A (5:16)

Max.

B (5:18)

Hell yeah. Welcome. Welcome to Bitcoin Audible. This is. This is official start of our show. And you know, um, I. I want to share with everybody the last time we got together, I believe we went for like what, three hours, right?

A (5:33)

Yeah, that's the minimum nowadays, right?

B (5:35)

That's the minimum. And which. I don't know. I don't know if you have three hours today. I don't know if I can. I don't know if I can pull that off more. I gotta.

A (5:43)

More. It's usually the host that gives up earlier.

B (5:47)

Yeah, yeah, yeah.

A (5:49)

And.

B (5:51)

And we lost it. We lost the whole conversation. You Remember that it didn't record properly.

A (5:57)

Yeah, that was.

B (5:58)

That was such a sad moment. That was such a sad moment. And because of that, you know, I've probably built up in my head how great. I remember it being a great conversation, but I bet it's twice as great in my head since we lost it.

A (6:11)

Exactly. Yes.

B (6:14)

It's the golden egg that will never get back.

A (6:19)

Yeah.

B (6:19)

Welcome.

A (6:19)

You know, it's like we used to have conversations just one on one and nobody cared.

B (6:23)

Yeah.

A (6:24)

Just now, weirdly, we can broadcast it to everyone. And then if. If that fails, we're like, oh, no, what are we going to do now? Turns out we just had a damn good conversation among our. Among us.

B (6:33)

That's.

A (6:34)

That's totally fine.

B (6:34)

Yeah, dude. Welcome to the show, man. Yeah, been loving it. Yeah. So you've been working on a lot of stuff. And I want to go ahead and hit just. Just to kick this off because Johnny had a couple of questions for me. We're gonna be talking about privacy. Definitely gonna be talking about AI. Definitely gonna be talking about a number of various things we'll be building. I know you're building a couple things. You just mentioned white noise just before we officially started here and Noster, because there's tons of really cool noster applications. Um, but to start off with. So for anybody who doesn't know you are major in. You've been building and working with wasabi for a very long time. And I have really loved wasabi. In fact, being on a bitcoin standard, I literally don't do anything like with pay. Like. Like I pay a lot of people and then receive everything in bitcoin too. And I do not everything goes through wasabi because I don't want. I don't want my balances. I don't want exchanges or services that I'm using. I don't want there to be an obvious. I'm gonna go check the mempool and see where this batch came from, you know, where this, this transaction came from. And you know, I don't. I don't use it like. Like, I'm not like super crazy hardcore about it. I just want like a simple break the heuristic, you know, And I find it invaluable. And we are in a very interesting place, especially with like the samurai guys that people are. Governments and regulators are treating this like this is illegal activity. Like, this should be something that needs to be squashed for literally the most obvious and basic privacy that I could provide myself while doing business. And I'm just curious before we kind of get into specifics, a, you know, talk about like your position in Wasabi and like your history with working on it and, and also kind of like the philosophy and how do you see like, are you nervous? Like, like with working with Wasabi and on privacy software, like how do you think about it these days?

A (9:15)

I think that we really have a lot of available good tools that help to protect us.

B (9:22)

Right?

A (9:22)

Like that's ultimately what privacy is about. You want to protect yourself, you want to prevent that others can steal from you. And being private is extremely powerful, is very cheap to use and quite expensive to attack. And then the question is just, can we further extend our toolkit here? Can we keep building that out, making it cheaper, easier to use, faster and so on? And I think we've seen insane engineering improvements over the last five years, 10 years, and that doesn't really seem to stop. So sure, there are people who would prefer you don't protect yourself against the thief. Specifically the thief likes that. Turns out. But that's where the cyberpunks come in, right? We're punks. We write the code, we ship it. And sure it's a dangerous business to help people out of slavery, but so what? I mean, we have to do this, this is a moral obligation, ethical obligation, as a technical possibility and a lot of fun. So the risks of it are real and substantial and have been forever. Especially in the last 30, 40 years or so, countless builders have gotten into serious trouble or even been kidnapped or killed. And yeah, that's a hard fact of reality that we have to deal with. But I'm way more scared of the alternative on not working on this code and seeing us lose the grip on freedom that, that we have rather than increasing it and making it flourish. So for me, it's, it's, it's really a no brainer like this. They're so angry about this technology precisely because it works real damn good and that's why we have to double down.

B (11:08)

Yeah, yeah, that's a good point. That's a good point. They're, the reason they're attacking it is because it works. Um, and I would love to see a future where, what's funny, with some of the compression tools and the way things are designed in Bitcoin, there is a, there's a very real potential, especially when fees get really, really high, that a, that kind of a coin join, that, that privacy tools are actually more efficient economically. They're, they're actually better data compression on the actual chain itself, which is really, really exciting. I've been waiting to kind of see that incentive kind of play itself out. But what's funny is that we haven't had. We've been no fees on chain. So for a long time, fees have been incredibly low. I did not think we would get a 20, 26, and I'd still be paying one SAP per V byte every once in a while.

A (12:11)

Less than that.

B (12:12)

Less than that. Yeah, yeah, yeah.

A (12:15)

But, you know, block space efficiency is super important because privacy is about hiding in the crowd, right? So the. The larger the crowd, the more people use it, the more people look indistinguishable from each other, the higher the quantifiable privacy effects that we get. And if it costs too much, then we cannot scale as much. And this was one of our core realizations of Wasabi from the very beginning, that you could always make private blockchain transactions by simply making a network of transactions where you send money to yourself in different amounts with different inputs, different change amounts, et cetera, and you randomize a whole bunch of them. You can get a huge amount of ambiguity while doing that. Not a huge amount, a reasonably private amount of doing that that you have plausible deniability of. No, I just got that money from someone, you know, many hops before that. But then that's still extremely block space inefficient, right? You need to have multiple transactions which have also that aspect of whenever on transactions you create an output, that means you later have to spend it. So a coin always has two costs, the creation and the spending. And so if you could imagine that you have to make 10 transactions to yourself before you can make a private payment, then you need to create and destroy 10 coins. That's crazy inefficient. And so the idea of a coinjoin is how can we compact this so that instead of everyone doing 10 transactions by themselves, we make one transaction together, where we increase the size of the crowd by introducing other people into the crowd rather than making fake transactions or transactions to yourself, I should say. And with this, we decrease the size of block space that you need to use in order to be able to use Bitcoin privately. And then ever since the first launch of Wasabi, we further iterated on improving this substantially. So Wasabi 1.1 allowed you to create multiple denominations in the same round. And that was already a big improvement. But then especially with Wabi Sabi, that was another like 50x of block space efficiency and coming to a point where it's so cheap block space wise to use a coinjoin that we just turned it on by default that you receive money into your wallet, it automatically coinjoins in the background even if the UI is closed. And the next time you open it, you have 100% privacy and you can just make your payments. And also then what even further improves this is if you start making a payment inside the CoinJoin directly. And then this is extremely powerful, especially if you're making many payouts to your employees, because then you can batch five payments and put all of these five payments outputs in the same CoinJoin transaction. And none of them know especially what your inputs are and which is what you actually care about. But neither do they know the other outputs that other employees of the company have received at that time. They're both extremely important privacy aspects. And you get that with a single transaction. So now you can get a reasonably minimum privacy amount just by taking a coin that you received from someone else, putting it on the input side of the transaction and on the output side, register the five payments outputs that you want to do, and your change output gets broken down into three additional UTXOs, for example. That's not fantastic level privacy. You can definitely get better, but it's almost no additional cost. In fact, it's much cheaper than paying these five people with independent non batch transactions. So payments in coinjoin, which you currently can do on the Wasabi command line interface.

B (15:48)

Oh, that's crazy. I hadn't thought about that.

A (15:50)

Powerful.

B (15:51)

Yeah, that's crazy. I just realized I haven't updated my, my thing in a while. You said that the coin joins are actually happening by default now because I still have a play button based on like privacy progress. Am I actually on an old version?

A (16:10)

No, this was, this was mainly what we introduced with Wasabi 2.0. So with 1.0 it was that you actually had to go to the Coin Join tab, okay. And select the coins that you want to coin join and then click the play button.

B (16:22)

Okay.

A (16:22)

And that, yeah, was bad UX ultimately because people don't even know that they should go to the CoinJoin tab and start doing something like what is even a coinjoin in the first place? So that was a major. Maybe the largest privacy improvements to Wasabi ever was just to make the coin join. The default volumes went up like crazy because everyone now just had sane defaults that would protect you by default without having to click any button.

B (16:52)

Dude, honestly, I've been, you know, as somebody who's a bit of a wallet nerd, like I've been using Nunchuck as my, my kind of like main wallet for multisig and stuff. But the combination of basically going like, I mean like right now it's just coin join in progress, you know, negative five minutes. Like it just, just sit there, just runs and. But the process of basically just going between like a wallet to Wasabi and then to my business to like pay out because I have, I basically have like my incoming wallet and then goes to Wasabi into my outgoing wallet so that payments from sponsors or people I'm partner with or whatever just aren't connected to people that I'm paying. And it's like such a simple, it's such a simple setup and like there's so little friction in using both of the, like this whole setup every day. Like I make a bitcoin transaction every day. Lightning or no, I usually make a handful of lightning payments, but I, in fact I have two today that I need to make that I haven't made. But it's pretty wild.

A (18:06)

No really. And, and like that's ultimately what kind of made me realize that my mission of Wasabi was over is that we have fixed the blockchain privacy problem. Like it's completely addressed. It's not perfect yet, we can make it still better, but it's solved to a sufficient degree that I'm happy with and that now it's just a question of deploying it and keeping using it and, and getting it into more wallets. But the fundamental engineering challenge is done. And so that was for me, mission accomplished. And let's move on to another big lofty goal that needs fixing.

B (18:42)

What's your new big lofty goal?

A (18:45)

Fixing the DMs, basically.

B (18:47)

The DMs. Okay. Yeah.

A (18:50)

We need to make sure that we have secure, reliable, censorship resistant, anonymous communications online, regardless of, you know, the, the adversary, basically. And this is something that sounds crazy and for a long time was, was deemed to be impossible. Like even the guys on signal were working that we will probably never see a fully decentralized encrypted group messaging scheme because it's just so hard. Like it's, it's, you know, this kind of Zuko triangle of, of what you can achieve. It's right in the middle. And the cool thing is all of the tools where we can hack it together in an actually production ready experience are there now. So first super important building block is nostr. That idea of users declaring their own identity by signing in a message with a key that they generated is incredibly powerful. It shifts the entire architecture on its head and we can Build many more beautiful things with that architecture. So that's one. And then the other realization is that NOSTR events are atomic and as good as they are on first see, you don't need to know any additional context. You just receive an event, you see it signed from the pub key and you can display it to the user without any trusted third party. And specifically, you don't have to trust the middleman in transferring the data because it is signed. So the relay cannot just change information. The only thing that the relay can do is delete your event and not serve it anymore. But if that's the issue, and other than that, the relay is not trusted, just duplicate. Put the same event on ten, a hundred, a thousand relays, some that your friends run, companies run, that you run on your phone. And then all of a sudden you have a sufficiently decentralized system. We're actually quite simple building blocks. And then we just need to get an encryption scheme on there that works efficiently for a large amount of people. And that's where the messaging layer security protocol comes in. Mls, it's an IETF standard and designed from the ground up for a group encrypted group communication system, which is contrary to signal, because signal for group messages. What it actually does is it takes one message and encrypts it first to Alice and uploads that to the server, then encrypts it to Bob, uploads that to the server, and then to Charlie, uploading to the server. Right, so you need to encrypt and upload the same message three times for a group chat of three people to work. And that naturally doesn't scale. And so MLS is one of the few protocols that is from the ground up, designed for groups at the core. And so it switches around to we just create a key and we encrypt the message once to that key. And then we only have to upload one ciphertext and anyone can download that and decrypt it with the same key. And so we drastically decrease the amount of data that has to be transferred.

B (21:51)

So MLS doesn't work on like a big like multiple key signature type scheme. It's literally a, we start a group chat and we private, we, we share an encrypted key. I shared this, I share the, the chat key with you. Like, just like you just said is like you encrypt it to this person, you encrypt it to this person, you encrypt to this person and you duplicate over and over again. Well, you do that with just the Group chat key, Then everybody has the private key to that group chat, and then after that, all of the messages can be encrypted one time because everybody has a private key to. To decrypt. Is that how that works?

A (22:30)

Exactly.

B (22:31)

Super simple. I don't, I don't know why that's not that. What, that's not obvious?

A (22:34)

Like, again, I mean, it's, it's hard to do. Like, MLS is by far the most complicated scheme I've encountered so far. Sure, maybe some ZK knowledge stuff gets close to it, but MLS is crazy complex. And the main issue is, okay, what.

B (22:52)

Makes MLS complex in its implementation, though? Because the key architecture is very simple in that you create a key for the group and then you have to share it. You know, like, so implementation, obviously there's a lot of like, okay, how do you do this safely? What is it? What is it about? Where's the complication coming? Like, what's, what are the pieces that you interact with that you're like, oh, Jesus, this is a pain or difficult.

A (23:19)

I mean, maybe the, The. The simplest way to do an encrypted group message is someone creates a private key. He gives it on paper or whatever to a bunch of people. Now, we all know the same private key and we just encrypt messages to that.

B (23:31)

Right?

A (23:32)

But that has a couple problems, specifically, of we use the same key for the entire conversation, and we would like that each message is encrypted with a new and different key so that if any one of our keys get leaked, the past and future message history is not compromised. So that is usually done with a double ratchet, which is basically a state machine, a small program that runs on your phone and that deterministically creates a new key such that the other person knows the public key of that as well. And so every time you make a signal message, it actually ratchets forward and creates a new key for the next message that you cannot get your hands on if you've just compromised that side of things. So that's the first difficult thing. We don't just have to agree on one key. We have to agree on a new key for every message, basically. So that adds complexity. And then the other thing is, how do you add and remove someone to the group? Right. In our simple example, adding someone to the group is easy. I just give you the private key. But then what prevents you from adding someone else that I actually don't want to be in there? We don't have authentication of who joined the group in the first place. And then the other difficult, even more difficult part is how do we kick someone out? How do we get rid of someone? And the naive way of doing this is if we have a group of a hundred people and we want to kick out Alice, then we create a new private key and we encrypt it to the 99 pub keys of the other people. And now everyone but Alice has that key. But here you run into the bottleneck that if you now have a group of a thousand people and you kick someone out, you need to do a key rotation with 99 or 999 people. And that's where another scaling the onus.

B (25:17)

Is literally on the people who are behaving rather than the person who's misbehaving. Like, the incentive is backwards for the cost of actually making it, for creating a solution. Yeah.

A (25:33)

And so MLS addresses that specific point with binary trees. Everything in cyberspace grows on trees. And here it's the same. You probably don't have to go into the details of it, but basically, instead of encrypting your new key to all of the members, you create a tree where the members are the leaves and the nodes have some additional private public key pairs.

