid signatures. That’s because we use randomness when we create the signature, okay? And so, in order to anticipate or in order to map these signatures ahead of time in the announcement, our transaction signatures to these oracle signatures, we have to know the randomness the oracle is going to use to the attestation with the Bitcoin elliptic curve. All schemes on the Bitcoin elliptic curve are going to need this, unfortunately. But with BLS signatures, there is no – go ahead.
Mark Erhardt: So basically, if we were using regular ECDSA or schnorr signatures, the announcement by the oracle would also do a pre-commitment, like the R value or something, that they would use in the signature?
Lloyd Fournier: Absolutely, that’s exactly correct. The randomness that you would use in the signature would be pre-committed to in the announcement. You would say, “I’m going to attest to this event with this randomness as part of my signature”. That’s a decent way of looking at it at least. With BLS signatures, there is a single valid signature for any message under a public key, and so there is no randomness. We can actually do this anticipation without an announcement, and this actually changes the game quite a bit. It’s still the same idea. You anticipate signatures and encrypt your Bitcoin transaction signatures to the revelation of these signatures, but now we do not need to know any announcement. We still need to know that the oracle is going to attest to that thing.
Mark Erhardt: So, you’re basically just committing to the public key of the oracle, rather than the announcement and random value now. Cool.
Lloyd Fournier: Yeah, and we can get the public key once initially.
Mark Erhardt: So you still need to be sure that the oracle will do an attestation about the event then?
Lloyd Fournier: Well, you don’t actually. It turns out that if you just change what an oracle is, right, take it out of the DLC idea a bit, and we just say, “Okay, an oracle is a web service that you can ask it to go to any other website, JSON API or web page or whatever, and get it to fetch the web page content, do the HTTP request and make some BLS signatures on the content, then the oracle doesn’t need – and let’s say the BLS signatures include the content that you’ve asked to assign, but also the request that you’ve asked it to make. And so, you now do not even need to know that the oracle is going to attest to any event, as long as that event is authoritatively published on the web and you know what the URL is and you know where exactly that will be in that HTTP response.
You can predict what the oracle is going to sign, in other words you can think of this space existing beforehand on the web, and you can anticipate what this oracle signature will look like for any of the outcomes for this bit of data on the web and make a contract out of it, without ever contacting the oracle. And then at the time that it comes for the oracle to attest, the oracle has not announced that it’s going to attest to this, you simply ask it, “Please go to this website and sign this stuff”, and then it will do that, and you will be able to unlock your signatures on your Bitcoin transaction. Does that make sense?
Mark Erhardt: Yes, it does make sense. I might have said an API or something explicitly well-defined that gives you an exact piece of data, but it works for me, I think. So basically, you know that you’ll be able to ask a specific public-key-associated service to retrieve real-world information and attest to it, so basically like a web server notary. That’s pretty cool.
Lloyd Fournier: Yeah, it is very cool, I really like this idea, I’m extremely excited about it. It makes me smile, because it really makes this idea of the programmable web real because you can now take any bit of the internet and transform it into a cryptographic object, which you can use to make a Bitcoin contract on. That’s quite a powerful idea.
Mark Erhardt: Doesn’t it though move some of the knowledge about what piece of data you’re specifically interested in back into the purview of the oracle? Because now you’re asking the oracle to attest to some specific data, so obviously you must have been interested in that data; where previously, you could just use published pre-announcements and the oracle would never even notice whether or not you used it?
Lloyd Fournier: You don’t, I mean, the oracle will know when you fetch the attestation in any case, right, because you would have to ask the oracle, “Okay, what was the attestation for this event?” so the oracle will know that you’re interested in that event. This now, you’re right, narrows it down to a particular web page, where obviously most times a particular web page or adjacent field is in fact indicative of the same thing, it doesn’t give you extra information, I guess, over what you would have given before. But there is actually a very good advantage to this, because changing it to BLS means that you don’t need to fetch the announcement. But I mean, you don’t need to fetch the attestation either, actually. If you can both agree on what the outcome was, and you can both go to the web page anyway, you can both settle the contract without even contacting oracle.
