TFTC: A Bitcoin Podcast - btc++ Austin 2025 Live Desk Day 1

Episode Date: May 12, 2025

Archive of the live desk at btc++ Austin 2025

We're live. We're live. We're live from Bitcoin Plus Plus, Menpool edition in Austin, Texas, sitting down with Average Gary and VNPRC. It's great to be here. I got it. Always happy to be here. Gentlemen, exciting week for this conference. Do we think Lisa created all the controversy around OpReturn to sell tickets to this? No comment. 100%. I heard she was Satoshi, actually uh the the lead up to this has been very exciting but i think as i've gotten here i was here yesterday we have an event going on at bitcoin park austin about energy and mining for me personally the conversation here versus over there is vastly different there's a lot of uh a lot of very passionate people on both sides of this

op return argument but i don't think we're going to solve that issue between the three of us today in this 15 minute conversation i'm more interested talking about what you guys are building with hash pool i think let's describe the problem you guys fork stratum v2 uh what we call it the implementation and you're creating a mining pool that incorporates e-cash mints to aggregate and account shares and then do payouts. Why don't we start by describing the problem that currently exists and how you guys are trying to solve it. Yeah, absolutely. So yeah, one of the issues with centralized mining pools is they're not terribly private and they don't do small scale payouts terribly well. You can go to a couple of mining pools that have lightning payouts. This is sort

of a heavyweight solution because you have to run a lightning node. eCache is a new technology uh that's coming up it's not new it's like the first first cryptocurrency in like 1980 by david chom it's a very old idea but it is relevant now because of bitcoin rediscovered and re-implemented yes rediscovered it's having a renaissance these days i think thank you for the uh technical correction well actually is what this all conference is all about i think yeah yeah nerds So, what eCash brings to the table is not only does it have extremely good privacy, it's got theoretically perfect privacy using cryptographic blinded signatures, but it also enables micro payouts, which sort of even smaller scale,

even more lightweight than even possible in Lightning. And that's not like a lot of people compare eCash and Lightning. I see them as complementary. eCash actually sits on top of Lightning. The Cashew spec, which we are implementing in HashPool, sits on top of Lightning. It builds on Lightning. It makes Lightning more extensible and more private. But sort of the primary thing that we're doing in HashPool is we're representing mining shares as eCash tokens.

So it enables us to build an accountless mining pool. Normally, for a mining pool, you create an account. Even in the best case scenario, like for example, if you mine to Ocean, my personal favorite mining pool, you still put in a Bitcoin address as your account. It's a pseudonymous account. You don't have to associate your name or KYC, but it is an account, which means that the mining pool has a database within their server, and they're tracking what shares you submit and what payouts are due to that account. eHash lets us sort of flip this. eHash is the eCash token represented by a mining share.

we can actually delete this database off of the mining pool side and instead issue a token. And this is an e-cash token, so it has fantastic privacy properties. And the way it works is when you submit a share, you get a token back. You can store it in your wallet, and you can trade it to someone else. This gives us a couple of really nice benefits. The mining pool does not know who is mining to it. It does not know who is trading these shares to whom,

and it also does not know who is redeeming these shares for Bitcoin payouts. Once the share window expires, the payouts are due. Could be someone who bought the eHash, could be the person who mined it. Nobody knows except that person themselves. It's a free and open hash rate marketplace basically with no accounts and permissionless. And one of the other things that I think it helps accomplish is if you look at the centralization of Bitcoin mining, one of the biggest things is economic scale the viable economic scale of a mining operation and historically we see that just you need fat bags to do fpps which i think is a dying payout

mechanism um and i'm not alone in that thought but you see these huge pools basically floating credit on bitcoin because that's what fpps is it's saying hey if you have bad luck mining we're going to make you whole because we know that you're a hash rate operator and you have bills to pay on a monthly basis and so we can shrink down the economic viability scale of those operations we can make it better so with hash pools that one of the concepts that i think is going to be really robust is you and all of your friends with you know 50 bid axes could get together and your local og can help float the bad variants of your pool and by the way you can upstream it to a pool like ocean and sort of the smaller economic viability becomes realizable and

that's at the pleb level but then also at you know your what is it bob uh the the rabbits and the horses right so like rabbits and horses why can't they run their own hash pools or do these smaller scale things and get somebody that's not you know the the giant bag holders that like amp pool and some of these other big pools foundry are running that are required to do fpps you can have smaller capital holders actually deploying resources to to float that bad luck variance right the mining in the mining business yeah so you can essentially get the same benefits of pay fpps and since that you're getting paid out pretty consistently a share gets you ehash tokens that you can then sell on a secondary market and so you're basically creating immediate liquidity for

your shares and instead of depending on the treasury of the fpps pool that you're mining to you can sell those shares on a secondary market to somebody who's going to take that variance risk In the secondary market, because e-cache, the way e-cache works, it's just data. It's a bare instrument. So the thing is the token. Because the way that works, I can send you a signal DM with e-hash in it or e-cache in it. So it's a very, very broad market that anybody can participate in. And you can send it an email or however you want.

So that's, I think, one of the biggest unlocks is the marketplaces for hash rate historically, to my understanding, have been pretty limited. But now it's like this will blow it way open. Yeah. One of the problems with FPPS payout mechanisms is that the pool assumes all of the risk. And this is sort of a toxic combination because they're also aggregating hash rate and pricing hash rate. They're doing all three of these roles, which just creates a massive centralization pressure so that we get fewer and fewer pools. They're bankrolling the operation with either they have a lot of money in the bank or they have a profitable business that they can just plow those profits into their mining pool and sort of dominate the market but this is really bad for bitcoin because you want more block template producers and not less block

template producers so one of the things we can do with ehash and other technologies as well we can push that risk down a layer from the pool to a robust marketplace of ehash traders and these people can you can have a professional trading class develop in a decentralized way and if each hash pool is submitting their own block template now we have a top level pool that doesn't price hash rate doesn't submit block templates all they do is aggregate hash rate and i think this is a far healthier mining pool stack which is ultimately where we need to go yeah correct me if i'm wrong but there's been this long running goal in hash rate markets to create these hash rate futures products. Yeah, there's a lot of people talking about ways to do this. Luxor has done it.

Biduda had an OTC desk that was doing it. I'm not sure if that's still operational. Bob was just here explaining that he's been working on this as well. And correct me if I'm wrong, but it seems like this is this is the path to actually get to those markets because you need need instant liquidity to those shares the whole problem up to this point has been like a two market set like a two-sided market problem where who wants to buy it and who's willing to sell at the same time if you make the shares more granularly and immediately spendable it seems like the potential for these hashrate futures markets to develop yeah they're extremely they can be extremely small in value so you get very very granular trades and they also

um shit i lost my train of thought one of the cool things i i just like the the hypothetical situation or like the mvp that we're looking at is you know you're mining upstream to ocean with your with your proxy pool your proxy hash pool and you're getting bolt 12 lightning payouts well a cashew mint right it rides over the lightning rails right so i can pay you in lightning over a cashew mint um and so with the bolt 12 payouts from ocean we get paid out right it comes inbound to our lightning node that's running on the cache you meant and now all the people that have been hashing with this hash pool are have the exact liquidity they need to go outbound to wherever they they want to hold their sats so if you don't like e-cash because it's you know the best shit

coin if you don't like it though you don't have to hold it you can go to your lightning node because it's interoperable with lightning or you can take it maybe you save up all your e-hash until you have enough for a utxo right so it's like this scaling text and i think a lot of people are exploring this with eHash and eCash in general, where it's kind of like a tiered custodial to self-custody transition? I remembered my thought. The other really, really big benefit of eHash over these other derivative instruments is it's very low level, close to the protocol, which makes it very liquid. If you're talking about like legal contracts, like some people have tried to build, you know, that brings in the human trust layer. If you're talking about DLCs,

you want to have now that's an on-chain contract so you're talking about larger amounts with ehash we can do very small amounts very close to the protocol very liquid yeah so i think one of the big problems hash rate derivatives markets to date is getting back to the two mark two-sided market problem is historically you just found one miner mining to a certain pool and a buyer of their hash rate in the future this would enable many to one in terms of seller to buyer many to many many in the cashew protocol if anybody's not on noster the cashew protocol and noster are very very tightly integrated like there's a whole bunch of support for it so all the all we've changed in cashew uh we're using the spec but all we've one of the things that we the only thing you need to

change is create a new currency unit right and and cashew already natively supports multiple currency units so instead of having sats we have the hash currency unit and so i had like a couple lines of code change to a cashew web wallet to let it use hash and later on tomorrow when i'm doing my demo i will live redeem e hash into a web wallet yeah no i was telling you uh made you a little uncomfortable but this morning when i was thinking about this interview i was in the shower and it's like the full circle like hash cash to e hash is a little bit poetic because for those who are unaware bitcoin mining algorithm runs on hash cash shop 256 invented by adam back before bitcoin yes but it seems like it's all coming full circle and these interoperable protocols whether it's

bitcoin at the protocol level lightning cashew mints uh and now noster as a communications layer like it's making all this possible yeah the problem with hash cash was it wasn't there was no way to create a limited supply but if you turn that into mining rewards now we can use bitcoin as the absolute hard money limited supply i think of uh i think of hash rate shares as sort of crude oil bitcoin is refined jet fuel the mining pool is the refinery you can mine up some crude oil and actually trade crude oil futures and just hold on to it and it'll turn into bitcoin yeah and so thinking about solving the overarching problem that exists at the mining pool layer right now which is fpps pools basically creating these large treasuries that other pools are forced

to proxy into because they need to reduce their own variance risk to make sure they don't go bankrupt. Seems like that's led to block template issues where the trade-off that you're making when you proxy into one of these bigger pools is you can leverage our treasury, but we make the block template. So solving the treasury issue first, do you think this leads to more robust and diversified block template creation at the pool layer? Yes. And I think the secret sauce is that we can give an economic incentive for people to produce their own block templates. People have been working on, Bitcoiners have been working on new protocols that allow you to submit your own block template as a miner. Stratum v2 is the first one. Datum came later. So these things are out there. The

question is, why would you run this? And I don't think we've had a good answer until now. But now if you can run a hash pool, you can actually run one of these decentralized hash rate markets, take a little fee off the top, and you are economically incentivized to run your own mining pool and submit your own block templates to the network. So as an individual miner, because you can get that fee by providing your own block template, that's the incentive? The incentive is for an operator of a hash pool instance to mine their own block templates. There's different ways to do it. There's different architectures.

I'm giving a talk tomorrow about different architectures for a mining pool stack. And one place I want to go to is I want to have an aggregator pool at the top level, like I described earlier, that only aggregates hash rate and does not produce block templates or price hash rate. And then we can have sort of sub pools, or I call them proxy pools, hanging off of that large aggregator pool. They each control one Coinbase output of the blocks that are produced, and they can do whatever they want with them.

In Hashpool in particular, you take that Coinbase, you spend it to the Mint, either in a Lightning transaction or however you want to get that money into the Mint, and then they can issue e-cash shares from that, which again is interoperable with the Lightning network. So you can take those sats as e-cash. You can take them as a Lightning payment to your Lightning wallet. Ultimately, if you're a big miner, I think you should be able to do on-chain as well. But we're going to start with the small miners.

It's all accounting. Like at the end of the day, it's all accounting. How do you account for the hash rate, the shares? How do you account for the variance of your luck in mining? And so I never thought I'd be working on accounting software, but this is where we're at. And even, you know, even the corporate big miners, you know, they have auditing and compliance requirements for publicly traded reasons. This is a way that you can have a third party auditor do a private blinded auditing of hash

rate. So if a big miner is running a hash pool for just internal purposes, they could have a third-party auditor audit cryptographically that all the hashes were accounted for. And this is one of the things in Cashew, like the hard problem with Cashew itself and just eCash in general is the liability side of things, right? Like how do you prove the liabilities? Well, you have to trigger a bank run. What that means is if you can get everybody that's participating in a hash pool to all come together and say, hey, here's all of my eHash, you can solve that. If the three of us are mining in a hash pool together, if we each take our percentage of the eHash and we all consolidate it together and the mint itself publishes all of the cryptographic proofs that it issued to us, we can have a 100% cryptographic reconciliation of these are the shares that I mined. here's the secrets that i use to create the blinded signatures and then here's the blinded

signatures from the pool side or the the mint side and you get this holistic automated way because it's just data yeah and so in terms of getting this adopted and making this the sort of go-to way in which individual miners interact with mining pools what's the order of operations in your mind um we're kind of uh like scoping out the prototype right now figuring out what the protocol looks like and building it out we got a uh a demo ready and gary's gonna demo it very alpha right now it's very alpha it's hacky but it works um which is what really counts i think the path to adoption is the first thing we want to do is build a uh a proxy pool that uses ecash and we're going to call it axe pool the idea is to give bidax runners somewhere to point

their hash rate because there's a big usability problem in bidax you want to onboard people to mining you say hey buy this buy this bidax it's 100 150 bucks you can start hashing today and people get real excited there's natural demand for this which is one of the reasons i started working on this but uh the next question they ask is where do i mine to and there's just not a good answer if you mine to one of the monolithic mining pools with the bidx you don't produce enough hash rate to to really ever be able to afford a payout on chain you can um do the hard thing and set up a a core lightning node provide liquid you gotta like peg in your bitcoin into the lightning network and start receiving um both 12 payouts on a few mining pools this is like a

this is a lot for someone you just onboarded a bitcoin who just heard about it and is want excited to want to get it started um i think we can solve that usability challenge using the cashew protocol and ecash protocol to enable people to just buy a bidx configure it to their bitcoin club runners bidx or axpool instance and then they just uh download a cashew wallet to their phone i think did i send you some some e-cash recently in nashville in nashville yes yeah how long did it take to get onboarded to uh cashew uh like 30 seconds 30 seconds so they can buy a bidx configure it install a cashew wallet in 30 seconds and start receiving sats to their cashew wallet basically instantaneously that's where we want to go i think that's the path to

adoption yeah small local meetups do you see pushback from the incumbent large mining pools or Or are they incentivized to adopt something like this as well, do you think? I have not seen any direct evidence of pushback. But there's no reason that we can't work together. All we need is a PPLNS pool to hang off of. Ocean is the only game in town right now. Well, you can hang off of an FPPS pool if you wanted to because what you're doing is you're taking the accounting and the risk

and taking it down a layer. So instead of putting it all in the pool, you as a hash pool operator, can you you can float all that risk or you can find a market buyer to float that risk for you and so it's reducing the scale so i think the bigger miners i don't know if this is a threat to them this is not i don't see this scaling up to you know double digit percent of the network hash rate it would be pretty sweet if they did but i think it'd be a bunch of individual hash pools doing that but it's a completely different business model that it's targeting right it's a smaller economic scale i think one one caveat i thought about you can technically

hang one of these off of an FPPS pool. I don't think it makes a lot of sense though, because as we discussed, the FPPS pool assumes the risk and prices the hash rate. So now we're going to pay a fee upstream to a monolithic pool to price our hash rate for us. And now we're going to add a hash rate derivatives instrument to price hash rate again after they've already priced it. It just doesn't make economic sense to me. And what is the lower threshold of miners participating in hash pools do you think is necessary to ensure proper decentralization at the mining pool layer like is it two percent like how many blocks per day would i think one percent is one percent is one block a day it's 144 blocks roughly every day so one percent would you get to one block a day

um is that enough i don't know uh you know how how are they reorging but i think we start small and we grow organically um one thing that i think is really in our favor we've got some really strong tailwinds and that is uh hash rate heating so if you think about it today large mining operations are paying money for electricity so that they can mine bitcoin so they have a positive cost to mine bitcoin if we start reusing all the electric heaters in the world talking pool Cool heaters, water heaters, HVAC systems, anything that's low-grade heat, meaning you don't need to boil water. This is a tremendous energy market. It's way bigger than all of the centralized, you know, large-scale mining operations out there.

I think what's going to happen in the future is the cost to mine Bitcoin will trend down and eventually go negative. and that really instills me with confidence that if we start small, we enable this technology, we let any pleb buy a BidX or buy an HVAC system and start taking those rewards as easily as they can with the UX improvements, I think we'll naturally just grow to take over the world and kind of deprecate the old centralized business model.

To put this in the context out there, I think Tyler, who wrote the book with Brains on Bitcoin heat reuse, the stat he threw out when we were all in Nashville earlier this year is if you take 2% of the HVAC market alone and convert some of that to Bitcoin mining, it would double hash rate alone, which is insane. It's insane. Well, the tailwinds too on the dev side is we're not building,

like we didn't just build this. It's a fork of Stratum V2. So Stratum V2 is the newer mining protocol. They have a reference implementation. We forked that and shoved CDK, which is another open source cashew development kit tool. And so we kind of brought these two forks of different projects together. And I see this across sort of the Bitcoin mining industry, especially with the 256 Foundation, what they're doing. There's a lot of open source dev coming to the mining industry itself.

And we're just one of the many projects that are sort of accelerating this. Well, thank you, gentlemen, for doing it. very important not only on the mining pool side of things i think it's critically important there but i think i've been loud about this on x on noster on the show for quite a bit i'm extremely bullish on the reemergence of e-cash on top of bitcoin i think the application that you guys are putting forth is an incredible validation of leveraging these interoperable protocols in an an open source way to do things that make bitcoin better um yeah that's what we're here for fix bitcoin take over the world yeah appreciate it any parting thoughts for the stream gentlemen

uh no uh be peaceful not harmless and that's uh intellectually and physically keep hashing keep learning keep hashing where can they find out more about hash pool hash pool.dev you can also follow uh us on twitter at btc hash pool awesome gentlemen thank you first of the day that was a great one yeah it's been an honor thank you really made my life easier we set this off on a good on a good foot i think so start strong yeah let's start strong all right if you guys need to go on i can ad lib it here and uh i'm sure i'll see you we'll see you around all right and in the conference thanks everybody funny faces thank you next interview all right that was anybody listening right now we just talked about if you're new to bitcoin

and you hopped into that conversation like what the hell are they talking about to break it down into three simple parts mining pool layer miners to make sure they get paid out on a consistent basis pool their hash and point it at a mining pool that aggregates that hash and does payouts over the years that has become somewhat centralizing in the sense that a lot of these pools are becoming large they're creating block templates and last year we found out that pools that we thought were separate from each other are actually just quasi child pools of large pools like f2 pool and so they're proxying in to leverage f2 pool or excuse me uh amp pools treasury and in return individual miners are saying all right ample you can make the block

template which could lead to censorship pressure pressure in the future what they're trying to solve by combining e-cash via the hash pool protocol is to solve that problem by enabling individual miners get paid out automatically in the sense that they get e-hash tokens that they can sell for bitcoin and not have to plug into a larger pool and it looks like we've got our next guest ready to come on super test net thank you for joining me sir you are welcome how are you marty doing well great how's your day been so far so good i've gotten to speak with all my heroes I got to see the great people who come to these Bitcoin++ events, got to talk to Nifty, and now I'm sitting next to Marty Bent.

So, day couldn't be better. What are you most interested to get out of these three days at Bitcoin++? The things I just mentioned. Yeah. I like to talk to all the smarties and learn what they're doing and get ideas from them. I was just talking to Brian Bishop in the other room, and he was like, Hey, you want to win the hackathon? Here's an idea that could win you the hackathon. and just stuff like that, you know,

just finding inspiration and sources of new ideas for making new things. What is the hackathon? What is the idea? Oh, his idea was a space shooter game where it pays you sats if you do really well. So, like, you control a spaceship and you're shooting other spaceships,

multiplayer thing. Okay. So that's his idea. Yeah. What are your thoughts on the OpReturn discussion that's been happening over the last couple of weeks? I'm on Team Pro Filters. keep the filters enhance the filters um let's try let's fight the spam instead of capitulating

that's that's my that's what i want to do anyway how would you describe the problem and why is that your conclusion um there's a bunch of i think something like 40 percent of transactions since the inscriptions drama are contained spam contain jpegs contain token data contain just things that aren't Bitcoin. And I believe that with more and more of this happening, the Bitcoin Core developers, I think, many of them are like, it's a losing battle. There's no way to recover from this situation,

so just stop fighting it, I think is what's happening. I guess that gets into the broader conversation of the subjective views on what is bitcoin what is it to be used for what is block space to be used for is it specifically for money i want to mention i don't think i'm a good person to ask about other people's opinions okay so if i recommend getting a bitcoin core developer on here and asking them what their opinion is rather than asking me who disagrees with them to state their opinion because i probably got it wrong uh i could see that but like in terms of what is your what should block space be used for is does bitcoin have a specific money use case

that was clearly sort of laid out when the protocol was launched, or can it be used for these other use cases like inscriptions as well? The thing that I most like to see Bitcoin used for is something called an economic transaction, which I define as a transaction that sends money from address A to address B in as few bytes as possible and doing nothing else. That's what I think it should mostly be used for.

I think it's also good, at least until we have something better than coin joints, I think you can also do coin joins. Coin joins are not very economical, at least not the way they're currently done. They actually take more space than they could to send money to whatever address you're trying to send money to. But I think that because of the privacy benefit you get from doing a coin join outweighs the fact that it's not very economical. So I'm willing to say, okay, that's fine. Some other things, I think multisigs are not as economical.

They can be now that we have Frost. You can do them in as few bytes as possible. But it's also easier to use regular multisig. So there's a few things that I think I am willing to make exceptions for. But generally, send your money in as few bytes as possible. And inscriptions and stuff is just adding bulk data that's not Bitcoin to your transaction. Get it out of here. What is the threat to the network if we don't get it out of here?

It makes Bitcoin harder to run. And if Bitcoin is harder to run, then legitimate users who might run the software don't because they're like, this is just a bunch of spam. I don't want to download a blockchain that's just a bunch of spam. So they don't run it. And then the network becomes controlled by the few people who do run the nodes. Outside of this discussion, what most interests you right now in Bitcoin? and I noticed that you've been, I don't want to say trolling, but you've been interacting with the Monero community and trying to...

I'm very interested in improving the privacy of the Lightning Network, and a lot of my recent projects focus on privacy tools. And one of the things that I think is currently true is that I think the Lightning Network offers better privacy tools than Monero does. So I try to illustrate that through... I've been trying to illustrate that through challenges with Monero people, where I say, how about this? You send me a Lightning payment.