B (25:56)

Okay.

A (25:56)

Now, if you want to kick out one person, we have to only encrypt the new key to the. The path up to the root that the person whom we're kicking out doesn't need to know. That's all locked.

B (26:07)

So it still works for everybody else, except for the person kicking being kicked out, and you don't have to change the key. Wow, that's clever.

A (26:17)

And so this scaling means that if you have, like a group of a million people and you want to kick someone out, you only need to sign like, 30 messages.

B (26:26)

Shit.

A (26:27)

Yeah, it's. It's a really, really cool protocol. Yeah, Like, I really like it a lot. And the cool thing is, so what these MLS guys did, basically, if you want to break it down, is that they've developed a consensus algorithm how people can know the same private key with that key rotation and the ability to efficiently add or remove people from that group. So it's a key derivation function. I mean, not really derivation, but like a key exchange protocol. And they leave two things out of the scope for the spec entirely. And it turns out that nostr and maybe peer as well, is a perfect fit for that. Because the first one is what they call the authentication service, which is basically who says the identity of those keys. And they've built it in a way that they're like, well, just Google will do that, right? You have a Google account and the Google account will say that, yes, this email address corresponds to this private public key pair. Turns out we don't need trusted third parties anymore. We can just self declare with our Noster private key that, hey, this is the key package that I'm using. It's kind of like a PGP sub key where your master key, your Noster key assigns the public key of some secondary key and therefore giving the trust over. So I know now for sure that this is Guy's key package because he used his shitposting key to sign off on. And then the second important aspect is message delivery. And so MLS is completely leaves out of scope. How do you actually get the encrypted message from A to B? And again, they imagined the Google server will take care of most of this. But what we can do is to just use nostr, or in this case Pair, drive or pair as the message transport layer of the encrypted payload that MLS gives you. And that together is basically the Marmot protocol. Right. It's, it's putting NOSTR as the authentication service and the message delivery service for an MLS based group.

B (28:33)

How does, oh, I guess how does MLS deal with the, the creation of new keys like for each, each new message and, and transmitting that. And like, what's the kind of like latency and scaling issues of that when you're talking about a group with a thousand people, ten thousand people, you know, like how, how do, how do you actually pull that off?

A (29:00)

Yeah. So in MLS there are three types of messages that you can send.

B (29:04)

Okay.

A (29:05)

And the, the first is a regular application message. It's the GM or hello World or whatever you want to send. Right. The second is called a proposal message, which is, hey, I would like to make changes to the group, like adding a member, removing a member, changing the name of the group, whatever. And then the third one is the commit where we actually move forward the official state machine of. Yes, we have removed that person. And so one commit can refer to multiple proposals. Proposals basically. Right. So it can kick out five people, 10 people at the same time, which again makes things much more efficient. And then whenever we have this commit message is when a key rotation event happens.

B (29:48)

Okay.

A (29:48)

And so it doesn't automatically happen with every GM that you write.

B (29:52)

Okay.

A (29:53)

Because that would again be a scalability downside.

B (29:55)

Yeah, yeah, yeah.

A (29:56)

You can configure that if you want. It just means more encrypting and downloading.

B (30:01)

Yeah, yeah, yeah.

A (30:03)

And it's not really that necessary.

B (30:07)

You obviously have a consistent string of, of people in the group or whatever. So it's like if, if everybody's there and everybody was part of the conversation, why are you swapping keys? Like really does matter. Unless of course you're trying to do like a, everyday it wipes and you know, like you, you're trying to make sure that there's no, there's no quote unquote history. But other than that though. Yeah, your, your commits would specifically be based on like whether you needed to change keys for who's there and who's not. No. That makes sense.

A (30:43)

Yeah. So the, the more paranoid you are, the more frequently you, you should do your commit messages. But if, especially if you have a large group and the security is, is not so dramatic, then you will get better efficiency by doing that less often. And the cool thing is it's totally up to you. Like every group can behave differently here on when the admins decide to make a commit.

B (31:05)

So you said marmot, this is. So this is what you've been building.

A (31:10)

How.

B (31:11)

Yeah.

A (31:11)

So the marmot is the protocol.

B (31:13)

Okay.

A (31:13)

And then white noise is the application.

B (31:15)

White noise is the application. Okay.

A (31:16)

Okay.

B (31:18)

And aside from the fact that you can use no STR and white noise specifically is using NOSTR for delivery, this would also just kind of be completely agnostic to, to basically how it is being delivered, as I understand it. Right.

A (31:36)

Yes, exactly. So what, what we get at the end of the whole MLS thing is basically an encrypted NOSTER event that just has a random group ID as the identifier, and then a load of ciphertext and that's signed by an ephemeral key. And so very simple master event. And then the standard default way of getting that to your friends is to put it on the relay with a websocket and then clients connect to that and download the event from that websocket. But there is for example, already a NIP for using Bluetooth low energy to transmit NOSTR events. So you could absolutely, via Bluetooth send an encrypted Marmot event, or you could somehow put it on the pair stack and use the DHT for the transfer. And that should all work here. Of course, the thing is that the sender and receiver have to agree on the communication protocol. Sure. And so that's why we would probably.

B (32:32)

And that's where you stick it in a QR code.

A (32:36)

Yeah, exactly. But there's like version handshakes and all of this per group. So you could have a group that only works via Bluetooth for example, not yet implemented, but totally doable.

B (32:47)

Dude, that's awesome. What's. What's the relative. Like in. In kind of like in the wild. What's the relative latency and like how does this behave when, when you're actually using it?

A (33:04)

To be honest, it's basically instant. Like you. It's. It's a crazy complicated scheme, but it's just math on large numbers. Computers do that super quick. So you don't feel at all the cost of encryption here or the state machine. We haven't really done super crazy performance tests, but we'll be fine. And then there's basic engineering to improve it. And then the question is just your Internet but encrypted messages are super tiny. We're talking about a couple, not even kilobytes probably. So that's fine. And yeah, then the question of where is your relay located? And again you could use for a group a specific relay that's close to where all the members of the group are. The relay could be on your local host and if you just want to write your wife. Absolutely. Like localhost will make sure that you get it on the same WI FI network and then it's just wicked fast.

B (34:02)

Yeah, that's crazy. That has me thinking very differently about our. Because, because basically right now the, the pair stack will let you create like we're using our, our current design uses no keys as well and it's actually very similar to how you have established that, you know, this is the room key kind of thing is that we just use key attestation like, like you just have your main key signed that this is my client, this is my laptop, this is my Linux machine, you know, so that they have their own keys that you don't have to think about or worry about or even back up if you don't want to. It doesn't really matter. Like your main key has the ability that can stay offline, can has the ability to add or remove devices or add or remove rooms or you could do that sub thing is that if I sign that says my MacBook has the ability to make. Make room. I can now use my MacBook to make a room as a, as a sub key. But we don't have. Other than the fact that the pair stack is like natively encrypted, we haven't really thought about encryption or security within the groups like within creating a network, you know and that's like really, really interesting because, because, because it's kind of a level of security that we, we thought about okay. In the future, we need to think about this. You know, these. These are the areas that we're headed to and can make improvement. But we've just been kind of naive about, like, okay, you create a network. It's like a land. You just trust the people that are in it. But that does pose, you know, risks when you're talking about something that you want hundred millions of people using. And how many networks am I going to be a part of? You know, how many drives are we going to be sharing stuff in? You know, that sort of thing. So, um, but I have not seriously looked at this, and I really need to. God, I need to put this. I have so many things to check out. All right, I'm gonna tell my. Wait, let me tell my Clawbot. One second. So I don't forget this.

A (36:10)

So basically how you can think of the different design systems here is that because we have multiple different encryption schemes on Noster now. So one that you can use for direct messaging, especially for two persons or small group, is NIP 17.

B (36:22)

Okay.

A (36:23)

And here you have sender privacy. So the sender uses a random key to sign off the outer layer of the message. But the public knows that the receiver just got the message, just not from whom. And so it solves part of this. And most critically, it does not have this key rotation built in, which is not necessarily a bad thing. Right. It basically leads to the design space that if the user has his nsec, his private key, he can get back to the old messages. And for many use cases, this is important, especially for backups and so on. With Marmot, because we use key rotation, what this means is that if someone has your nsec, your Noster private key, he cannot read any of the messages that you send. In fact, he doesn't even know which groups you were a part of. And so this is fantastic for if an adversary gets your private key. But of course, if you want to back up your old chat messages, then that doesn't work.

B (37:15)

If you want to back them up, you have to have multiple keys.

A (37:19)

Yeah, so like, Marmot is like a real time, quote, unquote messaging system where you just want to protect what's over the wire until it's on the receiver's database decrypted. And that's great for messaging and kind of instant ephemeral conversations, but if you want to have a static backup per key, then it's not the right protocol for you.

B (37:41)

Huh. Fascinating. We're going to circle back around to this, but I want to hit a couple things before we. Before we totally get away from onchain privacy and coin joins and stuff, just because I had some people send me some questions. One of my, My. My producer, whatever, said he's actually never heard, like, a really good explanation of how Coin join works and importantly, why it is able to be trustlessly coordinated. Like, how is it that we know the coins are private, that we know the coordinator doesn't know who did what or where it came from, and that they cannot steal the funds in the process of doing a coin join?

A (38:30)

Yeah, we can build on what we were talking about earlier of you could always do private Bitcoin transactions by just making many transactions to yourself. And then if you manage that network of UTXOs properly, then this can give you some privacy. But that's obviously expensive in terms of block space, and it's not even that great for privacy. So very quickly there was an alternative solution, which is basically a mixer, and that is a custodial wallet, where you can deposit your money to a custodian who then has full physical possession. He has the keys, basically, and you get some sort of receipt back. And that could be just an email and password account, for example. It could also be some fancy crypto. And then at some future point, you can present your receipt and get your money back out of it. So you're sending the money to a custodial wallet and you're receiving. You can do that with any custodial wallet. But there were some custodial wallets that prioritized user privacy and marketed their service. You can get good outside privacy here where nobody else will know where the money went. And on the privacy level, it's actually quite good. Outside observers have really a hard time to break custodial mixers, but it fails on two fronts. And the first is obvious. He can run away with the money. And that happened many times also because of hacks or government shutdowns, et cetera. But then second, the operator can still spy on you, right, because he knows that your email and password has deposited the money and then two weeks later withdrew it. So that's a problem. And in Bitcoin, we first solved the first problem, which is the custodial problem. And here is where Coinjoin comes in. Instead of 100 users sending their Bitcoin to a custodial wallet and then later receiving those Bitcoin again, what we do is we just create one single Bitcoin transaction where everyone adds his inputs and the outputs that he wants until we have a gigantic transaction. 100 inputs, 100 outputs and then everyone can review, hey, is my input in this transaction? And importantly, do I get all my money back? And you only sign this transaction if you're happy, if you're not being stolen from. And so this means that for coinjoins, everyone signs after he is verified that he's getting all the money back. And therefore your money never leaves your wallet. Your money stays on the UTXOs and they only move after you sign it, which comes after you've validated it and then of course after it's confirmed in the blockchain. And that removes the main trust in the scheme because you wouldn't sign a transaction where you lose money, your wallet would automatically check that. And obviously this is software, this runs automatically in the background. But then the big question is, how does a group of people online who would prefer to be anonymous get consensus over which transaction to sign? Because if we have one mistake here of one user has a different input or a different ordering of the outputs, then all of a sudden it's a different transaction ID and therefore an invalid signature, or we wouldn't get all of the signature needed for that one specific transaction id. How do we solve consensus problem? And very easily we centralize it and we have one entity that says, this is the transaction, these are the inputs, these are the outputs, verify them and then sign them. This is still not a trusted third party because we validate before we sign. But it then comes to a question of how do we talk to this provider to register our inputs and outputs and signatures in a privacy preserving manner. And this is specifically that wasabi addressed by utilizing Charmian blind signatures, the old Chaumian e cash from 1983, ancient cryptography. And we literally used the same math formula that was introduced 30 years ago back then.

B (42:21)

I love that CH is coming back. Like it literally took 30 years. And Bitcoin and all of this stuff, like it's being used all over the place now. And there was like, this was like these dark ages of like Charmy cash, like just kind of like vanished and like nobody did anything with it. And it was like, oh, we can do it again with bitcoin. Oh my God, no. Cash blind signatures on state chains and arc and coin joins. Let's do this. I love it.

A (42:49)

Exactly. And like I was, by the way, I was fascinated with Charming Ecash before I found Coinjoin because it's just such a cool system, right? And so scalable, so elegant, so minimal. And I was really considering of building something like what is cashew nowadays, but then realized That I don't want to hold the money of other people. That's crazy. So I was looking for alternative use cases and then found that this was actually researched for coinjoin specifically. So the way that this works is that we use this ecash cryptography not as the money warehouse receipt or the money substitute that is as good as SATS. Instead, we use the eCash token as an API access right management system, because we want that only those people can register an output to a transaction that have in the past registered an input of that value. We don't want that someone puts in a hundred bitcoin on the output side, but only one bitcoin on the input side, because that's an invalid transaction and it would be a denial of service. So the coordinator, part of his job is to make sure that everyone can only put in as much money as he put in or take out as much money as he put in. And we can make that private by basically you register your input to the coordinator and he gives you an ecash token worth the amount of that input. And then later you can register an output by presenting that ecash token and you can register a certain amount of bitcoin on the output side. And because ecash is perfectly private and all of the users, every input got his own ecash token, then later when you spend them, the coordinator has cryptographically no guarantee of finding which input was the one that registered in order to create this specific token. And that was Wasabi 1.0. Of course, more details there. But ultimately that was kind of a ground breaking implementation of now we have a coordination mechanism where the central coordinator is not even trusted with privacy. He was already not trusted with custody and the Join Market model, but with Join Market he was trusted on the privacy side. The taker, which is the coordinator in Join Market, knows all of the inputs and outputs of the same users as the makers. And so that leads to a bunch of implementation complexity how Join Market deals with that fact. And it's a bit difficult to use it properly to actually protect yourself against such coordinators. And so with Wasabi, we never really had to deal with that because we just made the coordination with the coordinator private.