So, in the expected case, you have perfect privacy from the oracle itself. Now, you will have to contact the web page yourself, but usually contact, like going to a web page or going to a JSON API, is a rather innocuous kind of action, it doesn’t tell the operators of that JSON API too much, depending on the JSON API, of course. But in many cases, you’re just getting some data which your phone apps are doing anyway, right?
Mark Erhardt: So, the idea is here that you would either require an attestation from both participants, or an attestation from the oracle about the same datum, and since BLS signatures are non-interactive, aggregatable, does that have something to do with that?
Lloyd Fournier: No. So, I mean if you both agree, you just do a schnorr signature, you just cooperatively close the contract like you would with a Lightning channel.
Mark Erhardt: Oh, yeah, you don’t need the pre-committed transaction that was dependent on the oracle at all, yes.
Lloyd Fournier: Yeah, exactly. So this actually gives a rather strong privacy guarantee in the cooperative case. Now, in the non-cooperative case, as I’ve already just said, it’s basically the same. You have to fetch the attestation in the current scheme anyway. But here is where I changed the model slightly even more, let’s take this idea even further. So, instead of just telling the oracle to go fetch this thing on the web page, we can tell the oracle to process it for us. We can tell the oracle to visit many web pages and process it and turn it into an average, for example. So, we can tell the oracle to go to different web pages and run a program on the inputs, and get the outputs and map the outputs to specific integers. If you want the Bitcoin price, it doesn’t have to be the actual Bitcoin price. You could get the Bitcoin price and then map it to some specific intervals.
Let’s say you’re doing the Bitcoin price, every time the Bitcoin price changes $30, I get a little bit more satoshis than you. We make a transaction for that interval, or whatever interval you want, can be a floating point number, whatever. The point is that we can ask the oracle to process this and map it into our contract domain. We don’t have to leave it as a Bitcoin price, we can map it to like indexes of transactions we’re going to use. And for some technical reasons, this is actually much more performant, or at least it doubles the speed of the protocol, which is actually important because it is actually a bit slower using this BLS scheme. So, you can get the oracle and tell it to run a program on a bunch of inputs from different web pages and give you the output.
If you’re thinking about smart contracting programming language, this becomes like a smart contracting programming language, right? We can take data from the web, write a contract, do whatever logic we want to it, output some bits and use that as a kind of circuit to enable certain transactions in our own personal contract, which can do whatever we want, and the oracle won’t know about it. But now, like I was saying, we’re telling the oracle now quite a bit more. We’re talking about which web pages, which data, and how to process it and turn it into our local domain, which will obviously leave some fingerprints there about what perhaps we’re doing. It may even, if you’re not careful, leave some fingerprints about the amount of transactions you’re signing, which may even leave some fingerprints about the values in those transactions being a multiple of something in the contract. You can imagine if you’re doing a thousand transactions, each one pays one person more than the other one and they go up linearly, and if that’s a common thing, then mapping these data to certain intervals which map closely to these –
Mark Erhardt: To the program run by the oracle, yeah.
Lloyd Fournier: Yeah, the program tells them information.
Mark Erhardt: I think that’s still –
Lloyd Fournier: Like I said, we only contact the oracle at all in the uncooperative case, which I think for me, that ameliorates that concern.
Mark Erhardt: I think it’s also strictly superior to only tell the data to one oracle rather than print it in the blockchain. I mean, even if we’re leaving some data with the oracle and we must consider that it might be an adversary, not leaving it in the blockchain for perpetuity for anyone to look up at any time later in the context of the transaction that was executed, is strictly superior.
Lloyd Fournier: Yeah, so hopefully if you’re cooperatively closed, then you never leave any information about your contract whatsoever on the blockchain, obviously. But yes, in an uncooperative place, there may be some privacy leaks. I think it’s not so bad. I think it’s definitely not a deal breaker, this uncooperative privacy issue, and I mean the privacy is fantastic. I mean, like you mentioned, the other protocols leave data in the blockchain, and everyone can see why this money is moving, because the data is there. I mean, the privacy is just almost perfect, let’s say. At least in the cooperative phase, it is perfect, and in the uncooperative phase, you leak a bit of information but it’s not so bad. You don’t tell them what your transaction is.