I will send you a Monero payment. We'll see who can trace it. And so far, everyone who's taken up the challenge, I've been able to trace my Monero payment. They have not been able to trace their Lightning payment. But a lot of the Monero people have started to say, we're not falling for this trick. We're not going to take the challenge.

So we'll see what happens. In terms of privacy on Lightning, what is the state of that right now? Well, I think it's best in class. I think it's better if you run your own node and are careful, then I think you get better privacy than you can get anywhere else. But there are improvements to be made, and one of the things that I am currently excited about is lengthening the number of hops on a route.

I think that I finally decided that the more hops you have on a route, the better your privacy gets, so I want to increase that, and that's one of my current projects is doing that. Is there a cost tradeoff there? Typically, yeah, you pay more for more hops. And what is the magnitude of that cost increase? It depends on what routing nodes are charging. Typically, they don't charge very much, but, yeah, it depends.

Yeah, I saw LM Big released his revenue from some of his nodes, or is there her nodes, a couple of weeks ago. I believe he was making like 0.05% APR on all the liquidity he or she provides the network. Okay, yeah. So you could be making that 0.05%, dear user, if you run a Lightning node. Do you think the combination of Lightning

and these e-cash protocols that are beginning to become more popular can increase the privacy? Yep. Yeah, I think one of the things that I'm really grateful for eCache for doing is bringing to my attention the concept of blinded servers, which that was the first time I think I'd ever seen one, was in eCache in general. So learning how those work, and now I try to implement them in other projects of mine, and I try to take inspiration from that. So thank you, Cashew, for bringing good ideas to the forefront, and FedEmit too. In terms of building on this good ideas concept,

what products do you think are desperately needed now to achieve the ultimate goal of better privacy in terms of consumer apps? What can app developers do to improve the UX or UI or the architecture they're leveraging to bring better privacy? First, I want to see, well, let's add a button that's like toggle this button

and you add privacy to your transaction. And then I want to see, well, let's just have that on by default And you don't even need to show it because it just happens automatically. But making tools to where you can improve your privacy is like a good first step. And so you make the tools where people can improve their privacy. Then you create wallets that give people the option to use those tools. And then you make wallets that do them automatically. It's privacy by default.

And you mentioned CoinJoins earlier. What are your thoughts on PayJoin development kit and potential of that being adopted by more exchanges? because obviously Bull Bitcoin is using it, I believe. Do you think at the protocol level, widespread adoption of PayJoin could improve inherent privacy on Chain Bitcoin? Yeah, I think it can. I haven't looked into PayJoin dev kit specifically,

but if it's like the other dev kits that have come out recently, great. One of the nice things about PayJoin is that it breaks the common input ownership heuristic, which is a privacy weakness that Bitcoin has. And another great thing about it is that its usage is undetectable. You can't tell if somebody is using it or not. So if, let's say, we get, I don't know, 20% of transactions using PayJoin, then you can no longer reliably say,

we know that the person who sent this input also owns the one right next to it, which is a common way that Bitcoin is traced today. So when you make these undetectable tools, it's really good because you make it so that as soon as a significant number of people are using them, all the anti-privacy heuristics fall apart. Or not all of them, a lot of them. That's something I'm really interested to see is do the exchanges because there is an economic argument to implement pay join as well because you can save on fees as an exchange that's batching many transactions on a day-to-day basis. And so obviously, historically in the industry, regulated exchanges have been a bit wary about sort of wandering into the privacy discussion. But I think now with PageWayne specifically, if you can add that economic argument that, hey, shareholders, investors, board members, we're actually going to save money by implementing something like this, and there's an economic reason to do it. And that's something that excites me is creating economic incentives to increase the privacy of users of the Bitcoin network.

There's that, and I think that's a great reason to do it. If you can save money by doing more private things, then wonderful. I think there's also a legal argument for it. In some countries, you have a legal requirement to safeguard your users' privacy. I think Europe, the Eurozone has this thing called the GDPR, and it says, like, you have to protect your users' privacy. So I wonder if exchanges could start doing coin joins and say, we're just following the law. Like, the law says that we're not supposed to know about our users' finances, and this is the way to do that in Bitcoin. I think that that would be an interesting legal avenue to approach it from.

I also think they probably would say, you're not allowed to have it not be private to you. You have to know where the money is coming from and going to. Oh, look who just showed up. Dusty. Thank you for joining us, sir. Dusty Detmer. It has been a while. I ran over here.

It has been a while. How have you been? I've been really good, man. Really good. Are you still working on splicing? Yeah. Yeah. Yes.

Right now he's working on catching his breath. Catch your breath. Get a sip of water. This is actually good. We were just talking about Batch Transact. Well, we were talking about coin joins, which are related to batch transactions, and how there's an economic incentive to do them as well as a privacy incentive. And I was like, I wish Dusty was here.

And then you showed up. That's something we've been talking about a lot, right? We've had coin joins for a long time. And if you scale things or graph them by cost, you have regular transactions cost this much, coin joins cost more, right? But they shouldn't. But they shouldn't. And, like, the idea is if we can get a protocol for merging transactions that's just generally used,

merging transactions can actually be cheaper. So, like, the scale of, like, join transactions costs the cheapest, transactions costs a little more, and then coin joins costs the most, right? And if you can get the incentive for joining transactions to be, say, we've got fees, you get a larger base of people joining that coin join-like thing, right? Yeah, in theory, let's say you have five people who all want to make a transaction on Saturday. It's actually cheaper for them to all make one transaction where they're all sending money in the same transaction. They get to share the same transaction like envelope and structure, and it saves them on fees.

By my count, the best you can get without signature aggregation techniques is about a 4% savings on the amount of money you can get if you have a lot of people joining a transaction with you. Which isn't, like, enormous groundbreaking savings, but it's not nothing. You know, 5%, 4% or so is pretty good. Is that the math on the overhead bytes for a transaction, 4%? Yeah, like, in a standard transaction, let's say it's 300 bytes. Eight of those are the version number. Eight of those are the lock time. And then there's, like, a couple other, like, individual bytes that say, like, how many inputs you have and how many outputs you have.

That all gets shared, and if you have five people doing five transactions, there's five copies of all that data. Whereas if they all do one transaction, there's one copy of all that data, and it ends up being about 4% of the total. Right, that sounds right. In my head, I was like, 5%, 10%, but that's probably me rounding up in my head. Yeah. Yeah. And so that's what we were just talking about, like how do you create the economic incentives to do this? And we were talking about exchanges implementing this, and super brought up the point.

These exchanges are sometimes by law told that you need to protect user privacy, and adding something like this makes a lot of sense. We were talking about PageJoin development kit, but let's get into what you're working on splicing and how that helps with what's going on here. So for anybody watching in now that may be new to this concept of splicing, let's just give a top-level 101 and dive into what you're working on. 101 on splicing. It's always the hardest version of the pitch to do,

But essentially, splicing is a technique for lightning where you can adjust the size of a channel bounce using an on-chain transaction. So this solves a lot of other problems with lightning. People often make multiple channels to adjust their balance. Splicing, essentially, its main goal is to solve that. But along the way, the way we wrote the splicing spec is it allows lightning nodes to join their splices together in an individual transaction. And this is the part where we're talking about. So in Lightning, essentially, splices are solving one problem with Lightning channels,

but they're also doing coin joins across Lightning nodes. And this is something I'm really excited about for the future is trying to get that coin join-like thing happening in Lightning and combining all the other stuff out there. Pay joins, other coin joins, regular transactions. And I think that's a big deal. One of the things you're doing, or one of the dependencies for creating batch transactions and coin joins

is some sort of coordination protocol or a coordination server that every participant in the batch transaction can talk to and say, here are the addresses I want to send from and here are the addresses I want to send to and here are my amounts, and then he can prepare a transaction. That role is typically one that CoinJoin coordinators charge money to do, and that's why CoinJoins cost more than they should. But I think one of the things he's working on

is like Lightning Nodes do this. They just have to do it. They are a multi-party protocol, and so the nodes need to talk to each other to decide how much and where they're going to send the money. So you leverage that, and Lightning Nodes can do it for much cheaper than CoinJoin coordinators can do.

Yeah. The Lightning Nodes are already coordinating this on Lightning Nodes. All we need is one new protocol, but they can coordinate for other transactions as well. And this doesn't have to be coordinating within their own channels since they can just pick it up on the Gossip Network

and say, oh, you're trying to do a channel rebalance. We are too. Let's put up UTXOs in a batch transaction. Lightning nodes are always on. They already have logic for dealing with UTXOs. They're perfect for coordinating coin joins. Yeah, this actually makes a lot, because that's been a big problem up until recently for pay join

is the liveliness for the receiver and the sender. And so Lightning just inherently solves that out of the box. Yeah, I really want to get pay join involved in the Lightning. Splice joins right and I think there's like a happy marriage that can happen there and so what What are the next steps that actually getting this implemented? So I've been thinking with us a lot and it's kind of like herding cats because you have paid joins coin joins other esoteric Transaction protocol not on cats though. Not not off cats

But it's just everyone's writing a protocol for their own thing, right? There's a page one protocol for page lines There's coin gen protocols other protocols that exists for doing these transactions what my ambitious goal is is to build a unified spec. If you're trying to do a transaction, do it in a certain way to join our spec, and then it'll automatically be joined to other things that are out there

using Lightning as the coordinators. So my goal is to try to build a spec like that that everyone likes and appreciates so that when somebody's building the page one or the next page one, they can be like, oh, I'll just adopt this existing spec, and then automatically it'll start joining

the whole process of the event of the end goal of getting all transactions merged together, right? Like, you could, in theory, see a world where each block is just one transaction. Yeah. Right, something like that. But how many days till simplest possible batch transaction on Lightning?

Simplest possible batch transaction. S-P-B-T. Minimum viable batch. That's a good question. Like, my time has really been focused lately on just finishing splicing. Yeah. We're right at the end.

So we're getting interoperability between Eclair and CLN. And once we get that, it's ratified in the spec. And then the bulk of my splicing is pretty much done at that point. I'm still going to do some maintenance stuff on it. And then my attention is going to go to that. And then as soon as I can write the simplest, what do you call it? Minimum viable batcher. Minimum viable batcher.

That's my next goal right there. Yeah. And so caveman brain over here, trying to understand. So the standardization of the coordination of the transactions, would that look something like everybody sort of just plugs into the Lightning Gossip Network and tries to see what's going on there? Those are good questions. I haven't answered them yet. One thing that I hope ends up happening is that you go to your wallet and you say,

I want to send $5 to Jeremy, and let's suppose it's an on-chain transaction, base layer transaction. Your wallet might be like, would you consider waiting? Like, I was getting an Uber this morning, and my Uber was like, would you consider using a shared Uber? You'll save $3. Right. You could do that same, but, you know, it'll take you an extra 10 minutes. You could do the same interface.

If you're willing to wait a little bit, then your transaction fee could go down from $10 to $8. I mean, we already see this on centralized exchanges. You get the ability. Do you need immediate transaction? Do you have a medium time preference, or are you willing to wait a day? Like, if so, it's free. Yeah, and exchanges are batching. doing that in a in a peer-to-peer way without without needing a centralized well I guess it

still be a centralized coordinator being able to do it as a protocol and have thousands of people offering to do this for you would be better than just saying well if I want to get cheap transactions I have to go through coinbase yeah I think there'd be they would say they said that being one centralized coordinator you have like thousands or tens of thousands of coordinators that you could you could pick from and the coordinators could use other coordinators as well there's no like exclusivity to it yeah so it's like any it's similar to like topology of noster where anybody can run a relay server anybody could yeah i guess so a server like this and just tap into it i gotta think about that analogy but sure yes um going back to fees

for a minute there's four percent you save on the transaction bytes for sure but there are other benefits that with this protocol that could provide particularly around change and like toxic change that coin joints have those can be dumped into lightning channels and if you start doing that you start removing outputs of the transaction you start getting much bigger savings from doing that kind of thing that's possible that's true yeah the it's only yeah there's there's more savings to be had than just the overhead yeah like four percent of the minimum savings you'll get you know it's possible to get a lot more than that depending on the situation i suppose the minimum is zero savings it's not going to cost more yeah uh you mentioned working on this with

CLN, AnyClear, L&D not mentioned. What's it like wrangling the different Lightning implementations to get something like this? Oh, I have no power to wrangle anybody. I mean, Lightning's a very decentralized thing, right? There's no centralized anything. I mean, I was working on splicing for a long time and nobody cared, and I kept going to spec meetings

like, hey, I got this. Oh, he cared, and they cared. But I got this splicing stuff. I'm working on this stuff. I'm looking for feedback on this new spec change. It's just crickets for a long time, right? And I like to think that me just working on it and making progress eventually got people to notice.

And obviously Claire's basically finished the reputation. LDK looks like, I don't know, half done maybe. They're making great strides. So, like, the only way I can try to get other notes to do it is just try to inspire them, which hopefully I've done some of. It looks like I have. I'm inspired. Let's go.

I'll tell you that, Dusty. I'm depressed. Aw. I'm just kidding. Well, I think it's important. I was on the Tom Woods show the other week, and that's why I brought up Monero because that was one of the questions And she asked me, like, people say Monero is private, therefore it's better money.

And the point I made, and I think you're validating it here, which make me feel good about my answer, is that these cryptocurrencies compete on monetary properties and privacy is a feature that can be added over time. So with Bitcoin, it's make sure it has good monetary properties, good monetary policy, and then slowly but surely over time, add privacy enhancing features in different layers. And it seems like it's happening. Is it happening at a pace that you're happy with? or do you worry that there's not enough urgency around adding these types of features? I hope to, like, I'm trying to recreate what I did with splicing, like, do this work

and inspire people to kind of join it as well. That's what I'm hoping to do with this. I don't see anybody else working on something similar to what I'm describing, like, how do we get a unified protocol for joining transactions? I think I'm the only one doing that. Maybe somebody out there that I'm not aware of

is also focused on that. I'd love to talk to them if they are. But I'm kind of looking at a long-term view, right? Like, I want to set it up in a way where I want to build stuff that helps a spec protocol for this, that even if no one focuses on it for five years, and five years from now someone's like, oh, we need this new blah, blah join thing, they'll be able to reference that spec, right? The goal is to get future developers to be like, oh, you know what? I should do this in a way where it can merge with everything else.

So I'm trying to, like, set the stage for the long run. In terms of urgency, I guess I don't really think about that very much. I view Bitcoin as a thing that's going to be around for hundreds and hundreds, thousands of years, and I want to make sure it gets right in the long run. Yeah. That's kind of where I'm focused. Yeah. It'll be interesting to see, as we've seen many times throughout Bitcoin's history,

is there some sort of event in terms of congestion at the protocol layer that forces people to think about these efficiencies? Oh, absolutely. Forces the discussion around that spec and implementing it. That's a great point. It's like right now fees are what, like nothing, right? So there's no urgency. As low as they can be.

But that's the time to build the stuff, right? Like build all the tooling for that now. And when exchanges are like, whoa, I'm paying $100 transaction, I wish I could make that $95, have all the stuff in line. And as a benefit of them saving those $5, they're going to also improve privacy for all of Bitcoin. So it's like that's really the goal of what I'm focused on here is creating the incentive to people that are moving lots of Bitcoin

but don't care about privacy, calm the privacy to save on fees. That's the goal. And the beauty of these types of designs too is that even if you're not using them, if you hit a critical scale of people using them, it helps the privacy of all Bitcoin users. Yes, 100%.

We were talking about the interoperability of protocol layer, lightning, eCache. Does eCache factor into any of the designs that you're working or the ideas? I haven't thought about that at all. That's a great thing to think about. I think what I really like is that to build a spec like this, it can't be built in isolation.

It needs people that care about it to also get involved in the spec, which is internally how we do Lightning. There's been a splicey spec for a long time, but as people implement it, they have changes. LDK has some changes they want lately that's coming through. Eclair has their own. So I think the goal I'm after is building a spec that makes the best sense possible to me

and then trying to get other people that are building things that can tie into it to give their inputs because they're going to have their own expertise and trade-offs that they're dealing with. What are your guys' thoughts about the broader conversation around Bitcoin in the mainstream? A lot of focus on corporate treasuries, Bitcoin strategic reserve. Do you think that's taking away from what's arguably more important, which is the technical viability of the protocol? Or do you think that's actually good, let people focus on that while we can work on these specs? Do you have any thoughts on that?

I didn't even hear the question. Thoughts on the broader mainstream conversation around Bitcoin. Focus on corporate treasuries. Yeah, it's a bunch of number-go-up people who have infiltrated Bitcoin. Let's get back to the founding principles of Bitcoin, which are like we can separate money from state, we can get rid of the control of corporations over our economy and restore the money to being something that the people control and the people own.

In order to do that, we need more people running nodes, not more sailors dumping billions into coinbase and i kind of agree with that like for me i bitcoin is for everybody and eventually everyone will use it that includes all the assholes and the people you don't like with different priorities right and like as a developer and like a builder for bitcoin i'm just focused on i'm going to build it the best it can be for the world and there's going to be all these moments of noise like right now it's michael sailor you know like and right now top returns now but maybe in like you know 10 20 years like china versus india have their own thoughts on bitcoin and that's the biggest news right i i

kind of just sort of it's gonna come and go you know it's like fame is fleeting right the the hot thing of the moment will disappear but the the quality work and the bones that make it the best it can be that'll stick around right that's got staying power so i kind of view it as like right now it's treasuries and sailor then later on the narrative will be something else right like bitcoin is for everybody people are going to use it for things we don't want someone's going to fund a war with it eventually right and i'm not a big war guy so but but it doesn't really matter right like it's just i i kind of tune it out i guess is what i'm trying to say yeah no it's um i find myself too i run a media business a meeting business a media business media business that

makes more sense you know what's getting more clicks right now number go up discussions you Right. It's hard to get people to care about the technical details, which is funny considering the technical details are incredibly important to the long-term viability of Bitcoin as distributed software. Does it feel different this cycle than last prior cycles? It certainly does. Yeah. Yeah.

So I've been in Bitcoin since 2013. There's definitely something that is a bit different this cycle in terms of everybody's paying. It's like we kicked the hornet's nest and the hornets came out. We're like, hey, this is pretty cool. I want to get some more of this, which is a bit weird. It is, yeah. I'd like to see more discourse in the Bitcoin community, in the media,

about the importance of knowing who controls your Bitcoin. It should be you. The user of Bitcoin should be the one who controls his own Bitcoin. Great time to learn how to run a node. Set up, spin one up, and start syncing the blockchain. This is a good time to do it. Well, in terms of that, like self-custody, spinning up your own node, setting up a wallet that pulls data from that node to verify that you're actually receiving the Bitcoin that you think you are, what sort of UX improvements can be made there to make that easier? because I've talked to a lot of people getting into Bitcoin,

and like it or not, a lot of people don't like storing passphrases. So is a model, obviously Block has a server in the BitKey model, but something like that where you have sort of a keyless backup setup something that we need to work towards? I feel like Phoenix is not getting enough hype on this exact front. Like they have a self-custody app. Phoenix D? Phoenix from the Eclair guys?

Phoenix Wallet. Phoenix Wallet. Phoenix D also is pretty similar. The server would be like the self-sovereign part, right? Well, the app is self-sovereign as well. Yeah. Like, you're not compiling it, but the app actually holds the funds. And with Phoenix, they've made it so you can actually do your own self-custody Lightning channels from the phone, like all natively.

And they've done it really smoothly, of course, using splicing as part of what's made it so awesome. But I feel like Phoenix is just like this silently not getting a lot of attention for being this technological marvel. I think they are getting attention. They're like the most popular Lightning wallet right now. Are they? I think so. Well, good. I guess I'm behind.

I'm glad to have them back in the United States. Right. Now they're back in the U.S. Yeah. That's the other thing. I was talking about coin joins earlier. Obviously, the Samurai developer case is ongoing. It seems like the prior administration was being a bit, I'd say, corrupt is the right word in terms of how they went after the Samurai devs, Kion and Bill.

I mean, it came out earlier this week that they literally went to the regulator and said, are these guys breaking money transmitter laws? They said no, and they went and arrested them anyway. Prosecutors went to the regulators and were like, are these guys violating money laws? No. Okay, let's sue them anyway. At least let's throw them in a cage. Right.

Oh, my God. And that's like in terms of, I think that's an interesting. And then from what I heard, I haven't read to see if all this is true, But the person who I saw, the statement going around is that the federal prosecutors then withheld that evidence when they were asked about it in court. Yes, yes. And said, we're not going to tell you. For a year. For a year they withheld that.

This is pure corruption. And it has real effects, right? Like developers start getting scared. Like just talking to developers, other Bitcoin developers, you can hear it. They're all kind of like, oh, maybe I should tread more lightly, you know. Well, it was right after that that Phoenix left the United States. Yeah. Now they're back, possibly because there's better news.

Well, that's the point I wanted to bring up. Is there a period of time, a window of time here where developers, particularly in the U.S., should be sprinting to build out as much sovereign freedom tech as possible while it seems like we have some air cover from this current administration? We've got four years. Build everything. You're talking about a sense of urgency. Yes. There we go.

I mean, I hope this, like, government anti-Bitcoin movement was, like, just temporary. I'm hoping in four years, even if it's a democratic government, they don't go back to that. And I don't know. They call it a gut feeling. I feel like it's not going to be the same. I think that he disagrees. I don't think we can rely on governments being nice to us. That has not been a very promising avenue.

Try and build stuff that they can't stop. Just stop putting us in jail for writing code. That's all I'm asking. I feel like it's a low bar. But if they do, you know, I really admire projects where the founders go to jail and the software still works. That's a good sign. Yeah.

So try and build stuff that if they can't stop it, then perhaps they will try less. Yeah. And I think also one idea in my head that could be a total fallacy is, like, if you're building code that you don't control, there's less incentive for the state to throw you in jail, right? Like if they jailed Satoshi, for example, he can't stop Bitcoin, right? If the government wanted to stop Bitcoin by jailing Satoshi, it's not going to work, right? I don't think Craig Wright did go to jail.