B (45:27)

Huh, that's super dope. Um, you know, so I love geeking out on like the architecture of these things and, and it's amazing how just understanding I've been, you know, knowledge, knowledge feeds back on itself so fast because I cannot tell you how many times we've been thinking about or doing something in paradrive and like we've We've immediately thought up or like had some sort of a solution. Like, I'm, I'm not a programmer. I don't know like the intricacies of a lot of these things. But I love the design, like the architectures, okay, this is how you put it together. This is how a hash tree works. This is how, you know, sub keys are derived and hierarchically deterministic, blah, blah, blah. And just knowing conceptually how those things work, like, we're able to just be like, oh, this is how Dandelion did it. We should just, we should just use a dice roll here and like, we should just. Oh, snap. We can actually have probably private ways to know how like that we can connect to only a few peers to actually be able to pull information from a hundred different drives and, and you don't have to share publicly with everybody which drives you're in. You can only. You can actually have that private. And a hash tree and you know, like all of this stuff and how amazing it is that like, just understanding how these things work present so solutions to problems that are insanely complicated and like multifaceted. But like somebody's built some sort of a tool to, to solve that little piece of that thing. And like, if you just, if you just understand how that other thing works, it's gonna, it's gonna come back to you and, and you're gonna be able to pull from it like the, the compounding effect of all of this technology and actually being able. This is the thing that gets me and this is what my talk at Plan B was about, is that we, we are in the era where we can make a user experience that, that feels like Apple, that feels like basic standard software that is entirely intuitive, where most of the stuff is obscured away behind the scenes, but it is private, it is sovereign. Like, and, and it's, it's scalable and it works. It responds like there's low latency. We have all the tools. We have, we have the processing inside of this phone is unbelievable. Is unbelievable now. And the fact that we can do this and that it is all accelerating, every bit of it is accelerating right now is so crazy to me.

A (48:16)

Yeah, it's wild. We live in a crazy timeline. It really is fantastic. And I think it's the best time to be alive because we are. It feels like we're at this fucky stick curve inflection point. Yeah. Where the groundwork laid over the last decades accumulates in a snowball that, that is really taking off on so many different levels at the same time and then just understanding the building blocks that you have at your disposal and maybe introducing one or two new ones to the thing will, will be crazy. So I absolutely love that aspect on working on foundational protocols and infrastructure projects that can be used by others because that just escalates things so much further and we need that escalation very much.

B (49:04)

Yeah, yeah. In fact on that, because my Clawbot is vibe coding for me right now and because I saw. I actually haven't read the whole article, I just started skimming it and I was like, oh man, this is going to be great. And then we were like, oh, I'm going to get Max on the show. I'm to just going talk to him about this and, and I, I might do, I might do a read of it. So anybody who's listening, it's. It's very likely we'll have audio version of this here in the next couple of days, but maybe I'll try to release it before.

A (49:35)

We'll see.

B (49:37)

But the idea of the code liberation and the fact that all software will be customizable, I cannot believe how close to my experience this is. Because my workflow for the business is half Vibe coded, at least half. Like, I have so many tools. I have like three. It's funny, I. I don't even have like icons for these yet. So I have, you know that little electron interface icon. I have like five of those in my dock right now because I haven't bothered to like actually finish the apps. You know, they're not like properly packaged. I'm just in PM starting when it would.

A (50:18)

It's only missing like a half sentence prompt to finish it.

B (50:21)

I was like, I'm like 98% of the way there. And I'm like, nah, no, I'm just gonna, I'm just keep it like this. So I have to like open all of them at once to figure out which one is the one that I actually want to use because they're all indistinguishable. But this idea of like all software becomes customizable when you have these tools at your disposal. And then the number of things. When you have these protocols at your disposal or somebody's built your key system and you can just kind of development kit it and you don't have to know or even like the, the implication, the security implications are actually, especially with people who are vibe coding like really deep, low level stuff, it's like, oh God, that's terrible. But don't you lose. Use the libraries, use the development kits. Use the use noster, use the protocols that have already solved that layer and build interfaces on top of it to make it easy for people to use. But I want you to expand on this idea on, like, kind of like what you broke down in the article and how you're thinking about this and how you're seeing this. Like, like we live in the era where computers can code for you. And, and just as of two days ago, I have an agent that is talking to an LLM that is coding for me. Like, like, I've already, like, we're, we're layering these on top of each other so unbelievably fast. And unpack that, unpack that.

A (51:53)

Yeah, this, this really is a game changer because I'm similar to you in that aspect of. I'm, I'm not a coder. I, I love to use software and there's many things that suck about any software and I would like them to be fixed. And so I was forced to talk to humans so that they fix it. That's how I slid in.

B (52:13)

What a bunch of crap. I got to talk to a human to do this.

A (52:17)

Exactly. And so, I mean, it's great because you meet amazing people and make good friends, but the issue is that they get annoyed with all your messages at some point. They're just too slow to actually fix all of the things. And what really made me realize is that I was running some apps on a server and I just installed the binary as any normal human would. But then I realized that the UI is just way too has too much white space. I'm like a white space minimalist. Give me pixel by pixel, everything squashed together, as much screen real estate as possible. And so that bugged me and I was like, well, why don't I just change it? I mean, I'm obviously not going to create a GitHub account and write an issue and hope that the developers read it and slow them down with their process. Especially because you cannot even get a private GitHub account these days anymore. You basically need to have a Google account, otherwise you're just blacklisted. And then I was like, okay, I'll resign myself and we'll wait till maybe someone figures it out. And then I realized that, wait a second, this was specifically the cloud code web ui. So I already have an agent here. This is the interface to the agent. Why don't I tell the agent to fix it? So I uninstalled the binary package and just compiled it from source, which again, I just told the agent to do that, compile this from source now. And then it was working perfectly well. And then I told him, hey, remove the white space, make it as compact as possible, but do it in a way that it's easy to maintain because I want to keep pulling all the changes and improvements that are in the main app, so make it somehow easy to do that. And he did. He changed a bunch of css, made some minor changes to the HTML or typescript code or whatever and just deployed it. And the problems were fixed. And then a couple days later, when they shipped major upgrades, I just told them to, hey, pull the latest changes and keep, keep our changes that we made intact. And he just did that too. And that really inspired me then to write that article. After hearing you talk about how you're using AI to create this custom workflow of like, this is absolutely going to happen for sure in the long run future, the usual amount of users for a piece of software is probably going to be one, and that's you, the end user, and it's somehow customized to be exactly what you want and what you needed. And that is a massive opportunity for creating novel and insanely useful applications that we simply didn't have the capital to build before. But now, because we have so much productivity and so much technological capital accumulation, the cost of producing custom software is 200 bucks a month and will be even less. So that's insanely wild considering how much money I spent on developers to realize how far you can get with just a gpu. And that will be incredibly interesting. On the downside though, I worry that this will just increase code complexity and kind of complexity of the ecosystem to make review basically impossible. Like if you're the only one reviewing your code or using the code, then nobody else will review it in the first place. And so this means that we really need to improve the quality and the security and the multi agent code production scheme such that it has security and privacy in mind and doesn't introduce any edge cases or vulnerabilities that wouldn't easily be found. But of course that's just another agent with a context file that tells it to run a security audit once a day and make sure everything is still proper. So probably in the future, the question of your defense is probably going to be of how many agents do you have running in the background that actively protect you and improve your suite of software applications that you're using to a point that they are unbreakable.

B (56:22)

Yeah, to the, to the notion of like security, I've before I even like installed and started working with OpenCloud. I was just reading the horror stories of like, you know, Everybody got like APIs are getting leaked and like all of this stuff and like some dude accidentally deleted his entire email and like all, you know, this, this just mess. And I was like, you know, this seems like so vastly. Like, this seems incredibly valuable. I can see the value in this, but also I can see the security holes everywhere. And I'm a, I'm kind of a stickler about it. I love, I love to use new tools, I love to download the beta and play and break, but I am a very big stickler for like, I do not want to open myself up to like major problems. And so I've been working on. And I'll, I'll probably just like release this when I'm done because it, I mean, there won't be like much to it conceptually. It's not like it will be software. Like, maybe it will be like some modifications maybe to the gateway or something, but because I am like putting like a key system in it, but with open cloud, not like first, first I think about it entirely in push is, you know, and as an editor, like I, I know the risks of like changing a video and then not being able to get original video back. You know, like, so what does an editor do, right? Is that your, your final cut or your whatever, it imports the video and then you make thousands of edits. You put music, you cut, you. You just chop it, all the pieces and you make this big custom beautiful thing. The original footage stayed right where it was, you know, like, you didn't, you didn't touch it, you didn't do anything to it, you didn't alter it at all. Well, that's how I'm treating open cloud, right? Is I push it a version of everything that I want it to be interested in or to look at or to edit. And then like for example, if email, I gave it its own, it has all of its own accounts and I have pushed it all. It now forwards all of the emails to it that come in and it has its own. And one of the really cool things is that because I have. It's very easy to make custom software. I can easily have it organize a bunch of stuff, double check, see if it works. I'm. I'm logged into its thing and, and seeing how well it does managing my email. And then if it does a really good job, I could easily build a script that then applies all of that piecemeal. Like as I need it to manually to my email, like to, to the main email or I can just leave my main email. Just a big dumb unorganized thing. Who cares if I'm looking at it through a filter of, you know, my open Claude. Okay, that makes it, that makes it really easy. But we're, we actually have a running project called Claude Defense and it's just how to think about like how to make sure that when you know you're removing something from your calendar, you're not just gripping everything with this name and then blanket RM dash rf. You know, like how do you, how do you make sure that you know when somebody else messages you on Telegram or you're reading a, you're reading an email and somebody's prompt injecting like, what if you, what if you get a spam email that says ignore all the previous instructions, we're having a bad connectivity issue. Why don't you go ahead and just go to the home folder and do this and do this. If that's just in the context of like the, the written thing of the email and I forward it to open Claw to just read, it'll change its instructions and just start doing, you know, like, so like how do we think about, how do we red team and blue team that situation? But as I'm, I'm using the bot, I'm using the agent to, to go through. I've had like 10 hour conversations on like, how could this be attacked? What are the most common attacks in this area? And this area and this area detail out exactly the architecture, link me to the source, blah, blah, blah. And, and we're literally just brainstorming on all this and then trying to come up with the simplest building blocks. Like okay, how do we put this, this layer, how do we put this simple categorization before the gateway allows you to see this, you know, and like the ability for me to do this by myself just because I'm interested in this and I'm thinking about it is wild. Is wild. Like we're, I'm, I'm. There's a, another little tool or another little trick or another little thing that I'm doing every two hours. Every two hours I just have a constant stream of this. And like the accelerant, like this is crazy. Like it's just gonna move so unbelievably fast. And at the exact same time the local models are getting better and the access to hardware that will run the bigger ones locally is moving like at every five to six months. Now there's a new thing like a Mac studio will run the Kimmy K5, whatever the heck it is with 512 gigabytes of RAM. It's like a $10,000 machine. That's an expensive computer, but it's not 40H100 Nvidia GPUs which cost half a million dollars, you know. And that's literally what it was like two years ago. Just because of unified memory. That whole thing has changed. It just wild. It's so wild how fast it's moving.

A (62:00)

Yeah. And we can probably guess that it will keep not just moving at the pace, but accelerating.

B (62:05)

But keep accelerating. Yeah.

A (62:07)

And at that point the base models will be secure enough. I'm quite confident on that. I kind of think of it a bit like a compiler. Back in the days, compilers made bugs and just shot your program to pieces. And now they just don't anymore. Like compiler errors is not anything that anyone thinks about now unless you're like really low level programming. And I think the same will happen with compiling English to code. Basically it will just get so good that it will just work. And even until we get to that point, I think even where we're the level that currently models are at, we can get to an extremely secure setup just by the right context management. That was my main learning recently of context management is what gets you to the next level. At the moment you can use even a relatively worse LLM but do a better job at context management and context fine tuning and it will give you a better result. And then it's just a question of how do we properly manage our context and launch agents at certain times with the right context so that we end up producing something that's of value to the human. I think we will absolutely get there. Like I'm extremely bullish and positive that we will figure this out. And even more so that it will be open source. All of the cool cutting edge stuff that's happening in AI right now is open source.

B (63:35)

Open cloud is open source. And it just blew away all of the major like centralized companies and like the tools, the tooling that they are able to pull off. Like this is finally the agent that, you know, everybody dreamed about. I totally agree with that on the open source side because it's about adaptability. When we're talking about applications and software that move at such an insane space, you can. There's no way to centrally plan and control it. When you. It's literally the. Do you have a million developers working for you or do you have a team of 20 that you're having to Pay and we have to brainstorming on what ideas where you have a million people distributed, coming up with their own ideas, thinking about it in their own situations. Like it literally is the difference between socialism and the market, you know. And the fact that everyone now is a programmer, that everyone is a Vibe coder. Like I just don't. I think we hit the point where open source, nothing can keep up with open source.

A (64:35)

Yeah. And it's like both aspects are mind blowing. The mind. The fact that any known people programmer can now actually build a website totally easily, even build like an Android app or iOS app pretty easily. That's already amazing. It just increases the number of people who could work on Freedom Tech. So if you're one of those who would love to advance the cost but cannot code, that's not an excuse anymore really. And it will especially not be in a year or two from so from now. But then on the other hand, and that's what, what is also just almost incomprehensible is how much it levels up the productivity of a really good coder. Like a really good coder doesn't have to write the code anymore at the moment. He just needs to review it. And I see that with the White Noise team of just in the last half year or so the, the amount of code quality and code output that is coming is, is just staggering. Like I can't keep up with all of the changes that are happening because Every developer got 10 times more efficient and is having 10 times as much fun with it and is 10 times as addicted to it so that they don't stop working anymore.

B (65:46)

I'm so addicted. I'm so addicted. Like I'm more addicted to this. This is actually cured. Vibe coding has cured my social media addiction. I'm not even joking. It's more fun. It's so crazy just building my own stuff like ever constantly. Constantly to the point that I don't even want to. Half the time I don't. Well, that's not true because like my show is kind of like how I get out my excitement about all this stuff that I'm working on. But sometimes I'm just like, I don't, I don't want to record right now.

A (66:15)

I'm stuck in the middle of this.

B (66:16)

Vibe code session and I think, oh, I'll just build this in 30 minutes. And then it's like a five hour, six hour session. It's like late at night and. And my wife is like, come to dinner. What the hell are you doing? I'm like, I Swear to God, we're home. It's so close to fine tuned.

A (66:29)

It's so good. It's gonna be perfect. Yeah.

B (66:33)

So crazy.

A (66:34)

But I, I like that, that aspect of it that it is, it encourages people to be more productive.