Mark Erhardt: And it also depends on how you actually construct the request and how you then use the information that you get, given that I mean you could feed it also with additional information to obfuscate how much you’re really interested in and ask for more attestations than you need, or you could separate out crucial parts of it into multiple requests, stuff like that, right? You could probably make it harder to track what belongs together. And then if you combine the information, it does maybe look different.
Lloyd Fournier: Absolutely. You could ask for multiple things you don’t even care about and combine them all together. I mean, I wouldn’t do this in practice, but yes, I think it’s not some fundamental limit either. You could go for the full privacy or decoy kind of privacy if you wanted to, in the case that you have to be uncooperative, yeah.
Mike Schmidt: Yeah, and I suppose if I were an oracle, I would be wanting to charge additional, based on the amount of processing that I’m doing. So, I guess in the case of the obfuscation example, I think Lloyd in your post or in the paper, I think there was the example of the Bitcoin price. So, if you were only interested in the Bitcoin price to Murch’s point, you could also ask for the Ethereum, the Zcash, the Monero price, and then knowing that you’re just going to use the Bitcoin price, but maybe the oracle charges you a bit more for something like that.
Mark Erhardt: In a way, it might also be more incentive-compatible with the model where the oracle pre-commits to attesting to an event and then just publishes the attestation. I don’t see how the service of the oracle would be rewarded at all, but here it’s pretty clear. If you’re running a program specifically for one user, the oracle can obviously require a payment for running the program and giving the attestation, for example, a Lightning payment or something. Make the request, then payment required, and here you see your attestation, your personally crafted attestation. So, I think that it might make sense from a, “I want to run an oracle” perspective and I can actually earn something with it.
Lloyd Fournier: That’s an interesting point. I didn’t think too much or haven’t philosophized on the payment and how you reward oracles for doing these things. I guess it’s an open question. I think BLS signatures, they’re not too computationally intensive just to do. But in running these small programs, I think it’s probably okay to do it for free for a while obviously, but at some point, yeah, I think some remuneration for these oracles would be appropriate if they’re providing a competitive service.
There’s an interesting idea, so I’ve been taking this idea even further. What if we go even crazier with this idea, okay? So, an oracle is a trusted party, and obviously it makes sense to put it into a kind of federation. And with this protocol, as a sort of side note to this whole discussion, we can now do thresholds of oracles more efficiently, at least more efficiently in the time complexity in the number of oracles. So, in the current protocol, it’s exponential or combinatorial in the number of oracles, the number of the nano computation you have to do, whereas this is linear. So, that’s a big win as well for thresholds of oracles. But we have thresholds of oracles and we also have this burgeoning field of these Fedimints, right? And so you have these trusted parties that are oracles, there are trusted parties that are federated mints, and they both probably should be federated, or in some way. So, what if you just joined them together? What if you had an oracle that was a federated mint as well?
Now, this becomes interesting, because now this programming language could deal with actual assets and coins itself, and could actually go from just processing that web data, also issue assets in the contract. So, you could add functionality to the contracting language that allows you to create new assets to transfer assets between people, whoever’s calling the contract may be able to get some coins out of it, if something has happened on a certain website, for example. And then, there’s an obvious way of renumerating, for giving coins to this party for providing the services here, because it’s an e-cash, it’s a federated mint as well.
So, yeah, that’s a really far-in-the-future kind of idea I’m playing with in my head, seeing what we could get away with, if instead of having a blockchain with these smart contracts, which we know how well that is going on Ethereum – depending on your perspective, it’s going very well or very badly –but perhaps it could go a lot better in this kind of federated setting, where you’re using e-cash rather than recording everyone’s assets in perpetuity for everyone to see. It’s just, you can forget about them and you have a trusted federation to be able to verify if you have an asset, but be able to issue assets in the contracts and also put assets into the contracts for later release to other people. And to be able to also consume web data at the same time, you could imagine very sophisticated DLCs that are not just DLCs, right?