No, I think I saw Peter Todd walking around earlier, actually. But, yeah, if you build stuff that can run without you, hopefully you're giving yourself some protections from being jailed. And that's what we're all doing here, right? It should be doing. What, in terms of what's being built right now, do you think there's an over-focus on any particular area within Bitcoin that is maybe not as high of a priority?

Yeah, we need more focus on splicing. There's a whole lot of focus recently on stuff that requires a coordinator. So I'm thinking ARC, Spark, and LSP software that all requires a coordinator to run it. And there's a big downside for that, which is that there's a choke point. They can squeeze that coordinator and make them stop doing it. So trying to find ways to make stuff that is coordinator-free is great. I agree with that, yeah.

So I've seen you post about this on the Internet over the last few weeks. You're not too excited about ARC specifically. Because personally, I've been talking about ARC seems pretty good. No, I'm very excited about ARC. I think it's really cool. But it is an example of something where there's a coordinator to choke. So that is a downside of it. But, yeah, it's still cool software. And so that would be the Arc service provider that's doing the rounds.

Yeah, of which currently there are two. Perhaps there will be more if the software gets released and people see it's profitable to run one. Maybe there will be more. That's another thing that helps. The second best thing to having no coordinator is to have thousands of them in different jurisdictions. And that's another thing with Arc specifically.

It seems like to scale beyond a certain degree and make it viable, not force liveliness on end users. There may need to be some sort of covenant proposal that gets merged into Bitcoin. Thoughts on covenants broadly? Yeah, I feel like people stopped talking about covenants, didn't they? Like maybe I'm just hyper-focused or selfishly. I just want Lightning version 2 to happen. and it just needs any of them to go through, right? Like, I'm a fan of APO because it's the simplest one.

It's probably, like, two lines of code. And then we can do Lightning version two. So for anybody who's listening who's like, what the hell is... So any prev out. Yeah. Like, what is it and what would it do to make Lightning... Just recite the two lines of code, please.

Yeah, recite the two lines of code. So, like, in every Bitcoin transaction, you have to sign it, right? And there's a little setting saying what kind of signatures is this applying to. So the standard one is just sign the entire transaction. There's a bunch of different options. You can say sign nothing, I don't care.

The other one is sign individual input and output. And the idea is just add it. This is in the original Bitcoin code. This setting is there. You're just adding one more option to that setting, saying that I want to sign the whole transaction, but don't make it dependent on where the money came from. So it allows you to essentially write,

if you have $0 in your bank account, write a check from that account, and then fund it later, right? And that one key step is what we need for Lightning version 2. And most covenant proposals can do that. APO is just the original one, and it's simple as hell that accomplishes that. It's also, that one's actually talked about in the Lightning white paper. They wanted to, yeah, they called it C hash no input back then. Ah, yeah.

But that is in there as like, this is how we do Lightning. And then we figured out a way to do it without it. So, you know, we moved forward without it. Right. But that's certainly a thing. And right now, like, Lightning version 2, it's, like, it's so much better. Like, there's been some, like, prototypes done. They're, like, it's, like, the code's simpler.

It's safer. It's more reliable. It's faster. The databases are smaller. It's just, it's the sexy Ferrari of what lighting can be. But we're sitting here, like, can't use it because we're waiting on this covenant stuff. And interestingly, like, if we knew no covenants are coming ever, then we'd get to work making Lightning version 2 work without those fixes. And it would suck, but we could make progress.

But because there's a possibility of it happening, it's like Lightning version 2 is just frozen. Like no progress can happen on it because we don't know what the core layer is going to look like. Yeah. Not only that, you think about the migration, right? I remember having conversations about the potential migration from hash time lock contracts to point time lock contracts. And the Lightning Network version 1 is built up to a certain scale where if we were to get Lightning version 2 to market, What would that transition of channels look like, and how would it happen? That's a really good question for InstaGibs.

He's here. He's the guy who really wrote the prototype for version 2. But my understanding is you can do version 2 with HTLCs. That would probably make the transition a lot easier. And then probably just slowly add PTLCs. Because PTLCs only work once everyone has it, right? So I imagine it would be HTLCs first, version 2, all the other benefits. Interop, I assume, would work.

But that's the devil's in the details of that one. I think it's also there's some upgrades where you could be like the next time you open a channel you can just use the new version the new like if your current channels are all HTLCs the next time you open one it can be a PTLC channel like there's ways to make them make a

smooth transition between them yeah and you can do that channels that support both probably yeah not I assume yeah what last question before we wrap up thank you guys for joining me it's been fascinating Donato thank you for having me um anywhere anybody out there watching this and thinking i want to get involved to help out on

these particular areas what what should they be doing to i recommend uh yelling at developers on the bitcoin mailing list that's the best thing to do what's your answer i really need help with my new like dream protocol it needs one it needs a marketing name so i could use marketing help i'm calling it the delayed transaction protocol terrible name yeah already right dtp i guess But this is the idea I've been talking about of joining all the different types of transactions, a protocol that can join them together. And I need to get everyone involved that's doing transactions, right? So it's a big coordination effort.

Wait, wait, wait. I gave you a better marketing name. Honey Badger. Honey Badger. Honey Badger. Honey Badger. I like Honey Badger. That's a good name.

That's a pretty good name. It's pretty good. All right. So we've solved that one. All right. We solved that one. And I just really want to be able to get all the people that are working on this kind of stuff, like pay join, other people working on esoteric transactions,

or even just regular ones, to be able to get us together and get a spec we can all agree on, that's really critical. It's a big coordination problem. I could use all the help I could get with that. Well, final, final last question. How important are events like Bitcoin++ and making stuff like that happen? Oh, super important. Man, it's like this kind of event is just full of Bitcoin developers

that are deep in developing on Bitcoin. And it's the perfect environment to talk to people and coordinate exactly things like this, among other things. I love bringing new people to conferences and introducing them to all of my heroes and then they can learn why they're my heroes and figure out ways to make all their cool ideas come true. So this is a great place for the sharing of ideas, for the coordination and the networking that gives you the leg up you need to make your next step. Well, super, Dusty. It's been a pleasure. It's been a pleasure.

Thank you guys for doing what you do. Thank you for joining me. You're welcome. Thank you, Marty. Have a great day. I'm sure I'll see you around Austin. No. No?

Yeah. All right. Let's be great. You too. Bye-bye. You're on the mainstream now. Oh, I'm on the mainstream. Welcome to the new audience. Go for it. Hey, anybody coming in from the Bitcoin Plus Plus official audience, welcome to the news desk live here at Bitcoin Plus Plus.

I'm running solo right now. Marks, do you want to hop back here with me? Sure. Yeah. We've got our next people coming on. so howdy howdy incredible day what's up it's going well so far yeah this is uh this is exciting it's my first live news desk so this is a new format for me so thank you for helping come take some of the pressure off you're doing well you're doing well yeah what are what are you excited

about this week particularly obviously opera turn obviously opera turns a big thing but what i like is that people are just discussing other things that are important too yeah right so like let's just not get mired and distracted by this one thing that's going on there's still plenty of other work to do yeah so and i'm hoping that we can build off of the covenant stuff from last year right that was it seemed like there was a lot of really good momentum and now it's kind of fallen by the wayside a little bit yeah yeah that's the uh that's what we're just talking like any prev out like it seems like any time the covenant discussion specifically gets to a certain point the community for lack of a better term gets a bit distracted and that's uh it's interesting

It's hard, and from my perspective, as somebody who's not a developer, investor in the space, running a media company, interviewing developers, it's hard to really get a grasp of where the signal is and where the focus should be. So conferences like this, for me, being able to have a conversation with Super and Dusty like that, it's extremely helpful. Because you think about where Bitcoin is, where it could be, particularly as it pertains to better privacy and better usability as a payments network. Like, the ideas they're talking about are extremely enticing in the sense of, I want that. Yeah. Yeah, definitely. And I also wonder if what we're seeing with OpReturn, if this is going to have an effect on being able to get covenants and other things through, right? Are we exposing what the current process is for getting new updates and upgrades to Bitcoin out there? And is that going to hamper it?

Like, do you think that this is going to delay things? like let's say that covenants were going to get approved and and merged in and everything are we setting ourselves back i think that's the latest stuff it's like taking aside the technical aspects of all this i'm really getting to what i think is the meat of the heated discussion over the last two weeks is this governance conversation of how do changes actually get through in bitcoin what is the governance structure i mean obviously the focus has been on the core repository but There are other implementations. There are policies within the core GitHub repo.

And I think there's just a misunderstanding between broader Bitcoin users and the development community about how individuals are supposed to interact with core's repo specifically. And I think there's just been a lot of talking past each other the last couple of weeks. Yeah. Yeah, definitely. It's really interesting that we have the mining summit going on over at Bitcoin Park Austin right now. very different types of conversations right it's almost like the two different factions have shown up and they're at two different conferences yeah in the same city and we need to get them together talking in person that's what parker if you're listening i told you to come over come on come

on the live news desk because we get parker over here luke dasher what are your thoughts on this opportunity because to me i said this on rabbit hole recap last week which you were on thank you for coming on that as well it seems like you're bailing me out on on the live streams quite frequently these days it makes sense if there is a subset of the network that is getting certain transactions uh that many other nodes are not that creates problems in the fee market which can create problems for second layer solutions like lightning particularly around fee estimation when you're trying to close channels and so to me it makes sense i'll just say this bluntly what i believe is like get rid of the limit people are making these transactions anyway uh if you're

going to mess up the p2p layer by not doing that uh i think that creates some weird economic sort of imbalances at the p2p layer of the network which should be solved yeah well and and i'm just a node runner i'm not a bitcoin developer i do use my own node as an economic player so like i do have a vote i guess if you say in the system all the discussions i've heard over the last week is it does seem like the drama around op return is not actually on what op return is preventing or not preventing i think it is more on the governance side of things because i think stuff is getting through anyways regardless of what the op return limit is and like you said we can't estimate fees correctly because even if even if all of us had like the strictest op return policy

on all of our nodes people are still going to go to directly to the miners and get their get their transaction in so i'm i'm more of the opinion of remove the limit and let's just let the free market decide and let's let fees determine what things get in and don't get into the blocks yeah it's gotten to a broader discussion uh discussion of what is bitcoin is it what should be the use cases of of bitcoin transactions should it be purely economic transactions sending UTXO from one address creating a new one in another or should we enable people to anchor data that isn't really pertinent to economic transactions into into the blockchain as well and my thought is it's possible to do that whether we like

it or not and we just have to live with the fact that it is possible and design around that if you want to talk about disabling that type of economic or that type of use case of anchoring data inscribing data into the blockchain i think that's a bigger discussion of changes that would need to be made to basically make that impossible which yeah as it pertains to op return it's simply saying this is happening it's possible we should just recognize it and increase the limit so that people could use it yeah and if this extra data is actually helping with scaling solutions it could be economic data right like we talk about bitcoin needs to be the best money well what if these extra you know bytes being included are helping

to be better money we can't decide that yeah right like we talk about the fed all the time how they can't orchestrate the global economy we can't orchestrate the global bitcoin economy and if there is a way to include data in these bytes that enables this layer to this layer to or a future layer to we haven't we haven't just figured out or maybe you know it's there's all sorts of stuff that is yet to be discovered and if we are just simply saying no we know what is spam and what is not listen to us like that that seems like a precarious situation to put ourselves in yeah so i think bitcoin will survive this uh oh yeah this tiff as well yes yes i agree now if we want to have a governance discussion on like how to run a github repo like i feel like that's

a separate thing but our return is getting mixed up in that yeah i think that may be the more important thing for really getting at it there's i it's been very clear there's been this pent-up sort of uh i don't want to say frustration but there's been pent-up pressure in terms of the bitcoin core github repo uh how it's maintained and what are the processes to get things merged and i don't think it's any fault of the way it's set up i just think there's a lack of there's a wall in terms of the communication between how that's being operated and how it's being communicated to broader users and i think that's easy like it makes sense to me developers are focused on what they're building right yeah and typically they're highly focused on the individual

projects they're working on and this has been a big discussion do they just need a pr arm to basically explain what's happening yeah yeah people understand it i think that's steve lee has been really big on that he's been trying to push for like pms that could be kind of the the face and the voice of some of these dev initiatives right to help to help give better pr if you will not you know public relations for for all of this to make it a little smoother yeah and then you have the whole discussion of separate implementations like knots la bitcoin like if you're angry with how core is handling their particular implementation i don't want to say go work on another one but like is the option to work on another one

important worthwhile ideal i think that's a big question that that needs to be answered yeah and again i think events like this getting people in meet space is just looking around at the crowd here at btc plus plus won't dox anybody who doesn't want to come on camera but it seems like there are many different individuals who have very strong opinions on both sides of the argument that are here so hopefully we can get together talk it out yeah yeah because it's easy to like lob grenades at each other online but you get here in meat space and uh you can sense people's like true intentions and you can sense that a lot of times we are more genuine than we come across like when we're on x we're just we're we're trying to kind of like further

further discussion in a different way um whereas you can get to the real meat of the issue here so um yeah i would love even on the stream if we could get some people on here that had different opinions that would be really helpful because we are just one side of it like it sounds like we both are kind of of of the opinion of just like remove the limit um and maybe people are watching right now and like really angry about that like let's get somebody on here to say no we we shouldn't remove the limit yeah and then go back to the discussion like the form factor the mediums of the discussion i think it's clear need to i don't want to say change but there needs to be different avenues created for these discussions x terrible place for them yeah

Short format, grenade throwing, GitHub, centralized, Microsoft owns them. As we see with the mailing list moved from the Linux over to Google, distribution over to Google, there have been some problems with Google. So it's literally what is the correct forum to be having these conversations? What is the decorum of those forums and who decides on all that? This is the beauty of working on a globally distributed network. Wouldn't the answer be Nostr at that point, right? Well, let's put the case forth.

Build some kind of mailing list type environment, but on Nostr. Because I think the problem with the mailing list is any one person at this conference is technically capable to spin up a new mailing list, but you still have the centralization part of it, right? Like they could potentially block people from participating in that conversation. So if we had some kind of Nostr relay, decentralized relay kind of thing with a Bitcoin dev mailing list, that would be interesting yeah what would that look like in your mind in terms of what a bitcoin

dev mailing list client yeah yeah some kind of some kind of client web client where um i don't know if it's like a web of trust kind of thing where people are are their their voice is included in the discussion if they have made some contributions here and there and and you can kind of prove like okay you've even if it's just one commit that's been included right merged in And maybe that gives your NPUB ability to be a part of the conversation. I don't know. I think there's something there. That's fascinating stuff.

All right, outside of this OpReturn, we're focusing too much on it now. What are you most interested to get out of this? I mean, selfishly, I'm here to help run this, but then also get as many people to sign up for Maple AI as possible. So I'm talking to devs about it when I can because I really think that, you know, we talk about how the cloud is a centralizing force, and sometimes it feels like it's kind of a psyop.

We could have had this really nice decentralized web, but we got the cloud instead of that. And secure enclaves are a way to kind of take back that capture that we've had from the cloud. And so we're starting with Maple and trying to give people a really good, solid AI that is not captured. But then beyond that is we're building OpenSecret

try and get people to build. The way that I'm phrasing it now is it's non-custodial ownership of your data, like self-custodial data, basically. So when users are using Maple or when devs build on OpenSecret, they are giving their users self-custody of their data. Yeah. And I think it's an incredibly creative way to solve this data storage issue on behalf of end users and app developers themselves because i think part of your pitch is that as an app developer that data is toxic right you don't want the responsibility of holding that data and securing it because at some point if you become popular enough becomes a huge honeypot so trying to create frameworks from which you just take that risk away seems incredibly beneficial to me yeah

and obviously direct parallels to bitcoin wallet apps right and custodians of bitcoin they don't to have that honeypot they don't want to have that that burden of having user funds and so we try to find self-custody solutions not even user funds like i've been thinking about ways of implementing open secret one simple use case even if you want to watch only wallet and you want to upload an xpub to watch certain addresses like this seems much more preferable to use something like open secret to upload that on a bear server yeah yeah that's true uh we have people who are trying to track their you know their bitcoin base price so that they can do their taxes and all their sales of bitcoin to pay for for everyday living expenses and so i can see that where i would want to put

here all my wallets that i'm tracking and just keep tracking before me like i don't want you to do anything with my funds you don't own my funds but i also don't want some other app developer knowing these are all my wallet addresses yeah correct me if i'm wrong some of these secure enclaves in the cloud are using zero-knowledge proofs some of the secure enclaves using zero knowledge proofs um that i don't know yeah no i don't know it could be wrong so i know that there are like web three platforms that are using secure enclaves to provide zero knowledge proof um like roll up type stuff um so no that's that's something i've looked at i must have dreamt that Yeah, hallucinated.

Or maybe you're right, and I need to look into it more. But yeah, no, we're using AWS Nitro for our cloud stuff for CPU, and then we've got GPUs with NVIDIA. Right now we're working on getting our own models up and running too because we're just using Lama 3.3 right now, but we really want to get on Lama 4. We want to get on some of the other deep-seq. Yeah, when deep-seq.

When deep-seq, that's my question. Yep. Not sure yet. We're still working on it. Our Lama 4 implementation got held up because we currently use a GPU provider in Europe, and there was a licensing problem where Lama 4 is not available in Europe.

Because of European regulations? Yeah, European regulations, and specifically, like, the multi-model, multi-agent stuff that they have. So we're working through that. But DeepSeek is hopefully around the corner. Yeah. I mean, we've talked about this a lot on the show and off air,

but this intersection of ai and bitcoin in two ways first using ai to build products as a founder or a project maintainer in the space and then to like the literal intersection of ai using bitcoin particularly in the agentic model world as somebody who's been really diving deep into using ai to help build a product like what do you think anybody working on projects out there should be leveraging to to accelerate what they're trying to do yeah and there's a there's a third element to it which is the energy side and i know that miners and they're they're working on that aspect right of like bitcoin mining and uh ai compute so i'll leave that set aside for now but when you're building like just just start using the tools um i i say i think i said this on rhr

but i i used to be someone who would just kind of tune out ai discussions because it's like i want to learn about bitcoin stop talking about ai but it's really infiltrated everything now so if you are a bitcoin developer or you're building an app you're doing anything in app development you should be looking at like how can i use ai to at least help help do a little bit of my work there's still some some training you have to do not not ai training but like training yourself with the tool chain and you have to review everything you can't just like take it on its word because it hallucinates a lot it messes things up it'll like destroy entire classes and build new classes so there's a lot going on there but but start playing with it and i would

just recommend like pick a small feature and just see okay can i have ai help me build this feature how does it go um and then from there you can start doing more and this is just like typical typical development workflow is like don't just try to bite off the whole apple at once just bite off one one portion of it um so i would say start there and then on the agent side of like agents paying agents there is some new stuff now where coinbase is doing their own version of l402 they're calling x402 yeah yeah let's talk about this yeah yeah and i i um i need to do more info they're pretending like it's a novel new idea yes they're pretending like it's new and of course they're doing it in their own coinbase way um but something that's interesting is i feel like

they're starting to kind of prove out like maybe this is an idea that should be pursued right so A little bit of validation. And I know that Visa is also jumping in. Or no, it was a Stripe. Shoot. Stripe's definitely getting into stable coins. Visa's involved too. They have a whole.

Visa, yeah. Because I'm on a live stream, it's either Visa or Stripe. One of them is actually getting into like tokenizing credit cards where they want to provide a similar kind of thing that we're talking about with eCash and agents. So there's something to look at there where like the traditional player are trying to get into the space, but using their existing rails. How would you tokenize a credit card?

Yeah, I'm trying to remember the discussion we had last week about it. But I think that they're trying to basically like virtualize a credit card transaction and split it in. And then you would have like a whole bunch of tokens that go back to a single credit card transaction. Okay. I might be totally half-based right now. So yeah, but there is, there's something,

there's a development going on there where the traditional players are trying to move into the space. Well, that's what Coinbase announced, what are they calling it, X402, and whether it's Coinbase, Visa, Stripe, whoever's doing the credit card thing, it just seems like they're trying to bolt on incumbent infrastructure into the HTTP layer, but as we know, HTTP 402 sat dormant for decades because the incumbent financial system is not a viable option. to to solve that it wasn't until something like bitcoin or other crypto assets came to to be that it's actually possible now and i i'm very interested to see because coinbase usdc is a big part right what they're doing with x402 i wonder are the agents going to decide what is the best money like you get these products in the market do you think at some point who knows AGI or ASI is necessary, but do the agents begin to prefer a certain form of currency,

digital currency over another? And will it be Bitcoin? Yeah. Well, so free market, let's put all of these solutions out there and let's make it possible. And then let's give the AI agents their choice to just say, Hey, you, rather than dictating, like, here's the technology you must use, you simply just say, I want you to accomplish this task. These are the tools you have to use, give them all the tools and let them go figure it out. Right. And I think very quickly, they will start to figure out what is the best route for them to flow um there's uh there's jordy working on few sats right um and he was trying to do a lot of l402 stuff now he's actually looking to be like the best implementation of also x402 so he's he's not trying to like go into kind of the the

shitcoin space but he's saying hey i want to be the provider for agents to pay for stuff and if they want to use lightning let them use lightning if they want to use coinbasing let them use that and if there's like a visa or a stripe or someone else is coming on he wants to provide all of those options to um to let to provide a free market for these agents to decide yeah that's the uh most exciting sort of intersection of a bitcoin as a medium of exchange right now i'm not sure if you saw paul toy's uh tweet from last week basically letting people in on what they've been working on uh and basically a statement he made at the end of his post was that pretty soon like ai agents are going to be uh conducting lightning transactions at a scale that orders a magnitude

over human usage and like is that are we going to find ourselves in a place where we realize that like these ai agents are actually using bitcoin as a medium of exchange more than more than humans rather quickly yeah bitcoin a peer-to-peer you know electronic cash system really we didn't know that was gonna be peer-to-peer between agents between ai we thought it'd be meat space type stuff that would be interesting uh i mean you think about just trying to pay with lightning right now like you scan someone's qr code scan an invoice and pay it and all the juggling and back and forth that has to happen agents can just coordinate that so much faster than we can and they don't have to sit around and say send me a new invoice this one expired it's like no

They're both there ready to go. Yeah. And, like, what does that do to, like, people have, we have the MicroStrategy, or the Strategy, excuse me, Conference for Corporations. Very different conferences, what's happening here today. But Saylor famously talks about, like, the total addressable market. I think he gave a presentation yesterday talking about, like,

$200 trillion is in sight. But I don't think people have really factored in the potential for the digital economy to expand at a pace and at a size that far exceeds what's happened to date with the internet age. How much economic activity will these agents be doing? I don't know for sure, but I imagine it's going to be massive and microtransactions too. Yeah. And most of us can only think about what we can see, right? We can't think about the stuff that doesn't even exist yet. And agents are going to be coming up with business models

and transaction types that we haven't even fathomed yet. So there's a lot that's going to happen, and we need to build the rails for it to be ready and be available for that. So I think it's too early to tell that AI agents paint each other with e-cash or with Bitcoin, something like that, Lightning. I think it's too early for us to just kick it to the side and say, oh, it's not going to work.