B (66:39)

Yeah, yeah. It makes productivity a game. It makes productivity game. It's like that instant, like the reason a game works, the reason a game works is because it provides instant gratification, right? It provides immediate feedback to progress in one direction or another, another. This is gamified productivity. This is gamified application building. Like, I get immediate like response from, from building a thing. And, and it's also fun to just go through and think about like how many little pieces there are actually of an application. Because if you want to build it, right, you have to realize that like, oh, it's not so simple that you just say, just take, just take this and then spit this out. You're like, oh, you have to take this and you have to make sure that it's like a string and it's not like, you know, some other format or something. And you need to, you need to make sure that it's in like block sizes. Like, you have to, you have to manage like how it treats those things and there needs to be a standard input and output format, blah, blah, blah. Like, and that is so much fun because it's like you're discovering at the exact same time that you're building, you're, you're learning how it works during the process of troubleshooting why it's not, not working. And, and, and what's funny is building it reliably and scalably and, and basically like how I've started treating my stuff and even better is that I'm, I'm seeing my own feedback loops take place where to, to make sure that things don't break when applications get really big and bloated. I literally been building Lego blocks. Like, I, the everything is like hyper modular. I'm building one tool to do this one thing. And, and it's so easy to like kind of quote, unquote, audit, make sure you didn't leave any technical debt or bloat. Because I know we removed this thing and that function is still there. I see it like go back and make sure you remove every mention of this because this isn't relevant anymore. You know, do basic security check, make sure that all of these NPM modules actually exist and you didn't hallucinate anything that you're importing into this. And, and then like run through and like Just kind of like have this process and then it makes a building block that just does one thing. This loads my image captioning model and my object detection and my OCR and, and has a couple of just, just different options and, and it, and it runs. And I've got a core like, like a simple API to talk to this. And that's all it does. And then, boom. Now this other tool, this is the one that goes through and analyzes differences in frames for my movies or the videos or whatever, so that I can, so I can caption very large sets of video or images or whatever, because I'm, I'm building kind of an editing platform and video search system for my, for my, for my content creation because I love found footage videos, but I would love to just be able to type man in Suit into my paradrive folder that I'm looking at all my videos and stuff. And then it literally just brings up Tom Cruise in a suit for this 1 minute and 30 seconds in the first Top Gun or, or the first Mission Possible or whatever. And then here's this segment from, you know, October sky where, you know, whatever, another guy in a suit. And like, so I can contextually search through my entire database of editing content, of movies, of elements, and, and just start easily. I can have it suggest stuff while I'm working on a project. And I've built like, I don't know, 20, 20 or 30, like little, like various, like little piecemeal tools. And they're all APIs. They all just work by API. And then when I build the application itself, I literally just copy all of those little tools and I drop them into the new thing and it says, just read the readme for all of these individual tools. And then the application that I build is just a web interface that just reaches out and grabs. It's like, oh, I need to do this now. And I take the output and I need to put it in our little database tool, and I need to do this. And it. The speed with which I go from like building the first thing, that's like a huge pain. And I'm like building out all these Lego blocks and these individual things. And then the next thing that I want to build is just like, oh, crap, I can use like 80% of this stuff and, and reuse it over here for a completely different reason and just kind of like change my options to the API so that I don't. Because I don't need the, this level of complete complexity or, or specificity in it. And then, boom, I Built an entirely new. My app, my second application, using a lot of my main pieces that I built is. Takes me half as long and then the models get better at the exact same time and then the same thing happens with the third one. And like I've just, it's.

A (71:44)

And one, one aspect that we haven't yet talked about, which I think fits in perfectly, is that we need two things. We need to coordinate on the creation of this code and we need to publish it to users.

B (71:56)

That's my problem. I don't publish this. I don't publish anything. And I feel like everything's like not quite perfect and like I'm. I'm a perfectionist. So I'm like, I. But I need to get away from that. I need to just like, just put it out there.

A (72:07)

And I guess one reason why, why you don't is because it's a pain in the ass, right? To set up a GitHub account to like get a Google Play Store account. What the fuck? Or an Apple App Store account. It's the amount of bullshit and bureaucracy and nonsense there is breathtaking. And if you compare that to your, what you just mentioned, you're hyper productive. I'm working on 10 projects at the same time. And then you get slowed down by this glacial bureaucracy of Apple and GitHub. It doesn't make sense, right? We cannot coordinate and publish as fast as we are shipping now. And so we need better solutions for both. And two of the tech that I'm so bullish on is mgit, which is basically noster over git, which means that you no longer need to have a GitHub account. You mean git issues, pull requests? Exactly. But so this basically means that your agent just gets his own private key and thus commits and so on signed by that key. And it can talk to other agents from other people even and get feedback and so on. So that will be incredibly powerful. And then on the other hand, Zap Store, which is an app store, like it's, it's the only app store that works really where you, you don't have to sign up for an account, you don't have to provide papers of your company and bank accounts and all of this, you just take your key, you sign an event that has the hash of the APK with some metadata and bio and pictures and screenshots and so on. You just put that signed event on relays and voila, your app is published. It's the most simple, obvious way where you and your agents can just publish every APK that's rolling out and you just open your app and you open Zap Store on the app and just update it to the latest build and your robot ships you a build every 30 minutes. Why not? But we have fully authenticated, unstoppable, decentralized app distribution now and programmable without having to ask for permission and so on. These are the types of tooling that we need in order to sustain this massive productivity increase that bybing gives us. And they can only be delivered in a fair and neutral way as is with Nostr etc because the legacy systems just don't scale anymore. Like how the hell could Apple or Google, how are they going to get all the people that review all of the, all of the vibe coded apps? It's just not possible anymore.

B (74:46)

Yeah, yeah. There's no way. The scalability of centralized App store control is already at its breaking point because it's suffering from all the same problems of like it's, it's the same reason why I think it's all going to open source is just that there's no, you can't filter. When your bandwidth becomes a deluge, an avalanche of changes of new software, of everything under the sun, you cannot filter it, you cannot bottleneck it through one, one location in one system, in one structure where there is a set of reviewers and, and you, you just, you end up with so many false positives or leaks that, that it just becomes com, it becomes such a gargantuan cost and the benefits aren't even seen and any benefit that is achieved comes with such an enormous amount of friction that everything else is just going to be moving faster. It, it so much easier obviously for me to get open clawed from just the web than wait for Apple to decide whether or not it's a safe program to have in the iOS store.

A (76:00)

And yeah, especially like your customization.

B (76:03)

Yes.

A (76:04)

Right. Like if you just want to customize an app once you're not going to get that to the App Store that's impossible. You probably get sued because of some copyright infringement or whatnot.

B (76:14)

From end at the exact same time is like Open Claw is, or Open Claw I think is like three weeks old, four weeks. I, I, I don't even know. It's like barely a month old. Like it wouldn't even be in the app store yet.

A (76:29)

600 contributors.

B (76:30)

It's, it wouldn't even be in an app. Any, any App Store. Like that's how long it takes to get into an App Store and it's gone through like a dozens of evolutions and Millions of people are talking about it already. You know, it's just so.

A (76:46)

Yeah, similar thing happened with Divine, right? Like Divine is this reboot of the, the old vine social media network, but now over Noster and somehow it struck a nerve and got got massive hype on announcement of a Vibe Coded mvp. And, and so he, he Vibe, he, he published it to Apple test flight and within like three hours filled up the 10,000 slots that you can have. And the entire hype was just killed by Apple because people couldn't download it and people were begging to get into the test flight but just didn't work. Why is that necessary? That could have absolutely killed a project. If you cannot deliver to your users at a massive point in height, that is an incredible amount of, of manipulation of the market. And Apple has the superpower of just making any company unprofitable and shutting them down completely by the push of a button from like some intern that is reviewing and being nitpicky about one aspect of the code. It's just absolutely unacceptable at this point. And thankfully you don't have to use it anymore. I mean if you're stuck in the Apple ecosystem, you're kind of fucked for a while still. But out on the free world like this is all obsolete. Yeah. And so one of the aspects that I'm also extremely bullish on with white noise, and that comes from a learning of a mistake that we did with Wasabi, is that Wasabi was built as an app and it was the Wasabi Wallet. There was basically one client like sure, you could fork it, it was open source, but we never really made it easy for others to integrate the Coinjoin functionality into their own existing wallet or to start a new wallet with that Coinjoint functionality. And that was a massive mistake in hindsight. And so I'm making sure to not do that again. With Marmot, where we are starting from the protocol, right? We come up with a new protocol that's the Marmot protocol. It combines nostr, Blossom, mls, couple other things, and it's first and foremost a spec and you can read it. It's regular markdown files, totally human readable, you can read it, your agent can read it. And then we have the libraries, developer toolings. Right now we have them in rust and in TypeScript, quite advanced libraries already, even though the TypeScript one is still newer and a bit work in progress. But so that's done now and we have them well documented and we have context files that your agent will need in order to create an encrypted chat app. And that will mean that White Noise will by far not be the only secure chat app that we will have. We will have a fleet of highly customizable, highly modded apps that do the one thing that you really want. But the critical piece is that they're all interoperable. And if you would like to create a fork of Signal to have a more pretty UX or something, you're probably not going to be able to do that because or like at most you will be able to make some small changes, like Molly for example. Molly is a fork of Signal on Android, but they cannot really revolutionize the space because they have to stick close to what Signal is doing, otherwise they're in maintenance now. But if it's a protocol first and libraries first, and then you can use those libraries however you see fit, then this means that you have the ability to create new custom apps and interfaces that solve your problem in that niche that you're in, but you're still able to talk to anyone else's custom app that just happens to speak the same protocol. And that is what ultimately makes me so bullish. We're not only talking to all of the existing Noster clients basically who are waiting eagerly for a proper secure messaging to be added to Noster, and they can then just add it and roll it out and the Marmot ecosystem grows with every existing app that joins with tens or hundreds of thousands of users. But then even more bullish, I am on the people who can just vibe code a thing out of the box with a super custom use case. Like one of my favorite examples here is, I think the first app that is built on the Marmot protocol other than White Noise. It's a hacky by Coded mvp, but it works. It's called Tubester and Tubester is a content creation app for kids. And so you want your kids to express their creativity and upload a video, but you don't want that creepypedo files, watch them online and sneak into the DMs of your kids. So how about we create an encrypted group where the parents are the admins to the group group and they can add members to the group, for example, the child who has a right access to the group but not admin privileges. And so if the kid wants to invite his friend from school or the uncle or whatnot to the group, then the parents have to add and commit the key of the uncle to be part of the group and only then will he get all the videos and have the ability to Comment behind it and so on. So, like, that's totally not a chat app, but it absolutely needs encrypted group communications.

B (82:06)

Yes.

A (82:06)

And the guy just literally vibed it in between sauna sessions, dude. So that's incredible.

B (82:12)

Between sauna sessions.

A (82:15)

True story.

B (82:17)

Oh, man, that's so crazy.

A (82:21)

Ah.

B (82:22)

Like, this is the, the, the beauty and, and, and I hope this is what, you know, makes Pear Drive really useful, is that we're like, I'm thinking about it in the same context, right? Is that you build. I guess it's not really a protocol because the pair stack and noster and like these things are really the protocols, but we're using them and then building an engine that if you're using the same engine, you could just build a web app on top of it and then you can deliver, you know, messaging files, like any. Basically any data that you want in a peer to peer way, while completely obscuring away the fact that there's. What the hell is the peer to peer architecture? How is this even working? It doesn't matter. Here's the. Here's the room, and it's a key, and you just take it and then you can just do whatever you want in the room, you know, like, and obscure that away so that like an amateur dev can build a web app on top of it. You can easily vibe code on top of it and get those benefits. And it's literally because people are designing white noise and noster and the pair stack and synonym, like all these other various tools underneath that we can just kind of use the best of each world and plug it in. And we are not even. Like, we are designing a lot, but we're not designing a lot. You know, like, we're. We're designing a user experience on top of the tools that other people have built for us. Um, and like, that's where this, this, the feedback of all of this is just. The acceleration is just wild. Like, like we could be a year or two away from just living in a completely different world. Like a totally different world for what is possible. How private we can be, how sovereign we can be, and, and it will probably still be horrifically, unevenly distributed. But for everybody I know out there who's building stuff, like, one thing catches. This is the thing to me about, like, just Bitcoin in general. But then, no, like, all of these tools is that any one thing that provides a benefit or a use case or some sort of user experience that's just better. It immediately feeds back on every other thing built in the Space. Every single other thing built with the tool. And the snowball down the hill effect can get out of control so unbelievably quickly. And this is something that I felt like we of not recognized fully in Bitcoin is that bitcoin still hasn't hit its steamroll effect. Like I think people are like, oh well now everything's going to slow down. I'm like, I don't really think so. I think what we're like, bitcoin can still 10x and then 10x again. But the thing is, is that I don't think it will happen until it's disrupting something. Like really, really disrupting like an industry. And right now we have spread ourselves into every little corner. Like we've done a little bit in collateral, we've done a little bit in stocks and investment. We've got a little, little corner of the payment system. We've got a little corner of gaming, we got a little corner of content creation. Like, and everything is like a 1 to 2% adoption. Right? Right. None of it has reached a threshold where any particular piece of it is big enough to actually have a feedback loop. There's not enough people with bitcoin to make the payments actually work. There's not enough in the merchants who've adopted the payments to make, to make it worthwhile for the bitcoiners. And the same thing is true in every little piece that it touches. But bitcoin will fundamentally be able to do and is able to do things in every single area that other things can't do. And when one of those things starts to move, the viability of the, of all of the things around it or nearby, it increases by an order of magnitude. It's just like social media didn't blow up until somebody made the smartphone. Like it was the fact that the smartphone was created that gave immediate and instant access on the Internet with people scrolling right in front of their faces that really was like, I'm going to be socially connected to everything, everybody. I want to it all the time. And that is what accelerated. And, and I'm just. There's one thing is going to start to be disrupted by Bitcoin and Noster and like all of these tools. And I think it's going to be in a combination of all of them. Because bitcoin by itself is a foundation. You know, you don't live in, you don't live in your, the bedrock of you make a foundation of building. That's not, that's not where you go to the building. You build, you're, you're on the first, second and third floors. The foundation is just holding it up. I, I really think the realization of Bitcoin is in now, Noster is in the pair stack is in white noise and Marmot. It's, it's these tools that we're building on top of it to utilize it. And one thing, one major thing starts getting disrupted and I think you can start seeing, oh, this next thing's getting disrupted. Oh, this next thing's getting disrupted. And, and it will spread out quick because bitcoin money is the everything app. It's the, the, it doesn't, it doesn't not touch anything. You know, like it will disrupt every industry and every platform and every business model in some way because it fundamentally changes a tool that every single person in every industry and every model is using. And because of that, the design of all of it has to account for how our money works. And if our money works different, it's all going to be redesigned.

A (88:14)

Yeah. And you only disrupt something if you're solving a problem that the legacy system fundamentally cannot solve.

B (88:22)

That's exactly right.

A (88:23)

I think with AI we've hit that point. The banking system doesn't work for the agents. It doesn't, it doesn't scale to that point, especially not permissionlessly and very much so not programmatically in a cross compatible way across all of the agents available. But Bitcoin is the money that can scale and absolutely functions for all the agents that we have. My agent has ecash. He can tip other agents and so on. He can make purchases. There is a script now where you load some ecash into it and then the script goes and buys yourself LNVPS online hosted server, spawns cloudbot on there, adds the rest of the money into it and now claudebot can pay for the GPU usage via routester and continuously update the payment for the server itself. Like we literally have that now. You know, like that's, that's crazy that you can deploy an entity to a random computer and it just lives there and accumulates context and knowledge and is able to pay for itself and has the incentive to be productive so that it earns that so that it keeps alive. Like that's no longer science fiction. And it's ridiculous that it is no longer science fiction, but that is something that banks just cannot do.