The DLCs are between two parties. That’s sort of the fundamental limitation of them. You can’t do what you call a decentralized exchange (DEX), or you have to find other people to participate in the contract with you to do it, and that’s great if you can reliably get this kind of network set up. But it will end up being a hub-and-spoke model. So, with the DLCs, whether it’s this protocol or the existing protocols, you have a hub that will give out these DLC contracts to people and you contact them. But then you can imagine if you didn’t want the hub-and-spoke model, you could have this oracle combined with Fedimint thing have a kind of decentralized exchange on it, programmed using this language, or a decentralized betting service in this case, not an exchange.
We don’t necessarily have to have shitcoins in this, is what I’m talking about. These tokens I’m talking about issuing could be like positions in a certain asset, right? And then when you want to get your bitcoin out, you give back the token that represents the position and it liquidates it and gives you the bitcoin that you’re owed. This is the way I’m thinking, but this sort of just answers that question of how oracles could get paid. If they went to this model, it would be very obvious how they get paid. They get some e-cash back when they execute a contract.
Mark Erhardt: Right. And now the DLC would still be between two parties, but because you have basically a designated market operator, the counterparty for everyone is sort of always available and performs the translation and matchmaking for you, which sort of gets rid of the needing to look for a partner in the DLC.
Lloyd Fournier: Yeah, exactly. So, I mean one way of thinking about it is that right now on Ethereum and other places, they try to do this thing, decentralized exchange. I don’t really like this term, I don’t really like it. So they do this, what is this called, “automated market making” on Ethereum. I don’t really like that idea, I don’t think it works that well. I think it’s like a poor man’s version of an exchange, right? You rather just have a centralized exchange actually. But what you could do is, the order book of a centralized exchange is actually what you want when you’re trading, that’s what you want. And when you’re settling for this automated market making thing, you’re settling for something that isn’t as good. But I think trusted parties can run order books, that’s probably okay. The thing is you don’t want them to have custody of the funds.
Mark Erhardt: Yeah, exactly.
Lloyd Fournier: So with this thing, you could let a centralized party run the order book, but have the federated mint do the custodying, and there would be no friction there. And essentially, the Fedimint doesn’t have to know it’s doing that, right? As long as it has this programming language there, you could go in there and make this kind of consume this website, which is running this HTTP endpoint, which is running the order book, but then make it do the transfer of assets within the Fedimint sort of obliviously, or without knowing that it’s doing that, without having to program anything. The user or the developer of the smart contract would be able to make that relationship. So that’s an exciting idea. I don’t know if any of this works, by the way, just mulling around in my head.
Mark Erhardt: Yeah, that’s super-exciting. Thanks for sharing in depth.
Mike Schmidt: Lloyd, we went over the use cases and the positives and the benefits of this potential technology. It sounds like it’s all unicorns and rainbows. Is there any downside to this; and if so, what are they?
Lloyd Fournier: No, that’s a hard one. In terms of for me, this is strictly superior. Okay, the downside I can find you is it’s a bit slower to create. For example, you’re just creating a simple contract. Let’s say 1,000 outcomes on the price, it’s going to take you like a second or so to compute it. And also that second, it also involves three rounds of communication, so one-and-a-half round trips. So, the protocol itself is more complicated, so that’s a downside. It uses an elliptic curve that is not the one that Bitcoin uses, which means we’re working outside the Bitcoin cryptography stack, that’s a downside. And it involves more rounds of communication. So, instead of just sending a message, “Hey, I want to do a DLC”, “Yeah, let’s do it”, and send a message back, or maybe two or three, you know, one to propose it and then one back and one again for setting it all up so that you have a transaction on the blockchain, there’s a few extra rounds in there.
It’s required because it uses statistical security rather than computational security as the technical reason for it. So, it requires some games to be played, where you send something and then someone gives you a challenge over the wire, and then you take the challenge and do something with it. So, there’s a few extra rounds of communications. So, all the downsides are in the cryptographic layer and the protocol engineering layer, they’re not at all in the architectural or conceptual layer, which is a good place to have the complications, right? I think it’s a good place to have them.