Like Visa has too much of a head start. They're too big of a company. Let's just keep building the technology and put it there. Because like you said, AI agents might surprise us and do things that we weren't expecting. Yeah, and I would put forth, too, anybody building on Bitcoin, another thing I don't think people appreciate is the maturation of the state of the interoperable network. Whether it be Bitcoin on-chain, throw liquid in there, lightning, e-cash bins, we've got ARKs coming to market as well, Spark. I think people severely underestimate the power of the maturation of that interoperability between all those networks.

First conversation we had today with VNPRC and Gary talking about hash pools, how you can use a combination of Bitcoin mining layer, literally mining shares and cashew mint and the lightning network to really solve a critical pain point in the mining pool layer of bitcoin and that is and i believe they use to a certain extent too or can use noster to a certain extent so you're combining three or four interoperable protocols to to create a user experience and a product and i don't think i think we're just on the precipice of really unlocking all the value that exists between these networks being interoperable with each other yeah i wasn't able to catch everything they were saying but is he talking about like take all of these solo miners that are at home

maybe heating their room or using the bid axis is there a way to use ecash to kind of give them an opportunity to sell their hash rate immediately so okay what you do in your mind you create a hash cash shot 256 mining share trillions of them sending them to a pool what hash pool would enable you to do is if you're sending it to a hash pool you're immediately getting e-hash shares in return which are e-cash tokens that can then be traded for bitcoin to somebody who wants to take a risk of holding that and potentially getting a bigger reward and so hoping that so then those are somehow tied to a block that gets mined? Yes, so once a block gets mined, you actually know what the payout is and the ash tokens get converted to Bitcoin. Wow, that's awesome. I believe that's

how it could be cool. But in terms of solving the problem that exists now at the mining pool layer with FPPS, creating the central force, this sort of solves that by enabling people to get paid out or get liquidity for their shares immediately and give that risk of the payout variance to somebody else who's willing to stomach it. Okay. Yeah. Interesting. And something that developers might not think about, right,

when they're building some kind of tool or some kind of protocol, but we need to build them to be flexible so that these kinds of solutions can come out of the woodwork, right? Give the tools to people so that they can go solve their problem and somebody else can solve their problem

and then they can creatively get together and solve a new problem. So I think it's great. I think it's a really cool, novel idea. Yeah. Yeah. Where did everybody go? I don't know. They all took off and went to lunch.

We're supposed to be on stream. We actually have lunch in a bag over there. Yeah. Okay. Oh, they're all across the stream. Everybody's at lunch. So apparently the two interviews we were supposed to have right now are not showing up. They went to lunch.

They went to lunch. They went to lunch. Okay. Yeah. Oh, you are one of them. All right. I mean, do we want to get going or do you want to wait until they're both here? Yeah, do you want to stay here, too, join?

Sure, yeah. Here, you can come hop in the middle and I'll move to the outside. Good to see you, sir. Nice to meet you. Thank you for joining us. Yeah, for sure. What about block templates and mining pool centralization? That's what we were just talking about.

Oh, damn. Good segue. We were literally just talking about hash pools. Oh, man. my ears are ringing all right so I guess let's just jump right into it what is your fascination with block templates and mining pool payouts um yeah you know like I think it was like a year ago um b10c posted a blog post that where he was like comparing merkle branches between a bunch of different mining pools and like he found like a whole bunch of them were matching up and um

Yeah, I saw that blog post and it really resonated with me because it's like if you're not like it's kind of kind of like a silent killer, like if you're not actually like looking like under the hood at these block templates and like comparing them, like this could be potentially something that kind of like, like blows up and becomes kind of like a centralizing force. so yeah that that really resonated with me so like i started kind of building this tool called stratum.work that like uh it goes out to all these different mining pools and it like pulls in block templates from all these different mining pools and it shows them in a table and it colorizes things and it like shows different fields about like the coinbase transaction and stuff like that and um yeah it just really gives the observer like the ability to kind of like compare things and like look at the data and try to like make correlations and find relationships and things like that so i've been doing that for like about a year and have been really getting like deep into

the weeds on like looking at historical transactions and like finding like like little errors in like block templates and stuff like that just to try to like correlate um pools and and sort of like figure out which ones are working together so if you had to uh thank you for putting these uh these charts together by the way we look at them on um our chart quite a bit and it is it is fascinating to to see this and shout out to um i always forget the name b10c he's the man um for really ringing the alarm i was probably this time last year yeah it was about that march april last year and it's pretty scary especially you have ampoule which is essentially a bit main offshoot creating the block template for many other pools which people up until that point assumed were

separate and doing their own thing but it became clear that those pools were in danger of going bankrupt if they didn't leverage the the central bank of and pool and their treasury and i don't think it's ever been explicitly admitted but it seems pretty clear the trade-off was you can leverage our treasury to pay out your your um your miners every day but we're giving you the block template and with that in mind like what is the state of block block template diversity right now like how many block templates different block templates are mined per day yeah well so um b10c actually did just put out a new blog post uh maybe last week or something i think it was called mining centralization in 2025 where um like his research and my research both kind of like confirm

um this uh like list of pools that are we're basically calling like ant pool and friends um uh and he does it like a really great job of um like mapping out in a line chart uh and kind of like putting a percentage to all of these like because you know like you look on mempool.space and like i love those guys but like their pie chart is based off of the like the ascii script sig of the block and then like the coinbase payouts but like under the hood there's like a lot of kind of like sharing that that goes on there and aren't those names self-reported they're self-reported so i mean uh yeah like antpool could just switch their script sig to say via btc and like mempool that space would pick that up as via BTC. I think

like, yeah, eventually, we would figure that out. Because like via BTC miners, and pool miners would would be seeing that they're like not getting paid for blocks that they say they're supposed to be getting paid for. But like, to the general observer, it would look, yeah, look like the data would be easily skewed. So Now, I pulled up B10C's 2025 Bitcoin mining centralization report, and we reported on this a couple weeks ago on Rabbit Hole Recap.

And correct me if I'm wrong, but it does seem like there was a market reaction to B10C raising the alarm bell, what you can see with Ample and Friends hash rate falling. Yeah, I think, like, he does mention in there that, like, the earlier data, like, to the left of the chart, could be like a little bit in inaccurate. I don't remember the reason why he said, but he kind of explains that a little bit there. But yeah, kind of hard to say if it was a market reaction, or if it was maybe just like a little bit of a data thing. But I think it's known, but I don't know

if it's pulled up on the stream. But like, maybe it was that one. There was one there that it was kind of like, kind of scary to see that like, at one point, the and pool and friends was up to like 42% of network hash rate. Yeah, I think it was this one that might have been that one. Yeah, yeah. 43 43. I mean, like, to me, anything over 33% is like a little bit concerning, because it's like one in every three blocks is coming from this like one block template. And here we see 43%. So yeah, it does seem to be trending down, which is a

good thing but it's kind of like ticking up at the end there so yeah i mean and pool and friends and founder right below your threshold of 33 currently at 31.9 percent um and so how do we solve this block template issue in your is it is it a technical problem or an economic problem i was making the point for a while that it's an economic problem of how mining pools particularly with an FPPS payout scheme operate because you need a big treasury to ensure that the payouts get there. Yeah, it's like probably incentive-driven,

kind of like what you're saying. Yeah, like I don't know how to exactly solve this, but I think some of the work that like Scott and the BIDX guys are probably, you know, like kind of like a good good segue into decentralizing hash rate but yeah I don't know like probably like talking more about miners creating their own block templates with something like stratum v2 or datum or gonna gonna definitely decentralize block template construction but yeah I don't have like

a really good solution it's kind of like a really multifaceted problem to try to off yeah andrew pulcher welcome to the stream thank you for joining us you are hopping into a discussion about block templates and centralization in the mining pool layer of bitcoin cool well thanks for having me do you have any particular thoughts on this subject i don't so stratum all the mining protocols i've always been kind of out of the loop on i i got into bitcoin in the early days when initially you needed a windows computer to uh to even compile bitcoin and then when it was possible to run it on linux you needed a graphics card everybody was gpu mining but i didn't want to put unfree drivers on my linux box so then i like didn't mind because of

that and then you needed asex and then at that point it was just too late so i was always just like excluded by non totally non-bitcoin factors so the uh what's funny we're at this bitcoin plus plus conference about mempools and i think that is a big topic that ties into mining centralization is the whole i mean obviously the discussion leading up to this has been about op return and the fact that you have a subsect of people transacting on the network that are transacting out of band directly with mining pools to get certain transactions non-standard transactions included in blocks and that's creating potentially create centralization at this mining pool layer too and i think would be curious to get both your thoughts on the

the op return limits and whether or not that actually solves that problem and how so sure um so there's been some great talks for those listening uh you should check out i don't know if the conference is live streamed or if they'll be uploaded soon uh but gloria and peter in particular uh articulated this kind of thing really well where as you said the mempool policy is pretty important um is kind of an essential feature of bitcoin censorship resistance right you really want your mempool policy to match what minor policy is because the mempool is acting as kind of a staging area for everything that goes into blocks but if there's a mismatch between what the mempool is doing what the miners are doing well the miners get paid the mempool

operators don't right so that mismatch turns into a denial of service vector for them because then people can kind of spam the you know do whatever they want to the mempool they're paying fees but if those transactions aren't what show up in blocks then the fees aren't they're not really paying the fees so all they're doing is taking resources and what we've seen in practice is that there are there's a lot of demand for putting data storage onto the blockchain right and miners are starting to operate these kind of out-of-band fee protocols and we see not just with the data but I think Gloria had a slide showing with our full RBF where if your mempool policy is not doesn't correspond to what's maximizing

profit for miners then in fact your mempool is out of sync and you have bad block propagation your compact blocks are well you need to do a bunch of retries and your block reconstruction rate is lower you're spending a lot of resources as a node operator so long story short i think we got to drop the limits right i think stuff like this where the opera turn data is not inherently like i mean maybe like philosophically is a bug in bitcoin or something but it's not technically a bug it's not like we could filter out and say no data on the chain and be successful if we could then probably we should put that in the consensus rule right but we can't in doing in a standardness rule

simply puts it out of sync with what the miners are doing and then that's wasting resources from mempool operators and then that encourages transactions to bypass the mempool it encourages or discourages people from operating a node that's relaying transactions and it's just this is bad for the network the the incentives are bad yeah and correct me if i'm wrong but one of the inefficiencies is if these transactions are going directly to mining pools are confirmed in a block for node operators that didn't have those transactions in their mempool they're forced to then go verify like on the go that that's an actual valid transaction and that takes up resources as well right yeah when i uh

in probably 2012 or 2013 there was kind of a neat effect where if you're running bitcoin before it was Bitcoin Core, it was just Bitcoin. Every 10 minutes, Netflix would freeze. And the reason was that you were downloading an entire megabyte block and then validating it all in these big chunks, in these one meg chunks. And we more or less solved the bandwidth problem with compact blocks, right, where you see a transaction. If you've already seen it when the block comes in, you just get this kind of like short signature of the transaction. And the validation problem We solved with, we just cast. You see the transaction, you validate it.

Then when it appears again in a block, it's got the same TXID. You don't need to check it again kind of thing. And if your mempool is out of sync with what blocks are being produced, you're back to the, like, Netflix is freezing every 10 minutes kind of scenario, right? So, yeah, it's not great. Yeah. What are both your thoughts on, I think, to me at least, observing this conversation around op return over the last couple weeks?

It seems to me, because I agree, I've said this last week on a show, with this in mind, you're sort of disrupting the P2P network and slowing down node operators that aren't seeing those transactions. We should try to make that as efficient as possible. But it seems like there is some sort of communication issue that has arisen. I think a lot of people have been talking past each other over the last week. Any thoughts on just how Bitcoin Core as a repository operates and what are the rules of actually interacting with the PRs in that repository

versus how people understand that to work? Yeah, there's a couple of different issues there. So one, regarding people talking past each other about opera terms in general, right? We kind of had a big argument around op returns at many points over the years. But five, six years ago, we were talking about op returns. And the story then was quite different in that we didn't have miners doing out-of-band fees at all.

So by having a policy, a mempool policy, that was blocking op returns, it did effectively block the op returns. And it also sent a signal to the community that this is something that the network kind of tries to discourage. probably shouldn't do this kind of thing and even if it maybe meant made economic sense to be like doing ordinals or whatever right then people wouldn't

because they were there were other chains for that and there wasn't really demand and as the community has grown the economic incentives take primacy and the technical maturity is there that allows miners to be doing these these out-of-band fee kind of things and blocks are consistently full so we have of an operating fee market and stuff. So the fight that we were having about op return, should it be 40 bytes, should it be 80 bytes kind of thing like six or seven years ago,

that's not the fight that we're having today. And I think a lot of people are still carrying on from there, right? And I think that might be part of the communication breakdown. Or even they feel like by allowing op returns, they're like conceding. They're like, this is us like giving up

on this fight kind of thing. But I would submit that that fight's over, right? Like, there's no fight. Boris, anything? So I've got a way more smooth, like, this is very much technical depth that I have. I'm still formulating my decision there. Keep going. Sure.

So I'll try to touch on the second thing you said, right, which is, like, what is Bitcoin Core's role in this, right? So why, it's kind of a funny situation as an open source developer when you're working in public on GitHub and people interpret the GitHub repository as like you can come in and just like post things and like complain and post vitriol and like start making demands of the project. And that's not a fair demand of an open source project, right? just because we're developing and collaborating in public doesn't mean necessarily that it's a town square where the public can kind of, like, come in and start conversations. But that's one perspective. Now, of course, everyone listening is like,

Bitcoin Core is not, like, an open-source project. It's just, like, people developing internally. Bitcoin Core's mempool policy kind of sets the de facto policy. I mean, in one sense, the policy that Bitcoin Core implements is what the majority of people are using. So, in fact, they set the policy for the network. But also, because they're the people with all the, like, cluster mempool developers

and, like, the people thinking about package relay and doing this stuff is really hard to think about and really complicated, they're also trustworthy, right? Like, when they implement something, even if I disagree with what they're doing, I personally don't have the mental bandwidth to go change it and do other things, right?

So when they make decisions that I... Don't like then then I understand the impulse to like try to step in and say like guys You've got all the geniuses like please won't you do something genius II like that's better for me, right? Like I get that impulse. I really do But at the same time these are like volunteers many of them have sponsorships and stuff But then like well, it's just a volunteer one step removed working on an open source project is technically very heavy and There's a lot of complicated

economic incentives at play as well as complicated technical problems at play and it's it's a difficult problem right um and it's frustrating to see like kind of twitter level discourse like coming into the development world right well that was we were talking about that earlier too like what is the the medium for these types of discussion i think twitter is terrible medium for it obviously there's been problems with the mailing list over the last couple of years um And is there a better forum for this? Does it have to be long-form conversations like this? Yeah, it's a good question, right?

So I think there needs to just somehow be some friction, just like a little bit of friction, right? Like you don't want to be like setting up censorship walls and stuff. But the mailing list actually has improved a fair bit since we brought in the new moderation rules, which is basically if you're a new poster to the Bitcoin mailing list and your message goes to a moderation queue, They're almost always unless you're like outright like posting LLM spam or like whatever your message will get through

but then there's a delay of a few hours or maybe a day or something and that tends to prevent conversations from It reduces the tendency for conversations become heated and like rapid back-and-forths and stuff and of course it's not perfect because most of the active posters are now white listed on the moderation they once you post a few times and like you stop going to the moderation queue, and if you want to have fights with each other, you can, and people do. Twitter is bad because it's instant and it's low volume. You're incentivized to post the shortest thing you can so you can get it out quickly. I don't know. I've been talking

about this a lot with other open source developers who are in some ways nostalgic for the early 2000 when we had like mailing lists where you're doing development on mailing lists because just signing up for a mailing list is actually a little bit of friction i know i just said like people have fights on there and stuff and there's a lot more friction to signing up for a mailing list and posting than there is friction to just tweeting and just adding whatever orgs and stuff that you want to um so um i don't know exactly but i feel like there should be a little bit of friction somehow yeah it's uh at the very least it's been fascinating to watch unfold but We don't have to belabor this point.

I'm sure many people are going to be talking about it quite a bit this week. Let's shift to high-level cryptography on Bitcoin, like cutting-edge stuff. What excites you? I've seen some projects working on Frost, MUSEG2. What is the state of that area of development in Bitcoin in your mind? And where, with Miniscript as well, where do you think it is and where it could go in the next couple of years? Yeah, so I work for Blockstream Research, and we do kind of two broad categories of stuff that we do in Bitcoin. And one side is cryptography.

This is Frost and Musig, and we're trying to get into quantum. We're trying to bring some quantum expertise and stuff. And sadly, that's not me. It's just this super cool stuff that excites me, but I feel like a totally non-technical person when I go into the quantum discussions. And the other side of Blockstream Research is the scripting stuff. So Miniscript, we're working on a language called Simplicity, which is not going to directly be on Bitcoin anytime soon, but it is designed for Bitcoin to fit into Bitcoin.

And then we think about and play with a number of kind of script extensions. So like Cat in particular is, well, my baby. But also CTV and TxHash and all these things that are floating around. I think they're all pretty cool. Yeah. And going to that CAT, CTV, TxHash, I think that's another discussion. It's funny, the op returns sort of took the air out of the room of the covenants discussion. What has, in your mind, I guess, what is the optimal covenant to get incorporated into Bitcoin if there is one? yeah um so i like cat because of the covenant solution it doesn't need to be optimal so cat

does a lot of things it was not very good at covenants but there's a worry if we had a covenant solution that wasn't very good at covenants there's a worry that then it makes it hard to get a good solution in right or maybe we do after a whole bunch of fighting then we have this vestigial like thing that that sucks forever the nice thing about cat is that it's small it does stuff that's not covenants and it shouldn't crowd out the rest of the covenant space so i like to give that answer because it's a totally like evasive answer right you say what's the optimal thing i'm like we don't need optimal like what are you talking about um but then what would the optimal um covenant solution look like well before you get to the optimal i guess it's important like what is the

goal of what is the goal of getting covenants in the first place what do they enable yeah they would they help support yeah that's a great question so what covenants are is the ability in bitcoin scripts for you to control what the transaction you're creating looks like control where where are the coins going under what conditions uh we have the scripting language that lets you do all through the cool mini scripty kind of things with time locks and htlcs and whatnot but in the end it's an all or nothing like you you're describing a set of conditions and once those conditions are met the coins can go anywhere and if you want to constrain where they're going like in the lightning network what you need to do is have a multi-signature and then

have all the counterparties who were like defined up front kind of agree that you know whatever protocol they're doing they agree to uh just only sign if the protocol has been followed and then you have all these like backup kind of kind of things so covenants let you enforce this kind of thing directly on chain and this simplifies a lot of stuff so all of these kind of what are called commit and refund style protocol so things like lightning where if one party is uncooperative if they post an old state if they refuse to sign something if they drop offline whatever there's a recourse right so something bad happens on chain and then there has to be like a backup

like take the coins back kind of thing and same with any form of atomic swap or like cross-chain swap kind of protocol, similar to Lightning, there's always the ability for one party to cheat initially, and then the counterparty has to undo the cheating. And this, for one thing, is inefficient, but it also prevents you from doing things where you really need the cheating to be impossible to begin with. So the classic example is vaults,

or maybe a more general thing is velocity limits, where I've got a wallet. I want to say the coins can't move except if they spend a day sitting in a staging area where I have time to react and pull my backup keys out and reset it if something happens so that way if my keys are stolen then then I'll have some ability to to do something about it with this claim and refund style things there's no I can't do that right so I for one thing there's no counterparty to like to be the one who's cheating the premise here is that like I'm the bad

guy and that my keys are stolen and my keys were used to do something bad and so there's no counterpart even if you could kind of define a counter party then once the coins have been stolen once the keys are compromised it's too late right like there isn't a refunding step there isn't like a bond or some sort of incentive it's I'm kind of babbling a bit because my point is that you just I can't construct a system that would work in the way that HDLC is enlightening and stuff work or arc or all of these cool multi-party protocols you can't construct such a system that would do vaults or do rate limiting so

that's in my mind that the biggest goal of covenants is to be able to say I have these coins, they can't move unless there's a delay, or maybe it's a low amount. Maybe I can delegate, I can say, well, here's this other signing key that I'm going to give to my kids. And with this signing key, they're only allowed to move a little bit of money, and then they have to send it all back to the original thing. And that's also a form of velocity limit, but it's kind of keyed to individual keys kind of thing. And you can imagine in a business context, right, you could like implement business logic, where you have keys that are eventually delegated. And in the end, like your accountants can go ahead and do petty cash kind of things on their own and

and more authorized people can do bigger things and stuff like that. So that's what I think of when I think about covenants. Yeah. And I like the way James O'Byrne described OpVault when he was working on that, which is basically create a simple structure that makes it almost impossible for exchanges to get hacked. In the business context, it's like create a transaction structure that really gives the user fallback and the ability to recover coins if they misplace them or some malicious actor tries to move them in a wrong way, which I think is very important. And in terms of incentivizing self-custody of Bitcoin, which I think is the end goal of a lot of people,

I think it goes a long way to solve that and make it easier for that and really strip away the fears that people have with securing private keys in the first place. Yeah, that's a great way to put it. It's all good. If we want to switch back over to block templates and mining, I'm all in. But this is like out of my realm here. I love the conversation, though.

this is great but like let's switch back to blockchain plates and mining real quick like have you looked in the hash pool at all i've been talking to the two guys here uh not to dox them or anything i think they were on earlier today so you can carry okay cool yeah i've been i've been trying to dig into that protocol to understand it it seems interesting yeah so this is a combination of cashew protocol and mining shares really you're trading hash cash for e-hash and creating a more liquid market for your mining shares immediately. So the idea is that you could solve the FPPS payout problem by giving smaller miners immediate liquidity

so they're not forced into these large FPPS pools. And then somebody's taking on the risk of those hash cash shares. So I actually think it could solve many problems, individual miner payouts. But then when it comes to hash rate derivatives, a market that's been trying to evolve for many years in Bitcoin, creating that two-sided marketplace immediately. It's a many-to-many instead of right now it's like one-to-one from a market structure perspective. Yeah, definitely interesting. One thing that I want to ask those guys is ensuring that there's liquidity within the mint so that when you try to redeem your shares, there's actually Bitcoin there. maybe there could be some kind of proof of reserve thing but uh yeah it seems interesting i think it's

yeah and that um i guess this actually gets into we can tie this all in together have a three-part like talking about covenants and all that uh in terms of like what has been unexplored that already exists today to scale bitcoin what are some of the the tools or areas that have little attention but probably should be explored more to see how far we can scale Bitcoin in its current state before moving on to something like Covenants or any other

potential future software? That's a fun question. So one observation is that in Taproot we have kind of this tree of scripts. So when you prior to Taproot if you wanted to use

Bitcoin scripts for mini script like things where you have so many keys and time locks and hash locks and stuff that are combined in some way, you needed to write out on chain when you spend your coins, you need to write out the entire script with all its conditions. You've got all these like if this and then otherwise this kind of thing. And like even the code that doesn't get executed, you have to post. And with Taproot, we instead have what's called a Merkle tree

that allows you to put every possibility kind of in parallel beside each other and you only reveal the one that you actually need. And this, if you've already got a script that naturally is this tree of if-elses, there's a clear application, but there's more interesting things where you can kind of think of every possible spending path as being a separate script. So as an example, you could say that you've got a 3 out of 5 multisig. okay then there are i think 10 different possibilities for that right it's five choose

three different possibilities for signers and so what you could do is use check multi-sig or now it's check sig ad right you could explicitly write out all of your keys and then provide signatures for the ones who are signing or you could break this out into the 10 possibilities and then have each one be a three of three with a different set of three keys and now when you have three signers the three of them just sign one of the branches that corresponds to the three that are available and you publish it. So what the blockchain sees is three keys and three signatures rather than saying five keys and three signatures.