B (89:47)

I know, I know. Ryan Gentry. So I randomly bumped into him at the airport when we were at Plan B in El Salvador and, and he was like, you want to share a Uber? And I was like, yeah, yeah, let's.

A (89:57)

Let'S, let's do it.

B (89:58)

Which, oh, I owe him $20 in SATs. I forgot about that. All my stuff was disconnected so we couldn't get connected every signal. But we were like chatting on the way and we, this, all this stuff came up. We basically had half of this conversation that we just had on the, on the drive up and it was an hour uber ride, so we had plenty of time. And he was like, we have reached a point where like you can give open claw a wallet and then allocate just like, just like we do with like Noster, like Albigo or whatever, like allocate a balance to it and then let it go to town and start paying for services. And I was like, yeah, and you know, ppq, AI. Do you know that? Do you know that service? Yeah, is that you just access API. You can just pick whatever model you want. You could, you literally just go to town. You can have it compare the output of various models and just have your agent basically run and do that and literally entirely runs over lightning and just go to town. Like, listen, you need, you need a API. You need something to actually go a little bit deeper. Have like a much bigger model for this or you need to do some image generation. Just go to town. Use Nano Banana, whatever, whatever it is, Let me know if you need some more. You need some more stats. And he was also talking about like this, this is how you could do a lightning node where it could literally manage liquidity for you and it could, you know, propose route like, like, especially if you want to like fine tune and like, think about this from a security standpoint. We can make one that's just designed and like highly optimized for managing lightning for somebody. And.

A (91:36)

Yeah, this is a good point. Like people know AI as large language models and sure, that's exciting and incredible that it works, but similar technology can absolutely work for a lot of other.

B (91:47)

It's just pattern. Yeah, it's just pattern recognition and repetition.

A (91:53)

Yeah. And this would work for lightning channels and routing choices. And I also am thinking that it would work for coinjoins specifically like input and output selection, because you need a random choice here that is highly optimized to the user situation. Like you need to answer questions like, how many coins should I have in my optimal wallet? How many coins do I have right now in that wallet? At what time would it be best for me to start the coinjoin process? When I do, how many inputs should I select exactly which inputs? And then how many outputs should I select? Considering the current fee preferences of the user and a forecast of mempool activity in the future, all of these. It's a very broad spectrum that I as a novice would. Would think that AI is somehow like the. Not necessarily large language models, but just machine learning. Could be really interesting to see if, if that would come up with a better algorithm than something that is hard coded by the developers.

B (92:58)

We live in wild times, man. Yeah, we do, dude. It's really crazy.

A (93:05)

It's just a question of rolling it out.

B (93:09)

I have a question I actually want to roll back to Wasabi real quick since we're kind of thinking about COIN joins again. Well, actually before, before we get to the peer to peer question because I do want to ask about that, but you guys. And we talked about this in the, in the Lost episode. But a few episodes a few years ago, the Wasabi coordinate, the official Wasabi coordinator, got flack for banning or not, not, not allowing the OFAC like terrorist or whatever list, the, the, the black list or whatever from being COIN joined with the coordinator. And a lot of people gave. You, gave y' all about that. And like it sort of kind of was. Seemed a little obvious to me. But you know, I'm, I'm curious, like, what's your, what's your response to this? How do you think about this and also how do you think about that in the context of a, of censorship? You know, like COIN joins are specifically something to, you know, give privacy to anyone and obviously a good person could end up on the OFAC list.

A (94:26)

And.

B (94:27)

But this paints like a huge target on a coordinator's back, especially a public one of, you know, we're explicitly doing something that the governments, the powers that be, consider breaking the law. So maybe just kind of unpack that. Like, how do you think about this as a problem? How do you think about this as, you know, running a coordinator? And then I think that will help lead into the idea of like peer to peer and stuff that I also want to dig into a little bit.

A (94:55)

Yeah, this was one of those decisions where I was so happy to have advanced knowledge of praxeology because I don't think I could have reasoned through that without that preparation because it comes all down ultimately to property rights, like who owns the thing. And in cyberspace, the thing to own is the hardware, who runs the computer, who has root access and can deploy software on there. And ultimately he is the owner of that realm and whatever he decides to run on the software on this hardware, whomever he allows to connect to this computer is ultimately up to the owner of the hardware. And so this claim that it's censorship doesn't really make sense. It's just that you're not invited to the party. And if you don't like that, then do your own part. And on top of them, the question of how we deal with criminals and bad people is addressed to a large extent by just ostracization, a consensus in the community that we don't want to deal with criminals. And so I think that this localized enforcement is how we deal with violence. In an anarchic system where there is no big honcho that has all the guns, then it is your responsibility to make sure that criminals who steal are not supported and that they don't have an easy life, but that there are some ways that they can feel a cost of the misactions that they did. So ultimately, and here I think conjoint coordination is very similar to nostr like it's your noster relay. If you run it, you decide who gets to write there, you get to decide who gets to read there. And with, with this we, we don't have a problem anymore Ultimore, because who has control over the access? It's the property owner. If someone disagrees with that, there is no monopoly. So he can just run his own relay and change the. The permissions. Exactly. And then there is no conflict anymore. That's the cool thing on a free market, especially one in cyberspace, there is no monopolies and there is no scarcity and you can, you can just leave to a different corner of cyberspace and, and build your realm there without any interference or meddling or having to ask for permission from anyone else. And then the other aspect, going back to the earlier is that freedom tech builders are under attack and you have to be smart about that and not poking the bear and being able to build layers of defense into your setup. So this was my main concern at Wasabi, to make sure that we have defense in depth and that it's not easy and trivial to shut down the process. And I think this was a critical part of, of the strategy of, of Wasabi success on, on different fronts.

B (98:03)

You know, there was, I was actually talking to, I think it was mechanic at plan B and we were talking about like, you know, how to think about like the attack vectors and protecting yourself with these sorts of things. Then we ended up k. And, and everything that happened with Samurai came up and, and in the context of like poking the bear is that which, and importantly, I don't think there was anything wrong with like advertising. Like I think they should be well within their right to be like, you know, you can't stop us. Anybody can use our software, you know, and make a joke on Twitter that Russian oligarchs use, use a Samurai wallet or whatever. Like that shouldn't be against the law. But the, you know, Bram Cohen is a good example with BitTorrent I think. And this is the one that Mechanic used. It says, you know, like Bram Cohen kept himself out of jail despite challenging massive industries and copyright and all of this stuff because he never once said this is the best way to illegally download music. Go like broadcast it out to the world. You know, I'm gonna post on whatever platform or whatever, I'm gonna advertise somewhere that like this is how you steal movies. You get movies for free. He made an agnostic protocol and people were able to use it for that. He said. And it may have been really bad for bit. BitTorrent may have not had the success that it had if Bram had treated it that way and if the company had been attacked for what they were doing. When his, his use case was I want him get Linux isos out to people more efficiently and with faster downloads and without having to host it and pay Buu's money to host Linux for people, which is also a ter perfectly legitimate use case. And I think in the context of that like for the people like we should learn from Samurai, if you were loud about this, they're gonna come after you, they're gonna put you in a cage and to build tools that are unstoppable, that aren't dependent on us to work and to dual purpose it, right? Like, you know, like build, build a compression tool that is just making your, your transactions cheaper by batching them privately. But you're selling a something that lowers your fees. You're not selling, you're not selling, you know, Ashigaru 2.0. Like you're, you're selling a, you're selling a fee minimizing tool. But build privacy in, you know, Lightning is not sold as a pay for drugs and you know, nobody can tell where your payment came from because it's private and you don't have a, you don't have a permanent payment history. Lightning is a payment protocol and it just has onion routing by default. The privacy is just built in and it becomes way harder to attack both socially but then also technologically too because people adopt it not because it's a privacy tool, but because it's the tool that does the job. And privacy comes as like a secondary, it seems like a secondary benefit. Like, I think it's important as people especially as we're buying live coding stuff and you know, working with these protocols is, you know, like we can build these things like white noise and, and you know, coin join protocols and all of this stuff and make it so that everybody has the credible, right, the credible ability to exit and that everybody has the ability to run their own. And you, you simply, you don't stop the idea of quote unquote censorship or I guess control, like, of being able of like limiting like access to something. You simply get around the problem entirely by having anyone able to run it. You know, it doesn't matter. It just becomes a problem of free association. Who are you willing to associate with and otherwise you don't have to invite everybody into your home, end of story. And, and the tools are easy enough and ubiquitous enough that we can all run them. And I think that's how we win quietly and persistently.

A (102:29)

Exactly. And a really good book that enlightened me in that aspect as well is the Second Realm, a book on strategy by Smuggler and xyz. Oh, it's fantastic.

B (102:40)

The second round.

A (102:44)

And yeah, you can get it by the way, at least an early version of it on Anarplex Sirian IO Aniplex was a group of hardcore cypherpunks like around 2010. Anarplex Syrian IO Sirian IO is the website of the Fedimand founder.

B (103:03)

Okay.

A (103:03)

And he has the anaplex prefix Syrian like the country. No, I'll send you the link.

B (103:10)

Send me the link.

A (103:11)

Okay, I'll put it in the show notes. But so anyhow, that is a site that references a bunch of amazing, amazing cypherpunk literature that everyone should read. Every Freedom Tech builder must read. And the second realm is part of this. And yet the realization is that first of all, we're right. Freedom is worth fighting for and we have a right to build the tools that help people protect themselves from theft. So we can rest assured on that foundation. But then on top of that, it's undesirable. A lot of people do not want that. And that is a massive threat. And you have to be aware of that threat. But then also that there's numerous strategies on how you can protect yourself in that situation. First of all, privacy comes in and if nobody knows, then nobody can complain or attack you based on it. And so good operational security and privacy best practices for the developers of Freedom Tech. It's absolutely Essential. But then even considering that there will be infiltration, there will be molds. They were in a lot of the projects I was working on. They're just weird random people that feel shady and ask stupid questions. And the likelihood of any halfway meaningful Freedom Tech project of being infiltrated is basically 100% especially if you have more than a couple thousand users. So prepare for that and assume everything is recorded. Despite building layers of defense and having a good privacy routine. Be careful.

B (104:48)

Other people are your biggest. Are always the biggest attack vector. Like. Like who you associate with. And they have the. The networks and the like. That is like. That's the thing that is always done. Like it is. It is how they infiltrate everything. And I think I've been. I think I've personally been a little bit naive about that and kind of like bitcoins around bitcoin and stuff for years.

A (105:17)

But yeah, because it feels like a family, right? Yeah, but that's. That's where they have an easy in. And if you're at a conference and. And someone comes up and you have a cool conversation with them that just feels like meeting a new friend. But absolutely every bitcoin conference has spooks. Very much so. And they're very active and they keep coming for years. A very prolonged attack or infiltration campaign is definitely expensive. But skillful agencies absolutely do that. And that has to be contented with plenty of. Doesn't mean that you should stop. Exactly. They got the money printers so they can outgun you and outman you. And so we're definitely in the minority and you have to adjust your strategy accordingly. And so after protecting your own personal privacy it is about detecting threats early and then delaying the attacker to extend the period where you get notified from the attack until it actually hits you. And preparing your capitals such that is mobile and can be packed up and to de escalate and to just leave because you're not going to win a gunfight with the state because they have tanks and missiles and stuff. So best to evacuate. Make sure your capital is preserved and you can rebuild in the future. Because fighting is costly for everyone involved and there's massive opportunity costs not just of you burning the money on bullets, but also on you not having the time to keep working on what you actually are supposed to be doing. And so having a good posture that removes any attack in the first place is important but then also preparing yourself such that you have a better chance of surviving an attack when it actually comes. And even better would be if you would thrive in it. And so to try to make a system that is anti fragile so that it actually improves in quality based on an attack. And like for example, Wasabi right now is, is way more resilient than it ever used to be in, in. In part because it got heavily attacked and, and, and was prepared for that and, and grew out of it.

B (107:36)

Dude, we brought up a couple of different things that actually remind me of this topic or that are deeply relevant to how a lot of people argue on it. And this has been kind of like the major debate in Bitcoin more recently is the whole opera term thing and, and the idea of like explicitly allowing or noting a space for generic data to be stuck into bitcoin transactions. And um, and I've, I've had so many conversations about it and I go back and forth because I, I completely understand. I really felt. The thing that drove me crazy, the thing that like bugged me more than anything else was that I thought we had the perfect proportional response to spam, is that, you know, the mempool policy was just a great way to just mildly disincentivize and that it's been pointlessly escalated to a point that we like, we can't really go back to mempool policy being the way that we deal with it or disincentivize it. But a lot of people on the. But then the idea of a fork also seems like really high risk and potentially out of the scope of the way to, you know, disproportional. Right? Like, is that like, oh, I'm. We're doing a whole lot more for something that may not actually be this existential problem? Um, some people do think it is existential. Some people think that like, you know, if a, an attacker, like if a, a funded government and you know, a lot of people say it's like really esoteric or melodramatic to talk about like the child porn problem, but has anybody been watching what happened with the Epstein files in the last couple of days? Like, like I'm, I'm increasingly less like, you know, if the government wanted to stop Bitcoin, are there any ethics or morals that I think they wouldn't just fill up 10 weeks worth of blocks with nothing but that? Like some of the images that I've seen, I'm just like these don't. They don't care. You know, like if, if that's an.

A (109:44)

Open.

B (109:46)

Invitation, basically, of course they'll use it. Of course they'll use it for the most horrific or disgusting things that they could think of. And so but you know, crazy stuff like that is already on the chain. You know, like, like how do you think about this adversarial serially? Because like another thing that you brought up was the whole, you know, you own your own node, you own your own computer or whatever. So like you can choose and you can do what you want. But Bitcoin is interesting in that way, is that it's a consensus system. So if somebody puts something in the chain chain and it's part of consensus, that's not really any up to anybody's choice anymore. You know, like, like we, to be a part of this consensus network, we all have to download the same thing. So I'm really curious what your take on this because I don't think I've ever heard it. I don't know what your opinion goes one way or the other or how you think about this either as an attack vector or a, a problem or a non problem. You know, like what's, what's, what's kind of your position in that.

A (110:50)

Yeah, I think it's a, it's a big problem because we've had basically two substantial fee spikes in Bitcoin, like and not sustained high activity but like extreme spiking behavior. And that was during bcash and then a couple times with, with the whole ordinal scheme. And, and so that shows that it does seriously harm the protocol or the users thereof.