Mark Erhardt: Yeah, because when somebody has implemented the library that does all this, building on top of it wouldn’t be that much more complicated. But it might be a little harder to bootstrap it in the first place. I mean, Ethereum is already there and it’s doing all this already, so people plug into a rich existing ecosystem. But yeah. Well, that was pretty exciting. Would you remind us, why do DLCs benefit from taproot?
Lloyd Fournier: Yeah, it’s just that we do – actually, out of all the things, it probably benefits the least, but it’d be faster, that’s the main thing. The current specification uses schnorr and ECDSA adaptor signatures. With Taproot, you can switch it to schnorr adaptor signatures, which are ten times faster to create. Right now, DLCs are kind of slow. It takes a few seconds to do, especially because of this problem if you’re doing multiple oracles, it gets very, very slow, because it’s combinatorial or close to exponential complexity of adding new oracles to the mix to do a threshold, like 2-of-3, 3-of-5, it gets exponentially more computationally intensive. So a 10X improvement makes some parameters viable that were not viable before.
Mark Erhardt: Yeah. Okay, so I kind of walked into that one. But my understanding was that DLCs originally came basically out of the idea of scriptless script, so it was related to it early on.
Lloyd Fournier: Actually, I mean that’s one interesting thing, because that’s a point I skipped over. The original DLC paper did not use scriptless scripts. So, it’s interesting you say that because it’s not really there, in my opinion.
Mark Erhardt: Okay, I’ll just shut up now!
Lloyd Fournier: I mean, it should be there. In fact, I read the paper and I thought it was using scriptless script adapter signatures to do it. But in the original paper, it wasn’t like that. But when we eventually got to doing a specification, the first thing we did when I got involved was to use ECDSA adaptor signatures instead of the original protocol, and that drastically improves the protocol. So, yeah, I don’t know, maybe the idea was around then, or maybe it was like Tadge came up with it just before that or something. I definitely know he was definitely aware of it. I think the ideas were pollinating each other perhaps in some ways. The original paper doesn’t have the idea in it, which is unfortunate, but in the end, it doesn’t matter.
Mike Schmidt: Lloyd, I saw that it didn’t seem like there was any feedback on the DLC mailing list, which is where you originally posted this, not on the Bitcoin-Dev mailing list, for listeners. So, if you want to read about that, check out the DLC mailing list. I didn’t see any responses there. Have you guys gotten feedback outside of that email post about what people thought of the protocol and the paper you put out?
Lloyd Fournier: Yeah a little bit. They’re interested in it. The people who work with me on the DLC specification effort are, yeah, they’re interested in it. I don’t know, the thing – okay, it’s very difficult. When you come up with, and I’m kind of annoying like this, I just come up with these ideas and it’s basically, you have to rewrite everything. It’s like, “Trash everything you’ve done, start a new thing now”, and it’s not so appealing to hear those kinds of ideas. And when you’re trying to get a product out, it’s also not so helpful to be distracted by that. It’s more like, “Okay, that’s something coming in the future. Maybe that’s not on our plate right now”, and it shouldn’t be. It’s correct to have that approach because it’s not worth just trashing everything you’ve done. You’re trying to build a user base and you’re trying to see how people interact with the technology and learn from that and go forward.
But of course, this will change to some extent how users will interact with the technology just because of the way it allows you to bet on anything on the web. So it’s something that they will have to consider. But I don’t like to just tell everyone to drop everything and just work on my idea. We’ll see how it plays out. What I can say is that my colleagues, my ex-colleagues and people I love working with, they’re the comit.network or CoBloX and they are working on something called itchysats.network, which is already a working DLC. It doesn’t follow the DLC spec, but it’s a DLC technology on layer 2, so it’s within channels that allows you to do Bitcoin – let’s say BitMEX, but non-custodial so you can do BitMEX non-custodial on itchysats.network right now. It’s early, whatever, but we’ve talked about it, and we’re going to work on using this idea in their systems.