And this is, well, already there's an efficiency gain, assuming that the two keys you save would have been larger than the proof that you have to do, but typically it would be. But what's more interesting here is that if you have a three of three, Or in general, if you have a multi-signature where everybody's participating, then the crypto story is much better. The cryptography story is much simpler as far as cooperatively producing a signature. So if you've got all these different branches that are each three of threes, different three of threes, you could use something like Mucig, or I guess Mucig2, to combine all of the three keys into one key

and all of the three signatures into one signature. So they have to interact. They don't have to interact to generate the keys, but they have to interact at signing time, of course. But they're already jointly signing a transaction, so presumably they're online. And in fact, with Taproot, we can do one better. So already we're down from having to post five keys

and only three signatures all the time to now we only have to post one key and one signature plus some sort of proof. Taproot does one better, where if your final spending condition is only a single key and signature, then you don't have to reveal any scripts at all you make your top level taproot key

be one of these joint keys with three participants and then if those are the three participants who are actually signing then you just reveal the key and the signature one key, one signature and this is tremendously smaller

than the naive three of five thing that also got great privacy and fungibility properties because one key, one signature looks identical whether you've got three of three or three of five or, you know, 50 or 60 or a normal wallet or a lightning HDLC that didn't go through one of the unhappy paths and stuff like that.

And this is all crypto that's pretty much deployed. I mean, Mucig 2 is... People use it. People use Mucig 2 in production. We've got a pull request to Rust SecP, which is a library I maintain that has not been merged after, like, eight or nine months. But I told them I would try to merge it this week. So it's not quite all there, but it's very close.

We've got to review the code. We've got to merge it kind of thing. Versus stuff like Frost, where there's still some complexity in defining what does the protocol look like? What are all of the different steps? Do we have a BIP? Do we have a specification? Then do we have a production-ready implementation?

Then do we have wallets that have embedded this kind of thing? MUSEc2 is pretty much there, right? But then going one further, we have stuff like Frost that let you directly do threshold signatures. Things like PTLCs, which let you do HTLC type, like chaining payment channels across each other, just using the interactive signatures. And so now we're just using straight taproot, one key, one signature, no scripts, and doing all sorts of cool stuff that looks identical. In fact, a bit smaller than an old school wallet with one key for one address. So cheaper, more private, less data intensive.

Yep, less CPU intensive as well. yeah yeah it was just fascinating was was it bit go when it comes to i believe they were using music to to pass information in the the nonce i think they came out of the paper like two years ago really pushing forward i think this idea i think it was yeah they've been kind of on the bleeding edge of like deploying things in real life yeah no but it's always um that's one thing as an observer as a user uh when it comes to these discussions about soft forks and covenants ire on the side of why don't we try what's at our fingertips now like is that a bad is that a bad approach to have you think there is a lot that's at our fingertips right now this is kind of the

approach that i take there's a counter argument right that this is kind of like avoiding problems saying like rather than dealing with the difficult kind of social problems of finding consensus on improving the bitcoin protocol in ways that it does need to be improved like right now you cannot do covenants no matter what all right you can with collider script and maybe we can talk about that but like right now you cannot do covenants in bitcoin in any reasonable way and we need consensus changes in order to make that possible and to get consensus changes there's social work to be done to actually build that consensus and to find out what the right protocol is how do we implement it how do we get it out there how do we deploy it and it's very

tempting to say well there's all this kind of low-hanging fruit here of like why don't I go work on scalability and why don't I go work on interactive signature protocols because then I don't have to touch the consensus system then I can just kind of build stuff and there's value to this right like it is important good stuff that improves scaling and privacy and fungibility but is also tempting to work on that stuff to avoid the hard problems that also need to be solved right so i guess on the hard problems question like how many hard problems need to be solved we know okay low-hanging fruit easy one unix timestamping bub like that needs to be solved

at some point yeah that's okay so that wasn't even on my radar but yeah so i if you just ask me without priming like what are the hard problems that need to be solved i would say right now in Bitcoin you can't do arbitrary computations right so we don't have something like opcat or big num arithmetic or kind of stuff that lets you do arbitrary just like general computation and we don't have covenants those are the two kind of expressivity features that we don't have you make a great point that also we have these like looming future bugs we have like the threat of quantum computers that we're probably going to have to address in our

lifetimes say um we have the uh the lock time running out a room bug which like will happen at a specific date right um this is an unsigned is it i think it's 2106 or something right it's the 2140 bug but it will trigger in bitcoin in 2106 not 2140 it's 2049 bug it'll trigger in that sounds right to me yeah um yeah and that one is some sense is easy because it's narrow and technical right so we have all these like kind of consensus cleanup issues um same like we should prevent 64 byte transactions and stuff and there's like maybe an open question of like has anybody like pre-signed like lock timed 64 byte transactions far in the future and we'd be censoring them and it's like no i'll go on the record just saying no nobody did that and like

if they did they were asking for it um but those things are narrow it's slow it's slow to fix them because there's still a lot of qa and a lot of um like defining exactly what to do and getting them out there but there's the social problem they're not too hard there the social problems are definitely hard about covenants they're difficult for general computation because general computation enables covenants by accident. They'll be hard for quantum computers because none of the quantum signature algorithms that are out

there have the great properties that elliptic curve signatures have, right? So we have to make things worse. I mean, if quantum computers can steal all the money then obviously anything's better than that, but they have to be worse than they are today in the post-quantum world.

How can you test quantum libraries if quantum technically doesn't exist yet? Is that a... This is, what a fun question, right? So the premise behind all cryptography is that you have kind of some sort of trap door kind of fun. You've got something that's easy to do and something that's hard to do. And what you want is easy to produce signatures if you know a key, and it's hard to produce signatures if you don't know the key.

We call that a forgery kind of thing. and there are kind of these limiting theorems in computer science to say that you can't just like literally make it impossible to do things if you have some certain knowledge but possible if you have the knowledge but impossible otherwise but what you can do is say if you can forge a signature without a key then you can use the same algorithm to solve some super hard computer science problem that we've tried for multiple decades to solve and we're pretty sure is impossible and an elliptic curve cryptography typically this boils down

to something called the discrete logarithm problem which is essentially the problem of reversing a public key to get a secret key and we've been trying for many decades some people have been trying for many decades to solve this and they haven't found any efficient way to do so and if they could break Bitcoin signatures then they could do so therefore they can't break Bitcoin signatures but we know that a quantum computer can right we understand the information theory behind the way that quantum computers operate and so we can devise quantum algorithms that would work if we could entangle enough qubits

and stuff and write algorithms that will will break these signatures so then in post quantum like how do we know that something secure post quantum so we obviously can't use these elliptic curves and these discrete log base signatures we have to come up with some other problem that's hard and it has to also be hard for quantum computers. And do we know of such problems? Well, breaking SHA-2, like reversing, like if I give you a SHA-2 hash, find me the preimage, that's almost certainly hard for a quantum computer, right? Why do I say that? Like if you prodded me and made me like justify this on a philosophy podcast, I'd have to throw

up my hands and say like well I mean we don't know but like I have a lot of confidence that a quantum computer is not going to break shot to okay can a quantum computer break unfortunately you can't make good signature schemes just with shot to you can make bad ones you can do lamp or a signature and then which are net signatures which have very large public keys and very large signatures so okay that's something and that would be secure almost certainly oh and they're one-time signatures if you sign twice you leak your key or part of key so like you got to be super careful about never RBF thing with the same

winternets key but if you want a good signature that like has a small public key and a small signature size and that's fast to validate these are all the things we need in Bitcoin then right now it seems like we can do this if you assume that like certain lattice based problems are really hard but are they I I don't know. It sure seems like they get solved in CS papers every couple of years, right? It's hard. So the broader cryptography community needs to go through this process of iteration and maturing to try to find better algorithms that will have the efficiency properties we need and also be secure.

And right now, all the attacks, because quantum computers don't exist at scale, all the attacks are like kind of these hypothetical mathematically described attacks that use this complicated information model and probably if people had a real quantum computers they'd be iterating faster you could like throw fuzzers at them and like whatever a quantum father looks like and we get a lot more iteration that way but like then it's too late if people are able to break stuff by like trying really hard with with real computers then then it's too late it's not a good answer but that's why that's why this

hard right like we've kind of got to make a bad decision here well i think uh it's unfortunate that bitcoin always gets signaled out when the threat of quantum comes up in the mainstream because it's like if this is a problem it's not a problem only for bitcoin right no for sure um it's a problem for for ssl so all of your like when you connect to a secure website the public key cryptography that's used to authenticate you to the website and to to uh handshake the encryption that would all break in the presence of a quantum computer the key exchange that you do when you like authenticate to a Wi-Fi network that would break in the presence of the quantum computer but the difference with

those things is that they're not consensus systems and they're for the most part they're kind of ephemeral right if I connect to my bank and it's secure while I'm connected then if somebody breaks my security and breaks that connection years later that's i mean that's not great but it's not the end of it doesn't allow them to then go take money from the bank because they're going to be like okay this session ended like eight years ago so why you can't you can't continue it and because they're not consensus system they can also iterate right so the bank can switch to like 20 different signature schemes right as they as they kind of get broken and like try to always stay one step ahead

and they can also kind of make worse it's because of that they can make bad decisions and it's not as bad as for bitcoin right in bitcoin if every transaction has these massive signatures that have like 100 kilobyte size keys that's horrible that's hundreds of kilobytes for every single key that's on the blockchain forever whereas your bank does it for some ephemeral sessions for a couple years until we find a better scheme whatever you know you waste some bandwidth for a few years and then it gets better so it's uh the problem they're harder for bitcoin i agree that's probably not why people sing but people like to single out bitcoin i think because it's big scheme that like a lot of people are jealous of but I mean but there are kind of good reasons

to single up Bitcoin it is uniquely hard for Bitcoin. This is probably a very naive question like when in terms of like data signature size and bandwidth issues if quantum computing exists would it alleviate those issues at the same time? Does that make sense? Yeah. Would we be okay with the trade-off for heavier signatures?

Probably. I believe, until quite recently, that if we had quantum computers, then we could do computational chemistry fantastically faster, because we could use the quantum computers to directly model the quantum mechanical interactions

of atoms as they form molecules. But then I talked to some computational chemists recently, and they were pretty skeptical about that claim. It's not the magical. That's like the slightly higher IQ version of believing that quantum computers do every computation in parallel universes. It's believing that quantum computers directly do quantum chemistry in their brains. But there are a lot of things they do. right so the um the way that they break elliptic curve cryptography and the way that they break rsa which works totally differently but the underlying algorithm they're doing is something called cycle

finding where you have some sort of function that repeats every two to the 200 times or every two to the 2000 times or whatever and well it's not exactly two to the 2000 it's some like large prime number or product of prime number that's kind of close to that and the problem is how do you find what exactly that period is, right? Or if you're, like, looking for a radio wavelength and trying to find it to, like, 500 decimal places, like, figure out an exact frequency, then quantum computers can do that, right?

They can figure out the period of these functions. And that building block is super general, right? There's all sorts of open computational problems, like optimizing logistics, the traveling salesman problem. So when you're a UPS driver trying to go to like a thousand houses in a day what is the optimal path to do that um or well if there's a whole bunch of houses there's so much density now there's a good chance you just like do the garbage man route because everybody has a package but imagine you're like doing higher

level logistics and you're going to like 50 cities then well you're probably not going to every single neighboring city right then you really got to do trade-offs and there's not good solutions to that uh we could do that much better um if you're a drone and you're doing like as a crow flies kind of things and suddenly this uh these kind of logistical problems uh become hard and much more meaningful as well right like if your drone can travel 10 percent less distance that's 10 percent less fuel you're using versus if the ups guy has to drive 10 percent further right that's probably just a rounding error something right it's not um he's also got to sit in traffic right he's also got a you know refill on gas and whatever but a drone has only finite distance

that it can go, and it's, you know, like, every 10% you can eke out, that's a huge win. Is a trade-off worth it? I mean, it would end my career. It would be kind of, I don't know, it's kind of an aesthetic sort of end, you know? I could be like, oh, hey, grandkids, you know, I used to work on cryptography until the field was shut down with the quantum computers, you know? and it would be kind of this cool like just thing that existed in one era

of cryptography I mean there is quantum cryptography and I guess I could learn that but like in my head I don't because it's too hard probably yes TLDR I think the trade off is worth well thank you gentlemen

both for joining thank you this has been incredible thank you for all the work that both of you are doing in the space I think a new speaker starting in there so if you guys want to get in there great thanks so much yeah i think it's your time i think it's peter appreciate it yeah okay and we're back at bitcoin plus plus sitting down with liam egan it's reminiscing i uh i remember you coming to bit devs a couple years ago talking about bulletproofs and i believe at that time

you weren't really into bitcoin is that correct or i mean i wasn't like working professionally on bitcoin but i've been sort of uh casually interested for for a long long time like um i remember my friend's dad was like mining bitcoin in like 2011 he worked at an fpga company he had all these old that's that was that's the first i remember of it but yeah so you got the six month window of fpga mining yeah yeah well he had like a bunch of them from work so yeah yeah so what uh let's discuss what you're working on now you're working at alpin labs working on roll-ups on bitcoin admittedly not as well versed on roll-ups as Maybe I should be, and I think that's probably due to some preconceived notions that have arisen in my own perspective

due to how they've been implemented on Ethereum, which seems to me, as an external observer, to be suboptimal in terms of creates this liquidity fracture. And for the Ethereum protocol specifically, it's pushed a lot of activity up, which has, I think, been to the detriment of their base chain. yeah so just to define things because people often use these terms in confusing and inconsistent ways like there's like this notion of a side chain which is like the most generic kind of term and then like a roll-up is like a layer two where you put the data of the layer two on the l1 and then you either use like an optimistic construction or a zk construction to prove

that you are running the roll-up correctly and move money into and out of this side system. And so, yeah, like, the way that they work on Ethereum is you have, well, I mean, it depends, right, on the roll-up, but, like, the kind of generic instruction is you have, like, for a ZK roll-up, a SNARK verifier that lives on-chain, and you can just interact with it

because it's, like, a smart contract, and we can't do that on Bitcoin because we can't verify snarks directly on Bitcoin. So like at Alpen, we're using BitVM or BitVM2 to like kind of optimistically verify a snark, which then verifies the correct execution of the rollout. And for those who are unaware of how BitVM works, what's the, how does that work?

Yeah, so BitVM is like, people sometimes use it to refer to several different things the version that we use these are all from like Robin Linus at all and so basically you

if you had no kind of limits on the amount of Bitcoin script you could put into a single transaction or block size or whatever you could write like what we have basically I think 750 megabyte Bitcoin script that verifies

a snark like a zero knowledge proof and if you could you could put that on bitcoin it would be verified as part of consensus and you know whatever but you can't so what we do is break it into pieces so it's like you run the program up until this point and then we at each point where it stops so you can run it for like four million operations and then you stop you commit to the state so you basically hash the state of the program at this point and then in the next chunk you start from that state again and so if I'm running the bit VM program where I'm proving that it's run correctly I commit to the hash of each state while I'm running the program and then you the observer like the watch

tower or challenger or whatever can also run the program and you see aha like this state that you committed to here is wrong if you run it from the previous state to here you get a different state and so you can then challenge me by putting that chunk of the big program on chain so instead of running the whole thing you just run this one little piece and you can show that this piece fails and the that can be verified as part of consensus so is it similar to Bitcoin mining in the sense that to build a block you need to reference the TX hash in the previous block to be able to do that so in order to make bit VM work as

As Bitcoin exists now, we need to pre-sign everything in order to have these, like, multi-stage sort of transaction graphs. If we had a covenant, we could avoid that, probably. I think the work on CTV Bitcoin is, or BitVM, is still ongoing. So it's unclear exactly what you can get rid of or not. but uh yeah so you like you pre-sign this whole like state transition graph and what is the ultimate goal of these these roll-ups what can they enable yeah so there's a pretty good article actually from um a dba it's like a vc basically like um bitcoin right now lacks um certain kinds of functionality like people have different

feelings about this but for example bitcoin doesn't have as robust privacy as you would be able to obtain in zcash or monero or any number of other like tornado cash these like other kind of privacy protocols that exist outside of bitcoin and from my perspective privacy is is crucially important for fungibility and like just the health of the system and anyway if we have a roll-up and we can verify a snark then on the roll-up we can kind of do whatever we want the snark is like a universal verification primitive so once we can do this one thing which is verify a snark whether it's with bitvm or something else then on the roll-up we can have strong privacy we could use like shielded csv which is a protocol that i co-authored with jonas nick and robin