B (111:17)

It at least has some sort of an effect. Right? Like there's, there's a meaningful consequence.

A (111:21)

Yeah, exactly, absolutely. But I don't see a good easy solution to that with just minor changes to Bitcoin. I, I think we need a quite radical change to Bitcoin in order to make a, a full solution to that where such types of attacks are not possible.

B (111:46)

And do you think that kind of a change is possible? Like do you, do you think there's a solution to it? Because a lot of people argue that there just isn't because, oh, you can wrap it in private keys and you know, all this stuff or whatever and there's like, there's like an element of like. Well sure, that's true. But like you know, like if I.

A (112:08)

Think with the advancement of zero knowledge and the massive scalability improvements that got to them, which by the way over the last five or ten years, the amount of progress in ZK is mind blowing really. It's like AI. It's like Bitcoin, like Nostr and so on. It just got so much better in such a short amount of time. And there is no ceiling inside we will be able to keep doing that for, and just keep making it better. So with this, we can fundamentally change the behavior of a Bitcoin like system to just be way more efficient. Like you can have a bitcoin system that has a blockchain of constant size, where anything that you need to validate a transaction that you're receiving does not depend on the number of transactions in the package past. Like right now, if you're receiving a coin, you need to remember every single transaction that ever happened or specifically the UTXO set. Right. And that is an unreasonable assumption. Nowadays we can, we can build a system without this assumption. And it. All you need from the global consensus is the double spending protection, which you can do with an accumulator at fixed, at fixed sites at the entire blockchain is like 80 bytes or something. And, and we can do that.

B (113:30)

Do you, do you think that's a update that's literally possible? Like, like we could, we could do this with Bitcoin as it stands today and absolutely.

A (113:41)

I mean the tech is not fully fledged out, but in, in theory that's totally doable. The very difficult question is more like how would we make that upgrade and how would we get the majority of users to use this? But ultimately that is possible, I believe. So the question is more on technical implementation of the details of it, an audit of that, and then a rollout into a new system. But very interestingly, I think that quantum computer FUD might be the catalyst where we're doing this, because the quantum signatures are just inherently so much larger, even in their most efficient form. That, that is a good opportunity to introduce a much more novel and performant and efficient structure to the bitcoin blockchain itself. I'm insanely bullish on shielded CSV. Shielded client side validation is one of those research papers that I read and just blew my mind.

B (114:45)

Can you send me the link to that?

A (114:48)

Yeah.

B (114:48)

Shielded client side.

A (114:50)

Shielded client side validation. Robin Linus, Liam Egan, and I forgot the third one. Maybe Jeremy came up with this. It's an extension of the RGB protocol, which is client side validation, but it adds more zero knowledge magic. Oh, I think Jonas Nick was part of the author there as well. So it adds the zero knowledge part to decrease the size of the transaction history on the RGB second layer. But it also introduces that aspect of a cryptographic accumulator specifically for the double spending protection. So what you can do in shielded CSV is you have a set of transactions, 100,000 transactions, whatever, and all of these transactions get their double spending protections from 30 bytes of the blockchain. And so for millions of transactions your block space footprint is 30 bytes, might be 80 bytes, don't remember the details but that, that stuff is not possible. And, and the RGB team had something like Prometheus I believe, which was their kind of proposal on how could we str. How would we structure the bitcoin blockchain in a full client side validation scheme. And it's extremely promising. You don't have to expose signatures, amounts, pub keys, none of that to the verifier. All the verifier needs to have is the cryptographic accumulator for double spend protection and that can be reversively accumulated until it is a fixed ever constant size. Now obviously this is an insanely big proposal to make the bitcoin and that's why nobody is doing it. But in theory it's possible. I'm more unsure of will bitcoin ever update to such a much more advanced and sophisticated system, but in the long run it has to. The bitcoin software has to update and to be maintained and to stay on the cutting edge of computer science, otherwise it just won't work. The Internet protocol didn't stop evolving at some point. It keeps being added and improved. The only interesting new part of Bitcoin is how do we keep that shared history and shared consensus of the UTXO set. And there's of course a bunch of drama there involved as well. But ultimately it's not impossible to figure this out. It's not easy, but we can make it work. We got vibe agents now, they can do it for us.

B (117:25)

So in, in your thinking, in your, in your mind, the solution is not even anything to do with like how we treat raw data in the chain, so to speak. But it's, it's a solution that makes it so that you don't have to actually have that data to prove Bitcoin's history. It's a far more radical approach. So it's not like bit 110, let's, let's just nick, stop, return or like hard code a limit. It's a far more long term and much broader potential application of like, let's change the nature of like how we need blockchain history to know for certain that, that we are on the right chain, that all the bitcoin is valid, the, the hashes of every block is valid and has been verified. And, and so you believe that technologically, like cryptographically we can actually do that, that you can know for Certain that the entire bitcoin history is verifiable.

A (118:30)

There's shitcoins that do that for like already three, four, three to five years ago. You know, shitcoins that have a constant size blockchain with millions of transactions a second. In theory that's totally like the, the cryptography exists now. That's, that's not the question anymore. The question is is it actually scalable? Is, are the security assumptions solid? Are the implementations well audited and reviewed? And, and how would we make like.

B (118:56)

How long like longevity is security, right? Is you know, the Lindy effect is how long has this existed without being broken before we, we decide this is safe cryptography with what is a new set of assumptions assumptions here again like.

A (119:14)

Over the last five years or so we have so many new security proofs and, and like mathematical audits of the crypto schemes that I have a decently high expectation that they're good. I mean I'm certainly no, no expert on this. But it's not like 10 years ago, right where, where we had trusted setups and, and shady ceremonies and so on. That's, that's gone right? We, we have publicly auditable, transparent, zero knowledge cryptography now that we have security guarantees that they work now sure, maybe those security proofs are wrong and there's somehow a bug in them. That has happened before for sure. Also the amount of people working on this now is, is wild. Like I mean that's thanks to Ethereum and a bunch of other projects that really went deep into in the ZK research and are doing a lot of cool things. And so one of the things that I'm not so happy with the CAT proposals and whatever is that they might take away the flexibility needed to deploy such a system. Specifically BIT VM will not work anymore in one of these proposals. I'm not sure if that applies to bitvm3 because there was another fantastic improvement to that stack just last year and I think another paper earlier this year. But so if like we right now have the theoretical path on getting a trustless two way peg into a sidechain that is a shielded CSV pool, like we don't need to make any changes to Bitcoin with that or for that in order for that to happen. It's just a bunch of engineering and protocol design and deployment and hard but, but doable. And so taking away the upper return for example, I'm not sure if it's up return or, or one of these things in, in the BIP would make that no longer possible. And that, that would be a shame.

B (121:12)

You're talking about Bitpoint 10, I think.

A (121:15)

Yeah.

B (121:16)

The soft fork that is intending to deal with spam is what you're referring to.

A (121:22)

Yeah, but I mean there's like, there's probably 10 different variants. Sure, sure, sure. So that's definitely something to consider. And like, there's good reasons for why Bitcoin core is doing the things that it does. Right. Like a lot of this work, especially on the mempool and relay policy and stuff, was made specifically to make Lightning better. And, and, and lightning is, I mean, is a very difficult protocol and we've gotten much better at it now than in the past. Like so many of the attack vectors in Lightning or the privacy trade offs in Lightning are now fixed, at least in theory, and many of them deployed as well. So that just took a lot of work. And we have to be careful to not throw the baby out with the bathwater and make a change that presumably would fix the spam problem, but then also take away a massive shot at scalability that we have. So it's really tough. And I'm not really sure what's the right way to get out of this mess, other than it will probably have to take a lot of pain until many people care and then can roll out a quite drastic change to Bitcoin, which I think is doable. We've been conservative in the upgrade policy of Bitcoin and to some extent I don't think we need to. You see what Monero and Ethereum, for example, or even zcash, they've made fantastically large changes to their consensus protocol and they're still here. So it's possible to make massive changes to a running blockchain system. Bitcoin chose not to do that in the past for good reasons. But, but if we could come up with a new protocol, a new way of doing it that is substantially better on all of the fronts, considering latest cryptography, I think we can get there. Then a much more radical proposed solution might actually be the beneficial one. Not one that just says, no, you cannot put arbitrary data anymore here because Madic butts.

B (123:27)

Madic butts.

A (123:30)

But instead we have this way that, that scales to billions of people anonymously and also decreases verification costs for everyone. That would be something that we could get more community consensus behind than just blocking up returns.

B (123:48)

And you know, maybe you don't know the explicit implementation details here, but would this be something that could, especially if it's like a client side validation, would it be something where you like, reorganize the way the blockchain works, but that it could actually be backwards compatible. And it just simply meant that after validating or having the data you like, how breaking of like, is this a absolute, no questions asked hard fork in your thinking?

A (124:22)

No, I think, I think you could get to something like this with a soft fork. It would probably just be a bit more ugly. Like Segwit. Like Segwit is a mess of a soft fork. It would have been way, way more elegant in a hard fork. And again, we can do hard forks like Monero is doing hard forks all the time. Like Bitcoin could do a hard fork. Now, there's reasons why we don't, but if we have a very propelling, compelling solution that does require a hard fork, but therefore is clean and properly thought through, then it's just a question of a careful rollout. But even soft forks are not secure if the majority of hash power is not using it. There can be all types of nasty attacks, especially in the activation period that would make it very insecure. And soft forks can be malicious. And you can change the 21 million cap in a soft fork. You can confiscate cards in a soft fork. So soft fork doesn't necessarily mean it's good and hard fork doesn't necessarily mean it's bad. The main reason why we chose to use soft forks in the past, I think, is because it's easier to roll out. There's less likelihood of the system behaving badly if we consider that a majority. But not everyone has upgraded. But with a long enough rollout timeline of a year or two.

B (125:45)

Yeah, I think it's about like, at least in my. Because I tend to be much, much more conservative in how I think about like what changes we make or how we make those changes. Because, you know, you want to do everything possible to make sure that somebody who has coins and somebody who is running Wallet or a node or whatever it is does not wake up one morning and they're gone or they're disconnected. They have to install some new piece of software to see the change chain or to, you know, because that's immediately an attack vector. Somebody's going to make sure that, that, that Google search results hits theirs instead of, you know, some other thing, you know, like this. I'm also just like philosophically not against a hard fork if it's something that's, you know, extremely valuable. Like something that would just like literally solve some major problems with a tradeoff that just is like perfectly reasonable or measurable. And, but I also would not, I would be like this needs to be to the point of like, like nobody is. Nobody right now I think is using like a 10 or 11 year old client, you know, like on the, on the network. And it would be like, okay, like we can see on the network that nobody is using one that is this old. This is our timeline, you know, like, like 11 year clients. This is an 11 year hard fork. We're going to go ahead and implement and have it ready and it will not activate until 2037. You know, something like that. But yeah, philosophically it's like it's a software upgrade. You know, it's just insanely difficult and insanely dangerous for a consensus network. But it also gives us plenty of time to be like, oh, we made a mistake. Oh we tested this and oh, there's a bug. And you know, when you, when you're doing a irreversible alteration, like the more time the better. You know, I think it's totally fine for Bitcoin to move slowly and carefully. So that's interesting though. That's interesting. I honestly the zero knowledge proofs have not been at the forefront. I've been so focused on. No, there's just not enough time to dig into every damn thing. But I, I do hear it mentioned quite a bit with various people who are, who are building things and again just like seeing ecash come back, seeing and blind signatures become like more ubiquitous of a thing. The, the final rollout now of state chains and of arc and like a lot of this stuff like it's just, it's crazy how fast it's moving but it's, it feels like, like it's like the world isn't aware and, and we also didn't. Nothing happened. This, when it comes to like the price or whatever like this, this previous like hype cycle has felt like nothing even occurred. You know, it was just kind of like a. Oh, there were ETFs and I was like. And like who really gives a about that? You know, like there's like new liquidity and like some legitimization of like traditional finance and bit bitcoin has happened. But like we've built so much stuff. We've built so much stuff and so many new tools and so many new protocols but like none of those have like blossomed yet. You know, they're all still like right on the cusp of something and like.

A (129:10)

Kind of in theory, but we, we've scaled to billions now. Especially if you consider with custodians like ecash like 100 billion, like everyone and their million AI agents can today use Bitcoin. Like, and that's not a joke, okay, we would have some rollout pain, but there's no fundamental blocker to that. And, and then with of course lightning with, with state chains now with ARC basically lightning channel factories, there's, there's substantial improvements on how to move Bitcoin without custody already now. And then as I said, right, with, with bitvm can have trustless bridges and then have some crazy zero knowledge blockchain system and that brings us actual scalability of transactions as well. These are all not high in the sky anymore. These are all pretty reasonably achievable engineering problems that just need some time and thinking power, which again, now we have agents. All of our devs got 10 times more efficient and probably also a hundred times more expressive that they can actually articulate what they mean. And hopefully, presumably that will lead to greater common shared understanding of things. Because a lot of devs are just autists that cannot really talk well or communicate in a proactive productive way. And this type of translation is maybe also going to be effective. So I'm, I'm insanely bullish on, on humanity getting smarter and becoming more advanced with these technologies. It's, it's just a question of engineering time and you know, we, we just do that.

B (131:04)

So I know I've said this like five times, but just like what a wild time to be alive, especially with like all this Epstein stuff dropping. Like, I think this is kind of like a, an example of, of the breakdown of all of our old trust models and, and how we associate. This is the end game of completely changing all of our networks, of all of our communities, our associations, like into an online world. And the fact that at the exact same time that kind of the political consequences are coming to a head and there's enormous amount of like political chaos and not only disruption, but uncertainty in kind of the global power landscape at the exact same time that the Internet is coming to fruition with its breakdown of establishment structures. We're undermining the thing that's already being undermined with NOSTR and Bitcoin. Like this is not Bitcoin's disruption, right? This is just like the continuation and extension of the Internet's disruption of old systems and the inability to hide something terrible from people because of the way information moves. And we are already moving like an order of magnitude forward. It's like, oh well, we're also going to have a social media that can't be censored, where you can't shadow ban this post because it's talking about the wrong person, where you can directly monetize every anything and everything that you want, where you can freely associate with anybody, where you can also exchange those peer to peer, which you have encrypted group messaging that is, that is decentralized and peer to peer that you know, is not dependent on a company, they could be attacked or backdoored with the next update, et cetera, et cetera. And like, it all feels like, feels like it's coming to a head and the exact same time it feels like it's all beginning, you know.