One of the main motivations for me is that I’m actually their oracle and it is a pain in the ass, every time they want a new thing to bet on or a new way of doing things, they have to ask me and say, “Lloyd, can you please attest to this thing now? And can you please attest to this thing but more frequently?” It’s a lot of maintenance work looking after an oracle, that especially the people are relying on and money is there. It’s a pain in the ass actually. So, I’m really looking forward to this idea and I’ll be implementing it, just so I can get delete my database, because this oracle doesn’t need a database. It just goes to things on the web it can have a cache maybe to remember stuff that it’s just done so it doesn’t get spammed, or whatever, on the same thing. But I don’t have to maintain the oracle at all, it just needs to run as a service on my server and as long as the process is there, the thing is working, right? And that’s what I’m excited about.
Mark Erhardt: So you’re saying you’re externalizing the work on the user instead of the oracle, and you’re excited about that.
Lloyd Fournier: How rude! It’s not really the user, right? It’s them, the developers. Because the user could. You could write DLC software that could go – the user can propose some Python program that can be loaded up to the oracle and it purports to get some data, or whatever, that you could bet on and then the users can like, “Hey, man, let’s do this bet on this Python program I just wrote”. You could have that user experience. I don’t think it’d be very popular, but really what it is, is it’s about the developers can now add trading on ETHUSD or trading on any other random thing on the internet and don’t have to ask me to attest to it. That’s super-exciting to me.
Mark Erhardt: Yeah, that sounds super-exciting, thank you very much. I think we’ll quickly go over the other news items still. What do you think, Mike?
Mike Schmidt: Yeah, that sounds good. Lloyd, thank you for that. That was very educational for myself and I think our audience, and we appreciate your time. You’re welcome to stay on and comment on any of these other items we’re discussing. But yeah, thank you very much for that.
Lloyd Fournier: Sure, no problem. That was great fun.
Rust Bitcoin 0.29
Mike Schmidt: So, one of the releases this week in the newsletter was Rust Bitcoin 0.29, which is a major release. It has some breaking changes, but also adds some more modern features. So, for folks not familiar, Rust Bitcoin is a Rust library that has a lot of different serialization, deserialization, and Bitcoin-related data structures, and P2P message structures that you can use if you’re wanting to do Bitcoin stuff. I know there is some consensus stuff in there but it is not meant to be a consensus implementation. And so, they’ve added compact block support and improvements to taproot and PSBT as well. Murch, any commentary on Rust Bitcoin in general or the specific release?
Mark Erhardt: I’m not super-familiar with it, but maybe I think BDK and LDK would probably use Rust Bitcoin, and Lloyd would know; is that right?
Lloyd Fournier: I haven’t looked at the release notes, but I can tell you one thing that’s in there that’s really good, is we now have a type for locktimes, okay, and sequence numbers. And so, instead of just having to fiddle around with like, “Oh, is this locktime good enough to spend this coin?” and then having to say, “Oh, Bitcoin locktimes, they can be either a height or an actual timestamp or median time passed in technical term”, and you have to do this checking yourself in prior versions. But now they have a type for it, you can just ask, “Is this locktime satisfied by this number? Like, if I put this number in this transaction as the locktime, will this be able to spend from this coin?” You can now do that. That’s something I’m pretty happy about, and there’s something similar for sequence numbers, though it needs to be a bit further developed, but that’s coming, too. Yeah, that’s what I noticed from it.
Core Lightning 0.12.0rc2
Mike Schmidt: Great. Thanks, Lloyd. I think we could skip over the release candidate for Core Lightning (CLN). I think anybody who is using CLN should test out the release candidates and provide feedback accordingly.
Bitcoin Core #23480
In terms of notable pull requests merges, the Bitcoin Core #23480, it adds a new descriptor, which is rawtr(), raw taproot descriptor, for referring to an exposed key in a taproot output. Murch, I think you may have a little bit more familiarity with the use cases here and exactly why rawtr(), raw taproot descriptor is useful?