Linus we could also implement things like bitcoin backed loans and like all of this could be done in a way that doesn't rely on or it relies less on trusted central parties so right now a lot of these things if you wanted to do them with bitcoin you'd have to use a custodial service or something else and I don't know I just the vision for me in terms of roll-ups personally is enabling like robust privacy on bitcoin but also just any number of other things what are some of the trade-offs that come with this model with um are they're enabling more stuff on bitcoin or with bitvm like as an end user with bitvm like are there custody trade-offs are there oh yeah so um it it depends what you compare it to so the advantage of bitvm is that it sort of works

like for some definition of works right now with no changes um it requires what we call like a one out of n assumption this is a little like confusing sometimes i find people are confused by it so you could build a roll-up with a multi-sig right now like um i they don't call it a roll-up but like essentially like liquid right you have a federation and there's some subset of people have to sign in order to move the money into or out of the side chain that's fine this has a liveness safety trade-off so you you are assuming that some subset of the signers are not colluding to be evil and when you make the set of like colluding signers smaller or like larger like you require more people to collude to be evil the system becomes more secure but then if any of us

people go offline then the system can deadlock and you can't do anything what bitvm enables is all n parties that set up the system must collude in order to steal the money and if any one of them is live the system will still remain live so this is the trade-off of bitvm now if we compare this to like a snark verifier on ethereum it's fundamentally different on on ethereum you could verify a snark as part of consensus right now so it just reduces to the l1 consensus so if we enabled opcat for example we could verify a snark directly on Bitcoin as part of consensus that would offer a significantly greater degree of security as compared to bit VM but would require

changing Bitcoin and then another just trade-off I want to mention which is sometimes ignored is like the complexity like it's very complicated the system that we're trying to build and work with. So, you know, it requires a lot of eyes and work and audits and whatnot, so people should keep that in mind as well. Is the hope that by proving this use case via BitBM could create some support for a Covenant solution like OpCat or CTV, whatever it may be that would then enable um you to do what you're doing in a much more efficient

easier manner and alternatively if even if that does happen but a software doesn't emerge can the scale using bit vm as it uh as it in its current state yeah um i think that like bit vm does sort of force the question so to speak because there's like a lot of ambiguity about the trade-offs with different soft forks and bit vm i think sharpens the discussion because it's like we we're going to do this now maybe we could do it better if we did something else but it's no longer the case like we're blocked on this like i don't know kind of vague political vagaries um but the yeah so i don't know i mean i think i think that the system overall would be would be better i i mean essentially this this is what it comes down to in my opinion if you think that

verifying snarks on bitcoin is something we should do or should be able to do then i think long term we should have a better way of doing it than bitvm if if you disagree with that then like it's it's different you're in a tough spot i think but like i'm not sure i can convince you of that but if you're willing to accept that premise then i think that like we should pursue solutions that are less complicated and have better trust assumptions than bitvm and uh we we are i don't know if people are aware of jeremy rubin's like garbled circuit construction i've been working on like the same kind of thing in parallel and hope to publish something soon um about that but it's like an alternative to bitvm that also doesn't require modifying bitcoin but it has many of the

same trade-offs you have to like pre-sign stuff to simulate covenants and and so on so yeah we were talking about this before we hopped back on air but i believe one of the big problems with roll-ups on ethereum is this liquidity fracturing and is that something from a design perspective you guys are thinking about it alpin yeah definitely um so this is like this is absolutely a big problem and there's a lot of people who work on on this on ethereum there's like espresso and i don't know to be totally honest i don't fully grok the complexity of this problem i'm more of like focused on snarks but at alpin um our goal is to have like this kind of neutral orchestration layer that sits between bitcoin and the alpin roll-up

so the alpin roll-up will like be evm but you could imagine that a different execution environment as we call it would use the same bridge and so because of the extremely high complexity of bitvm getting people to use this like shared kind of infrastructure would hopefully make liquidity fragmentation issues like less severe and by building this like with the kind of knowledge of how things have gone in other ecosystems from the get-go hopefully avoid the problems does that coordination layer create a centralization effect in any way well i mean there is this sort of bridge that everybody's using so in that sense it's centralized but because of the properties of bitvm we can have a large number of people

guarantee the security of the bridge um so you know i i would say yes i mean i think the vision is that the the orchestration layer is neutral so it's not like you know it's kind of like shared infrastructure that people use um and it's not centralized in the sense that it's like controlled by a single party but it is sort of used by like one avenue that people are going through that yeah yeah and because of the high cost that it does kind of i think make sense yeah fascinating yeah um so beyond beyond this we just had andrew pull sure on just talking about like other areas within bitcoin where you have things that are available today that aren't being leveraged as much we mentioned frost music too um maybe

combining something like protocol layer bitcoin lightning network e-cash schemes to create these these setups like have you have you looked into the e-cash development there and are there ways you could solve these problems using that which problems do you like roll-ups or well just getting the same sort of usability that we're trying to attain via roll-ups in terms of more private more scriptability and do do great financial transactions in my opinion a lot of the there's a lot of people working on different pieces of these sorts of things this is going to be very vague but it makes sense to me like in different places in different ecosystems lots of people are working on different parts of something and it seems to me

like there's a sort of confluent element to a lot of the work that it like i think shielded csv is kind of in that direction towards like this confluent solution that like combines scalability and privacy and uses the blockchain sort of as it should be for minimal data coordination like Sean Bowe recently had a cool he's one of the Zcash

sort of architects core devs and so he has a thing that's like I think related and like anyway to answer the question directly about Ecash I think that Ecash has like several really nice

properties it scales very well It doesn't have any kind of on-chain footprint for internal transfers to the eCache pool or Mint. But the fundamental problem with eCache is that it is not auditable. You cannot, as the Mint, prove that you're behaving honestly, at least in the current design. I think that this is okay for certain use cases. but for other use cases it's um you know not feasible i mean in some ways like bitcoin and then later zero zero coin zero cash we're like kind of trying to solve the auditability problem of ecash which precedes all of these technologies by like several decades um so uh yeah i think you

can use ecash for stuff but in my opinion if for like private decentralized permissionless payments you kind of want something auditable like where you can be certain that people aren't sort of doing secret inflation or stealing from you like you can see the money moving into and out of the mint but you can't be certain that there aren't like a bunch of outstanding tokens or something

yeah I guess the the solution to the auto problem with these e-cash solutions today to simply just like manufacture runs on the mint yeah a certain cadence to make sure that they're acting honestly yeah I mean like

i think it's like a good pragmatic solution in certain cases but like i don't think you could replace like monero with e-cash it just it wouldn't work because it's it's fundamentally more trusted more trusted yeah you have to trust that somebody isn't isn't doing the secret inflation even if you do periodically check and the frequency of the checks i suspect is is inversely related to the privacy of the system like if you have to keep opening up your money like i don't know i mean i don't really understand all the specifics of how these things work but i think it's possible for what it's worth to get similar levels of scalability and avoiding like publishing data on chain with something like shielded csv like these recursive validity proofs for

transactions that are verified client side um and then combined with this i think i don't know what it's called his thing but sean bow's thing or zk coins where you can forget the history like there's some version of this that combines all of the nice properties of verifiability and like ephemeral data and and all like i don't know i think i think this exists or we will eventually someone will eventually find it yeah that's i mean a big part of the broader discussion over the last two weeks has been like what should block space be used for is it this include inclusion of arbitrary data to do things like mint inscriptions or ordinals or i guess with the data and bitvm be considered arbitrary but it's enabling what some would argue to be monetary

use cases on a different layer yeah yeah i mean we are very much using um bit or um like bitcoin for data availability um like the terminology of roll-ups like some people say that roll-ups turn everything into a data availability problem where it's like the proof would so for example right in bitcoin you have all these signatures in a block you could replace all the signatures in the entire block with one snark that just checks all of them because it doesn't actually matter what the signature is it just matters that it exists and so um like we're we can remove certain data but then other data needs to be there right like the transfer itself you need to know what it is because otherwise you don't know what the state of the system is so

um yeah i i don't know i mean this sort of stuff is very very political i think probably too much political but one perspective that was interesting that i guess i hadn't really thought about or hasn't been sort of stated so cleanly was like the reason why there's a block space limit is is multifaceted but a big part of it is the the time to verify a block it's not actually the data itself like network bandwidth and storage are relatively cheap the the hard part is verifying a block so if you made a block a hundred times larger you you would you'd be in like the solana situation where you need like a rack to like run a node so it's too too expensive but for making data available it doesn't cost anything to verify because it's not verified as part of consensus you

know maybe you're verifying a single snark for its validity but even then i mean you don't have to do that every time so yeah yeah we're thinking about that's fascinating i think that's the one thing i've said multiple times today it just seems like a lot of people are talking past each other and there is clearly a miscommunication problem or a communication problem maybe it's not even miscommunication just like how to the developer community working on these different protocols and these different repositories clearly communicate how they operate and what they're doing and how that affects end users it's fascinating yeah i don't know i think i think people are often nicer in person yeah i've experienced that today at this conference it

seems pretty obvious uh liam thank you for hopping on i know we just grabbed you thank you Keep crushing it. It's fun to see how involved you are in the space considering where you were when you came to BitDevs a couple years ago. Yeah, yeah. It's been a fun journey. Thanks.

All right, thank you. All right. We are going to transition. I've got to run over to Bitcoin Park Austin to lead a panel there about I don't know what it's about I have to check it on my way over there I'm moderating though

it's easy to moderate and so I'm going to hand over the mic to Marks from Open Secret who's going to round out the day of interviews here he's going to take it for the next hour or so but this is fun I hope you guys are enjoying this this is the first time I've ever done a news desk

a live news desk new medium trying it out ad-libbing something I'm not used to but we're here and i think i just want to reiterate what liam just said uh i can see it up close and personal people are much nicer in person than they are in the internet when you're forced to look somebody in the eyes and have a conversation i think the uh the shit slinging reduces significantly marks yo i will get out of this seat so you can take it over thank you sure yeah thank you for taking over for me it's been a fun day yeah it has are you ready to ad

i know you've been live streaming and you've been doing freedom tech yeah mine i usually like plan ahead of time and then i do a half hour of live streams this will be totally new i'm jumping in the deep end right now well that's right we've got a few i definitely want to talk to gloria myself at some point we've got gloria over there jameson lop yeah i know jameson and nifty were supposed to come on together yeah joint conversation um but yeah do you want to go grab gloria and bring her over see if she wants to chat yeah either way okay yeah or jameson looks like he's ready to come over you can yes it's sort of free-flowing over here it's free-flowing jameson i hate to do this we're you just missed out we're transitioning hosts from myself to

marks so marks is going to come take my spot it's good to see you sir it has been a while we're talking we were just mentioning how uh people are much nicer in person than they are We can come back later We could have Jameson and Nifty chat later We could do that We got Jameson on here Does one of you want to come up here and join Jameson and Marks? I think we're probably talking about different things

What are you talking about with Jameson? We were going to talk about Nifty was supposed to be on with him at this time slot We could talk about me being the main character Are you the main character? Are you the main character? According to some people. Okay.

Oh, is it? Oh. Logan just said that. Marks, I'll let you take over. We're going to talk about Jameson being the main character. I'm going to hop over to Bitcoin Commons. We'll be back. All right.

What's up, James? one talk down two to go okay cool I know that nifty was gonna join here as well but I don't know where she's at so anyways so you had one talk already you said? Yes, I just gave my presentation entitled the user is the enemy. Okay, a short version of this is that, you know, self custody is great, but it comes with a lot of responsibility. And, you know, over the past decade or so, we've done a great job improving the security of self custody setups for people. And the problem, though, is that I think we've got a blind spot where we've really been focused on security from external attackers you know hackers so on and so forth

but we haven't done a great job protecting the user from themselves okay so so are you talking about um so we have like the big infamous ethereum hack that happened recently right where that was an outside attacker but it was also from within the house where they messed up in a way right so does that play into this yeah i mean that's one potential issue uh you know that was that was a culmination of a number of different problems, both at the protocol level and the hardware device level and so on and so forth. Though I would argue part of the problem with that also is just that in that situation, they were essentially using free software. I don't think that they were really paying any security experts to evaluate the security posture of what the security model

was that it was actually offering to them and so I think it was a good takeaway though that just because you have multiple keys like multi-sig is not panacea it's a good start but since I've been doing multi-sig for a decade I've seen a number of different times where a multi-sig setup has been architected in such a way that it's really no better than a single sig setup. And this was one of those times like, single point of failure for multiple reasons within problems within that technology stack that they were using. Okay, is that because like, so say you're talking to an average

person at home who wants to do multi sig, if they're keeping all the keys in the same location? Yeah, then it's just a matter of inputting that happened to Yeah, yeah, it's It's like the strength from a multi-sig setup doesn't come just from having multiple keys and needing multiple signatures. It comes from using a diversity of different hardware, software, and security models around each key. And it's this diversity creates strength. that you think of it as like overlapping armored segments um where if if like if all of if all of the segments are the exact same then uh some some uh potential exploit or weakness that will affect

one will affect all of them and the reason you want diversity is you know that things will go wrong but if something goes wrong you want it not to be catastrophic you you want it to only affect like one of the keys and not all of them true okay would you give similar advice to somebody who's at a company trying to protect a corporate treasury versus somebody at home just doing their own thing ultimately like it comes down to like how much value are you protecting right is this a corporate treasury where the the corporation would be in dire financial straits If you lost that Bitcoin, you need to put a lot of resources into thinking through the security around it. And a similar thing with like the individual.

If I generally categorize it in like three different categories, there's like pocket money. You know, if it's like the amount that you would walk around in your wallet with cash and you wouldn't really miss it if it went poof, then sure. Just like keep it on a hot wallet, whatever. You lose it, it's not the end of the world. Then there's, like, okay, I have, like, a small savings investment amount. You should at least spend $100 on a hardware device, you know, because taking those keys off of the Internet will protect you from, like, all of the hackers, external attackers and stuff.

But then it's, like, is this a significant investment? Is this a large portion of your net worth? Is this generational wealth? If so, then you want to protect it pretty much at all costs. And if you want to protect something at all costs, then you want to eliminate any potential single point of failure that could cause a catastrophe. And that's where CASA is really aiming to serve that market, where you're willing to put in the time and the effort to actually be thoughtful about the security and be paranoid and adversarial in your thinking so that you can protect against everything. And as I was just talking about a few minutes ago, that includes protecting against yourself. Yeah. So Casa offers a multi-sig where you can do more than three, right? You can do

larger multi-sig. Yeah. So, uh, we support generally both two of three and three of five setups. And, uh, is there a time when you'd recommend one over the other? Like, especially if it's a, an at-home person who's trying to secure their family generational wealth? Yeah. I mean, I think two of three makes sense. I basically say if you're a whole coiner, you should probably at least have like a two of three set up so that you have a decent amount of redundancy. If it's large amounts, like if we're talking millions of dollars, then it starts to make sense of going even more distributed.

And the whole point of this is to actually make it harder to spend those coins. Yeah, for sure. It's harder both for you to spend and therefore harder for an attacker to spend and harder for some sort of disaster to strike that would compromise a sufficient number of them. Yeah, definitely. Okay, so what about, let's say that you've done the best you can to get the right wallets, to get geographic separation,

whatever things you need. Do you see a place for things like insurance products that we see happening to insure Bitcoin? And I don't know if Casa's going into that. You don't have to talk about future stuff with that if you don't want to. But, like, where do you see insurance playing a role into this? Yeah. Well, and, you know, I think that, like, what you're seeing with Anchor Watch makes sense because the insurer is only going to want to insure something that is pretty low risk.

And so the setup that they put you into is already very low risk in the first place. and the main reason why it's even insurable is because it's such a robust setup that it even you know even if you like blow up all of your keys there's still a recovery path there's a way to get out of that and so that that is something that we think is totally worth exploring you know there's a number of different ways that you can go with it it's problematic for casa because we We don't do AML-KYC, and we don't really want to. And so that's not really compatible with a sort of legally insured framework. And I think what will be interesting and what we've been thinking through

and talking to various providers about is, you know, are there ways? It's really just a question of, like, who holds these other escape hatch keys? Like, who's going to be a sort of recovery participant to help get you out of a catastrophic situation? Protocol doesn't care. It could be anybody. The bigger question, I think, comes down to, you know, legal issues around who's comfortable doing it, what's the risk and responsibility around, for example, being a a one of in uh key recovery holders for third parties it's a fairly unexplored space yeah

okay um do you see so talk about for a corporate treasury kind of attrition of of employees or attrition of executives and having to rotate keys and stuff what do you recommend it's good good safe practices for that kind of a scenario where you've got an executive who was one of the signers and either leaves on good terms or leaves on bad terms and maybe is takes their key with them or something like what what do you do protect that kind of situation so you know there's a number of different ways that that can go uh you know one interesting thing about distributing keys to different members of an organization is that you don't necessarily actually have to give them access

to the private key material. What you can do is you can essentially load the key onto a hardware device that doesn't allow exporting and give that hardware device to the person, and then they can sign with it, but they can't actually get access to the key material, and that can give you a slightly different security model So do they have the pin to the device maybe or a biometric, something like that?

Yeah, they can unlock it and sign with it. Yeah, so that's one way that you can make it less likely that you might need to actually do a key rotation. So then, of course, if it's a voluntary termination and they just turn in the hardware device, then you may feel comfortable enough that you don't actually feel like you need to do a key rotation. if it's involuntary or the you know device is lost or you know they just run away then you may want to do one anyways just in case right and what if there's a rift 50 50 rift with the executive team yeah but you know that's no different from any sort of corporate governance issue right But so governance is a tricky thing to think about.

And now when you're adding that into having control over bearer assets, you just have to be even more careful. You really just talking about how far do we need to distribute trust amongst like the officers of an organization such that we can limit the potential for catastrophe if there's some sort of governance crisis. Yeah, and then is there a layered structure? You talk about having a hot wallet and having your cold storage that it's hard to spend.

Would there be something where the executive team has a cold storage amount that they have access to, but then maybe the board of directors has a more hardened treasury that their signer's on so you can kind of layer your risk that way? Absolutely. That'd be interesting. Well, speaking of governance, I wanted to ask you, you know, obviously a hot topic right now is like Bitcoin Core and that kind of stuff. Do you have any thoughts that you'd like to discuss at all or with kind of not just the OpReturn stuff, but maybe how Core is being run right now and how this is going?

And yeah, well, I mean, I think there's a lot of disconnects, mainly because a lot of people who don't participate in Bitcoin Core don't understand how the organization operates. I'm an outsider and don't really know. Like I use Bitcoin, I run my own node, I connect to it myself. So I'm not a developer. So I'd love to like understand more how it works, maybe in your mind how it works and how it should work, possibly. yeah so you know for example one of the big disconnects and this has happened a number of times in the past isn't by no means the first time that people have been afraid the Bitcoin core is going rogue against the will of the users and

all this other stuff I had a tweet about this it was basically you know from time to time someone will get upset about how a given issue is seems to be going in a discussion on bitcoin core or the developer list or whatever and if they feel like it's not going well they may take their grievance externally to social media and then what happens is a lot of people who are on social media but are not familiar with bitcoin core governance may feel like it's wrong you know perhaps people are being censored so on and so forth or they believe that this is a direct democracy and because a github repository isn't you know open to the public, essentially, that anyone who goes and comments on a given issue is essentially

recording their vote. And they don't understand that that's not how Bitcoin Core operates. This is it's a meritocracy. It's not a democracy. The volume of opinions is irrelevant. It's the quality of opinions that get judged. And so there's also this signal and noise issue. And this is why, for example, this particular PR got locked because so many people were coming in and it's not that there were too many people. Like more contribution is always preferable if the contribution is high quality. If the people who are coming in are contributing novel perspectives and ideas and arguments for or against whatever is happening. And so this is the signal-to-noise thing of you have to understand that this is also essentially a workspace. There are people who are trying to coordinate.

This is a decentralized organization, and this is the centralized shelling point that people are using as their... Slide on over here. Why don't you switch microphones? You can give Nifty that one. We've got them numbered here. Finish your thought, and then, yeah, we'll talk her in. So this is a workspace and if there's too much noise, if there's high noise and low signal that is occurring in this communication channel, then it makes sense for censorship of low quality noise to happen is basically a denial of service safeguard, right? It is the filtering of Bitcoin Core against low quality noise that is not contributing to this discussion.

Yeah, I mean, we're experiencing that in this exact moment right now where this session just got out and there's all these people walking around talking and a lot of noise in here. It's kind of hard to have a conversation. I like your concept of this decentralized workplace. And if I picture like an actual company office and somebody come in and start ranting and railing and disturbing people, I can see how that would be, you know, get in the way of truthful conversation and real work at the end. Yeah. Nifty, welcome. Hey, how's it going, guys? Good.

How you doing? Good. We're doing really good. How's the conference? What are the vibes so far? Vibes are really chill. Good. I think we're all hanging out in the mempool today.

It's in, you know, nice weather, lots of sun. Nobody's peed in the mempool yet. that we know of yeah yeah i was over at the mining summit yesterday over at bitcoin park austin the new name and in the men's room it was very crowded and somebody walked in and said i wish the men pool was as full as this yeah oh yeah well that's cool um yeah so we were just finishing up a conversation about governance and bitcoin core oh yeah you have any thoughts there you want to share on that topic honestly not a whole lot i don't think about it that much i mean My experience with Bitcoin Core is I run it and remember to update it occasionally.

Sometimes they have proposals that I'll go look at. I don't think I've weighed in on anything recently, though. I don't know. If there's something I want in Core, I usually submit a pull request. I don't have any pull requests that have gotten accepted yet, though. As the main character for the past week, which was a fascinating turn of events for me because, you know, when I went to comment on that pull request, Satreia didn't even cross through my mind.

And that's because Satreia doesn't really have anything to do with that. I mean, yes, there was some discussion on the mailing list about that, but, like, Satreia doesn't care. Satreia doesn't need any of these changes. They're not pushing for any of it. It's just tangentially related for one edge case with their fraud challenge mechanism. But anyways, that's an aside.

I'm glad you point that out real quick because I think a lot of people have latched on thinking Citraea is trying to attack Bitcoin Core, right? They don't need it though. I think the thing is... Yeah, so I'm glad you called that out. Yeah, I think the important thing about the Citraea stuff is they can make their project work without any changes to Bitcoin, which is part of how they've architected their project is we're building this thing, we're excited about it, and here's our architecture.

And this proposed change with the Operture and stuff will make one small part of their projects different. But it doesn't really make it better for them. It's about making it better for Bitcoin. Right, yeah. But sort of as a meta to all of this, and what kicked off a lot of the crisis was, of course, that I supposedly had a conflict of interest. Okay. Oh, yeah, that's right. Yes.

It's dirty VC money. Yeah, yeah, yeah. And so, you know, what I've been telling people is I have a financial interest in Cetrella. I have a much, much larger financial interest in Bitcoin. That's a good point. I argue that I actually don't have a conflict of interest with Cetrella because this doesn't meaningfully affect Cetrella. But even beyond that, even if you don't want to believe any of that, you know, part of the reason for the moderation action that happened was that commenting on other people and their motivations is considered off-topic.

And so I think that that's an interesting rabbit hole to go down because it's very clear that a lot of people on social media think that it's actually highly relevant for people to disclose their incentives. And I disagree with that. And the main reason is you have to think about it adversarially. There is no way for us to possibly know what everyone's incentives are. We can ask people to be honest. We can ask people to disclose. There's no way for us to actually know if they're being honest. The reason why this all got blown up is because I have disclosed it. My name was in the press release. It's on my website that I'm an investor, so on and so forth. But you can only ever know your personal incentives.