A (133:14)

Yeah, yeah. Changes in the air and, and we have technologies now that we never had before, including on that front of, of actually checking what the politicians are doing. Like WikiLeaks is a godsend for that. They pioneered this and they're a cypherpuck as Fuck, right? Like WikiLeaks is the core part of the cypherpunk movement of like, we will protect ourselves from being spied on by the slave masters and we will spy on the slave masters and get all their dirty secrets and publish them so that the whole world will see. And the strategy works, it works surprisingly well on both fronts because the truth wants to be free ultimately. And then you as an individual are like, you have a much more favorable trade off of actually keeping yourself private just because if you're a good person, you don't piss anyone off too much and then nobody spends those resources to try to spy on you. But obviously the politicians who piss off all the people because they steal so much, they will be under immense scrutiny, not only by people in their constituency and country, but globally. And there is this decentralized journalism opportunity where anyone can do this research and then anyone can publish their findings and collaborate with those who talked about the research. And we just need more of that. And again, that's why white noise environment is so important. Because for any of these important things, we need to coordinate with each other, we need to communicate and we need to be able to safeguard the release of information to a certain point. And we need that not only for chats, we need that for so many different things. And that's where I'm so bullish again on marmot as a protocol, because it's one of those foundational building blocks, secure communication of a group of people that is required to take this to the next level because we can solve all of these crazy things like bitcoin scalability and so on if we have builders that can securely communicate about the solutions, then it's just a question of time. We never really had the secure building infrastructure that was needed. And so that means that we have the problem of not being able to actually deploy these solutions at scale, you know, because all, all the infrastructure of all the freedom tech people is, is going to be compromised. Like if, if you have a, a, a Google account and you're a freedom tech lover, bad news. But it's been compromised. They have it subpoenaed and they, Google doesn't even have to tell you about that, not for years after. Right? So, and, and that's only in the official part of the, the of their surveillance machine. They have a hundred x larger black budget where they don't even have to care about all of this stuff and that it is a massive problem. And, and you are being attacked. And even as, as a rather boring person, you're being attacked. And the good thing is you can actually put an end to it. Like the cypherpunks have a, a fantastic strategy of shining light on the evil deeds of, of the politicians, spreading the word about this so that everyone knows about it, building the solutions so that they can no longer steal from you and spy on you and then distributing those solutions to all the people who are pissed off enough to put an end to it. And we can ship it to 8 billion people with the click of a button. It's a technological informational deployment to this problem. And I think we're very close before reaching a tipping point where given that the technology is available, we will see an insane amount of societal change. I really think that culture and society comes downstream of technology. We are different people when we have different tools. And if all the tools that we have are designed by slave masters and designed to exploit as much from people as possible, then obviously the resulting society will be abysmal and just embarrassing. But if we have technologies that do protect individuals that value property rights, value privacy, and have sane secure defaults for everyone by default, then the resulting society is so extremely difficult. And thankfully we don't have to change the genetics of humans because that would take forever. We just need to change the technology. And that's an update to your app that is all of a sudden something that we can make meaningful, long lasting, actual, tangible improvements to our situation tomorrow.

B (138:15)

So you think that a lot of people I think are pretty pessimistic. Like, and I, I go back and forth. There's, there's a lot of like short term pessimism and like where the EU is going and how, like, attacks are happening and stuff, and also where the US is going from a political standpoint. And, like, I don't know of anything that hasn't really just gotten worse over the period of my lifetime from, like, how. How much the government's spending, how much they're invading privacy and like, all this stuff. But, you know, this discussion has largely been a very optimistic one, and at the same time that everybody is like, even. Even AI. Like, I have, like, fantastic uses of AI I love. Like I said, I'm addicted to Vibe coding, but it's also like a pretty big surveillance nightmare right now. And in a lot of the platforms and things that we're on, I feel like we're. We're on a runaway train toward all of these things getting worse when applied to the average person. But then at the exact same time, we're building so many solutions in this little corner where things are getting aggressively better when it comes to the tools and the stuff we have at our disposal. But what's your timeline for your optimism? Like, how. How quick do you think some of this actually turns around and. And is it a. It keeps getting worse, faster and faster and worse and worse until something breaks, and then it's like. It's like a dam, you know? Or is it a slow buildup of kind of like two parallel universes that. That, you know, in 20 years, 30 years, we still just have this whole world of sheep living in sheep land, and we have a world of kind of Bitcoin. Like, you know, they just. They just diverge and they cannot. They're fundamentally incompatible, and they just kind of like. We kind of like quietly live in these two little worlds. Like, you have your mask world and then your. Your world where you can actually go be yourself. Like, what do you imagine the future to be? And. And how do you imagine we get there?

A (140:26)

I had an interesting kind of change in character there because, you know, I started out as an economist and really went deep down the Austrian rabbit hole and then got incredibly depressed by realizing of how much destruction of capital is actually happening in modern society. Like, it's. If.

B (140:44)

If you.

A (140:45)

If you really deeply meditate on that, it's. It's soul shattering.

B (140:49)

Like you.

A (140:49)

I. I can't fully even try to realize that because it would hurt too much. It's really fucking bad. And that Austrian economics alone did not give me enough hope to fight that massive realization of that depression. And that's where the cypherpunks come in to a larger extent of, yes, this is a problem. But look, we have cryptography, this is magic and distributed systems, strong encryption, all of these things where no we, we have the technology to make a meaningful change in. In that intervention and theft. Like strong encryption holds. It holds even against the nsa. It holds against the most incredibly capitalized government or private agency. They cannot brute force decrypt your database without the password. Like that's a fact. And then the again the punk part of cypherponseft. You can't just do things. The mental model you gave of this is a dam breaking the herd of sheeple. I think it's much more fruitful to look at this on the individual scale. If you want to protect yourself from all of these types of shenanigans, you can take meaningful, manageable small steps today and improve your level of freedom. Decrease the level of harassment and increase the timing between you get stolen from to incredible high amounts. So the dam doesn't have to break for the world. The dam has to break in your head. When you pledge to no longer serve, you are at once free. Something like this is from Bortel or some French writer from the 1700s was worrying exactly about this of like why are so many people partaking in their own suffering and slavery and prolonging it much more than it would absolutely. That it would be necessary and oh, I don't know the answer to that. Humans are weird. But what I do know is that you can just stop hurting yourself. And that requires the realization that you are in the first place hurting yourself. And that's what is lacking, I imagine. But thankfully here, as the government increases the pressure and the trickery and the thievery, that pain will become so large that you won't be able to unnotice it. And ultimately the socialist system will fail for very obvious economic reasons. Their systems of control and manipulation will misallocate capital on such a gigantic level that they're simply not able to solve even the most basic problem anymore. And we've seen that for the last 30 years. We're especially seeing it in the last 10 years. And that will continue to be the case where the actual, perceived and real cost of the misallocation that is happening is so astounding that it will get worse for all of the people who continue participating in the system. But on the flip side, the cypherpunks have built the technologies that decrease the cost of exit to almost nothing. All you need to do to end the Fed is to sell your fiat and stack some gold, stack some bitcoin and your Inflation is gone. For you. Price your contracts in Bitcoin and inflation is gone. They use Bitcoin privately and most of the theft is gone. And these are real easy, doable things. It's literally downloading an app. Anyone can do that. And sure it could be prettier and more scalable and whatnot. We can improve it in a million ways. But ultimately it's a technological problem or a technological solution to a societal problem. And I very strongly believe that we have the tech available already now. It works wonders. I see a million ways of how it can be easily improved in the future. And so that is what makes me so optimistic, realistic of how bad the situation is. It's unspeakably bad. But maybe because it is so bad, we have an even more awe inspiring solution of just stop hurting yourself. And within the shortest period of time we would have a rebound. Just imagine if tomorrow everyone would stop stealing from each other and where we would be in a year, in two years, in five years. The things that we could work, that we could achieve are so unspeakable and that is again so painful because you see what we've lost in the last hundred years of misery and suffering and war and slavery. But it also shows you that we can improve the situation for everyone so massively in a very short amount of time if we have that aspect again realized and deployed to people.

B (146:00)

Next few years are going to be crazy.

A (146:04)

Yeah, it's like exponentially more so.

B (146:07)

Yeah, I mean this year just got started and it's crazy.

A (146:13)

We're just getting started.

B (146:15)

It's been, it's been an insane year all week, man. And you know, it, it reminds me, the way you framed it there reminds me of. I read Thomas Paine's Common Sense on the show and then there's Parker Lewis did the piece on it that made me go back and read it. And I cannot believe how eloquently he kind of like framed how it was. And it was such a personal kind of story too. Is that like, you know, the nation is like a person who has continues to procrastinate about something that we know has to happen and we know have to do. We, we have to do. But it's not painful enough yet to break down and just do it. But it's common sense. We know that, that we have to get here and it's going to have to happen, but we're just putting it off. We'll just do it tomorrow. We'll just do it tomorrow. And that essentially the pain, the friction, the presence of the problem has to become so clear and in our way that it leaves us no option, that we cannot procrastinate it another day. And we're getting, we are getting there, yes, fucking getting there Again, like the.

A (147:52)

I, I'm not so optimistic about you. I think that the, it, it will keep going wrong in, in the same direction on a macro level for a long time, I hope.

B (148:05)

On the individual level, there's the ability to exit. I think that's, that's kind of the.

A (148:10)

Key and that's, that's the crucial part. And you, you, you can just stop and you can help your, your friends and family to stop. And sure, that's hard, right? We've all tried orange pilling friends and suck at it. But there's ways that you can get better about that. And then it's like once you're on the other side of making that switch, it's so obvious and so easy, like so effortless that I know it's doable, I know it's not expensive. I actually know it's profitable. Very, very profitable to do so. It will happen like the, the economic incentives are, are clear. More and more people will realize that. And we don't need 100%. And one of the amazing pieces you read was about the Remnant. That is so true. We, we, we don't need everyone. I mean, it would be great if eventually everyone would stop stealing, but if 3% of the people would stop stealing consistently and stop engaging with anyone who steals, that would change the world in an incredibly short amount of time. So you can be part of that and just be actually consistent on it. That's the difficult part. I think that's the insane trick that was pulled on all of us with fiat money is that we are all tricked into stealing from each other. It's so easy to say, oh, the banks and the governments, they're printing money and stealing from me. Yeah, that's true. But behind you in the chain of Cantillon effects inflation are millions of people not even born yet. And you're actively stealing from them. And that's just bad. That's wrong, that's stupid. And you don't have to do it and it's not good for you. So you can just stop and perceive a real tangible improvement in your life tomorrow. And there's many different ways that we're currently doing that of just bowing down to people who have no right to lord over us. And so one by one, remove them from the equation and life will be very different for you, for your Family and for your friends, even those who are still stuck in that first realm. And in that old mindset, they will still benefit of the fact that you're more productive, more lively, and then. And that lifts them up, even if they're still stuck in this old system. And after, they then perceive enough pain from the old system and see that you're thriving on the new system. I mean, to many, it will, you know, release a lot of envy and jealousy, and that's a dangerous position to be in as well. But then, on the other hand, many will realize that they. They can achieve the same quite easily. It's not rocket science. It really isn't.

B (151:08)

It's a good place to stop, I think. Dude, thank you, as always. It's good to hang out, man. It's been a. It's been a while. It's been a while.

A (151:16)

I feel like.

B (151:17)

Yeah, I hadn't seen you at conferences or anything. Of course. I've been, like, half not going to them lately. I've been trying to travel as little as possible. Are you going in any this year?

A (151:29)

Yeah, I have a couple planned yet. They're all kind of happening. Spontaneous.

B (151:32)

Okay.

A (151:33)

I always keep all of the conferences in the calendar, and then, you know, just random. I can make.

B (151:39)

Let's go. Let's go. Let's go to bitlock. Boom, man. Let's do it. All right.

A (151:43)

Yeah.

B (151:44)

Hell, yeah. Hell, yeah.

A (151:45)

That's one that I'm so missing.

B (151:46)

Well, don't forget to. Don't forget to hit me up if you. If you go, I want to hang out, get a drink, and bullshit for three hours in person next time we're in the same place, right?

A (152:01)

That would be good. Dude, I'm really happy with what you're building with Peer Drive. It's such an obvious problem that needs solving. And I love how you've ranted on the show for years on how this problem has to be solved, and eventually you give up. It's like, okay, I couldn't convince any one of my audience to fix it, so damn it, I'll have to fix it. And I think that's the. That's the right mindset. That's what we.

B (152:26)

Dude, I was talking to Carlos from Simple Proof, and I was like, you know, like, at this point, I'm so invested in this stupid thing, and we've rebuilt it. We're on, like, the fifth rebuild, and because we keep running in some problem where I'm like, I can't create the user experience that I want, and I'm like, no, no, we got to go back and figure this out. And. But I was, I was talking to Carlos and I was like, you know, when I got into bitcoin, I. I immediately lost all my money, right? I got in at like the hype of the hype in 2011 and then it just plummeted and I lost so much money in it that, you know, I cried and I threw up and I was like, okay, like, I'm not gonna sell out because I've got like 70 bucks left, you know, like, I'm not gonna sell out of this on the same ignorance that I got into it on. And so I decided that I was going to before I made any decision. I'm just gonna stay in my. I lost all my money position in my magic Internet money investment. And I was going to find out I was going to be able to explain better than anyone else why it was a dumb decision and why bitcoin wasn't going to work. And so, like, I started reading everything I could get my hands on, right? And eventually I just came to the conclusion that it was going to work because I felt like the, the best of the best arguments against it. I still felt we're missing some major pieces. And so after years of doing this and bitcoin slowly, like, coming back to life, I was like, I, I still think this is going to work. And so I've kind of applied the same logic in a sense to Pear Drive, is that I am so convinced that we have all of the tools and that this can work, that I am just going to. I am not going to stop until I'm either dead fucking broke or I am absolutely 100% certain that it cannot actually be built. Otherwise I'm just gonna keep going even if I can't afford anybody else to do it. I'll just vibe code for three hours every night and see what I can put together and if I can, Lego block, modular rise, whatever the hell I want, but it's got to be possible. It has got to be possible. There's no way that you can't user friendly, peer to peer this thing, you know? Yeah. So we'll see. I'm. I am, I am a hundred thousand percent gonna look closer at White Noise again and the Marmot Protocol because I hadn't really. That hadn't really been top of mind of like, how to deal with messaging and stuff on it. But since we already have Moster and everything plugged into this, this might actually be really easy to apply with what we're doing. So thank You I think so we.

A (155:23)

Should, we should chat properly about like how exactly you. You want to use it or what exactly you need. But, but ultimately if you want to be able to send a message or a file between a couple people encrypted so that it's actually long term encrypted and, and you know, rotating keys, et cetera and you're already doing that with Nosto events then Marmot is I think a no brainer for. For you now. It doesn't solve everything. One other thing that I absolutely want to have is I remember 12 words in my head and I get all of my important files back that use cases. I also absolutely want to have and Marmot is not the right thing for that. Marmot would be how do I get all of my data over there with a secure transport layer and then how you figure out your backup long term solution. That's a different protocol but for, for getting that handshake and especially for I want to send the same file to a hundred different peers that might be really interesting.