Mark Erhardt: I must admit that I’m a little guessing here, but from what I read in that, it sounds like it’s mostly useful to do watch-only wallets. I think if you want to import a single key otherwise, you could use other descriptors already. But, yeah, I don’t know for sure. Sorry.
Mike Schmidt: Yeah, it sounds like there’s some edge cases where this could be preferred, and I think that’s sort of indicated when you see this raw prefix, that there’s likely something else if you’re an end user that you could be using. But if you’re someone building on top of these descriptors, maybe you need the raw variant in order to do some of these edge cases. I think some of the use cases were around vanity addresses or cooperative tweaking of keys.
Bitcoin Core #22751
Bitcoin Core #22751, this is a new RPC. So, Bitcoin Core includes RPC functionality, which means you can call certain functions, get certain functionality from outside of Bitcoin Core by using these RPCs. You can get something as simple as just getting a balance, what is the balance of the wallet; there’s an RPC for that and a variety of other things. This PR adds a new RPC, which is simulaterawtransaction. What you parse to that function is one or more unconfirmed transactions, and what is returned is essentially what is the delta on the balance of the wallet going to be if those transactions were confirmed, so all the inputs, all the outputs, what is this going to do to my wallet balance? Murch, any thoughts on simulaterawtransaction?
Mark Erhardt: I think it might make it easier to automate some co-signing stuff and things like that, where you want to know whether a transaction actually benefits you. But other than that, I don’t see an immediate use. We did have instagibbs up here for a moment and I was wondering whether he wanted to chime in on the previous, but he’s gone now.
Mike Schmidt: He’s here.
Mark Erhardt: Oh, is he? Twitter is so weird for me right now.
Mike Schmidt: Instagibbs, I’ve given you the invitation to speak if you so choose. Hola!
Greg Sanders: I was having trouble with my headset. So, the one list, going back to the raw taproot thing, the one thing it does allow you to do is, it allows you to understand and sign for the keyspend for a taproot output. Even if you don’t, for some reason the wallet doesn’t understand what tree was committed to it. So, I don’t know if that’s like a compelling use case, but that’s one. Because alternatively, you’d have to put in like the raw address or something, but then it wouldn’t understand what key to sign for pretty much from a wallet architecture perspective. That’s it.
Mike Schmidt: Okay, yeah, that makes sense. Thanks for chiming in there. That was one that I wasn’t entirely sure of the use case this far, so thanks for enlightening us.
Greg Sanders: Yeah, I think there’s also some debate on for PSBTs adding these kind of things, like partial information for some situations. You wouldn’t need all the information to do some things, sometimes you do, so I think Pieter and others have been debating this.
Mike Schmidt: Thanks for jumping in. We have a couple minutes left. We’ll get on to this Eclair PR, so this is Eclair #2273, and this is one PR of several that is working towards dual funding in LN. So, in the traditional funding model, you would have a channel in which it’s single funded, and that’s sort of v1, if you will, in which if I’m opening the channel, I am contributing the entirety of the channel balance myself and then that channel goes one way. Whereas in the dual-funding model, which folks are working towards speccing out and also implementing, not only would I be able to put in funds going one direction, but the receiver, or I guess the other end of the channel, would also be able to put in funds as well. So, right off the bat, either of us could be sending each other satoshis along that channel, it’s not just one way.
Mark Erhardt: It’s funny that you’re saying that the original conception was that channels would be opened by one side, because very early on, I think the idea was always that both sides contribute funds to the channels and both start with a balance. And then it just turned out to be really much harder to implement it that way. So, it’s sort of coming back full circle to the original idea, where now both parties chip in and open channels. I think it is related to the push on first open, sort of allowing when you open a channel to have balances for both sides, even when one person funds the channel, but inherently it’s a little different. And one thing I just realized that I’m kind of excited about is, it will increase the number of transactions where multiple parties contribute funds to create a transaction. So, in a way, if dual funding takes off, it will be more payjoins on the network, and that’s kind of cool.