You can speculate about other people's incentives. But I think that doing so is a waste of time, especially from the Bitcoin Core organization point of view. This is why the guidelines for discussing things on Bitcoin Core are to focus on the idea and not the people behind the idea. We have a lot of controversial people in the Bitcoin space, and if we focus on the people rather than the ideas that they're proposing, then we'll really be ossifying sooner rather than later. Yeah, I think that's true, but I have to admit that understanding people's incentive structures has been very useful for me to understanding what they're trying to communicate. because sometimes i feel like when communication comes out and comes across part of what can make it difficult for me to understand what they're saying and why they're saying it is understanding motivation behind it so i mean i think i have to i will say like no i don't think financial interest

is always or in many cases is usually not especially technically minded people the reason for stating a thing but usually there is some motivation for saying the thing right and so figuring out what the motivation is hopefully it's a technical one but sometimes i don't know sometimes i've seen people not in bitcoin core like more broadly and like business or you know more governance style things where decisions are trying to be made i have seen situations where it's like why is this person behaving in this way that just seems a little off or like i can't figure out why they're behaving that way and in those cases figuring out what the incentive structure is usually explains the behavior i think that the best

example of that was actually from the block size wars and segwit and how we did not understand for several years that really the reason that bitmain was against segwit was because it screwed up their covert asic boost but you know even even with us knowing that it screwed up their covert asic boost in no way had any impact upon the technical technical arguments like for segwit i guess so like when i think about like you know governance or broader questions about why people are motivated to like particularly advance a viewpoint though like independent the technical stuff it's always i find it usually it's very it's good to know what the incentives are because sometimes that can explain why you ever meet resistance when you're like well the technical

says that xyz is the best like why aren't you supporting that maybe like i mean okay like you know a lot of the bitcoin core i think a lot of bitcoin core arguments are technical some of them I would say are also taste judgments, like judgments of taste, and so that is slightly preferential, right? Well, that's the main reason why this is such a contentious debate, is because it's not a technical debate, or let me put it another way, it's a technical debate versus a taste and ideological debate, and so that's why I think that never the twain shall meet. These two different sides are talking past each other, because the, you know, I would say the majority of the technical people are in the camp where they're distasteful of a lot of the

arbitrary data use on Bitcoin. And then a lot of people who are prioritizing the ideology of what Bitcoin should be for don't really care so much about the technical arguments because they think we should be fighting regardless of how difficult the fight is. Well, because Bitcoin is both technical but also an economic project, how much of that economic discussion should come into the technical discussion right because if we're deploying software to a whole cloud cluster it's not bitcoin it's just regular software we have to discuss what are all the knock-on offense effects of this algorithm that we're deploying right so how much of that those economic effects should we bring into the conversation with tweaking some things i think like ideally a lot

so i think some things to maybe are good to know to keep in mind around a lot of the technical discussions and a lot of the hang-ups that i think we run into in core development around proposals to change it not entirely but a lot of them are about like the ability for the network to stay up and running in a way that any like actors who want to come in and maybe submit a lot of transactions try and take the network down that sort of thing like maybe this is my like a little bit skewed perspective but i feel like a lot of conversations especially a way that what people are focused on in particular is um just making sure that bitcoin as a protocol is robust against like abuse to some extent and when you map that back to like okay how does that represent like

the value it's like well bitcoin being a robust network is the way that we preserve the value of being able to use it right um so i think like like yeah i think questions of value do come into it But I think every person who holds Bitcoin would agree that the network staying up and staying robust to attacks against it is the most important thing in terms of value preservation. yeah i mean i think it's it's also interesting that um the longest streak that we had of fees eclipsing subsidy happened because of token mints that you know a lot of people find detestable and i'm not i don't really care about like the tokens or the nfts or the semi any of this other stuff but i am interested from an economic standpoint in allowing people to experiment to find whatever the most valuable uses of block space are now you would think

hopefully that the most economically valuable uses of block space other than of course just securing large amounts of bitcoin would be for like anchoring second layer networks into bitcoin and to date that hasn't really happened but we also have not had a lot of great tooling to allow people to build permissionless uh second layer networks and that all ties back into you know some of the the latest debates and how i find it very interesting that a lot of people at least who are in my replies are basically saying that you know citra as a second layer network is is all just shit coins like you can use it for a number of different things but if the shit coins are more willing to pay for that block space than anyone else like yeah it's like your subjective judgment

is that this is shit coining and the objective judgment will be whatever the fees are that they're willing to pay for i'm sort of like i i see shit coinery as being quite tragic and that to some extent you must acquire bitcoin to create shit coins on the block space does that make sense But the Bitcoin that you acquire goes to miners as fees to something like the majority of the Bitcoin that that shit coinery that's happening on Bitcoin that it's being involved in is acquiring enough value out of the ecosystem that's buying the shit coins, converting that into Bitcoin and handing it to the miners. Yeah, I think they've collected like six thousand Bitcoin from ordinals related stuff over the past couple of years. That's not nothing. That's an insane number. And so to me, I see that as like that's like a wealth redistribution from shit coin communities.

to the mining community to some extent right like they're taking value out of their system and having to pay it to miners in order to make these protocols work and at the end of the day like you know maybe it's like i just think kind of tragic that it's not staying in those communities that have that value and why are they doing it well they're doing it speculation like it's a casino right they're trying to hope that they're able to pump up their thing or whatever um yeah i know well and with those financial incentives what role do miners play in this whole discussion are they pushing for one side or the other and like what role do they have in bitcoin core technical discussions in general that's an it's an interesting thing because i've been seeing this

term ideological miners uh more and more frequently lately and personally when i read that i read uh you know economically illogical miners or you know basically people who are willing to mine and not maximize their profit. And I think that there will always be some people out there like that, but that it doesn't quite work. When you're talking about the current state of mining, which is highly institutionalized in a lot of public companies, these are organizations that have fiduciary responsibility

to their shareholders to maximize value. So expecting any of them to be ideological and to basically leave Bitcoin on the table, I think is you're just dreaming. But I think that's I mean, I think you're talking earlier about economic incentives and core. I feel like the economic incentives and miners is way more important to a large extent, mostly because it's like, OK, you know, you get back to this like filtering. Like there's certain transactions that people are ideologically opposed to because of the content of that transaction being something that they don't think should be entered into the permanent record Bitcoin record. To some extent, that is a taste over economics. right like it's a it's a taste decision that's outweighing your economic interest assuming that your economic interest is to collect as many satoshis as possible right

yeah okay yeah um so then in that regard uh oh man i lost i lost my question i had there so right now then if if opera turn opera turn limit were to be removed right which it looks It's like, has the PR been merged yet? It's being merged? No. No, no, no. There's no time. We haven't had the debate yet.

What I see on X, yeah, we have another debate. What I see on X is that it's already been done. It's like being pushed and forced on me. The debate happens at 5.30 Central tonight, so tune into the live stream. Live stream? Okay. But this is also another false narrative. I feel like the people who are trying to create a lot of drama because they were unhappy with

what they perceived to be the direction that the conversation was going in, they have injected this sense of urgency i have i've seen no sense of urgency from anyone who is like pro uh removing the limit um and and i would if anyone has a ton of time to burn go look at ava chow's five hour uh live stream from yesterday that goes through the entire decade-long history of this debate This is not a new debate. First, Peter Todd had this exact same PR like two years ago, and it got closed just because there wasn't enough interest. But it goes all the way back to 2014,

and the same people were actually in the pull request comments of off-return changes back in 2014. Okay. That's promising that we've had some devs around that long, right? I know it sounds like devs are dropping like flies, but it makes me feel good that we still have some that are willing to come back and comment like that. Does it actually feel like devs are dropping like flies?

That's just the narrative that I hear. It felt like that in the Craig Wright era when people were getting sued. That's actually why I started Bitcoin Plus Plus is because I felt like there was this narrative that nothing was happening and all the fun people were leaving. And I was like, I know lots of fun people

working on cool projects in Bitcoin. One of them was Casey Rotemore, interestingly, who was working on the inscription stuff at the time. Talk about a main character for a while, yeah. Yeah, so he was at the first Bitcoin Plus Plus, the very first ever like inscriptions workshop happened here in Austin like four years ago now which is kind of funny um and here we are like three years later and it's like talking about its impact on the mempool anyways but um that was the narrative for a while I feel like I mean it's

interesting to me that you still feel like that's something that you hear out in the ecosystem um I'm kind of wondering as like a developer community it's not what I've been seeing but you know what I see traveling around going to conferences is quite different than I think what gets represented especially in american media around bitcoin i feel at the very least uh a number of the more prominent devs have dropped out of the more public eye yeah which i can also understand uh you know wrench attack risk legal risk so on so forth yeah especially as that number keeps going up um and i've heard it on both sides of this argument where people say you know we need devs to be working on Bitcoin, and if we harass them over something small like this, what

incentive do they have to keep coming back and working on Bitcoin? So, Jameson and I are going to have a debate tomorrow night at 530 Central on the Bitcoin Plus Plus live stream, and we'll be talking about ossification, and I think this question about getting new devs in Bitcoin, really, you know, it's like, what are they supposed to be working on? Like, is Bitcoin a finished protocol? Like, a lot of the existing work, I mean, you could almost argue this is opportune stuff. One, has been fairly contentious for the amount of code that it's actually involving.

And two, like, it's really an optimization. Like, it's not really, like, a lot of the stuff that's going into Bitcoin Core, accepting a good number of projects that we had some core devs here this morning talking about, Gloria talking about Package Relay, InstaGibs or GrapeStanders talking about, I think he worked on Ephemeral Anchors. I don't know if that's what he talked about. And then Peter Willa talking about Cluster Mempool. those are solid engineering projects

that require kind of reworking the code base but a lot of the stuff that gets more contentious things is really like 12 lines of code like a few lines of a configuration well I mean look the block size wars are about one line of code and so it's like okay we're going to add more developers

like what are they working on like is it anyways I'm sure there's core devs out there who are like Lisa here's the entire road map of all the things we want to change and improve etc but But, yeah, anyways, so we'll be debating ossification tomorrow. Okay, two great debates, one tonight and then one tomorrow night.

Tomorrow, same time, yeah. Yeah, I also wonder, you bring up, like, cluster of mempools and stuff, but I wonder if the ones that get out and flare up on social media are the simple ones because they're easy for people to understand, right? And it's easy to generate talking points about it, right? You don't have to go in depth. Yeah, I think they do tend to be more taste-oriented because everyone has a taste.

To some extent, I think people are more likely to have an emotional feeling about shitcoinery, for example. Yeah. And, yeah, and that's what they want to talk about, is, like, why is shitcoining bad? And so, yeah, I don't know. No, but, I mean, it is rather amusing to me that a lot of these more taste-oriented things

are basically about, you know, how can we censor people on this censorship-resistant network? because we don't like what you're doing okay here's a spicy question um is is extending the block subsidy like peter todd has has famously proposed is that a question of taste so for those of you who are watching basically peter todd has proposed that we remove the 21 million coin limit and continue to basically have a perpetual issuance of bitcoin for every block so at some point it's supposed to go to zero this would change it so that it doesn't go to zero just every Bitcoin block forever would continue to produce more new Bitcoin that gets added to the supply.

Would it continue to have every four years and just always approach zero but never get to zero? No, it wouldn't approach zero. It would need to be a tail emission like Monero, I think, has tail emission. Let's say 500,000 sats a block. Yeah, so technically the inflation rate would still be trending towards zero forever if you had a flat tail emission. Right. But, I mean, I think it's more of an economic argument of what is really necessary for thermodynamic security. And, I mean, I have my own thoughts on that.

And I had a talk a few weeks ago where I gave a concept I called GoldieBlox, which is essentially looking at the block size limit not from a data throughput perspective, but from an economic perspective of what if instead of thinking of block size as how we constrain the system and how much of this common good there is to go around, but rather we look at it from what is the demand for block space? What are people willing to pay for block space? How is that trending? And should we adjust the supply of block space based upon the demand for block space,

similar to how we adjust the mining difficulty target based upon the current available hashrate. Oh, now you're talking about, like, didn't Ethereum do this? They have, like, a variable, like... They have a slight boost ability. I think you can basically double the amount of gas in a block depending upon, like, what the last block was. The demand curve, right, yeah.

But it's not highly variable. Got it. And Monero has a similar type of thing where you can go up to as much as double of the median block size over the past 100 blocks, but it becomes quadratically more expensive to do so. But I talk about both of those in my talk and why I think they're interesting

and maybe we can learn something from them, but they're not what I would consider an all-encompassing solution. Where can people find that talk that you put out? Is it on YouTube? It's on YouTube. It's on my website on lop.net. Cool.

and then you can all cancel me again yeah well i think we've had a good conversation do we want to keep going or do we need to get the next people on i can see who else is next what happened is y'all started we did oh okay got it yeah is bob i can see where bob's at was he supposed to be here in like 10 minutes from now or how early are we we're like 3 15 we got 15 early um yeah if gloria's around she can come hop on i love it so i scheduled for these conversations i scheduled two people to come hang out and sort of randomly like i was like me and lop should come because we'll be debating tomorrow so that's kind of i feel like a fun teaser for the event next

week but um i've gotten a lot of people that i just sort of randomly put together they're like or they're like, oh, I don't want to talk with this person. I'm like, no, no, that means you need to sit down and have a chat with her. Oh, man, you've got to get mechanic on here. I know, I'd love to get some of the ocean guys over here to chat. He won't talk to me at all. They're here. They're walking around.

Yeah, I think I see it. We're on the panel together, aren't we? So I know Luke Dash Jr. is like at 3.30 or something. Yeah, who's on the panel today? Not me. So Walton put this together. and I'm not sure. I think he's got seven or eight people.

We have just enough chairs. Ooh. So it's going to be... That's going to be rowdy. That's a big panel. Yeah, I think the... I was joking earlier, but maybe I shouldn't joke about it,

is you've got about five minutes worth of talking, maybe less, so, like, you should come up with your best insults to put out there, because any other points you try to make, you're probably not going to have time to come around. You're not putting Shinobi on stage, are you? I think Shinobi's going on stage.

Oh, boy, it's going to be all insults. When I say you, it's Walton, who, like, had full reign over this panel, so... optimized for maximum spiciness i think that's exactly what it was going for yeah and will you have mechanic up there too mechanic i believe is going on stage two yeah and i think like so for the debate debate tonight we're advertising it as two versus one my understanding is that they found a second person for peter todd side i think so mechanic and chris guida are going to be talking with i don't remember who peter todd and another person were arguing in favor of the

off return limit i think okay okay i think yeah well it should be really interesting i'm glad you're live streaming it because i know there's a lot of people who want to see that yeah i'm really excited about it um yeah so we're going live with that in a bit um is there anyone you'd want to talk to mark i mean there's just yeah i feel like i'm kind of LARPing here yeah okay yeah she walked by for like it went into the room okay all right so should i go track down some people you guys want to yeah yeah send some people over here i can just kind of filibuster for a bit on my own okay yeah sounds good talk about your project yeah sure you're coming to riga in august yes i am very excited to have uh marks at riga so bitcoin plus plus has another four

events planned for the i'm shilling now but um we got another four events planned this year this is our our yearly austin event but we'll be in riga uh right ahead of baltic honey badger in august we'll be there the 7th and 8th marks is um graciously uh decided to come out and talk about his project open secrets um is that the name you guys use open secret yeah and then maple ai is like our main product right now we're trying to bring privacy and ai together and um it's really where a lot of people are giving away their data to these large corporations and don't even think about what that means right yeah and then they enable memory features chat gpt and grok so yeah now we're gonna start remembering everything and they think oh this is great for

me but really it's kind of the opposite right like i don't want to be giving that all over and having them remember that a friend was telling me a really interesting who's more in the ai industry is telling me an interesting story and i'm gonna get the characters wrong but i think it was something like one of the ai teams let's say anthropic for example i might be wrong about that was using notion as their note-taking thing and they turned on the ai features and then found out as their internal like company stuff and the ai features for notion were supplied by none other than one of their competitors which is oops open ai um so i don't know i think they ended up changing platforms or turning off the features and I might have the characters wrong but it was basically

that configuration where they were using some tools as a management software that then was sending all of their data to their competitors about what they were doing and planning question exactly I think what you guys are working on is how much of that was open AI reading or having access to or maybe you know they don't even realize it gets ingested into the um the model and then someone anyone can ask oh what is you know what are anthropics plans for whatever and the models like oh I happen to have like information about that that I can provide you I don't know what kind of filtering they're doing internally with them but one way I think about LLMs is really big they're really big search

engines right yeah all it is is a search engine that's returning the format that you get the information back is different than kind of I would say v1 of search engines which were links so it used to be typing keywords and it provides a series of links for their websites that you then have to go and figure out where the information you were looking for is embedded inside that link page. LLMs are slightly different in that what they're doing

is they've gone through and scanned all the web pages and the result that you get back is basically a generated essay based on the questions that you've asked it. So it's still like a search engine and it still has the same ingestion process as any search engine. So anytime that you,

And one of the things about LLMs, sorry, I'm like on a tangent now, but one of the things about LLMs is when you're asking questions to it, that in turn can become part of what it then uses to look through, right? So if you're providing data or context to the LLM, there's always a possibility that that might get ingested

into the search engine that then becomes available for other people to search through, right? And so one of the amazing things about Mark's project with the TryMaple, is that .ai? Yeah, TryMaple.ai, yeah. Is that you guys are using some state-of-the-art encryption such that when people are talking to the AI, right, it doesn't respond back with data that anyone else can read. It's private to you, right? Yeah, so we call it self-custody of your data where we generate a private key for you using a secure enclave.

And then we publish our server code open source, and people can do a verifiable build of that code. And then they can run it against the code that's going in the secure enclave. So they can know that we don't have any backdoors in there. Cool. Once that process is established, now everything is, yeah, it's encrypted end-to-end. So your chat on your device, it's encrypted locally and then sent off to the GPU, and the GPU is also running a secure enclave. So, yeah, at that point, we aren't recording your data.

We're not training on your data. We can't read what you're chatting about. Yeah, which is cool, and it's quite different than my understanding of how the other ones work. No, I mean, I see LLMs as, instead of a search engine for websites, It's a search engine of human knowledge, and it just happens to use websites to back that up. It's more like an interactive Wikipedia in a way, sort of. Yeah, except it can auto-gen itself.

Sometimes it makes sense, sometimes it doesn't. Anyways, that's a whole separate thing. Great, well, thanks for chatting. Yeah, thanks for hanging out. Also, check out the hats if you're on the live stream. We've got two great hats. We've got some Knott's hats and some Bitcoin Core hats show your support for your favorite node

at shop.btcpp.dev yeah no floresta no but I can add it we can add that as an option are you running floresta no but I basically did a review of all of the

bitcoin client implementations recently because really a lot of them don't have much if any contribution or maintenance most of them are dead interesting floresta is number if this is very rough but like from looking at developer activity it's bitcoin core and floresta and like btcd do you include rust bitcoin in your account yep okay and it's not interesting because okay yeah it's interesting there were a bunch of block

stream developers a few years ago that were very active on rust bitcoin but maybe not as much recently priorities change maybe that's what we need more developers in bitcoin for is working on alternate clients yeah there we go it seems like a win for everybody yeah it's true when bitcoin plus plus like uh what a client edition or something interesting stuff in cool well thank you mark thanks thanks for coming on all right we'll see you later a home miner that i run to heat my room in the winter uh mostly my office and i love it because my house is on propane propane is really expensive and so um but i actually have solar panels on the roof so it's like it's awesome during the winter i can run this little space heater in my room and it's just an s9

in there right yeah um but i got really excited about bidax when that launched um the thing that has been confusing to me even though i've been following this space and stuff and i'm sure there are people who haven't been following it that are reluctant to is like bidax is an open source project you got soul satoshi you have a bunch of different people so like help us understand what how to make sense of all of it like we're so used to having like the trusted source where we go buy something from them and it shows up right like this is a little more distributed and diversified so maybe kind of walk through yeah that landscape yeah so there definitely is quite a bit of confusion on that a lot of people think you know we invented bidx and this

is our project which isn't true we love the project and we advocate for it but there's several manufacturers in the space started with one that we're aware of and that's been a huge part of it with the OSMU group and so we started with them and basically it starts there and then just trickles out so every device that's sold or even bought has a five dollar donation back to the group to further you know just give back to people go full-time and just do developing and furthering the project more and more so that's where it all starts and then now we have bidax.org and all of the verified distributors so there's manufacturers and

you just really have to do your research that your seller is getting it from reliable manufacturers and just always give back is the is the biggest part okay so how do we know to trust the website that has the verified manufacturers on it so that is bidox.org that's their original or not original because it has been updated but that is verified by scott he puts everything on there and he controls pretty much all of it okay so we trust scott i'm i'm in the trust game right so like i'm all about how do we know like how do we verify everything don't trust verify kind of stuff so so um you know how do we know scott is being genuine and what he's putting on there and Is that just, we just kind of watch the market for a while and make sure that everything's

running well? Like, what are your thoughts there? That's a very good question, and I think that actually was brought up today on Twitter, and I'm probably not the best to answer in this scenario and get all my facts right, but that is the most important thing is that you verify yourself, and I know there's a lot of things in the works to make everything much more transparent as far as the project itself and then from start to finish and I know it's being worked on but the hard part is trust and I'm probably not the best one to answer right now you're good and what is your role at

Sola Satoshi then so I'm a owner it's just three people right now it's just me everyone knows Matt I'm pretty much you know behind the scenes kind of person and it's me and my wife for the other two so pretty small team just trying to scream from the rooftops okay and are you the one like filling the orders are you yeah because you know it's me and my wife so you probably mailed some stuff to me i've got a couple bit axes so yeah i probably can't remember it's been a couple thousand you've done a lot yeah yeah that's great no that's really cool um and so where where do you see this going like where do you where do you see bid axe heading what does the roadmap look like and we're just for a decentralized open source mining it's really endless um you're going to

be seeing a lot more different projects if you will that's a fork of the original bitax you know the gamma super ultras that we all started with and we're going to be integrating into more multi chip hardwares module and try to increase that tera hash while decreasing the price for it as well that's the biggest thing with these single chip miners so i think it's just going to get cheaper more reliable easier access to get stuff and just going to keep spreading and then And how does, are there commercial actors that come in and potentially partner up with Solo Satoshi or with other manufacturers to build products? Like, let's say in my house, I've got a central heater in my house that's propane. I would love to switch that over to Bitcoin miners doing electric mining if I could.

Are those discussions happening or how do you see that potential opportunity? Yeah, there's actually a lot of talk, especially in the heat pump space. And I think we're really going to start seeing a lot more integration of that. There is already talks about it. The key is just getting an open source project into a closed source type situation in your house, which is absolutely doable. We do that in software all the time, right? So it's cool to see some of that transferring over to hardware. Yeah, I think a recent one I've seen was someone actually had bid axes connected to their water pipes.

Like, the actual ASIC was wrapped around the copper pipe so that the water was keeping it cool. So, there's definitely going to be uses coming out, and I really just can't wait to see it in every fridge, dishwasher, every appliance. I mean, a computer can run it. So, I think the possibilities are endless, and the roads are going to be really awesome to ride. Okay. And what are you personally looking to do to grow Solo Satoshi? So me and Matt still have full-time jobs. My wife is the only one that's full-time currently.