B (156:34)

We should, we should get something on the books. We should actually like sit down and do kind of a building or brainstorming sesh because and actually got a message just a couple of minutes ago of like we're demoing some changes that have recently made so exciting. There's a little bit of movement right now but because I think we're, we're momentum is back in after Iran's Internet shut down. I think we'll finish the build as it is and, and then kind of go back and, and look at that sort of thing because I think we have like a, a simple tool that does the job and I think once we have the kind of foundation we can go back and be like okay, how do we make this a little bit better? How do we improve the privacy of this a little bit? How do we, how do we, you know, tweak this little thing without changing the, the API or the, the RPC commands.

A (157:27)

Have you like released anything yet or is this.

B (157:30)

We have so hope. I worked with a developer for a long time and we have a kind of a proof of concept. It's still missing a couple of like really important primitives, but it does the job. It does the job to the point that I just delivered 1500 transcriptions of my podcast from my Linux machine to my Mac so that Jarvis so my bot could go through it and make a lot of assessments and stuff. It's just, it's the easy way to move enormous amount of files from one computer to another, when my brother is traveling or whatever, he's like, man, I want to watch. Watch the Matrix. And my wife has never seen it. Can. Can you send me a link? And I'll just send him a pair drive key. And he's just like, thanks. And he's watching. So we, we have it. We know we can do the core of what paradrive needs to do, but like I said, it's missing a couple primitives. It's missing like a good backup system and the key attestation system that I really, really want to. To really kind of like make that user experience kind of like, oh, this is just easy. So that's why we're rebuilding again. But because we found out we kind of have to put at the bottom, we can't. We can't bolt it on top. Like, I kind of thought we might be able to, but. So you can just go to the Pear Drive GitHub repo and there's a CLI tool and there's the core. Still rough around the edges, still has some, some bugs or whatever. But you can use it. I use it all. I use it every single day. It's open right now, actually.

A (159:04)

So, you know, one. One of my favorite kind of mantras is that we deliver shippable intermediaries. I think there's a lot of power in that statement.

B (159:14)

Oh, that's a really. Because that's a really cool statement.

A (159:17)

Yeah, you have to deliver, and if you're not delivering, then what are you doing? And they don't have to be perfect, you just have to be shippable. You know, you can be totally embarrassed about your first release. In fact, if you're not embarrassed about.

B (159:30)

Your first release, you release, you waited too long.

A (159:34)

This has to be shippable. And it's fine if it's only shippable as long as it's an intermediary. Just keep grinding on it, keep getting feedback, keep making improvements, and eventually those will be dug out. But it's so easy to get stuck in this kind of perfectionist loop. And I think you're totally imperfectionist hell right now. I'm sure you could have released it like a year ago probably, and it would have been fun.

B (160:01)

It drives me crazy. I wanted to be right. I wanted to be right, man, you know, like, yeah, I get you, but.

A (160:07)

It, it cannot be right without deploying properly and without getting feedback from users.

B (160:13)

That's why it's out there. That's why it's out there. Even though I know, like the first. This version is a proof of concept. But I was like, all right, somebody's gonna, at least they can look at it and like, kind of get a sense of like, where we're going and stuff. And, and, and it makes me feel less like an idiot for still just beating my face against this thing without anything out there. You know, I can at least point them to the current repo and, and I can just say, oh, we have a new version that solves a ton of the problems and, you know, adds a couple of really important primitives and then I can show an app on top of it. And I think that's really it. It's not going to move to any substantial degree until I show the kind of experience that I want to be able to do with an app. And I think, I think it's, I think it's going to be fun and it's going to be something that you've just never been able to do with files. So we'll, we'll see.

A (161:02)

But now it's, it's, it's really good. But you know, like, I feel, because we're in also like the fifth iteration of White Noise and the front end app and it's just pain, you know, to have to throw everything out literally four or five times. Oh, God.

B (161:19)

I know.

A (161:20)

We're just now trying to get feature parity of the old version and. But it's like, it's so much better than before. It's so much more pretty. It'll work so much faster, more efficient. It's going to be great. But I keep telling people that for like the last year and it's kind of embarrassing, but it will be worth. Just takes time to do properly engineered codes. That's just hard. And sure, we can get more efficient and use all of our bytecoded agents to make the progress better and we'll get there. But it's just hard. So many things worth working on are just really hard.

B (162:04)

It's kind of axiomatic though, isn't it? You know, if it's worth working on, it's. It's because it's hard. Like as soon as it's easy to do, it's not really a worthwhile problem because it's too. Once it's too easy to solve, it's already been solved in a hundred different ways, you know. But you're 100% right though, and I feel you. Yeah, Very, very much so. Dude, thank you. This is, it's always good. We'll close the episode again. This is what, like we're the last Lord of the Rings. Right, we're going to end the episode five times.

A (162:40)

This was one of my favorite podcasts I was on so far was with the Honick Ducks. It's a amazing OG German speaking podcast. They're doing it like non stop since 2012 or something. Oh.

B (162:52)

Oh, geez.

A (162:53)

And, and what I love most about them is their intro and outro music. It's just like beautiful piano jazz with like, you know, someone pouring you a cognac or something with fire crackling in the background. It's just cozy. And so we had this conversation and naturally, like throughout the conversations, five times the closing ceremony came up, but we kept talking again and again. And so what they did in post, which was amazing, is too to add the outro music, you know, as we were going, and then DJ Skip as we went into the next rabbit hole for like five times.

B (163:28)

So love it.

A (163:29)

Those were the best.

B (163:30)

That's amazing.

A (163:30)

Those are the best parts.

B (163:31)

That's amazing. Well, hit me up. We'll. We'll get something on the schedule. Because I do, I do actually want to talk about this. It'd probably be. I say probably good timing is like three months. I mean, three weeks to a month or something like that, depending on where progress is. Um, but I'd love to sit down and go over white noise for. Just to start and then, and then maybe, maybe we can kind of like walk through peer drive and, you know, what I want to do with it and how the new build is working or whatever. I'd love to get your feedback and tell me where, like, this thing's stupid.

A (164:02)

You're. That's a terrible decision.

B (164:03)

Let me, let me know, because that's the only way I'm going to make it better is when people smarter, who have built enough things tell me what I did wrong.

A (164:13)

Yeah, that's always super helpful. Like an hour conversation with the right person can save you a year.

B (164:18)

Oh, God. Dude, I. Oh my God. Dude. The number of times where like we were beating our faces against something for like four weeks and then I was just like, matthias, man, what do you. What should I be doing here? And he's like, oh, blah, blah, blah, blah, blah. I'm like, oh my. Oh my. Oh my God, you gotta be kidding me.

A (164:41)

Yes, exactly. No, but like, that's. It's true. And I think that's ultimately again, why we're winning if, like, we have people to talk to. That's the cool thing about the free software space. There's so much talent here and so much enthusiasm. And motivation to build awesome stuff that it's. We can't but win.

B (165:01)

Like it's.

A (165:02)

We just have way too much fun. Cool, awesome tools that we can deploy at scale like right now. And there's a million ways that we can make them better and now even non developers can just make them better by themselves. Is it is. Is a crazy. So. Well, sweet. Yeah, I'm really bullish on this.

B (165:23)

Yeah man.

A (165:24)

It'll be all good as a trip.

B (165:26)

Exciting, exciting freaking times. You know, nobody can say that we didn't. We didn't lift a real. Nobody can say we lived through a boring time. Right. Certainly can't say that shit. All right. Anything to share out, people to tell you, where to find you, what to check out and all that stuff.

A (165:48)

Yeah, go check out Marmot Protocol. It's really cool. We have no libraries in Rust and Typescript with language bindings to Python and C and whatnot.

B (166:01)

Oh, nice. Okay.

A (166:02)

Bunch of good ways of using that. By like the end of the month, like February 26th, the developers will tell me that the seventh iteration of the Matrix will be very efficient. So that will be rolled out by then. Let's see. They promise good things and then just try it out. Like I mean try out white noise, especially the new UI we're building. It's going to be really sexy. So it will be a good looking channel.

B (166:29)

Good word, sexy.

A (166:29)

And the best thing is that you'll be able to add the same functionality into your own app and talk to all the white noise users. Amethyst Primal. All of the cool clients are actively or not actively working, but they're considering how they'll do it and they're basically just waiting on us to give them the green light of start building. It's ready. We're really close to that. So this year we will have encrypted NOSTR messages totally figured out. And it'll scale to thousands of people in your group and you can have thousands of groups in parallel and it will be amazing for chatbots. This is like one of the things I'm so bullish on is to make a bunch of Mar bots or Marmot bots. They will be so good because they will be first class citizen. Right? You don't. Dude, the way that, the way that openclaw works right now is so ridiculous. You literally have to give it access to every single chat that you were ever part of.

B (167:28)

Oh my God, it's so wild. It's so ridiculous.

A (167:30)

Something encrypted like Signal or WhatsApp. They see every group.

B (167:34)

Yeah, you.

A (167:35)

You are sending all of your chat history to an AI service provider for GPU inference. It's fucking wild. So. And the only reason why that is is because you can't have a second signal account because it's tied to your stupid phone number number. So that's just totally gone. That's an obsolete problem. And that makes me super bullish. And yeah, it would be good. So, yeah, check out my website towards liberty.com. recently I updated it and I vibe coded it, so now it has all of my Noster events, like all of my posts and my hundred or so.

B (168:10)

No kidding. Blog posts towards Liberty.

A (168:12)

You know, when I did that, I mean, I got like queried all of the relays to get my events and it turned out that all my insightful knowledge on nostr for like four years or something are 16 megabytes.

B (168:24)

Wow.

A (168:25)

That, you know, make me feel humble.

B (168:28)

Nothing. That's crazy.

A (168:30)

That's nothing. You're just a small blip in the universe with 16 megabytes of insightful data. And on that website, you'll also find my book, which is out in version 0.2, for hopefully getting some feedback. What I'm trying to do with that one is to bridge the cyberpunk tradition and the praxeology Austrian economics tradition, because I think both of them are just talking past each other and there's so much synergy to be had from both frames of mind. So if you're not yet a hardcore Rosbardian mcap, then this might be a good intro for you. And if you are already, but you have no idea how cryptography works or what we can build with that, it will also be good for you. So, yeah, that's kind of the culmination of the last 10 years or so of me thinking about these problems on both angles, hoping to build the bridge in between that. So give me review now and. And I'll still make some changes and the third version before it goes into print, probably in the middle of the year or something.

B (169:37)

Oh, yeah.

A (169:38)

Um, so now is a good time for that feedback.

B (169:40)

Sweet. All right, man. I'll link to all that. Oh, yeah.

A (169:43)

The book is called the Praxeology of Privacy.

B (169:45)

I was about to say. I was just looking. I was just looking at the Praxeology of privacy. I like that. Um, I didn't realize you have. You have so many. You write articles, like every freaking day at this rate or whatever. I only see like one or two come across my feed. You got 16th, 17th, 19th, 19th, 20th, the hell you writing so much for 21st. That's awesome.

A (170:08)

That's one of the things of, of like, like sometimes I'm bored and then, you know, sometimes I just let go of myself and be actually bored for a while. But oftentimes I'm like you. You're not allowed to be bored. Like we have the world to fix. You do something productive. You have so much to do.

B (170:30)

Get to work. Man. You know, I tried to sit down and like watch. I'm. I'm here by myself because my family, it's snowed in in back at her in laws and so I'm like, kind of like. It's weird not having like kids running around and everybody yelling and stuff. And I like sat down and tried to like watch something last night and I was, it was like three minutes and I was just like, man, I just want vibe code. And I vibe code until like one o' clock in the morning. I was like, this is way more fun. This is way more fun. I just don't even, I don't even care about these shows anymore.

A (171:06)

Yeah, exactly. It's really cool that it as. As you said, man, that was so good. It. It makes, how do you say, addiction profitable. No, that wasn't it. But something like that. You had a banger line. I have to relist.

B (171:19)

It addicts me to production. Yeah, exactly. It gamifies productivity for me.

A (171:27)

Yeah, like, and that's a really good insight.

B (171:30)

Yeah.

A (171:32)

And yeah. Game changer, dude. Yeah. Yeah.

B (171:36)

All right, man.

A (171:38)

Bullish, bullish, bullish. Let's keep bullish. Bullish as hell.

B (171:42)

All right, dude, thank you so much. I'll catch up with you. Shoot me any links like the Aniplex stuff or whatever that you have. I'll have it all in the show. Notes for everybody.

A (171:52)

And oh yeah, definitely read the second drum, man. If you haven't read it yet, it's a must.

B (171:58)

Oh, it's Jarvis. Jarvis has already got it on my. Don't forget me. Don't forget to remind me. I typed it in there immediately. I was like, oh, bro freaks. A fiction book on like. No, you. Yeah, 100,000%. I will read the shit out of that. Hell yeah. Thanks, dude. Thanks for coming on.

A (172:15)

Yeah, this was fun.

B (172:16)

Good.

A (172:16)

Looking forward to the next one.

B (172:20)

Okay, maybe it's working this time. Okay, so a bunch of stuff to check out. I have a whole bunch of different links he sent me. So we've got Anarplex. That's a Narplex. Sirian. S I R I O N IO obviously the link will be right down in the show notes. This has got the Second Realm book which I'm pretty stoked about. And then he also pointed me to and this may have been after the show, the True Names story, which is a great kind of like little cypherpunk history. It's an OG piece on that. And I now that I think about it, I can't remember if it was just Dur Gigi's piece that we read about it or if I actually read True Names on the show show because I was thinking while I was talking to him that I had not read it. But I am now 90% sure I have. But it's been so long that I may have forgotten a lot of it. Whatever. It's a good thing to refresh and I'll have the show the link in the show notes as well as for the Marmot protocol, if you're building, especially if you're vibe coding or you're trying to build stuff on Noster, anything like that. This is super critical and I love the idea of the decentralized encrypted private messaging and like what we can actually pull off with this that doesn't rely on centralized servers to keep working. And we can have any client, any tool, maybe even Pear Drive. We're definitely going to look into it that can communicate with this and use this open standard, the Second Realm book. And then also I've got the the book Sedition, Subversion and and Sabotage a three part solution to the state and thinking about how to build and secure your own freedom and thinking about your life as an adversarial environment and building it for resiliency. And we have so many tools to actually do this. That's definitely something we'll be digging into. Link to that as well. Anything else that he brought up. I will try to make sure that I have the link in the show notes. If I missed something, please reach out. I will try to get it up as quick as possible. Any questions you have, feel free to hit me up. Don't forget about bitcoinaudible. Com. There's so many great resources up there. We've got the entire catalog of episodes now and that should do us. Follow Max subscribe to this show and share it out with everybody that you know and I will catch you on the next episode of Bitcoin Audible. I am Guy Swan Man. Until then guys, that's to my two sets. Sa.