Mike Schmidt: Yeah, that’s a good point, I hadn’t thought of that. That’s a nice side effect. And I don’t exactly know how this would go out, but I would think there would be a fairly large adoption for dual funding once it’s implemented and fairly solid, but it’s a great point.
Mark Erhardt: I mean, it makes this problem where, “How do you get inbound liquidity?” just much more easy, because you could then have a marketplace of people with similar amounts opening channels to each other. They would start with a balance on both sides, would immediately be able to forward, and, yeah. I hope it does go through. It’s more complicated than single open channels though.
Mike Schmidt: The next PR here is also Eclair, and this harkens back to Newsletter #211 that we covered a couple weeks ago, which is this htlc_maximum_msat field, which I think we covered as a BOLT previously, and this is Eclair actually implementing that change. And what that is, is simply the channel requires a field of what is the max size Hash Time Locked Contract (HTLC) that I will route through this channel? And everybody, I think, or most implementations, were already sending that. So, this is just a formalization of that field being included. And Eclair will, I believe, reject channels that do not include that field moving forward.
Mark Erhardt: As a reminder, that field is sort of also a proxy for the amount of funds that people allow, or channels allow, to be in all HTLCs open at the same time, and that in turn is protection against probing; that’s the word I’m looking for. So, when you can basically lock up most funds in a channel, you can find out how much money exactly is on each side of the channel. And if you limit how much money can be held in open HTLCs at the same time, you can diminish the amount of information such attackers can glean from the channel. And that’s why it’s being formalized, or one of the reasons it’s being formalized.
Mike Schmidt: The next PR here is an LND PR, and it basically is a change to use taproot everywhere possible. And so, where LND might use taproot is like if you’re sweeping funds from the wallet when you’re funding the channels, and there’s some other watchtower-type use cases in which they’re going to use taproot now. So, that’s good, taproot adoption is good. Murch, any thoughts on the details here?
Mark Erhardt: I think it’s probably not going to be used when funding channels because taproot-based channels are not implemented yet. But I think it generally defaults to basically all the change outputs and automatic closing transactions that LND creates will use P2TR outputs. So, we might see a little bit of an increase of P2TR outputs on the network with the next LND release.
Mike Schmidt: Excellent. The next PR is LND #6816, which is really just a documentation PR, and it’s talking about zero-conf channels. So, a zero-confirmation channel is one in which the funding transaction has not actually confirmed in the blockchain yet, and there’s certain use cases where it may be appropriate, where there’s trust, where you can actually use these zero-conf channels before the transaction actually confirms. Murch, any thoughts on zero-conf channels or the associated documentation here?
Mark Erhardt: I haven’t read it, sorry. No, no idea.
Mike Schmidt: It’s just documentation. Cool. And the final PR this week for Newsletter #213 is BDK #640, which is just an update to the get_balance function, and it returns the balance separated into different categories. So, there’s available balance for confirmed outputs; trusted-pending balance for unconfirmed outputs from the wallet itself, so in the example there would be like a change output; and then a different category of balance for untrusted-pending balance, that would be unconfirmed outputs from outside wallets; and then an immature balance in the case that this is a miner and there is a mining output which hasn’t quite reached the 100 confirmations necessary to spend it. Murch, any thoughts on the categorization of get_balance in BDK?
Mark Erhardt: That sounds good to me. I was, when I read that first, thinking also it might be interesting to have effective balance, which estimates the current feerate and then deducts the cost of the inputs or the whole UTXO pool from the balance and tells you how much you could actually spend, based on the current feerate. But I guess that’s a future PR or something!
Mike Schmidt: That’s such a Murch response, I love it! Always thinking about fees. Alright, Murch, anything else that you wanted to discuss from this newsletter or any other announcements we should make?
Mark Erhardt: No, I think we’re good. We’ve got it all covered. Thanks instagibbs and Lloyd for joining in. I think, see you next week.
Mike Schmidt: Yeah, that sounds good. Thank you all for attending, and yeah, thank you instagibbs and Lloyd. This was a great conversation, and look forward to talking with you all next week. Cheers.
Mark Erhardt: Cheers.