And so this year we're hoping to go full-time so that we can just put everything we've got into it even more because there's just not enough time in the day. So it's all about creating those relationships and growing and really just speaking for the people in the project and just keep growing because it's just a full-circle deal. you sell more you find more blocks people buy more and then every sale there's more donations going back to the community to keep growing it yeah okay there's a there's a bit of a meme going on with marty he's not here right now but i don't know if you watch rabbit hole recap or anything

but uh marty show you're talking about um quantum physics and if you look at your bid axe then it's going to mine blocks because it's you're like helping out by staring at it so so we start our bitaxes now yeah trying trying to influence it to actually mine a block so yeah well yeah we should do like a live stream everyone come watch yeah we're gonna stare at their bitaxes together have a coordinated thing just post it out and be like all right this day at this time we're all going to look at our bitaxes yeah get them to mine something yeah no that's great so um yeah i mean what do you want people to know about solo satoshi and what what can we do to kind of help man um You can always come purchase on our website, solosatoshi.com.

We're open to collaborations. We're trying to work with as many people as possible. So we do apologize if there's delay in response on some of those things. But that's what it's all about is just growing our network and trying to spread the word and get all these different projects and, you know, get them fuel so that they can get the capital to really push them to market and go into mass production type thing. And so that's really what we take pride in is just trying to help be the voice for people and spread the word. OK, on the software side, we have a lot of software developers that watch the stream right now. Is there any like software things that could be built? Oh, God. Firmware like what what kind of stuff would you like to see come into existence with people who are either really seasoned developers or people who are using AI to vibe code?

What would be interesting to you? I would definitely recommend everyone to check out the OSMU group, the Open Source Miners United. They have a Discord, they have GitHub, and they can always use help. They can always use developers on every project that's on there. Definitely recommend taking a look at all of them. But they could always use help, and anything is definitely appreciated. So there's nothing I would say in particular because, I mean, it all needs help, right? Everything needs fuel, but the best place to start would definitely be with the OSMU group.

Yeah. What's the website for that? Just osmu.org? Actually, I do believe that's actually pulling up a blank for me right now because I'm always just on the Discord. All right, yeah. So do whatever you do to search the Internet or use AI, Perplexity, whatever it is. GitHub, it is all posted on GitHub, all the different projects. It's just easier, at least for me, to track, you know, what's going on in the individual projects as far as people running into issues and just conversations.

So it's really cool to see. Cool. Right on. And if people want to follow you online, what's your Twitter handle, X handle? Solosatoshi is our main Twitter page. And then my name, Hunter Solosatoshi, and then Matt, he's on there as well, and even Kelsey. So that's the whole team. That's it.

Great. Small businesses. A lot of small businesses around here. Oh, yeah. that's great awesome anything else you want to share i'm good okay thanks for coming on appreciate it you did great yep i appreciate it's the first time for everything okay that's right well there's gonna be many more in your future yeah that's right cool yeah thanks all right yeah so um we got our next guest here hey how you doing come on in yeah i'm marks

grab a microphone here pick up pick up whatever you want okay welcome are we live we are live we're live we're streaming on x we're streaming on youtube okay and it's probably being recorded too i assume but um cool i don't know you at all you probably don't know me at all so uh i was on marty ben's bobcats the tfc is that you okay no so it is marty but he had to go to another event briefly he had to go be on a panel so i stepped in for him for now gotcha yeah so i was on the podcast a while back talking about braid pool so braid pool is a decentralized mining pool project i'm working on um basically they're they're kind of three decentralized mining pool or four decentralized mining pool projects here at this conference uh we just heard talks from

colpreet talking about p2 bull v2 and hydropool uh we just had a talk by the ocean guys talking about the datum protocol there's a uh cashew e-cash mining pool and i'm the fourth uh braid pool which uses a new direct basically graph based consensus mechanism and what I'm probably the most interesting thing for this conference is the idea of a committed mem pool and what that is is you know the talk for around the around the fireplace has been make your own blocks I I personally think this is kind of a bad idea making blocks and deciding transactions is a shitty problem and nobody wants to do it it is a legal liability it's a risk in every

jurisdiction, somebody's going to come to you and say, don't mind this transaction, whether it's a political enemy or crime or, you know, donations to Canadian truckers. And, you know, the finance rules that exist are based around banks and banks know who their customers are. Right. They are capable of blocking transactions and they're capable of seizing funds. Bitcoin miners can't do any of those things. Right. So it's totally not appropriate to put those rules on Bitcoin miners nonetheless because of the history people will try people are trying but we need to get people out of that game and so the problem is that nobody wants to make block templates and in ethereum there are two kinds of block templates right there's compliant block block templates

and there's a mev meval block templates right one is fraud and the other is completely destroying the idea of a censorship resistant network so we don't want either of those to come to bitcoin And when you have a problem like this where no one wants to do a particular thing, the solution is not to get, you know, to try and find somebody to do it. The problem is to make everybody do it. And so what we're going to do instead is have this idea of a committed mempool. So the way it works is every share that is submitted to the pool, and this is a decentralized pool, so it's a blockchain just like Bitcoin. There's no central server. There's no central custody.

But every time somebody submits a share, they can also tack on a couple of Bitcoin transactions. And then what you do is you look through the history and you add up all the Bitcoin transactions in that history. These are now all committed with proof of work, right? And then taking those, that set, I construct a block out of it using any deterministic algorithm. We're just going to use Bitcoin D for that. And not only have I taken the responsibility for deciding the block out of your hands, um i all this also has a huge advantage for a decentralized mining pool in that it can be independently computed by all nodes so i don't even have to tell you what my block template is

because you already know okay um which means that's that that means that the shares can be very small i don't have to tell you i just have to tell you what my parents are in the share chain uh the parent beads or blocks in the share chain um and then it can be independently reconstructed so naive question like a share is that a uh hash i'm basically it's a proof of work it's called a weak block. So it is in every way the same as a Bitcoin block. It just doesn't meet the proof of work target. So in order to validate a share, I really need the entire block. I need to know all the transactions in it. I need to know the Coinbase, and I need to do the proof of work, right? So several of these projects like Hydropool, P2Pool V2, and Ocean are only

validating the Coinbase, and they're not validating the set of transactions. And there's a lot of weird things you can do if you have control over the set of transactions. You can construct blocks that have all kinds of nasty stuff in it that is not allowed by the peer-to-peer relay rules. Quadratic hashing, and there's a whole list of things that vulnerabilities we've solved by making sure the peer-to-peer network doesn't relay these bad transactions. But if I don't know what you're mining until I see your block, it'd be very easy for someone to be mining that. Or let's say one guy's mining empty blocks, right? And I'm mining a profit maximizing block. You know, there's a social contract there that if we're both operating the same pool, we're both

of doing the same thing either we're you know profit maximizing or not and maybe a third guy is you know mining monkey jpegs and taking fees out of band for this right these three guys are not the same and it's not appropriate for them to be agreeing to pay each other the same amount for the same amount of work because they're earning different amounts and their network problems too right like there are various reasons that a block might not get published uh once it's found i mean let's say one guy's mining with a um you know a bed axe on a raspberry Pi. And he's capable of sending these little tiny shares really easily. But now he's got to make a four megabyte block. He got lucky and just found a block. Now this Raspberry Pi, which is on a

dial-up line, has to upload this four megabytes of stuff. A, it's going to take him a lot longer to create it and validate it. But now he has to upload it. And it takes a long time to get to everybody. This guy has a wildly different orphan rate than everybody else. While he's busy doing this um somebody else might mind a block right and then all of that work he did uh isn't worth as much as the guy who has a well-connected node right yeah so the miners are an equivalent here so so this this idea of a committed committed mempool makes them equivalent um so a they're all mining on exactly the same block template it's not exactly the same but i know what it is and we've all agreed to some social contract that says here are the rules by which we make the block template

We're going to pick the highest fee transactions first. We'll use the cluster mempool work that Peter Willa talked about earlier today to make sure we're getting the highest fee rate and the most profit possible. Second of all, the actual amount of communication to the miner is very small. It's only a couple kilobytes every time he sends or receives a share. So this keeps the network bandwidth low. And when a block is found, it's not the responsibility of that miner to broadcast it. So in both P2Pv2 and Ocean, when a block is found, the miner who found it has to broadcast it.

He may be capable, he may be not. In Ocean, if they've spot-checked that particular block, the pool could broadcast it as well. So that's slightly better. Now two entities are responsible for broadcasting it. But in Braidpool, every single node will be capable of broadcasting it. Because it will be computed. Once I receive your share, I'll compute it and broadcast it myself. Okay, and you said earlier that when you put a share into it,

a couple transactions are, like, attached to it. Yeah. How do they decide which transactions are attached to that? That is a nasty problem I am not looking forward to having to solve. Okay, yeah. Because does that turn out to, like, every person that wants to spend Bitcoin now has to, like, contribute shares? Well, there are already – no, no, no, no, no, not at all.

It works exactly – so there is a standard mempool. So the two transactions that are attached to a share will be taken from Bitcoin's standard mempool. Okay. So, yeah, they have to get those transactions. somewhere but it doesn't have to come from there it could come from anywhere um but once we've committed to them uh it will be mined by some future miner within the pool okay um and yeah there there's a there can be a different set of rules around what those transactions are

um i don't want to be as opinionated about this as the ocean guys are um filters don't work uh monkey jpegs are bad uh but at the same time like we're probably gonna just gonna go with what the what the peer-to-peer relay rules are on Bitcoin. You know, if Bitcoin will accept this into its mempool, we'll take it. Yeah. So what do you consider the standard mempool to be then? Because, like, there's a lot of different mempools around. Well, it's the one that Bitcoin relays.

Okay. And, you know, there's going to be a lot more conversation at this conference about expanding that. We'll go with whatever the consensus is. I don't think it makes any sense to substantially modify that. Although, Braidpool is going to be open source software. So if you want to fork it and mine monkey JPEGs amongst you and a bunch of your friends, you can do so. I don't like it, but I can't stop you.

And where are you at in the progress of Braidpool? It's still fairly early days. Most of the work I've done so far has been like statistics and analysis and the consensus algorithm and trying to get all that right, the payout mechanism. I have just engaged the Summer of Bitcoin project. So I got a grant from Spiral to work on this at the end of last year. This is a very old idea, by the way. I first gave a talk about this in 2015 at Scaling Bitcoin Hong Kong, and it's been a backburner project for a long time.

But as of the end of last year, I am full time on it. I engaged the Summer of Bitcoin project, which is funded by Brink, I believe, and they give a stipend to college students to work for the summer. I have selected seven students from the Summer of Bitcoin project to work on this for the summer. So I might be crazy. I got to herd these cats. But hopefully by the end of the summer, we will have a working Node software.

Those seven are divided. Four are working on the basic Node software and the committed mempool, and three are working on a mining dashboard to make it the most awesomest and everyest possible way we can. Okay. And just a side question, are you using AI to help you write your code? Are they going to use AI to help them write? Oh, that's a whole question.

If you're getting college students, they're probably all vibe coding now. So I have a fourth project here. I've been using AI. I haven't sat down and written a ton of code in a while. I'm a developer by training, but it's been a bit for me, and I decided to get in touch with these new tools and see what they can do. So I started fooling around with it myself. I bought a couple of GPUs.

I've been running all the local models. I've been using AIDR and Lama.Vim, which is a fill-in-the-middle completion AI tool, which is actually really cool. So I got to talking about this with the students, And people were more interested in that than all the Bitcoin stuff I proposed. So I threw out a proposal to build an AI-assisted coding tool for Bitcoin projects. Okay. And I got a ton of applications from the Summer of Bitcoin.

Yeah. And so in the process, we have partnered with a number of other organizations. So the Bitcoin Dev Project has joined forces with us, as has the Python BTC Utils Project. So we now have three sponsor organizations. The Bitcoin Dev Project is responsible for the ChatBTC website. So if you Google ChatBTC, you can go talk to an AI about Bitcoin. That project is about two years old, and the architecture behind it is dated. So we're going to update that using what's called RAG,

Retrieval Automated Generated Generation. So basically you feed a bunch of context to the AI so it doesn't hallucinate. And as many people who are listening to this probably know, if you go ask any Bitcoin question of any AI, they answer terribly. It's going to be wrong. It seems they are only trained on Twitter and nothing else. So we're going to augment that by all of the Bitcoin dev mailing lists, all of Stack Exchange, all of Bitcoin Talk, all of that.

So that's already done. The ChatBTC guys have that. The Bitcoin dev project have that. So we're going to expand upon that, improve upon it, and release a new version of ChatBTC. See, at the same time, we're going to start adding code to that. So we're going to ingest every single code base for every single Bitcoin project we can find and put that into this retrieval system so that when you ask a question about Segwit,

it says, hey, I know seven different implementations of Segwit in different languages. And when you're writing a wallet, it will be able to critique that based upon that context of seven different other projects which already implemented Segwit. So this ultimately is going to be an AI coding tool for Bitcoin. That is also going to come out of this as a side project. Okay. Do you think there needs to be any waiting aside? So if you have seven different segwit implementations, but one is more widely used and viewed as to be like better What should that factor in into how you're? Putting in the rag you're not training the model, but you're using rag to help

Yeah, we decided not to do training because it's very computationally sensitive and we don't have a ton of resources, right? We're not Nvidia or chat So, yeah, we're essentially going to use a RAG for this and local models. We're focusing on local models, right, rather than the big ones so that you can run it yourself. And it's a good question. I don't know how I would rank implementations. Right. It's, you know, I've thrown out the suggestion of doing multiple rounds with the AIs here.

So, given these four different sources, rank them in terms of relevance or rank them in terms of quality or things like that. So imagine you have a coding assistant that's always running in the background as you're writing your code, right? So right now I'm using fill in the middle completion. What it does is as I'm typing, it tries to guess what the rest of the line is. And I found that to be actually quite good. The llama.vim that I'm using essentially keeps a running context. So not only does it add your current file, but it may add things you've cut and pasted, may add other files you've looked at. And we can augment that by all kinds of things.

But that context, the KV cache, the context window, is essentially already populated. So I'm not giving it a whole bunch of data, and then it has to ingest it, and I have to wait for a response. It's just sitting there waiting for me to give it a query, right? Or waiting for me to fill in a line. And, you know, it just grows over time and gets more and more knowledge about your project.

And it's a good question. I am myself confused as to how I'm going to use this or what we're going to do with it. Yeah. So we do a project we have called Maple AI. Try Maple.AI. We're trying to do local privacy, but we do it in the cloud. We use secure enclaves. And so you can chat with a chat GPT type interface. So you have like an H100 running in the cloud rather than your local GPUs, right? But we're trying to figure out how to do RAG. And one of the problems is it's really easy to influence AI, right? You can push it around and like any context you give it, it now weights that as like something that's super important. And you're like, I just told you this one thing. I've seen the opposite. I've given it context and had to completely ignore it. Interesting. Yeah, see, it's kind of a hit or miss then maybe,

but like we're seeing, yeah, you give it a little bit and suddenly it latches onto that detail and like wants to bring it up in every conversation. It depends on which AI. So, you know, I mean, there's two of the things we're going to do here is take a look at the local AIs where we can run, right? So, QWEN 3 was just released.

5.4 Reasoning was just released. We might have a new DeepSeek soon. They're talking about 2.5. Yeah, I mean, like they're getting faster and they're getting smaller at an alarming rate. So, my thesis here is that you're going to be able to run 03 on a local machine by the end of the year and arguably almost already can right so they're getting better at understanding context but i've given them context they've completely ignored

before okay um so you know it's not a foreground conclusion that if i feed it the right answer it will give it back to me right especially if i give it a lot of context so one of our students wants to look at long context models so like a million token uh models yeah if i give it lots in lots of contexts, what can I do with that? Well, like the new Lama 4, I think it's 10 million or something is what they're claiming. Yeah. So if we go ahead and fill that with every piece of information we can find, is that good or bad?

You know, can it still find the answer? It's a needle in the haystack problem, right? So yeah, we'll see. Another of the students has scraped StackExchange, the Bitcoin StackExchange, for question-answer pairs. And so one of the things I've told them is that we need an evaluation benchmark. We need to know whether we can answer questions correctly in the first place, whether the model can retrieve the context we gave it, and whether it can reason about it, or maybe it can rank context, or maybe we have a two-step process where one AI looks at the context and tries to decide what's useful, and a second AI answers the question. So, yeah, all these will be part of that project.

We're going to do a lot of experimentation and exploration. And hopefully by the end of the summer, we will have essentially the product of this hopefully will be a RAG that you can download. So this would be a multi-gigabyte database you can download, plug it into your local AI, and plug it into your coding setup. That would be awesome. Maybe you can use something like AIDR to generate code, but these smaller models are not that awesome, but they're getting better. But yeah, I mean, if you're using AIDR and you say, write a function for me that does this, and it knows you're using Rust Bitcoin, and it knows what the APIs are in Rust Bitcoin, knows what those functions are,

we can prevent it from hallucinating parameters and hallucinating command line arguments and all kinds of things. But will it be useful for generating code? Don't know. Like I said, I'm using fill in the middle right now. Pretty impressed with it. I want to take it to the next level.

And the thing that I see that's missing from things like Cursor and Windsurf is if you had an AI assistant that was always running, sitting there watching what you do and could run queries in the background in between you typing, right? So, on the one hand, you can say, I'm just going to send a prompt up to OpenAI and get an answer. Okay, it takes a while to get a response.

But if it's already got everything in its context, you know, it's slowly like filling up this context. It's like, oh, you called this function. Well, let me look up the, you know, in the background while you're working on something. It goes out, finds the documentation for that function, finds the library downloads the source code for that library generates an abstract syntax tree for the entire library and it has all that as context now as you're writing code further it

it can do a number of interesting things that you might not even know how to ask if you're asking an open ai but it you know it might tell you like as you're writing it might tell you oh um you know this parameter you're passing needs to be borrowed uh or you know know etc you know who knows what what things but it it could probably determine that in the background and then give you that feedback like as you're coding yeah i think this is interesting because one of the complaints i hear a lot was one we need more developers in bitcoin but two bitcoin is so hard to understand that no developer understands the entire project and so i see ai you're talking about maybe it's not going to be writing bitcoin code very well but helping us

understand the code yeah and of all the projects but also like core and other things i can see that being very valuable yeah absolutely a good contribution i mean like i always have 17 tabs open you know i know i'm using this library and i know i need to call this function okay how do i call this function what do i need to pass it um you know your ai in the background could be figuring all that out for you right so rather than having a browser tab open it could say oh the first argument's a string second argument's a pointer to uh to avoid or whatever you know and it knows all that right as you're typing it um you know it can suggest oh i think you're trying to pass this object in there right there there are other tools there's one I use

called you complete me and it is related to tree sitter if you know what that is this creates an abstract syntax tree out of your code so it knows what the local variables are it knows that you're writing a function in this class it knows what what's in scope and will suggest to you things that are in scope like if you have you start typing the name of a variable it'll suggest the name of that variable the fill in the middle completion models do the same thing this is kind of a more precise way of doing that and I want to make it even more precise so that it knows not only does it is able to kind of guess the variables I'm using but it also knows the syntax and some structure around it and can kind of

take it to the next level okay cool I'm still trying to figure what that looks like to be honest but just imagine there's a coding assistant running on your GPU sitting in the background watching everything you do has access to all these resources um and can suggest things write code um and fill things in for you yeah is a laptop gpu gpu going to be powerful enough for this kind of stuff it really depends on the models eventually um yeah i mean i i have a my laptop died recently i had to get a new one but i had a um amd 6800 with 12 gigs on my last laptop um i ran a surprising number of models on that they were quite good these are small ones like 3bs or something you're not running

Yeah, well, up to 8B or so, but these are quantized, too, and quantized introduces down syndrome in these things. But, you know, we, I don't know. Again, they're getting smaller every day. I bought a 24 gig 7900 XTX for this purpose, so I can run much larger models now. But, you know, the thesis is that they're going to keep getting smaller. And the hardware requirements, if we evaluated it today, will be wildly different six months from now. So we'll build a tool now that will run on a 24-gig card today, and maybe it'll run on a 12-gig card in six months. Maybe run on a 4-gig card six months after that.

That's kind of the trajectory I see. Yeah. What is that? Is that Moore's Law or Murphy's Law? Moore's Law, right? No, it's... I'm trying to remember. What's the one where like...

Yeah, Moore's Law. Yeah, yeah. Moore's Law. No, it's not what's going on. what's going on is that, you know, the OpenAI guys said, this takes a lot of compute. So we're going to corner the market on compute. We're going to buy every GPU we can and price everyone out of the market such that a single, you know, B2200 processor from NVIDIA costs $50,000. It should cost $50,000. It costs $50,000 because a bunch of VCs gave a bunch of idiot money to OpenAI and

OpenAI says, we're going to corner the market. Meanwhile, the Chinese said, screw that. There's a lot of efficiency gains we can find, right? They came up, you know, DeepSeq came up with reinforcement learning. You know, there's a lot more kind of open source models and they're all kind of improving each other's models, going back and forth. DeepSeq added reasoning. And this has been improved upon in the last couple of months in a lot of different ways. So yeah, I think there are a lot more efficiency gains to be had as well. This isn't the end. We're going to see a lot more efficient models okay great well we got on a big tangent there sorry yeah no you're good no i love talking to ai all day um any final words you want to give about braid pool or uh your summer bitcoin

stuff you have coming up or what um it's gonna be chaos on my discord if anybody wants to join come join the chaos it'll be fun i gotta hurt a lot of cats it's open source development you know trying to teach these kids open source development it's not like i gave you a project plan you got to go execute it no it's like find that it's you want to scratch and if you want to scratch and show up and write some code for us. My grant sponsor, Spiral, has expressed interest in hiring another full-time developer on this. So if you really find this compelling and want to join full-time,

there is a possibility to get another person to grant on this project. Okay. Best way, if someone wants to get that, do they just start contributing and kind of do proof of work that way? Yes. Or what should they do? That's always the way in this ecosystem and in open source in general. And yeah, reach out to me, of course. Let me know that you have that in mind.

Great. Well, hey, it was great chatting with you. appreciate it thank you much okay we'll see you all right i think that might be the end for today right is this the sign-off moment do i need to like sing the star-spangled banner or something that's how they used to sign off tv peace and love